Exercise Dose in Clinical Practice

Livelihood of the human species has depended on routine physical activity (PA) for thousands of our years. Until recent times, the successful procurement of basic life-sustaining commodities such as food and water required us to move on a regular basis. We therefore evolved the capacity to walk and to run, oftentimes over long distances, and our survival depended on our ability to do so. In parallel with the development of adaptations that facilitated PA were those that promoted fuel storage and conservation of energy. The balance between acquired traits that enabled PA and those that supported nutrient storage and encouraged rest and recovery has, until recently, served our species well.

Once a necessary part of life, routine PA is no longer a requirement in contemporary developed society. Technology obviating the need to walk, including cars, escalators, elevators, and moving walkways, are a ubiquitous part of daily life. In addition, numerous societal influences, both at home and in the workplace, promote sedentary living at the expense of physically active work and play. This transition represents a paradigm shift for the human species with potentially grave consequences, including the epidemic of obesity and the burden of chronic diseases attributable to physical inactivity. However, for reasons including the pursuits of health, fitness, and recreation, a substantial number of people voluntarily choose to engage in some form of PA, ranging from low-intensity walking to extreme endurance sports. Thus, the PA “dose” to which our species is exposed has never been so variable and so complex.

Heterogeneity in PA habits is relevant to clinical cardiovascular practice on a patient-by-patient basis. Routine PA is an effective way to increase longevity and to reduce the risk of cardiovascular disease (CVD). It is now standard of care to counsel sedentary or inadequately active patients to increase PA for both the primary and secondary prevention of CVD. In addition, recent data suggesting that high levels of PA may lead to significant cardiovascular consequences have led highly active patients and their clinicians to question the risk/benefit ratio of high exercise dose exposure. This review was designed to address the concept of exercise dose in contemporary cardiovascular practice with an emphasis on the relationship between PA and key response outcomes. We begin with an overview of fundamental exercise dose principles, highlight data describing the relationship between PA and specific clinically relevant outcomes, address recent data examining the impact of exercise at both ends of the dose spectrum, and conclude with a brief discussion of integrating the exercise dose concept into clinical practice. This review focuses on the effects of exercise on cardiovascular risk factor modification and overall mortality because the impact of exercise on CVD prevalence, CVD mortality, and secondary prevention, including the role of formal cardiac rehabilitation, has recently been discussed elsewhere.^1,2

Exercise Dose: Historical Considerations

Although the concept that PA habits affect health and longevity dates back to the writings of ancient scholars, including Hippocrates and Galen,³ most of what is now known about this topic stems from recent decades. In 1953, Professor J.N. Morris and colleagues^4,5 from the United Kingdom published tandem landmark studies examining the interactions between occupational PA and incident coronary heart disease and longevity. In this observational work, London streetcar conductors, who spent their working time walking up and down aisles and double-decker staircases, had approximately half the rate of coronary heart disease of substantially more sedentary streetcar drivers. Morris and colleagues^6,7 observed similar results among alternative vocations, including postal workers and civil servants.

As occupational activity declined as a proportion of an individual’s total PA, researchers shifted their focus away from examining this domain, instead focusing on quantifying PA exposure during to leisure-time activity or across the entire day.⁸ With this new focus, large, prospective, cohort studies, including the Harvard Alumni Health Study and the Women’s Health Initiative, have subsequently further clarified the relationship between exercise dose and both cardiovascular health and longevity.^9–11 These epidemiological studies relied on a common observational approach in which exercise exposure, conventionally obtained with standardized self-reported survey tools, were quantified in large numbers of people who were followed up with respect to clinical outcomes over extended time periods. These cohorts, when stratified as a function of habitual PA practices, thus provide valuable information about the relationship between exercise dose and clinical outcomes. In aggregate, epidemiology literature has evaluated the associations between PA and key health outcomes among large numbers of participants across heterogeneous populations and thus serves as the invaluable foundation for current public health guidelines.

However, it must be acknowledged that data derived from epidemiological investigation have several key limitations. First, they may be subject to unmeasured bias and confounding that preclude definitive determination of causality. For example, subsequent investigations by Heady et al¹² found that streetcar drivers had, on average, greater waist circumference as derived from uniform measurements than conductors. Although subsequent analyses adjusted for this identified confounder and still found important differences in coronary heart disease,¹³ other differences between the 2 occupations likely remained unidentified. Second, epidemiological studies typically rely on the use of self-reported PA habits, which may be inaccurate despite the use of the most rigorous survey tools. Finally, epidemiological data sets typically contain very small numbers of people who routinely participate in the highest levels of exercise and thus are almost universally underpowered to examine the upper end of the exercise dose-response curve.

The randomized, controlled trial (RCT) design, the gold-standard approach to clinical hypothesis testing, has the potential to overcome many of the limitations of observational work. The RCT approach, as discussed in subsequent sections, has been successfully applied to examine the biological impact of PA. RCTs are, however, among the most resource-intense, logistically complex study designs. Therefore, all exercise RCTs to date have used the rapidly responsive surrogate markers of disease risk or burden (ie, plasma lipids, physical fitness, myocardial structure) rather than hard clinical end points. Exercise RCTs have therefore greatly enhanced our understanding of exercise physiology, perhaps delineating mechanisms that underlie the outcome trends reported in comparatively larger observational work, but they provide little information about the impact of exercise on definitive health outcomes such as mortality. The design and implementation of exercise RCTs powered to examine hard clinical outcomes, although a logistic and resource-intense challenge, represents an important future goal for the broad scientific community.

