Menu
Menu
Menu
Menu
Background: The unfinished burden of poor maternal and child health contributes to the quadruple burden of disease in South Africa with the direct and indirect effects of the COVID-19 pandemic yet to be fully documented. Objective: To investigate the indirect effects of COVID-19 on maternal and child health in different geographical regions and relative wealth quintiles. Methods: We estimated the effects of COVID-19 on maternal and child health from April 2020 to June 2021. We estimated this by calculating mean changes across facilities, relative wealth index (RWI) quintiles, geographical areas and provinces. To account for confounding by underlying seasonal or linear trends, we subsequently fitted a segmented fixed effect panel model. Results: A total of 4956 public sector facilities were included in the analysis. Between April and September 2020, full immunisation and first dose of measles declined by 6.99% and 2.44%, respectively. In the follow-up months, measles first dose increased by 4.88% while full immunisation remained negative (−0.65%) especially in poorer quintiles. At facility level, the mean change in incidence and mortality due to pneumonia, diarrhoea and severe acute malnutrition was negative. Change in first antenatal visits, delivery by 15–19-year olds, delivery by C-section and maternal mortality was positive but not significant. Conclusion: COVID-19 disrupted utilisation of child health services. While reduction in child health services at the start of the pandemic was followed by an increase in subsequent months, the recovery was not uniform across different quintiles and geographical areas. This study highlights the disproportionate impact of the pandemic and the need for targeted interventions to improve utilisation of health services.
This is a quantitative analysis of the indirect effects of COVID-19 on selected maternal and child health indicators in South Africa. We used data from the district health information system (DHIS2). The DHIS2 captures data on key indicators for routine monitoring and evaluation of healthcare provision in South Africa’s public facilities. Public facilities included fixed and mobile clinics (customised motor vehicle that travels to communities to provide healthcare services), community health centres as well as hospitals funded by the government. Approximately 61% of the population access care in public health facilities [19]. Data was collected on a monthly basis from all public facilities in all nine provinces. For this analysis, data from January 2018 to June 2021 were collected using Excel Microsoft Office Software (version plus 2016). The main outcome variables are listed in Table 1. We selected these variables because they are routinely collected to monitor utilisation of maternal and child health services in the public sector. Other variables included the relative wealth index (quintile 1: poorest, quintile 2: poor, quintile 3: middle, quintile 4: wealthy, quintile 5: wealthiest), geography (urban, peri-urban, and rural) and the nine provinces. Outcome variables and their definitions. The impact of the lockdown restrictions on the different outcomes was estimated first by calculating changes in the outcomes across facilities, relative wealth index (RWI) quintiles and geographical areas and provinces. Subsequently, to account for confounding by underlying seasonal or linear trends, we fitted a segmented fixed effect panel model. The analysis was stratified by RWI index, geographic area (urban/rural) and province. The RWI was adopted from Chi et al. (2021) who constructed the index using standardised set of questions from household surveys. Using housing characteristics such as roof material, rooms in house, floor material, and water supply, an RWI was calculated by taking the first principal component of the questions [20]. We adopted this approach because the estimates were more granular making it possible to match each facility to an RWI within a 5 km radius. The regression took the following general form: Where yft is the outcome indicator, e.g., full immunisation, α0 is the intercept, t indexes time starting from the beginning of the sample, i.e. January 2018, PreTrendt is a variable reflecting the number of months since January 2018, PostCovidt is the number of months since the first COVID wave (i.e. PostCovidt=1 in April 2020, PostCovidt=2 in May 2020 and so on), βi are the fitted model parameters, γi are the fitted seasonal coefficients and Monthi represents a corresponding month indicator (with December as the reference category). νf is a facility indicator which represent the unobserved underlying heterogeneity that the fixed effect model caters for, and ϵft is an idiosyncratic mean zero error term. We estimated the impact first from April 2020 to September 2020 and then extended the analysis to June 2021. We used Python (version 3.6) software for data cleaning and analysis.
N/A
