Systematic review highlights the dearth of evidence on exercise in reducing mortality

We know that people who exercise a lot live longer.  But how does exercise match up with traditional drug treatments?  An open-access systematic review in the BMJ synthesised the experimental evidence for exercise compared with drug treatments on mortality.  This paper garnered a lot of media attention.  Does the hype match up with what the evidence actually says?

Clinical question

What is the comparative effectiveness of exercise and drug treatments in reducing mortality?


Reading a big book

First the reviewers identified conditions for which there was evidence for exercise.  Then they compared exercise with drug treatments in those conditions.

The literature search looked for meta-analyses of RCTs of exercise interventions. For those conditions with evidence about the effectiveness of exercise, they carried out additional searches for meta-analyses of drug interventions and for RCTs that directly compared exercise with drug treatments.

As this was a complex analysis of meta-analyses of very different studies, a host of statistical techniques were deployed to assess the relative treatment effects and potential impact of heterogeneity.


Four meta-analyses were found that looked at the effect of exercise on mortality.  These focused on secondary prevention of coronary heart disease, stroke, heart failure and pre-diabetes.  For these conditions, 12 meta-analyses were found on the various drug interventions.

No RCTs were found that directly compared exercise with drug treatments for these conditions.

In coronary heart disease, statins, beta-blockers, ACE inhibitors and antiplatelets were all effective at reducing mortality.  The data for exercise were not statistically significant.

walking and cycling

The review was limited by the variety of different exercise interventions and lack of data on effectiveness

In stroke, exercise was significantly beneficial, albeit based on a small data set.  No evidence of benefit was found in pre-diabetes.

The overall conclusion of the study was that exercise could potentially offer similar benefits to drug treatments, but there isn’t enough evidence to be sure.  Although these results are promising, the reviewers commented:

these findings should be interpreted with caution given the scarceness of data and the different settings involved.


  • The literature search was limited to Medline or the Cochrane Library
  • It’s not clear whether there was blind, independent data extraction
  • The data for stroke should be interpreted with caution because of the small event rates in the trials cited
  • There was a stark contrast between the quantity of data on drugs compared with exercise
  • There was also important limitations in how exercise interventions were described and delivered.


Naci H, Ioannidis JP.  Comparative effectiveness of exercise and drug interventions on mortality outcomes: metaepidemiological study.  BMJ 2013;347:f5577.

Media coverage: