Comparison of patient exit interviews with unannounced standardised patients for assessing HIV service delivery in Zambia: a study nested within a cluster randomised trial

Objectives To compare unannounced standardised patient approach (eg, mystery clients) with typical exit interviews for assessing patient experiences in HIV care (eg, unfriendly providers, long waiting times). We hypothesise standardised patients would report more negative experiences than typical exit interviews affected by social desirability bias. Setting Cross-sectional surveys in 16 government-operated HIV primary care clinics in Lusaka, Zambia providing antiretroviral therapy (ART). Participants 3526 participants aged ≥18 years receiving ART participated in the exit surveys between August 2019 and November 2021. Intervention Systematic sample (every nth file) of patients in clinic waiting area willing to be trained received pre-visit training and post-visit interviews. Providers were unaware of trained patients. Outcome measures We compared patient experience among patients who received brief training prior to their care visit (explaining each patient experience construct in the exit survey, being anonymous, without manipulating behaviour) with those who did not undergo training on the survey prior to their visit. Results Among 3526 participants who participated in exit surveys, 2415 were untrained (56% female, median age 40 (IQR: 32–47)) and 1111 were trained (50% female, median age 37 (IQR: 31–45)). Compared with untrained, trained patients were more likely to report a negative care experience overall (adjusted prevalence ratio (aPR) for aggregate sum score: 1.64 (95% CI: 1.39 to 1.94)), with a greater proportion reporting feeling unwelcome by providers (aPR: 1.71 (95% CI: 1.20 to 2.44)) and witnessing providers behaving rude (aPR: 2.28 (95% CI: 1.63 to 3.19)). Conclusion Trained patients were more likely to identify suboptimal care. They may have understood the items solicited better or felt empowered to be more critical. We trained existing patients, unlike studies that use ‘standardised patients’ drawn from outside the patient population. This low-cost strategy could improve patient-centred service delivery elsewhere.