Effect of the COVID-19 Lockdown on Mobile Payments for Maternal Health: Regression Discontinuity Analysis.

Background: The COVID-19 pandemic resulted in the unprecedented popularity of digital financial services for contactless payments and government cash transfer programs to mitigate the economic effects of the pandemic. The effect of the pandemic on the use of digital financial services for health in low- and middle-income countries, however, is poorly understood.

Objective: This study aimed to assess the effect of the first COVID-19 lockdown on the use of a mobile maternal health wallet, with a particular focus on delineating the age-dependent differential effects, and draw conclusions on the effect of lockdown measures on the use of digital health services.

Methods: We analyzed 819,840 person-days of health wallet use data from 3416 women who used health care at 25 public sector primary care facilities and 4 hospitals in Antananarivo, Madagascar, between January 1 and August 27, 2020. We collected data on savings, payments, and voucher use at the point of care. To estimate the effects of the first COVID-19 lockdown in Madagascar, we used regression discontinuity analysis around the starting day of the first COVID-19 lockdown on March 23, 2020. We determined the bandwidth using a data-driven method for unbiased bandwidth selection and used modified Poisson regression for binary variables to estimate risk ratios as lockdown effect sizes.

Results: We recorded 3719 saving events, 1572 payment events, and 3144 use events of electronic vouchers. The first COVID-19 lockdown in Madagascar reduced mobile money savings by 58.5% (P<.001), payments by 45.8% (P<.001), and voucher use by 49.6% (P<.001). Voucher use recovered to the extrapolated prelockdown counterfactual after 214 days, while savings and payments did not cross the extrapolated prelockdown counterfactual. The recovery duration after the lockdown differed by age group. Women aged >30 years recovered substantially faster, returning to prelockdown rates after 34, 226, and 77 days for savings, payments, and voucher use, respectively. Younger women aged <25 years did not return to baseline values. The results remained robust in sensitivity analyses using ±20 days of the optimal bandwidth.

Conclusions: The COVID-19 lockdown greatly reduced the use of mobile money in the health sector, affecting savings, payments, and voucher use. Savings were the most significantly reduced, implying that the lockdown affected women's expectations of future health care use. Declines in payments and voucher use indicated decreased actual health care use caused by the lockdown. These effects are crucial since many maternal and child health care services cannot be delayed, as the potential benefits will be lost or diminished. To mitigate the adverse impacts of lockdowns on maternal health service use, digital health services could be leveraged to provide access to telemedicine and enhance user communication with clear information on available health care access options and adherence to safety protocols.