Exercise Dose: Common Definitions

PA is defined as any bodily movement resulting from the contraction of skeletal muscle that increases energy expenditure above the basal level.¹⁴ Exercise, as a subcategory of PA, is defined as any planned and structured action with the objective of improving or maintaining physical fitness or health. All exercise involves some combination of isometric (static) and isotonic (dynamic) stress, which serves as the basis for the physiological classification of competitive sports. In clinical practice, exercise is most commonly divided into the broad subgroups of aerobic/endurance (ie, running, walking) and resistance (ie, weight lifting) activity, although many sport and exercise modalities integrate both physiological disciplines. Although data describing the cardiovascular impact of resistance exercise are mounting, this review focuses exclusively on aerobic/endurance exercise, given the relatively well-developed concept of exercise dose response for this modality. Within the broad category of aerobic/endurance exercise, common modalities include walking/running, cycling, rowing, and swimming. Although methods of measuring exercise exposure vary across these different forms of exercise, universal concepts for the quantification of exercise dose are described below.

Quantification of exercise exposure is typically accomplished with the concept of exercise dose. At the most basic conceptual level, exercise dose is determined by 3 discrete variables, duration, frequency, and intensity, and in aggregate, exercise dose can be described as the product of these 3 parameters. Duration reflects the amount of time accrued in a single exercise session and, for aerobic/endurance exercise, is most often characterized as minutes or hours. Frequency captures the number of exercise sessions over more extended periods (ie, days, weeks, or months). In sum, these 2 parameters reflect the total amount of time spent in exercise over a given period.

Exercise intensity, a relatively more complex concept, is typically quantified in absolute terms as the metabolic cost of an exercise session or in relative terms as the performance of a given activity as a function of some percentage of measurable maximal capacity (Table 1). For aerobic/endurance exercise, absolute exercise intensity is commonly measured in kilocalories burned per unit of time or in metabolic equivalents (METs; Table 1). Using standardized estimates of METs or kcal expenditure, these metrics of absolute intensity are useful for the estimation of exercise dose in community-based populations and thus are commonly used in large observational studies of exercise dose. A limitation of absolute intensity metrics is their innate inability to account for the large variability in fitness that exists across individuals. For example, intensity exercise of 5 METs may simultaneously represent a peak exercise effort for a patient with advanced CVD and a relatively easy effort for a competitive athlete.

In contrast, indexes of relative exercise intensity account for differences in individual fitness levels by defining intensity as a percentage of some peak or maximal physiological parameters (ie, heart rate and oxygen consumption; Table 1). In the example provided above, exercise performed at a relative intensity of 75% peak oxygen consumption (V.o₂) may translate into a treadmill walk at 3 mph for a patient with advanced CVD and a treadmill run at 8 mph for a competitive athlete. The use of relative exercise intensity is commonplace in clinical practice for the generation of exercise prescription and in high-quality observational and interventional exercise studies. Relative exercise intensity metrics may be preferable to absolute intensity metrics when they can be applied with rigor. However, it must be emphasized that effective application of a relative intensity metric requires accurate determination of a peak value for the metric of choice, which typically involves some form of laboratory- or field-based exercise assessment. This is typically a resource-intensive process that may not be feasible for large population-based studies. Although the use of equations or predictive algorithms for metrics such as peak heart rate have been developed, the application of these tools is often associated with considerable inaccuracy.

Both metrics of intensity can be used effectively to measure cumulative exercise dose by multiplying by exercise duration and frequency. When exercise is measured in terms of absolute intensity, this translates to total dose expressed as kilocalories per week, MET-hours per week, or MET-minutes per week (Table 1). These values are commonly used in epidemiological studies of exercise dose because they can be extrapolated from self-reported activity patterns, but they have proved difficult to apply in clinical practice during the assessment of exercise habits and the prescription of exercise interventions. Conversely, when exercise is measured in terms of relative intensity, total dose is instead expressed as time per week spent in light/moderate/vigorous exercise. These terms may be more patient-friendly but, as noted above, require some knowledge of an individual patient’s physiology and exercise capacity. Both approaches have intrinsic strengths and limitations and consequently may be more or less appropriate in specific clinical and research settings. As detailed below, both the total exercise dose and the relative contribution of its 3 cardinal components may affect the relationship between exercise exposure and a given clinical or physiological outcome.

Exercise Dose and Clinical Outcomes

Numerous observational studies and several well-designed exercise intervention studies have examined the impact of exercise dose on cardiometabolic risk factors, physical fitness, and mortality. Careful examination of the existing data suggests that the optimal dose of aerobic exercise, more specifically the exercise dose required to produce the most benefit, varies considerably across key outcomes of clinical relevance. In clinical practice, this means that the prescription of an exercise dose may best be accomplished by identifying a specific therapeutic target for each individual patient.^15–28

Plasma Lipids

The impact of aerobic exercise training on serum lipoproteins has been studied extensively. Representative exercise intervention studies that have specifically examined the aspects of exercise dose are summarized in Table 2. The response of high-density lipoprotein (HDL) to exercise training appears to be modest but favorable. A 2007 meta-analysis of 25 exercise intervention RCTs reported a mean net increase in HDL of ≈2.5 mg/dL.²⁹ In this study, the estimated minimal weekly exercise energy expenditure and duration required for this HDL response were 900 kcal and 120 minutes, respectively, which approximates the minimum exercise dose recommended by current PA guidelines of at least 150 minutes of moderate-intensity exercise or 75 minutes of vigorous-intensity exercise weekly.^14,30 A 2007 meta-analysis demonstrated that HDL augmentation was not significantly affected by exercise interventions using training sessions of <30 minutes, was more robust among subjects with higher initial total cholesterol and lower body mass index, and was not associated with exercise frequency or intensity.²⁹ This meta-analysis contained relatively few studies examining exercise interventions exceeding the 2000-kcal/wk range and thus does not provide sufficient data to clarify the HDL–exercise dose relationship above this level of energy expenditure. A prior observational study examining changes in lipids over a 20-day road race suggested that exercise at this high dose may translate into a greater HDL rise than reported with more moderate levels of exercise.³¹ The concept that an exercise dose that is well in excess of PA recommendations may incrementally improve lipids above more modest exercise exposure is supported by observational data characterizing lipid profiles among endurance athletes.^32–34

