Joinpoint Help Manual

4.5.0.1

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Joinpoint
The Joinpoint Regression Program is a Windows-based statistical software package that analyzes
joinpoint models. The software enables the user to test whether or not an apparent change in trend is
statistically significant.
Joinpoint fits the selected trend data (e.g., cancer rates) into the simplest joinpoint model that the data
allow. The resulting graph is like the figure below, where several different lines are connected together
at the " joinpoints."

Joinpoint also allows the user to view one graph for each joinpoint model, from the model with the
minimum number of joinpoints to the model with maximum number of joinpoints.
You may use any software to create data files to be analyzed by Joinpoint (see Input Data File). The
SEER*Stat software provides an easy mechanism for creating data files containing cancer rates or
counts that can be analyzed by Joinpoint. The controls on the Joinpoint Specifications tab will be set
automatically if you load data files exported from SEER*Stat. To learn more about the SEER*Stat
software visit the SEER*Stat Web site at:

http://seer.cancer.gov/seerstat/

Related content

System Requirements
Citation
Technical Support
Setting The Joinpoint Parameters

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Setting The Joinpoint Parameters
When using Joinpoint, work on each tab in sequence - from left to right, and from top to bottom within
each tab.

Input File Tab

On the Input File tab you will describe the data file to be analyzed by the Joinpoint program
and select the options for the modeling of the data.

Method and Parameters Tab

Use the Method and Parameters tab to specify the modeling method, the criteria used to
determine the locations of the joinpoints, the model selection method, average annual percent
change parameters, autocorrelated errors options, and whether to exclude any cohorts from
the analysis.

Advanced Analysis Tools Tab

Use the Advanced Analysis T ools tab to set options for pairwise comparisons, Jump
Model/Comparability Ratio Model or multi group clustering. In order to set these options you
need to have selected the radio button for:

You can save all your settings in a Joinpoint session file by selecting File > Save As... from the top of
the main window and choosing a name for the session file.

When you have made all of your selections on the Joinpoint tabs, click to execute the Joinpoint
session.
Note, the session file is saved as part of the output file, so you can select Output > Retrieve Session
to retrieve the session that was used to create the output, even if that session file wasn' t specifically
saved.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Input File Tab
The purpose of the Input File tab is to set up your regression analysis by identifying the input data file,
selecting the model type, selecting and defining the dependent, independent, and any by variables,
selecting the heteroscedastic errors option.

Related content
Input Data File
Dependent Variable
Independent Variable
By-variables
How to Compute Age-Adjusted Rates in Joinpoint
Log Transformation
Heteroscedastic Errors Option (Weighted Least Squares)
Delay Adjusted Statistics
Data Grid
Data Files Containing Cohorts with an Unequal Number of Observations

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Input Data File
The input data file contains a dependent variable and an independent variable or covariate. The
dependent variable values can either be directly provided in the input data file or Joinpoint can
calculate the values if their components are provided. For example, Joinpoint can calculate age-adjusted
rates if the associated count, population, standard population, and adjustment variable (typically age
group) are provided.
The dependent variable may be an age-adjusted rate, crude rate, count, percentage, proportion, or other
numeric value. The independent variable is typically the year. The data file may also contain standard
errors (for rates) and by-variables such as age group, sex, or race. These variables (or columns of data)
do not have to be in any particular order, but the data file (or rows of data) must be sorted by the by-
variables and the independent variable. You identify the dependent variable by specifying whether it is
calculated or provided, the type of variable to be modeled (i.e., crude rate, percent), and then selecting
the appropriate items on the right (i.e. count, population, rate).
SEER*Stat provides an easy mechanism for creating Joinpoint data files and setting the controls on the
Joinpoint Input File tab. In SEER*Stat, select " Matrix | Export" when viewing the SEER*Stat matrix to
create a file in the format of a Joinpoint Input Data File. In Joinpoint, select " New”>”Session…" from the
File menu and then select the SEER*Stat Dictionary file. The input data file name, variable names and
locations, and file format information will be loaded from the SEER*Stat Dictionary.
The Heteroscedastic Errors Option is automatically set from the SEER*Stat Dictionary, depending on
whether or not the standard errors are included. If they are not included, Constant Variance
(Homoscedasticity) is selected. If they are included, Standard Error is selected.

Related content
Format of the Joinpoint Input Data File
Using SEER*Stat to Create Joinpoint Input Data Files
Data Errors That Stop An Analysis

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Format of the Joinpoint Input Data File
Each record in the data file has the following layout:
Variable(1) Variable(2) . . . . . . . Variable(k)
The records must be sorted by: by-variables, independent variable.
File Contains Column Headers
Joinpoint can process data files where the first record contains the names of each variable in the
data file. The names should be delimited with the same field delimiter used in the rest of the file. If
the first record in your data file contains the variable names, please check the " File contains column
headers" checkbox. Once this is checked, the data grid on the Input File tab will be updated
accordingly.
Field Delimiters
The variables may be separated by a single character. This field delimiter may be a tab, space,
comma, or semi-colon.
Missing Characters
Missing values may be represented by a space, a period, or the text string NA.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Using SEER*Stat to Create Joinpoint Input Data
Files
Joinpoint requires a specific file format for the input data file. Therefore, the following export options
must be used in SEER*Stat when creating data files to be used in Joinpoint:
The data file must be a text file (use a .txt extension or a gzipped text file (use a .gz extension)
The Output Variables option must be set to " Numeric Representation"
The Line Delimiter option must be set to " DOS/Windows (CR/LF)"
The Field Delimiter may be any one of the choices. However, if commas are used as the field
delimiter then the option to " Remove All Thousand Separators (Commas)" must be checked.
" Remove Flags (Footnote Characters)" must be selected.

The independent variable (year) must be an individual value. Therefore, we recommend you create a
user-defined variable in SEER*Stat that does not contain any ranges. For example, when creating a new
" Year of Diagnosis" variable remove the range variable (1973-2004) for all years combined. When using
a SEER*Stat dictionary file (*.DIC), any independent variable value with a dash (*-*) located in its format
label will not be used in the analysis. The label will be updated to contain the following text: (not used
in calculations). This functionality was added so that users would not have to reproduce their SEER*Stat
analyses in order to remove totals or sub-totals in their independent variable.
The records for each by-group must be contiguous. If N = number of records for the first by-group, lines 1
through N must contain all values for the first by-group. Within each by-group, the records must be
sorted by the independent variable (this is typically year).
To be sure that the sort order is correct, on the SEER*Stat T able tab, put all table variables in the row
dimension. Make the independent variable (year) the last variable in the list of row variables.
Alternatively, you could set the independent variable as the column variable.
The SEER Cancer Statistics Review (CSR) uses the same Joinpoint Regression Program. Their analysis
uses the " Standard Error" setting for the Heteroscedastic Errors Option with the standard error of the
rate used as an estimate of the standard deviation. Using these options, and the current default number
of permutations (4499), if the Joinpoint Regression Program chooses the model with 0 joinpoints, the
annual percentage rate change will agree with the calculation of this value given in the CSR.
Note: to obtain the SEER*Stat Program visit the SEER web page at:

http://seer.cancer.gov/seerstat/

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Dependent Variable
The dependent, or response, variable is the variable being tested in the model. You can either provide
your dependent variable values in the input data set or you can have Joinpoint calculate them from other
variables in the input data set (e.g. a crude rate can be calculated from variables with the count and the
population).

Run Type
This option should be selected if you want Joinpoint to calculate your dependent variable.
Calculated Joinpoint can calculate Crude Rates, Age-Adjusted Rates, Proportions and Percentages. The
From table below documents which variables are required for each type of dependent variable.
Data File Please note that if Joinpoint calculates the dependent variable, it will also compute the
associated standard errors.
Provided This option should be selected if you are providing Joinpoint with your dependent variable
In Data values. For some analysis types, the standard error is also required. Please reference the
File table below for more information.
Delay This option is only available if a SEER*Stat dictionary (DIC) file is used for the session that
Adjusted contains delay adjusted statistics. Joinpoint will read in and analyze the delay statistics for
Data each cohort. See Delay Adjusted Statistics for more information.
Delay & This option is only available if a SEER*Stat dictionary (DIC) file is used for the session that
Non-Delay contains delay adjusted statistics. Joinpoint will read in and analyze both the delay and
Data non-delay statistics for each cohort. See Delay Adjusted Statistics for more information.

Types of Dependent Variables Accepted by Joinpoint:

The following dependent variable types can be analyzed by Joinpoint:

A count is typically a number of “events” (i.e. cancer cases or deaths. Joinpoint will accept
Count
counts in fractional form (e.g. 410.375).

A crude rate is the number of cases (or deaths) occurring in a specified population, usually
expressed as the number of cases per 100,000 population at risk.
Crude Joinpoint will accept counts and populations in fractional form (e.g., 410.375).
Rate

An age-adjusted rate is a weighted average of crude rates, where the crude rates are
calculated for different age groups and the weights are the proportions of persons in the
corresponding age groups of a standard population. The age-adjusted rate for an age group
comprised of the ages x through y is calculated using the following formula.

Where count is the number of cases in the ith age group, popi is the relevant population for
Age-
the same age group, and stdpopi is the standard population for the same age group.
Adjusted
Rate If count i and popi are both zero, Joinpoint treats that rate as zero. Joinpoint will not allow
the unusual event where popi is zero but count i is non-zero.
If count i is greater than zero and popi is zero, Joinpoint will set popi to be equal to count i
in order to compute the interim crude rate.
Joinpoint will accept counts, populations and standard populations in fractional form (e.g.,
410.375).
For a detailed explanation on how to compute age-adjusted rates, see How to Compute Age-
Adjusted Rates.

A proportion is a ratio of a numerator to a denominator where the result is between 0 and 1.

Joinpoint will not analyze cohorts that have proportions greater than 1 when such
proportions are calculated. Jointpoint will accept numerators and denominators in
Proportion fractional form (e.g., 410.375).
Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017
 A percent is a fraction or ratio with 100 understood as the denominator. Joinpoint will not
allow percentage values greater than 100 when such percentages are calculated.
Jointpoint will accept numerators and denominators in fractional form (e.g., 410.375).
Percent

Select this category for all other types of numeric variables that do not fall into one of the
Other
previous categories (i.e. temperatures over time).

Standard Error Calculations:

When Joinpoint computes the Dependent Variable, the following formulas will be used to compute the
associated standard errors.

Standard Error for Crude Rate

This calculation assumes that the counts have Poisson distributions.

Standard Error for an Age-Adjusted Rate

This calculation assumes that the counts have Poisson distributions. Suppose that the age-adjusted
rate is for an age group comprised of ages x through y.

Standard Error for Proportions and Percentages

For consistency, assume that percents are transformed to proportions before analysis.

Where pi is the proportion and ni is the sample size of the ith observation.

Edits performed on the Dependent Variable:

When a session is executed, Joinpoint will first edit the input data file. For a list and
description of all data errors and warnings produced by Joinpoint, please see the " Job Execution
- Errors and Warnings" help section.

Variables:
One or more of the following variables will need to be specified in the input data file depending on the
Calculated/Provided, Type, and Heteroscedastic Errors Options selections:
Count Variable
Population Variable
Age-Adjusted Rate
Standard Error
Adjustment Variable
Standard Population
Numerator Variable – this variable replaces the “Count” variable when appropriate
Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017
 Denominator Variable – this variable replaces the “Population” variable when appropriate
Joinpoint does allow any variable value to be provided in fractional form (e.g. 410.375).

Heteroscedastic Required Required Required Required

Calculate T ype
Errors Variables Variables Variables Variables
Constant
No Count Count
Variance

Standard
No Count Standard Error Count
Error

No Count Poisson Variance Count

Constant
No Crude Rates Rate
Variance
Standard
No Crude Rates Standard Error Rate
Error
Crude Poisson
n/a
Rates Variance
Age-
Constant
No Adjusted Rate
Variance
Rates
Age-
Standard
No Adjusted Standard Error Rate
Error
Rates
Age-
No Adjusted Poisson Variance n/a
Rates
Constant
No Proportion Proportion
Variance
Standard
No Proportion Standard Error Proportion
Error
No Proportion Poisson Variance n/a

Constant
No Percent Percent
Variance

Standard
No Percent Standard Error Percent
Error

No Percent Poisson Variance n/a

Constant
No Other Other
Variance
Standard
No Other Standard Error Other
Error
No Other Poisson Variance n/a
Constant
Yes Count n/a
Variance

Yes Count Standard Error n/a

Yes Count Poisson Variance n/a
Constant
Yes Crude Rates Count Population
Variance
Yes Crude Rates Standard Error Count Population

Yes Crude Rates Poisson Variance Count Population

Age-
Constant
Yes Adjusted Count Population Age Group Standard Pop
Variance
Rates

Age-
Yes Adjusted Standard Error Count Population Age Group Standard Pop
Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017
 Rates
Age-
Yes Adjusted Poisson Variance n/a
Rates
Constant
Yes Proportion Count Population
Variance

Yes Proportion Standard Error Count Population

Yes Proportion Poisson Variance n/a
Constant
Yes Percent Count Population
Variance
Yes Percent Standard Error Count Population

Yes Percent Poisson Variance n/a

Constant
Yes Other n/a
Variance
Yes Other Standard Error n/a
Yes Other Poisson Variance n/a

Related content

Data Errors That Stop An Analysis

Data Warnings Displayed by Joinpoint

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Independent Variable
The independent variable is defined on the Input File tab by its name or column position. These controls
will automatically be set when the SEER*Stat Export Dictionary is loaded. If you created your data file
with software other than SEER*Stat you must set these controls.
The data file must be sorted first by all by-variables and then the independent variable, unless you want
Joinpoint to calculate age-adjusted rates -- in which the data file would also be sorted by the adjustment
variable. For a detailed description of how age-adjusted rates are calculated and what the required
variable order, please go to the How to Compute Age-Adjusted Rates help section.
The independent variable or covariate is typically the year of diagnosis or year of death. You are limited
to a 40-character label for the independent variable name.
The " Shift Data Points by" option allows all the values for the independent variable to be shifted up by a
fixed value. This is done by simply entering a value for " Shift Data Points by" . For example, if your
independent variable is years (input as 1975, 1976,...), but you would like these points to represented on
the graph at the midpoint of the years (1975.5, 1976.5, ...), then you would enter the value 0.5 for this
option. Shifting the data points will change the location of the joinpoints and the intercepts but will not
change the slopes or APCs.
Note: this value cannot be greater than the maximum interval between data points.
One reason for employing this option would be because each data point represents a summary of data
collected over a time interval. For example, cancer incidence or mortality data is often collected over
the course of a year, and is usually entered as a whole year value, e.g. 1990, 1991. Instead, one may
want to shift by half a year so that the data point is represented as the midpoint of the interval, e.g.
1990.5, 1991.5). This is especially important if joinpoints are allowed to occur at places other than the
data points (either in continuous time using Hudson' s algorithm, or using a grid search where grid points
are allowed between data points). If the data points are not shifted, the results may be counterintuitive.
For example, without a shift, a joinpoint located at 1989.50 represents the beginning of 1990, and
1990.49 represents the end of 1990. If the points are shifted by half a year, then 1990.00 represents the
start of the year and 1990.99 represents the end of the year.

Related content

Define Format
Data Warnings Displayed by Joinpoint
Data Errors That Stop An Analysis

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Define Format
Joinpoint allows the independent variable to be renamed and its values to be recoded. If the
variable is recoded, the recoded values are used in the Joinpoint calculations.

The following attributes can be set for the Independent variable:

Variable name:
The name must be of reasonable length and it is recommended that the label be concise (short) for
ease of viewing in the output.
Independent Variable Coding Information (Value = Label):
An example of an Independent Variable format would be:
0 = 1975
1 = 1976
2 = 1977
Etc.
Labels MUST be numeric. If non-numeric values are used, then Joinpoint will not process any cohort
that contains the associated base ‘value’ (value=label). Labels must also be a reasonable length and
it is recommended that the label be concise for ease of viewing in the output.
When using a SEER*Stat dictionary file (*.DIC), any independent variable value with a dash (*-*)
located in its format label will not be used in the analysis. The label will be updated to contain the
following text: “not used in calculations”. This functionality was added so that users would not have
to reproduce their SEER*Stat analyses in order to remove totals or sub-totals in their independent
variable.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

By-variables
A separate analysis will be performed for each by-group. The by-variable information is automatically
loaded when you open the SEER*Stat Export dictionary associated with your data file. If you created
your data file with software other than SEER*Stat, you must manually add by-variables. To add a by-
variable, click the Add... button, select the desired variable, and click OK.

Related content
Format of the Joinpoint Input Data File
Define Format

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Define Format
Joinpoint allows by-variables to be renamed and labels to be assigned to each of their associated
values. If the format for a variable is changed, the new format will be saved with the session and when
the session is executed, the resultant output will reflect the changes.
The following attributes can be set for each By-variable:
Variable name:
The name must be of reasonable length and it is recommended that the label be concise (short) for
ease of viewing in the output.
Coding Information (Value = Label):
Each By-variable must consist of a series of values. Each value can be assigned a label. Users are
allowed to change the label for any value. If no labels were initially found for the by-variable, then
the label for each value is defaulted to the associated value.
A By-variable format example would be:
1 = Male and Female
2 = Male
3 = Female
Labels may have any meaningful text including numbers, letters, and special characters. They must
also be a reasonable length and it is recommended that the label be concise for ease of viewing in
the output.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

How to Compute Age-Adjusted Rates in Joinpoint
For a detailed explanation on how to calculate age-adjusted rates, please refer to the following web
site:
http://seer.cancer.gov/seerstat/tutorials/aarates/step1.html
For Joinpoint to compute age-adjusted rates, each input data record must have the following variables IN
THIS ORDER:

1. All cohort-defining variables

2. Independent variable
3. Adjustment variable (this is typically an age variable)
4. Count
5. Population
6. Standard Population

The following requirements will be applied to each cohort when age-adjusted rates are computed by
Joinpoint:
The data has to be sorted using the following variable order:
All cohort defining variables
Independent variable
Adjustment variable
Count values must be greater than or equal to zero.
Population values must be greater than or equal to zero
If the population value is zero, the associated count must also be zero.
The count must not be greater than the associated population.
Standard Populations values must be greater than zero.
Standard Populations must be identical for each adjustment variable value.
The Adjustment Variable must have the same categories (in the same order) for each By
Variable/Independent Variable combination.

Below is an example of an input data file correctly structured for Joinpoint to compute age-adjusted
rates.The data file contains one by variable (sex:1 = Male, 2 = Female) and the independent variable is
Year (values are 1980 thru 2000 inclusive).For each Sex/Year combination, there is a complete set of
adjustment variable values (00 years to 85+ years). Please note that the standard population appears on
every data record and the value corresponds to the associated age group.

US 2000
By Variable Independent Adjustment
Count Population Standard
(Sex) Variable (Year) Variable (Age)
Populations

1 1980 00 years 29 139,879 3,794,901

1 1980 01-04 years 87 553,189 15,191,619

1 1980 05-09 years 67 736,212 19,919,840

1 1980 10-14 years 71 770,999 20,056,779

1 1980 15-19 years 87 651,309 19,819,518

1 1980 20-24 years 177 639,159 18,257,225

1 1980 25-29 years 290 676,354 17,722,067

1 1980 30-34 years 657 736,557 19,511,370

1 1980 35-39 years 1,072 724,826 22,179,956

1 1980 40-44 years 1,691 700,200 22,479,229

1 1980 45-49 years 2,428 617,437 19,805,793

1 1980 50-54 years 2,931 516,541 17,224,359

1 1980 55-59 years 2,881 361,170 13,307,234

1 1980 60-64 years 2,817 259,440 10,654,272

1 1980 65-69 years 2,817 206,204 9,409,940
Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017
1 1980 70-74 years 2,744 172,087 8,725,574

1 1980 75-79 years 2,634 142,958 7,414,559

1 1980 80-84 years 1,884 99,654 4,900,234

1 1980 85+ years 1,701 92,692 4,259,173

1 1981 00 years 34 141,456 3,794,901

1 1981 01-04 years 97 562,059 15,191,619

1 1981 05-09 years 72 738,302 19,919,840

... ... ... ... ... ...

1 2000 80-84 years 2,176 100,249 4,900,234

1 2000 85+ years 2,001 94,875 4,259,173

1 2000 80-84 years 2,176 100,249 4,900,234

2 1980 00 years 37 159,829 3,794,901

2 1980 01-04 years 101 562,001 15,191,619

... ... ... ... ... ...

2 2000 85+ years 2,521 96,228 4,259,173

Below is an example of how an age-adjusted rate is computed. The age-adjusted rate is 400.3 and is the
sum of the interim values over all age groups.The interim value in each row is equal to (Crude Rate * Age
Distribution of Std Pop).

