International Journal of Sports Science 2016, 6(6): 219-229

DOI: 10.5923/j.sports.20160606.04

Traditional vs. Undulating Periodization in the Context of

Muscular Strength and Hypertrophy: A Meta-Analysis
Leonardo C. Caldas1,*, Lucas Guimarães-Ferreira1, Michael J. Duncan2,
André S. Leopoldo1, Ana Paula L. Leopoldo1, Wellington Lunz1

1
Center of Physical Education and Sport, Federal University of Espírito Santo (UFES), Vitória, Brazil
2
Department of Biomolecular and Sports Science, Coventry University, Coventry, UK

Abstract There is currently no consensus regarding the best periodization model to increase muscle strength and
hypertrophy. However, most recently the undulating periodization (UP) has been believed to be better than traditional
periodization (TP) model. The purpose of this study was to conduct a meta-analysis to investigate whether the UP is superior
to TP to induce gains in strength and hypertrophy. Studies were searched in databases covering three idioms. Twenty-five
studies met the inclusion criteria, allowing 72 and 6 effects sizes (ES) to strength performance and hypertrophy, respectively,
covering 400 males and 192 females. Duration of training protocols ranged 6 to 16 weeks. Continuous data from maximum
strength (1RM), isometric, power, and muscular endurance (RMs) tests were collected. Additionally, only gold-standard
measures were included for hypertrophy assessment. Statistical analyses were performed using dedicated software for
meta-analysis. Considering the pooled data, UP was significantly (P = 0.005) larger than TP model to 1RM test (ES= 0.22;
95% IC= 0.07, 0.38). However, there was no significant difference (P > 0.05) between UP vs. TP for power (ES= -0.04; 95%
IC= -0.29, 0.22), RMs (ES = 0.20; 95% IC = -0.07, 0.48), isometric strength (ES = -0.13; 95% IC = -0.50, 0.24), and
hypertrophy (ES = 0.32; 95% IC = -0.07, 0.71). UP model seems better than TP model to improve maximum strength
performance, but not to power, muscular endurance, isometric strength, and muscle hypertrophy.
Keywords Muscular strength, Resistance training, Weight training, Power training

with high volume and low intensity and progressively

1. Introduction shifting to high intensity and low volume throughout the
training program [6]. It comprises training phases with 4 to 6
It has long been recognized that periodized strength weeks of duration with different goals (e.g., hypertrophy,
training induces higher gains to maximum strength, power, strength, and power or peak) [7].
and muscular endurance compared to nonperiodized training On the other hand, the volume and intensity in the UP
[1, 2]. Considering that neuromuscular system adapts model vary more frequently, and a number of neuromuscular
quickly to a resistance-training program, it has been accepted components are integrated into a same-phase [6, 7]. UP is
that changes in training program are required for continuous structured in training zones (e.g., strength, hypertrophy, and
gains in strength and muscle mass [1, 3]. However, the most muscular endurance) which must fluctuate daily, weekly [8,
effective strength training periodization model has yet to be 9], or biweekly [6, 10].
elucidated. Some researchers have believed that UP is more effective
Although there has been criticism about the historic than TP to improve strength and others parameters [3, 11, 12].
empiricism of periodization programs [4], the traditional (TP) It is hypothesized that more frequently varying in the
and undulating periodization (UP) have been the most training stimulus could result in an optimization of the stress
frequently used models to improve muscular strength and and recovery on the neuromuscular components [10, 12, 13].
hypertrophy. The TP is termed 'linear' program, despite it is There are, however, counterarguments regarding the
also characterized by undulations in both volume and hypothetical superiority of the UP model [9, 14]. In fact,
intensity across each micro- and mesocycles [5]. The TP other researchers found better results in favor to the TP
model is characterized by a division of the entire seasonal regarding maximum strength improvement [15, 16].
program into smaller periods and training unit [5], initiating Taken collectively, there appears to be no consensus
regarding which model of periodization is most suitable and
* Corresponding author:
[email protected] (Leonardo C. Caldas)
effective in order to improve strength and muscle
Published online at http://journal.sapub.org/sports hypertrophy. In these circumstances, a meta-analytical
Copyright © 2016 Scientific & Academic Publishing. All Rights Reserved review is essential to support the decision-making; it is
220 Leonardo C. Caldas et al.: Traditional vs. Undulating Periodization in the
Context of Muscular Strength and Hypertrophy: A Meta-Analysis

