Data for Can lay health workers support the management of hypertension? Findings of a cluster randomised trial in South Africa

A pragmatic cluster randomised trial with a process evaluation in eight rural clinics assessed the effect of adding two LHWs supporting nurses in providing chronic disease care in each intervention clinic over 18 months. Control clinics continued with usual care. The main outcome measure was the change in the difference of percentage of clinic users who had elevated cardiovascular risk associated with high blood pressure (BP) before and after the intervention, as measured by two cross-sectional population surveys.

Results There was no improvement in BP control among users of intervention clinics as compared with control clinics. However, the LHWs improved clinic functioning, including overall attendance, and attendance on the correct day. All clinics faced numerous challenges, including rapidly increasing number of users of chronic care, unreliable BP machines and cuffs, intermittent drug shortages and insufficient space.

Conclusion LHWs improved the process of providing care but improved BP control required improved clinical care by nurses which was compromised by large and increasing numbers of patients, the dominance of the vertically funded HIV programme and the poor standards of equipment in clinics.<br><br>Data provided consists of: 1. A Stata data file: nkateko trial final data.dta 2. Two pdf files with the before and after survey questionnaires: a.

Nkateko-Surveillance-V6-20130927.pdf (first survey) b. Nkateko-Surveillance-V10-20150819.pdf (second survey) The samples for both surveys were drawn from the census database maintained by the Agincourt Health and Demographic Surveillance System and demographic data were drawn down from the census and added to sample file. These variables are usually indicated by the variable label.

Apart from a few created variables such as age and age group, all the other data come from the two surveys and the variable names indicate which question in the survey the data comes from. The variable “survey_round” indicates which survey information is drawn from.