Crude US Standard Age Distribution of Interim

Age Count Population
Rate Populations Std Pop Value
00 years 29 139,879 20.7 3,794,901 0.013818 0.29
01-04
87 553,189 15.7 15,191,619 0.055316 0.87
years
05-09
67 736,212 9.1 19,919,840 0.072532 0.66
years
10-14
71 770,999 9.2 20,056,779 0.073031 0.67
years
15-19
87 651,309 13.4 19,819,518 0.072167 0.96
years
20-24
177 639,159 27.7 18,257,225 0.066478 1.84
years
25-29
290 676,354 42.9 17,722,067 0.064530 2.77
years
30-34
657 736,557 89.2 19,511,370 0.071045 6.34
years
35-39
1,072 724,826 147.9 22,179,956 0.080762 11.94
years

40-44
1,691 700,200 241.5 22,479,229 0.081852 19.77
years

45-49
2,428 617,437 393.2 19,805,793 0.072117 28.36
years

50-54
2,931 516,541 567.4 17,224,359 0.062718 35.59
years

55-59
2,881 361,170 797.7 13,307,234 0.048454 38.65
years

60-64
2,817 259,440 1,083.8 10,654,272 0.038794 42.12
years

65-69
2,817 206,204 1,366.1 9,409,940 0.034264 46.81
years
Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017
 70-74
2,744 172,087 1,594.5 8,725,574 0.031772 50.66
years
75-79
2,634 142,958 1,842.5 7,414,559 0.026998 49.74
years
80-84
1,884 99,654 1,890.5 4,900,234 0.017843 33.73
years
85+
1,701 92,692 1,839.4 4,259,173 0.015509 28.53
years
All Ages 274,633,642 1.000000 400.3

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Log Transformation
Select a linear or log-linear model, based on whether or not you would like the dependent variable to be
log-transformed.
No - Linear Model: y = x' beta + e
Yes - Log-linear Model: ln(y) = x' beta + e
For a recommendation on which model to select for your analysis, please reference this FAQ.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Heteroscedastic Errors Option (Weighted Least
Squares)
Select one of the three Heteroscedastic Errors Options. You must then specify the variable(s) required
for that option.
The error random variable in a model is homoscedastic if the variance of the error is constant;
otherwise, the error is heteroscedastic. Often the homoscedastic assumption is violated, particularly
when the variance of the error varies with time. The option allows the user to choose between a model
where errors are assumed to have constant variance (homoscedasticity) and a model where nonconstant
variance (heteroscedasticity) is assumed.
Heteroscedasticity is handled by Joinpoint using weighted least squares (WLS). The weights in WLS
are the reciprocal of the variance and can be specified in several ways. The “Standard Error (Provided)”
option allows the user full flexibility to specify the standard error at each time period. These standard
errors can come from the output of SEER*Stat.
The last option, “Poisson variance”, estimates the non-constant variance by assuming the dependent
variable counts follow a Poisson distribution. If “Crude Rate” is selected, you must choose to have
Joinpoint calculate the rate and therefore provide both a count and a population.

Related content
Constant Variance (Homoscedasticity)
Standard Error (Provided)
Poisson Variance

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Constant Variance (Homoscedasticity)
This selection assumes the random errors in the regression model are homoscedastic (have constant
variance) and estimates the regression coefficients by ordinary least squares for both model ln (y) = xb
and model y = xb.

Adding 0.5 to Zero Counts

When analyzing counts or computing crude rates, Joinpoint will add 0.5 to any zero count under the
circumstances listed here.

Related content
Standard Error (Provided)
Poisson Variance

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Standard Error (Provided)
This is the default selection. It assumes that the random errors are heteroscedastic (have non-constant
variance) and estimates the regression coefficients by weighted least squares, where weights at each
point are:
For model y = xb

w = 1/v, where v is the square of the std dev that has been input for that point.
For model ln (y) = xb

w = (y2)/v, where y2 is the square of the response for that point and v is the square of the std dev
that has been input for that point. (Motivated by delta method.)

Related content
Constant Variance (Homoscedasticity)
Poisson Variance

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Poisson Variance
With Provided Count
This selection assumes the dependent variable is y = c, where c is the adjusted count, which equals
either:
The count, if the count is greater than zero.
The count plus one half, if the count is equal to zero.
Assume the random errors are Poisson, and estimate the regression coefficients by weighted least
squares, where weights at each point are:
For model y = xb
w = 1/c, where c is the adjusted count for that point.
For model ln(y) = xb
w = c, where c is the adjusted count for that point. (Motivated by delta method.)

With Calculated Crude Rate

This selection assumes the dependent variable is y = c/p, where c is the adjusted count and p is the
population. The adjusted count will be equal to either:
The count, if the count is greater than zero.
The count plus one half, if the count is equal to zero.
Assume the random errors are Poisson, and estimate the regression coefficients by weighted least
squares, where weights at each point are:
For model y = xb
w = p2/c, where c is the adjusted count and p2 is the square of the population for that point..
For model ln(y) = xb
w = c, where c is the adjusted count for that point. (Motivated by delta method.)

Adding 0.5 to Zero Counts

When analyzing counts or computing crude rates, Joinpoint will add 0.5 to any zero count under the
circumstances listed here.

Related content

Constant Variance (Homoscedasticity)

Standard Error (Provided)

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Delay Adjusted Statistics
Joinpoint has the ability to process both delay and non-delay adjusted statistics exported from
SEER*Stat. If a new Joinpoint session is created using a SEER*Stat dictionary (DIC) file that contains
delay adjusted statistics, two new " Run Types" will be available within the Dependent Variable section
on the Input File tab. The following options will be provided within this control:
Delay Adjusted Data – by selecting this option, Joinpoint will only read in and analyze the selected
delay adjusted statistic.
Delay & Non-Delay Data – by selecting this option, Joinpoint will read in and analyze both the delay
and non-delay statistics for each cohort. The delay and non-delay model graphs and associated Joinpoint
statistics will appear together for each cohort.

For information about delay-adjusted adjusted incidence data, please go to the NCI website.
.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Data Grid
The Data Grid provides users with a snapshot of their data file and shows them how Joinpoint will parse
the file.
Once a user selects their input data file, the Data Grid will be populated with the contents of the file.
The contents of the grid are updated whenever the user adjusts the Delimiters, Missing Character, or
File Contains Column Headers controls.
Users should use the Delimiters, Missing Character, and File Contains Column Headers controls until
their data is correctly displayed in the Data Grid. Once the appropriate settings have been made, then
users can go and set the other variables (independent, dependent, and by) and controls on the tab.
Users can select to display only a set number of records from their input file. The choices to display the
following number of lines: 5, 10, 20, 50, or 100.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Method and Parameters Tab
The purpose of the Method and Parameters tab is to specify the:
modeling method (Grid Search or Hudson' s)
constraints on the location(s) of the joinpoints
number of joinpoints
autocorrelated errors options
model selection method (Permutation Test, BIC, or Modified BIC)
average annual percent change options
exclude specific cohorts from the analysis

Related content
Method – Grid Search or Hudson' s
Number of Joinpoints
Autocorrelated Errors Option
Average Annual Percent Change (AAPC)
AAPC Confidence Intervals
Specify AAPC Ranges
How Joinpoint Selects the Final Model
Include/Exclude Select Cohorts
Minimum Number of Observations Required

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Data Files Containing Cohorts with an Unequal
Number of Observations
As of version 4.4.0.0, Joinpoint can process data files that contain cohorts with an unequal
number of observations.

Reading the Input File and Populating the Independent Variable:

When all of the input variables on the Input File tab of the session have been specified, Joinpoint
will read the data file. The Independent variable format will be populated with all the unique
variable values found in the data file and the Maximum Number of Joinpoints (on the Method and
Parameters tab) will be assigned based on the cohort with the maximum number of observations. If
cohorts with a varying number of observations are located, Joinpoint will not initially indicate that
the file contains cohorts of this type. Only when the session is executed will Joinpoint indicate if
there are cohorts with an unequal number of observations.
Executing the Session
When the session is executed, a dialog will appear indicating that cohorts with fewer observations
have been found. Joinpoint will display a total count of all cohorts whose maximum number of
Joinpoints will follow the session maximum. All other cohorts will be organized by their total number
of observations and will be assigned a lower maximum number of Joinpoints. That maximum number
of Joinpoints will be based off of this table: Default Maximum Number of Joinpoints. The dialog will
allow users to cancel the job or to continue.
What happens when the maximum number of Joinpoints is enough for all cohorts?
If you have specified a Maximum Number of Joinpoints that is low enough to satisfy the default
maximum requirement for all cohorts, then (upon execution of the session) a message indicating that
cohorts with an unequal number of observations were located and that no adjustment to the
Maximum Number of Joinpoints for any cohort will be made.
Example:
For example, suppose you have a data file with the following three cohorts:

1. Male and Female 24 observations (data lines)

2. Male 21 observations (data lines)
3. Female 12 observations (data lines)

In this example, Joinpoint will default the Maximum Number of Joinpoints to 4 (please see the
Default Maximum Number of Joinpoints). When the session is executed Joinpoint will indicate that
cohorts with few observations than required to run with a setting of 4 Joinpoints were located. The
following will be displayed on the warning dialog:
1 cohort with 24 observations. Maximum Joinpoints will remain set to 4.
1 cohort with 21 observations. Maximum Joinpoints will be set to 3.
- Male
1 cohort with 12 observations. Maximum Joinpoints will be set to 2.
- Female

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Method – Grid Search or Hudson's
Joinpoint allows two different methods for model fitting - Grid Search or Hudson' s. The Grid Search has a
finite number of discrete locations that are tested to find the best model fit, while Hudson' s allows for
continuous fitting.

As of Joinpoint version 4.5.0.0, Hudson' s method is unavailable.

Related content
Grid Search Method - Details
Hudson' s Method – Details

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Grid Search Method - Details
New for Version 4.5.0.0
For version 4.5.0.0, we have made some changes to the layout and values for Grid Search. But don’t
worry, everything still works much as you remember from older versions. The old layout was a little
confusing with parameters being shared between methods when they really shouldn’t be. The new
interface and values are more precise and specific to each method. Here’s what’s been changed and why:
The “Number of Observations” controls have been moved under Grid Search. In the old interface,
the number of observations controls were used for both Grid Search and Hudson’s. But this
resulted in some ambiguity for the parameter values. So we’ve moved the controls under Grid
Search to make it clear that these are specific only to the Grid Search Method. Hudson’s will
have its own parameter controls (in time).
The description for the “Number of Observations” controls has changed. The old descriptions
used for these controls were not accurate or clear for all possible cases. We’ve reworded the
descriptions for these controls so that they are accurate in all situations.
The default values for “Number of Observations” have changed. You may notice that the default
values for the number of observations are different than in previous versions. That is correct, but
the new values mean the same thing. Due to the change in the descriptions for each control, the
old default values were no longer correct. The new default values are accurate and obtain the
same results as the old default values. If you load a session from an older version of Joinpoint,
these parameters will automatically be updated to the new values. In this way, if you re-run an
old session, you’ll get the exact same results.
The “Number of Joinpoints” controls have been moved after the method selection. The controls
are now in the correct order. The maximum number of joinpoints is actually dependent on the
settings for “Number of Observations”, so it makes sense for the “Number of Joinpoints” controls
to come after the “Number of Observations” controls.

How Grid Search Works

Using the default settings, the Grid Search Method only allows the joinpoints to occur exactly at
observations. This does not, however, find the best fit. A better fit can be achieved by using a finer grid -
by changing the setting for " Number of points to place between adjacent observed x values in the grid
search" to something larger than the default of zero. So, the Grid Search Method creates a " grid" of all
possible locations for joinpoints specified by the settings, and calculates the SSE at each one to find the
best possible fit. With lower values for “Number of points to place between...”, this method is
computationally more efficient.

Min number of observations from a joinpoint to either end of the data

(excluding the first or last joinpoint if it falls on an observation)
Use this parameter to keep joinpoints from being placed too close to the end points. This value must be
at least two (2) and no more than nine (9), but the default is set to two (2). So, for example if this value
is set to 2 and the data is annual from 1990-2014, then the first possible Joinpoint is 1992.

Min number of observations between two joinpoints (excluding any joinpoint if

it falls on an observation)
Use this parameter to keep joinpoints from getting too close together. This value can be set as low as
two (2) and as high as nine (9), but the default is set to two (2). For example, if this value is set to 2, the
data is annual from 1990-2014, and there was a joinpoint at 2000, then the closest possible other
joinpoints would be at either 1997 or 2003.
The permutation tests are valid for any of the allowable choices for the above two settings (the minimum
number of observations from a joinpoint to either end of the data, and the minimum number of
observations between two joinpoints); however, some statistics (the standard error of the slope
parameters and the associated p-values) cannot always be calculated when there are not at least three
observations on a line segment (excluding observations at the joinpoints). See the statistical notes
about Estimated Regression Coefficients (Beta). Similarly, some statistics cannot be calculated if a line
segment in the joinpoint model is an exact fit.

Number of points to place between adjacent observed x values in the grid

search

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

To allow joinpoints to occur between observations, specify the number of points to place between
adjacent observed x-values in the grid search as 1-9. So, for example, if this value is set to 3, and the data
is annual from 1990-2014, then the joinpoints could occur at values like 1995.25, 2000.5, or 2010.75, as
well as on the observations. The minimum allowed value is 0 and the maximum is 9. Note that a fine grid
search with this value set at >5 may take a long time to process. An alternative to this would be to
use Hudson' s Method.

Related content

Hudson' s Method – Details

New default settings for Grid Search
Minimum Number of Observations Required

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Hudson's Method – Details
As of Joinpoint version 4.5.0.0, Hudson' s method is unavailable.
In version 4.5.0.0, Hudson’s Algorithm (used to estimate the exact best fitting location of joinpoints in
continuous time) was temporarily disabled. It was determined that the algorithms for setting how close
joinpoints can be to each other and to the ends of the series were not working properly when using the
Hudson’s Algorithm. When examining this issue in detail, it was determined that there were some
complicated interactions between these settings and the use of the Hudson’s Algorithm. We were not
fully able to work out these issues in time for this release. Since version 4.5.0.0 had some important
other updates, we decided to temporarily disable Hudson’s Algorithm. We apologize for any
inconvenience this may cause.
In previous versions, the placement of joinpoints was more constrained than it should have been given
the settings. Also the description of the settings was incorrect for how they were used by Hudson’s
Method. We have now separated the settings for Grid Search and Hudson’s Method to make this clearer.

Related content

Grid Search Method - Details

Early Stopping Options
Hudson' s Method Disabled

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Number of Joinpoints
Enter the minimum and maximum number of joinpoints to fit where:
0 = min = max = 9 when the Grid Search method is selected.
-or-
0 = min = max = 4 when Hudson' s method is selected.
As of Version 3.5, the default value for the maximum number of joinpoints depends on the number of data
points. This value can be changed by the user, subject to having a minimum number of data points
necessary to satisfy two pre-set rules:
A joinpoint cannot occur within a user-specified number (default: 3) of data points from the
beginning or end of a series.
There must be at least a user-specified number (default: 4) of data points between two
joinpoints.
The default maximum number of joinpoints is a recommendation based on the same metrics used to
determine the minimum number of data points (or conversely the maximum number of joinpoints for a
given number of data points). The default is based on the following recommendations:
At least seven data points should be observed in order to consider allowing a joinpoint.
There should be, on average, at least four data points between consecutive joinpoints.
These algorithmic recommendations lead to the following default maximum number of joinpoints.

Number of Data Points Default Maximum Number of Joinpoints

0-6 0
7 -11 1

12 -16 2

17 - 21 3
22 -26 4

27+ 5

Note: due to computational intensity, the default for the maximum number of joinpoints is capped at 5
when the Grid Search Method is selected and is capped at 4 when the Hudson' s Method is selected.The
Grid Search allows a maximum up to 9, but those runs could take quite a long time to complete.The
maximum for Hudson' s is 4.

Related content

Data Errors That Stop An Analysis

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Autocorrelated Errors Option
If you select " Fit an uncorrelated errors model" , the program assumes the random errors in the
regression model are uncorrelated and estimates the regression coefficients by ordinary least squares
(unless the errors are heteroscedastic; see Heteroscedastic Errors Option).
If you select " Fit an autocorrelated errors model based on the data" , the autocorrelation parameter will
be estimated separately for each by-group using the method described in Section 2.3 of Kim et al.
(2000). Under this option, the autocorrelation parameter is estimated for the model with the default
maximum number of joinpoints or the maximum number of joinpoints set by a user.
If you select " Fit an autocorrelated errors model with parameter =" , you must input an autocorrelation
parameter (usually between -1 and 1) which represents the correlation between adjacent points. The
program then assumes the random errors are autocorrelated and estimates the regression coefficients
by weighted least squares. The autocorrelation model is based on the assumption that corr(ei, ej) = φ |i-j|
where ei and ej are the errors corresponding to the i th and j th points and φ is the autocorrelation
parameter chosen. This option makes sense only with equally spaced points.
Although the autocorrelation may be estimated from the data, correcting for autocorrelation with this
estimate may seriously reduce the power to detect joinpoints (see Section 3 of Kim et al. (2000)). We
found in our simulations in Table IV of that paper that adjusting for autocorrelation was helpful in
maintaining proper size of the tests of joinpoints when there was large autocorrelation. We also found
that if there was no autocorrelation then the adjustment seriously affected the power of the test to
detect joinpoints. For example we see in Table IV (b) with φ = 0, the power goes from 90% to 68%. This
is because it is difficult to differentiate between autocorrelation and joinpoints in a model.
If you suspect that your data are positively autocorrelated, we suggest using the " Fit an autocorrelated
errors model with parameter =" option to see how sensitive your results are to changes in
autocorrelation. The option should be used as follows:

1. Fit the model with the uncorrelated errors option.

2. If the user suspects that there is positive autocorrelation in the data, then repeat the analysis
trying several values of the autocorrelation parameter, say for example 0.1, 0.2, and 0.3. If the
results are very similar with different values of the autocorrelation parameter, then the user knows
their results will still hold if there is autocorrelation present. If the results change as the
autocorrelation parameter changes, then the user may end up presenting the series of results, to
show how the results depend on different assumptions about the autocorrelation.

If you suspect negative autocorrelation, the uncorrelated errors model will suffice (see Kim et al., 2000).
Note: only the uncorrelated errors model can be used with the Pairwise Comparison.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Average Annual Percent Change (AAPC)
While Joinpoint computes the trend in segments whose start and end are determined to best fit the data,
sometimes it is useful to summarize the trend over a fixed predetermined interval. The AAPC is a method
which uses the underlying Joinpoint model to compute a summary measure over a fixed pre-specified
interval.
Annual Percent Change (APC) is one way to characterize trends in cancer rates over time. With this
approach, the cancer rates are assumed to change at a constant percentage of the rate of the previous
year. For example, if the APC is 1%, and the rate is 50 per 100,000 in 1990, the rate is 50 x 1.01 = 50.5
in 1991 and 50.5 x 1.01 = 51.005 in 1992. Rates that change at a constant percentage every year change
linearly on a log scale. For this reason, to estimate the APC for a series of data, the following regression
model is used:

One advantage of characterizing trends this way is that it is a measure that is comparable across scales,
for both rare and common cancers. For example, it is reasonable to think that rates for a rare cancer and
a common cancer could both change at 1% per year, but it is not reasonable to think that a rare cancer
and a common cancer would change in the same increments on an absolute (or arithmetic) scale. That is,
a cancer with a rate of 100 per 100,000 could be changing by 2 per 100,000 every year, but a cancer with
a rate of 1 per 100,000 would probably not change in the same increments.
It is not always reasonable to expect that a single APC can accurately characterize the trend over an
entire series of data. The joinpoint model uses statistical criteria to determine when and how often the
APC changes. For cancer rates, it is fit using joined log-linear segments, so each segment can be
characterized using an APC. For example, cancer rates may rise gradually for a period of several years,
rise sharply for several years after that, and then drop gradually for the next several years. Finding the
joinpoint model that best fits the data allows us to determine how long the APC remained constant, and
when it changed.
Average Annual Percent Change (AAPC) is a summary measure of the trend over a pre-specified fixed
interval. It allows us to use a single number to describe the average APCs over a period of multiple
years. It is valid even if the joinpoint model indicates that there were changes in trends during those
years. It is computed as a weighted average of the APCs from the joinpoint model, with the weights
equal to the length of the APC interval.
How is the AAPC Computed?
Can We Compare AAPCs for Two Independent Groups?
What are the Relative Advantages and Disadvantages of Reporting an AAPC over APCs?
What is the Advantage of Reporting an AAPC over an APC Computed by Fitting a Single Line (on a log
Scale) to the Data?

How is the AAPC Computed?

