Scand J Med Sci Sports 2013: ••: ••–•• © 2013 John Wiley & Sons A/S.

doi: 10.1111/sms.12047 Published by Blackwell Publishing Ltd

Review

Effect of training cessation on muscular performance: A

meta-analysis
L. Bosquet1,2,3, N. Berryman1,2,3, O. Dupuy1,2,3, S. Mekary2,3, D. Arvisais4, L. Bherer3, I. Mujika5
1
Faculty of Sport Sciences, Laboratoire MOVE (EA 6413), University of Poitiers, Poitiers, France, 2Department of Kinesiology,
University of Montreal, Montreal, Quebec, Canada, 3Montreal Geriatric Institute, Laboratoire LESCA, Montreal, Quebec, Canada,
4
Direction of Libraries, University of Montreal, Montreal, Quebec, Canada, 5Faculty of Medicine and Odontology, Department of
Physiology, University of the Basque Country, Leioa, Basque Country, Spain
Corresponding author: Laurent Bosquet, Faculty of Sport Sciences, Laboratoire MOVE (EA 3813), University of Poitiers, 8, Jean
Monnet Road, 86000 Poitiers – France. Tel: +33 0 549 453 340, Fax: +33 0 549 453 396, E-mail: laurent.bosquet@univ-poitiers.fr
Accepted for publication 13 December 2012

The purpose of this study was to assess the effect of resis- CI) = -0.46 (-0.54 to -0.37), P < 0.01], [maximal power;
tance training cessation on strength performance through SMD (95% CI) = -0.20 (-0.28 to -0.13), P < 0.01]. A dose–
a meta-analysis. Seven databases were searched from response relationship between the amplitude of SMD and
which 103 of 284 potential studies met inclusion criteria. the duration of training cessation was identified. The
Training status, sex, age, and the duration of training effect of resistance training cessation was found to be
cessation were used as moderators. Standardized mean larger in older people (> 65 years old). The effect was also
difference (SMD) in muscular performance was calcu- larger in inactive people for maximal force and maximal
lated and weighted by the inverse of variance to calculate power when compared with recreational athletes. Resis-
an overall effect and its 95% confidence interval (CI). tance training cessation decreases all components of mus-
Results indicated a detrimental effect of resistance train- cular strength. The magnitude of the effect differs
ing cessation on all components of muscular perfor- according to training status, age or the duration of train-
mance: [submaximal strength; SMD (95% CI) = -0.62 ing cessation.
(-0.80 to -0.45), P < 0.01], [maximal force; SMD (95%

Muscular strength is a major determinant of sport per- appears that, even if recommendations are clear regard-
formance, both in explosive (Delecluse, 1997) and long- ing the beneficial effects of strength training, adherence
duration events (Saunders et al., 2004), as well as an to those programs are still a challenge (Andersen, 2011).
important contributor to functional performance and Identifying the kinetics of strength loss once resistance
health in older adults (Moreland et al., 2004; Hurley training ceases is important to design successful tapers
et al., 2011). The capacity of the skeletal muscle to gen- and return to optimal fitness for competitive athletes, and
erate a high level of force is a complex interplay between more generally for the individualization of exercise
several factors, including muscle fiber type (Gollnick & training prescriptions whatever the characteristics of the
Matoba, 1984), muscle cross-sectional area (Jones et al., population. The literature examining this issue is very
2008), muscle architecture (Aagaard et al., 2001), and heterogeneous in terms of training/training cessation
neural drive to the muscle (Gandevia, 2001). Resistance characteristics, muscular strength tests and measures and
training is a safe and effective intervention to improve population characteristics. Although there is consensus
these determinants and increase muscular strength, among narrative reviews that training cessation leads
whatever age and sex (Falk & Tenenbaum, 1996; Latham more or less rapidly to detraining (Mujika & Padilla,
et al., 2004; Ratamess et al., 2009). However, training- 2000a,b, 2001c,d), methodological heterogeneity does
induced adaptations are transitory and may disappear not allow to make direct comparisons between studies or
when the training stimulus is withdrawn, thus leading to specify the overall detraining effect according to sex,
to detraining. Detraining has been defined as the partial age, training status, or other relevant variables such as
or complete loss of training-induced anatomical, physi- the duration of training cessation.
ological, and functional adaptations, as a consequence The aim of this study was therefore to assess the
of training cessation (Mujika & Padilla, 2000a). The effects of complete resistance training cessation on
reasons for such a scenario are numerous in an individu- the different expressions of muscular strength, including
al’s life, e.g., illness, injury, travel, loss of motivation, maximal force, maximal power, and submaximal
or post-season break in competitive athletes. Also, it strength, through a meta-analysis of the available

1
Bosquet et al.

Fig. 1. Flow chart of the study selection process.

