Physical Activity and Feelings of Energy and Fatigue
- First Online:
- Cite this article as:
- Puetz, T.W. Sports Med (2006) 36: 767. doi:10.2165/00007256-200636090-00004
- 282 Views
Approximately 20% of adults worldwide report persistent fatigue. Physical activity is a healthful behaviour that has promise for combating feelings of fatigue and low energy. This article summarises the epidemiological literature that examined the association between physical activity and feelings of energy and fatigue. Twelve population-based studies conducted between January 1945 and February 2005 that concurrently measured physical activity and feelings of energy and fatigue were located. All of the studies suggested that there was an association between physical activity and a reduced risk of experiencing feelings of low energy and fatigue when active adults were compared with sedentary peers (odds ratio = 0.61; 95% CI 0.52, 0.72). The effect was heterogeneous and varied according to study design and the energy/fatigue measure used in the study. Because epidemiological comparisons cannot establish direction of causality, standard criteria for evaluating strength of evidence in epidemiological studies (i.e. strength of association, temporal sequence, consistency, dose response and biological plausibility) were used to judge whether the observed association between physical activity and feelings of energy and fatigue suggest causality in the absence of adequate experimental evidence. There was agreement among the studies suggesting a strong, consistent, temporally appropriate dose-response relationship between physical activity and feelings of energy and fatigue. No compelling evidence has confirmed any plausible biological mechanisms that explain the apparent protective effect of physical activity against feelings of low energy and fatigue. Nonetheless, the epidemiological evidence is sufficiently strong to justify better controlled prospective cohort studies and randomised controlled trials.
Physical activity is a healthful behaviour that has promise for combating feelings of low energy and fatigue. Fatigue is a major public health concern in which approximately one in five adults worldwide report persistent fatigue. This statistic is slightly higher in the US where fatigue is reported by 24% of the population with women having 1.5 times the risk of males for being fatigued.[3,4] It has been estimated the lifetime prevalence of unexplained fatigue lasting ≥2 weeks is about 14% and that approximately 9% of people at any point in time are experiencing fatigue of >6 months’ duration. Of the individuals experiencing unexplained fatigue in their lifetime, approximately 20% will likely have recurrent/chronic fatigue over many years.
A consensus definition of fatigue has yet to be accepted across the research community. With no known biological markers and diverse proposed causes of feelings of fatigue, defining the construct is problematic. It is probable that feelings of energy and fatigue are multidimensional, with emotional, behavioural and cognitive components. In fact, feelings of energy and fatigue have been conceived in various ways including as symptoms, moods, dimensions of cognitive effort and aspects of quality of life.[7,8] However, for our purposes, the focus is on energy and fatigue conceptualised as mood states. Energy, when defined as a mood state, refers to the subjective feeling of having the capacity to complete mental or physical activities. Fatigue, when defined as a mood state, refers to the subjective feeling of having a reduced capacity to complete mental and physical activities.
Despite the anecdotal reports of increased feelings of ‘vigour’ and ‘pep’ after periods of physical activity, research concerning the association between physical activity and feelings of energy and fatigue is relatively new. The delay in this area of research is likely related to research design and measurement issues. Much of the epidemiological research in this area has utilised cross-sectional designs that limit the interpretations that can be made in the absence of sufficient prospective cohort and randomised controlled studies. Difficulty in measuring fatigue also may have delayed the accumulation of evidence that could help better understand the relationship between physical activity and feelings of energy and fatigue. For example, the prevalence of fatigue ranged from 3.7% to 31.3% in one study depending on whether fatigue was assessed as a clinically diagnosed problem or a persistent symptom during the preceding 2 weeks.
At least 12 epidemiological studies have concurrently measured physical activity and fatigue. Many of these studies have used measures of fatigue with uncertain validity, while others have used the 36-Item Short-Form Health Survey (SF-36) vitality scale. The SF-36 vitality scale is a predominant energy and fatigue measure in scientific literature having been cited in >2500 studies. The vitality scale contains item content and scoring suggestive that its scores measure the bipolar mood of energy/fatigue. Unlike many single-item fatigue scales used in epidemiological research, the SF-36 vitality scale has correlational and experimental evidence that supports interpreting the vitality scores as a valid measure of the frequency of monthly feelings of energy and fatigue. Such evidence has shown vitality scores to have convergent and discriminate validity with other quality-of-life energy/fatigue subscale scores, the ability to discriminate among adult patient groups differing in severity of fatigue-related medical and psychiatric conditions and generalisability to population samples in almost 12 countries.[11–14]
quantify the magnitude of the effect of physical activity on feelings of energy and fatigue in the population;
identify the extent to which the study design and energy/fatigue measure moderate the effect;
judge the strength of the epidemiological evidence that physical activity reduces the risk of feelings of low energy and fatigue.