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

AAPC is derived by first estimating the underlying joinpoint model that best fits the data. The
accompanying figure shows the joinpoint model for prostate cancer incidence from 1975-2003 (from the
2005 data submission), which found joinpoints in 1988, 1992, and 1995. (This model is fit under the
default joinpoint parameters). The AAPC over any fixed interval is calculated using a weighted average
of the slope coefficients of the underlying joinpoint regression line with the weights equal to the length
of each segment over the interval. The final step of the calculation transforms the weighted average of
slope coefficients to an annual percent change. If we denote bis as the slope coefficients for each
segment in the desired range of years, and the wis as the length of each segment in the range of years,
then:
APCi = { Exp(bi) - 1 } x 100
and

In the prostate cancer example, to compute the AAPC from 1994 to 2003, we first note that an APC of -
10.7 runs for 1 year (with a slope coefficient of -0.113), while an APC of 0.7 runs for 8 years (with a slope
coefficient of 0.007). Thus, the AAPC is computed as:
{ Exp( ( 1 x (-0.113) + 8 x 0.007 ) / 9 ) - 1 } x 100 = -0.6.
The AAPC for a year range which is entirely within a single joinpoint segment is equal to the APC for that
segment, e.g., the AAPC for 1999 to 2003 is computed as:
{ Exp( ( 4 x 0.007 ) / 4 ) - 1 } x 100 = 0.7.
The AAPC for the entire range of data is computed as:
{ Exp( (13 x 0.026 + 4 x 0.151 + 3 x (-0.113) + 8 x 0.007 ) / 28 ) - 1 } x 100 = 2.4.
Parametric Confidence Interval for T rue AAPC
Denote the normalized weight as

and rewrite the AAPC as

An approximate 100(1- α)% confidence interval for the true average annual percent change is

, where

are the lower and upper confidence limits of the interval, Ζα is the α th quantile of the standard normal

distribution, and denotes the estimate variance of bi obtained from the fit of the joinpoint model.
In the Joinpoint software, the AAPC confidence interval is based on the normal distribution, and the APC
confidence interval is based on a t distribution. If an AAPC lies entirely within a single joinpoint
Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017
segment, the AAPC is equal to the APC for that segment. To obtain consistency between the APC and
AAPC confidence intervals in this situation, the confidence interval for the AAPC has been modified to be
identical to that used for the APC using the t distribution instead of the normal distribution. For details
on the confidence intervals for the APC, see: Estimated APC.
Empirical Quantile Confidence Interval for T rue AAPC
Motivated by a conservative tendency of the asymptotic confidence interval for the true AAPC, a new
method called the empirical quantile method is implemented in V 4.2. as an improved confidence
interval. The idea of the empirical quantile method is to generate resampled data by (i) generating
resampled residuals as the inverse function values of the uniform random numbers over (0,1) where the
function is the empirical distribution function of the original residuals and then (ii) adding resampled
residuals to the original fit. For each resampled data, the model fit was made and the AAPCs are
estimated. Then, the 100(α/2)th and 100(1-α/2)th percentiles of the resampled AAPC values are obtained
as the lower and upper limits of the 100(1-α)% empirical quantile confidence interval for the true AAPC.
For more information on the empirical quantile confidence interval, please contact technical support.
Confidence Interval and Hypothesis T est
If the confidence interval contains zero, then there is no evidence to reject the null hypothesis that the
true AAPC is zero at the significance level of α otherwise, we reject the null hypothesis in favor of the
alternative hypothesis that the true AAPC is different from zero.
[Return to the top]

Can we compare AAPCs for two independent groups?

The true AAPCs for two independent groups, for example, males and females, can be compared by using
the following approximate 100(1-α)% confidence interval with the estimated difference of AAPC(1) -
AAPC(2), where

Note that an approximate 100(1-α)% confidence interval for φ (1) - φ (2) is (dL(α),dU(α)), where

Then, using a Taylor series expansion, an approximate 100(1-α)% confidence interval for the difference
between the two true average annual percent change rates can be obtained as where

[Return to the top]

What are the relative advantages and disadvantages of reporting an

AAPC over APCs?
Reporting an APC for each joinpoint segment provides a complete characterization of the trend over
time. However, sometimes a summary measure over a fixed interval may be desirable. The statistical
power to determine if an APC is different from 0 is a function of the length of the interval. Thus, a short
segment rising at a steep rate may not be statistically significant. Comparing the last segment of two
series (e.g. males and females) sometimes yields seemingly contradictory results when the segments are
of very different lengths. Comparing AAPCs of equal lengths from both series is usually a more
meaningful comparison. For example, the delay-adjusted thyroid cancer incidence trend (using data from
1975-2002) was rising for males at 2.2% per year from 1980-2000 (and is characterized as rising since it
is statistically significant) and at 11.4% per year from 2000 to 2002 (characterized as a non-significant
change because the APC is not statistically significant). However, for females it was rising at 5.3% per
year from 1993-2002 (characterized as rising since it is statistically significant). Because the last
segment for males is relatively short, it introduces uncertainty, and we arrive at the conclusion that
recent rates for females are rising, while for males we are uncertain if the rates are rising. To make the
Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017
comparison between males and females more comparable, it is useful to compute the AAPC over the
same fixed interval for both series. The AAPC for 1993-2002 is 4.2% for males and 5.3% for females
(each characterized as rising since they are both statistically significant). The AAPC for 1998-2002 is
6.8% for males and 5.3% for females (each characterized as rising since they are both statistically
significant).
Rather than reporting the APC for the final segment for a long list of cancer sites, there may be
advantages to reporting the AAPCs over specified fixed intervals. If space permits, reporting both the
AAPC and the final segment APC gives an even more complete picture, since each give a somewhat
different perspective.
[Return to the top]

What is the advantage of reporting an AAPC over an APC computed

by fitting a single line (on a log scale) to the data?
Prior to the development of the joinpoint and AAPC methodology, to characterize a trend over a fixed
interval, a single regression line (on a log scale) over the fixed interval was fit, and the slope coefficient
was then transformed to an APC. This older methodology has two disadvantages over the AAPC. First,
the older methodology assumes linearity of the trend (on a log scale) over the interval, while the AAPC
does not. Secondly, the AAPC can be used to characterize a short segment based on a joinpoint model fit
over a much longer series. This is especially advantageous for situations when the data are sparse (e.g.
a rare cancer or data from a small geographic area). In these cases, because of the variability of the
underlying data, the older method might estimate an APC for the period 1996-2005 with a very wide
confidence interval. The AAPC from 1996-2005 might be based on an underlying joinpoint model using
data from 1975-2005, and thus the resulting AAPC would usually be more stable.
For any additional questions about the AAPC, please contact Joinpoint Technical Support. More technical
details about the AAPC are available in the following journal article:
Clegg LX, Hankey BF, Tiwari R, Feuer EJ, Edwards BK. Estimating average annual percent change in
trend analysis. Statistics in Medicine 2009; 28(29): 3670-82. [Abstract]
[Return to the top]

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

AAPC Confidence Intervals
Introduction to Parametric Confidence Interval for True AAPC
In the Joinpoint software, the parametric confidence interval for the true AAPC is based on the normal
distribution, and the APC confidence interval is based on a t distribution. If an AAPC lies entirely within
a single joinpoint segment, the AAPC is equal to the APC for that segment. To obtain consistency
between the APC and AAPC confidence intervals in this situation, the confidence interval for the AAPC
has been modified to be identical to that used for the APC using the t distribution instead of the normal
distribution. For details regarding these AAPC confidence intervals, see Parametric Confidence Interval
for True AAPC. For details regarding the confidence intervals for the APC, see: Estimated APC.

Introduction to Empirical Quantile Confidence Interval for True AAPC

Motivated by a conservative tendency of the asymptotic confidence interval for the true AAPC, a new
method called the empirical quantile method is implemented in V. 4.2.0.0 as an improved confidence
interval. The idea of the empirical quantile method is to generate resampled data by (i) generating
resampled residuals as the inverse function values of the uniform random numbers over (0,1) where the
function is the empirical distribution function of the original residuals and then (ii) adding resampled
residuals to the original fit. For each resampled data, the model fit was made and the AAPCs are
estimated. Then, the 100(a/2)th and 100(1-a/2)th percentiles of the resampled AAPC values are obtained
as the lower and upper limits of the 100(1-a)% empirical quantile confidence interval for the true AAPC.
For details regarding these AAPC confidence intervals, see Empirical Quantile Confidence Interval for
True AAPC.

Related content
Average Annual Percent Change (AAPC)

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Specify AAPC Ranges
The AAPC Confidence Intervals can be calculated using the Parametric option or the Empirical
Quantile option. To specify the AAPC ranges, click the Specify AAPC Ranges... button. From the
Specify AAPC Ranges window, you can select to display the AAPC for the entire range of data, the last
N observations, or other specified ranges.
When the AAPC Confidence Intervals Parametric option is selected, the AAPCs can be computed
after the Joinpoint calculations are performed, using the APCs. With this option, the AAPCs can be
modified in the output by following the following procedure, without having to rerun the session.

1. When viewing the Joinpoint output, to specify which ranges to display use the Output > Specify
AAPC Ranges... menu item from the top of the main Joinpoint Output window.
2. From the Specify AAPC Ranges window, you can select to display the AAPC for the entire range of
data, the last N observations, or other specified ranges.

When the AAPC Confidence Intervals Empirical Quantile option is selected, the AAPCs cannot be
computed after the Joinpoint calculations are performed.

Related content

Average Annual Percent Change (AAPC)

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

How Joinpoint Selects the Final Model
Joinpoint selects the optimal model using three different methods.
The first method uses the sequence of permutation tests to ensure that the approximate probability of
overall Type I error is less than the specified significance level (also called the alpha level, default =
.05). Assuming that the default value of the minimum number of joinpoints is 0, " the overall Type I error"
is the probability of incorrectly concluding that the underlying model has one or more joinpoints when,
in fact, the true underlying model has no joinpoints.
The second method is based on the Bayesian Information Criterion (BIC). The value of BIC is the
loglikelihood value with penalizing the cost of extra parameters. The model with the minimum value of
BIC is selected as the optimal model.
The third method is Modified BIC - a modification of traditional BIC proposed to improve its performance.
Here are the details for the three different methods:
Permutation Test
Bayesian Information Criteria (BIC)
Modified BIC

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Permutation Test
The program performs multiple tests to select the number of joinpoints, using the Bonferroni correction
for multiple testing. Set the overall significance level for multiple testing.
The program performs permutation tests to select the number of joinpoints. Since fitting all N! possible
permutations of the data would take too long, the program takes a Monte Carlo sample of these N! data
sets, using a random number generator. Specify the size of the Monte Carlo sample of permuted data
sets.

Related content
How Joinpoint Conducts Permutation Testing
Permutation Test Details
Joinpoint p-values
Early Stopping Options

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

How Joinpoint Conducts Permutation Testing
In the Joinpoint Regression Program, the permutation test is used repeatedly for testing between two
different joinpoint models, a simpler model with fewer joinpoints called the null model, and a more
complicated model called the alternative model. The alternative model fits better because it is more
complicated.
The question for the test is: does it fit much better than would be expected by chance. To test this
statistically, we calculate a ratio, SSE N/SSE A, where SSE N is the sum of squared errors (SSE) from the
null model and SSE A is the SSE from the alternative model. Values of the ratio close to 1 mean that the
alternative is not much better than the null model, while larger values mean that the alternative is much
better.
In order to decide how much larger a ratio needs to be to be statistically significant, we use the
permutation method. In this method we randomly permute (that is, shuffle) the errors (also called the
residuals) from the null model and add them back onto the modeled values from the null model to create
a permutation data set. Then we calculate the ratio for the permutation data set.
If the true model was the null model, we would expect that about half of the ratios calculated
from the permutation data set would be less than the original one.
If the true model were the alternative model, we would expect that after permuting the errors
most of the new ratios would be less than the original ratio. In other words the permuted data
set would look less like the alternative model than the original data.
So we reject the null model (or null hypothesis) if less than a certain proportion of the ratios are greater
than or equal to the original ratio.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Permutation Test Details
First, the user specifies MIN as the minimum number of joinpoints and MAX as the maximum number of
joinpoints on the Method and Parameters tab. Then the program uses a sequence of permutation tests to
select the final model. Each one of the permutation tests performs a test of the null hypothesis H 0:
number of joinpoints=ka against the alternative H a: number of joinpoints=kb where ka< kb. The procedure
begins with ka = MIN and kb = MAX. If the null is rejected, then increase ka by 1; otherwise, decrease kb

by 1. The procedure continues until ka= kb and the final value of is the selected number of
Joinpoints.

Significance level of each individual test in a sequential testing

procedure
Because multiple tests are performed, the significance level of each test is adjusted to control the
overall type I error at specified a level (0.05). Before Version 3.0, Bonferroni adjustment was used, i.e.,
a1=a/(MAX - MIN) and if the individual test p-value is less than a1, the null is rejected.
The Bonferroni adjustment is conservative because the actual overall significance level is usually less
than the nominal level a. Starting with Version 3.0, the new adjustment procedure controls the overall
over-fitting error probabilities,

under a. Let k denote the number of joinpoints and a(ka; kb) be the significance level of each individual
test H0 : k = ka vs. H1 : k = kb. The new procedure set a(ka; kb) = a/(MAX - ka). Notice that the individual
significance level depends on the number of joinpoints ka under the null. Consider an example where
MIN = 0 and MAX= 4. The new procedure has the following properties:

If we like to bound these over-fitting probabilities by a, then we can assign different values for each
a(ka; kb) . That means, we can achieve a better power by setting

Overall Significance Level

Set the significance level to be used for the permutation test.

Number Of Randomly Permuted Data Sets

The minimum allowed number of permutations is 1000. The maximum number is 10,000. For greater
consistency in the p-values obtained if one were to change the seed for each run, we strongly
recommend running the program for at least 4499 permutations. You may use a smaller number of
permutations to speed up the calculations, but the permutation test may produce less consistent
results. Unless you have a good understanding of the implications of changing the number of
permutations, we recommend against it.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Joinpoint p-values
The Joinpoint Regression Program performs a series of hypothesis tests that test the null hypothesis of
ka joinpoints against the alternative hypothesis of kb joinpoints, where ka and kb change for each
hypothesis test. Each p-value corresponds to this type of test. The p-value is an estimate of the
probability, under the assumption that there are only ka joinpoints, of observing data that look more like
a joinpoint model with greater than ka joinpoints than the data that we have in fact observed.
As the permutation test is a randomization test, it depends on the random number generator. For greater
consistency in the permutation test p-values obtained if one were to change the seed for each run, we
suggest running the program for at least 4499 permutations. For this reason, the default number of
permutations is now 4499 in the current version of the Joinpoint Regression Program. Choice of the
number of permutations selected by the user is a trade off between computer time and consistency of the
p-values obtained.

Related content
Random Number Generation

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Random Number Generation
The program performs permutation tests to select the number of joinpoints. Since fitting all N! possible
permutations of the data would take too long, the program takes a Monte Carlo sample of these N! data
sets, using a random number generator to calculate p-values for a series of permutation tests.
Here we discuss the implications of the choice of the number of permutation data sets, say N. The
program runs faster with smaller values of N, but it gives better precision for the p-value with larger
values of N. In addition, a larger N reduces the probability that another analysis of the same data might
get a different answer when run with different random number generator seeds.
Computer programs can produce pseudo-random numbers through algorithms that mimic randomness,
which we use to shuffle or permute the errors. The algorithms use a seed or seeds to start the algorithm.
These seeds can be used to produce repeatable pseudo-random numbers. The problem of two analyses
obtaining different answers from the same data is addressed by this program by specifying default
random number generator seeds. Thus, as long as no parameters are changed (including the random
number generator seed and N), repeats of the analyses will produce the same results. Otherwise, two
runs of the same analysis using different seeds could get different answers.
To get an idea how results would change for someone using different random number generator seeds,
we list some confidence intervals for p-values below. For example, with N=999 Monte Carlo samples if
you obtained a p-value of .04 from the program there is an approximately 99% chance that another
researcher repeating the analysis with N very large (i.e., an ideal situation with N -> infinity) would
obtain a p-value between .025 and .0577.
N=99
lower 99% ci
estimate upper 99%
ci
0.0000 0.01 0.0521

0.0034 0.04 0.1065

0.0069 0.05 0.1218

0.0111 0.06 0.1364

0.0325 0.10 0.1910

0.2702 0.40 0.5281

N=999
lower 99% ci
estimate upper 99%
ci
0.0031 0.01 0.0199
0.0250 0.04 0.0577

0.0331 0.05 0.0694

0.0415 0.06 0.0810

0.0762 0.10 0.1259

0.3595 0.40 0.4402

N=9999
lower 99% ci
estimate upper 99%
ci
0.0075 0.01 0.0127
0.0350 0.04 0.0452

0.0445 0.05 0.0558

0.0540 0.06 0.0663

0.0923 0.10 0.1079

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017
0.3873 0.40 0.4127

N=99999
lower 99% ci
estimate upper 99%
ci
0.0092 0.01 0.0108
0.0384 0.04 0.04116

0.0482 0.05 0.0518

0.0581 0.06 0.0620

0.0978 0.10 0.1025

0.3960 0.40 0.4040

Typically, you should allow the computer to use the default seeds for the random number generator. By
using these default seeds, one can duplicate results from a previous Joinpoint session even though the
program uses Monte Carlo sampling. Although not recommended in general, the Joinpoint Regression
Program allows one to change the default seed by selecting Session > Preferences from the File menu.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Early Stopping Options
Since Hudson' s Method is computationally intensive, Early Stopping Options have been added so that
not all permutations need to be analyzed. The options are:
Fixed - all permuted data sets are analyzed, with the default being 4499.
B-Value - the maximum number of permuted data sets to be analyzed is determined by the
significance level specified, using a less conservative approach. See Fay et al. for details (Fay,
MP, Kim, H-J, and Hachey, M. (2007) " On using Truncated Sequential Probability Ratio Test
Boundaries for Monte Carlo Implementation of Hypothesis Tests" (Journal of Computational and
Graphical Statistics 16, 946-967).
Curtailed - the maximum number of permuted data sets to be analyzed is determined by the
significance level specified, using a more conservative approach.
This option is not currently available for the Grid Search Method but may be added in a future version of
the software.

Related content

References

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Bayesian Information Criterion (BIC) Details
The equation for computing the BIC for a k-joinpoint model is:
BIC(k) = ln{SSE(k)/#Obs} + {#Parm(k) /#Obs} * ln(#Obs),
where SSE is the sum of squared errors of the k-joinpoint regression model, #Parm(k)=2*(k+1) is the
number of parameters of the k-joinpoint model and #Obs is the number of observations.
The k-joinpoint model with the minimum value of BIC(k) is selected as the final model.

Related content
Data Errors That Stop An Analysis

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Modified BIC
Zhang and Siegmund (2007, Biometrics) discussed that in the context of change-point problems, the
traditional BIC does not satisfy the technical assumptions of Schwarz (1978, Annals of Statistics) and
proposed a modification to improve its performance. The MBIC in Joinpoint regression is derived as an
asymptotic approximation of the Bayes factor and is of the form:

where n is the number of observations, Γ(z) is the gamma function:

and

for a k-joinpoint model with the values of the independent variable

and the joinpoints estimated as . Note that a+ = max (a,0).

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Include/Exclude Select Cohorts
The “Include/Exclude Select Cohorts” option should be used when you want Joinpoint to analyze a subset
of the cohorts found in your input data file. With this feature, you can select those cohorts you wish to
analyze and those that you do not.
Regardless of the number of cohorts you select to include in your analysis, all cohorts will still be
present in the output data file. Those that were “excluded” from the analysis, will be marked with an
and labeled as “Excluded By User”. Excluded cohorts will only display the observed data values in the
graph and data grids.
If one or more cohorts are excluded when you execute your session, Joinpoint will display those cohorts
in the “Issues Found in Cohort Data” window. The cohorts will be marked as “User Selected to Exclude
This Cohort” and the user can choose to cancel the job.

Selecting Cohorts to include using the “By-Variable” categories:

Choose By-Variables to Include
This list will contain all categories for each of your by-variables. You can use this list to include
or exclude cohort categories (groups) in your Joinpoint analysis.
Options:

Add By-Group –This

button will add all cohorts
containing the selected
By-Variable category to
the “Cohorts to Be
Included” list.
Remove By-Group – This
button will remove all
cohorts containing the
selected By-Variable
category from the
“Cohorts to Be Included”
list.

Selecting Individual Cohorts to Include:

Choose Individual Cohorts to Include
This list will contain all the individual cohorts found in your input data file. By using this list,
you can include or exclude individual cohorts from your Joinpoint analysis.
Options:
Add All – this button will add all items in the “Choose Individual Cohorts to Include” list to
the “Cohorts to Be Included” list.
Add -> - this button will add any selected items in the “Choose Individual Cohorts to Include”
list to the “Cohorts to Be Included” list.
<- Remove – this button will remove all selected items from the “Cohorts to Be Included”
list.
Remove All – this button will remove all items in the “Cohorts to Be Included” list.

Cohorts T hat Will Be Modeled By Joinpoint:

Cohorts to Be Included: This list will contain all cohorts to be analyzed by Joinpoint. If the
list is empty, then the Joinpoint session cannot be executed. There must be one or more
cohorts listed.

Total Possible Cohorts: In the lower left-hand corner of the “Select Cohorts To Run” dialog will be
the number of cohorts found in your input data file.
Total Number of Cohorts to be Processed: In the lower right-hand corner of the Select Cohorts To
Run” dialog will be the total number of cohorts you have selected to have Joinpoint analyze.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Minimum Number of Observations Required
The minimum number of observations required to run a Joinpoint Analysis (using the Grid Search method)
is determined by the following formula:
Min. Num. Obs. = (2 * NumObsEnd) + (MaxJP -1) * NumObsBetween + MaxJP
NumObsEnd: The minimum number of observations from a Joinpoint to either end of the data,
excluding the first or last Joinpoint if it falls on an observation.
NumObsBetween: The minimum number of observations between two joinpoints, excluding any
Joinpoint if it falls on an observation.
MaxJP: The maximum number of joinpoints, determined by subtracting the user defined minimum
number of joinpoints from the user defined maximum number of joinpoints.

If you receive an error message stating that there are too few observations to fit the specified model,
check these settings on the Method and Parameters tab and adjust them to fit this formula.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Advanced Analysis Tools Tab
This is where you can select your:
Pairwise Comparison options
Jump Model and Comparability Ratio Model options
Multi-group clustering (will be available in the future).

