Skip to main content


Hypertension, drug treatment, and sub-group analysis: a case for updating Cochrane Handbook?

Date and Location




Sunday 22 September 2013 - 13:30 - 15:00


Presenting author and contact person

Presenting author


Contact person

Abstract text
Background: Mild hypertension has been recently precisely defined as blood pressure (BP) level of systolic 140-159 mmHg and/or diastolic 90-99 mmHg. It represents about half of the hypertensive population. Several old trials suggested that antihypertensive drugs were beneficial “in mild hypertension”. These trials recruited also participants with higher BP levels. We published in 2012 a Cochrane review focused specifically on mild hypertension participants without a previous cardiovascular event. Our review casted doubts about the rationale of currently large prescription habits. Objectives: Our interpretation contradicts the Cochrane Handbook recommendation for subgroup analyses. We discuss here the need for updating the Handbook. Methods: Systematic sub-group review of RCTs. Results: Data on >8,000 people followed for 5 years suggested no reduction in total cardiovascular events, RR 0.97 [0.72, 1.32], in stroke, RR 0.51 [0.24, 1.08] or in mortality RR 0.85 [0.63, 1.15]. This amount of data was limited compared to that available in more severe hypertension and in secondary or high risk prevention, in which a significant benefit was abundantly demonstrated. There was no significant interaction of treatment effect estimates between mild hypertensive and other hypertension categories. Two interpretations were possible : 1) the lack of interaction authorizes generalizing the results to mild hypertensive patients: this is the usual recommended way to interpret the results from subgroup analyses; 2) the lack of evidence within the mild hypertension subgroup prevents prescribing these drugs in these individuals. We adopted this latter interpretation, for two main reasons: 1) The numerical importance of the subgroup, in fact the majority of hypertensive patients; and 2) the lesser benefit to be expected in these lower risk individuals, opening largely the possibility of unfavorable benefit to risk ratio. Conclusions:The lack of certainty regarding the effect of treatment within a numerically dominant sub-group should be disclosed to the patient.