The impact of aerobic exercise training on serum low-density lipoprotein (LDL) is not consistent across available studies. Two large meta-analytic studies found that HDL but not LDL was significantly improved by exercise training when confounding variables (diet, body weight, etc) were considered.^35,36 Conversely, a series of meta-analyses by Kelley et al^37–40 found that routine exercise produced a clinically favorable but small reduction in LDL of 4 to 6 mg/dL among various adult populations. Similarly, recreational middle-aged male runners participating in an 18-week jogging program in preparation for a marathon road race experienced a 5% reduction in total LDL.⁴¹ An important, well-designed, randomized, controlled study (Studies of Targeted Risk Reduction Intervention Through Defined Exercise [STRRIDE]) examined the impact of several doses of moderate- and high-intensity aerobic exercise (walking and jogging, respectively; Table 2). Despite negligible changes in total LDL, study participants demonstrated significant reductions in both the size and number of atherogenic LDL subfractions that were greater in those jogging ≈17 miles/wk than in those walking or jogging ≈11 miles/wk, suggestive of a significant dose response.^15,42

The response of triglycerides to exercise training interventions has been similarly inconsistent and highly variable. A meta-analysis examining aerobic exercise training in an unselected adult population demonstrated decreases in triglycerides of 5 to 38 mg/dL or 4% to 37%,³⁵ although other studies have shown smaller reductions in triglycerides.^37–40 The greater heterogeneity in the response of triglycerides compared with HDL may be partially explained by the fact that triglyceride levels are affected acutely by the last exercise session to a greater degree than HDL.⁴³ Results from STRRIDE suggested a significant dose response for triglycerides, with higher total amounts of exercise (≈17 miles of jogging versus ≈11 miles of walking or jogging) resulting in greater triglyceride reduction. Interestingly, STRRIDE also suggested that moderate-intensity compared with high-intensity exercise, even when total exercise dose (ie, energy expenditure) is kept consistent, may result in a greater initial decrement in triglycerides and a more sustained reduction after training cessation.^15,42 However, the notion that exercise intensity is a key determinant of triglyceride response has not been a consistent finding. The majority of studies have shown that when a reduction in triglycerides is observed, overall exercise dose is the most important determinant.³⁵

In summary, routine exercise has clearly been demonstrated to have a beneficial impact on serum lipids. The effect of exercise appears strongest and most consistent for HDL, whereas the impact of exercise appears both less consistent and smaller in magnitude for LDL, subfractionated lipoproteins, and triglycerides. Although the magnitude of the impact of routine exercise on lipids at the individual patient level is small, lowering of total cholesterol by as little as 10% through either dietary or pharmacological intervention has been shown to result in a 27% reduction in incident CVD.⁴⁴ Although no such analysis exists for the specific impact of exercise on lipids, these data suggest that the small changes in serum lipids with exercise may have the potential to reduce the risk of CVD and CVD-related mortality significantly when translated across broader populations. Both cross-sectional data and exercise intervention studies appear to conclude that the minimum exercise dose required to produce favorable changes in serum lipids approximates that recommended by current PA guidelines and that incremental benefit may be achieved with exercise doses that exceed these levels. At the present time, there does not appears to be an upper level of exercise dose above which deleterious effects on serum lipids occur; available epidemiological data support a continuous incremental beneficial impact on HDL through doses of exercise far in excess of current PA guidelines. However, the relationships between high-dose exercise, lipids, and complementary indexes of CVD risk (eg, inflammatory biomarkers) have not yet been examined in controlled, longitudinal studies.

Blood Pressure

The relationship between exercise training and blood pressure (BP) has been the topic of extensive study. In large, observational studies, greater amounts of leisure-time PA have been associated with a reduction in incident hypertension,^45–48 although this relationship has been less consistent among women and blacks. Controlled exercise intervention studies and meta-analyses that have examined the BP response to exercise training suggest a modest but clinically significant effect. Specifically, exercise training appears to translate into a 3– to 5–mm Hg reduction in systolic BP (Δ2%–4%) and 2– to 4–mm Hg reduction in diastolic BP (Δ2%–3%).^49–53 BP reduction attributable to exercise is generally greater among hypertensive individuals (reductions of 6–8 mm Hg in systolic BP and 5–6 mm Hg in diastolic BP) and slightly less in normotensive individuals (reductions of 2–3 mm Hg in SBP and 1–2 mm Hg in DBP).^49,50,54

There does not appear to be a clear minimal exercise dose threshold for BP lowering or any clear independent effect of the key components of exercise dose: exercise frequency, bout duration, or intensity. One small, controlled study in hypertensive older men suggested that lower-intensity exercise was associated with a greater reduction in SBP than higher-intensity exercise (<70% V.o₂max, Δ20 mm Hg versus >70% V.o2max, Δ 8 mm Hg; Table 2).²¹ The finding that low-intensity exercise was superior to high-intensity exercise for lowering BP supported prior animal model data.^55,56 It is noteworthy that although this concept has persisted in the literature for years, neither subsequent controlled exercise intervention studies that have directly compared different exercise intensities (Table 2) nor any meta-analytic studies have replicated these results in humans. On the contrary, several studies comparing high-intensity interval-based training with moderate continuous-intensity exercise have found that high-intensity training is at best superior or at least equivalent to similar amounts of moderate-intensity continuous training for BP lowering.^19,20,24 In summary, the impact of exercise on BP has been clearly demonstrated in studies that have used exercise doses in the range of current PA recommendations.^14,30 However, a clear dose response has not been demonstrated, and there is neither definitive evidence to support a minimum exercise threshold for BP lowering nor any data to support an upper limit of exercise dose beyond which the BP reduction is diminished.