Related content
Pairwise Comparison
Jump Model / Comparability Ratio Model
Multi-Group Clustering

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Pairwise Comparison
The main goal of the comparability test is to compare two sets of trend data whose mean functions are
represented by joinpoint regression. Specific interests are testing:

1. whether two Joinpoint regression functions are identical (test of coincidence) or

2. whether the two regression mean functions are parallel (test of parallelism)

and the details can be found in Kim et al. (2004):

H-J. Kim, M. P. Fay, B. Yu, M. J. Barrett and E. J. Feuer (2004), Comparability of Segmented Line
Regression Models, Biometrics, 1005-1014.

Consider the Joinpoint regression mean function at the j th year of the i th group, x = xij,

where ki is the unknown number of joinpoints, the T' s are the unknown joinpoints, the β' s and the δs are
the regression coefficients, and a+ = a for a > 0 and 0 otherwise. In general, the two groups may have
different numbers of joinpoints, say ki for the i th group, but the program fits both groups with a larger
model with kmax joinpoints.
Computation Notes:
These tests are selected using the Pairwise Comparison option under the Advanced Analysis
Tools tab.
Using these tests could substantially increase the computation time.
Autocorrelated errors models are not available with these tests.
All 2-way combinations of the innermost by-variable are tested.

What are the hypotheses?

The null hypotheses for the test of coincidence and the test of parallelism are

and

respectively.

What is the test static?

The test statistic with k = kmax is

where RSS denotes the residual sum of squares obtained from the least squares fitting, and d1,k and d2,k
are appropriate degrees of freedom.

How is the kmax chosen?

The recommended data driven choice of kmax is the largest number of joinpoints estimated under the null
and alternative models:

where is the number of joinpoints estimated for the two groups combined under the null model, and
and are the numbers of Joinpoints estimated separately.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

How is the p-value estimated?
The p-value of the test is estimated using the permutation distribution of the test statistic. The
residuals, obtained under the null model, are permuted to generate the permutation distribution of the
test statistic, and the p-value is estimated as the proportion of the permutation data sets whose test
statistic values are greater than or equal to the original test statistic value.

Significance Level
Set the significance level for the test of coincidence or the test of parallelism.

Max Number Of Randomly Permuted Data Sets

The program performs a permutation procedure for the test of coincidence or the test of parallelism.
Since fitting all N! possible permutations of the data would take too long, the program takes a Monte
Carlo sample of these N! data sets, using a random number generator. Specify the size of the Monte
Carlo sample of permuted data sets.
The minimum allowed number of permutations is 1000. The maximum number is 10,000. For greater
consistency in the p-values obtained if one were to change the seed for each run, we strongly
recommend running the program for at least 4499 permutations. You may use a smaller number of
permutations to speed up the calculations, but the permutation test may produce less consistent
results. Unless you have a good understanding of the implications of changing the number of
permutations, we recommend against it.

Related content
Data Errors That Stop An Analysis

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Jump Model / Comparability Ratio Model
T he Jump Model / Comparability Ratio Model in the Joinpoint software provides a direct
estimation of trend data (e.g. cancer rates) where there is a systematic scale change, which
causes a “jump” in the rates, but is assumed not to affect the underlying trend.

NOTES:
T he Jump and Comparability Ratio models are available in Joinpoint as of version
4.4.0.0, but are labeled as an "alpha" feature and considered to be still in
development.

Models
The jump and comparability ratio model in the Joinpoint software provide a direct estimation of trend
data (e.g. cancer rates) where there is a coding change, which causes a “jump” in the rates, but is
assumed not to affect the underlying trend. The user supplies the location of the last data point before
the coding change occurs. The software automatically locates the discontinuity or “jump” halfway
between this last data point and the next one. The software currently only allows for a single “jump”.
For example, the ICD coding changes (Anderson et al. 2001) from ICD-9 to ICD-10 for classification rules
for selecting underlying causes of death. The last year that ICD-9 was used was 1998, and ICD-10 was
implemented starting in 1999. If the user had entered annual data, they would enter 1998, and the
software would place the “jump” at 1998.5.
There are two different options for how a model can be implemented to allow for a coding change:
Comparability Ratio Model: In some cases a “double coding” study has been conducted, where a
certain number of cases have been coded under both the old and new systems. In such cases a
“comparability ratio” and its standard error can be externally entered, where the comparability ratio is
defined as:

For example, the National Center for Health Statistics (NCHS) maintains a web page on comparability
ratios derived from double coding studies, and has a report (Anderson et al.(PDF)) which contains the
comparability ratios and associated standard errors for a long list of underlying causes of death. In
other cases (for example, changes in how stage of disease is coded for cancer), there may be a number of
years where both staging systems were used simultaneously, and the comparability ratio can be derived
from these years of overlap. In the comparability ratio model, the data before the jump is multiplied by
the comparability ratio (and new standard errors are computed using the delta method utilizing both the
standard error of the data point and the standard error of the comparability ratio). The new series is fit
using a standard joinpoint model, and the transformed data points and fitted values are transformed
back to the original coding (by dividing by the comparability ratio) prior to graphing.
Jump Model: In other situations, there may be no “double coding” studies available. In these
situations, the “jump” is a parameter in the model (rather than entered externally) and is estimated
directly. It is the underlying assumption of this model (i.e. that the same trend continues before and
after the jump) that allows this type of model to be estimated. For the jump model, the user only has to
enter the location of the last data point before the coding change occurs. The model can be described as
below. For y = log r, where r is the rate at a given time, and x is the time, assume:

where are unknown change-points, s is the known location of the coding change, are independent
errors, the symbol a+ represents the a for a>0 and 0 if a≤0, I is an indicator function equaling to 1 if the
condition is satisfied, ’s , ’s, ’s and γ are the model parameters to be estimated. The parameter γ
represents the jump and exp(γ ) represents the ratio of rates coded under the new coding scheme divided
by rates coded under the old coding scheme (i.e. the comparability ratio estimated from the jump model).
An example is provided to demonstrate how incorporating a coding change (even one that is relatively
modest) can change the overall conclusions about the trends. For melanoma, the published ICD-9 to ICD-
10 comparability ratio is 0.9677 (SE = 0.0032, 95% CI (0.9614, 0.9741)). Figure 1 shows US melanoma
mortality for all races and both genders from 1992 through 2014 using the standard joinpoint model, the
comparability ratio model, and the jump model. The standard joinpoint model found no joinpoints, and
shows a flat trend with a non-significant APC of -0.06% per year. A comparability ratio less than one (i.e.
0.9677) forces a sudden drop in the trend line between 1998 and 1999. With this shift, the comparability
ratio model shows a joinpoint in 2010 with a significant rise of 0.30% per year prior to 2010 and a
significant fall from 2010 to 2014 of 1.43% per year. The jump model estimates a similar comparability
Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017
ratio of 0.9444, and finds a joinpoint at 2009 with a significant rise of 0.50% per year prior to the
joinpoint, and a significant decline of 1.19% after. These are qualitatively different results when the
coding change from ICD-9 to ICD-10 is taken into account.
Motivated by this example, the complete set of US melanoma mortality results in NCI’s 2017 Cancer
Statistics Review, with results from the standard joinpoint model, jump model and comparability ratio
model, can be found on the Melanoma example.
Figure 1. Standard joinpoint model, jump model, and comparability ratio model for all races
and both genders US mortality for melanoma, 1992-2014. T he estimate of the comparability
ratio estimated from the jump model is 0.9444 with standard error = 0.0116 (the estimate of
the comparability ratio is statistically different than 1). T he comparability ratio (input from a
double coding study) is 0.9677 with standard error = 0.0032 (the comparability ratio is
statistically different than 1).

Input Parameters
For the jump model, the user supplies:
The location where the coding changes start to occur.
All other parameters set in the standard joinpoint regression model such as minimum and
maximum number of joinpoints, permutation methods, etc.
For the comparability ratio model, the user supplies:
The location where the coding changes start to occur.
The size of the comparability ratio. The ratio is limited to [0.01, 100].
The variance of the comparability ratio. The variance is limited to [0, 100]. If you use the
homoscedastic error option for Joinpoint (i.e. no standard errors) the variance of the
comparability ratio will be set to " NA" (not applicable) and will not be able to be
changed. If the heteroscedastic error option is used, the user may enter zero if they do
not have an estimate of the variance of the comparability ratio, but they are encouraged
to enter the variance if they have one. All other parameters set in the standard joinpoint
regression model such as minimum and maximum number of joinpoints, permutation
methods, etc.
Restrictions:
The jump point must be 4 or more data points from either end of the data.
The jump and comparability ratio models can only be run with the Grid Search method.
The jump and comparability ratio models can only be run when the Log Transformation is
set to Yes.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Output
Both the jump and comparability ratio models will output all the items/statistics produced by the
standard joinpoint model.
The jump model will produce additional statistics related to the jump value and also provide an
estimate of the comparability ratio: comparability ratio = exp(jump value). While in almost all
situations the comparability ratio and the jump values should both be either statistically
different from 0 and 1 respectively (since they are related by a transformation, occasionally, the
comparability ratio and jump value do not show the same statistical significance. This minor
discrepancy rarely happens. It is due to an approximation from the delta method applied to
finding the confidence interval of the comparability ratio of the jump model.

Which model to use?

Considerations of which model to use could include:
No “double coding” study may be available, in which case the jump model is the only option.
The “double coding” study on which the comparability ratio is estimated usually is conducted
using data from calendar years close to when the coding change occurred. However, the actual
ratio may vary as one gets further from the year the coding change occurred. The jump model
implicitly uses all of the years before and after the coding change to estimate a best fitting
jump.
The population for which the “double coding” study was conducted may differ from the population
for your data series (e.g. the double coding study may have been conducted for all races and your
data series may be for blacks, or the double coding study may have been conducted in one cancer
registry, but the data series is for a different registry). The jump model has some advantages in
this case because it is estimated directly using the data series of interest.
A joinpoint may be close to the location of the jump. In this case, the estimate of the size of the
" jump" in the jump model may be partially confounded with the slope before and after the
joinpoint. For example, a series for non-Hispanic white males for oral cavity and pharynx cancer
mortality is shown in Figures 2 thru 4. The Standard Joinpoint model is shown in Figure 2 and
displays an annual percent change (APC) of -1.77% from 1992 through 2005 and a non-
statistically significant APC of 0.63% from 2005 to 2013. The comparability ratio (estimated
from a double coding study) is 0.9603 and shows an APC of -1.36% from 1992 through 2005, and a
non-statistically significant APC of 0.53% from 2005 to 2013 (Figure 3). The Jump model (Figure
4) estimated a comparability ratio of 0.8844 which is further from the null value of 1 as
compared to the value from the comparability ratio model (0.9603). An examination of this model
shows a joinpoint at 1997 which is very close to the coding change at 1998.5. The upward APC
segment from 1997 through 2002 is only made possible by the large compensating downward
jump, and appears to be a spurious result.

Figure 2. Standard joinpoint model for White non-Hispanic Male US Mortality for Oral Cavity
and Pharynx Cancers, 1992-2013.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Figure 3. Comparability ratio model for White non-Hispanic Male US Mortality for Oral Cavity
and Pharynx Cancers, 1992-2013. T he comparability ratio (input from a double coding study)
is 0.9603 with standard error = 0.0039.

Figure 4. Jump model for White non-Hispanic Males US Mortality for Oral Cavity and Pharynx
Cancers, 1992-2013. T he estimate of the comparability ratio is 0.8844 with standard error =
0.0311.
Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017
 The underlying variability of the data may make estimation of a small or modest jump size
impossible. In small sub-populations (e.g. API, AI/AN, rare cancer sites, or small geographic
areas), such situations may occur. Since the jump is a parameter estimated in the jump model, a
test can be conducted if the jump is statistically greater than zero. In cases where the jump size
is insignificant, the comparability ratio model may be the better choice. Even in situations where
the jump is statistically significant, if there is large variability, and the comparability ratio is
small, one should be wary of estimates of the jump which differ widely from the comparability
ratio.
Fitting both models is usually a good idea. The “safer bet” is usually the comparability ratio
model since the jump model can occasionally produce anomalous results. However, the jump
model can offer better estimates if the double coding study was estimated from a limited range
of years or from a population which differs from the data series being modeled. In many cases
where the estimates from the two models are similar, the jump may be preferred.
In cases where no double coding studies exist, one should be cautious in accepting the results of
the jump model. The analyst should evaluate the size of the underlying variability of the data,
and should be suspicious of joinpoint segments which start or end close to the jump location,
and the slope of the segment seems to be “compensating” for the size of the jump.

In general, it is best to do a careful examination of the models using the criteria above before deciding
which model to select. An algorithmic approach may be desired in the case of many data sequences to be
analyzed. See Melanoma for an example of an algorithmic approach.

Display
The software allows viewing of a graph for each joinpoint model, from the model with the minimum
number of joinpoints to the model with the maximum number of joinpoints. The graph will show a
“jump” at the location where the coding change occurs. The software also allows for graphs of
different models and/or cohorts to be combined.
Citation & Reference
For more technical details on the Jump Model / Comparability Ratio Model, please contact
technical support.
Anderson RN; Miniño AM, H.D.R.H., Comparability of Cause of Death Between ICD–9 and ICD–10:
Preliminary Estimates. National Vital Statistics Reports, 2001. 49(2): p. 2001-1120.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Multi-Group Clustering
Future Enhancement

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Preferences
File Preferences
Access the File Preferences window by selecting Preferences from the File menu. Make changes to the
fields as necessary. When you are finished, click OK.

Default Data Directory

The default directory to be used when browsing for your data from a Joinpoint session can be set
here. In order to set the directory, use the browse button and select the directory to be used as the
default.

Display Splashscreen on startup

When the Joinpoint program is started, a splashscreen will automatically appear for 5 seconds. This
option can be used to display or hide the screen.

Maximum Number of T hreads

With the introduction of processing multiple cohorts simultaneously (threading), an option to adjust
the number of threads is provided. The number of threads that Joinpoint will try to use when running
a session can be set here. There number cannot be set above the maximum for the computer that
Joinpoint is installed on.
A reason to lower the number of threads is to provide other concurrently running Windows programs
more resources under Windows. The more threads that Joinpoint tries to use, the slower other
concurrently running Windows programs might run.

T urn on all Joinpoint warning messages

Joinpoint has various ' nag' warning messages that appear to users to warn them of potential
problems. A ' nag' warning is a warning that users can choose to turn off and when turned off it will
remain off until this option is adjusted. When this button is clicked, all ' nag' warning messages
previously turned off will be turned back on.

Session Preferences
Access the Session Preferences window by selecting Session Preferences from the Session menu (only
available when a session is open). Make changes to the fields as necessary. When you are finished, click
OK.

Alpha Levels - Joinpoint Locations

The significance level used to test the Type I error when selecting a model. The default value for this
field is 0.05. Alpha level values must be greater than 0 and less than 1.

Alpha Levels - APCs

The significance level used to test whether an APC value is significantly different from zero. The
default value for this field is 0.05. Alpha level values must be greater than 0 and less than 1.

Alpha Levels - AAPCs

The significance level used to test whether an AAPC value is significantly different from zero. The
default value for this field is 0.05. Alpha level values must be greater than 0 and less than 1.

Alpha Levels - Jump Value and Comparability Ratio

The significance level used to test whether a Jump Value or Comparability Ratio is significantly
different from zero. The default value for this field is 0.05. Alpha level values must be greater than 0
and less than 1.

Random Number Generation

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Typically, you should allow the computer to use the default seeds for the random number generator.
By using these default seeds, one can duplicate results from a previous Joinpoint session even
though the program uses Monte Carlo sampling. Although not recommended in general, the Joinpoint
Regression Program allows one to change a default seed here.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Statistical Notes
Let K min and K max be the minimum and maximum for the number of joinpoints, respectively. First the
program goes through each of the k-joinpoint models, K min ≤ k ≤ K max. For each of the models, the
program estimates the regression parameters with the smallest sum of squared error (SSE, or smallest
weighted SSE). Statistics related to each of the k-joinpoint models are discussed in the selections under
Related Items. The sequential permutation test procedure to choose the best joinpoint model is
discussed in detail in Kim et al. (2000), and only briefly described here.

Related content
Statistics Related to the k-joinpoint Model
Significance Level of each Individual Test in a Sequential Testing Procedure
References

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Statistics Related to the k-joinpoint Model
Several statistics related to the k-joinpoint model are described here. All statistics come directly from
Lerman (1980).

Related content
Notation
Parameterizations
SSE and MSE
Estimated Joinpoints
Estimated Regression Coefficients (Beta)
Estimated APC
AAPC Comparison

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Notation
n is the total number of data points.
k is the number of joinpoints in the model.
p is the total number of parameters in the model, including the joinpoint parameters. For our
models p=2k+2.
Q k is the sum of squared errors (SSE) from the model that minimizes SSE with k joinpoints. If a
weighted analysis is done (anything but heteroscedastic errors option= constant variance) then
Q k represents the minimum weighted SSE.
Q x,j,k is the (weighted) SSE from the model that minimizes (weighted) SSE with k joinpoints and
with the jth joinpoint occurring at x.
F-1a,b(p) is the pth quantile of the F distribution with a and b degrees of freedom.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Parameterizations
The program outputs parameter estimates from two different parameterizations: the " general
changepoint" parameterization (GCP), and the " standard" parameterization (SP) of Kim et al. (2000).
The standard parameterization is (see Kim, et al. 2000, equation 1),

where (a)+= a if a > 0 and 0 otherwise.

The general changepoint parameterization is,

where I(A) is the indicator function for {A}, τ0 = min(x), and τk+1 = max(x), and under the constraint that
E[y|x] is continuous at τj.
For the relationship between the parameterizations see Table 1 and Appendix A.
Table 1: Parameter Transformations for Different Models
Output Label Standard General Changepoint

Intercept 1

Intercept j, j ≥ 2

Slope 1

Slope j, j ≥ 2

Slope j - Slope (j-1), j ≥ 2

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Appendix A
Relationship Between Standard and General Changepoint
Parameterizations
Rewrite standard,

Rewrite GCP,

Standard General Changepoint Parameterizations

β = β01
β = β11
δj = β1,j+1 - β1j

τj

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

SSE and MSE
Sum of squared errors (SSE) is actually the weighted sum of squared errors if the heteroscedastic errors
option is not equal to constant variance. The mean squared error (MSE) is the SSE divided by the degrees
of freedom for the errors for the constrained model, which is n-2(k+1).
The minimum SSE for a k-joinpoint model is calculated using Lerman' s grid-search method (1980) based
on Kim et al' s standard parametrization (Equation 1). The corresponding values for (τ1,...,τk) and (β0,β1,
δ1,...,δk) are the estimates of joinpoints and regression coefficients, respectively.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Degrees of Freedom
When the Grid Search Method is used, the degree of freedom for the estimated regression coefficient is
d=n*-2(k+1), where k is the number of joinpoints and n* is the effective number of data points after
deleting offending observations, data points that are on joinpoints.
When Hudson' s Method is used for a continuous fitting, the degree of freedom is adjusted in a
continuous manner instead of subtracting the offending observations. The idea is to delete, with higher
probabilities, the observations whose x-values are closer to joinpoints, while no more than one
observation is deleted around each joinpoint. See Kim et al. (Kim, H.-J., Yu, B. and Feuer, E.J. (2008)
" Inference in segmented line regression: A simulation study" , Journal of Statistical Computation and
Simulation 78(11), 1087-1103) for details.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Estimated Joinpoints
If k > 0 then the output lists the estimated joinpoints. The associated confidence intervals (CI) come
from Lerman (1980) equation 6 using C2α . Explicitly, the100(1-α)% confidence interval for the jth of k
joinpoints includes all values of x from the grid such that Q x,j,k ≤ C2α , where

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Estimated Regression Coefficients (Beta)
The output is a combination of the two parameterizations (see Table 1). The estimates of (β0,β1,δ1,...,δk)
come from the grid-search method. The estimates of (β01,β11,...,β0,k+1,β1,k+1) are calculated based on
Table 1.
However, the standard errors of the regression coefficients are estimated under the GP model (Equation
(2)) without continuity constraints. Following Lerman' s implementation, (Lerman; 3rd paragraph, page
79, 1980; Feder, Section 4, page 69, 1975), the data points that are on the joinpoints are deleted. Then
conditioned on the partition implied by the estimated joinpoints (τ1,...,τk) , the standard errors of
(β01,β11,...,β0,k+1,β1,k+1) are calculated using unconstrained least square for each segment. If there are
segments with zero or one observation (not including the joinpoints at both ends), then a generalized
inverse is used to calculated the covariance matrix. The standard error of the difference in slopes, δj, is
the square root of the sum of the squared standard errors (variance) for the two consecutive slopes β1j
and β1,j+1

The test statistic (U) is the parameter estimate divided by the standard error. The test statistic has a t
distribution with d degrees of freedom where d is defined as follows. Let nJ be the number of data points
that are on joinpoints. The effective number of data points is n* =n-nJ. Let k0 and k1 be the number of
segments with zero or one observation. The effective number of parameters (rank of the design matrix) is
p* =2(k+1)-2k0-k1 and the degrees of freedom d=n* -p* . For the default option where the minimum number
of data points between two joinpoints (excluding any joinpoint that falls on an observation) is two,
k0=k1=0 and d=n* -2(k+1). For testing H 0: β = 0 the p-value is calculated as 2{1-t d( |U| )}, where t d is a t
distribution with d degrees of freedom.
If a standard error cannot be calculated, then the associated statistics will not be displayed.

