Comparing the rate of serious adverse events in hypertension trials with rates of similar events in routine clinical practice: are older people in trials representative of the target population?

Problem

The optimal treatment of hypertension in older people in uncertain. Part of this uncertainty stems from concerns that randomised controlled trials, on which treatment guidelines are based, are not representative of older people. Representativeness of ‘standard’ antihypertensive trials is often questioned, with limited recruitment of older-people. Some trials specifically recruit older participants to address this. If such older-people’s trials are representative, we would expect rates of hospitalisation and death in each trial to be similar to community rates, and higher than rates in standard trials.

Approach

We identified trials of Renin-Angiotensin-Aldosterone system (RAAS) drugs for hypertension using the clinicaltrials.gov registry, and divided these into standard trials (recruiting a general adult population) and those focusing on older people (minimum age 60 or higher). Serious Adverse Events (SAEs) are routinely included in trial reports and are predominantly accounted for by all-cause hospitalisations and death. We extracted SAE rates in older-people’s and standard trials, adjusting for trial characteristics (phase/drug/comparison/outcome). Using primary care data from SAIL databank, we identified a community cohort of adults with hypertension commencing similar drugs to obtain an expected rate of hospitalisations/deaths, and compared this to observed SAE rates in each trial.

Findings

We included 110 trials: 11 older-people’s trials exclusively recruited people over 60 years; 99 standard trials also included younger people. Older-people’s trials had higher SAEs rate than standard trials (IRR 1.74, 95% CI 1.03-2.92). However, when comparing the SAE rate in the trials to the rate of hospitalisations or deaths in people with hypertension starting RAAS drugs in the community, the hospitalisation and death rate in the community was much greater than the rate of SAEs reported in standard trials (ratio 4.17 (95% CI 3.45-5.26)) and older-people’s trials (4.76 (95% CI 2.86-8.33)), adjusting for age and sex.

Consequences

Our findings demonstrate that people in hypertension trials experience substantially lower rates of adverse health outcomes than people with hypertension treated with similar drugs in the community. This adds weight to the body of evidence showing that hypertension trials are under-representative of their target populations. However, our findings also add nuance to this statement, as trials focussing on older people do have a significantly higher rate of SAEs than standard trials. Therefore, trials focussing on older people do, at least in part, reflect the increased risk of adverse outcomes seen in older populations. Our findings also indicate that SAE rates should be considered as a novel metric with which to assess the representativeness of trial populations, through comparison with the incidence of similar events in routine clinical care. Such an approach could facilitate more direct quantification of the consequences of trial under-representativeness.