particularly useful because it is a quantitative, impartial and applicable only for the studies of Kok [20]; Kok, et al. [21])
reproducible technique that permits inclusion of multiple to prevent multiple publication bias [18].
scientific studies into the analysis. Thus, the sample number After the literature search, the following sequence was
is increased, standard error (SE) decreased and statistical carried out: Firstly, the studies were saved electronically for
power enhanced, providing a more robust overview of any further reading and evaluation. Secondly, abstracts were read
given area. and those that did clearly not meet inclusion criteria were
Therefore, using a systematic review followed by excluded. Thirdly, the abstracts showing sufficient
meta-analysis we aimed to investigate the effects of two information covering the inclusion criteria and had no reason
periodization models (UP or TP) on strength, power, for exclusion were filed to further full text reading. Fourthly,
muscular endurance and hypertrophy. after full text reading, the studies that fulfilled the inclusion
criteria were selected for review, and studies that did not
meet all inclusion criteria were excluded. Finally, after all
2. Method full text reading of each study, the reference lists were also
searched for any additional papers that were not found by our
2.1. Experimental Approach to the Problem
search strategy.
This systematic review and meta-analysis was performed
in agreement with the steps advocated by Cochrane 2.3. Coding and Classifying Variables
Handbook [17]. The question we wanted to answer was: The key categories coded of each study included were:
Which is the best periodization model to improve strength (a) Identification of the studies (authors, title, year
performance and muscular hypertrophy: traditional or publication, journal, objectives); (b) Methodological quality
undulating periodization? characteristics (strategy of randomization and masking,
The electronic searches included two recognized database study design, allocation concealment, intervention
(PubMed and Scopus) considering their scientific relevance, monitoring, follow-up loss); (c) Sample characteristics
quantity of studies and language (English), and additionally (age, sex, weight, height, number, fitness level, concurrent
a database (Scientific Electronic Library Online; SciELO) training, intervention type); (d) Methodological
that comprises more than 10 emerging countries in science characteristics (type of training, testing, measurement
covering different languages (Portuguese, Spanish, and technique, control type, statistical analysis); (e) Training
English). This strategy was performed to minimize bias from characteristics (duration, frequency, number of sets, interval
database and English language [18, 19]. between sets, exercises selected and muscles trained, volume
and intensity work); (f) Results (anthropometrics and
2.2. Procedures
performance in the strength tests). The main outcome coded
We used the following search strategy in PubMed, which was the results given by strength tests.
included conjugated terms and their required position on the All data extracted were registered in a datasheet. The
paper: (undulating [Title/abstract]) OR nonlinear mean, deviation (SD or SE) and sample number from pre-
periodization [Title/Abstract]) AND training [Title/Abstract]) and post-intervention were extracted for each group included
AND Humans [Mesch]). Regarding Scopus we used the by studies. We considered teenagers who aged ≥ 10 and < 20
conjugated terms (TITLE-ABS-KEY (undulating) OR years-old, adults between ≥ 20 and < 60 years-old and
TITLE-ABS-KEY (nonlinear) OR TITLE-ABS-KEY elderly ≥ 60 years-old [22]. We classified participants in
(non-linear) AND TITLE-ABS-KEY (periodization). And each study as either experienced or not experienced in
we used the following search strategy to the SciELO: resistance training based on the information provided by the
(ab:(undulating)) OR (ab:(no linear)) AND (ab:(training)). authors of each study. In one study [23] was stated that
These terms were searched in Portuguese, Spanish, and participant had experience in strength training, however it
English. For all databases there were no limit to data, age, or was not informed how much experience (e.g., months or
gender. years of training). In addition, the participant from this same
The inclusion criteria was as follows: Original and study were absent of training for two months. Therefore, we
interventional studies with healthy human submitted to coded this sample as without experience. It was not possible
strength training; presence of one or more dynamic and/or to code one study [24] because the participants with and
static tests used to evaluate strength (maximal, endurance, without experience were grouped together.
and/or power); gold standard imaging techniques (magnetic Strength performance tests were coded in maximum
resonance, computerized tomography or ultrasound) to strength (one-repetition maximum; 1RM), muscular
assess muscle hypertrophy; the aim stated by authors was to endurance (repetitions maximum; RMs), power, and
compare UP vs. TP; continuous outcome presented as mean, isometric strength. The data for 1RM were split into four
standard deviation or standard error, and sample number. categories: Bench press, leg press, squat, and other 1RM
Conference abstracts were excluded due to insufficient tests. This decision was taken because bench press was used
methodological description. Results repeated in more than in all studies that evaluated upper body strength, and leg
one publication form were included just once (This was press and squat were used by the most of studies that
 International Journal of Sports Science 2016, 6(6): 219-229 221

evaluated lower body strength. Other 1RM tests were For these reasons, we developed a specific qualitative
included in a unique subgroup. Of note, some studies utilized evaluation considering the main methodological issues
more than one test that could be included in this last highlighted from above-mentioned documents and other
subgroup. For these reason, to prevent the possibility that related specifically to exercise-based interventions. Thus, we
one or few studies could skew the overall results, only one considered 8 criteria, as follows: (1) Study design
test from each study was chosen randomly. Thus, the characteristics: if the sample was randomized (2)
meta-analysis to 1RM was completed to bench press, leg Participants’ health condition: if sample participant had
press, squat, and other tests separately (as subgroup) and some risk factors (e.g., tobacco use, diabetes, hypertension,
pooled. cardiovascular diseases, obesity); (3) Masking method: if the
For muscular endurance tests (RMs tests), two subgroups experimental design was blinded or double-blinded; (4)
were created: Bench press and lower body exercises. Again, Monitoring of training: if a expert professional accompanied
bench press was chosen because it was used for all studies. the training intervention; (5) Loss to follow-up: if there were
Several exercises were used to lower body, being most of sample withdrawals or dropouts; (6) Concurrent activities: if
them were leg press, squat, and leg extension. Therefore, we the sample was involved in others modalities of training; (7)
decided to cluster and describe them as lower body segment Dietary prescription: if there was monitoring of food and
RMs test. Some studies used more than one test that could be supplementation intake; (8) Equated work: if the total work
included in this last subgroup. In this case, we randomly of training was matched to both periodization models. In this
included a single test for each study. The meta-analysis to case, we accepted work described in Joules or as load x
RMs test was completed to bench press and lower body repetition.
segment RMs test separately (as subgroup) and pooled.
Regarding power tests, two subgroups were created: 2.5. Sensitivity Analyses
Jumping and throwing. For jumping, the more common test Three sensitivity analyses were conducted. First, we
was vertical jump (either countermovement or the squat analysed if the results were affected by training status. In this
jump). For throwing, the majority of studies employed the case, participants considered by authors as experienced or
bench press or throw of medicine ball. In this case, the inexperienced to strength training were analysed separately.
meta-analysis was conducted only as pooled because of the Second, we evaluated if the results were affected by
small number of data to the subgroups analysis. Just three conceptualization of periodization adopted for the authors. In
studies evaluated isometric strength [14, 25, 26]. Thus, a this analysis, we included or excluded all those studies that in
meta-analysis was conducted on the data from these studies. our own understanding could or not be classified as UP,
The hypertrophy measures were obtained from the arm independently of the author’s interpretation. In this regard,
[12, 26] or thigh [20, 21, 27, 28]. Nonetheless, the analyses we conceptualized the UP as a model of training that is
were only carried out using the data together, independently structured in different zones of training (e.g., muscular
of the segment evaluated. However, when elbow flexors and endurance, hypertrophy, maximum strength, and power)
extensors were measured by the same study [12, 26], only with frequent variations in the volume and/or intensity
one data was randomly included to avoid the double effect during periods (i.e. daily or weekly) that are repeated
size from the same design. through the training. Moreover, we interpreted the TP as a
Of note, only post training data for both UP and TP models model of training characterized by phases lasting 3 to 4
were employed to perform this meta-analysis. However, this weeks of duration, with each phase distinguished by a
strategy would only be possible if pre training data for both specific objective. It begins emphasizing volume and later
UP and TP models had been matched. Therefore, we first intensity. These conceptual interpretations are analogous to
performed a meta-analysis to the UP and TP using only pre other authors [6-9, 34].
training data, and we confirmed that the data from both Thus, taking our own conceptualization into account, we
interventions were statistically equals (as shown in Results). could include the following studies [11-14, 20, 21, 23-29,
Prudently, we just did not include data from Arm Curl 1RM 35-41]. On the other hand, we had do exclude the following
test from one study [29] in the post training meta-analysis studies [15, 16, 42-44].
because we found statistic difference between UP vs. TP Third, we analysed whether the matching training volume
groups at pre training. In addition, all 25 studies included in could influence the results. In this case, we included only
meta-analysis found that both UP and/or TP models were studies that work was matched.
sufficient to induce performance gain on the most of the tests
as compared with the pre training data. 2.6. Statistical Data Analysis
All statistical analysis was processed using the RevMan
2.4. Assessment of Study Quality Manager software [45], which is a software available for free
The assessment of study quality in general has been download (http://tech.cochrane.org/revman/download). All
accessed by strategies that scoring system are included or are math algorithm used here can be assessed in details in
not included [30-33]. However, these proposals are 'help/statistical algorithms in revman', into the RevMan
dedicated to clinical trials, but not to exercise interventions. software. Briefly, data of the pre and post intervention from
222 Leonardo C. Caldas et al.: Traditional vs. Undulating Periodization in the
Context of Muscular Strength and Hypertrophy: A Meta-Analysis