Related content
Data Errors That Stop An Analysis

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Estimated APC
When the model option on the Specifications tab is ln(y)=xb, then the output calculates the estimated
annual percentage rate change (APC). For any segment with slope β, the APC is 100{ exp(β) -1 }. The
100(1-α)% confidence limits are:

where d is the degrees of freedom and s is the standard error for the slope listed in the output (i.e., from
the unconstrained linear models), and t d-1(q) is the qth quantile of a t distribution with d degrees of
freedom.
The p-value for a two-sided test that the true APC is zero is calculated based on a t distribution.

Related content

Estimated Regression Coefficients (Beta)

Average Annual Percent Change (AAPC)

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

AAPC Comparison
When the test for pairwise comparison is selected, and the two groups are not parallel or coincident, the
AAPCs for the two groups are tested to determine if they are statistically different. For details, see
Average Annual Percent Change (AAPC).

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Significance Level of each Individual Test in a
Sequential Testing Procedure
The Joinpoint program uses a sequence of " permutation" tests to select the final model. Each one tests
the null hypothesis H 0: k = ka against the alternative hypothesis H 1: k = kb. The procedure begins with
ka= K min and kb= K max. If the null is rejected, then increase ka by 1; otherwise, decrease kb by 1. Let

be the final selected number of Joinpoints.

Because multiple tests are performed, Bonferroni adjustment is used to ensure that the approximate
overall type I error is less than the specified significance level (significance level is also called the α-
level, default α=.05). Each of these permutation test are carried out a significance level of α 1=α/(K max-
K min), i.e., if the p-value < α 1, then it rejects the null.
The Bonferroni adjustment is conservative because the actual overall significance level is usually less
than the nominal level α. The new adjustment procedure controls the overall over-fitting probabilities.

Let α(ka,kb) be the significance level of each individual test H 0: k=ka vs. H 1: k=kb. The new procedure
set α(ka,kb)=α/(K max-ka). Notice that the individual significance level depends on the number of
joinpoints ka under the null. Consider an example where K min= 0 and K max= 4. The new procedure has
the following properties:

If we like to bound these over-fitting probabilities by α, then we can assign different values for each
α(ka,kb). That means, we can achieve a better power by setting
α(0,4)= α(0,3)= α(0,2)= α(0,1)= α/4
α(1,4)= α(1,3)= α(1,2)= α/3
α(2,4)= α(2,3)= α/2
α(3,4)= α

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

References
Basic Method
Kim, H-J, Fay, M.P., Feuer, E.J., and Midthune, D.N. (2000) " Permutation Tests for Joinpoint Regression
with Applications to Cancer Rates" , Statistics in Medicine 19, 335-351. (correction: 2001;20:655).
Correction to Table 1(a) of Kim, et al. is provided as a PDF at
http://surveillance.cancer.gov/documents/joinpoint/table1.pdf.

This is a paper where joinpoint regression is applied to describe cancer rates and the
permutation test is proposed to determine the number of significant joinpoints. The grid
search of Lerman (1980) was used to fit the segmented regression function and the p-value
of each permutation test is estimated using Monte Carlo methods, and the overall
asymptotic significance level is maintained through a Bonferroni adjustment.

Background Papers for Basic Method

Feder, P.I. (1975) " On Asymptotic Distribution Theory in Segmented Regression Problems: Identified
Case" , Annals of Statistics 3, 49-83.
Feder studied asymptotic properties of the least squares estimators in multi-segment regression and
proved that under some technical conditions on the independent variable, the least squares
estimators are consistent and asymptotically normal.
Lerman, P.M. (1980) " Fitting Segmented Regression Models by Grid Search" , Applied Statistics 29: 77-
84.
Lerman proposed the grid search method to fit segmented line regression where the joinpoint
estimates occur at discrete grid points, and studied asymptotic inference using asymptotic normality
proved by Feder (1975).
Hudson, D. (1966) " Fitting segmented curves whose join points have to be estimated" , Journal of the
American Statistical Association 61, 1097-1129.

Hudson proposed a procedure to fit a segmented regression curve whose joinpoints can be
estimated anywhere in the data range. In the first stage, the model is fit to every feasible
partition of the data without imposing continuity across the change-points. We then
examine the locations of the intersection points of the least square lines. When the
intersection point of the two adjacent least squares lines is not between the two data
points which separate the segments, we make an adjustment in the fit by mathematically
imposing the continuity constraints at one of the two boundary data points. The final
estimates are obtained by searching for the global minimum of residual sum of squares.

Fitting Joinpoints in Continuous T ime

Yu, B., Barrett, M., Kim, H-J, and Feuer, E.J. (2007) " Estimating Joinpoints in Continuous Time Scale for
Multiple Change-Point Models" , Computational Statistics and Data Analysis 51, 2420-2427.

In this paper, we extend the Hudson' s continuous fitting method to a multiple joinpoint
model and discuss some practical issues in the implementation. We also compare
computational efficiencies of the Lerman' s grid search and the Hudson' s continuous fitting.

Early Stopping Rules for the Permutation T est

Fay, M.P., Kim, H-J, and Hachey, M. (2007) " On using Truncated Sequential Probability Ratio Test
Boundaries for Monte Carlo Implementation of Hypothesis Tests" , Journal of Computational and
Graphical Statistics 16 (4), 946-967.

Joinpoint selects a final model conducting a series of permutation tests and we can save
computation time by using sequential stopping boundaries. This paper proposes a
truncated sequential probability ratio test boundary to stop resampling when the early
replications indicate a large enough or small enough p-value, and studies its properties.

Comparing T wo Joinpoint Regression Lines

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Kim, H-J, Fay, M.P., Yu, Binbing, Barrett, M.J., and Feuer, E.J. (2004) " Comparability of segmented line
regression models" , Biometrics 60, 1005-1014.

We propose a procedure to compare two segmented line regression functions, especially to

test (i) whether two segmented line regression functions are identical or (ii) whether the
two mean functions are parallel allowing different intercepts. A general form of the test
statistic is described and then the permutation procedure is proposed to estimate the p-
value of the comparability test.

AAPC
Clegg, L.X., Hankey, B.F., Tiwari, R., Feuer, E.J., Edwards, B.K. (2009) " Estimating average annual
percent change in trend analysis" , Statistics in Medicine 28(29): 3670-8.
Studies on the performance of Joinpoint
Kim, H-J, Yu, B., and Feuer, E.J. (2008) " Inference in segmented line regression: A simulation study" ,
Journal of Statistical Computation and Simulation 78:11, 1087-1103.
Via simulations, this paper empirically examines small sample behavior of asymptotic confidence
intervals and tests, based on Feder (1975)' s asymptotic normality of least squares estimators in
joinpoint regression, studies how the two fitting methods, the grid search and the Hudson' s
continuous fitting algorithm, affect these inferential procedures and also assesses the robustness of
the asymptotic inferential procedures.
Kim, H-J, Yu, B., and Feuer, E.J. (2009) " Selecting the number of change-points in segmented line
regression" , Statistica Sinica 19(2):597-609.

In this paper, we show that under some conditions, the number of joinpoints selected by
the permutation procedure of Joinpoint is consistent. Via simulations, the permutation
procedure is compared with some information-based criteria such as Bayesian Information
Criterion (BIC).

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Executing the Joinpoint Program
Once you have set all controls in the Joinpoint Regression Program, execute the program by clicking
on the toolbar or clicking Run on the menu bar.
A progress meter will be shown on the screen while the Joinpoint calculation engine processes the data
and generates the output. Click Cancel if you would like to halt the calculations before they are
complete.
Once the processing is complete, an output window will automatically be displayed. You will need to
save this output if you would like to access it again at a later time.

Related content
Data Files Containing Cohorts with an Unequal Number of Observations
Processing Multiple Cohorts Simultaneously (Threading)
Viewing Results
Statistical Notes

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Processing Multiple Cohorts Simultaneously
(Threading)
Because the processing of each data cohort in Joinpoint is independent of the others, the ability to
process multiple cohorts simultaneously was introduced into the program starting with Version 4.1.0.
This new functionality could greatly decrease the time it takes to execute sessions in Joinpoint. Also due
to this change, the job progress window has been adjusted to display the status of each cohort being
processed at any given time. As in the past, there will also be an overall job progress meter displayed
during the execution of any Joinpoint session.
Please note that this feature will not work when executing Joinpoint under Windows XP. Windows XP
users will only be able to process one cohort at a time (similar to previous versions of Joinpoint).

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Input Data – Errors and Warnings
When a session is executed, Joinpoint will first conduct a series of edits on the data. Some edit failures
will cause the job to stop, while others will produce warnings.

Related content
Data Errors That Stop An Analysis
Data Warnings Displayed by Joinpoint

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Data Errors That Stop An Analysis
If there are too few data columns in the input data file.
Error Message: Too few data columns in Input Data File.
Reason(s) triggered: This is triggered when the variable positions set in the session go beyond
the number of variables found in any given data line from the input data file. Most likely the
input data file (or data file name) was changed and now it does not match up with what was used
to set up the original session.

If Joinpoint is computing age-adjusted rates and encounters a by-group with a different number of
observations from the previous by-group.
Error Message: You cannot calculate age-adjusted rates from a data file with differing number of
observations per cohort!

If the shift-value specified by the user is greater than the largest interval between independent
variable values.
Error Message: Error with shift value: The shift value cannot be larger than the largest
independent variable interval.

If the Independent variable has a format and Joinpoint encounters a value that is NOT in the
format.
Error Message: " ???" is not a valid value for the independent variable.

If the independent variable has a value that is non-numeric.

Error Message: Error in Data File: Joinpoint cannot process by-groups where independent
variable values are non-numeric.

If two or more independent variable values are found to be equal (within a given cohort).
Error Message: Error in Data File: Joinpoint cannot process by-groups where two or more
independent variable values are equal.

For comparison runs, if the inner-most by variable has only one grouping.
Error Message: You cannot test for pair-wise differences when the inner-most by variable only
has one category.

When age-adjusted rates are computed, if the standard population value used to compute an age-
adjusted rate is zero.
Error Message: Standard Population denominator is zero.

If there are one or more cohorts that do not have enough data points for the specified job
parameters. The following formula is used to determine the minimum number of required data
points for any given cohort:

Min Required Observations = (2 * Min Number of Obs from a Joinpoint to either end of data) +
((Max Joinpoint - 1)*(min num of observations between two joinpoints)) + Max Joinpoint;
Error Message: Too few observations to fit specified model. Please verify that your data is sorted
correctly.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Data Warnings Displayed by Joinpoint
When a session is executed, users will first be notified if any of the following problems exist in
the cohort data:

General Warnings

If the analysis is Log Linear and the dependent variable is a “Count”, each time a dependent
variable value of zero is encountered, 0.5 will be added to the value. The following WARNING
will be displayed once per analysis execution:
Warning Message: There are records with a value of zero for count, 0.5 will be
added to each zero count value.

Cohort-Specific Warnings

If any of the following problems exits, a special window will appear which will list all the
errors for each cohort. From the window, users can either end the job or continue it. If they
choose to continue the job, any cohort with one or more of the following problems will not be
analyzed.

If a record is found with the dependent variable value missing.

Warning Message: Joinpoint cannot process records with missing dependent variable
values. Joinpoint will not analyze by-groups that have this error.

If the dependent variable value is 0 and it is a log linear run, then the dependent variable
cannot have records with the value of zero.

If a record is found with the independent variable value missing.

Warning Message: Joinpoint cannot process records with no independent variable.
Joinpoint will not analyze by-groups that have this error.

If a record is found where the dependent variable value is non-numeric.

Warning Message: Error in Data File: Joinpoint cannot process records with
independent variable values that are non-numeric.

For analyses where standard errors are read in/used:

If the SE is missing - Warning Message: Joinpoint cannot process records with missing
weight values. Joinpoint will not analyze by-groups that have this error.
If the SE is less than or equal to zero - Warning Message: Joinpoint cannot process
records with weight variable <= 0. Joinpoint will not analyze by-groups that have this
error.

When crude rates are calculated by Joinpoint, the following will trigger a warning and the
cohort will not be analyzed by Joinpoint:

There were one or more count values that were non-numeric, missing, or less than zero.
There were one or more population values that were non-numeric, missing, or less than
zero.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

 There were one or more population values of zero.
If the standard error is required for the analysis and one or more count values are zero,
then the associated SEs will be zero. A zero SE is not allowed.
If it is a Log Linear analysis and a count value of zero is found.

When age-adjusted rates are calculated by Joinpoint, the following will trigger a warning
message and the cohort will not be analyzed by Joinpoint:

There were one or more count values that were non-numeric, missing, or less than than
zero.
There were one or more population values that were non-numeric, missing, or less than
zero.
There were one or more standard population values that were non-numeric, missing, or
less than zero.
The total of the standard population does not match the total obtained from the first
cohort.
If it is a Log Linear analysis and the final age-adjusted rate value is zero.
One or more of the following adjustment variable problems exist:

The adjustment variable only has 1 category/group.

The records for the cohort are not evenly divisible by the number of adjustment
variable groups.

When proportions are calculated by Joinpoint, the following will trigger an error:
There were one or more numerator or denominator values that were non-numeric or
negative.
There are one or more records where the denominator was zero.
There are one or more proportions that are less than zero.
There are one or more proportions that are greater than 1.0 (numerator greater than
denominator).

When percentages are calculated by Joinpoint, the following will trigger a warning:

There were one or more numerator or denominator values that were non-numeric or
negative.
There were one or more records where the denominator was zero.
There are one or more percentages that are less than zero.
There are one or more percentages that are greater than 100.00 (numerator greater
than denominator).

Related content
Dependent Variable
Independent Variable

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Viewing Results
This window displays the Output results of your joinpoint analysis. The various tabs (Graph, Data, Model
Estimates, Trends, Model Selection, or Comparison) allow you to view graphs, view the graph data, or
view the other analysis results. You may also access both the graph data and the report information by
exporting these for use by other software packages (Excel, SAS, etc.).
The Cohort Tree control on the left of the Output Window selects which By-Group and Model (i.e. the 1
joinpoint model or the 3 joinpoint model) to display. The bottom of this window displays the Rate setting
and the version of Joinpoint that this output was created in.
The Joinpoint calculation engine displays the output window but doesn' t automatically save or create a
permanent output file. You can save the output as a single binary file and/or you can export the graph
data and the report information as text files. JPO (Joinpoint Output) is the extension Joinpoint uses for
the output files.
Navigating through the different cohorts: Cohort Tree
How different session types are displayed: Viewing Output from the Different Analysis Types
Tabs you can view on this window are:
Graph
Data
Model Estimates
Trends
Model Selection
Comparison

Possible options from this window are (most are accessible from the Output menu):
Saving the results - From the menu, select File > Save As... Output files are saved with a JPO
file extension.
Retrieve the Session - See Retrieving a Session.
View Multiple Joinpoint Models - primarily used to view multiple cohort model results on one
graph. See View Multiple Joinpoint Models
Specify AAPC ranges - See Specify AAPC Ranges.
Display Options - See Display Options.
Display Numeric Values - This option will display cohort numbers instead of names in the cohort
selection control at the top of the window.
Exporting the results to a Text file - See Exporting Results to Text.
Exporting the results to an Excel file - See Exporting Results to Excel.
Printing the results - See Print Options.
Viewing the Output properties - See Output Properties.

Related content

Graph Tab
Data Tab
Model Estimates Tab
Trends Tab
Model Selection Tab
Comparison Tab
Retrieving a Session
Specify AAPC Ranges
Display Options
Exporting Results to Text
Exporting Results to Excel
Print Options
Output Properties

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Cohort Tree
T he Cohort T ree allows you to select the cohort to view in the Output Window.

On the left-hand side of the Output window is the Cohort Tree interface. The Cohort Tree interface
displays the number of join point models generated for each unique combination of By-Variable values
(which is called a cohort). By selecting check boxes in this list, you can display the associated graph and
data for that joinpoint model.
Search
At the top of the Cohort Tree is a box with the text “Search…” in it. When you type text in this box, the
Cohort Tree will highlight all the cohort labels that contain that text (case insensitive). The list will also
jump down to the first cohort label containing the text. Next to this search box, the X button will clear
your search, and the two green arrow buttons will allow you to traverse through the matching search
items. To the right of the green arrows will be a count of the number of cohorts containing the text
entered in the search box.
T ree List
The Cohort Tree allows you to select which joinpoint model you wish to display for a given cohort. Just
select a check box to view the model results for that cohort. Only one item can be selected at a time.
Several icons may also be displayed:
This is the Final Selected Model.
This is the selected model for a grouping not chosen.
This cohort does not have any joinpoints to display or has invalid data.
This cohort has been excluded by the user.

Final Selected Model Only ( )

When this option is checked, only the Final Selected Model for each cohort will be displayed in the
cohort tree.
View Multiple Joinpoint Models
This option will allow you to select more than one model for display at the same time. This is useful if
you want to compare the graphs of multiple joinpoint models within the same cohort, or even to compare
model results from different cohorts. For more information, please see View Multiple Joinpoint Models.
Related Content
 Graph Tab
 Data Tab
 Model Estimates Tab
 Trends Tab
 Model Selection Tab
 Comparison Tab

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Graph Tab
This tab displays a graph of the output data. It is a scatter plot graph with dots for each data point, and
lines between join points.
Possible options from this tab are:
Change graph settings (Titles, Legend, X-Axis, Y-Axis, Colors, Symbols) - See Display Options.
Export the graph to a file - See Exporting Results to Text.
Print the graph - See Print Options.

Related content
Data Tab
Model Estimates Tab
Trends Tab
Model Selection Tab
Comparison Tab
Specify AAPC Ranges

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Data Tab
This tab displays the actual data that is graphed on the graph tab. All X and Y values are listed including
the modeled Y value and the Joinpoint locations.
Possible options from this tab are:
Change decimal point settings or whether APCs are shown - See Display Options.
Export the data to a text file - See Exporting Results to Text.
Export the data to an Excel file - See Exporting Results to Excel.
Print the data - See Print Options.

Related content
Graph Tab
Model Estimates Tab
Trends Tab
Model Selection Tab
Comparison Tab
Specify AAPC Ranges

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Model Estimates Tab
This tab displays the specifications for each of the models run in the analysis. The final, selected model
is first displayed.
Possible options from this tab are:
Change which statistics are displayed, or to change the number of decimal places - See Display
Options.
Export the data to a text file - See Exporting Results to Text.
Export the data to an Excel file - See Exporting Results to Excel.
Print the data - See Print Options.

Related content
Graph Tab
Data Tab
Trends Tab
Model Selection Tab
Comparison Tab
Specify AAPC Ranges

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Trends Tab
This tab displays the estimated joinpoints, APCs, and AAPCs for each of the models run in the analysis.
Note: this window is only displayed if the Log Transformation is Yes { ln(y) = xb }.
Possible options from this tab are:
Change which statistics are displayed, or to change the number of decimal places - See Display
Options.
Change which AAPCs are displayed - See Specify AAPC Ranges.
Export the data to a text file - See Exporting Results to Text.
Export the data to an Excel file - See Exporting Results to Excel.
Print the data - See Print Options.

Related content
Graph Tab
Data Tab
Model Estimates Tab
Model Selection Tab
Comparison Tab
Specify AAPC Ranges
Average Annual Percent Change (AAPC)

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Model Selection Tab
Results for the specified Model Selection Method will be displayed on this tab. The information
presented here provides details on how the best, selected model was chosen.
Please note that when you are on this tab, the model (# Joinpoints) controls are disabled. This is
because the information about all the models are displayed side-by-side on this tab.
Possible options from this tab are:
Change the number of decimal places displayed - See Display Options.
Export the data to a text file - See Exporting Results to Text.
Export the data to an Excel file - See Exporting Results to Excel.
Print the data - See Print Options.

Related content
Graph Tab
Data Tab
Model Estimates Tab
Trends Tab
Comparison Tab
Specify AAPC Ranges

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Comparison Tab
The Comparison Tab of the Joinpoint Output Window displays the results from the Pairwise Comparison
Test.
Please note that the Test for Parallelism or Coincidence is only applicable for the model with the
maximum number of joinpoints. The values on this tab will remain the same for the entire cohort,
regardless of the specific joinpoint model you have chosen.
Possible options from this tab are:
Change the number of decimal places displayed - See Display Options.
Export the data to a text file - See Exporting Results to Text.
Export the data to an Excel file - See Exporting Results to Excel.
Print the data - See Print Options.

Related content
Pairwise Comparison
Graph Tab
Data Tab
Model Estimates Tab
Trends Tab
Model Selection Tab
Specify AAPC Ranges
Parallel Pairwise Comparison and Number of Joinpoints

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Retrieving a Session
The Joinpoint session file is saved as part of the output, so you can select Output > Retrieve Session
from the top of the main window to retrieve the session that was used to create the output, even if that
session file wasn' t specifically saved. This is a convenient way to see exactly what settings were used
to create a particular output file.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Specify AAPC Ranges
When the AAPC Confidence Intervals - Parametric option is selected for the session, the AAPCs can
be computed after the Joinpoint calculations are performed, using the APCs. With this option, the AAPCs
can be modified in the output without having to rerun the session. When viewing the Joinpoint output, to
specify which ranges to display use the Output > Specify AAPC Ranges... menu item from the top of
the main Joinpoint Output window. On this window you can select to display the AAPC for the entire
range of data, the last N observations, or other specified ranges.
When the AAPC Confidence Intervals - Empirical Quantile option is selected for the session, the
AAPCs cannot be computed after the Joinpoint calculations are performed. The Specify AAPC Ranges
window will be unavailable in this case.

