This study evaluates the impact of a quality improvement (QI) intervention on maternal and child healthcare services in seven primary healthcare (PHC) clinics, in a rural setting of KwaZulu-Natal, South Africa.

INTRODUCTION TO QUALITY IMPROVEMENT Quality Improvement (QI), defined as "systematic and continuous actions that lead to measurable improvement in health care services and the health status of targeted patient groups" (IOM 2014) has roots in the consumer industry as far back as the 1920s. Data-driven structured process improvements have their origins in the consumer industry, particularly motor vehicles, and Avedis Donabedian first described a model for healthcare quality improvement in the 1960s which reflects the same process change model utilised in the consumer industry: the key elements being structures, processes and outcomes. Don Berwick and colleagues founded the Institute for Healthcare Improvement (IHI) in the 1990s, now a leading institution on QI worldwide. The QI approach or 'form' is structured and uses specific QI tools, whereas its application or 'function' is varied depending on selected process interventions and local context. Given the move towards rigorously improving quality of health care worldwide, QI methodology is increasingly gaining popularity not only in high-income settings but also in low- and middle-income countries (LMIC) including South Africa through Department of Health (DoH) commitments to improving quality of health care and reducing disparities thereof.

EVIDENCE BASE FOR QI Whereas QI is increasingly used worldwide including resource-rich settings and resource-limited settings, there is a paucity of scientific evidence assessing causal impact of QI on health outcomes, particularly the gold standard randomized controlled trial. Studies in resource-limited settings suggest a beneficial effect of QI on measured healthcare outcomes however none assessed QI rigorously as a single intervention.

GAPS IN THE PMTCT CASCADE AND INFANT NUTRITION Despite worldwide rollout of antiretroviral therapy (ART) and efforts to prevent mother-to-child transmission (PMTCT), there are still large gaps in coverage with approximately 240,000 infants vertically infected with HIV worldwide, 89% of whom were from the WHO Africa region. South Africa has a very high prevalence of HIV, with a national average of 30% amongst antenatal care (ANC) clients. In the Africa Centre Demographic Surveillance Area (DSA) located within the Hlabisa sub-district of uMkhanyakude district, the HIV prevalence amongst females of reproductive age (15-49 years) ranged from 20-45% in 2011.KwaZulu-Natal province had the highest prevalence of HIV amongst ANC clients of 37.4% in 2012. Furthermore, although KwaZulu-Natal (KZN) had the highest rate of ART coverage amongst HIV-infected pregnant women eligible for ART, the uMkhanyakude district in KZN had ART coverage of only 73% during the same period, far behind the national target of 90%.

Studies have shown inadequate virologic suppression amongst HIV-infected pregnant and breastfeeding (PBF) mothers, and inadequate repeat HIV testing of HIV-negative mothers during the PBF period. Surveillance of PBF mothers in Kenya, Malawi and South Africa demonstrated an HIV seroconversion rate of approximately 4%, illustrating the need to repeat HIV testing through pregnancy and breastfeeding in order to minimize risk of MTCT. Moreover the rate of virologic suppression amongst HIV-infected PBF women varied from 27% (in Kenya) to 72% in Malawi in the same study, and underscores the importance of virologic monitoring to reduce MTCT through virologic suppression. A study in Kenya demonstrated a repeat HIV testing rate of approximately 23% in HIV negative mothers. Given the risk of seroconversion during PBF, the World Health Organisation recommends repeat HIV screening during PBF in high prevalence settings. Although SA has achieved a commendable reduction in MTCT to 2.7% in 2011 at 6 weeks of age in line with the UNAIDS call for virtual elimination of MTCT by 2015, these gaps in PMTCT coverage increase the risk of undetected maternal HIV seroconversion and inadequate virologic suppression and therefore PMTCT achievements to date may be reversed without adequate intervention. Furthermore, given the fertility rate of the population, an MTCT rate of 2.7% with an antenatal HIV prevalence of 30% the number of new infant infections per year in South Africa is concerning high. Preliminary data from the Africa Centre suggest that less than 40% HIV-negative pregnant women undergo re-testing for HIV during pregnancy, whilst approximately 10-20% on lifelong ART (initiated prior to or during the current pregnancy) have virologic failure.