literature. We also carried out exploratory subgroup included study using the following moderators: training status
analyses to determine whether population characteristics (competitive athletes, recreational athletes, or inactive people), sex
(male, female, both), age (<65 years old, !65 years old), limb
or training/training cessation characteristics were out- (upper, lower), duration of training and training cessation, and type
comes that may influence the magnitude of the effect. of muscular performance (maximal force, maximal power, sub-
maximal strength). Measures of maximal force included 1 to 5 RM
Material and methods during isoinertial contractions [constant weight lifted at a volun-
tary speed (Verdijk et al., 2009)], peak torque during isometric
Literature search strategy
dynamometry, and peak torque during isokinetic dynamometry at
The databases EBM reviews/CCRCT (1991 to 4th quarter 2011), 30 to 60°/s. Measures of maximal power included vertical jump
Embase (1980 to 2011 weeks 50) Kinpubs (1947 to 2011), Physi- height, sprint performance, peak power during a force–velocity
cal Education Index (1970 to 2011), PubMed (1950 to 2011), test, and peak torque during isokinetic dynamometry at 120 to
SportDiscus (1830 to 2011), and Web of Science (1970 to 2011) 240°/s. Measures of submaximal strength included 6 to 12 RM
were searched using the terms (detraining OR deconditioning OR during isoinertial contractions, time to exhaustion during isometric
training cessation) AND [(one repetition maximum OR 1 RM OR dynamometry, and total work during an isokinetic fatigue test. An
max$ strength OR max$ force) OR (power OR jump$ OR force- interval scale was used for the coding of performance and duration
velocity) OR (muscular endurance OR RM)] for English-language of training and training cessation, while a nominal scale was used
and French-language randomized controlled trials, crossover for the coding of the other moderators. The duration of training
trials, repeated-measure studies, theses, and dissertations. The ref- cessation was a posteriori divided in seven categories: <7 days, 8
erence lists of the articles obtained were searched manually to to 14 days, 15 to 28 days, 29 to 56 days, 57 to 112 days, 113 to 224
obtain further studies not identified electronically. This led to the days, and >224 days. Any disagreement between both reviewers
identification of 284 potential studies for inclusion in the analysis was discussed in a consensus meeting, and unresolved items were
(Fig. 1). taken to a third reviewer (S. M.) for resolution.

Selection criteria
Studies were eligible for inclusion if (a) they implemented a train- Statistical analysis
ing intervention followed by a training cessation period and gave Standardized mean differences (SMDs) for each study were cal-
relevant details about the procedures, including the type and dura- culated using Hedges’ g (Hedges 1981). In the studies that used
tion of training as well as the duration of training cessation; (b) the multiple measures of muscular performance, a single composite
outcome included valid tests and measures of the upper or lower SMD was calculated (Borenstein et al., 2009). Considering that
limb muscular performance in healthy humans; and (c) the paper the effect of training cessation on muscular performance may
reported the number of participants and all the necessary data to differ according to the training status, age or other moderators, we
calculate effect sizes. Studies were excluded if they presented a priori decided to use a random-effects model. Standardized
results reported in a previous publication. mean differences were weighted by the inverse of variance to
calculate an overall effect and its 95% confidence interval (CI).
Cohen’s criteria were used to interpret the magnitude of SMD:
Coding for the studies <0.5, small; 0.5 to 0.8, moderate; and >0.8, large (Cohen, 1988).
Two independent reviewers who were blind to authors, affiliations, Statistical heterogeneity, which refers to the percentage of the
and the publishing journal (N. B. and O. D.) read and coded each variability between studies that is due to clinical and methodologi-