both UP and TP were processed. The effect sizes of each sessions/week. Most of studies were ≤ 12 weeks of training
group were calculated using the standardised mean duration (80%), adults sample (84%), participants
difference (SMD). The SMD of each study was processed by experienced in strength training (60%), RMs as intensity
a random-effect model. The weight of each and pooled SMD parameter (64%; most of them covered by ≤ 15 RMs) and ≤ 5
were established by the inverse variance method. sets/exercise (92%). Table 1 summarizes the mains
When studies presented standard error of the mean (SE), methodological characteristics from the studies.
we converted it in standard deviation (SD) using the equation
�� = �� � √�. The heterogeneity of data were assessed by 3.2. Methodological Quality of Studies
the chi-square (χ²), and the heterogeneity-induced variability The majority of studies (68%) described that participants
percentage of the effect sizes was accessed by I-square (I²) were randomly allocated into each group, while others
[46]. The 95% confidence intervals (95% CI) for each SMD studies paired (20%), or allocated the participants by
and for all pooled estimating were calculated, and the convenience sampling (4%), or did not describe any
maximum type I error was 5% (P ≤ .05). allocation strategy (8%). Regarding clinical conditions, 72%
of the studies described that the samples were composed by
129 Abstracts were found in database, healthy participants, one study (4%) was composed by obese
PubMed, SciELO, and Scopus. participants and 24% did not describe the clinical condition.
None of the studies described any strategy of masking during
the performance tests. Thus, they were not blinded or
21 studies maintained after abstracts double-blinded. In most of studies the monitoring of training
reading and a rapid assessment of full was used (80%), while three studies (16%) did not describe
text. information about this issue, and one (4%) highlighted that
no monitoring of training was used. An expressive number of
studies (40%) did not provide information about loss of data
7 studies could be included after at follow-up. Five (20%) described that there were no
searching in the reference list from the withdrawals or dropouts, and 40% of studies described loss
studies previously accessed. of one or more participants. In addition, almost one third of
studies (28%) did not describe anything about concurrent
28 studies included and activities. In 32% of the studies the authors stated that
archived. participants did some concurrent training and 40% of the
studies stated that participants performed only strength
Careful reading of the 28 articles
training.
archived.
Regarding dietary prescription, the majority of studies
2 studies excluded due absence (64%) did not prescribe any diet, 28% did not assess or
of the data required and 1 control the dietary intake, and just 8% incorporated some
study excluded due using a form of dietary monitoring. In relation to the training work
reverse linear preriodization. equalization, 20 studies (80%) from a total of 25 affirmed to
have equated the work. However, only 11 studies (44%)
25 studies had yours data accessed and showed data that could allow us confirm this methodological
included in the meta-analysis. statement. From these 11 studies, it just 9 of them (36%)
calculated the work using ‘load x repetitions’. The data
Figure 1. Procedure for selection of the studies and decision-marking regarding to the methodological quality of studies are
inclusion and exclusion
showed on the Table 2.

3.3. Measures of Strength Performance and

3. Results Hypertrophy
3.1. Descriptive Data At first, we performed a meta-analysis from the
Figure 1 depicts the algorithm of searching and selection. pre-intervention results to confirm if the data from
Three studies were excluded after reading all full texts. The participants allocated into UP and TP groups were
first one [47] because the mean and SD in pre- or post homogeneous before training intervention. As expected, all
intervention could not be obtained, other [48] due absence of results from the performance tests were statistically similar
results about strength or hypertrophy, and the last one [49] (all Z-value were ≤ 1.20 and all P-value were ≥ 0.23), as
because absence of TP group. We were able to include 25 follows: Maximum strength (effect size = 0.08; 95% IC =
studies, allowing 72 effects sizes to strength performance -0.05, 0.20;), muscular endurance (effect size = 0.01; 95%
and 6 effects sizes to hypertrophy. It amounted to 592 IC = -0.23, 0.26), power (effect size = 0.07; 95% IC = -0.19,
participants (400 male; 192 female). The intervention 0.33), isometric strength (effect size = -0.03; 95% IC = -0.40,
duration ranged from 6 to 16 weeks, with 2 to 4 0.35), and hypertrophy (effect size = -0.04; 95% IC = - 0.40,
 International Journal of Sports Science 2016, 6(6): 219-229 223

