Lecture No #08
STUDY DESIGNS (II)
(Experimental Study Design)
Dr Hafsa Paracha, PT
DPT, MSBE (DOW)
Research Coordinator,
Dow Institute of Physical Medicine and Rehabilitation,
Dow University of Health Sciences
[email protected]
OBJECTIVES:
At the end of the lecture, students will be able to know:
What are experimental study designs?
What are the principles of experimental study design?
What are the Clinical Trials?
What are Major clinical Research Study designs?
What are the Single Subject Study Designs?
EXPERIMENTAL OR
INTERVENTIONAL STUDY
DESIGN
In experimental study designs, Researchers assign patients to intervention and
control/comparison groups in an attempt to isolate the effects of the intervention.[1]
Experimental study is plan usually consist of selection of treatments whose effects are to be
determined and assignment of treatment to experimental units(the process of randomization)
along with collection of data for analysis.[2]
PRINCIPLES OF
EXPERIMENTAL DESIGN:
Following are the principles of experimental study designs:
I. Randomization
II. Control.
RANDOMIZATION:
“It is the process of assigning treatments to experimental units or subjects by some random
process with the help of random number table.”
BLOCK RANDOMIZATION:
Subjects are divided into blocks, then randomization is carried out in each block.
RANDOM NUMBER TABLE:
• The preferred method for generating a simple random sample is to use random numbers that
are provided in a table or generated by a computer.[3]
• Moving in either a horizontal or vertical direction. [3]
CONTROL:
Phases in clinical trials are divided into groups; one group is experimental group which is given
the treatment under investigation and the other is control group which is treated exactly in the
same except that the treatment is not given.[2]
In experimental study, the group receives the treatment is called intervention group. [2]
Control groups are of either of two: the non-treated control group or placebo control group. [2]
Non-Treated Control group: No treatment is given in the control group.
Placebo control group: Any dummy treatment like actual treatment is given to the subjects.
CLINICAL RESEARCH
DESIGNS
They are classified into Single-case/Individual based experimental design and group based
experimental design. [2]
SINGLE CASE EXPERIMENTAL STUDY DESIGNS:
Single case experimental study designs addresses themselves primarily to question of
differences, Is there difference between treatment A and treatment B. [2]
Single-subject designs were developed largely by behaviorists, influenced by scientists such as
BF Skinner, Sidman, Bijou, and others to demonstrate the influence of setting and other
intervention variables upon the performance of research participants, and to document the
individual variability of participants’ performance in response to these variables.[3]
Single case experimental study designs are frequently called as interrupted time series deigns
because measurements are repeated at specific time interval.[3]
Following are the types of individual based designs given by Kazdin:
I. A---- B single case design
II. A---B---A single case design
III. Mulitple- Baseline Designs
IV. Alternating Treatment designs.
V. Interaction Designs
VI. Changing Criterion designs
WHEN TO USE A SINGLE
SUBJECT STUDY DESIGN
The scientist-practitioner may use single-subject designs in any of several different situations. [3]
A single-subject design may be used where withholding treatment is considered unethical, or
random assignment of subjects may not be possible. [3]
Single-subject designs may be used where it is too difficult to get enough participants for a
study. [3]
The most frequent reason for using a single subject design is to obtain detailed information
about factors, such as participant features, setting descriptions, and specific intervention
procedures on the variability of a participant’s performance. [3]
CHARACTERISTICS OF
SINGLE SUBJECT STUDY
DESIGN
Single-subject designs have several key characteristics: [3]
1. Baseline assessment: An extended baseline assessment provides a picture of how the subject is
currently performing or “behaving,” prior to introduction of the intervention)
2. Stability of performance: All subjects will have some variation in performance. Change in conditions
(e.g., introduction of the intervention after a baseline assessment) is typically not introduced to a
subject until a subject’s performance has stabilized.
