Effect of health-system continuous quality improvement on long-term retention in HIV care among pregnant and breastfeeding women in the Democratic Republic of Congo: A cluster-randomized trial

Background. Despite relatively high coverage of interventions to prevent vertical transmission, HIV incidence among children born to women living with HIV (WLH) remains above elimination targets, partly because of substantial disengagement from antiretroviral therapy (ART) during the postpartum period. We evaluated the effectiveness of a data-driven continuous quality improvement (CQI) intervention in improving retention in care, virological suppression, and prevention of vertical transmission among pregnant and breastfeeding WLH receiving ART in maternal and child health (MCH) clinics. Methods and findings. This pragmatic cluster-randomized trial was conducted in 105 MCH clinics--three high-volume clinics in each of the 35 health zones (HZs) of Kinshasa Province, Democratic Republic of the Congo. During the pre-randomization phase (November 2016-December 2017), routine data collection using prevention of mother-to-child transmission registers was strengthened in all participating clinics. In January 2018, HZs were randomized 1:1 to implement CQI or continue the standard of care. In intervention HZs, quality improvement (QI) teams were established at the HZ and clinic levels. Indicators selected collaboratively with QI teams were calculated from routine program data and shared quarterly. Through July 2019, QI teams met quarterly to identify service-delivery challenges and develop strategies to improve outcomes, with an emphasis on retention. Thereafter, responsibility for producing indicators was transferred to QI teams, and meetings continued virtually through WhatsApp. Pregnant and breastfeeding WLH and their infants receiving care at participating clinics between November 2016 and July 2019 were enrolled and followed for at least 24 months. Primary outcomes were loss to follow-up (LTFU), viral suppression (<1,000 copies/mL), and a positive HIV test among live-born infants. Overall, 2,376 women were enrolled, including 1,217 (51.2%) before randomization. The analysis of CQI effectiveness included 1,900 participants who remained in care at randomization or enrolled thereafter: 1,103 (58.1%) in control HZs and 797 (41.9%) in intervention HZs. At 6, 12, and 24 months, LTFU was lower in the intervention group than in the control group: 18.6% versus 21.6% (adjusted risk difference [aRD], -5.2 percentage points; 95% CI, -9.6 to -0.8), 29.2% versus 31.0% (aRD, -3.9; 95% CI, -10.2 to 2.4), and 50.1% versus 55.5% (aRD, -7.1; 95% CI, -15.2 to 1.0), respectively. Estimated effects were larger in PEPFAR-supported and urban clinics, although interaction tests were not statistically significant. Viral-load results were available for 38.2%, 50.4%, and 73.5% of eligible participants at 6, 12, and 24 months, respectively. Neither viral-load-result availability nor viral suppression differed between groups. Infant HIV-test results were available for 8.1%, 32.7%, 34.2%, and 28.5% of eligible infants at 6 weeks and 6, 12, and 24 months, respectively. Conclusions. CQI implementation produced a modest improvement in retention, with the clearest evidence at 6 months and favorable but imprecise estimates through 24 months. Limited availability of maternal viral-load and infant HIV-test results precluded definitive conclusions about virological suppression and vertical transmission. Local quality-improvement interventions may strengthen processes within frontline teams control but cannot alone overcome failures across an interconnected care system. Future studies should align interventions operational reach with all health-system levels responsible for the intended outcomes.