Related content

Average Annual Percent Change (AAPC)

AAPC Confidence Intervals

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Display Options
While displaying the Output Results from a joinpoint run, you may want to configure how the information
is displayed. From the Main Menu, select Output > Options... to show the Display Options window.
The Display Options window allows you to customize the output that is being displayed in the Output
Results window. Each of the tabs on this window correspond to a tab on the Output Results window.
Options include:
Graph T ab
The Titles and Labels for the graph.
Whether to show the legend and where to place it.
Show APC Ranges.
Sizing for the X and Y Axis.
The colors and symbols to use when displaying the graph.

Data T ab
Which statistics to display.
The number of decimal places to show for values.

Model Estimates
Which statistics to display.
The number of decimal places to show for values.

T rends
Which statistics to display.
The number of decimal places to show for values.

Model Selection
The number of decimal places to show for values.

Comparison
The number of decimal places to show for values.

Note, the options you select will be saved with that particular output file (if you save the file after
changing the options). If you would like your preferences to always be the default for any new Joinpoint
runs, click on Set as Default after you have set the options the way you desire. In the future, you can
select Load Defaults to bring back your saved defaults. On the Graph tab, only the Legend and Colors
and Symbols sections are saved as defaults.

Related content

Graph Tab
Data Tab
Model Estimates Tab
Trends Tab
Model Selection Tab
Comparison Tab

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Exporting Results to Text
You may access both the graph data and the report information by exporting these as text files for use
with other software packages (Excel, SAS, etc.).
Notes:
Only what is displayed in the Output Window will be exported. For example, if only 1 decimal
place is chosen, only that many decimal places will be in the exported file. Or, if the " Regression
Coefficients" box is unchecked, those statistics will not be included in the exported file.
This option will not be available if the " View Multiple Joinpoint Models" option on the output
viewer is selected.

To export information (data) from a Joinpoint run, first select Output > Export > T ext... from the main
menu and then, depending on the data you desire, select one or more of the following:
Export All – Choose this option to automatically select all of the options below.
Graphs – Choose this option to export the graphs as an image file (.bmp or .jpg). Each graph is
saved as a separate file using the base name and location you specify, plus the specific cohort
name for that graph. You do have the option of exporting graphs in high resolution. Please see
details below.
Data – Choose this option to export the observed and predicted values for each observation.
Model Estimates – Choose this option to export the model estimates and related information.
Jump Model – This feature will be available in a future version of Joinpoint.
APCs – Choose this option to export the APCs.
AAPCs – Choose this option to export the AAPCs.
Model Selection – Choose this option to export information about the model selection process.
Comparison – Choose this option to export information about the Pairwise Comparison test.

Once you have selected one or more of the above options, you will need to address the following:
Output By-variables as – When exporting By-Variables, Joinpoint can export the numeric
representation of the By-variable or the By-variable label (either in quotes or not).
Line Delimiter – Select the appropriate line delimiter for the platform on which you will be
working with the exported file.
Field Delimiter – Each variable field in an export line will be separated by a delimiter. Joinpoint
offers three types of field delimiters: Tab, Comma, Semicolon.
Graphs – When exporting graphs, you have the option of exporting the currently displayed graph,
the graph for the final selected model of each cohort, or the graph for each cohort/model
combination. There is also an option to export the graphs in High Resolution. Once the high
resolution export option is checked, you have the ability to adjust the number of pixels involved
with the width and height of each graph image.
Export Models – Joinpoint can export all of the joinpoint models per cohort or it can export only
the final selected joinpoint model.
Missing Characters – Sometimes data values can be missing. Joinpoint offers several export
options for missing values. For SAS users, a period is considered a missing value.
Remove Flags Marking the Joinpoints - Whether to include a column with Joinpoint flags.
Remove Flags Marking the Best Models - Whether to include an asterisk indicating which
model is the best one.
Output All Models in the Same Column - Whether to include a column indicating the model.
Output Variable Names Before Each By-Group - Redisplays the column headers for each new
By-Group.
Load Defaults - This option will load your default settings for this window (your defaults are
automatically loaded when the window is displayed). The selected options and their filenames
are not included in the default settings.
Set As Default - This option will save all of your settings in this window (excluding the options
and filenames). This button is only enabled after you make changes to your settings.

Related content
Graph Tab
Data Tab
Model Estimates Tab
Trends Tab
Model Selection Tab
Comparison Tab
Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017
Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017
Exporting Results to Excel
You may export your report information to an Excel file. Each section of data will be present in the Excel
file as a separate worksheet.
Notes:
You must have the Excel application installed on your computer for this function to work.
This option will not be available if the " View Multiple Joinpoint Models" option on the output
viewer is selected.

To export information (data) from a Joinpoint run, first select Output > Export > Excel... from the main
menu and then, depending on the data you desire, select one or more of the following:
Filename – The name of the excel file to create. You can press the Browse button to choose a
location for the file.
Export All – Choose this option to automatically select all of the options below.
Data – Choose this option to export the observed and predicted values for each observation.
Model Estimates – Choose this option to export the model estimates and related information.
Jump Model – This feature will be available in a future version of Joinpoint.
APCs – Choose this option to export the APCs.
AAPCs – Choose this option to export the AAPCs.
Model Selection – Choose this option to export information about the model selection process.
Comparison - Choose this option to export data about your Pairwise Comparison Test or Multi-
Group Clustering.

Once you have selected one or more of the above options, you will need to address the following:
Output By-Variables As – When exporting By-Variables, Joinpoint can export the numeric
representation of the By-variable or the By-variable label (either in quotes or not).
Export Models – Joinpoint can export all of the joinpoint models per cohort or it can export only
the final selected joinpoint model.
Missing Characters – Sometimes data values can be missing. Joinpoint offers several export
options for missing values. For SAS users, a period is considered a missing value.
Remove Flags Marking the Joinpoints - Whether to include a column with Joinpoint flags.
Remove Flags Marking the Best Models - Whether to include an asterisk indicating which
model is the best one.
Output All Models in the Same Column - Whether to include a column indicating the model.
Output Variable Names Before Each By-Group - Redisplays the column headers for each new
By-Group.
Load Defaults - This option will load your default settings for this window (your defaults are
automatically loaded when the window is displayed). The Filename and Sections to Export are
not included in the default settings.
Set As Default - This option will save all of your settings in this window (excluding the
Filename and the Sections to Export). This button is only enabled after you make changes to your
settings.

Related content
Graph Tab
Data Tab
Model Estimates Tab
Trends Tab
Model Selection Tab
Comparison Tab

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Print Options
While viewing the Output Results from a joinpoint run, you may want to print out this information. From
the Main Menu, select Output > Print... to display the Print Options window. From this window, you
can configure what information you want to display in your report.
Please note that this option will not be available if the " View Multiple Joinpoint Models" option is
selected.
Options to choose from are:
Cohort Selection - Whether to print the currently displayed cohort, or all of the result cohorts.
Model Selection - Whether to print the currently displayed model, the final selected model, or
all models. All models is not available for Comparison analyses. Only the final selected model is
available when all cohorts are selected.
Sections to Print
Most options are related to which output tab information you want to display from the
results.
Session Parameters - Includes all of the session parameters used to generate the
output results. The parameters will be printed at the beginning of the report.
Graphs - If you select the " Graph" option from the " Sections to Print" , you can choose to output
high resolution graphs.

Clicking the Print button on this window will display a report viewer window. This window will allow you
to view your report before you actually print it. Options from this window are to scroll through the
pages, send the report to a printer, setup the print page, or to save the report to an Excel, PDF, or Word
file.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Output Properties
Select Output > Properties from the main menu to see the Output Properties window. This displays
information on the settings used to create the active output file. Information such as the Joinpoint
version, the Input Data Filename, By Groups, Creation Time, Date Submitted, Output Filename, and more
is displayed.
If you want to re-run the models with different settings, a convenient way is to retrieve the session from
the output file.

Related content
Retrieving a Session

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Viewing Output from the Different Analysis Types
T here are a number of different analysis types you can execute in Joinpoint. Within the
Output window, each is displayed slightly differently.

Standard Joinpoint Regression Analysis

A standard regression analysis is one where no special options are selected and is the most common
type of Joinpoint analysis. Within the Output Window, a standard analysis is displayed in the Cohort
Tree as a list of By-Variable combinations where each resultant cohort contains the model results for
the joinpoint numbers requested by the user.
Pairwise Comparison Analysis
A Pairwise Comparison is performed by selecting the “Pairwise Comparison” option on the Advanced
Analysis Tools tab of the Session window. The results of a Pairwise Comparison analysis will be
displayed in a similar fashion to a Standard Analysis except that an additional By-Variable level
named “Coincident / Not Coincident” or “Parallel / Not Parallel” will be present. This added level will
be above the last By-Variable. Within Coincident or Parallel you can select only one model. However,
under Not Coincident or Not Parallel, you can select one model from each of the lowest level By-
Variables. For either selection, you will get two graphs and two groups of data displayed.
Details on the Pairwise Comparison Analysis may be found here.
Delay / Non-Delay Data Analysis
A Delay / Non-Delay analysis is performed when “Delay & Non-Delay Data” is selected for the Run
Type on the Input File tab of the Session window. In addition to the normal By-Variables, a new
lowest level By-Variable is created for Delay and Non-Delay. You can select one model from Delay,
and one model from Non-Delay. Both models will be displayed on the graph at the same time, and in
each of the data grids.
More information on Delay Adjusted Statistics may be found here.
Jump Model / Comparability Ratio / Standard Joinpoint Analysis
A Jump Model / Comparability Ratio analysis is performed when “Jump Model” or “Comparability
Ratio” is selected for the Jump Model / Comparability Ratio on the Advanced Analysis Tools tab of the
Session window. On the Output Window, a new lowest level By-Variable will be created for each of
your selections. For example, if you selected Jump Model and Include Standard Joinpoint Analysis,
each of the joinpoint models would be broken out by Jump Model and Standard Joinpoint Model. You
can select one model from each different type to display.
Details on the Jump Model / Comparability Ratio Analysis may be found here.
Delay / Non-Delay with Jump Model / Comparability Ratio / Standard Joinpoint Analysis
The Jump Model / Comparability Ratio / Standard Joinpoint Analysis options can be used with Delay /
Non-Delay data. When this occurs, each Delay / Non-Delay cohort data set will have results for each
of the selected analysis types (Jump Model, Comparability Ratio, and Standard Joinpoint Analysis).
The cohort tree will reflect this.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Viewing Multiple Joinpoint Models
As of Version 4.4.0.0, users can select to consolidate the results of multiple Joinpoint
models. T his option is particularly useful for viewing multiple cohort model results on one
graph.

For most analysis types produced by Joinpoint, the Cohort Tree will provide this option (at the bottom of
the tree control). The View Multiple Joinpoint Models enables a different display mode in the Output
Window. This mode allows you to select as many models as you want to display on the graph and
associated output tabs. When this option is selected, the following rules apply:

1. When viewing multiple joinpoint models from the same cohort, all line segments from the same
model will have the same color. Each model will be assigned a unique color.
2. If you select model results from multiple cohorts, then all model segment lines from the same
cohort will get the same color.
3. You cannot Export or Print when viewing multiple joinpoint models.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Technical Requirements/Help
Citation

Technical Support

System Requirements

Frequently Asked Questions

FAQ' s

Joinpoint Revision History

What are the most recent changes to the Joinpoint program?

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Citation
The joinpoint regression model and permutation tests for identifying changes in trend are described in:

Methods Citation:
Kim HJ, Fay MP, Feuer EJ, Midthune DN. " Permutation tests for joinpoint regression with
applications to cancer rates" Statistics in Medicine 2000; 19:335-351: (correction: 2001;20:655).
The present document describes the use of the Joinpoint Regression Program for Windows.

Software Citation:

Joinpoint Regression Program, Version 4.5.0.1 - June 2017; Statistical Methodology and Applications
Branch, Surveillance Research Program, National Cancer Institute.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Technical Support
For a list of common questions, please check the Frequently Asked Questions (FAQs) list.
For technical support beyond the help system or Web site, please use the On-line Technical Support
Request Form.
Note: if you contact Joinpoint technical support for help with a specific analysis, it helps if you can
attach a copy of your data file and Joinpoint session (.jps) or output (.jpo) file.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

System Requirements
The system requirements to run Joinpoint are a personal computer with at least:
Pentium-based PC
32 MB application RAM
Approximately 10 MBs hard disk space
A 32-bit or 64-bit Microsoft Windows Operating System – Windows 7 or higher
Screen resolution set to 800 by 600 pixels or greater
.NET Framework 4.0 or higher (this will be installed with Joinpoint if necessary)
The use of Excel features (Excel as an input data file, exporting to Excel) requires Office 2010 or
later to be installed

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Frequently Asked Questions
- For more help using Joinpoint check the Surveillance Research Program Web site for Joinpoint:
Frequently Asked Questions.

Citation & References

1. Is there a suggested citation for Joinpoint?

2. Are there any reference papers that can help me understand the methodology?

Software Updates and Installation

3. What are the most recent changes to the Joinpoint program?

4. How can I replicate previous results with a newer version of Joinpoint?

5. Can Joinpoint run on a Macintosh PC?

Model Estimation and Selection Methods

6. Why doesn' t the Joinpoint program give me the best possible fit? I can see other models
with more joinpoints that would fit better. Exactly how does the program decide which
tests to perform and which joinpoint model is the final model?
7. Describe the permutation test used here.

a. How many permuted data sets should I use?

b. What does the p-value mean for Joinpoint?
c. Why are new significance levels being used in Version 3.0 and succeeding releases?

8. What is the Bayesian Information Criterion method for selecting the best model?
9. Why should I use the Permutation Test (PT) when Bayesian Information Criterion (BIC) is
so much faster?
10. Should I use the linear or log-linear model?
11. When running the test for pairwise differences with the Coincident option, should I use
a weighted or unweighted model?
12. Does the test of parallelism of two series require the series to be uncorrelated?
13. For a statistical test of parallelism between two groups, how is the assumed number of
joinpoints (which is displayed on the comparison tab) determined?
14. Can the Joinpoint software estimate the autocorrelation parameter?

15. Why did the default settings for Grid Search change?
16. Why is the Hudson' s method disabled?

Applications of Joinpoint
17. I used the program on SEER data and do not get the same answers as in the SEER Cancer
Statistics Review. Why?
18. When I fit a Joinpoint model to find the rate trend, why doesn' t the APC for one of the segments
match the APC found by SEER*Stat for the corresponding segment?
19. Sometimes, the change in the slopes between two segments is not statistically significant (p-
value>0.05) from the t-test, but there is joinpoint between two segments or vice versa. Why?
20. Sometimes, the APC for one segment is significantly different from zero, but when an extra
joinpoint in the segment is determined by the Joinpoint software, neither APCs for the two
consecutive segments are significant. Why?
21. Can I do Joinpoint regressions where the time intervals are not equally spaced or the time points
are not integer?
22. Can Joinpoint be used to model proportions or percents?
23. How does the Joinpoint software deal with zeros in the dependent variable?
24. Can I use Joinpoint to project or extrapolate into the future?

Using Joinpoint Results

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

 25. How do I import Joinpoint' s graphs into other software such as Word, Excel, or PowerPoint?
26. How do I print Joinpoint results?
27. How do I save Joinpoint results to PDF or Word?
28. How do I export Joinpoint results to a text file or to Excel?
29. How do I export high resolution graphs?

Data Sources
30. When I use Excel as my data source, why do I get " ####" in the data?

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Citations
Is there a suggested citation for Joinpoint?

Answer:

Both a methods and software citation can be found in the Citation section here.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

References
Are there any reference papers that can help me understand the methodology?

Answer:

Reference papers are listed here.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Joinpoint Revision History
What are the most recent changes to the Joinpoint program?

Answer:

An outline of the most recent changes to the Joinpoint program can be found here.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Replicating Previous Results
How can I replicate previous results with a newer version of Joinpoint?

Answer:

Because of changes in statistical methodology, computational issues, and the random number generator
from one version to the next, the same version of the program should be used again to exactly replicate
results. If you would like prior versions of Joinpoint to replicate previous results, please request them by
contacting Joinpoint Technical Support.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Mac Installation
Can Joinpoint run on a Macintosh PC?

Answer:

Joinpoint was designed to run under a Microsoft Windows operating system. Some users have reported
successfully running Joinpoint on a Macintosh using a windows emulator, but we do not support that
operating system.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Selecting the Final Model
Why doesn't the joinpoint program give me the best possible fit? I can see other models with
more joinpoints that would fit better. Exactly how does the program decide which tests to
perform and which joinpoint model is the final model?

Answer:
As with many statistical models, if you add more parameters you get a better fit. The same is true with
joinpoint models. What the program does is to try to choose the smallest number of joinpoints such that
if one more joinpoint is added, the improvement is not statistically significant. Thus, in the final model
you may interpret each of the joinpoints and its corresponding changes in trend as significant.
Joinpoint selects the final model using two different methods: Permutation Test and Bayesian
Information Criterion (BIC). First, the user specifies MIN as the minimum number of joinpoints and MAX
as the maximum number of joinpoints on the Method and Parameters tab.
Then the program uses a sequence of permutation tests to select the final model. Each one of the
permutation tests performs a test of the null hypothesis H 0: number of joinpoints = ka against the
alternative H a: number of joinpoints=kb where K a < K b. The procedure begins with ka = MIN and kb =
MAX. If the null is rejected, then increase ka by 1; otherwise, decrease kb by 1. The procedure continues
until ka = kb and the final value of is the selected number of joinpoints.
The second method is based on the Bayesian Information Criterion (BIC). The value of BIC is the
loglikelihood value penalized by the cost of extra parameters. The model with the minimum value of BIC
is selected as the optimal model.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Permutation Test Details
Describe the permutation test used here.

Answer:

In this program, the permutation test is used repeatedly for testing between two different joinpoint
models, a simpler model with fewer joinpoints called the null model and a more complicated model
called the alternative model. The alternative model fits better because it is more complicated.
The question for the test is: does it fit much better than would be expected by chance. To test this
statistically, we calculate a ratio, T=SSEN/SSEA, where SSEN is the sum of squared errors (SSE) from the
null model and SSEA is the SSE from the alternative model. Values of the ratio (T) close to 1 mean that
the alternative is not much better than the null model, while larger values mean that the alternative is
much better.
In order to decide how much larger a ratio needs to be to be statistically significant, we use the
permutation method, since it is not possible to drive this analytically. In this method, we randomly
permute (that is, shuffle) the errors (also called the residuals) from the null model and add them back
onto the modeled values from the null model to create a permutation data set. Then we calculate the
ratio T for the permutation data set and measure how much evidence the data provide against the null
hypothesis by estimating the proportion of the permutation data sets whose T values are at least as
extreme as the one we observed with the original data set. This proportion is called the p-value in
statistical testing.
If the true model was the null model (i.e. the truth being the simpler model), we would expect that on
average, about half of the T ratios computed from the permutation data sets would be greater than the
one derived from the original data (i.e. the distribution of the p-value under the null is uniform over the
interval from 0 to 1).
If the true model was the alternative model, we would expect that after permuting the errors most of the
new T ratios would be less than the original T ratio and thus the p-value would be small. In other words,
the permuted data set would look less like the alternative model than the original data.
So we reject the null model (or null hypothesis) if less than a certain proportion of the T ratios are
greater than or equal to the original T ratio. That is, we reject the null hypothesis if the p-value is small.
For more specific details, see Kim HJ, Fay MP, Feuer EJ, Midthune DN. Permutation Tests for Joinpoint
Regression with Applications to Cancer Rates. Stat Med 2000;19:335-351. To request a reprint, email
Mr. Reggie Taborn for a copy at: [email protected]

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Number of Permuted Data Sets
How many permuted data sets should I use?

Answer:

For greater consistency in the p-values obtained if one were to change the seed for each run, we suggest
running the program for at least 4499 permutations. For this reason, the default number of permutations
is now 4499 in the current version of Joinpoint (it was 999 in previous versions).
If the maximum number of possible joinpoints is set to 3 and the minimum is set to 0, then there are 3
statistical tests each conducted at the Bonferonni adjusted cutoff significance level of .05/3 = .0167.
The value 4499 was chosen so that if you obtained a p-value of .0167 using one seed with 4499
permutations, then, assuming the number of possible permutations is large, the complete run using all
possible permutations would have approximately a 99% chance of the p-value being between .0120 and
.0220 (length of confidence interval = .0100), and approximately a 95% chance of the p-value being
between .0129 and .0206 (length of confidence interval = .0077). Choice of the number of permutations
selected by the user is a tradeoff between computer time and consistency of the p-values obtained.
The Joinpoint program uses Monte Carlo simulation to calculate p-values for a series of permutation
tests. See Permutation Test Details for details on the permutation tests performed.
Here we discuss the implications of the choice of the number of permutation data sets (N). The program
runs faster with smaller values of N, but it gives better precision for the p-value with larger values of N.
In addition, a larger N reduces the probability that another analysis of the same data might get a
different answer when run with different random number generator seeds. (Computer programs produce
pseudo-random numbers through algorithms that mimic randomness, which we use to shuffle or permute
the errors. The algorithms use a seed or seeds to start the algorithm. These seeds can be used to
produce repeatable pseudo-random numbers.)
The problem of two analyses obtaining different answers from the same data is addressed by this
program by specifying default random number generator seeds. Thus, as long as no parameters are
changed (including the random number generator seed and N), repeats of the analyses will produce the
same results. Otherwise, two runs of the same analysis except with different seeds could get different
answers.
To get an idea how results would change for someone using different random number generator seeds,
we list some confidence intervals for p-values below.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017
p-values
What does the p-value mean for Joinpoint?

