Forty-two studies (over 111,866 participants), 20 on primary prevention and 22 on secondary prevention were included.
1. Morbidity and mortality (total and CVD-related).
Total mortality was reported as an end-point in 19 studies (n=7 primary prevention and n=12 secondary prevention studies); CVD mortality in 18 studies (n=7 primary prevention and n=11 secondary prevention studies); and morbidity in 17 studies (n=7 primary prevention and n=10 secondary prevention studies). The NNTs for morbidity and mortality (total and CVD) varied widely and most confidence intervals showed an uncertain effect (up to infinity).
For secondary prevention, four studies showed positive effects on total and CVD mortality and one additional study on CVD mortality only. In addition, a positive effect was shown on morbidity in three studies.
Two multifactorial primary prevention studies showed a decrease in morbidity but no effect on mortality was observed.
2. Blood pressure.
Nineteen studies (n=8 primary prevention and n=11 secondary prevention) reported changes in systolic blood pressure and one additional study (multifactorial, primary prevention) reported only changes in diastolic blood pressure. All of the multifactorial primary prevention studies reported a decrease in systolic pressure favouring the intervention group, and the two single primary prevention studies showed mixed results. Overall, the diastolic blood pressure findings showed similar decreases. The pooled mean net change in systolic blood pressure was -3.0mmHg for the intervention and -1.4mmHg for the control; and for diastolic blood pressure -4.2mmHg for the intervention and -3.1mmHg for the control. This was not statistically significant and only clinically modest.
All except one of the 12 secondary prevention studies (n=3 single intervention and n=8 multifactorial studies) showed decreases in both systolic and diastolic blood pressure favouring the intervention group. The pooled mean net change in systolic blood pressure was -2.8mmHg for the intervention and +0.6mmHg for the control; and for diastolic blood pressure -2.9mmHg for the intervention and -1.0mmHg for the control. This was not statistically significant and only clinically modest.
3. Cholesterol.
Twenty-one studies (n=9 primary prevention and n=12 secondary prevention) examined changes in cholesterol levels. Levels decreased in the intervention group of all seven multifactorial primary prevention studies, with two studies reaching clinical significance. One of these studies also reached statistical significance. The one single intervention study also showed a decrease. The mean net change in cholesterol for the multifactorial intervention was -0.36mmol/L (p=0.08) and for all studies was -0.33mmol/L (p=0.08) taking into account the single primary intervention study.
Data from three of the 12 secondary prevention studies could not be evaluated. Two single intervention studies showed a decrease favouring the intervention group while the other single intervention study showed no change. Six out of seven of the multifactorial intervention studies showed a decrease in favour of the intervention group. Overall, the pooled mean net change in cholesterol for multifactorial studies was -0.43 for the intervention and -0.07mmol/L for the control (p=0.007).
4. Weight loss.
Nineteen studies looked at weight loss as an outcome measure. One of the two single primary intervention studies reported weight reduction among the calorie-restricted group. In three of the five multifactorial primary prevention studies a significant weight loss was reported in favour of the intervention group, with the exception of one study group in one trial (mean weight change was -0.9kg vs. +1.2kg, p=0.023, pooled mean net change in weight was >-1kg).
Twelve secondary prevention studies looked at weight loss. Weight loss was reported in the intervention group of seven out of eight multifactorial intervention studies, but only significant in six of the studies. However, the absolute weight loss was less than 1kg and the pooled mean change for intervention participants was -0.6kg vs. +1.3kg for the controls (p=0.026). One single intervention exercise study showed a clinically but not statistically significant decrease in weight favouring the intervention group, as did the remaining multifactorial study.
5. Smoking.
Eighteen studies looking at interventions aimed at smoking cessation. Four of the studies were effective (one person stopped smoking for every 4-14 smokers targeted), but NNTs could not be calculated because of missing data in four of the studies.
6. Alcohol consumption.
One primary and three secondary multifactorial interventions used reported a decrease in alcohol consumption as an outcome. Only one study reported a statistically significant difference between study groups in favour of the intervention group (decrease of 37% vs. 5%, p=0.02). A further three single intervention studies used reduction in heavy drinking as an outcome measure. Two larger studies showed a decrease in the intervention group (statistical significance not stated) and the final small study reported a non-significant difference.
7. Sodium decrease.
Two primary prevention studies showed a significant reduction in sodium excretion, favouring the intervention groups. Six secondary prevention studies looked at sodium decrease. One study had insufficient data for analysis. One reported decreases favouring the intervention group and presented findings for obese and non-obese participants separately. Two multifactorial studies reported statistically significant changes favouring the intervention group, whilst the remaining two studies failed to show any changes in sodium excretion.
8. Changes in exercise habits.
Four studies were of primary prevention and 10 used multifactorial interventions. The outcome measures varied considerably preventing statistical pooling. However, ten of the 16 studies reported positive outcomes.
Eight studies used exercise as an intervention with morbidity and mortality as end points. Two showed an effect on CVD mortality (one on total mortality as well) and one on CVD morbidity.
Quality of the studies.
The size and quality of the studies varied. The type of randomisation was seldom stated. Dropout rates were given in most studies and drop-outs were excluded from the analyses. The results of the studies were reported using either net differences or intergroup differences. Baseline values were not available in any of the studies. Finally, statistically significant differences were well reported, but non significant results were not always mentioned.