[Comparison of different methods in the assessment of the ratio of hypertensive patients responding to antihypertensive treatment]

Identification of a responder to an antihypertensive treatment is helpful and also very difficult in an open uncontrolled study. The existence of many arbitrary methods to define a responder, the placebo effect and the regression to the mean phenomenon coexist and give spurious interpretation. We analysed an open study recruiting 6,530 hypertensive patients (DBPor = 95 mmHg at entry), treated with the same antihypertensive drug during 3 months. Mean initial and at 3 months blood pressures were respectively 178/101 mmHg and 151/86 mmHg. We observed significant variation in the % of responders (20 to 80%) to a drug when applying different methods. Placebo effect and regression to the mean lead to an overestimation. Without a placebo controlled design, we suggest to increase the BP level for inclusion to determine the % of responders with the standard goal pressure (DBP90 mmHg) and to use Oldham's method for adjusting for the regression to the mean unless the coefficient of correlation found is around zero and if upper level of DBP is not selected for inclusion.