Insulin and Glucose Metabolism

Exercise increases insulin sensitivity and non–insulin-mediated skeletal muscle glucose metabolism, and epidemiological data suggest that exercise training improves metabolic heath.^57,58 Several large RCTs have examined the impact of exercise on the prevention or treatment of metabolic syndrome (MS) and diabetes mellitus (DM). However, most have included exercise as part of a more comprehensive “lifestyle intervention” that typically includes dietary changes and weight loss, thereby making it difficult to examine the independent effects of exercise. Two such large studies, the Finnish Diabetes Prevention Study⁵⁹ and the US Diabetes Prevention Study,⁶⁰ prescribed exercise doses of 240 and 150 min/wk of moderate-intensity exercise, respectively. Post hoc analyses of both studies indicated that weight loss rather than exercise was the most important determinant of incident DM risk reduction. In the US Diabetes Prevention study, exercise was not significantly associated with reduced risk of DM when adjusted for weight change.⁶¹ However, participants in the Finnish study, who increased their exercise exposure the most, even after adjustment for weight loss and dietary changes, appeared to reduce the risk of DM significantly more than those whose exercise habits did not change.⁶²

It appears that increasing exercise doses lead to incremental improvements in markers of metabolic health. In a post hoc analysis of the STRRIDE study, the prevalence of MS declined from 41% to 27% among those who completed an exercise intervention but remained unchanged in the control group (Table 2). Furthermore, the “high-amount, high-intensity” group, whose overall gross exercise energy expenditure was highest, had the greater improvement in MS characteristics compared with the lower-dose exercise groups.²⁵ In similar fashion, a recent study of healthy subjects found that the dose of prescribed exercise was directly associated with improvements in insulin sensitivity in a graded fashion without evidence for a minimum required threshold or maximal effect.⁶³

Although it appears clear that there is a dose response for exercise and insulin and glucose metabolism, there are conflicting data on the optimal exercise intensity for improving metabolic health.²⁵ In the STRRIDE study, moderate-intensity exercise (ie, walking) improved MS to a greater degree than high-intensity exercise (ie, jogging) of a similar total dose (≈11 miles/wk, ≈1,100 kcal/wk). As discussed above, the potential superiority of moderate-intensity exercise for attenuating incident MS was similarly shown for serum triglycerides. This raises the possibility that moderate-intensity exercise uniquely affects the metabolic pathways that link insulin resistance, impaired postprandial lipolysis, and triglyceride excursion.⁶⁴ In contrast, other studies in both healthy subjects and those with MS have found that high-intensity exercise, including steady-state and interval-based training, is superior to moderate-intensity exercise of equivalent energy expenditure in improving insulin sensitivity.^22–24 In sum, the above studies confirm a beneficial association between exercise and metabolic health but provide no clear conclusions about which exercise intensity optimizes this relationship.

In summary, comprehensive lifestyle interventions, including exercise, diet, and associated weight loss, have clearly been shown to be beneficial in preventing MS and DM. The exercise programs that have been specifically studied in large trials are similar to that in the current PA guidelines, although the bulk of available data suggest that a higher dose of exercise results in greater improvements in insulin sensitivity.

Body Mass

Weight loss is among the most popular reasons that exercise is prescribed because it is indisputable that changes in body mass are driven by changes in net energy balance. To what degree a given exercise intervention will favorably affects energy balance depends on concomitant changes in caloric intake. Numerous organizations have published guidelines that address the role of exercise for the prevention and treatment of obesity.^65–67 Using cross-sectional data describing the PA levels associated with a normal body mass index, these organizations recommend an exercise dose that is approximately double that of the current PA guidelines for weight loss or the prevention of weight regain.

Data from several recent RCTs estimate that a weekly exercise dose ranging from 13 to 26 MET-hours weekly is required for minimal weight loss (1%–3%) or weight stability.^26–28,68 There appears to be a direct dose-response relationship between exercise and body mass. The STRRIDE study found that although those who exercised at a “high” amount (jogging ≈17 miles/wk) experienced small but significant reductions in weight (−3%) and body fat (−7%), those participating in “low” amounts of exercise (walking or jogging ≈11 miles/wk) had stable weight and visceral adiposity.²⁷ Two other RCTs, both that prescribed a single exercise dose to all subjects and then analyzed results on the basis of tertiles of PA as determined by variable compliance, found that higher exercise dose was associated with greater weight and body fat loss.^26,28 It is noteworthy that these studies did not concomitantly use specific diet interventions and that the authors of one of these studies calculated that subjects should have lost almost 8 kg instead of the observed 1 to 2 kg on the basis of the energy expenditure of the exercise program..²⁸ To achieve substantial weight loss (>5% decrease in body mass), 2 approaches have been proposed: a concomitant dietary intervention (either maintenance of pre-exercise intake or reduction in intake)^69,70 or exercise of a sufficient dose to expend a high amount of calories, at least 26 MET-h/wk (eg, walking at 3 mph for 60 minutes daily or jogging at 6 mph for 20 minutes daily).^71,72

In summary, the minimum exercise dose required to maintain optimal weight or to stimulate weight loss appears to be substantially higher than the minimum dose required to favorably affect other CVD risk factors (lipids, BP, insulin sensitivity). There is a clear dose response with exercise and weight loss. Available data suggest that a dose of at least approximately double that of the current PA guidelines is necessary to reliably establish and maintain a healthy weight and that more activity is required to achieve more substantial weight loss. In clinical practice, we routinely emphasize the importance of coupling exercise with deliberate control of caloric intake for patients who seek to reduce body mass.