2
 Training cessation and strength performance
cal heterogeneity rather than the sampling error, was assessed by Table 1. Effect of training cessation on maximal force according to popu-
the I2 statistic (Borenstein et al., 2009). According to Higgins et al. lation characteristics and limb
(2003), I2 values of 25%, 50%, and 75% represent low, medium,
and high heterogeneity, respectively. The presence of medium or Moderator SMD* 95% CI z P
high heterogeneity may provoke further investigation through sub-
group analysis of moderator variables, even if the overall effect is Age
considered nonsignificant (Higgins & Green, 2006). Weighted <65 years old -0.31 -0.40 to -0.21 -6.40 <0.01
SMDs and standard errors were then calculated for each category !65 years old -0.76† -0.90 to -0.62 -10.89 <0.01
within moderator variables, as well as 95% CIs to determine Sex
Males -0.49 -0.61 to -0.38 -8.45 <0.01
whether each SMD was different from 0. A Q-test based on the
Females -0.45 -0.70 to -0.19 -3.45 <0.01
analysis of variance was performed to test the null hypothesis that
Training status
the effect of training cessation was similar between the categories Inactive people -0.55‡ -0.65 to -0.44 -10.26 <0.01
of a moderator variable (Borenstein et al., 2009). When the null Recreational athletes -0.31 -0.45 to -0.16 -4.18 <0.01
hypothesis was rejected, pairwise comparisons were performed Competitive athletes -0.29 -0.69 to 0.11 -1.41 0.16
with a Z-test, the alpha value for significance being adjusted Limb
according to the procedure of Bonferroni (Vincent & Weir, 2012). Upper -0.81 -1.06 to -0.56 -6.29 <0.01
The criterion level for significance was set at P < 0.016 (i.e., Lower -0.88 -1.14 to -0.63 -6.77 <0.01
0.05/3) for training status, which corresponds to a z-score higher
than 2.41, and at P < 0.0024 (i.e., 0.05/21) for the duration of *SMD: < 0.5, small; 0.5 to 0.8, moderate; and > 0.8, large.
training cessation, which corresponds to a z-score higher than †
Different from < 65 years old (P < 0.01).
3.04. Finally, a metaregression was performed to assess the rela- ‡
Different from recreational athletes (P < 0.01).
tionship between the duration of training cessation and muscular
SMD, standardized mean difference; CI, confidence interval.
performance. All calculations were made with Comprehensive
Meta-analysis (http://www.meta-analysis.com).
Table 2. Effect of training cessation on maximal power according to popu-
lation characteristics and limb
Results
Moderator SMD* 95% CI z P
Overall results
The literature search allowed to identify 284 potentially Age
relevant publications spanning from 1956 to 2011, of <65 years old -0.18 -0.26 to -0.10 -4.46 <0.01
!65 years old -0.46† -0.72 to -0.21 -3.55 <0.01
which 103 met all inclusion criteria. The most common Sex
reasons for exclusion were (a) the presence of pathologi- Males -0.13 -0.23 to -0.03 -2.49 <0.01
cal populations, (b) the absence of training/training ces- Females -0.28 -0.46 to -0.09 -2.95 <0.01
Training status
sation interventions, (c) the absence of upper or lower Inactive people -0.34‡ -0.47 to -0.22 -5.32 <0.01
limb muscular performance assessment, and (d) the lack Recreational athletes -0.09 -0.21 to 0.02 -1.64 0.10
of adequate information to calculate SMDs. The overall Competitive athletes -0.20 -0.38 to -0.02 -2.15 <0.05
SMD indicated a detrimental effect of training cessation Limb
Upper -0.37 -0.61 to -0.14 -3.15 <0.01
on all components of muscular performance, since we Lower -0.26 -0.48 to -0.05 -2.44 <0.01
found a moderate decrease in submaximal strength
[SMD (95% CI) = -0.62 (-0.80 to -0.45), P < 0.01, *SMD: < 0.5, small; 0.5 to 0.8, moderate; and > 0.8, large.
I2 = 33.0%] and a small decrease in maximal force †
Different from < 65 years old (P < 0.01).
[SMD (95% CI) = -0.46 (-0.54 to -0.37), P < 0.01, ‡
Different from recreational athletes (P < 0.01).
I2 = 75.6%], and maximal power [SMD (95% SMD, standardized mean difference; CI, confidence interval.
CI) = -0.20 (-0.28 to -0.13), P < 0.01, I2 = 69.9%]. The
presence of medium to large statistical heterogeneity for of training cessation was also larger in inactive people
maximal power and maximal force justified the sub- for maximal force (z = 2.67, adjusted P < 0.05) and
group analysis of moderator variables. Similar analyses maximal power (z = 2.99, adjusted P < 0.05) when com-
were performed for submaximal strength in an explor- pared with recreational athletes, but not for submaximal
atory manner, given that I2 was less than 50%. strength (z "1.19, adjusted P > 0.05). Finally, we did not
find any difference between males and females, or
between upper and lower limb, whatever the type of
Moderating variables: population characteristics and muscular performance (z "1.40, P > 0.05).
limb
The potential effect of population characteristics and
limb on the magnitude of the decrease in maximal force, Moderating variables: training/training cessation
maximal power, and submaximal strength is presented in characteristics
Tables 1, 2, and 3, respectively. Regarding the type of training performed before the
The effect of training cessation was found to be larger training cessation, it has to be mentioned that out of the
in older people (!65 years old) for maximal force 103 studies included in this meta-analysis, only 19 pro-
(z = 5.38, P < 0.01), maximal power (z = 2.03, P < 0.05), posed a strength training protocol that was not based
and submaximal strength (z = 2.00, P < 0.05). The effect on submaximal/hypertrophy prescription guidelines. In

3
Bosquet et al.
Table 3. Effect of training cessation on submaximal strength according to tionship is visually evident, the difference between cat-
population characteristics. egories was significant only for durations longer than 8
Moderator SMD* 95% CI z P (submaximal strength) or 16 weeks (maximal force and
maximal power).
Age
<65 years old -0.48 -0.70 to -0.26 -4.27 <0.01
!65 years old -0.85† -1.13 to -0.57 -5.88 <0.01 Discussion
Sex
Males -0.61 -0.89 to -0.32 -4.11 <0.01 The aim of this study was to assess the effect of training
Females -0.68 -1.07 to -0.29 -3.42 <0.01 cessation on maximal force, maximal power, and sub-
Training status
Inactive people -0.61 -0.81 to -0.41 -6.08 <0.01
maximal strength through a systematic review of the
Recreational athletes -0.79 -1.24 to -0.34 -3.45 <0.01 literature and meta-analysis. We found a moderate
Competitive athletes -0.16 -1.10 to 0.79 -0.32 0.75 decrease in submaximal strength and a small decrease in
Limb maximal force and maximal power. This detrimental
Upper -0.77 -1.08 to -0.46 -4.86 <0.01
Lower -0.83 -1.14 to -0.52 -5.26 <0.01
effect was found to differ according to the duration of
training cessation, age, and training status, but was
*SMD: < 0.5, small; 0.5 to 0.8, moderate; and > 0.8, large. not influenced by sex or the characteristics of previous
†
Different from < 65 years old (P < 0.05). training.
SMD, standardized mean difference; CI, confidence interval.

