Using administrative and RHIS data from ten countries, we assessed the effect of the COVID-19 pandemic on a spectrum of health services. We estimated the immediate effect after the declaration of the pandemic on 11 March 2020 and assessed whether services had returned to pre-pandemic levels by the last quarter of 2020. We found declines of varying magnitude and duration in every country. Effects were heterogeneous across countries, and we found no clear patterns in disruptions by country income group or according to the severity of COVID-19 epidemics. The health systems most affected included those in Chile, Haiti, Mexico, Nepal and South Africa. By contrast, Ethiopia and South Korea, which represent the poorest and richest countries, respectively, in our analysis, were among the least affected by health service disruptions.
The magnitude of health service disruptions at national levels also did not appear to be directly driven by COVID-19 severity. Of the ten countries included, six reported fewer than 2,000 cumulative cases per million in 2020 and even fewer deaths (Supplementary Table 3). Only 41 total cases were reported in Laos in 2020. Chile, Mexico, Nepal and South Africa faced higher COVID-19 caseloads, with peaks in June or July (or late October in Nepal). However, health service disruptions were largest in April and May 2020 in all countries, suggesting that they were not caused by overburdened health systems but rather by a combination of policy responses and demand-side factors. Several reasons for reduced healthcare use appeared common across countries: fear of contagion, inability to pay for healthcare due to loss of employment or remuneration, intentional suspension of routine care to leave room for patients with COVID-19, the redeployment of health workers or hospitals to COVID-19 care and prevention and the barriers imposed by COVID-19 lockdowns. Whether the type of COVID-19 response (for example, elimination versus steady-state strategies) or the stringency and length of COVID-19-related lockdowns were associated with the magnitude of disruptions remains unclear and should be investigated further.
On the other hand, we found patterns in disruptions according to the type of health service. Outpatient visits and hospital-based services (including emergency room visits, inpatient admissions, trauma care, accidents and surgeries) declined in every country reporting them, and these disruptions often persisted throughout the period analyzed. Other studies also reported declining inpatient admissions during the COVID-19 pandemic16,24. These declines may be explained, in part, by a reduction in need. For example, decreased mobility and bans on alcohol sale in some places have led to fewer accidents and a lower need for trauma care22,25,26. Social distancing and mask wearing might have also contributed to reduced spread of infectious diseases. However, the reduction in need is unlikely to account for the entire magnitude of decline. Much of the disruption in tertiary care might reflect that many hospitals were converted into COVID-19 treatment centers and suspended or postponed other services. The prioritization of COVID-19 care also disrupted the availability of intensive care beds, medical supplies and technology for services other than COVID-19. Hospitals also tend to be in urban areas that appear to have been more affected by service disruptions than rural areas (Supplementary Table 4). Declines in emergency room visits may also result from people delaying or foregoing urgent care. For example, studies from France and England suggest that people with chest pains and other symptoms of a myocardial infarction have been reluctant to go to hospitals during the pandemic, leading to a reduction in patients admitted with ST-segment elevation myocardial infarction and an increase in out-of-hospital deaths27,28. Persistent disruptions in hospital services could have important consequences, including exacerbating the already high unmet need for surgical care in LMICs29,30.
Overall, preventive care, such as routine childhood immunizations, screenings and testing, were among the most affected services. Although some of these services can potentially be delayed for a short time, our estimates indicate that many were not fully restored by the end of 2020. After the declaration of the pandemic, there were declines in child vaccinations of more than 10% in Chile, Haiti, Laos, Mexico, Nepal and South Africa (out of eight countries reporting these data). Several of these countries, in particular Laos and Nepal, were able to resume most vaccinations by the end of the year (Fig. 4). However, our estimates for the number of vaccinations missed from April to December 2020 show that not all vaccinations initially delayed were ultimately given (Fig. 5 and Supplementary Table 2). Observed effects were different across vaccine types. This is likely due to differences in vaccine schedules and delivery modes. For example, BCG is delivered at birth and generally followed the same declines as facility-based childbirth. We reported on five common vaccines (BCG, pentavalent, measles, pneumococcal and rotavirus). Other immunizations with different schedules or distribution modes might have been differently affected, including vaccines against the human papilloma virus, which was likely affected by school closures. Globally, DTP3 and MCV1 vaccination coverage is estimated to have fallen by more than 7% in 2020 compared to expected coverage in the absence of the pandemic31. A total of 66 countries also reported postponing at least one vaccination campaign in early 2020, and only 25 reinstated them by the end of the year32. These disruptions are expected to lead to future outbreaks of measles and other vaccine-preventable diseases and to an increase in child deaths8.
We also found large and persisting declines in breast and cervical cancer screening. Declines in cancer screenings and routine diagnostic work have been reported globally33,34,35,36. In England, breast cancer diagnostic delays are projected to increase 5-year mortality by 8–10%35. Chile and Mexico risk facing similar increases in breast cancer mortality over the next 5 years.