2.1 Literature Search
Literature was identified by searching Current Contents, Google Scholar, PsycInfo, Web of Science and PubMed data bases from January 1945 to February 2005 using the terms ‘energy’, ‘fatigue’, ‘quality of life’, ‘SF-36’ and ‘vitality’ combined either with ‘exercise’ and ‘physical activity’ or with ‘epidemiology’, ‘population study’, ‘prevalence’ or ‘incidence’. Population-based epidemiological papers concerning physical activity and feelings of energy and fatigue were identified and reference lists of these articles were examined for any additional relevant sources. Investigations that measured vital exhaustion were excluded. Thus, this review focused on studies that examined feelings of energy and fatigue per se and not the construct of vital exhaustion that by definition includes feelings of hopelessness, loss of libido and increased irritability in addition to feelings of low energy and fatigue.
2.2 Study Characteristics
Twelve studies consisting of 137 351 subjects were included in the review.[3,16–26] The 12 studies had a median sample size of 4563 (range: 779–56 510). The age (mean ± SD) of the sample was 49.4 ± 10 years. Women comprised approximately 88% of the sample. To account for the potentially confounding effects of sex, age and body mass index (BMI), these variables were controlled for in ten, eight and five of the studies, respectively. Although, one of the 12 studies derived an effect from a large sample of patients with chronic diseases, the energy/fatigue scores weighted by sample size (mean ± SD) were within the expected range, T-score of 47.4 ± 2.6.
2.3 Effect Size Calculations
When possible, odds ratios were taken or calculated from studies so that scores were interpreted relative to the sedentary sample. The sedentary sample was classified as those individuals with the lowest work and/or leisure time physical activity levels. In all 12 studies, the physical activity levels of sedentary samples fell below the suggested exercise guidelines of the American College of Sports Medicine, the US Center for Disease Control and the US Surgeon General for health-related benefits (i.e. moderate-intensity exercise, equivalent to a brisk walk for at least 20–30 minutes, ≥3 days per week).[28,29] For continuous data, effect sizes were calculated by subtracting the mean score of the physically active group from the mean score of the sedentary group and dividing the difference by the pooled standard deviation. For studies in which precise standard deviations were not reported, the standard deviation was drawn from published norms.[16–18] Effect sizes were calculated so that increases in feelings of energy and decreases in feelings of fatigue resulted in positive effect sizes. Standardised mean difference effect sizes were then converted into an odds ratio equivalent for direct comparisons of effects between studies.[27,30] In studies in which multiple effects could be obtained (e.g. studies with separate results for men and women or that involved more than one physical activity grouping), effects were averaged together so that only one value contributed to the analysis.
2.4 Statistical Analysis
An SPSS macro was used via SPSS version 13.0 (SPSS Inc., Chicago, IL, USA) to calculate the aggregated mean effect size, the associated 95% confidence interval and the sampling error variance using a random effects model. Heterogeneity was indicated if the sum of squares of each effect about the weighted mean effect (QT) reached a significance level of p < 0.05 and the sampling error accounted for <75% of the observed variance. The weighted ‘fail safe N’ (i.e. N+) was computed to estimate the hypothetical number of unpublished or unretrieved studies of null effect and mean weight necessary to reduce the significance of the mean effect to 0.05.
Two variables, study design and energy/fatigue measure, were examined in the investigation. Although other moderator variables were considered, these variables were selected based on theoretical grounds as likely independent moderators and on the distribution of data in that adequate information was available to make meaningful interpretations. Furthermore, with the relatively small number of effects, the inclusion of too many variables could severely reduce the statistical power of the analysis limiting valid interpretations of moderating effects.[33,34] The study design variable was coded into two categories, cross-sectional and prospective cohort designs. The energy/fatigue measure variable also was coded into two categories, SF-36 vitality scale and other energy and fatigue measures.