0.35). Chi-square test (χ²) did not find any difference (all using the techniques previously described in methods (5 of
P-value were ≥ 0.66) and all I2 were zero. them used ultrasonography and only 1 used magnetic
Regarding the subgroups analysis for 1RM tests resonance).
pos-intervention, there was no statistical difference between
UP vs. TP on bench press (Z = 1.60, df = 19, P = 0.11), squat 3.4. Sensitivity Analyses
(Z = 0.75, df = 9, P = 0.45), leg press (Z =1.84, df = 8, P = Firstly, we analyzed if training status could affect the
0.07), and other 1RM tests (Z = 1.21, df = 5, P = 0.23). results of meta-analysis. Participants were categorized as
However, when the analysis was made pooling all data, it experienced and inexperienced to strength training in
was found a significant difference in favoring UP (Z = 2.80, according to the classification made by authors. Altogether,
df = 44, P = 0.005; effect size = 0.22; 95% IC= 0.07, 0.38; results were not changed after this analysis (Z = 1.71, df = 15,
Figure 2). P = 0.09).
In relation to RMs tests, there was no difference between The second analysis included data from studies according
UP vs. TP for bench press (Z = 0.64, df = 3, P = 0.52) and to our own conceptualization of periodization (as explained
lower body segment (Z = 1.25, df = 5, P = 0.21) either as before). In general, the results were not significantly changed,
subgroup or pooled (Z = 1.43, df = 9, P = 0.15). No however a significant effect size for RMs tests was also
difference was found between UP vs. TP in power tests, found in direction to the UP model (effect size = 0.35; 95%
either as subgroup (jumping, Z = 0.79, df = 6, P = 0.43; IC = 0.02, 0.67; Z = 2.10, df = 5, P = 0.04).
throwing, Z = 0.45; df = 4, P = 0.65) or pooled (Z = 0.27; The last analysis was performed using data from those
df = 11, P = 0.79). Similar result was found to isometric tests studies that the training work was matched. The statistical
(Z = 0.69, df = 4, P = 0.49) (Table 3). results were not altered. However, the effect size was
In relation to hypertrophy, there was no difference approximately 2-fold larger in direction to the UP model
between UP vs. TP (Z = 1.59, df = 5, P = 0.11; effect size = when we considered only work matching data for 1RM tests
0.32; 95%IC= -0.07, 0.71). It is important to mention that (effect size = 0.47; 95% IC = 0.17, 0. 77; Z = 3.05, df = 14,
only 6 studies [12, 20, 21, 26-28] measured hypertrophy P = 0.002).
Table 1. Characteristics of the subjects and studies included in the meta-ananlysis

Age Training Duration

Author (ye ar) se x classification Training le ve l Inte nsity Se ts days/we e ks (we e ks)
Apel et al. (2011) 15 M Adult T rained 57-80%1RM 3-6 4 (A + B) 12
Baker et al. (1994) 35 M Adult T rained 3-10RMs 3-5 3 (2 x A and 1 x B) 12
Bartolomei et al. (2015) 36 F Adult T rained 65-93%1RM 3-5 3 (A + B + C) 10
Buford et al. (2007)23 Both Adult Not trained 80-90%1RM 3 3 9
Caldwell, A.M. (2004)24 Both Adult Both 4-8RMs 3 3 (2 x A and 1 x B) 6
Foschini et al. (2010) 37 Both Adolescent Not trained 6-20RMs 3 3 14
Hartmann et al. (2009)25 M Adult T rained 3-25RMs 5 3 14
Hoffman et al. (2003) 42 M ND T rained 2-10RMs 3 2 12
Hoffman et al. (2009) 38 M Adult T rained 1-12RMs 3-5 4 (A + B) 15
Kok, L.Y. (2006) 20 F Adult T rained 30-90%1RM 3-4 3 12
Kok et al. (2009) 21 F Adult Not trained 30-90%1RM 3-4 3 9
Lima et al. (2012) 16 F Adult Not trained 15-30RMs 3 4 (A + B) 12
Marques et al. (2011) 44 M Adolescent T rained 50-85%1RM 1-4 2 8
Miranda et al. (2011) 39 M Adult T rained 4-10RMs 3 4 (A + B) 12
Monteiro et al. (2009) 40 M Adult T rained 4-15RMs 3-4 4 (A + B) 12
Painter et al. (2012) 14 Both Adult T rained 3-12RMs and 65-95%1RM 3 3 10
Peterson et al. (2008) 13 M Adult T rained 30-100%1RM 2-9 3 9
Prestes et al. (2009)43 M Adult T rained 6-12RMs 3 4 (A + B) 12
Prestes et al. (2015)29 F Older Not trained 6-14RMs 3 2 16
Rhea et al. (2002) 11 M Adult T rained 4-8RMs 3 3 12
Rhea et al. (2003) 41 Both Adult T rained 15-25RMs 3 2 15
Simão et al. (2012) 12 M Adult Not trained 3-15RMs 2-4 2 12
Spineti et al. (2013) 26 M Adult Not trained 3-15RMs 2-4 2 12
Souza et al. (2014) 27 M Adult Not trained 6-12RMs 2-4 2 6
Ullrich et al. (2015)28 F Adult Not trained 60-80%MVC 5 2 14
Note. Adult = ≥ 20 year and < 60 anos ; Adolescent = ≥ 10 ano < 20 anos; ND = Non described; RM= Repetition maximum; (A + B) = Program with
alternating sessions A and B; (2 x A and 1 x B) = program with alternating sessions, session A 2 times a week and B 1 time per week; (A + B + C) =
Program with three different training sessions.
224 Leonardo C. Caldas et al.: Traditional vs. Undulating Periodization in the
Context of Muscular Strength and Hypertrophy: A Meta-Analysis

Table 2. Assessing the quality of studies considering criteria of methodological rigor