other words, most of the training interventions were Effect of training cessation on maximal force
similar with regards to the training programs/objectives. Maximal force represents the peak force or peak torque
For these 19 other studies, different training methods reached during a maximal voluntary contraction. It is
were used (plyometrics, maximal strength, electrostimu- often considered fundamental for both athletic perfor-
lation, vibration). It also has to be mentioned that 24 mance and a healthy lifestyle (Abernethy et al., 1995;
studies out of the 103 implemented a training program Kraemer et al., 2002). Overall SMD revealed a small
using a combination of methods (hypertrophy, maximal decrease in maximal force once training ceases. It is
force, and power development). Even though there is a worth noting that this decrement grew with the duration
rationale to analyse training cessation effects separately of training cessation. As shown in Fig. 2, the decrease in
based on each specific training intervention, training maximal force became significant from the third week of
programs were not included as a moderator. This inactivity, and its magnitude increased thereafter as a
decision was based upon the fact that a great majority function of time. Many physiological factors may be
of studies presented similar training intervention involved in this process. They are typically classified as
(submaximal/hypertrophy). Also, a minority of studies central (or neural) and peripheral (or morphological)
used significantly different training prescriptions making factors. Central factors refer to motor unit recruitment
it statistically irrelevant to consider all these variables as and synchronization, firing frequency, and intermuscular
separate moderators. In contrast, the duration of training coordination (Cormie et al., 2011a). Central adaptations
programs could differ widely between studies. We per- occur rapidly with training and are thought to explain the
formed a meta-regression analysis that did not reveal any greatest part of short-term strength gains in previously
relationship between the exact duration of training and untrained individuals (Moritani & deVries, 1979; Hak-
the magnitude of the effect of training cessation on mus- kinen, 1989; Folland & Williams, 2007). Peripheral
cular performance, as the slope was not different from 0. factors refer to muscle fiber type and architecture, as
In contrast, the slope of the relationship between the well as tendon properties (Cormie et al., 2011a).
magnitude of the effect of training cessation and the Although the cellular adaptations that subtend muscle
exact duration of training cessation was significantly hypertrophy may occur early in a training program
different from 0 (z !2.27, P < 0.05), whatever the type (DeFreitas et al., 2011), it is generally considered that
of muscular performance, thus suggesting a close the relative contribution of morphological adaptations
association between both variables. Exact duration of increases gradually as training proceeds (Narici et al.,
training cessation was a posteriori divided in seven cat- 1996; Folland & Williams, 2007), with an increasing
egories. The Q-test based on the analysis of variance role of the endocrine system (Crewther et al., 2006,
allowed us to reject the null hypothesis that the effect of 2011). Although it was beyond the scope of this meta-
training cessation was similar between these categories, analysis to study specifically these underlying factors,
whatever the type of muscular performance (P < 0.05). one may hypothesize that this sequence of events also
Weighted SMDs and significant pairwise comparisons exists in the disadaptation process, the factors underpin-
are presented in Figs 2–4. The effect of training cessa- ning the continuous decrease in maximal force being
tion became statistically significant between the third mainly central during the first weeks of training cessa-
and fourth week for maximal force, maximal power, and tion, and mainly peripheral afterwards. This hypothesis
submaximal strength. Although a dose–response rela- is in accordance with the data published by the group of

4
 Training cessation and strength performance

Fig. 2. Dose–response curve for the effect of the duration of training cessation on maximal force. (a) Different from standardized mean
differences computed for " 112 days of training cessation.

Fig. 3. Dose–response curve for the effect of the duration of training cessation on maximal power. (a) Different from the standardized
mean difference computed for " 14 days of training cessation.

Häkkinen (Hakkinen & Komi, 1983; Hakkinen et al., lar strength. Maximal power represents the ability to
2000), who reported a rapid decrease (of small ampli- produce high amounts of force over a short period of
tude) in the neural activation once training ceases, fol- time, and plays a crucial role in many athletic events
lowed by a muscular atrophy when this period of (Mero et al., 1992; Duthie et al., 2003; Stolen et al.,
inactivity exceeds several weeks. 2005). Considering that training cessation results in
a significant reduction of maximal force, it would be
expected to reduce maximal power as well. However,
Effect of training cessation on maximal power maximal power is also determined by factors related
Although they are often used as synonymous, maximal to velocity that are independent from maximal force
force and maximal power are different facets of muscu- (Kraemer et al., 2002; Cormie et al., 2011b). Therefore,

5
Bosquet et al.

Fig. 4. Dose–response curve for the effect of the duration of training cessation on submaximal strength. (a) Different from the
standardized mean difference computed for " 7 days of training cessation. (b) Different from standardized mean differences computed
for " 112 days of training cessation.

depending on the effect of training cessation on these contribute to explain the difference in the effect of train-
factors, the rates of decline of maximal power and ing cessation on maximal force and maximal power, this
maximal force are not necessarily the same. Indeed, overshoot of IIX muscle fibers is probably central since
our meta-analysis showed that the magnitude of the the resulting increase in maximal velocity may compen-
effect of training cessation on maximal power was sate for the loss in maximal force to maintain maximal
smaller than that observed for maximal force. This dif- power.
ference between both muscular properties concerned
overall SMD, but also the kinetics of the disadaptation
process. As shown in Figs 2 and 3, the effect of training Effect of training cessation on submaximal strength
cessation on maximal force and maximal power was Submaximal strength represents the ability of the neuro-
quite similar during the first weeks, but although an muscular system to sustain a high fraction of maximal
improvement may be expected in maximal power after force for a long period of time or a high number of
short-term training cessation (i.e., 2 weeks or less) in repetitions. This specific ability is particularly important
relation with recovery from training-induced neuromus- in the maintenance of autonomy in older adults (Hunter
cular fatigue, this was less probable in maximal force. et al., 2004), but also in many long-duration sporting
However, there appears to be dissociation after 16 weeks events such as cycling or triathlon (Marcora et al., 2008).
of inactivity since we found a large decrease in maximal We found a moderate decrease in submaximal strength
force while maximal power was not different from the once training ceases. The negative impact of exercise
previous weeks. Andersen and Aagaard (2000) observed cessation duration on submaximal strength was bigger
a decrease in the proportion of IIb muscle fibers in the than on maximal force and maximal power. Physiologi-
vastus lateralis of healthy young males after a 3-month cal factors related to oxygen transport and energy pro-
training period. Of all muscle fibers, type IIb represented duction should be added to the neural and morphological
10.2 # 2.5% at pretraining measurement time. After 38 factors previously discussed to explain the detrimental
resistance training sessions within a 90-day period, this effect of training cessation on muscular force and power.
proportion decreased to 4.1 # 1.2%. Surprisingly, this The rapid decrease in blood volume that is observed very
proportion increased to 18.8 # 3.5% after 3 months of shortly once training ceases (Houmard et al., 1992) is the
training cessation. Andersen et al. (2005) later showed starting point of a cascade of events leading to a decrease
that this detraining-induced overshoot in IIX muscle in cardiac output (Coyle et al., 1984, 1985). Training
fiber proportion was accompanied by an increase in the cessation is also associated with a greater reliance on
electrically evoked twitch rate of force development, and glucose for energy provision that is concomitant with a
in the maximal unloaded knee extension velocity and rapid decrease in muscle glycogen stores (Costill et al.,
power, while cross-sectional area and peak torque 1985; Mikines et al., 1989) and a rapid decrease in the
decreased to baseline level. Although other factors may activity of oxidative enzymes such as citrate synthase,

