To identify the features of an optimal exercise programme, in terms of the type of exercise, intensity, frequency, and duration that would maximise the training-induced decrease in blood-pressure (BP).

MEDLINE, EMBASE and the Science Citation Index were searched from 1980 to 1995. Previous review articles and the bibliographies of the retrieved studies were also examined. Only artciles written in the English language were sought.

Study designs of evaluations included in the review

Randomised controlled trials (RCTs) with explicit statement of random allocation, with BP as a primary or secondary outcome measure, and a description of the intensity of the exercise programme, were included. Trials with a crossover design were included provided that the order of treatments was randomised and the order effects were not significant.

Specific interventions included in the review

Evaluations of aerobic or resistance training programmes of at least 4 weeks' duration were included. For the purposes of the analysis, the trials were categorised as one of the following: programmes of walking, jogging or running; cycling using a bicycle or cycle ergometer; cycling combined with walking, jogging or running; and resistance training. The duration of the programmes across the trials ranged from 4 to 52 weeks. The control participants were non-exercising; no other details were provided. Trials of exercise programmes combined with dietary interventions were excluded.

Participants included in the review

The participants were sedentary adults of either gender, who were either normotensive or hypertensive, and were healthy apart from hypertension. The age range across the trials was 18 to 79 years.

Outcomes assessed in the review

Changes in systolic and diastolic BP as a result of exercise were assessed.

How were decisions on the relevance of primary studies made?

The authors do not state how the papers were selected for the review, or how many of the authors performed the selection.

Selection bias was controlled at entry to the trial. A 3-point rating scale was applied to each included trial, based on an existing assessment tool (see Other Publications of Related Interest). The process of validity assessment was not described, i.e. how many of the reviewers were involved, and whether the assessment was conducted independently.

The categories of data extracted were listed in the paper. However, the process of data extraction was not described, i.e. how many of the reviewers were involved, and whether the extraction was conducted independently. An effect size was calculated for each trial, by computing the variance of the difference (in mmHg) between the mean change in BP (baseline minus final value, assuming unpaired values) in the exercise and control groups.

How were the studies combined?

The effects of exercise on systolic and diastolic BP were assessed independently. The pooled effect sizes were estimated using both fixed-effect and random-effects models, weighted according to the inverse of the variance of the effect size, and associated 95% confidence intervals (CIs) were calculated.

How were differences between studies investigated?

Tests of heterogeneity were performed using the Mantel-Haenszel method. Funnel plots were constructed to assess the impact of publication bias, with changes in systolic and diastolic BP (in separate plots) plotted against sample size.

Twenty-nine RCTs with 1,533 participants were included.

Data from 27 trials analysed with the fixed-effect model showed that aerobic exercise training resulted in decreases of 4.7 mmHg (95% CI: 4.4, 5.0) for systolic BP and 3.1 mmHg (95% CI: 3.0, 3.3) for diastolic BP, above those achieved in control groups. However, there was significant heterogeneity for both comparisons. Indirect comparisons showed no difference in the effect of the intensity or frequency of the aerobic exercise programme on the decrease in BP.

In 5 trials directly comparing high with low exercise intensity, the decrease was -0.7 mmHg (95% CI: -1.9, 0.4) for systolic BP and 0.1 mmHg (95% CI: -0.7, 1.0) for diastolic BP.

One trial comparing exercise, seven times versus three times per week, found differences of 5.0 mmHg (95% CI: 2.5, 7.5) for systolic BP and 2.0 mmHg (95% CI: -0.2, 4.2) for diastolic BP.

In 3 small trials of resistance training, the decrease in systolic BP was 0.4 mm Hg (95% CI: -0.5, 1.4), whilst that in diastolic BP was 1.5 mmHg (95% CI: 0.8, 2.3).

When the data were reanalysed using a random-effects model, the effect sizes were not quantitatively different from above; however, the associated 95% CIs were consistently wider. The funnel plots indicated that publication bias may be present, since the plots were skewed towards more trials showing larger reductions in BP without the expected number of small trials showing unchanged or increased BP.

Aerobic exercise training was effective in producing a small but statistically-significant decrease in both systolic and diastolic BPs, and these effects appeared to be independent of the intensity or frequency of exercise. Unfortunately, these changes in BP are unlikely to be of any clinical significance amongst most hypertensive patients. In addition, the results should be interpreted cautiously due to the significant heterogeneity between the trials. There was insufficient evidence to make any definitive conclusions about the effects of resistance training.

The research question and the selection criteria for the trials were clearly explained, with details of the primary material tabulated. The methods used for pooling the data were appropriate. The methods used for assessing study validity were clearly described; however, only selection bias was considered as a criterion, and it would have been informative if other aspects of study quality had also been assessed and reported.

The search strategy was limited in that specialist sources (i.e. those pertaining to exercise) do not appear to have been accessed, and the search was restricted to English language articles only. Therefore, it is possible that relevant material may have been missed. However, the authors did address the issue of publication bias, and reported an estimation of the possible effects of this. More information on the review process would have been useful, i.e. how many of the reviewers were involved, whether decisions were made independently, and how any disagreements were resolved.

Overall, the points in the authors' discussion and the conclusions are appropriate. In addition, the authors rightly draw attention to the significant heterogeneity between the trials, and the issue of whether the estimated decreases in BP were of clinical significance. Separate analyses of normotensive and hypertensive participants may have been useful.

The authors state that further research is required to identify the optimal frequency of exercise training.

Sackett DL, Oxman AD, editors. Cochrane Collaboration Handbook. Oxford: The Cochrane Collaboration; 1995.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.