) 36

) 24

) 5

) 44
09 2

) 42

) 38
) 37
15

) 23
5

04
4) 3

) 16

11
20

03

09
5

10

) 21
20

) 20
20
1) 1

07

20
20

20
.(

20

.(
.(

12
99

20
al

06

09
Author (year)

.(

.(

.(
al
.(
.M
01

20
(1

.(

al

al

al
et

20

20
al

et
(2

.(
al
l.

et
et

et

.(
ei

.(
et

n
ta

al
.

l,
om

et
al

.Y
an

al

es
an

an
el

ni

et
re

rd

dw
et

qu
,L
tm

et
hi

fm

fm
ol

a
ke

fo
l

rt

sc

ar
ok

ok
al

m
pe

ar

of

of
Bu
Ba

Ba

Fo
C

Li

M
H

K
A

Experimental
R P R R R R P CR R P P R ND
Design
Clinical
Healthy ND ND Healthy Healthy Obese ND Healthy ND Healthy Healthy Healthy Healthy
condition
Masking? ND ND ND ND ND ND ND ND ND ND ND ND ND
Monitoring of
Not ND ND Yes Yes Yes Yes Yes Yes Yes Yes Yes ND
Intervention?
Loss to follow-
Yes Yes ND Yes ND ND Not ND Not Yes Not ND ND
up?
Concurrent
Not ND Not Not Not Yes ND Yes Yes Not Not ND Yes
training?
Prescription
ND ND ND ND ND Yes ND ND ND Not Not ND ND
food?
Work was Yes Yes Yes Yes Yes
NS NS NS ND ND ND NS ND
equated? (n. Rep) (Rep X L) (n. Rep) (Rep X L) (Rep X L)
Note. Gray squares depicture the most desirable designs to assist in the visual interpretation; R = Randomized; P = Paired; ND = Not describle; NS = Not
shown; CR = Casual random; RLeg = Legs from the same subjet were randomized to different periodization models; (Rep x L) = Reps X Load; (n. Rep.) =
Number of repetitions. (Continued)

Table 2. Assessing the quality of studies considering criteria of methodological rigor. (continued)
g q y g g g ( )
40
39

3
8) 1
14

) 43

) 26

) 28
5) 2

) 12
)

) 27
)

) 11

) 41
09
11

)
12

00

13
09
0

15
20

12
01

14
(2

(2

02

03

20
20

20
(2

Author (year)
.(

20
(2

20
l.

.(
20

20
l.

.(
al

l.

.(
a

.(
l.

l.

al
a

.(

.(

al
et
et

ta

ta
et

al

al
et

al

al

et

et
o

se

se
da

on

et

et
ir

et

et

h
te
te
an

et
te

te

ão
rs

ic
a

a
in

uz
on

in
es

es
te
ir

he

he

llr
m
Pa

Sp
Pe

Pr

Pr

So
M

Si
R

U
Experimental
R R P R R ND R R R R R Rleg
Design
Clinical
Healthy Healthy Healthy Healthy ND Healthy ND Healthy Healthy Healthy Healthy Healthy
condition
Masking? ND ND ND ND ND ND ND ND ND ND ND Not
Monitoring of
Yes ND Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Intervention?
Loss to follow-
Not ND Yes ND ND Yes ND Yes Not Yes Yes Yes
up?
Concurrent
Not ND Yes Yes ND ND ND Not Yes Yes Not Not
training?
Prescription
Not ND Not ND ND Not ND ND Yes ND Not Not
food?
Work was Yes Not Yes Yes Yes Yes
NS NS NS NS NS NS
equated? (Rep X L) (Rep X L) (Rep X L) (Rep X L) (Rep X L) (Rep X L)
Note. Gray squares depicture the most desirable designs to assist in the visual interpretation; R = Randomized; P = Paired; ND = Not describle; NS = Not
shown; CR = Casual random; RLeg = Legs from the same subjet were randomized to different periodization models; (Rep x L) = Reps X Load; (n. Rep.) =
Number of repetitions.

Table 3. Effect size obtained for muscular endurance tests, power and isometric strength
p g
Effe ct siz e TP vs UP
Variable s me asure d n (df) Te st Z (P) C hi² (P) I²
[confide nce inte rval 95%]
RMs Bench press 100 (3) 0.18 [-0.37, 0.73] 0.64 (0.52) 5.59 (0.13) 46%
RMs Lown body 160 (5) 0.21 [-0.12, 0.55] 1.25 (0.21) 5.54 (0.35) 10%
RMs total 260 (9) 0.20 [-0.07, 0.48] 1.43 (0.15) 11.15 (0.27) 19%

Power (jumping) 126 (6) -0.14 [-0.49, 0.21] 0.79 (0.43) 2.20 (0.90) 0%
Power (throwing) 109 (4) 0.09 [-0.29, 0.47] 0.45 (0.65) 3.54 (0.47) 0%
Power T otal 235 (11) -0.04 [-0.29, 0.22] 0.27 (0.79) 6.48 (0.84) 0%

Isometric test 113 (4) -0.13 [-0.50, 0.24] 0.69 (0.49) 1.36 (0.85) 0%

Hypertrophy 117 (5) 0.32 [-0.07, 0.71] 1.59 (0.11) 5.60 (0.35) 11%
Note. df = degree of freedom; T P for traditional model of periodization; UP = undulating model of
periodization, RMs = repetitions maximum.
 International Journal of Sports Science 2016, 6(6): 219-229 225

Figure 2. Forest polt of maximum strength tests (IRM tests) presented as subgroup (Bench press, Leg press, Squat, Otheres tests) and pooled.
SD = standard desviation; Std = standardized
226 Leonardo C. Caldas et al.: Traditional vs. Undulating Periodization in the
Context of Muscular Strength and Hypertrophy: A Meta-Analysis