6
 Training cessation and strength performance
succinate dehydrogenase and malate dehydrogenase larger muscle mass in males since the force to cross-
(Coyle et al., 1984, 1985). All together, these disadapta- sectional area ratio, the number of muscle fibers, and the
tions clearly compromise oxygen transport, aerobic characteristics of motor units are not different between
energy production, and submaximal strength. Through males and females (Miller et al., 1993). Interestingly,
an additive effect to neural and morphological disadap- some data suggest that training-induced improvement in
tations, they probably contribute to the larger decrease maximal force mainly depends on muscular hypertrophy
we found in submaximal strength in comparison with in males, and nonmuscular (possibly neural) adaptations
maximal force and maximal power. in females (Hakkinen et al., 2001; Delmonico et al.,
2005). Considering this sex specificity in the adaptation
to resistance training, the question of a sex specificity in
Moderating variables the response to training cessation deserves attention. As
Senescence induces both neural and morphological shown in Tables 1–3, we did not find any difference in
changes that have a detrimental effect on muscular the magnitude of decrease in maximal force, maximal
strength (Manini & Clark, 2012). In fact, maximal force power, and submaximal strength between males and
and maximal power have been shown to decrease from females. Although the relative weight of central and
the fourth decade by approximately 2% and 4%, respec- peripheral factors probably differs between males
tively (Bosco & Komi, 1980; Bassey et al., 1992; and females, the effect of training cessation on muscular
Skelton et al., 1994; Phillips, 2007). This age-related strength is similar.
muscle weakness, also called dynapenia (Manini & An important issue when we aim at assessing the
Clark, 2012), has been associated with an increased risk effect of training cessation on muscular strength is the
of falls (Moreland et al., 2004) and with adverse physi- dose of physical activity that will be maintained by
ological changes that may predispose elderly people to the participants in the duration of the training cessation
osteoporosis, atherosclerosis, diabetes, and other chronic period. In fact, depending on the duration, intensity, and
diseases and functional limitations (Hyatt et al., 1990). frequency of this physical activity, the stimulus could be
Strength training has been shown to be an effective inter- high enough to maintain training-induced neural and
vention to counteract these adverse effects (Hurley et al., morphological adaptations. In this sense, if we consider
2011). However, considering the dynapenia phenom- that legs are used in a greater extent than arms in daily
enon (Manini & Clark, 2012), it could be argued that physical activity (walking, stair climbing, cycling, and
older adults are more vulnerable to the detrimental so on), one could hypothesize that the magnitude of
effects associated with strength training cessation. In this decrease in muscular strength when training ceases is
study, we arbitrarily set the limit between adults and larger for the upper limb when compared with the lower
seniors at 65 years old. As shown in Tables 1–3, we limb. Contrarily to this hypothesis, we did not found any
actually found a larger magnitude of decrease in older effect of limb, whatever the component of muscular
people, whatever the expression of muscular strength strength, thus suggesting that daily physical activity does
(i.e., maximal force, maximal power or submaximal not reach the level required to maintain training-induced
strength). The mechanisms underlying this larger adaptations when the duration of training cessation
decrease are probably a combination of neural and mor- exceeds a given level.
phological factors. The difficulty to maintain muscle As discussed before, there is a time sequence in the
mass is probably involved (Goodpaster et al., 2006), but adaptation to strength training. Neural adaptations,
the relative weight of central factors is certainly more which refer to motor unit recruitment, firing frequency,
important than usually thought (Manini & Clark, 2012). and intermuscular coordination (Cormie et al., 2011a)
The larger rate of decline of maximal power after the are thought to explain the greatest part of short-term
fourth decade when compared with maximal force strength gains (Moritani & deVries, 1979; Hakkinen,
(Bosco & Komi, 1980; Bassey et al., 1992; Skelton 1989; Narici et al., 1996; Folland & Williams, 2007),
et al., 1994; Phillips, 2007) goes in this sense. Part of the while morphological adaptations, which refer to muscle
larger training cessation effect in the older population fiber type and architecture, as well as tendon properties
could also be related to a more sedentary lifestyle. Alto- (Cormie et al., 2011a) are thought to explain the greatest
gether, these results underscore the importance of fol- part of long-term strength gains (Moritani & deVries,
lowing a regular and uninterrupted strength training 1979; Hakkinen, 1989; Narici et al., 1996; Folland &
program in elderly people. The larger decrease in mus- Williams, 2007). Training status, which is closely linked
cular strength when training ceases, associated to a to training history, directly determines the type of adap-
decreased adaptation capacity when compared with tations that subtends strength gains, and probably the
healthy adults (Staron et al., 1990; Charette et al., 1991) speed of reversibility. In fact, it is reasonable to think
may accelerate dynapenia and functional limitation. that adaptations induced by an 8- to 12-week training
Females generally have lower muscular strength than program in a previously untrained individual will disap-
males (Miller et al., 1993; Martel et al., 2006). The pear more rapidly than adaptations obtained after several
greatest part of this sex difference is attributable to a months to several years of training in recreational or