Physical Fitness

Numerous observational studies suggest that there are inverse relationships between cardiorespiratory fitness and both all-cause and CVD-attributable mortality.^73–75 For example, the adjusted hazard ratio for death declined by 12% for every 1-MET increase in peak exercise capacity among older male veterans followed up over 20 years.⁷⁵ Fitness can be improved by increasing exercise dose through the augmentation of any the principal dose components (frequency, intensity, or duration)^76,77 among people of all ages and both sexes.^78,79 For example, the STRRIDE study demonstrated that an exercise program consisting of variable exercise intensity at fixed total exercise dose (ie, jogging versus walking ≈11 miles/wk) resulted in equivalent improvements in peak V.o₂, whereas a program with variable dose but fixed intensity (ie, walking ≈17 versus ≈11 miles/wk) led to greater improvement in peak V.o₂ in the group who walked more.⁷⁷ Physical fitness is determined by numerous factors other than exercise dose, including age, sex, body mass, and genetics.⁸⁰ The individual response to exercise training therefore varies widely, as evidenced by the observation that 8 months of standardized exercise exposure (jogging 17 miles/wk) led to changes in peak V.o₂ that ranged from −5% to +70%.⁷⁷ In addition, the relationship between objectively measured fitness and exercise dose is further complicated by exercise modality, particularly among the very fit who may train primarily with a form of exercise (Nordic skiing, swimming, etc) that may not easily be reproduced during laboratory-based testing.

Finally, it noteworthy that physical fitness appears to be far more responsive to exercise training than any of the more traditional cardiovascular risk factors discussed above. This observation may in part reconcile the finding that routine PA attenuates the risk of CVD and mortality more than can be explained by its impact on risk factor modulation. Specifically, an analysis of data from the Women’s Health Study suggested that exercise-mediated modification of traditional risk factors explains only 59% of the inverse association between PA and cardiovascular events.⁸¹ This finding suggests that we have much to learn about the protective effects of exercise and that a sizable portion of the overall beneficial effect of exercise appears to be delivered through mechanisms that remain elusive.⁸² Therefore, understanding the relationship between traditional risk factors and exercise dose is not adequate to describe the dose-response curve for exercise and mortality. It is clear that there is a dose-response relationship between exercise and fitness, although to what degree exercise dose prescriptions should target objective physical fitness goals remains unknown.

Mortality

To date, no RCTs have been designed to examine the impact of PA on mortality. However, a compelling indirect relationship between routine PA habits and mortality has been clearly demonstrated. For example, the Harvard Alumni Health Study, which began in 1962 with the enrollment of ≈17 000 healthy male alumni and prospectively followed up this cohort for >30 years, demonstrated a direct and graded relationship between exercise dose (from 500 to >3500 kcal/ wk) and mortality. Importantly, the most marked reduction in risk was associated with moving from the <500- to the 500- to 999-kcal/wk PA category. Numerous other high-quality observational studies have similarly shown that increasing routine PA is associated with reductions in CVD risk and mortality in a dose-response fashion.^{11,81,83–89}

Exercise Versus Total PA: 30 Minutes Versus 10 000 Steps

Exercise, an important subset of PA, is relatively easy to quantify and thus has historically been the primary focus in studies examining PA. There is now growing interest in defining the relationship between total daily PA, including but not limited to dedicated exercise dose, and key health outcomes. The concept that “nonexercise” PA (nonmotorized commuting, occupational activity, etc) beneficially affects health has been demonstrated by several recent studies that show strong associations between time spent sitting and mortality⁹⁰ and routine standing breaks interspersed within typically seated routine activities of daily living.⁹¹ In addition, mortality risk among individuals who routinely meet or exceed current PA guidelines through structured exercise may be significantly moderated by PA habits during the nonexercise portions of the day.^92,93

Advances in our understanding of total PA exposure have been facilitated by technology. Commercially available fitness monitors designed to estimate daily energy expenditure or to measure daily step counts have become widely available. Although most of these devices accurately measure step counts during exercise that involves walking or running, they may be less accurate for step accrual during nonexercise PA and may significantly underestimate and overestimate daily energy expenditure without predictable variation patterns.⁹⁴ In addition, step counts provide no information about the intensity during which steps are accumulated and thus are not useful for accurately determine daily energy expenditure or exercise dose as defined in research settings. In clinical practice, a common recommendation is to aim for 10 000 steps per day. This recommendation originates from a Japanese pedometer developed in the 1960s that was named and marketed as the “10 000 steps meter,” not from any rigorous science linking this step count with clinical outcomes. Although there is now emerging literature supporting a link between daily PA in this approximate range and beneficial health outcomes,⁹⁵ more work is required to define the optimal daily step count and other similar metrics of nonexercise PA. In the interim, we support the use of fitness monitors in clinical practice because patients often find that they provide motivation that translates into substantive increases in overall PA dose.⁹⁶