3. Continuous assessment: Continuous assessment allows us to determine what are normally occurring
fluctuations in a patient’s performance versus those changes in performance that are due to the
introduction of the intervention
4. Use of different phases: Single-subject designs have at least two levels of the independent variable:
typically, a baseline phase and an intervention phase.
In single-subject designs, three data points are considered the minimum necessary to determine a subject’s
performance pattern, particularly the trend of performance (i.e., an increasing trend, a decreasing trend, or
staying the same); however, three data points may be insufficient to determine a subject’s pattern of
performance variation. Thus, more than three baseline data points are often observed in single subject
studies.
BASIC NOMENCLATURE
PHASE A :By convention, a phase in which no intervention has been implemented is represented
by the letter A. These typically include baseline (i.e., before any treatment) and any phases after
implementation of treatment in which no treatment is present (e.g., treatment withdrawal). [3]
PHASE B:Phases in which an initial treatment is present are represented by the letter B. When
the patient begins receiving the intervention, that phase would be represented by the letter B. [3]
If treatment is withdrawn, the phase is represented by A, and if treatment is reintroduced, the
designation is a B.
PHASE C: If a second (i.e., different) intervention is introduced, it would receive the designation
C. [3]
A- B Single Case Design:
The most basic of single-subject designs, the A-B design, has been described as the foundation of
single-subject research, even though it is considered a pre-experimental design.[3]
An A-B design consists of a baseline phase, A, followed by a treatment phase, B . [3]
The researcher collects multiple measures over time during the baseline phase to document the
patient’s status before implementation of the treatment .[3]
One weakness of the A-B design is that it does not control for events or extraneous factors that
might act simultaneously with the implementation of the treatment. If present, such factors, rather
than the treatment, might be the cause of any changes seen in the dependent variables . [3]
EXAMPLE
Marklund and Klässbo10 used an A-B design to investigate the effects of a 2-week trial of intensive
constraint-induced movement therapy (CIMT) for the lower extremity (LE) with five persons who had
experienced a stroke.
Both the baseline (A) and intervention (B) consisted of daily activities for 6 hours a day, 5 days a week, for
2 weeks each, with assessments occurring thrice weekly, and again at 3- and 6-months after the study.
During the A phase the participants engaged in non-organized training.
During the B phase the participants had their uninvolved LE immobilized while they engaged in a variety
of LE training activities, including sit to stand and return, stair training, and walking on uneven surfaces
indoors and outdoors. Assessment measures included the 6-minute walk test, step test, Timed Up and
Go, Fugl-Meyer subitems for the LE, weight distribution during standing, and walking ability.
If differences in any of the performance measures were seen during the B phase, a simple A-B design
would not control for historical influences, such as a change in spasticity medications, that might
influence lower extremity performance during the study
A-B-A SINGLE CASE DESIGN:
A-B-A designs, are characterized by at least one implementation and withdrawal of treatment
over the course of the study. [3]
In an A-B-A design, a baseline phase (A) is followed by treatment phase (B) and then by
another baseline (A). [3]
They are designed to answer the important clinical question, “Does this treatment work?”
Other variants either reverse the order of treatments and baselines (B-A-B) or have multiple
cycles of application and withdrawal of treatment (A-B-A-B). [3]
When studying a phenomenon that is expected to change as the intervention is implemented
and withdrawn, these designs provide good control of extraneous factors by enabling the
researcher to determine whether the hypothesized change occurs each time the treatment is
applied or withdrawn.
EXAMPLE:
Cross and Tyson, using an A-B-A design, examined the effects of a slider shoe on the walking
speed, walking efficiency, and Physiological Cost Index of four persons with a hemiplegic gait.
In this instance, the authors expected the participants’ performance to return to baseline levels
when the intervention was withdrawn. They reported that while all four participants’
performance returned toward baseline when the slider shoe was removed, only one
participant’s performance, subject 1, in the withdrawal phase actually returned to baseline
levels .