7
Bosquet et al.
competitive athletes. In line with this assertion, we found ness of the analyses. Training-induced adaptations
a larger decrease in maximal force and maximal power depend on a number of moderators and their interaction.
in previously inactive people when compared with rec- Some of them have been coded in this study, such as age,
reational athletes. Surprisingly, we found no difference training status, or training characteristics. However, it
with competitive athletes. One of the main reasons is was not possible to address the interactions between
probably the complexity of the training stimulus and its these moderators, although it may well be the corner-
corollary, the adaptation process. In fact, most competi- stone of success, particularly in competitive athletes.
tive athletes are using a block periodization that plans an Also, in a training cessation protocol with humans, it
alternance between training methods. Contrarily to pre- is very difficult to control the intensity/volume/type of
viously inactive people and to an important proportion of physical activity performed by the participants during
recreational athletes, the type of adaptation (i.e., central the training cessation period. Even though clear instruc-
vs peripheral) is mainly a consequence of the training tions were given to the participants and reported by the
methods and their periodization rather than training authors to avoid any form of resistance training during
experience. It should also be kept in mind that the rela- the training cessation protocol, it is not impossible that
tive weight of strength training in the overall training some participants chose to ignore these recommenda-
load is less important for athletes than inactive people tions for different reasons. As mentioned in the previous
since they have many other technical-tactical or condi- paragraph, this limitation is probably more important
tioning sessions in their training plan. The maintenance with an athlete population.
of a high physical activity level despite the cessation of
strength training probably accounts for the difference we
found with inactive people. A meta-analysis such as that Perspective
performed in this study does not provide the precision The purpose of this investigation was to assess the
required to address these specific issues. However, it effects of training cessation on the different expressions
provides a conceptual framework that may be useful to of muscular strength, including maximal force, maximal
design successful tapers since the knowledge of strength power, and submaximal strength, by means of a system-
loss kinetics allows to plan more precisely the moment atic review of the literature and a meta-analysis. We
when the resistance training load should be decreased to found a moderate decrease in submaximal strength and a
peak for a given competition. small decrease in maximal force and maximal power.
This detrimental effect was found to differ according to
Limits the duration of training cessation, age, and training
status, but was not influenced by sex, limb, or the char-
The meta-analysis methodology allows to quantify the
acteristics of previous training. This meta-analysis pro-
size of effects across a number of independent empirical
vides a framework that can be useful for the optimization
studies while simultaneously eliminating inherent biases
of taper strategies and return to fitness in competitive
in the research (Hagger, 2006). This does not mean
athletes, and more generally for exercise prescription in
however that it is free from bias. Publication and, to a
the general population.
lesser extent, language restriction bias are expected to
inflate estimates of the effect (Moher et al., 1999). Care
Key words: detraining, maximal force, maximal power,
was therefore taken to control these sources of bias as far
submaximal strength, aging.
as possible. Three databases we used in our literature
search (Kinpubs, Physical Education Index, and Sport-
Discus) covered theses and dissertations, thus allowing Funding
the access to this “gray literature” (i.e., literature that is
difficult to locate or retrieve; Moher et al., 1999). Our No funding from any organization was received for this
literature search was restricted to English- and French- work.
languages studies. Nevertheless, with the exception of a
paper published in Japanese but with an English
abstract,(Tsuyama et al., 2005), we did not find addi- Supporting Information
tional relevant reports when extending our search to
Additional Supporting Information may be found in the
studies in all languages (with the Web of Science data-
online version of this article:
base). Some limitations that were specific to the topic of
this meta-analysis have probably restricted the thorough- References included in the meta-analysis.