An important component on the agenda for reducing maternal morbidity and mortality is preventing unwanted pregnancies through better uptake of family planning methods. However limitations to contraceptive uptake include lack of knowledge and lack of access in resource-limited settings. According to UNFPA estimates, contraceptive prevalence in South Africa is approximately 65% despite free access to most methods.

Although replacement feeding of HIV-exposed infants might mitigate the risk of MTCT, in resource-limited settings with unreliable access to safe drinking water the high risk of childhood illnesses such as diarrhoea and pneumonia is associated with high rates of mortality. As under 5 childhood mortality in these settings can be reduced significantly by exclusive breastfeeding (EBF), the WHO recommends EBF for HIV-infected and HIV-uninfected mothers, whilst the importance of ART and virologic suppression are further emphasised. These recommendations are endorsed by the South African DoH however stigma associated with EBF by HIV-infected mothers often becomes a barrier to uptake of EBF in general. South Africa had the lowest rates of EBF in the world in 2012 at just 8%.

New South African DoH ART guidelines were implemented in January 2015, notably changing PMTCT guidance to Option B+ (lifelong ART for all pregnant women regardless of CD4 count). HIV screening every 3 months during PBF is also recommended due to the risk of seroconversions given the high prevalence of HIV in South Africa. Furthermore, viral load (VL) testing of PBF on ART is recommended, immediately for those already on ART prior to pregnancy and every 6 months thereafter (assuming VL results are <1000 copies/mL); newly diagnosed PBF women should receive a VL test 3 months after initiating ART and 6 monthly thereafter (assuming VL results are <1000 copies/mL).

THE CONTEXT OF ANTENATAL AND POSTNATAL CARE SERVICES The Africa Centre for Population Health is located in the Hlabisa sub-district of uMkhanyakude district in rural KwaZulu-Natal South Africa. The Africa Centre Demographic Surveillance Area (DSA) covers an area of 438 km2 and is home to the Africa Centre Household Demographic Surveillance System (HDSS), covering approximately 90,000 people from 11,000 households annually. The HDSS is a longitudinal cohort from the early 2000s and collects individual and household information on mortality, fertility, migration, health status and socio-economic data. This provides a nuanced understanding of the study population and enables linking of datasets.

There are 7 primary healthcare clinic (PHC) facilities which are nurse-led in the DSA, providing ANC, PNC and other medical services, and one local district hospital (Hlabisa hospital) which offers some tertiary care services including routine and emergency obstetric care. The number of deliveries occurring in Hlabisa sub-district is estimated at about 3000 per annum.

The local hospital also provides medical doctor support to the PHCs when available. As part of the National Strategic Plan to improve quality of healthcare services, the DoH has rolled out District Clinical Specialist Teams (DCST) including a family physician, paediatrician, anaesthetist, midwife and obstetrician. However the success of staff recruitment to the DCST has been varied. Finally, a local scoping study (unpublished data) of healthcare staff demonstrated gaps in training, supervision and retention of staff with consequent impact on healthcare worker motivation and performance.

STUDY HYPOTHESIS AND JUSTIFICATION Given the nature of overstretched health services, gaps in the PMTCT cascade and infant nutrition illustrated above, a QI intervention is considered most appropriate for optimising clinic processes and therefore healthcare outcomes. It is anticipated that through a real time data-driven approach, process outcomes can be continuously monitored, thereby motivating long term sustainability of changes in clinical practice and improving provider and patient satisfaction with services. Furthermore, QI is a strategy the South African DoH is committed to, as part of improving quality of health services and has worked on several projects with the Centre for Rural Health (CRH), University of KwaZulu-Natal.

The investigators believe this is the first study to causally assess the impact of QI on specified health care outcomes, utilizing a randomised controlled trial (stepped-wedge) study design.

Inclusion Criteria:

• Currently postpartum: immediately post delivery OR 3-6 days postpartum OR 6 weeks postpartum; AND
• Lives in Africa Centre Demographic Surveillance Area (DSA); AND/OR
• Attended antenatal care (ANC) or postnatal care (PNC) in study intervention clinics

Exclusion Criteria:

• Below 18 years of age;
• Not postpartum;
• Does not live within Africa Centre DSA; AND
• Did not attend ANC or PNC at study intervention clinics

Masking was only until randomisation status identified at the start of each intervention step.