Current Exercise Recommendations: The Data Behind the Dose

As evidence supporting the link between exercise and favorable health outcomes continued to mount, 3 independently generated PA guidelines were published in the 1990s to 2000s. Writing groups representing the American College of Sports Medicine and the Centers for Disease Control, the National Institutes of Health, and the US Department of Health and Human Services (HHS) each concluded that routine moderate-intensity exercise was an effective means to reduce the overall risk of chronic disease. The consistent recommendation across these initial guidelines was that all people should engage in at least 30 minutes of moderate-intensity exercise on most, preferably all, days of the week.^97–99 This approximate recommended dose remains the cornerstone of public health PA guidelines. The most recent guidelines by the HHS in 2008 and the World Health Organization in 2010 suggest a total of at least 150 minutes of weekly moderate-intensity activity (3–6 METs) or ≈450 to 900 MET-min/wk.^14,100 More recent guidelines by the American Heart Association and American College of Cardiology recommended a similar total amount of exercise, although divided into sessions of longer duration (40 versus 30 minutes) and performed less frequently (3–4 rather than 5 sessions per week), on the basis of their review of the literature specific to reduction in BP and improvement in lipids.³⁰ Walking at 3 mph (3.3 METs) for 150 minutes achieves the minimum target of ≈500 MET-minutes that all of these guidelines have in common. However, for those who choose higher-intensity activity (>6 METs), the HHS guidelines endorse a shorter total duration of exercise. For example, jogging at 6 mph (10 METs) for a weekly total of 75 minutes amounts to 750 MET-minutes, thus meeting the dose target. It is noteworthy that the HHS guidelines recognize an exercise dose of 450 to 900 MET-min/wk as the minimum necessary to promote health and acknowledge the potential role of higher total doses and intensities of exercise for individuals with specific goals, including fitness optimization or weight management.

A strong literature base supports the 2008 HHS recommendation for an exercise dose of 450 to 900 MET-min/wk. Authors of these guidelines reviewed a total of 73 studies published from 1995 to 2007, of which the majority were prospective, observational studies.¹⁴ This aggregate cohort included >1 million subjects with substantial age, ethnicity, and sex heterogeneity and captured ≈140 000 deaths over a median follow-up of ≈12 years. The overall relative risk of death after adjustment for measured confounders was 0.69 (ie, 31% risk reduction for death) when the most active subjects were compared with the least active subjects. Findings were similar for men and women, and risk reduction (0.56) was even greater in individuals >65 years old.^101–109

The specific dose recommendation of 150 minutes of moderate-intensity weekly exercise was based on the observation that 2 to 2.5 hours of moderate-intensity PA appeared to be the minimum amount required reduce the all-cause mortality rate. However, careful inspection of the exercise dose–mortality relationship does not demonstrate a meaningful inflection point at this specific dose of exercise (Figure 1). On the basis of data from 11 large, epidemiological studies that each contained at least 5 levels of exercise dose and in which exercise dose could be converted to a common metric of hours per week of moderate to vigorous exercise, the exercise dose-response mortality curve appears curvilinear, with the largest risk reduction occurring at the lowest end of the exercise dose spectrum. Compared with an essentially sedentary lifestyle (<0.5 h/wk of moderate to vigorous exercise), 1.5 hours of moderate- to vigorous-intensity exercise is associated with a 20% risk reduction in mortality. To attain an additional 20% risk reduction (for a total of 40% risk reduction in mortality), an additional 5.5 hours of moderate to vigorous exercise was required for a total of 7 h/wk. At the range of exercise examined (up to 7 h/wk of moderate to vigorous exercise), mortality risk continued to decrease as dose increased. Although increasing exercise doses were associated with diminishing returns, there was no exercise dose at which more did not yield progressive benefit. It is important to note that the methods used to produce these estimates preclude any statistical evaluation of the shape of the curve. Therefore, these data do not permit a determination of the minimum dose required for benefit or the maximum dose at which point the added benefit of exercise may end. Since the publication of the 2008 HHS guidelines, several important, large, epidemiological studies have further addressed both ends of the exercise spectrum, as discussed below (Table 3).^110–112

• Download figure
• Download PowerPoint

Low-Dose Exercise: “Some Is Better Than None”

Current PA recommendations were designed to optimize health outcomes, not to determine minimum exercise doses that afford tangible benefit. It is becoming increasingly recognized that relatively small doses of routine exercise have the potential to confer significant health benefit. In the Nurse’s Health Study, a study of middle-aged and older women, a significant reduction in all-cause mortality was found with as little as 1 to 1.9 hours of weekly moderate to vigorous exercise (adjusted relative risk [RR], 0.82; 95% confidence interval [CI], 0.76–0.89), and further increases in exercise dose of up to ≥7 h/wk were associated with only minimal (and statistically insignificant) additional risk reduction.¹⁰⁶ A study of elderly men similarly found that as little as 1 hour of walking or cycling per week was associated with reduced all-cause mortality risk (adjusted relative risk, 0.69; 95% CI, 0.58–0.88), whereas another investigation of elderly people found that even being an “occasional” exerciser (<1 time per week) conferred a mortality risk reduction of 28%.¹⁰⁴ The risk of CVD (coronary disease and stroke) has recently been shown to be similarly reduced by low-dose exercise in older adults, with those walking as little as 6 to 12 blocks per week having significantly reduced risk of incident CVD (hazard ratio, 0.7; 95% CI, 0.58–0.83).¹¹³

Several recent large, prospective, cohort studies in adult populations with broader age ranges have found benefit from small doses of exercise. In a large cohort (661 137 participants pooled from 6 individual studies) with 116 686 deaths over a median follow-up of 14.2 years, small amounts of exercise (0.1–7.5 MET-h/wk) conferred a 20% relative mortality risk reduction (hazard ratio, 0.80; 95% CI, 0.78–0.82) compared with a sedentary lifestyle.¹¹² This study also found that relative death risk continued declining by an additional 9% with exercise levels approximating current guidelines (7.5–15 MET-hours) and an additional 9% with doses of exercise that exceed guideline recommendations by 3 to 5 times (39% risk reduction, 22.5–45 MET-hours). In a separate study of Taiwanese adults, a cohort suggested by the authors as having lower baseline PA than Western adults, subguideline exercise doses (4.5 MET-hours or 90 minutes of moderate activity per week) were associated with a significant reduction in all-cause mortality risk of 14% (relative risk, 0.86; 95% CI, 0.81–0.91) and improvement in life expectancy by 3 years.¹¹⁰ In summary, exercise doses well below those proposed by current guidelines appear to have a significant impact on health and longevity. In clinical practice, this translates into the important concept that it is the least active individuals who stand to benefit the most from even the smallest increments in exercise and PA.