The moderator variables were entered into a weighted least squares multiple linear regression analysis to determine their independent effects (p < 0.05) on variation in effect size. An SPSS macro was used via SPSS version 13.0 for the analysis which employed a mixed effects model to account for between study heterogeneity associated with both study-level sampling error and random effects variance. The effects were weighed by the inverse of their variance and re-estimated with the random effects variance component added. Significant moderators in the regression analysis were decomposed using a random effects model to compute effect sizes and 95% confidence intervals.
All of the effects showed a positive association between physical activity and a reduced risk of experiencing feelings of low energy and fatigue. The mean odds ratio (95% confidence interval) was 0.61 (0.52, 0.72). The significant effect between physical activity and feelings of energy and fatigue (z = 6.14, p < 0.001) was heterogeneous (QT(11) = 188.43, p < 0.001; sampling error accounted for 22.1% of the observed variance). The fail safe N+ was 5.
3.1 Moderator Analyses
The overall regression model (QR) was significantly related to effect size (QR(2) = 83.83, p < 0.001, R2 = 0.44). In the model, both study design (β = 0.30, z = 9.09, p < 0.001) and energy/fatigue measure (β = 0.15, z = 2.00, p = 0.045) were independently related to effect size. Decomposition of the study design variable showed that the strength of the relationship between physical activity and the reduced risk of experiencing feelings of low energy and fatigue (odds ratio; 95% CI) was attenuated in investigations that used a prospective cohort design (0.68; 0.54, 0.85) compared with those that used a cross-sectional design (0.56; 0.47, 0.68). Decomposition of the energy/fatigue variable showed that the strength of the relationship between physical activity and the reduced risk of experiencing feelings of low energy and fatigue was greater in investigations that used the SF-36 vitality scale (0.59; 0.49, 0.72) compared with investigations that used an energy/fatigue scale other than the SF-36 vitality scale (0.65; 0.50, 0.85).
Physical activity is associated with about a 40% reduced risk in experiencing feelings of low energy and fatigue when physically active adults are compared with their sedentary counterparts. Caution must be taken in the interpretation of these results because the overall effect of physical activity on feelings of energy and fatigue was heterogeneous and moderated by the study design and energy/fatigue measure. Thus, the current state of evidence must be interpreted in terms of both study design and measurement of energy and fatigue. Epidemiological research also does not allow for causal inferences to be made in the absence of adequate experimental evidence. Therefore, the strength of evidence associating physical activity with a reduced risk in feelings of low energy and fatigue must be examined in relation to standard criteria (i.e. strength of association, temporal sequence, consistency, dose response and biological plausibility) before valid conclusions can be drawn about a potential cause-effect relationship. Examination of both the state of evidence and the strength of evidence will help clarify the details surrounding the relationship between physical activity and feelings of energy and fatigue, thus allowing for a more valid interpretation of the of the literature.
4.1 The State of Evidence
Study design and measurement issues are two major factors moderating the relationship between physical activity and feelings of energy and fatigue. Over 40% of the variance in the regression model was explained by these two variables. Therefore, before the current state of evidence examining the association between physical activity and feelings of energy and fatigue can be validly interpreted, the impact of the study design and energy/fatigue measure on this relationship must be addressed.
4.1.1 Summary of Data Based on Study Design
Epidemiological studies can be examined within the framework of a research taxonomy in which the hierarchy of the taxonomy is based on what the study design can and cannot explain. For example, cross-sectional designs examine the presence or absence of a condition/disease at a particular time. Because outcome and exposure are ascertained at the same time, the temporal relationship between the two is unclear. Prospective cohort studies, however, proceed in logical sequence from exposure to outcome, thus accounting for temporal sequence. This advantage places the prospective cohort design higher in the research hierarchy than cross-sectional studies. Such advantages and limitations associated with research design must be considered before valid interpretations of epidemiological evidence can occur.