4. Discussion considering the importance of bias control, we also created

strategies to incorporate consolidated and emergent
The main finding of this systematic review followed by databases in order to reach studies published in three
meta-analysis was that UP model was statistically better than different languages.
TP to improve maximum strength performance when pooled A classical concern in meta-analytical studies refers to the
effects sizes were analyzed. The effect size was still heterogeneity [17, 46, 52]. On the other hand, it has also been
enhanced in direction to the UP when only data from studies recognized that heterogeneity is inevitable since the
using equated work were analyzed. In addition, we also methodological diversities are inherent to the studies
found a significant effect size for repetition maximum tests included in any meta-analysis [46]. The chi-squared test has
in favor of UP when we used our own concept of frequently been used to assess heterogeneity [17], with a low
periodization. However, the UP was not significantly better P-value (e.g., < 0.05) suggesting heterogeneity. In our
than TP model regarding to power, muscular endurance, meta-analysis the chi-square was statistically significant for
isometric strength, and muscular hypertrophy. Considering some 1RM analysis; that is, exactly where we have found
that we only found difference to the pooled data, it is possible significant effect size in favoring to UP model. Even
that the small number of data-points to this latter parameters considering that chi-square can be problematic (and
may, at least partly, explain the lack of statistical difference. irrelevant) if many studies are included into a meta-analysis
The UP model has been believed by some researchers to [17], this result may be a concern.
offer a better stimulus-recovery relation through of the Although there is no consensus about the I2 interpretation
strength training as compared to the TP [10, 12, 13, 26]. [53], it has been used to describe the variability caused by
However, it is not consensual. In fact, it was also already heterogeneity [46]. The I2 < 40% is not an indicator of
found that UP induced a worse stimulus-recovery relation significant heterogeneity [17, 53], and we have only found
[15]. It is important to mention that all studies conducted to I2 > 40% for some 1RM tests (leg press; other 1RM tests).
date and included in our meta-analysis were of short-term, Other important concern in meta-analysis refers to the
thus it is not possible to make interpretation about long-term methodological quality of the studies. There is no single
periodization. strategy for this evaluation. The most of methods employed,
The possible mechanisms (e.g., neurophysiologic and/or either qualitative or quantitative, was created for a clinical
morphologic) that could explain why UP was better than TP context [30-33]. However, strength-training interventions
to maximum strength performance do not seem easy to have their own peculiarities. For example, it is very difficult
explain. to carry out a blind or double-blind study in strength training,
None of the studies included in our meta-analysis have but is decisive that strength training progression and
been investigated the potential mechanisms. It is well-known technique be monitored by a specialized professional. In
that the capacity to generate muscle strength are dependent addition, it is clearly possible and essential that researchers
from neural and physiological ways. Apparently, the have concern and make strategies regarding sample
explanation could not be attributed to muscle morphological randomization, clinical condition, withdrawals or dropouts,
alterations, since we have not found any difference (or even concurrent activities, monitoring of diet and equalization of
tendency to difference) regarding to muscle hypertrophy. the work training. As a consequence, the use of previously
Thus, a hypothesis that must be tested is whether UP model employed metrics to gauge methodological quality could be
could induce better neural adaptation as, for example, insufficient in the context of the exercise training. For this
increasing central motor drive and motoneuron excitability, reason, we adapted a qualitative strategy from others
inhibiting presynaptic Ia afferent [50] or even inducing faster [30-33].
neural adaptations [43]. However, this hypothesis will need In our view, the studies included in our meta-analysis had
be addressed in the future. reasonable quality in relation to the randomization strategies,
Effect sizes (or Cohen’s d) of 0.2, 0.5 and 0.8 have been clinical condition, and monitoring of training. The concerns
considered as small, medium, and large, respectively; Of about work matching were also reasonable; however, several
note, this classification has been criticized mainly when the data descriptions necessaries to confirm this point were
95% confidence interval and P-value are not considered [51]. insufficient. The studies had poor information about loss of
In the present meta-analysis the effect sizes (or standardized follow-up and were mainly deficient to masking methods. In
mean deviation; SMD), with P-value lesser than 5%, ranged addition, there were a great number of studies with
0.22 – 0.47 for maximum strength performance, before and concurrent training and missing control to dietary
after sensitivity analyses. These effects sizes denote that the prescription. We did not include information regarding of
mean score of participants in the UP groups would be around drugs use (e.g., steroids) and ergogenic aids in our qualitative
the 58th percentile to 69th percentile in relation to the TP analysis, but it is an important issue in strength training.
group [51]. Nevertheless, no studies included in our meta-analysis
We used a software dedicated exclusively to process described information about drugs use. Consequently, we
meta-analysis and to prevent unrealistic outputs we followed cannot completely dismiss the possibility of these all
the steps advocated by an important worldwide reference in methodological faults aforementioned had biased part of the
systematic review and meta-analysis [17, 52]. In addition, results. Thus, further studies must to consider such issues.
 International Journal of Sports Science 2016, 6(6): 219-229 227

