Epidemiological transitions are occurring throughout Africa. To inform public health programs and policies, longitudinal cohorts investigating non-communicable diseases are needed. However, loss-to-follow up is a major problem. In preparation for a longitudinal study, we conducted a randomized controlled trial to test communication-based retention strategies (message content and delivery methods) among a pilot cohort of South African healthcare workers (n = 1536; median age = 36; women = 1270). Two messaging formats across three delivery modes were tested. Response rates were analyzed by intervention, survey return date and method using chi-square tests and univariate logistic regression. Sixty-seven of 238 (17.4%) control group participants and 238 of 1152 (24.6%) intervention group participants were retained (OR 1.54: CI 1.15–2.07; P = 0.004). Odds of being retained were 1.68 times greater for participants who received regular contact and themed messages compared to control (CI 1.22–2.32; P = 0.001). Neither health status nor clinical condition affected response rates (P>0.05). Time-to-first contact did not impact response rates (P>0.05). Message content and delivery method influenced response rates compared to the control, however no difference was found between intervention groups. Although greater retention is required for valid cohort studies, these findings are the first to quantitatively assess retention factors in Africa
S1 Appendix. Follow-up questionnaire sample.
S1 Table. Differences in health status between responders and non-responders.
S2 Table. Differences in method of survey return between interventions.
S3 Table. Differences in response by email provided and method of survey return.
S4 Table. Differences in length of time to first contact between responders and non-responders.
S5 Table. Differences in response within interventions associated with response to first Madmaker.