Thus, only subject 1 presents good evidence that the treatment, and not some extraneous
factor, was the cause of the changes in walking speed. Alternatively, there were carry-over
effects for the other three participants, even though their performance fell off of the levels
observed during the B phase.
MULTIPLE BASELINE
DESIGNS
Multiple baseline designs, like withdrawal designs, answer the question of whether or not a treatment is
effective. They avoid the ethical dilemma of withholding an effective treatment . [3]
Multiple baseline designs, and its variations can be used to: . [3]
Examine intervention effects with the different behaviors (“multiple baseline across behaviors” [e.g., heel-
strike gait, straight swing through of the leg, and reciprocal arm swing])
Across several different patients (“multiple baseline across participants” [e.g., the same treatment with three
different patients]),
With the patient’s same behavior, but across different settings (“multiple baseline across settings” [e.g., heel
strike gait on a tile floor, carpeted floor]),
The same behavior but across different persons (“multiple baseline across therapists” [e.g., different
therapists delivering the same treatment to the same patient]).
Traditionally, a minimum of three different behaviors, settings, patients, or therapists is considered necessary
for demonstration of intervention effectiveness using this design.
ALTERNATING TREATMENT
DESIGN:
Alternating-treatment designs, sometimes referred to as simultaneous treatments designs,
include the use of different treatments, each administered independently of the other, within
the same phase .[3]
The designs answer the clinical question “Which of two (or sometimes more) treatments is
better?” The treatments are rapidly alternated based on a random pattern .[3]
EXAMPLE
Washington and associates used a different variation of the alternating-treatment design to
study the impact of a contoured foam seat on sitting alignment and upper extremity function of
infants with neuro-motor impairments.
Alignment and upper extremity function were measured with the children seated in a standard
highchair during the baseline phase.
During the intervention phase, two different interventions were presented each day: use of a
thin foam liner within the highchair and use of a contoured foam seat within the highchair.
Notation for this study might be “A-B,C” with the comma indicating that the two conditions (B
and C) were alternated within a single phase.
INTERACTION DESIGN:
Interaction designs are used to evaluate the effect of different combinations of treatments. .[3]
They answer the clinical question, “Which aspect(s) of a treatment program is/are the agent(s)
of therapeutic change?” .[3]
EXAMPLE
Embrey and associates examination of the effects of neurodevelopmental treatment and
orthoses on knee flexion during gait is another example of a design examining interaction
effects.
The design of this study is A-B-A-BC-A, with neurodevelopmental treatment (NDT) as the B
intervention and BC the combination of NDT and orthoses treatments. In this study, the
researchers apparently were not interested in the effect of the orthoses alone, hence the
absence of an isolated C treatment.
CHANGING-CRITERION
DESIGN
This set of single-subject designs may be considered as a design that has great utility for clinical
practice .[3]
The changing-criterion design answers the clinical question, “Does changing the criterion for
success lead to improved performance?” .[3]
Thus, the question is critical for rehabilitation therapies that include reinforcement for desired
behaviors at defined levels so that the desired behavior is likely to continue.[3]
The changing criterion design begins with a baseline phase (A), followed by the intervention
phase (B) .[3]
The distinguishing feature of the changing-criterion design is that the participant’s performance
level is pre-established, and then the intervention is conducted with the purpose of achieving
the pre-established criterion .[3]
Once that criterion is achieved, the participant’s performance level is reset to a more stringent
level, and once again the intervention is implemented with the purpose of achieving the newly
preset criterion. Thereafter, in subsequent sessions, the performance criterion continues to be
reset, and the intervention is implemented with the purpose of achieving the criterion . [3]
REFERENCES:
1. https://www.ncbi.nlm.nih.gov/books/NBK470342/.
2. Biostatistics and Research Methods by Muhammad Ibrahim
3. Rehabilitation Research (Principles and Applications), 3rdEdition by Russell, Jay Lubinsky and
Elizabeth Domholdt.(Published in 2004 by Saunders).
