Abstract
There are substantial logistical difficulties in conducting community surveys of minority ethnic group populations. Primary care lists have been identified as an important potential resource but the representativeness of samples derived through this method has received little evaluation. In a community survey of psychiatric morbidity, African-Caribbean people aged 55-75 were identified by practice staff from registration lists for seven primary care teams in south London. The sensitivity of the process was evaluated by contacting a random sample of people whose ethnicity was not known. Participants aged 65-75 (n = 174) were also compared to a similarly aged group sampled through household enumeration (n = 34) with respect to demographic factors, risk factors for vascular disease, depression and cognitive function. For those with correct addresses, the identified group was estimated to include 72% of the potentially eligible population. Only 8% of contacted people were found not to be eligible in terms of ethnicity. Compared to the household enumeration sample, the primary care sample had marginally higher socio-economic status but was similar with respect to all other measured characteristics. Primary care list sampling with staff-assigned ethnicity therefore appeared highly specific, reasonably sensitive, and did not seem to introduce substantial bias for this population.