The beneficial effect of pharmacological treatment of hypertension is well established. However, the outcome may depend on the specific antihypertensive drug (AHD) used. Adverse outcomes, with respect to morbidity and mortality, were reported for short-acting nifedipine in 1995 (reference 1). In the next year, an increased cancer incidence was observed in users of calcium channel blockers (CCBs) including verapamil, diltiazem and nifedipine (reference 2). Furthermore, a recent meta-analysis of randomised clinical trials found a significantly lower preventive effectiveness against myocardial infarction and congestive heart failure among patients allocated to intermediate- and long-acting CCBs than among those on other AHDs (reference 3). The aim of the present study was to compare mortality in hypertensive patients on CCBs with that in hypertensive patients on other AHDs.

In the County of Skaraborg, Sweden, an intervention program to improve the control of hypertension in the population was launched in 1977. The project included recommendations for detection, diagnosis, treatment and follow-up of hypertensive patients. Special out-patient clinics were established in the primary care units in six communities including Skövde.

All patients at the Skövde out-patient clinic having ongoing treatment with AHDs and with a diagnosis of hypertension who underwent an annual check-up during 1989 or 1990 were selected for this study.

Clinical information including co-morbidity was derived from the electronic patient records. The Swedish mortality register was used to identify fatal events and cause of death from 1989 through 1995.

Patients on CCBs at inclusion was classified as users of CCBs. Patients on other AHDs at inclusion, but not on CCBs, were classified as users of other AHDs.

Proportional hazard models (Cox regression analysis) were used to estimate and compare hazard ratios for mortality in CCB users and in users of other AHDs. Baseline differences in the presence of possible risk factors were accounted for by multivariate adjustment. Thus, hazard ratios were adjusted for sex and age (Model A) and for sex, age, number of AHDs used, systolic blood pressure, history of myocardial infarction, angina pectoris, cerebrovascular disease, intermittent claudication, cancer, and type 2 diabetes (Model B).

Basic characteristics are given in Table 1. Patients on CCBs were more often men, were more often smokers, more often had a history of angina pectoris, myocardial infarction and type 2 diabetes, were on a greater number of AHDs, and had marginally lower blood pressure.

After adjustment for differences in age and sex distribution, mortality from all causes, from IHD and stroke combined, and for causes other than cancer, IHD and stroke combined, was significantly higher in CCB users than in users of other AHDs; hazard ratios were 3.0 (95% CI 1.7 - 5.3), 3.2 (95% CI 1.3 - 7.8) and 4.0 (95% CI 1.3 – 12), respectively.

After adjustment for differences in age and sex distribution, smoking habits, number of antihypertensive drugs used, systolic blood pressure, history of myocardial infarction, angina pectoris, cerebrovascular disease, intermittent claudication, cancer and type 2 diabetes, overall mortality was significantly higher in CCB users; adjusted hazard ratio 2.6 (95% CI 1.3 - 5.0). For mortality from cancer, the difference was of borderline significance; hazard ratio 3.2 (95% CI 1.0 – 11). Mortality from IHD and stroke combined did not differ significantly after adjustment for possible confounders; hazard ratio 2.6 (95% CI 0.9 - 7.9).

Like several previous studies, the present study indicated a less favourable outcome in hypertensive patients on CCBs than in those on other AHDs; overall mortality was significantly higher in those on CCBs. Although CCB users had a higher level of risk factors at baseline, the hazard ratios were only marginally influenced by adjustment for these possible confounders, including co-morbidity.

Although the results were adjusted for co-morbidity and other possible confounders, it cannot be excluded that residual confounding in some way affected the results. Thus, the higher average number of AHDs used by the CCB-users may indicate more advanced hypertension. However, the hazard ratios changed only marginally when adjusted for possible confounders (Table 2, Model B).

Due to highly systematic journal records employed at the out-patient clinic, there is no ground to assume inferior validity of the information on co-morbidity. Accordingly, confounding by indication is a less probable explanation of the differences in mortality.

A second limitation is that patients were classified as users of CCBs and other AHDs, respectively, based only on status at inclusion. In some patients, the actual treatment may have been discontinued, or additional AHDs may have been added. It is quite possible that patients classified as users of other AHDs may have had CCBs added later. Thereby, differences in mortality rates due to the different treatments may have been underestimated.

It appears that CCBs have little or no advantages over other AHDs. Rather, both clinical trials and observational studies, including the present one, indicate that CCBs may confer higher mortality and cardiovascular morbidity than other AHDs. Therefore, CCBs should not be regarded as first-line drugs unless efficacy, effectiveness and risk equality with other AHDs is proven by future studies.

Table 2. Cause-specific mortality (Hazard Ratios and 95% Confidence Intervals) in 164 users of CCBs and in 555 users of other AHDs as reference.

1. Model A adjusted for sex and age.

2. Model B adjusted for sex, age, smoking, number of AHDs used, systolic blood pressure, history of myocardial infarction, angina pectoris, cerebrovascular disease, intermittent claudication, cancer and type 2 diabetes.