key: cord-274688-cr1rvy8u authors: Jewell, Britta L; Mudimu, Edinah; Stover, John; ten Brink, Debra; Phillips, Andrew N; Smith, Jennifer A; Martin-Hughes, Rowan; Teng, Yu; Glaubius, Robert; Mahiane, Severin Guy; Bansi-Matharu, Loveleen; Taramusi, Isaac; Chagoma, Newton; Morrison, Michelle; Doherty, Meg; Marsh, Kimberly; Bershteyn, Anna; Hallett, Timothy B; Kelly, Sherrie L title: Potential effects of disruption to HIV programmes in sub-Saharan Africa caused by COVID-19: results from multiple mathematical models date: 2020-08-06 journal: Lancet HIV DOI: 10.1016/s2352-3018(20)30211-3 sha: doc_id: 274688 cord_uid: cr1rvy8u BACKGROUND: The COVID-19 pandemic could lead to disruptions to provision of HIV services for people living with HIV and those at risk of acquiring HIV in sub-Saharan Africa, where UNAIDS estimated that more than two-thirds of the approximately 38 million people living with HIV resided in 2018. We aimed to predict the potential effects of such disruptions on HIV-related deaths and new infections in sub-Saharan Africa. METHODS: In this modelling study, we used five well described models of HIV epidemics (Goals, Optima HIV, HIV Synthesis, an Imperial College London model, and Epidemiological MODeling software [EMOD]) to estimate the effect of various potential disruptions to HIV prevention, testing, and treatment services on HIV-related deaths and new infections in sub-Saharan Africa lasting 6 months over 1 year from April 1, 2020. We considered scenarios in which disruptions affected 20%, 50%, and 100% of the population. FINDINGS: A 6-month interruption of supply of antiretroviral therapy (ART) drugs across 50% of the population of people living with HIV who are on treatment would be expected to lead to a 1·63 times (median across models; range 1·39–1·87) increase in HIV-related deaths over a 1-year period compared with no disruption. In sub-Saharan Africa, this increase amounts to a median excess of HIV deaths, across all model estimates, of 296 000 (range 229 023–420 000) if such a high level of disruption occurred. Interruption of ART would increase mother-to-child transmission of HIV by approximately 1·6 times. Although an interruption in the supply of ART drugs would have the largest impact of any potential disruptions, effects of poorer clinical care due to overstretched health facilities, interruptions of supply of other drugs such as co-trimoxazole, and suspension of HIV testing would all have a substantial effect on population-level mortality (up to a 1·06 times increase in HIV-related deaths over a 1-year period due to disruptions affecting 50% of the population compared with no disruption). Interruption to condom supplies and peer education would make populations more susceptible to increases in HIV incidence, although physical distancing measures could lead to reductions in risky sexual behaviour (up to 1·19 times increase in new HIV infections over a 1-year period if 50% of people are affected). INTERPRETATION: During the COVID-19 pandemic, the primary priority for governments, donors, suppliers, and communities should focus on maintaining uninterrupted supply of ART drugs for people with HIV to avoid additional HIV-related deaths. The provision of other HIV prevention measures is also important to prevent any increase in HIV incidence. FUNDING: Bill & Melinda Gates Foundation. Increase in death rate in people with AIDS diseases due to overstretched health system 1.00 (1. See footnotes to Table 2 in main paper. Table A2 . Predicted average relative annual increase in HIV mortality and incidence in 5 years from 1 April 2020 in countries in sub-Saharan Africa that would result from a 6-month disruption of specific HIV services for 50% of the population, with 95% uncertainty bounds. Assuming no change in sexual behaviour associated with the period of disruption. For adults and children, except where stated. Individual national estimates were generated, and respective values aggregated for the two subregions (Eastern and Southern Africa and Western and Central Africa) and for sub-Saharan Africa, with estimates extrapolated to represent regional burden of people living with HIV for each region. The impact on drug resistance as a result of this disruption and additional COVID-19-related deaths among people living with HIV (PLHIV) were not considered. The HIV-related mortality rate was applied to those on antiretroviral therapy (ART) who were removed from ART due to this disruption following empirical observations from the SMART Study Group (1). CD4 counts declined rapidly as a result, and gradually returned to pre-ART levels after interruption. Once ART coverage was resumed following the disruption, mortality rate for those on ART were reapplied. Times for CD4 progression used in the Optima HIV model are listed in Table A4 . The figure below illustrates the PLHIV on ART and their subsequent CD4 count in the baseline versus the no ART for 6-month scenario in 2020 for an example country. For ART disruption, this is for the default "off ART" mortality rate, CD4 counts dropped even more dramatically following the parameter informed by the SMART Study (1), as used herein. The default off ART mortality rate (see table A2 ) was applied for generating CD4 counts shown in figure A1 ; however, mortality rate off ART used in this study followed SMART Study findings, which would have showed even more dramatic CD4 reductions particularly in 2021. Figure A2 illustrates the difference in progression of CD4 counts among adults living with HIV either on or off ART for an example country in sub-Saharan Africa. The HIV-related death rate for people living with HIV who were on ART, but who stopped treatment due to potential disruption on the antiretroviral (ARV) drug supply as a result of the COVID-19 pandemic was modeled to represent findings from the SMART Study (1). The SMART trial showed that immune recovery during ARV therapy tended to be rapidly lost after interruption of treatment, with a median loss of 187 CD4 cells/mm 