Sensitivity analyses have classically been recommended difference to other variables (power, RMs, and isometric
to test the robustness of meta-analyses findings [54, 55]. In strength tests; hypertrophy).
this regard, one of our concerns was related to the equality of Regarding the studies included by Harries et al. [56], it is
training work between groups inserted in UP and TP. In fact, interesting to mention that they found 1RM bench press data
some researchers have been speculated that different total for the Rhea et al. study [41], while we only found data
training volume could explain the differences of results regarding leg extension in this same study. They also
between UP and TP [11, 21, 35]. Our sensitivity analyses included 1RM bench press data from the Apel et al. study
showed that equalization of total work did not alter the [15], however in this study mentioned we only found data
results. However, the effect size for pooled 1RM tests was from 10RMs, but not 1RM. Unlike us, they included a study
2-fold greater in favor to UP model when only studies (Franchini et al. [49]) whose authors used a reverse linear
matched by total work were analyzed. This result confirmed periodization.
the advantages of the UP model for maximum strength and, In conclusion, our results suggest that UP is better than TP
in the same time, it illustrates that equalization of total work to improve maximum strength performance. The same
must be considered as intervening confounding variable in conclusion cannot be made for power, muscular endurance,
studies investigating periodization. isometric strength, and muscular hypertrophy. However, a
Another theme considered in the sensitivity analyses was number of methodological faults from the studies and the
related to the interpretation of the periodization short-term periodization request some caution before trying
conceptualization. Initially, the UP or TP models were to extrapolate the results.
classified in conformity that stated by the authors. However,
we found similar experimental designs with different
classifications reported by the authors [16, 23, 43], and
sometimes we found designs that could not be easily
classified as UP or TP. Different nomenclatures and
REFERENCES
interpretations regarding the periodization models [11, 14, [1] Fleck SJ. Periodized Strength Training: A Critical Review. J
15, 21, 23, 25, 38, 43] could induce conceptual confusions. Strength Cond Res 1999; 13: 82-89.
Thus, we standardized the UP and TP conceptualizations to [2] Rhea MR, Alderman BL. A meta-analysis of periodized
select and include the studies in our sensitivity analyses. versus nonperiodized strength and power training programs.
Thereafter, we found a larger effect size for maximum Res Q Exerc Sport 2004; 75: 413–422.
strength and muscular endurance tests in favor of UP model. [3] American College of Sports Medicine. American College of
These findings are sufficient to suggest that inappropriate Sports Medicine position stand. Progression models in
conceptualization of periodization may influence the resistance training for healthy adults. Med Sci Sports Exerc
robustness of the results. At the same time, it must encourage 2009; 41: 687–708.
the scientific community to establish conceptual [4] Kiely J. Periodization paradigms in the 21st century:
standardizations regarding the periodization and follow evidence-led or tradition-driven? Int J Sports Physiol Perform
them. 2012; 7: 242–250.
Other important concern is about the short-term nature
[5] Issurin VB. New horizons for the methodology and
from the studies. Indeed, the longest training duration physiology of training periodization. Sports Med 2010; 40:
included in our meta-analysis was 16 weeks. This is rather 189–206.
problematic since periodization should be thought for a
long-term. Thus, it is currently not possible to draw [6] Fleck SJ. Non-linear periodization for general fitness &
athletes. J Hum Kinet 2011; 29A: 41–45.
conclusions regarding the effectiveness of ‘long-term
periodization’. [7] Ratamess N. ACSM’s Foundations of strength training and
Harries, et al. [56] also did a meta-analysis regarding the conditioning. Lippincott. Indianapolis USA, 2012.
TP and UP model. However, there are some important [8] Baechle TR, Earle RW. Essentials of strength training and
differences between our both studies concerning to the conditioning. 3rd ed. Human Kinetics, 2008.
methods and results. We included more studies (47%) than
Harries et al. [56]. They only analyzed data from 1RM tests, [9] Bradley-Popovich GE. Nonlinear Versus Linear
Periodization Models. Strength Cond J 2001; 23: 42-44.
while we also analyzed data from power, muscular
endurance, isometric strength tests, and hypertrophy using [10] Poliquin C. FOOTBALL: Five steps to increasing the
data from robust techniques. Unlike them, we also performed effectiveness of your strength training program. Natl Strength
sensitivity analyses that are particularly important to expand Cond Assoc J 1988; 10: 34-39.
the data interpretation. Harries et al. [56] did not find [11] Rhea MR, Ball SD, Phillips WT, et al. A comparison of linear
differences between TP vs. UP, while we found that the UP and daily undulating periodized programs with equated
was significantly better than TP to improve maximum volume and intensity for strength. J Strength Cond Res 2002;
16: 250–255.
strength performance. The main explanation for this
conflicting result is the larger number of effects sizes that we [12] Simão R, Spineti J, de Salles BF, et al. Comparison Between
have included in our study. Despite it, we did not find Nonlinear and Linear Periodized Resistance Training. J
228 Leonardo C. Caldas et al.: Traditional vs. Undulating Periodization in the
Context of Muscular Strength and Hypertrophy: A Meta-Analysis

Strength Cond Res 2012; 26: 1389–1395. [28] Ullrich B, Holzinger S, Soleimani M, et al. Neuromuscular
Responses to 14 Weeks of Traditional and Daily Undulating
[13] Peterson MD, Dodd DJ, Alvar BA, et al. Undulation Training Resistance Training. Int J Sports Med 2015; 554–562.
for Development of Hierarchical Fitness and Improved
Firefighter Job Performance. J Strength Cond Res 2008; 22: [29] Prestes J, da Cunha Nascimento D, Tibana RA, et al.
1683–1695. Understanding the individual responsiveness to resistance
training periodization. Age (Dordr) 2015; 37: 55.
[14] Painter KB, Haff GG, Ramsey MW, et al. Strength gains:
block versus daily undulating periodization weight training [30] Doi S a. R, Thalib L. A quality-effects model for
among track and field athletes. Int J Sports Physiol Perform meta-analysis. Epidemiology 2008; 19: 94–100.
2012; 7: 161–169.
[31] Jadad AR, Moore RA, Carroll D, et al. Assessing the quality
[15] Apel JM, Lacey RM, Kell RT. A comparison of traditional of reports of randomized clinical trials: Is blinding necessary?
and weekly undulating periodized strength training programs Control Clin Trials 1996; 17: 1–12.
with total volume and intensity equated. J Strength Cond Res
2011; 25: 694–703. [32] Maher CG, Sherrington C, Herbert RD, et al. Reliability of
the PEDro scale for rating quality of randomized controlled
[16] Lima C, Boullosa D, Frollini A, et al. Linear and daily trials. Phys Ther 2003; 83: 713–721.
undulating resistance training periodizations have differential
beneficial effects in young sedentary women. Int J Sports [33] Oxman AD, Guyatt GH. Validation of an index of the quality
Med 2012; 33: 723–727. of review articles. J Clin Epidemiol 1991; 44: 1271–1278.

[17] Higgins JPT, Green S, editors. Cochrane Handbook for [34] Fleck SJ, Kraemer WJ. Fundamentos do treinamento de força
Systematic Reviews of Interventions 4.2.5. The Cochrane muscular. Porto Alegre, Brasil, 2006.
Library, Chichester, UK, 2005.
[35] Baker D, Willson G, Carlyon R. Periodization: The effect on
[18] Egger M, Smith GD. Meta-analysis bias in location and strength of manipulating volume and intensity. J Strength
selection of studies. BMJ 1988; 316: 61–66. Cond Res 1994; 8: 235–242.