8
 Training cessation and strength performance
References
Aagaard P, Andersen JL, Dyhre-Poulsen Med Sci Sports Exerc 1985: 17: Visser M, Newman AB. The loss of
P, Leffers AM, Wagner A, Magnusson 339–343. skeletal muscle strength, mass, and
SP, Halkjaer-Kristensen J, Simonsen Coyle EF, Martin WH, Bloomfield SA, quality in older adults: the health, aging
EB. A mechanism for increased Lowry OH, Holloszy JO. Effects of and body composition study. J Gerontol
contractile strength of human pennate detraining on responses to submaximal A Biol Sci Med Sci 2006: 61:
muscle in response to strength training: exercise. J Appl Physiol 1985: 59: 1059–1064.
changes in muscle architecture. J 853–859. Hagger M. Meta analysis in sport and
Physiol 2001: 534: 613–623. Coyle EF, Martin WH, Sinacore DR, exercise research: review, recent
Abernethy P, Wilson G, Logan P. Joyner MJ, Hagberg JM, Holloszy JO. developments, and recommandations.
Strength and power assessment. Issues, Time course of loss of adaptations after Eur J Sport Sci 2006: 6: 103–115.
controversies and challenges. Sports stopping prolonged intense endurance Hakkinen K. Neuromuscular and
Med 1995: 19: 401–417. training. J Appl Physiol 1984: 57: hormonal adaptations during strength
Andersen JL, Aagaard P. Myosin heavy 1857–1864. and power training. A review. J Sports
chain IIX overshoot in human skeletal Crewther B, Cronin J, Keogh J. Possible Med Phys Fitness 1989: 29: 9–26.
muscle. Muscle Nerve 2000: 23: stimuli for strength and power Hakkinen K, Alen M, Kallinen M,
1095–1104. adaptations: acute metabolic responses. Newton RU, Kraemer WJ.
Andersen LL. Influence of psychosocial Sports Med 2006: 36: 65–78. Neuromuscular adaptation during
work environment on adherence to Crewther BT, Cook C, Cardinale M, prolonged strength training, detraining
workplace exercise. J Occup Environ Weatherby RP, Lowe T. Two emerging and re-strength-training in middle-aged
Med 2011: 53: 182–184. concepts for elite athletes: the and elderly people. Eur J Appl Physiol
Andersen LL, Andersen JL, Magnusson short-term effects of testosterone and 2000: 83: 51–62.
SP, Suetta C, Madsen JL, Christensen cortisol on the neuromuscular system Hakkinen K, Komi PV.
LR, Aagaard P. Changes in the human and the dose-response training role of Electromyographic changes during
muscle force-velocity relationship in these endogenous hormones. Sports strength training and detraining.
response to resistance training and Med 2011: 41: 103–123. Med Sci Sports Exerc 1983: 15:
subsequent detraining. J Appl Physiol DeFreitas JM, Beck TW, Stock MS, 455–460.
2005: 99: 87–94. Dillon MA, Kasishke PR 2nd. An Hakkinen K, Kraemer WJ, Newton RU,
Bassey EJ, Fiatarone MA, O’Neill EF, examination of the time course of Alen M. Changes in electromyographic
Kelly M, Evans WJ, Lipsitz LA. Leg training-induced skeletal muscle activity, muscle fibre and force
extensor power and functional hypertrophy. Eur J Appl Physiol 2011: production characteristics during heavy
performance in very old men and 111: 2785–2790. resistance/power strength training in
women. Clin Sci (Lond) 1992: 82: Delecluse C. Influence of strength training middle-aged and older men and
321–327. on sprint running performance. Sports women. Acta Physiol Scand 2001: 171:
Borenstein M, Hedges L, Higgins J, Med 1997: 24: 147–156. 51–62.
Rothstein H. Introduction to Delmonico MJ, Kostek MC, Doldo NA, Hedges L. Distribution theory for Glass’s
meta-analysis. Chichester, WS: John Hand BD, Bailey JA, Rabon-Stith KM, estimator of effect size and related
Wiley and Sons, 2009: 1–422. Conway JM, Carignan CR, Lang J, estimators. J Educ Stat 1981: 6:
Bosco C, Komi PV. Influence of aging on Hurley BF. Effects of 107–128.
the mechanical behavior of leg extensor moderate-velocity strength training on Higgins J, Green S. Cochrane handbook
muscles. Eur J Appl Physiol Occup peak muscle power and movement for systematic reviews of interventions.
Physiol 1980: 45: 209–219. velocity: do women respond differently Oxford: The Cochrane Collaboration,
Charette SL, McEvoy L, Pyka G, than men? J Appl Physiol 2005: 99: 2006.
Snow-Harter C, Guido D, Wiswell RA, 1712–1718. Higgins JP, Thompson SG, Deeks JJ,
Marcus R. Muscle hypertrophy Duthie G, Pyne D, Hooper S. Applied Altman DG. Measuring inconsistency
response to resistance trining in older physiology and game analysis of rugby in meta-analyses. BMJ 2003: 327:
women. J Appl Physiol 1991: 70: union. Sports Med 2003: 33: 973–991. 557–560.
1912–1916. Falk B, Tenenbaum G. The effectiveness Houmard JA, Hortobagyi T, Johns RA,
Cohen J. Statistical power analysis for the of resistance training in children. A Bruno NJ, Nutte CC, Shinebarger MH,
behavioral sciences. Hillsdale, NJ: meta-analysis. Sports Med 1996: 22: Welborn JW. Effect of short term
Lawrence Erlbaum Associates, 1988: 176–186. training cessation on performance
1–567. Folland JP, Williams AG. The adaptations meausres in distance runners. Int J
Cormie P, McGuigan MR, Newton RU. to strength training: morphological and Sports Med 1992: 13: 572–576.
Developing maximal neuromuscular neurological contributions to increased Hunter GR, McCarthy JP, Bamman MM.
power: part 1 – biological basis of strength. Sports Med 2007: 37: Effects of resistance training on older
maximal power production. Sports Med 145–168. adults. Sports Med 2004: 34: 329–348.
2011a: 41: 17–38. Gandevia SC. Spinal and supraspinal Hurley BF, Hanson ED, Sheaff AK.
Cormie P, McGuigan MR, Newton RU. factors in human muscle fatigue. Strength training as a countermeasure
Developing maximal neuromuscular Physiol Rev 2001: 81: 1725–1789. to aging muscle and chronic disease.
power: part 2 – training considerations Gollnick PD, Matoba H. The muscle fiber Sports Med 2011: 41: 289–306.
for improving maximal power composition of skeletal muscle as a Hyatt RH, Whitelaw MN, Bhat A, Scott
production. Sports Med 2011b: 41: predictor of athletic success. An S, Maxwell JD. Association of muscle
125–146. overview. Am J Sports Med 1984: 12: strength with functional status of
Costill DL, Fink WJ, Hargreaves M, King 212–217. elderly people. Age Ageing 1990: 19:
DS, Thomas R, Fielding R. Metabolic Goodpaster BH, Park SW, Harris TB, 330–336.
characteristics of skeletal muscle during Kritchevsky SB, Nevitt M, Schwartz Jones EJ, Bishop PA, Woods AK, Green
detraining from competitive swimming. AV, Simonsick EM, Tylavsky FA, JM. Cross-sectional area and muscular