It is clear that study design has an impact on the effect of physical activity on feelings of energy and fatigue. Because of the inherent weakness of cross-sectional design to account for temporal changes, the effects associated with cross-sectional studies may be inflated compared with prospective cohort design, which precludes this limitation. Despite the prospective cohort designs, advantages over cross-sectional studies, selection and confounding biases still exist with the prospective cohort design. These methodological limitations associated with the current evidence will remain until randomised controlled trials are introduced into the literature. Although study design is important to address, another methodology issue beyond that of study design can have a notable impact on the relationship between feelings of energy and fatigue in epidemiology research, namely the energy/fatigue measure used in the study.
4.1.2 Summary of Data Based on Energy/Fatigue Measure
In epidemiological research examining the relationship between physical activity and feelings of energy and fatigue, the issue of measurement becomes a key issue on two fronts. There is no ‘gold standard’ for measuring physical activity or feelings of energy and fatigue. Thus, both the exposure and outcome variables must be assessed with imperfect measures. This issue forces careful interpretation of results. The problem of establishing the validity of physical activity instruments has been a recognised, yet still unresolved, issue in epidemiological research for some time. However, the problem of establishing the validity of energy and fatigue measures in epidemiological research has just recently become a focus in the areas of medicine and mental health. Thirty or more fatigue scales have proliferated in the clinical/scientific community and no two scales measure the construct of energy and fatigue exactly the same.[8,38] Thus, epidemiological research has been inundated with energy and fatigue measures ranging widely in their ability to offer valid interpretations of the construct.
Five of the 12 epidemiological studies examining the relationship between physical activity and feelings of energy and fatigue have used measures of energy or fatigue with questionable validity.[3,21,22,24,25] These studies were associated with a 35% reduced risk in experiencing feelings of low energy and fatigue when physically active adults were compared with sedentary peers. This is in contrast to the seven epidemiological studies that used the SF-36 vitality scale.[16–20,23,26] Studies using the SF-36 vitality scale were associated with a 41% reduced risk in experiencing feelings of low energy and fatigue when physically active individuals were compared with their sedentary counterparts. The larger effect associated with the SF-36 vitality scale was likely related to the fact that the vitality scale is a multi-item scale focusing specifically on the frequency of feelings of fatigue experienced over the last month. Other less well established measures were either single-item measures that lacked sensitivity or were multi-item measures that confounded fatigue with other psychological constructs such as anxiety or depression.
The epidemiological evidence associated with questionable measures of energy and fatigue is difficult to interpret because of several measurement limitations. These issues include the use of single-item scales without defined population norms to arbitrarily dichotomise groups,[3,22,25] the use of multi-item scales with validated population norms, but confounded by the multidimensional aspects of the fatigue measure (i.e. using physical inactivity as both a dependent and independent measure) and the use of multi-item scales that confound the construct of fatigue with other constructs such as stress or anxiety. Because these measures do not have adequate validity evidence to support the interpretation of their scores, it becomes difficult to conclude whether physical inactivity is, or is not, truly related to feelings of energy and fatigue. The SF-36 vitality scale, on the other hand, provides stronger evidence for the relationship between physical activity and feelings of energy and fatigue because of its ability to provide a more valid interpretation of scores.
Although studies incorporating the SF-36 vitality scale and prospective cohort design appear to provide the most valid representation of the association between physical activity and feelings of energy and fatigue, these studies are still limited in that epidemiological comparisons between physically active and inactive groups cannot establish direction of causality. The nature of the current state of epidemiological evidence, even when taking into consideration the study design and measurement issues, does not permit a defensible statement about the cause-effect relationship of physical activity and feeling of energy and fatigue. Given the evidence supports a valid association between physical activity and feelings of energy and fatigue, one additional question remains: are the observed associations causal?
4.2 The Strength of Evidence
4.2.1 Strength of Association
In order to meet the criteria for the strength of association between the risk factor and the outcome, there must be a large and clinically meaningful difference in disease or disorder risk between those exposed and those not exposed to a risk factor in order to establish causality. The cumulative evidence of the observational studies using both cross-sectional and prospective cohort designs shows about a 40% reduction in risk of experiencing feelings of low energy and fatigue when active adults are compared with sedentary peers. In those studies that utilised the SF-36 vitality scale, the evidence showed a 41% reduction in risk of experiencing feelings of low energy and fatigue for physically active groups compared with their sedentary counterparts. These results with the SF-36 vitality scale can be interpreted as suggesting increased levels of physical activity lead to increases in the frequency of feelings of energy and decreases in the frequency of feelings of fatigue. The criteria for the strength of association between physical inactivity and feelings of low energy and fatigue appears to have been met; however, more studies with ideal methods (e.g. random sampling and better physical activity measures) would strengthen the evidence.