High-Dose Exercise: “More May Not Be Better but Is Not Necessarily Bad”

A growing segment of the global population chooses to participate in high levels of strenuous exercise for myriad reasons, including athletic performance, socialization, and weight management. Epidemiological data examining prior elite athletes, a group that by definition engages in exceptionally high exercise doses with respect to frequency, duration, and intensity during their first few decades of life, consistently document desirable late-life outcomes. Specifically, previously elite athletes appear to use comparatively less hospital resources¹¹⁴; require fewer asthma, cardiovascular, and anti-inflammatory medications¹¹⁵; and live longer than nonathletic counterparts.^116–118 Despite these compelling prior studies, there has been recent concern that high levels of strenuous exercise may do more harm than good. It has been suggested that the upper end of the exercise dose–response curve for mortality may be reverse J or even U shaped with the highest exercise doses reducing or completely eliminating the tangible health benefits afforded by lower doses. The contemporary basis for this largely theoretical concern stems from several lines of evidence, including recently published epidemiological, exercise physiology/cardiac biomarker, and cardiac structure/function data, as recently summarized and debated.^119,120

Assessment of mortality risk for the upper end of exercise dose is hindered by the fact that most population-based cohort studies have relatively few individuals who exercise at doses that substantially exceed PA recommendations. In the Copenhagen City Heart Study, an observational study that demonstrated favorable longevity among light- and moderate-dose joggers, individuals who were “strenuous” regular joggers had a higher hazard ratio for death (hazard ratio, 1.97, 0.48–8.14) than the reference group of sedentary control subjects. This observation raised concerns about a possible U-shaped association between mortality and jogging dose.¹²¹ However, as reflected by the wide and overlapping CIs, this conclusion is not supported by statistical significance owing to low representation of strenuous joggers in the study group (4%) and a correspondingly small number of deaths (n=2) during the follow-up period.

Two recent studies have attempted to more accurately define the risk of mortality at the upper end of exercise dose. The Aerobics Center Longitudinal Study evaluated all-cause and CVD mortality in >13 000 runners over a mean follow-up of 15 years.¹¹¹ Using quintiles of weekly running dose and thereby analyzing groups with equal participant numbers, the investigators found that the jogging dose and mortality curve was J shaped, not U shaped. This finding reinforces the notion that light to moderate doses of exercise have a substantial positive impact on health but that continued dose escalation appears neither incrementally better nor worse. Similarly, Arem et al¹¹² recently pooled 6 separate studies with exercise dose data from the National Cancer Institute Cohort Consortium to amass a cohort of >600 000 participants. The most active group (n=4077) reported exercise levels that were ≈10 times the levels recommended by HHS guidelines (ie, 75 MET-h/wk of activity, >176 min/wk of running) and were found to have lower rates of mortality than sedentary people and mortality rates statistically similar to those who were exposed to more moderate exercise doses. Careful inspection of the group at the highest exercise dose in this study reveals a slight trend toward increasing mortality and a widening of the CI around this point estimate. Although this may again reflect a relatively smaller sample size compared the group at the next lowest exercise dose (40–75 MET-h/wk of activity, n=18 831), it is possible if not probable that there is a heterogeneous response to such high levels of exercise. In summary, the majority of currently available data suggest that high-dose exercise is associated with lower risks of CVD and mortality than a sedentary lifestyle. Whether some portion of the protective benefits of exercise is lost at the highest ends of the dose spectrum remains uncertain. If benefit attenuation at high exercise doses does indeed occur, there is an emerging list of potential mechanistic mediators that have arisen from recent observational work.

It has been suggested that high-dose exercise may cause or accelerate coronary atherosclerosis on the basis of 1 study that demonstrated higher coronary artery calcium scores in older marathon runners than among sedentary control subjects matched for Framingham Risk Score.¹²² It is notable that more than half of the marathon runners in this study were former smokers and thus had a plausible explanation for the presence of coronary atherosclerosis independently of marathon training. In our opinion, it is unlikely that high-dose vigorous exercise initiated the process of coronary atherosclerosis but very possible that it potentiated the atherosclerotic process by inducing repetitive hemodynamic arterial shear stress or a chronic proinflammatory state. To what degree high-dose exercise modulates the underlying biology of atherosclerosis remains a complex and poorly understood topic.

There have now been numerous observational reports of cardiac troponin elevation after long-duration endurance races.¹²³ These observations have led to the hypothesis that such sporting events, particularly when engaged in repetitively over many years, may precipitate permanent myocardial damage in the form of fibrosis.^124,125 However, studies pairing cardiac imaging with necrosis marker assessment have not consistently shown an association between troponin elevation and acute or chronic myocardial dysfunction,^126–129 and carefully controlled laboratory-based exercise testing has demonstrated that acute exercise-induced troponin elevation is a nearly ubiquitous phenomenon among healthy subjects.^130,131 Benign and perhaps adaptive explanations for exercise-induced troponin elevation and alternative pathogenic mechanisms for ventricular fibrosis among athletes have been proposed.¹³² It is most likely that high-dose exercise injures the heart muscle infrequently and only among athletes with an additional susceptibility factor such as subclinical myocarditis, surreptitious performance-enhancing drug use, or occult genetic susceptibility.