9
Bosquet et al.
strength: a brief review. Sports Med strength and muscle fiber muscles. Appl Physiol Nutr Metab
2008: 38: 987–994. characteristics. Eur J Appl Physiol 2007: 32: 1198–1205.
Kraemer WJ, Adams K, Cafarelli E, Occup Physiol 1993: 66: 254–262. Ratamess N, Alvar B, Evetoch T, Housh
Dudley GA, Dooly C, Feigenbaum MS, Moher D, Cook D, Eastwood S, Olkin I, T, Kibler W, Kraemer W, Triplett N.
Fleck SJ, Franklin B, Fry AC, Hoffman Renni D, Stroup DF, QUOROM Group. American College of Sports Medicine
JR, Newton RU, Potteiger J, Stone Improving the quality of reports of position stand. Progression models in
MH, Ratamess NA, Triplett-McBride meta-analyses of randomised controlled resistance training for healthy adults.
T; American College of Sports trials: the QUOROM statement. Lancet Med Sci Sports Exerc 2009: 41:
Medicine. ACSM position stand on 1999: 354: 1896–1900. 687–708.
progression models in resistance Moreland JD, Richardson JA, Goldsmith Saunders PU, Pyne DB, Telford RD,
training for healthy adults. Med Sci CH, Clase CM. Muscle weakness and Hawley JA. Factors affecting running
Sports Exerc 2002: 34: 364–380. falls in older adults: a systematic economy in trained distance runners.
Latham NK, Bennett DA, Stretton CM, review and meta-analysis. J Am Geriatr Sports Med 2004: 34: 465–485.
Anderson CS. Systematic review of Soc 2004: 52: 1121–1129. Skelton DA, Greig CA, Davies JM,
progressive resistance strength training Moritani T, deVries HA. Neural factors Young A. Strength, power and related
in older adults. J Gerontol A Biol Sci versus hypertrophy in the time course functional ability of healthy people
Med Sci 2004: 59: 48–61. of muscle strength gain. Am J Phys aged 65–89 years. Age Ageing 1994:
Manini TM, Clark BC. Dynapenia and Med 1979: 58: 115–130. 23: 371–377.
aging: an update. J Gerontol A Biol Sci Mujika I, Padilla S. Detraining: loss of a Staron RS, Malicky ES, Leonardi MJ,
Med Sci 2012: 67: 28–40. training induced physiological and Falkel JE, Hagerman FC, Dudley GA.
Marcora SM, Bosio A, de Morree HM. performance adaptation. Part I. Sports Muscle hypertrophy and fast fiber type
Locomotor muscle fatigue increases Med 2000a: 30: 79–87. conversions in heavy resistance-trained
cardiorespiratory responses and reduces Mujika I, Padilla S. Detraining: loss of women. Eur J Appl Physiol Occup
performance during intense cycling training-induced physiological and Physiol 1990: 60: 71–79.
exercise independently from metabolic performance adaptations: Part II. Sports Stolen T, Chamari K, Castagna C,
stress. Am J Physiol Regul Integr Med 2000b: 30: 145–154. Wisloff U. Physiology of soccer: an
Comp Physiol 2008: 294: R874–R883. Mujika I, Padilla S. Cardiorespiratory and update. Sports Med 2005: 35: 501–
Martel GF, Roth SM, Ivey FM, Lemmer metabolic characteristics of detraining 536.
JT, Tracy BL, Hurlbut DE, Metter EJ, in humans. Med Sci Sports Exerc Tsuyama K, Yamamoto Y, Nakazato K,
Hurley BF, Rogers MA. Age and sex 2001a: 33: 413–421. Nakajima H. Effects of dynamic neck
affect human muscle fibre adaptations Mujika I, Padilla S. Muscular muscle training on strength and
to heavy-resistance strength training. charateristics of detraining in humans. cross-sectional area of neck muscles in
Exp Physiol 2006: 91: 457–464. Med Sci Sports Exerc 2001b: 33: judo athletes. Jpn J Phys Fit Sports
Mero A, Komi PV, Gregor RJ. 1297–1303. Med 2005: 54: 249–258.
Biomechanics of sprint running. Sports Narici MV, Hoppeler H, Kayser B, Verdijk LB, van Loon L, Meijer K,
Med 1992: 13: 376–392. Landoni L, Claassen H, Gavardi C, Savelberg HH. One-repetition
Mikines KJ, Sonne B, Tronier B, Galbo Conti M, Cerretelli P. Human maximum strength test represents a
H. Effects of training and detraining on quadriceps cross-sectional area, torque valid means to assess leg strength in
dose-response relationship between and neural activation during 6 months vivo in humans. J Sports Sci 2009: 27:
glucose and insulin secretion. Am J strength training. Acta Physiol Scand 59–68.
Physiol 1989: 256: E588–E596. 1996: 157: 175–186. Vincent W, Weir J. Statistics in
Miller AE, MacDougall JD, Tarnopolsky Phillips SM. Resistance exercise: good for kinesiology, 4th edn. Champaign, IL:
MA, Sale DG. Gender differences in more than just Grandma and Grandpa’s Human Kinetics, 2012: 1–392.

10