4.2.2 Temporal Sequence
In order to meet the criteria for appropriate temporal sequence between the risk factor and the outcome, exposure to a risk factor must precede development of the disease or disorder in order to establish causality. There must also be adequate time between the exposure to the risk factor and the development of the disease or disorder. Large prospective cohort studies, which demonstrate the appropriate temporal sequence, have ranged in duration from 1 to 5 years and have shown about a 32% reduced risk of experiencing feelings of low energy and fatigue among those people who become physically active compared with those individuals who maintain a sedentary lifestyle.[17,18,24–26] The criteria for appropriate temporal sequence appear to have been met based on the few prospective cohort studies that have examined the association between physical inactivity and feelings of low energy and fatigue.
It also appears the association between chronic exercise and low energy and fatigue mood states is alterable with exercise adoption or exercise cessation. This means that the association between a risk factor and disease or disorder will change over time by altering variables associated with the risk factor. Only one epidemiological study has examined the alterability of feeling of energy and with the adoption or cessation of exercise. In the study, those individuals who reported exercise cessation during a 3-year observation period had greater decreases in feelings of energy than those who remained sedentary, those who remained physically active or those who reported exercise adoption during that same period. The greatest positive changes in feelings of energy occurred among those who reported exercise adoption during the 3-year period, and these changes were similar to those in the group that remained physically active. This effect was independent of baseline SF-36 physical component summary score, marital status, BMI and recent life events.
The criteria for appropriate temporal sequence appear to have been met based on the prospective cohort studies that have examined the association between physical inactivity and feelings of low energy and fatigue. Furthermore, it appears this relationship is alterable with increases or decreases in physical activity level. Additional longitudinal research needs to be conducted before stronger conclusions can be made. In particular, there is a need to examine day-to-day and week-to-week changes in feelings of energy and fatigue with exercise adoption in order to better understand temporal sequence.
In order to meet the criteria for the consistency of the relationship between the risk factor and the outcome, the observed association between a risk factor and a disease or disorder must always be observed when the risk factor is present. Thus, the relationship should be similar among groups and across studies. Improvements in energy and fatigue mood states after chronic physical activity appear to be consistent across a range of populations and conditions. The epidemiological evidence indicates a consistent relationship across age, nationalities (i.e. American,[3,18,19,23] Australian,[17,20] Dutch,[24,26] Finnish, Israeli, Japanese, Norwegian), modes of exercise and medical conditions.
In general, there appears to be a consistent relationship between physical activity and feelings of energy and fatigue between men and women.[16,21,24,26] However, half of the studies suggesting this consistency have been cross-sectional in design. It is of note that the one study that did combine cross-sectional and prospective cohort designs showed inconsistencies in this relationship between men and women with the prospective design. In the study, both cross-sectional and longitudinal data were collected from a sample of Dutch men and women aged 20–59 years. The cross-sectional association at baseline and follow-up showed a consistent positive association between moderately intense leisure time physical activity and feelings of energy in both men and women. However, the longitudinal data found a statistically significant relationship in the change in total leisure time physical activity and change in feelings of energy and fatigue in men, but not women.
The criteria for the consistency of the relationship have been met in studies with different ages, nationalities, medical conditions and modes of exercise, but more research needs to be conducted in order to understand the consistency of the relationship between men and women using prospective cohort designs.
4.2.4 Dose Response
In order to meet the criteria for a dose-response relationship between the risk factor and the outcome, the risk of disease or disorder associated with a risk factor is greater with stronger exposure to the risk factor. Seven of the 12 available studies allow for an examination of a potential dose-response relationship between physical activity and feelings of low energy and fatigue.[16,17,19,20,23,24,26] All of these studies show a negatively accelerating dose-response relationship between chronic exercise and feelings of low energy and fatigue, suggesting that as individuals increase physical activity they decrease their risk of experiencing feelings of low energy and fatigue. This negatively accelerating dose response appears to be consistent across the frequency, intensity, duration and total volume (i.e. kilocalories expended) of physical activity. This evidence suggests that as individuals increase physical activity, they also increase the frequency of feelings of energy and decrease the frequency of feelings of fatigue. These finding are in agreement with a previous review suggesting that fatigued, sedentary individuals who adopt a moderate exercise programme should reap the greatest psychological benefit of increased feelings of energy and reduced feelings of fatigue.