Answer:

The Joinpoint Regression Program performs a series of hypothesis tests that test the null hypothesis of
ka joinpoints against the alternative hypothesis of kb joinpoints, where ka and kb change for each
hypothesis test (ka< kb). Each p-value corresponds to this type of test. The p-value is an estimate of the
probability of observing data at least as extreme as the data that we have, in fact, observed under the
assumption that the null hypothesis (i.e. smaller number of joinpoints) is true.
As the permutation test is a randomization test, it depends on the random number generator. For greater
consistency in the permutation test p-values obtained if one were to change the seed for each run, we
suggest running the program for at least 4499 permutations. For this reason, the default number of
permutations is now 4499 in the current version of the Joinpoint Regression Program. Choice of the
number of permutations selected by the user is a tradeoff between computer time and consistency of the
p-values obtained.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Joinpoint Revision History
What are the most recent changes to the Joinpoint program?

Version 4.5.0.1 (released June 12, 2017)

Bug Fixes
As of version 4.5.0.0, users can set the Alpha levels for the Joinpoint Locations, APCs, AAPCs,
and Jump Value and Comparability Ratio. Upon installation of that version, the default values for
each were set incorrectly to 0.5. The default values should have been 0.05. Installing the
new version of Joinpoint (4.5.0.1) will change any 0.5 Alpha setting to 0.05.
The significance level footnote on the Model Selection tab referenced the wrong alpha level.
This has been corrected.

New Features
The value 100 has been added to the list of possible “Rates per” (population at risk) when
computing Crude or Age-Adjusted Rates.

Version 4.5.0.0 (released May 17, 2017)

New Features
Two tabs on the Session Window have been renamed. “Advanced” has become “Method and
Parameters”. “Comparison” is now “Advanced Analysis Tools”. Several controls on these tabs have
been moved. More details are below.
The Grid Search now has specific parameters for controlling Joinpoint placement. The default values
for Grid Search’s “Minimum number of observations from a joinpoint to either end of the data” and
“Minimum number of observations between two joinpoints” have been changed. For more details on
this change, please go to the Help section.
Hudson’s Method has been disabled.
The Pairwise Comparison has two new parameters: Significance Level and Max Number of Randomly
Permuted Data Sets.
Alpha Values for APCs, AAPCs, and Jump Value and Comparability Ratio have been added to the
Session Preferences. Session Preferences can now be defaulted to different values.
For Jump Model and Comparability Ratio Model analyses, the “CR Estimate” is now displayed in the
Graph Legend on the Output Window. Added CR Estimate precision setting to the Display Options
window.
An APC/Slope precision setting has been added to the Display Options window.
Jump Model and Comparability Ratio display and precision options have been added to the Display
Options window.
The layout of the Model Selection and Comparison tab settings in the Display Options window has
been updated.
Added the Alpha Level to the footer of the Model Selection tab on the Output Window.
Bug Fixes
If the Jump Model option was run, the Comparability Ratio table (located on the Model Estimates
output tab) contained an incorrect statistic. For the Jump Model, the column labeled “Variance of CR
(a/b)” contained the standard error of the comparability ratio value produced from the Jump Model,
not the variance. The variance column has been replaced by a column containing the standard error
of the comparability ratio.
The Comparability Ratio table, located on the Model Estimates tab when the Jump Model or
Comparability Ratio Model are run, has been corrected with the following additional adjustments:
The Comparability Ratio Model now produces a Lower and Upper CI, Test Statistic and p-value.
These are computed using a normal (z) distribution.
Several footnotes have been added to the table to better describe the statistics displayed.
If all values of a by-variable weren’t present in the first grouping for the by-variable, Joinpoint
wouldn’t execute due to encountering an unknown value for the by-variable.
The fields for the Comparability Ratio model export were in the wrong position.
0.5 is added to zero counts only under the following conditions:
The Run Type is “Provided in Data File”, the Type of Variable is “Count”, and the
Heteroscedastic Errors Option is “Poisson Variance”.
The Run Type is “Provided in Data File”, the Type of Variable is “Count”, the Heteroscedastic
Errors Option is “Constant Variance”, and the Log Transformation is “YES”.
The Run Type is “Calculated from Data File”, the Type of Variable is “Crude Rate”, and the
Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017
 Heteroscedastic Errors Option is “Poisson Variance”.
The Run Type is “Calculated from Data File”, the Type of Variable is “Crude Rate”, the
Heteroscedastic Errors Option is “Constant Variance”, and the Log Transformation is “YES”.
A message is displayed in the Output Window for each cohort that had 0.5 added to one or more zero
counts.
A warning message is now displayed when the number of permutations for the Permutation Test is
set below 4499. The minimum number of permutations allowed is now 1000.
Fixed a problem where APC and AAPC exports were adding an extra field delimiter at the end of each
row.
Corrected the display of the Dependent Value column headers on the Data Tab of the Output
Window.
Added restriction that you can only use Poisson Variance if using Provided Count or Calculating
Crude Rate.
The Maximum number of joinpoints was not being validated on the correct Grid Search settings.
Fixed a problem where grids on the Output Window were not always left aligned.
Fixed a crash in the Output Window when the Non-Delay cohort of a Delay \ Non-Delay session was a
bad cohort.
The Comparability Ratio grid on the Output Window will display an explanation footer when an
incomputable value is displayed.
For Jump and Comparability Ratio runs:
The Last Observation of Old Coding must be at least 4 or more data points from either end of the
data.
If you use Constant Variance, the Variance of Comparability Ratio will be set to N/A.
You cannot set Log Transformation to No.

Version 4.4.0.0 (released January 4, 2017)

New Features:
Session Options:
Jump Model and Comparability Ratio calculation options have been added, which allows
analysts to estimate trends even when there is a coding change in the data series. For
further information on these new options please refer to the help manual.
The ability to analyze only select cohorts from the input data.
Joinpoint can now analyze data files containing cohorts with an unequal number of
observations.
Output Viewer Options:
The output viewer window now contains a tree view for selecting cohorts.
Multiple cohort models can be viewed on a single graph. All other output tab information
is also updated to contain all of the selected cohort models.
AAPC ranges can be specified using the output menu.
Joinpoint can now export high resolution graphs.
By Variables now allow text values.
Joinpoint now has an on-line tech support request form.

Bug Fixes:
Logarithmic scale display on the Output Window graph sometimes displayed incorrect Y axis
values.
In rare occurrences, the graph axis positions in the Output Window did not show all observed
data values.
JPO files over 1 megabyte would sometimes cause loading errors.
Dictionaries with a variable format of “labels” would show as blank in the Format Editor.
Some session properties were not being displayed in the Output Properties window.
Joinpoint will not let users add or remove variables when a SEER*Stat Dictionary file is used.

Version 4.3.1.0 (released April 19, 2016)

Bug Fixes:
Several library files were missing from the 4.3.0.0 installation. Those library files prohibited
some users from reading or exporting to Excel and generating reports.
A bug in the By-Variable format generator was fixed.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

New Features:
The Joinpoint session now has an input data tab. The new tab combines the Data Import Wizard
and the session Specifications Tab.
AAPC ranges can now be set on the Advanced Tab.
Joinpoint now links to an on-line help system. If you do not have internet connectivity, a PDF
version of the help system is installed with Joinpoint.
Matrix export was expanded to export to Excel.
Joinpoint now has 3 sample sessions. These sessions can be loaded from the menu File->Sample
Joinpoint Sessions. These new sample sessions have replaced the “Blank Session” option.

Bug Fixes:
Various bugs were fixed in the matrix export.

Version 4.2.0.2 (released June 23, 2015)

The data import wizard was updated to allow non-integer values.

When computing age-adjusted rates, Joinpoint will now skip cohorts when the total population
for the cohort is zero.

Version 4.2.0.1 (released May 6, 2015)

The data import wizard was updated to correctly handle files containing the information needed for
Joinpoint to compute age-adjusted rates.

Version 4.2.0.0 (released April 29, 2015)

New Features:
Joinpoint now provides a new method for computing AAPC Confidence Intervals. The new method
is called the Empirical Quantile method. The previous AAPC CI method is still available and is
called the Parametric method.
When performing the AAPC comparison and using the parametric method to calculate the AAPC
Confidence Intervals, a new method has been implemented to calculate the confidence interval
for the difference between the two true AAPCs. The AAPC comparison cannot be performed when
using the Empirical Quantile method to calculate the AAPC Confidence Intervals.
Joinpoint can now process Delay and Non-Delay Adjusted statistics output by SEER*Stat. For
SEER*Stat exports containing both delay and non-delay statistics, Joinpoint can graph both
together for each cohort.
When computing crude or age-adjusted rates, Joinpoint now provides an option for users to
display their rates per a certain population (e.g. 100,000 population at risk).

Bug Fixes:
There was a problem with the handling of a specific data file via the import wizard that was
identified and resolved.
A bug that prevented users from defining their independent variable values was resolved.

Version 4.1.1.5 (released February 26, 2015)

A bug preventing the use of CSV files was fixed

Version 4.1.1.4 (released February 19, 2015)

The p-value for a two-sided test that the true APC is calculated based on a t distribution.

Version 4.1.1.3 (released December 29, 2014)

A bug was fixed involving the text file import wizard.

Version 4.1.1.2 (released December 15, 2014)

A small bug was fixed to prevent Joinpoint from terminating its calculations under certain
Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017
circumstances.

Version 4.1.1.1 (released October 7, 2014)

A small bug was fixed to prevent Joinpoint from terminating its calculations under certain
circumstances.

Version 4.1.1 (released August 5, 2014)

Bug fixes:
In pairwise comparison runs, all values in the AAPC comparison table were being flagged as
statistically significant, whether or not that was actually the case. This problem has been
resolved and the program now flags only those that are significant.
The X and Y axis labels in the output graphs were at times not being displayed. This problem has
been resolved.

Version 4.1.0 (released April 25, 2014)

Enhancements:
Joinpoint now has the ability to process multiple cohorts simultaneously (threading). Joinpoint
will inspect the PC it is running on and determine the number of cohorts that can be processed
concurrently. Many times this is equal to the number of CPU cores on the PC. Processing multiple
cohorts will decrease the time it takes Joinpoint to run a session. Please note that this feature
will not work when executing Joinpoint under Windows XP. Windows XP users will only be able to
process one cohort at a time (similar to previous versions of Joinpoint).
Users can adjust the maximum number of threads (cohorts to process concurrently) that Joinpoint
will attempt to use when executing a session. Joinpoint will automatically set the number of
threads to use to the maximum possible for the PC it is running on. Users can then lower the
number of threads if they want. This option is provided via the File->Preferences menu.
A new session execution progress window has been created. The window will now show the
progress of each cohort that it is currently processing. When multiple cohorts are being
processed concurrently, a progress bar for each will appear in the progress window.
The APC and AAPC tables on the Trends tab in the output have two additional statistics: Test
Statistics and P-Value.
The slope can now be displayed in the data table when the Log Transformation is set to NO. The
report printing was also updated to reflect this change.

Bug Fixes:
The AAPC Confidence intervals will only be computed if the variance of each associated APC is
available.
The graphing of X-axis values where the range between the minimum and maximum value was
less than one was updated to display 10 axis points.
The correlation and covariance tables on the Model Estimates tab in the Joinpoint output were
computed only if unconstrained estimates with offending data points deleted is available.

Version 4.0.4 (released May 6, 2013)

The program was recompiled so that it would be compatible with Windows XP.

Version 4.0.3 (released April 29, 2013)

Enhancements:
If standard errors are used in the calculations, then they will be displayed on the Data tab
(whether they are provided or computed). Users have the ability to change the display precision
via the output options.
The APC footnote character was changed from * to ^.
The By Variable “Delete…” button was renamed to “Remove…”
When using a SEER*Stat dictionary file (*.DIC), any independent variable value with a dash (“-“)
located in its format label will not be used in the analysis. The label will be updated to contain
the following text: (not used in calculations). This functionality was added so that users would
not have to reproduce their SEER*Stat analyses in order to remove totals or sub-totals in their
independent variable.
Graph Display Option enhancements:

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

 The legend can be moved to the Right, Left, Top, or Bottom of the graph.
X and Y axis label precision can be changed.
The locations of each Joinpoint can be displayed on the graph (e.g. “Joinpoint 1: 1981”)
The legend can be adjusted to display the APC ranges.
The modeled line for each cohort can be set to a single color (e.g. Blue). In comparison
runs, the second cohort modeled line will be Green.
Added a Print button to the main Joinpoint toolbar.
File associations have been enabled for Joinpoint. Joinpoint will automatically open when either
a Joinpoint session (*.JPS) or output matrix (*.JPO) file name is double-clicked.
Users can now drag and drop files onto the Joinpoint program and it is a file type Joinpoint
recognizes it will open it up. Session (*.JPS) and Output (*.JPO) files will automatically open. If a
.TXT file is dropped onto Joinpoint, this will trigger the Data Import Wizard. If a SEER*Stat DIC
file is dropped, this will trigger the session open process.
In the output window, depending on which Model Selection Method was specified by the user, a
tab with one of the following three labels was shown: “Permutation Tests”, “BIC”, or “Modified
BIC”. Those three labels have been replaced by the following label: “Model Selection”. This
modification was made to standardize the labeling in the Joinpoint application.
A standard way of labeling the best, selected model was implanted into the Joinpoint
application. The label “Final Selected Model” will now appear instead of labels such as ‘Best Fit”
and “Best Model”. This modification was made to standardize the labeling in the Joinpoint
application.
Bug Fixes:
The Trends tab on the output form will not be displayed when the Log Transformation option is
set to No. This is because no APCs or AAPCs are produced when this option is selected. Export
and Output Options have also been similarly updated.
When having Joinpoint compute Crude Rates while using the Poisson Variance Heteroscedastic
Errors Option, there was the potential to have Joinpoint read in the wrong data column for the
counts. This has been resolved.
The covariance and correlation matrices for pairwise comparisons were not always refreshing
themselves correctly when turning them on and off.

Version 4.0.1 (released January 9, 2013)

Fixed a bug dealing with the decimal symbol for international users.

Version 4.0 (released December 17, 2012)

Joinpoint can now calculate crude rates, age-adjusted rates, proportions, and percentages if
their components are provided in the input data file. Associated standard errors can also be
computed. Please see the Dependent Variable section for more details.
The following modifications were made to the Joinpoint Session:
The new Comparison tab was created. The tab contains the parameters related to
Pairwise comparisons and Clustering. Clustering is a future enhancement.
The “Test for pairwise differences between by-groups” control (originally on the
Specifications tab) was moved to the Comparison tab and renamed “Pairwise
Comparison”. In order to produce Comparison analyses, users must also use the
“Comparison Type” control on the Comparison tab.
The Specifications tab was reorganized. Some items have been moved to other parts of
the screen and have more descriptive labels (i.e. “Model” was moved from the upper left
to the lower right and is labeled “Log Transformation”).
The Dependent Variable section (on the Specifications tab) was expanded to
accommodate Joinpoint’s new ability to compute various statistics (as previously
mentioned). Users can now specify whether the Dependent variable will be calculated or
provided and what specific type of variable it is (Count, Crude Rate, Age-Adjusted Rate,
Proportion, Percentage, or Other).
The Heteroscedastic Errors Option was reorganized. The two Poisson options in the
previous version of Joinpoint are now combined into one. “Poisson Variance using Crude
Rate” is now specified by selecting Dependent Variable=Calculated, Type=Crude Rate,
and Heteroscedastic Errors Option=Poisson Variance. “Poisson Variance using Count” is
now specified by selecting Dependent Variable=Provided, Type=Count, and
Heteroscedastic Errors Option=Poisson Variance.
A new report viewer has been implemented. Reports can be printed or exported to PDF, Word, or
Excel.
The Autocorrelation Parameter is now displayed on the Model Estimates tab of the Output
window when “Fit an automated errors model based on the data” is selected for the
Autocorrelated Errors Options.
Errors found in the input data file are now reported by cohort. When a session is executed, if any
data errors are found a special window is displayed with the errors found for each cohort.
Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017
Joinpoint is now compiled under Microsoft’s Visual Studio. The installation of Joinpoint includes
Microsoft’s .NET framework 4.0.
All user-specific options are now saved in a parameter file instead of the Window’s registry.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

New Significance Levels
Why are new significance levels being used in Version 3.0 and succeeding releases?

Answer:

The Joinpoint software uses a series of permutation tests to determine the number of joinpoints. Prior to
Version 3.0, the software used the Bonferroni adjustment to control the error probability of each of the
multiple tests. The Bonferroni adjustment has been shown to be conservative and the procedure tends to
select fewer joinpoints than it should. The procedure with new significance levels controls the over-
fitting probability and it is superior to the traditional Bonferroni adjustment.
Details: The Bonferroni adjustment is conservative because the actual overall significance level is
usually less than the nominal level α. Starting with Version 3.0, the new adjustment procedure controls
the overall over-fitting error probabilities, P(k >ka | k = ka), ka = MIN,...,MAX-1 , under α.
Let k denote the number of joinpoints and α(ka; kb) be the significance level of each individual test H 0:k
= ka vs. H a:k = kb.
The new procedure set α(ka; kb) = α/(MAX - ka).
Notice that the individual significance level depends on the number of joinpoints ka under the null.
Consider an example where MIN = 0 and MAX= 4. The new procedure has the following properties:

P(k > 0 | k = 0) ≤ α(0,4) + α(0,3) + α(0,2) + α(0,1);

P(k > 1 | k = 1) ≤ α(1,4) + α(1,3) + α(1,2);

P(k > 2 | k = 2) ≤ α(2,4) + α(2,3);

P(k > 3 | k = 3) ≤ α(3,4).

If we like to bound these over-fitting probabilities by α, then we can assign different values for each
α(ka; kb) . That means, we can achieve a better power by setting

α(0,4) = α(0,3) = α(0,2) = α(0,1) = α / 4;

α(1,4) = α(1,3) = α(1,2) = α / 3;

α(2,4) = α(2,3) = α / 2;

α(3,4) = α

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Bayesian Information Criterion (BIC) Details
What is the Bayesian Information Criterion method for selecting the best model?

Answer:

Permutation test (PT) and BIC are two different approaches to selecting the optimal number of
joinpoints. The Permutation test approach uses a sequence of permutation tests to determine the best
number of joinpoints. The PT approach controls the error probability of selecting the wrong model at a
certain level (i.e. 0.05), whereas the BIC approach finds the model with the best fit by penalizing the
cost of extra parameters. The models picked by BIC tend to fit the data well but are less parsimonious.
The applications have shown that the PT approach worked well for cancer incidence and mortality data.
The equation for computing the BIC for a k-joinpoint model is:

BIC(k) = ln{SSE(k)/#Obs} + {#Parm(k) /#Obs} * ln(#Obs),

where SSE(k) is the sum of squared errors of the k-joinpoint regression model, #Parm(k)=2*(k+1) is the
number of parameters of the k-joinpoint model and #Obs is the number of observations.
The k-joinpoint model with the minimum value of BIC(k) is selected as the final model.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Permutation Test vs. Bayesian Information
Criterion
Why should I use the Permutation T est (PT ) when Bayesian Information Criterion (BIC) is so
much faster?

Answer:

PT and BIC are two different approaches to selecting the optimal number of joinpoints.
The Permutation test approach uses a sequence of permutation tests to determine the best number of
joinpoints. The PT approach controls the error probability of selecting the wrong model at a certain level
(i.e. 0.05), whereas the BIC approach finds the model with the best fit by penalizing the cost of extra
parameters.
The models picked by BIC tend to fit the data well but are less parsimonious. The applications have
shown that the PT approach worked well for cancer incidence and mortality data.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Linear or Log-linear Model
Should I use the linear or log-linear model?

Answer:

The linear or log-linear model can be chosen depending on how linear the observed rates or the
logarithm of the observed rates are over time. In order to check the goodness of the chosen model, a
user can test for normality of the residuals obtained under the linear or the log-linear fit. Select a model
whose residual analysis indicates a better fit, regarding the model assumptions of normality, linearity,
equal variance, and independence. One reason for using a log transformation for cancer rates is that
they arise from a Poisson distribution which is skewed especially when the cancer is rare or the rates
come from a small population. The log transformation is a standard way to make this skewed distribution
approximate a normal distribution. Rates for common cancers or which come from a large population can
be approximated as arising from a normal distribution without a transformation.
One motivation for using the log-linear model for cancer rates regardless if they are rare or not is the
ease of interpretation. Under a log-linear model the rates change at a constant percent per year (i.e. a
fixed annual percent change - APC), while for a linear model the rates change at a constant fixed amount
per year. When comparing trends across age groups or across cancer sites where the rates are very
different, the advantage of a log-linear model is that the APC is a metric which makes sense to compare
across widely different scales. For example, a rare cancer and a common cancer may change at the same
annual percent per year, but it is highly unlikely that they would change at the same fixed amount per
year (e.g. if the rates were declining, the rare cancer rate would quickly become negative!).

