Understanding the impact of interruptions to HIV services during the COVID-19 pandemic: A modelling study

Abstract

Background: There is concern that the COVID-19 pandemic could severely disrupt HIV services in sub-Saharan Africa. However, it is difficult to determine priorities for maintaining different elements of existing HIV services given widespread uncertainty.

Methods: We explore the impact of disruptions on HIV outcomes in South Africa, Malawi, Zimbabwe, and Uganda using a mathematical model, examine how impact is affected by model assumptions, and compare potential HIV deaths to those that may be caused by COVID-19 in the same settings.

Findings: The most important determinant of HIV-related mortality is an interruption to antiretroviral treatment (ART) supply. A three-month interruption for 40% of those on ART could cause a similar number of additional deaths as those that might be saved from COVID-19 through social distancing. An interruption for more than 6-90% of individuals on ART for nine months could cause the number of HIV deaths to exceed the number of COVID-19 deaths, depending on the COVID-19 projection. However, if ART supply is maintained, but new treatment, voluntary medical male circumcision, and pre-exposure prophylaxis initiations cease for 3 months and condom use is reduced, increases in HIV deaths would be limited to <2% over five years, although this could still be accompanied by a 7% increase in new HIV infections.

Interpretation: HIV deaths could increase substantially during the COVID-19 pandemic under reasonable worst-case assumptions about interruptions to HIV services. It is a priority in high-burden countries to ensure continuity of ART during the pandemic.

Funding: Bill & Melinda Gates Foundation.

Keywords: Antiretroviral therapy; COVID-19; HIV; Mathematical modelling.

Fig. 1

Potential changes in HIV incidence (A) and deaths (B) following a hypothetical three-month disruption of HIV services. Orange shading indicates the period during which the three-month disruptions occur. Baseline (no COVID-19 epidemic occurs) = dark blue line; Reduced contact rates = orange line; No new VMMCs = yellow line; Viral suppression decreases by 10% = purple line; No new ART initiations = green line; Condom use reduced by 50% = light blue line; Interruption in ART for 40% of individuals = dark red line. Panel insets in (A) show a magnified view of HIV incidence patterns from 2020.5 to 2021.

Fig. 2

Projected direct and indirect deaths over time for South Africa (A), Malawi (B), Zimbabwe (C), and Uganda (D), under different assumptions about the COVID-19 epidemic in Walker et al. and the impact on HIV services (Table 1), for a three-month service interruption. The purple bars for COVID-19 deaths represent estimates from Walker et al. The whiskers represent estimates from Cabore et al. and Pearson et al. for an in an unmitigated epidemic, and Pearson et al. only for an epidemic with social distancing and enhanced social distancing from Table 2; no estimates from Cabore et al. were available for these scenarios [11,12]. Deaths averted by implementing social distancing and enhanced social distancing are from Walker et al. Yellow bars represent HIV deaths in the ‘interruption of supply’ scenario; red bars represent HIV deaths in the ‘managed disruption’ scenario; blue bars represent HIV deaths in the ‘managed pause’ scenario (all from Table 1). Confidence intervals for the HIV mortality scenarios represent uncertainty in the mortality risk for individuals with interrupted ART, ranging from a mean monthly mortality risk of 0.10% to 0.44%.

Fig. 3

The excess HIV deaths over 5 years caused by the disruptions, as a percentage of the total number of HIV deaths expected without the disruptions. Results give the mean across the four countries considered – South Africa, Malawi, Zimbabwe, and Uganda. Plots shown give this metric with respect to the duration of interruption in months (vertical axes) and the proportion of people living with HIV (PLHIV) on antiretroviral therapy (ART) affected by the interruption (horizontal axes). The panels show the results under three different assumptions about the average net monthly mortality risk for those experience an interruption in ART supply: (A) 0.10%, (B) 0.24%, and (C) 0.44%. Contour lines show where the total number of deaths is greater than the deaths expected for COVID-19 epidemics, averaged across the four countries, for three different models of COVID-19 (Cabore et al., Walker et al., and Pearson et al. [11]). Light red lines correspond to the mean number of COVID-19 deaths in the four countries in Cabore et al.; orange and blue lines correspond to the mean number of COVID-19 deaths with enhanced social distancing in the four countries in Walker et al. and Pearson et al., respectively. Dashed lines correspond to the base case scenario of 40% of individuals experiencing a three-month interruption of ART.