[19] Jüni P, Holenstein F, Sterne J, et al. Direction and impact of [36] Bartolomei S, Stout JR, Fukuda DH, et al. Block vs. weekly
language bias in meta-analyses of controlled trials: empirical undulating periodized resistance training programs in women.
study. Int J Epidemiol 2002; 31: 115–123. J Strength Cond Res 2015; 29: 2679–2687.

[20] Kok L. Comparing linear and undulating periodisation for [37] Foschini D, Araújo RC, Bacurau RFP, et al. Treatment of
improving and maintaining muscular strength qualities in obese adolescents: The influence of periodization models and
women. Doctor Thesis, The University of Western Australia, ACE genotype. Obesity 2010; 18: 766–772.
2006. http://research-repository.uwa.edu.au/files/3230528/K
ok_Lian-Yee_2006.pdf (2006). [38] Hoffman JR, Ratamess NA, Klatt M, et al. Comparison
between different off-season resistance training programs in
[21] Kok L-Y, Hamer PW, Bishop DJ. Enhancing muscular Division III American college football players. J Strength
qualities in untrained women. Med Sci Sport Exerc 2009; 41: Cond Res 2009; 23: 11–19.
1797–1807.
[39] Miranda F, Simão R, Rhea M, et al. Effects of linear vs. daily
[22] Fagundes AA, de Barros DC, Duar HA, et al. Food and undulatory periodized resistance training on maximal and
Nutritional Surveillance - Sisvan. Basic Guidance for Data submaximal strength gains. J Strength Cond Res 2011; 25:
Collection, Processing and Analysis and information on 1824–1830.
Health Services. Brasília - DF: Health Miinistry, 2004.
[40] Monteiro AG, Aoki MS, Evangelista AL, et al. Nonlinear
[23] Buford TW, Rossi SJ, Smith DB, et al. A comparison of periodization maximizes strength gains in split resistance
periodization models during nine weeks with equated volume training routines. J Strength Cond Res 2009; 23: 1321–1326.
and intensity for strength. J Strength Cond Res 2007; 21:
1245–1250. [41] Rhea MR, Phillips WT, Burkett LN, et al. A Comparison of
Linear and Daily Undulating Periodized Programs With
[24] Caldwell A. A comparison of linear and daily undulating Equated Volume and Intensity for Local Muscular Endurance.
periodized strength training programs. Master Thesis, J Strength Cond Res 2003; 17: 82-89.
Tennessee State, 2004. http://dc.etsu.edu/etd/954.
[42] Hoffman JR, Wendell M, Cooper J, et al. Comparison
[25] Hartmann H, Bob A, Wirth K, et al. Effects of different between linear and nonlinear in-season training programs in
periodization models on rate of force development and power freshman football players. J Strength Cond Res 2003; 17:
ability of the upper extremity. J Strength Cond Res 2009; 23: 561–565.
1921–1932.
[43] Prestes J, Frollini AB, de Lima C, et al. Comparison between
[26] Spineti J, Figueiredo T, Salles BF de, et al. Comparison linear and daily undulating periodized resistance training to
between different periodization models on muscular strength increase strength. J Strength Cond Res 2009; 23: 2437–2442.
and thickness in a muscle group sequence. Rev Bras Med do
Esporte 2013; 19: 280–286. [44] Marques MC, Casimiro FLM, Marinho DA, et al. Training
and detraining effects on strength parameters in young
[27] Souza EO, Ugrinowitsch C, Tricoli V, et al. Early adaptations volleyball players: volume distribution implications. Motriz
to six weeks of non-periodized and periodized strength Rev Educ Física UNESP 2011; 17: 235–243.
training regimens in recreational males. J Sports Sci Med
2014; 13: 604–609. [45] Review Manager (2012). RevMan (Version 5.2). [Computer
program]. Retrieved from
 International Journal of Sports Science 2016, 6(6): 219-229 229

http://tech.cochrane.org/revman/download. [51] McGough JJ, Faraone S V. Estimating the size of treatment

effects: moving beyond p values. Psychiatry (Edgmont) 2009;
[46] Higgins JPT, Thompson SG, Deeks JJ, et al. Measuring 6: 21–29.
inconsistency in meta-analyses. BMJ 2003; 327: 557–560.
[52] Egger M, Smith GD, Phillips a N. Meta-analysis: principles
[47] Vanni a. C, Meyer F, da Veiga a. DR, et al. Comparison of the and procedures. BMJ 1997; 315: 1533–1537.
effects of two resistance training regimens on muscular and
bone responses in premenopausal women. Osteoporos Int [53] Center for Reviews and Dissemination (CRD). Systematic
2010; 21: 1537–1544. reviews: CRD’s guidance for undertaking reviews in health
care. University of York, 2009.
[48] Inoue DS, De Mello MT, Foschini D, et al. Linear and
undulating periodized strength plus aerobic training promote [54] Egger M, Smith GD. Misleading meta-analysis. BMJ 1995;
similar benefits and lead to improvement of insulin resistance 310: 753–754.
on obese adolescents. J Diabetes Complications 2015; 29:
258–264. [55] Sutton a J, Song F, Gilbody SM, et al. Modelling publication
bias in meta-analysis: a review. Stat Methods Med Res 2000;
[49] Franchini E, Branco BM, Agostinho MF, et al. Influence of 9: 421–445.
linear and undulating strength periodization on physical
fitness, physiological, and performance responses to [56] Harries SK, Lubans DR, Callister R. Systematic review and
simulated judo matches. J Strength Cond Res 2015; 29: meta-analysis of linear and undulating periodized resistance
358–367. training programs on muscular strength. J Strength Cond Res
2015; 29: 1113–1125.
[50] Aagaard P, Simonsen EB, Andersen JL, et al. Neural
adaptation to resistance training: changes in evoked V-wave
and H-reflex responses. J Appl Physiol 2002; 92: 2309–18.