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Pairwise Differences with Coincident Option
When running the test for pairwise differences with the Coincident option, should I use a
weighted or unweighted model?

Answer:

If the weights for the two cohorts are fairly different, the combined model will be heavily influenced by
the cohort with the larger weight, and the results may not be as expected since the combined fit will
closely mimic the level and the number and location of joinpoints for that individual cohort. In this case
the statistical algorithm is appropriately weighting the series that is more reliable. In some cases,
however, it may be more appropriate to weight each series equally, even though one series has a larger
variance. For example, it may be appropriate to weight two racial/ethnic groups equally if the goal is to
find the best fit for two groups ignoring the fact that one series is more reliable than the other. In such a
case, running the unweighted model using the Heteroscedastic Errors Option of " Constant Variance"
may be more appropriate. These same considerations are less obvious but can be relevant when fitting a
parallel model. In this case, level is not an issue (since each cohort has its own level), but the fit of the
number and location of joinpoints will be heavily influenced by the larger cohort, unless an unweighted
analysis is used. There is no " correct" answer, but careful consideration should be given to the overall
purpose of the analysis in these situations.
For more details, see Pairwise Comparison in the Joinpoint help system.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Correlation In Two Series
Does the test of parallelism of two series require the series to be uncorrelated?

Answer:

The permutation test for parallelism assumes exchangeability in the pair of residuals at each time
period. When errors of the two series are correlated, this assumption is violated, so the permutation p-
value will not be accurate. Errors in series are correlated, for example, when the two series come from
the same population or sub-population measured from the same survey (e.g. obesity and diabetes for
white males from the same survey) or the same cancer registry (e.g. white male colorectal and prostate
cancer rates from the same registry). Errors in series may be considered uncorrelated when the
observations in the two series come from different samples of the same population (e.g. obesity for
white males measured from one national survey and diabetes for white males measured from a different
national survey).
One way to test for parallelism in two series with correlated errors is to compute a new time series, the
difference between the original two series. The series are parallel if the difference time series has a
constant mean. One way this can be checked is by fitting a simple linear regression and testing whether
the slope coefficient is zero.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Parallel Pairwise Comparison and Number of
Joinpoints
For a statistical test of parallelism between two groups, how is the assumed number of
joinpoints (which is displayed on the comparison tab) determined?

Answer:
When conducting the test of parallelism between two groups, the true number of joinpoints for each
group under the null hypothesis (H 0) of parallelism and alternative hypothesis (H A) lack of parallelism
are unknown. As described in Kim, et al. (Biometrics, 2004) the test of parallelism is conducted under
K max joinpoints, where K max is greater than or equal to the number joinpoints for both groups under both
H 0 and H A. Simulations presented in Kim et al. have shown that a reasonable choice for K max is the
maximum of the estimated number of joinpoints for each group fit separately, and for the two groups fit
together under the assumption of parallelism. For example, if the Joinpoint software selects 2
joinpoints for males and 0 joinpoints for females fit separately, and 1 joinpoint for males and females fit
under an assumption of parallelism, then the parallelism test is conducted under an assumption of
Max(2,0,1) = 2 joinpoints.
Kim, H-J, Fay, M.P., Yu, Binbing, Barrett, M.J., and Feuer, E.J. (2004) " Comparability of segmented line
regression models" , Biometrics 60, 1005-1014

Example:

T he Comparison tab shows "2 Joinpoints", when the cohort was deemed "parallel" and the
parallel best fit was 1 joinpoint.
For the Pairwise comparison, you have the following three best-fit models:
Male – Joinpoint selects 2 Joinpoints.
Female – Joinpoint selects 0 Joinpoints.
Combined Male and Female – Joinpoint selects 1 Joinpoint.

In the Pairwise Comparison help, there is a section labeled “How is the Kmax chosen?” That section
describes the process of how Joinpoint determines the “2 Joinpoints” on the Comparison tab. This
selection is based on the 3 models above. The number shown on the comparison tab (“2 Joinpoints”) is
the maximum number of Joinpoints in the above 3 models. So, it is the max of 2, 0, and 1.
The test for parallelism is performed AFTER the three models are fit. So, the number of joinpoints
displayed on the comparison tab is not necessarily the number of joinpoints selected for the model fit for
the combined data. Below are the screen shots of the three models and the test for parallelism.

Male – Joinpoint selects 2 Joinpoints, Female – Joinpoint selects 0 Joinpoints:

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Combined Male and Female – Joinpoint selects 1 Joinpoint:

T est for Parallelism:

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Estimating the Autocorrelation Parameter
Can the Joinpoint software estimate the autocorrelation parameter?

Answer:

Yes, starting with Version 3.5, the software can estimate the autocorrelation parameter.
If you select " Fit an autocorrelated errors model based on the data" , the autocorrelation parameter will
be estimated separately for each by-group using the method described in Section 2.3 of Kim et al.
(2000). Under this option, the autocorrelation parameter is estimated for the model with the default
maximum number of joinpoints or the maximum number of joinpoints set by a user.
Although the autocorrelation may be estimated from the data, correcting for autocorrelation with this
estimate may seriously reduce the power to detect joinpoints (see Section 3 of Kim et al. (2000)). We
found in our simulations in Table IV of that paper that adjusting for autocorrelation was helpful in
maintaining proper size of the tests of joinpoints when there was large autocorrelation. We also found
that if there was no autocorrelation then the adjustment seriously affected the power of the test to
detect joinpoints. For example we see in Table IV (b) with φ = 0, the power goes from 90% to 68%. This
is because it is difficult to differentiate between autocorrelation and joinpoints in a model.
If you suspect that your data are positively autocorrelated, we suggest using the " Fit an autocorrelated
errors model with parameter =" option to see how sensitive your results are to changes in
autocorrelation. The option should be used as follows:

1. Fit the model with the uncorrelated errors option.

2. If the user suspects that there is positive autocorrelation in the data, then repeat the analysis
trying several values of the autocorrelation parameter, say for example 0.1, 0.2, and 0.3. If the
results are very similar with different values of the autocorrelation parameter, then the user knows
their results will still hold if there is autocorrelation present. If the results change as the
autocorrelation parameter changes, then the user may end up presenting the series of results, to
show how the results depend on different assumptions about the autocorrelation.

If you suspect negative autocorrelation, the uncorrelated errors model will suffice (see Kim et al., 2000).

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

New default settings for Grid Search
Why did the default settings for Grid Search change?

Answer
In versions prior to 4.5.0.0, the default settings for Grid Search were:
Minimum number of observations from a Joinpoint to either end of the data (including the first or
last Joinpoint if it falls on an observation): 3
Minimum number of observations between two joinpoints (including any joinpoint that falls on an
observation): 4
Number of points to place between adjacent observed x values in the grid search: 0

Starting in version 4.5.0.0, the default settings for Grid Search are now:
Min. number of observations from a Joinpoint to either end of the data (excluding first or last
joinpoint if it falls on an observation): 2
Min. number of observations between two joinpoints (excluding any joinpoint if it falls on an
observation): 2
Number of points to place between adjacent observed x values in the grid search: 0

As you can see, the description for these parameters has changed slightly. These new default settings
are identical in results to what you used to get with 3, 4, 0. We made these value and text changes to
more accurately describe what was going on in the Grid Search algorithm. For example, if you used to
enter 4, 5, 0, then the new values will be 3, 3, 0. If you load an old session with the previous values,
they will automatically be updated to the corrected values.
We have also separated these settings from those of Hudson’s Method to make it clear that they function
differently for each method.

Related content

Grid Search Method - Details

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Hudson's Method Disabled
Why is the Hudson's Method disabled?

Answer
In version 4.5.0.0, Hudson’s Algorithm (used to estimate the exact best fitting location of joinpoints in
continuous time) was temporarily disabled. It was determined that the algorithms for setting how close
joinpoints can be to each other and to the ends of the series, were not working properly when using the
Hudson’s Algorithm. When examining this issue in detail, it was determined that there were some
complicated interactions between these settings and the use of the Hudson’s Algorithm. We were not
fully able to work out these issues in time for this release. Since version 4.5.0.0 had some important
other updates, we decided to temporarily disable Hudson’s Algorithm. We apologize for any
inconvenience this may cause.
But what about sessions that were created in the past with Hudson’s Algorithm? Were they wrong? The
answer is no. The parameters supplied to Hudson’s Algorithm were just not being used correctly.
Hudson’s was incorrectly bounding the location of potential Joinpoints more than should have been done
with the given parameters. Hudson’s should have considered a wider area for potential Joinpoint
placement. The use of the same parameters for both Grid Search and Hudson’s Method in older versions
of Joinpoint led to Hudson’s being forced to use parameters that weren’t specific for its own use.

Related content

Hudson' s Method – Details

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Discrepancies with SEER Cancer Statistics
Review (CSR)
I used the program on SEER data and do not get the same answers as in the SEER Cancer
Statistics Review. Why?

Answer:

The SEER Cancer Statistics Review (CSR) uses the same Joinpoint Regression Program. Their analysis
uses the " Input Standard Error of Response" setting for the Heteroscedastic Errors Option using the
standard error of the rate as calculated by SEER*Stat. Using these options and the current default
number of permutations (4499), if the Joinpoint Regression Program chooses the model with 0
joinpoints, the annual percentage rate change will agree with the calculation of this value given in the
CSR.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Joinpoint APCs not matching SEER*Stat APCs
When I fit a Joinpoint model to find the rate trend, why doesn't the APC for one of the
segments match the APC found by SEER*Stat for the corresponding segment?

Answer:

The Joinpoint model assumes that the trend is continuous at the joinpoint, whereas the APC for the
corresponding segment calculated by SEER*Stat does not impose the continuity constraint with the
consecutive segment. Hence, when there are one or more joinpoints, the APCs from Joinpoint and from
SEER*Stat are different, but they should be close to each other. When there are zero joinpoints, the
APCs from Joinpoint and SEER*Stat will be the same, except for differences due to the precision level of
the data.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Non-Significant Change in Slopes
Sometimes, the change in the slopes between two segments is not statistically significant (p-
value>0.05) from the t-test, but there is a joinpoint between the two segments or vice
versa. Why?

Answer:
The Joinpoint program shows the estimates of the regression coefficients, i.e., intercepts and slopes,
and the changes of the slopes. The p-values of the slope changes are calculated from the t-test based on
asymptotic normality. Based on a p-value greater than 0.05, one might say the two slopes are not
statistically different and hence conclude that the two segments are the same. But using the p-value
from a t-test is not as accurate as that from the permutation test, since the t-test is an asymptotic test
and the variances are calculated using the information matrix conditional on the estimated joinpoints
without imposing the continuity constraint and omitting the offending observations. The software (with
the permutation test procedure) does not require the asymptotic normality and maintains the correct
Type I error probability level and hence the number of joinpoints determined by the software is more
reliable.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Consecutive Non-Significant Segments
Sometimes, the APC for one segment is significantly different from zero, but when an extra
joinpoint in the segment is determined by the Joinpoint software, neither APCs for the two
consecutive segments are significant. Why?

Answer:
First, the test of APC is based on asymptotic t-test and the number of joinpoints is determined by
permutation test, which is more reliable. When a segment is divided into two segments by a joinpoint,
there are less data points in each segment, which usually decreases the power of each test. An omnibus
test, which combines both segments, is a more powerful test to detect the overall trend for both
segments.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Non-Equally Spaced Time Intervals or Non-
Integer Time Points
Can I do Joinpoint regressions where the time intervals are not equally spaced or the time
points are not integers?

Answer:

Yes. The Joinpoint software since Version 3.0 allows unequally spaced time intervals and non-integer
time points.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Modeling Proportions & Percents
Can Joinpoint be used to model proportions or percents?

Answer:

Joinpoint can be used to model proportions or percents, and for consistency, assume that percents are
transformed to proportions before analysis.
For confidence intervals of the model parameters and related p-values, Joinpoint assumes that the data
being analyzed arises from a normal distribution or the number of observations is large enough to use
the asymptotic normality of the estimated model parameters.
If ni is the sample size for the ith observation and is large enough that both ni pi and ni(1-pi) are larger
than or equal to 10, then the sample proportions can be considered to asymptotically follow a normal
distribution with standard error equal to .
If the proportions arise from a complex survey, then the standard error from a complex survey analysis
statistical package (e.g. SUDAAN) can be used. If ni is not large enough for a normal approximation (that
is, either ni pi or ni(1-pi) or both are smaller than 10), then the distribution may be skewed, and Joinpoint
results based on asymptotic normality may not be accurate unless the number of observations is large.
To correct heteroscedasticity, one would want to incorporate the standard errors of the proportions and
the standard error of can be used. These are the standard errors that are automatically
computed and incorporated when proportions are calculated within Joinpoint. If ni is not available, but
all the proportions are in approximately the same range, and the sample sizes are known to be
approximately the same, then it may not be necessary to enter the standard errors (i.e. in this case all
the standard errors are approximately the same, which is the implicit assumption in Joinpoint if no
standard errors are entered).

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Zeros in the Dependent Variable
How does the Joinpoint software deal with zeros in the dependent variable?

Answer:

It depends on the settings that are selected:

If all records for a cohort have zero counts, then the cohort is NOT processed.
If some of the records for a cohort have zero counts, then Joinpoint will add 0.5 to zero counts in
the following scenarios:

There must be one or more records in the cohort with a count greater than zero in order for
Joinpoint to adjust zero counts within the cohort. When a session is executed, Joinpoint will scan
the input file and notify users of any cohort that will have a count adjusted. Users will have the
option to cancel the job. If the session is executed, the output will display a message for each
cohort that had one or more counts adjusted.
For all other situations, the cohort is NOT processed. In this case, a message will be displayed
by Joinpoint and the executed job cancelled.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Project or Extrapolate Into the Future
Can I use Joinpoint to project or extrapolate into the future?

Answer:

While the Joinpoint program itself does not do projections/extrapolations, it is not unreasonable to use
Joinpoint to project a few years into the future. To do so, one would need to extrapolate the final
segment, and compute the standard errors of the projection, using the betas displayed in the output. The
Joinpoint Team has had some success using Joinpoint for projections (validated against actual values
when the data is rolled back several years so the projected values can be validated against actual
values). However, a projection using Joinpoint may or may not be more accurate than other methods (e.g.
time series methods), depending on the situation. Long term projections using Joinpoint are not advised.
Also, AAPCs, since they are summary measures over an interval, are not as appropriate to use for
projections as is the APC of the final segment.
References which used Joinpoint for projections are:
Chen HS, Portier K, Ghosh K, Naishadham D, Kim HJ, Zhu L, Pickle LW, Krapcho M, Scoppa S, Jemal A,
Feuer EJ. Predicting US- and state-level cancer counts for the current calendar year: Part I: evaluation of
temporal projection methods for mortality. Cancer 2012 Feb 15;118(4):1091-9. [Full Text ]
Zhu L, Pickle LW, Ghosh K, Naishadham D, Portier K, Chen HS, Kim HJ, Zou Z, Cucinelli J, Kohler B,
Edwards BK, King J, Feuer EJ, Jemal A. Predicting US- and state-level cancer counts for the current
calendar year: Part II: evaluation of spatiotemporal projection methods for incidence. Cancer 2012 Feb
15;118(4):1100-9. [Full Text ]

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Importing Graphs into Graphics Software
How do I import Joinpoint's graphs into other software such as Word, Excel, or PowerPoint?

Answer:

The graphs themselves, or the data to produce the graphs, can be exported from Joinpoint into a format
which can be used by other software packages. You have the following options:
I. Save the graph as a BMP or JPEG file
Option 1: Right click on the graph and a menu will appear. Select the “Export Graph...” menu
item. A window will appear where you can specify the file name, the type to save it as (BMP or
JPEG image), and the directory to save the file in. Once these are specified, click the Save
button.
Option 2: Use the Output > Export... menu item. Once the menu item is selected, the Export
dialog window will appear. Click the “Displayed Graph” option and then use the associated
“Browse” button to select a file name and location to save the file. When browsing, you will be
able to set the file type (BMP or JPEG) at the bottom of the window. Once the file name and
location have been specified and you have returned to the main Export dialog window, click on
the Ok button and the graph will be exported and saved to file.
Then, the saved file can be imported to other software like any other image or picture would be.
II. Save the data coordinates (X, Y-observed, Y-fitted) used for creating the graph
Option 1: Copy and paste the coordinates from Joinpoint into your software (Excel, PowerPoint,
etc.). To accomplish this, please perform the following steps:
Go to the data tab.
Copy the contents of the data grid to the Windows clip board. To do this, follow these steps:
Right-click on the data grid and a menu will appear.
Select the “Select Current Grid” option. Once you do that, the entire grid will turn
blue (indicating the grid is selected).
Right-click on the grid again, and select the “Copy” menu item. The data grid
contents have now been copied to the Windows clip board.
Paste the copied data tab coordinates into your software (Excel, etc.). For Excel: Open
Excel, go to the spreadsheet where you want the information copied, right-click on a cell in
the spreadsheet and select Paste (or use Ctrl-V).
Option 2: Use the Output> Export... menu item. Select to export the contents of the data tab
to a file. The file can be imported into other graphics software packages. For a more detailed
explanation of how to use this option, please see the following FAQ: Exporting Results.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Printing Results
How do I print Joinpoint results?

Answer:

Joinpoint can print both session parameters and output results. In order to print either, please follow
these steps:

1. Make sure the output or session window you want to print is the active window.
2. Click the print icon ( ) shown in the main Joinpoint toolbar or select the File > Print... menu item
3. If you are printing a session:
A Report Viewer window will appear showing you the ‘Print Preview’ of the session
parameters.
Select the print icon ( ) shown in the upper toolbar to print the report.
4. If you are printing output results:
A window will appear providing you with Print Options.
Select the Cohort, Model, and Output information you wish to print.
Click the Print button.
A report viewer window will appear containing your selected output.
The number of pages to be printed will appear in the upper left-hand corner of the Report
Viewer window. Please review this number to verify that your specified Print Options do not
print an unexpectedly large number of pages.
Select the print icon ( ) shown in the upper toolbar to print the report.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Saving Results to PDF or Word
How do I save Joinpoint results to PDF or Word?

Answer:

Joinpoint results from all of the tabs in the Output Window can be exported to PDF or Word via the
Report Viewer. You must follow the same steps as you would to print results, but instead of sending the
results to a printer, you can direct Joinpoint to save the results to PDF or Word. Please follow the steps
below:
Make sure the output or session window you want to print has focus.
Click on the print icon ( ) shown in the main Joinpoint toolbar or select the File > Print... menu
item.
A window will appear providing you with Print Options.
Select the Cohort, Model, and Output information you wish to print.
Click the Print button.
A report viewer window will appear containing a ‘Print Preview’ of the information you have
selected to print.
The results displayed in the report viewer can now be exported to PDF or Word. At the top of the
report viewer window click on the Export icon ( ). If you place the mouse over the icon, a ‘hint’
will be displayed saying ‘Export’.
Once the icon is pressed, a menu will appear under it. From this menu, select the output type you
want: PDF or Word.
Once an output type is selected, you will be able to specify a file name and location to save the
resultant file.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Exporting Results
How do I export Joinpoint results to a text file or to Excel?

Answer:

Joinpoint results can be saved to a text file in order to be used as input into other software or saved as
an Excel spreadsheet. To export the Joinpoint results, please follow these steps:
T o export to text files:
Make sure the output window containing the results you want to export has focus.
Select the Output > Export > T ext... menu item.
A window will appear providing you with various export options.
Click the check boxes associated with the information that you wish to export, specify the file
name associated with each selected item, and define the various other output options.
Click the Ok button.
T o export to an Excel spreadsheet:
Make sure the output window containing the results you want to export has focus.
Select the Output > Export > Excel... menu item.
A window will appear providing you with various export options.
Click the check boxes associated with the information that you wish to export. Each export
selection will be placed in its own tab within the spreadsheet.
Click the Ok button.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Exporting High Resolution Graphs
How do I export high resolution graphs?

Answer:

There are two ways to export high resolution graphs from Joinpoint:
Option 1: If you want to export one graph at a time, you can right-click on the graph and select
" Export High Resolution Graph" .
Option 2: If you want to export multiple graphs, use the " Output" -> " Export" -> " Text" menu
option. On the Export Window, select the " Graphs" checkbox at the top of the window. Next,
select the " Use High Resolution Graph" option and set the pixel size of the graph image. See
Exporting Results in the help for more information.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017

Using Excel as a Data Source
When I use Excel as my data source, why do I get "#####" in the data?

Answer:

When using Excel as your data source, Joinpoint will import the values from the Excel spreadsheet
exactly as they look. It does not take the actual value in the cell, it takes the formatted display value.
For example, in your Excel spreadsheet you have a cell with the value 5.1234. You have formatted the
column it’s in to display just 1 decimal place. So in the spreadsheet you see 5.1. When you import this
data, Joinpoint will import the value as 5.1. The data displayed in the grid on the Input File tab of your
Joinpoint session will reflect this.
The most common reason to get “######” in a field is that the column is too small. You just need to
resize your column large enough to display all of the data in the column. You then need to save your
Excel spreadsheet, and then reimport it into Joinpoint.

Joinpoint Help Manual 4.5.0.1 | https://surveillance.cancer.gov/help/joinpoint/ | 12 June 2017