Abstract
BACKGROUND: With the double burden of rising chronic non-communicable diseases (NCDs) and persistent infectious diseases facing sub-Saharan Africa, integrated health service delivery strategies among resource-poor countries are needed. Our study explored the post-trial sustainability of a health system intervention to improve NCD care, introduced during a cluster randomised trial between 2013 and 2016 in Uganda, focusing on hypertension (HT) and type-2 diabetes mellitus (DM) services. In 2020, 19 of 38 primary care health facilities (HFs) that constituted the trial's original intervention arm until 2016 and 3 of 6 referral HFs that also received the intervention then, were evaluated on i) their facility performance (FPS) through health worker knowledge, and service availability and readiness (SAR), and ii) the quality-of-patient-care-and-experience (QoCE) received. METHODS: Cross-sectional data from the original trial (2016) and our study (2020) were compared. FPS included a clinical knowledge test with 222 health workers: 131 (2016) and 91 (2020) and a five-element SAR assessment of all 22 HFs. QoCE assessment was performed among 420 patients: 88 (2016) and 332 (2020). Using a pair-matched approach, FPS and QoCE summary scores were compared. Linear and random effects Tobit regression models were also analysed. RESULTS: The mean aggregate facility performance (FPS) in 2020 was lower than in 2016: 70.2 (95%CI = 66.0-74.5) vs. 74.8 (95%CI = 71.3-78.3) respectively, with no significant difference (p = 0.18). Mean scores declined in 4 of 5 SAR elements. Overall FPS was negatively affected by rural or urban HF location relative to peri-urban HFs (p < 0.01). FPS was not independently predicted but patient club functionality showed weak association (p = 0.09). QoCE declined slightly to 8.7 (95%CI = 8.4-91) in 2020 vs 9.5 (95%CI = 9.1-9.9) in 2016 (p = 0.02) while the proportion of patients receiving adequate quality care also declined slightly to 88.2% from 98.5% respectively, with no statistical difference (p = 0.20). Only the parent district weakly predicted QoCE (p = 0.05). CONCLUSIONS: Four years after the end of research-related support, overall facility performance had declined as expected because of the interrupted supplies and a decline in regular supervision. However, both service availability and readiness and quality of HT/DM care were surprisingly well preserved. Sustainability of an NCD intervention in similar settings may remain achievable despite the funding instability following a trial's end but organisational measures to prepare for the post-trial phase should be taken early on in the intervention process.