3 HIV-related mortality off ART This below is from the appendix to reference 1 describing the Viral load and CD4 count changes during ART interruption. https://www.thelancet.com/cms/10.1016/S2352-3018(19)30400-X/attachment/3398d3d8-5988-4171-aa74-2f68ce877725/mmc1.pdf Risk of AIDS death is related to the most recent CD4 count, viral load and age, as described in that appendix. Viral load returns to previous maximum viral load (vmax) in 3 months and adopts natural history changes thereafter. CD4 rate of decline returns to natural history changes (i.e. those in ART naïve patients) after 9 months, unless the count remains > 200 above the CD4 nadir Rate of CD4 count decline depends on current viral load. c(t) is the CD4 count at time t, cmin(t) is the CD4 count nadir measured by time t and cc(t-1) is the change in CD4 count from t-1 to t. This is broadly based on evidence from a number of analyses of the effects of ART interruption (1-5 below) The Imperial College London Model is a deterministic, compartmental model of HIV transmission and progression, and has been described previously [1, 2] . Briefly, the model is stratified by age, sex, behavioural risk group (including partner change rate, condom use, and frequency of sex acts), and male circumcision status, with population growth in accordance with country-specific projections. HIV This analysis simulates the effect of 3-and 6-month interruptions of different HIV services starting mid-2020 as a result of the COVID-19 pandemic. In all scenarios of interruptions to HIV services, there is assumed to be no change to sexual contact rates, but a scenario of a 10% reduction in sexual contacts for all risk groups is modelled separately. Each type of disruption is assumed to occur independently so as to illustrate the individual effect of the respective disruption. However, in reality, multiple disruptions might occur simultaneously. Results are most sensitive to the assumption about mortality for HIV-infected individuals who stop ART. Our results are therefore presented using three different estimates of mortality, with the central result using the 'medium' assumption in Supplementary Table 1 . The SMART trial found a 3% risk at 12 months of either death of an opportunistic infection for those with interrupted ART [3] . This also implies a mean survival time approximately equivalent to that for HIVpositive persons who have never been on ART [4] . Upper bound 0.44% 5.28% This is the hypothetical worst-case scenario in which many persons deteriorate more rapidly. The EMOD model source code, parameter definitions and ranges, and description/documentation can be found at https://github.com/InstituteForDiseaseModeling/EMOD. Analyses are based upon a model calibrated to the HIV epidemic in South Africa and described elsewhere. [1] [2] [3] [4] Briefly, the model is parameterized with epidemiologic data including population size, fertility, mortality, voluntary male circumcision coverage, health seeking and sexual behaviour. South Africa data on age and sex specific HIV prevalence, ART coverage, population size and HIV incidence were used to calibrate the model. Calibration was performed using a parallel simultaneous perturbation optimization (PSPO) algorithm. 5 Roulette resampling in proportion to the likelihood of each simulation was used to select 250 model parameter sets. The model's fit to adult HIV prevalence and number receiving ART are shown in Figure A5 (gray lines: individual simulations; blue line: average of 250 simulations; black dots/lines: data from HIV surveys). To estimate the number of excess deaths in countries and regions throughout sub-Saharan Africa, we assumed that the number of deaths caused by a 6-month interruption in ART was proportional to the number of people virally suppressed on ART in a given country or region. A limitation of this assumption is that it does not take into account regional differences in the vulnerability of PLHIV with viral load suppression to death from an ART interruption, which could arise due to differences in the age distribution, CD4 count, health and nutritional status, and exposure to infections such as tuberculosis. Data used to estimate mortality by country and region are shown in Supplementary Table 1 Table A10 . Estimation of excess HIV deaths in 1 year due to 6 months interruption of ART with 95% uncertainty bounds. Updated assessment of risks and benefits of dolutegravir versus efavirenz in new antiretroviral treatment initiators in sub-Saharan Africa: modelling to inform treatment guidelines Interruption of HAART in HIV clinical practice. Results from the ICONA study Interruption and discontinuation of HART in the MACS Reasons for modification and discontinuation of antiretrovirals: results from a single treatment centre Safety of long-term interruption of successful antiretroviral therapy: the ATHENA cohort study Predictors of initial CD4 decline after antiretroviral treatment interruption in the SMART study The new role of antiretrovirals in combination HIV prevention: a mathematical modelling analysis Maximising HIV prevention by balancing the opportunities of today with the promises of tomorrow: a modelling study The Strategies for Management of Antiretroviral Therapy (SMART) Study Group, CD4+ Count-Guided Interruption of Antiretroviral Treatment Time from HIV seroconversion to death: a collaborative analysis of eight studies in six low and middle-income countries before highly active antiretroviral therapy Implementation and applications of EMOD, an individual-based multi-disease modeling platform Sexual partnership age pairings and risk of HIV acquisition in rural South Africa Targeting and vaccine durability are key for population-level impact and cost-effectiveness of a pox-protein HIV vaccine regimen in South Africa The future of a partially effective HIV vaccine: assessing limitations at the population level Application of a second-order stochastic optimization algorithm for fitting stochastic epidemiological models