Abstract & Commentary
Take Two Sneakers and Call Me in the Morning: Drugs, Exercise, and Mortality
By Howell Sasser, PhD
Associate, Performance Measurement, American College of Physicians, Philadelphia, PA
Dr Sasser reports no financial relationships relevant to this field of study.
Synopsis: Investigators combined multiple meta-analyses to assess the relative effectiveness of drug therapy and exercise in reducing mortality from specific causes. Exercise produced a comparable or better effect for all causes.
Source: Naci H, Ioannides JP. Comparative effectiveness of exercise and drug interventions on mortality outcomes:
Metaepidemiological study. BMJ 2013;347:f5577.
In this rather unusual paper, the investigators conducted a meta-epidemiological study, in which multiple meta-analyses are combined to produce an even larger population, ideally yielding even more stable statistical effects. As with meta-analysis, this technique permits the assessment of the relative effects of therapies even if they were not compared head-to-head in individual trials. The investigators identified four meta-analyses summarizing clinical trials of the effect of exercise, and 12 summarizing clinical trials of the effect of drug therapy. They also found and included three reports of clinical trials of exercise published after the most recent meta-analysis.
The studies included in the meta-analyses covered four disease processes — coronary heart disease, heart failure, stroke, and pre-diabetes (a collective term for patients with some but not all defining diagnostic elements of diabetes) — and all had mortality as the outcome of interest. Only one meta-analysis (of pre-diabetes) compared exercise and drug therapy head-to-head. The summed number of study participants randomly assigned to drug therapies (for all conditions) was much larger than the number randomized to exercise (175,043 vs 10,123).
Several drug classes were included for each disease category (for instance, statins, ß-blockers, ACE inhibitors, and antiplatelets for coronary heart disease). In some cases, these were compared head-to-head, but the common denominator in all cases was comparison with a placebo control. The paper presents drug class-specific results, but for simplicity, the results presented here are those summing over all drug classes for a specific condition. Exercise was structured for some diseases (for example, cardiac rehabilitation for coronary artery disease), but was not standardized as to what activities were included or how adherence was monitored. Studies of exercise were of necessity not blinded or placebo controlled. The results of these studies are presented as dichotomous (yes/no) variables.
Summary results are shown in Table 1. Drug therapy was significantly better than placebo control in reducing mortality in coronary heart disease and heart failure, and exercise was significantly better than control in reducing mortality after stroke. Neither drug therapy nor exercise produced a statistically significant reduction in pre-diabetes mortality as compared with control. Most interestingly, exercise was as good or better than drug therapy in reducing mortality in all disease categories, and this difference reached statistical significance for stroke.
Table 1. Change in Mortality Risk with Drug and Exercise Interventions
|
Drug Therapy vs Control |
Exercise vs Control |
Drug vs Exercise |
| Coronary Heart Disease |
↓ S |
↓ NS |
↔ OR 0.94;
95% CI, 0.80-1.11 |
| Heart Failure |
↓ S |
↓ NS |
↔ OR 0.99
95% CI, 0.73-1.36 |
| Stroke |
↔ NS |
↓ S |
↓ with Exercise
OR 8.66
95% CI, 1.28-245.10* |
| Pre-Diabetes |
↔ NS |
↓ S |
↓ with Exercise
OR 1.43
95% CI, 0.81-3.11 |
S = Statistically significant, NS = Statistically non-significant
*Confidence interval is wide because of small numbers in the exercise studies.
Commentary
The size of the populations assembled and the sophistication of the analytical methods leave little to criticize in the way this study was conducted. However, there are a few "ecological" factors that the authors themselves mention and that bear repeating. First, there is little practical or ethical possibility of testing either drug therapy or exercise in perfect isolation. Those taking almost any medication exercise to some extent. There is good evidence that the effects of exercise are cumulative and that exercise need not be organized or concentrated during the day.1 In the same way, those exercising can and do take medications that produce benefits similar or complementary to those of exercise. This means that any positive effect seen in a meta-analysis of either intervention should probably be thought of as a composite of both.
Second, exercise is necessarily a nebulous concept. It is not uncommon to hear physicians and physical therapists speak about "exercise prescriptions," but these are usually personalized to a greater degree than is typical with most medications. The exercise studies included in this report were heterogeneous. There was variation in the mode, frequency, intensity, and duration of exercise, as well as the disease-specific goals it was intended to achieve. By comparison, although the drug studies included multiple agents for each condition, they were relatively homogeneous with respect to how the therapeutic agent was "packaged" and delivered.
Timing also affects comparability. Most participants in the heart disease, heart failure, and pre-diabetes studies had similar disease severity and received the drug or exercise interventions at a roughly comparable time in the disease course. However, the stroke study participants varied widely in severity and this had a significant effect on when they received the study interventions, particularly exercise.
While these factors likely reflect natural conditions, they make a dose-per-dose comparison of drugs and exercise imprecise. This is especially important to bear in mind when interpreting the results for stroke. Exercise was introduced later in the post-acute period, on average, than were drug therapies. This may have predisposed the stroke-exercise group to be healthier — and more likely to survive — simply because they lived long enough to be in the study. On the other hand, it is improbable that those in the exercise group received no potentially beneficial medications before being randomly assigned to exercise. This may make the groups more comparable than is apparent given the simplification of details required for meta-analysis.
What implications does this study have for physicians and patients? It clearly is not a call to abandon statins in favor of aerobics. At the same time, it is a useful reminder that any therapy functions in a matrix of life and lifestyle circumstances, many of which also modify one another and play independent roles in promoting or lessening health. All assets and liabilities should be included when a physician and patient discuss a therapeutic strategy. Benefit from drugs and exercise, used together, may be additive or perhaps even more than the sum of the individual therapeutic parts.
The finding that exercise is at least as effective as drugs in reducing mortality is important, but is only part of the picture. Exercise also plays a role in controlling morbidity. Maintenance of mobility, balance, and coordination has a significant effect on quality of life. This benefit should not be overlooked even if a change in mortality risk is not the main therapeutic goal.
Summary Points
- Drug therapy was significantly better than placebo in reducing mortality in coronary heart disease and heart failure, and exercise was significantly better than non-exercise control in reducing mortality after stroke. Some exercise study participants may have received drug therapy as well.
- Neither drugs nor exercise showed a significant benefit in reducing mortality among those with pre-diabetes.
Reference
1. 2008 Physical Activity Guidelines for Americans. U.S. Department of Health and Human Services. http://www.health.gov/paguidelines/guidelines. Accessed April 9, 2014.