Although it remains unclear whether vigorous, long-term endurance exercise may accelerate atherosclerosis or cause adverse cardiac remodeling in some people, it is well-established that acute bouts of physical exercise transiently increase the risk of sudden death.^133,134 The risk of sudden death during exercise appears to be highest among people harboring underlying genetic or acquired CVD,^135–137 and preliminary observational data suggest that vigorous exercise may accelerate the phenotypic expression of arrhythmogenic right ventricular cardiomyopathy.^138,139 These findings emphasize the potential importance of conservative exercise dose recommendations among patients with established CVD in secondary prevention clinical settings. As recently reviewed, cardiac rehabilitation programs provide a useful resource for starting and tailoring an exercise program in patients with established CVD.² However, the risk of an acute cardiac event during exercise appears coupled with an overall reduction in rates of sudden death among those who engage in the highest levels of PA and exercise. This apparent exercise paradox must be considered when counseling avid exercisers about the inherent risks and benefits of high-dose exercise.

Clinical Implications: Exercise Dose in the Clinical Practice

Careful consideration of data defining the relationships between exercise dose and clinical outcomes is required for meaningful application of exercise dose counseling in clinical cardiovascular practice. In our clinical cardiology practice, we encounter the entire spectrum of PA, from completely sedentary patients to highly active elite athletes who seek care in our sports cardiology program. We assess PA habits and exercise dose exposure with each patient on a routine basis. As previously suggested,¹⁴⁰ we endorse the use of a PA vital sign that captures each element of exercise dose, including intensity, duration, and frequency. These data can be accurately ascertained, rapidly quantified, and recorded in the medical record. We recommend using data reflected in the PA vital sign to classify patients into 1 of 3 PA categories: routinely fails to meet PA recommendations, routinely meets PA recommendations, and routinely exceeds PA recommendations. An algorithm for downstream management considerations based on this simple classification scheme is shown in Figure 2. Evaluating PA in this manner during every clinical visit may be difficult because of competing priorities. Thus, we advocate addressing PA at least annually and more often in those who continually fail to meet PA recommendations.

• Download figure
• Download PowerPoint

Patients who routinely fail to meet PA recommendations represent the clinician’s greatest opportunity for clinical outcome–directed exercise dose counseling. This discussion is often best initiated by addressing the numerous risks associated with physical inactivity. On a patient-by-patient basis, it is crucial to address both perceived and objective barriers to performing adequate doses of PA. Perceived barriers to increasing PA, including time constraints, lack of motivation, and fear of injury or adverse outcomes, are common but amenable to revision through discussion. Objective barriers, including orthopedic limitations and physiological limitations imposed by medications or comorbid disease, are similarly common and represent areas for directed interventions in the form of referral to clinicians with musculoskeletal expertise and tailored pharmacotherapy with an emphasis on maximizing exercise capacity. Following these steps, it becomes possible to devise a PA/exercise dose recommendation that is geared to eventually meeting current PA recommendations. As discussed above, the use of wearable fitness trackers has become increasingly common. These devices provide data defining PA habits of acceptable accuracy for some types of activity,¹⁴¹ although it remains to be seen whether they can effect long-term change in exercise habits. In our experience, the use of wearable devices coupled with the provision of quantitative exercise prescriptions is often helpful to motivate patients to meet their individualized PA goals. This process often requires multiple clinic visits and may best be accomplished by a team approach that includes physicians and advanced care providers, including nurses, physician assistants, and exercise physiologists.

Patients who routinely meet PA recommendations deserve positive reinforcement. The process of reinforcement almost uniformly contributes to positive patient-provider interactions and ultimately may facilitate long-term PA recommendation compliance. We encourage routine discussions about key ingredients for success and subsequent documentation of these findings. In our experience, a priori knowledge of patient PA patterns is of tremendous value in subsequent clinic interactions when PA/exercise habits may have regressed to suboptimal levels.

Patients who routinely exceed PA recommendations are increasingly common both in general cardiovascular practice and in specialized sports cardiology programs. These patients often present with distinct psychosocial profiles, medical concerns, and atypical disease presentation. Despite the fact that routine exposure to high exercise doses confers relative protection from CVD, it must be emphasized that no level of exercise confers immunity. It is essential to avoid the tendency to minimize the importance of subjective complaints, particularly those of exertional symptoms and relative reductions in exercise, among patients who are capable of high levels of exercise. In this patient population, atypical symptoms and small but significant decrements in exercise capacity are often indicators of evolving CVDs.

With patients who chose to exceed PA recommendations, we routinely discuss the concept of diminishing returns with respect to health outcomes. Practically, exercising at the extreme high end of the exercise dose spectrum is not required for health optimization. We routinely approach patients who choose to exercise at high doses with an open discussion about both theoretical and data-driven risks and benefits. In essence, this conversation is about potential tradeoffs, including balancing negative attributes such as increased risk of atrial fibrillation with positive attributes such as traditional risk factor management and in some cases overall quality of life. In our experience, patients who chose to exercise at high doses do not uniformly adapt other healthy lifestyle choices. Therefore, we encourage active screening for unhealthy dietary intake, excessive alcohol consumption, and unchecked exposure to psychological stress, as we would in any other patient population.

Conclusions

Decades of scientific inquiry have led to the indisputable fact that routine exercise or high levels of PA confer positive cardiovascular health benefits. Exercise and PA, much like medications used in clinical practice, are best measured and prescribed by consideration of dose, which is a function of 3 principal attributes: intensity, duration, and frequency. PA guidelines are based on a solid epidemiological foundation, and numerous RCTs have begun to delineate the mechanisms by which exercise leads to health and longevity. In clinical cardiovascular practice, exercise dose and PA habits can and should be addressed with each patient with an ultimate goal of individualized counseling and exercise prescription. Although progress has been made, there is much to be learned. Refinements in our understanding of how exercise dose across the spectrum affects cardiovascular health are needed. Future gains will best be accomplished by the use of complementary strategies that include prioritized scientific funding, widespread application of technology designed to measure exercise dose both in clinical trials and in real-world living, and focused translational work geared toward delineating cellular and biochemical responses to exercise.

Footnotes