These seven studies provide information about the dose-response relationship between physical activity and feelings of low energy and fatigue. The greatest risk for feelings of low energy and fatigue is found among sedentary individuals. The greatest reductions in risk have been found when inactive individuals are compared with those who engage in a small amount of physical activity. These observations are consistent with current American College of Sports Medicine, the US Center for Disease Control and the US Surgeon General recommendations regarding the health benefits of engaging in a modest amount of regular physical activity.[28,29]
4.2.5 Biological Plausibility
In order to meet the criteria for biological plausibility, the observed association between a risk factor and disease or disorder outcome must be explainable by existing knowledge about possible biological mechanisms of the disease or disorder. In general, the biological basis for feelings of energy and fatigue is poorly understood. Moods result from neural activity of the brain. The specific brain mechanisms that generate the moods of energy and fatigue are unknown, but monoamines, histamine, acetylcholine, glutamate and GABA-mediated neurotransmission have been implicated.[39–41] There is evidence that physical activity can alter these neurotransmitters and neuromodulators.
Perhaps more importantly than just understanding the neurotransmitters and neuromodulators involved in generating moods of energy and fatigue is examining how these chemical messengers regulate hypothetically malfunctioning neurological circuits.[39,41] The exact neuronal basis of energy and fatigue has been difficult to associate with specific brain areas and circuitry. However, some sound conjectures can be made in this regard and to the extent that generalisations can be made about brain areas that could be associated with mental and physical feelings of fatigue. For example, brain cortical areas, such as the dorsolateral prefrontal cortex and CNS components regulating motor functioning, such as the striatum, cerebellum and spinal cord, could be reasonable candidates in mediating moods of mental and physical fatigue, respectively.[39,41]
Biological plausibility of physical inactivity causing feelings of low energy and fatigue appears possible, but currently no evidence exists examining the effects of chronic exercise training on brain biological mechanisms associated with feelings of energy and fatigue. Examining hypothetically malfunctioning brain circuitry and the regulatory effects of specific neurotransmitters and neuromodulators on these circuits may facilitate our understanding of the general neurobiological mechanisms of moods of energy and fatigue. With this in mind, the role of physical activity in altering moods of energy and fatigue should be pursued in that physical activity can act as a vehicle for altering neurotransmitters and neuromodulators, leading to regulatory influences on possible brain circuitry involved in moods of energy and fatigue.
Epidemiological studies have shown that people who are physically active in their leisure time have about a 40% reduced risk of experiencing feelings of low energy and fatigue compared with sedentary comparison groups. This is similar to other more prominent psychological disorders such as anxiety and depression. For example, epidemiological studies comparing physically active individuals to their sedentary counterparts show about a 50% reduced risk in experiencing depression.
It appears that study design and the energy/fatigue measure used in a study can moderate the effect of physical activity on feelings of energy and fatigue. This review suggests that prospective cohort designs that utilise well established energy and fatigue measures such as the SF-36 vitality scale currently provide the most valid interpretation of the effects of physical activity on feelings of energy and fatigue. Future studies should take into consideration these two methodological design issues.
Examination of the epidemiological evidence using Mill’s cannons as a judge for whether the relationship between inactivity and feelings of low energy and fatigue might be causal suggests a strong agreement among the studies. There appears to be a strong, consistent, temporally appropriate dose-response relationship between chronic exercise and feelings of energy and fatigue. However, evidence to confirm hypothesised plausible mechanisms of a protective effect of physical activity against feelings of low energy and fatigue is as yet uncompelling. Nonetheless, the epidemiological evidence is sufficiently strong to justify better controlled prospective cohort studies, randomised controlled trials and other investigations aimed at understanding how physical activity improves feelings of energy and fatigue.
The author would like to thank Derek Hales, Patrick O’Connor and Rod Dishman for their assistance in preparing this manuscript.
No sources of funding were used to assist in the preparation of this review. The author has no conflicts of interest that are directly relevant to the content of this review.