TB case detection declined by 28–66% in Ghana, Nepal and South Africa and remained lower than pre-COVID-19 by the end of 2020. With symptoms similar to COVID-19, such as a cough, fever and breathing difficulties, many people with TB symptoms might have opted to stay home or could have been mistakenly diagnosed with COVID-19 (ref. 37). The Global Fund to fight AIDS, TB and Malaria estimates that TB and HIV testing declined by 18–22% in countries supported by the fund. We found even larger declines in HIV testing in Nepal. An increase in untreated TB or HIV could have far-reaching consequences10,38. It is unclear whether social distancing may have contributed to reduced TB or HIV transmission. More time spent indoors in crowded households could increase TB transmission.
In contrast, across four countries, we found that the number of people on ART was virtually unaffected during the pandemic. Our findings are consistent with evidence that ART provision was generally maintained during the South African lockdown, whereas HIV testing and ART initiations declined12,39,40. Differentiated service delivery (DSD) programs for HIV, where drugs are distributed in decentralized locations, might explain the resilience of ART provision during the pandemic. Unlike traditional care models where visits are frequent and exclusively at the health facility, DSD models entail modifying the location for care (for example, to venues in the community), the frequency of visits (for example, biannually) and the cadre providing the services41.
Visits for malaria declined by 9% and 10% in Ghana and Thailand, respectively, but returned to pre-pandemic levels by the end of 2020. These short-term disruptions could still have led to an increase in malaria deaths, particularly if prevention activities (such as bed nets and insecticide spraying) were also disrupted42,43.
We found declines in diabetes or hypertension visits of more than 20% in Chile, Haiti, Mexico, Nepal, South Africa and Thailand. Similar disruptions have been reported elsewhere16,44,45,46. Some countries, including Chile, Mexico, South Africa and Thailand, reported implementing strategies to maintain drug adherence during the pandemic for people with these two conditions, such as online refills, community drug delivery or external pick-up points44,47. However, it is unclear whether they have been successful in maintaining drug adherence, as our data cover only the number of in-person visits conducted. Hypertension and diabetes management has been a particular challenge for LMIC health systems where the burden of uncontrolled diseases is high48,49,50. The pandemic could prompt policymakers to rethink the frequency of visits required and consider adopting principles of DSD to meet the needs of people living with these conditions51. South Africa has adopted such a strategy through the Central Chronic Medicines Dispensing and Distribution program52.
We also found large declines in in-person mental health services in Chile and Mexico. Only three countries reported on mental healthcare: Chile, Mexico and South Korea. In Mexico, the indicator was for mental healthcare after an attempted suicide, whereas Chile and South Korea reported on routine mental health consultations. In May 2020, the government of Chile established a digital mental healthcare platform (‘Saludable mente’) to address the rise in mental health disorders during the pandemic. Other countries also integrated mental health interventions, such as telephone hotlines to support frontline health workers and the general population, to their COVID-19 response, including Mexico and South Africa. These programs may have helped mitigate the impact of reduced in-person care, but there is little evidence to date on their effectiveness53. The increase in depressive and anxiety disorders reported globally during the pandemic calls for the urgent need to strengthen mental health systems54.
Reproductive and maternal healthcare was generally more resilient compared to other services. Only two health systems— Chile and Mexico’s public sectors—had large declines in contraceptive provision (52% and 87%, respectively). Although some public sector users may have switched to the private sector or to pharmacies for contraceptives, the unmet need for contraception appears to have increased across Latin America and the Caribbean during the COVID-19 pandemic55. Frequent contraceptive shortages were also reported. By contrast, family planning visits declined by only 4% and 14% in Nepal and Haiti, respectively, despite large disruptions in other services. This finding is consistent with other studies from low- and lower-middle-income countries (including from household surveys) that found relatively small changes in use of family planning services during the pandemic11,13,15,16. Economic uncertainties during the pandemic may also have led to an increased demand for contraception15.
The number of facility-based deliveries declined substantially in Haiti, Nepal, Mexico and South Africa but were relatively stable in the other six countries reporting. Other studies also found mixed results for the effect of the pandemic on facility-based deliveries11,13,14,16,24,56. Reasons for this likely vary by country. At the Mexican IMSS, many hospitals were converted into COVID-19 treatment centers, and many pregnant women were redirected to the private sector for childbirth (sometimes at their own cost)44. In South Africa, the dataset contained information from all public and private hospitals (in the KwaZulu-Natal province only). Thus, the 11% decline in facility-based deliveries likely reflects an increase in home births. Ethiopia, one of the countries with the lowest rate of facility deliveries, had a 3% decline in facility deliveries (not statistically significant). However, this estimate likely hides sub-national disruptions. One study using household survey data found a decline in hospital births in urban areas only57. Similarly in Haiti and Nepal, more women might have opted to give birth at home or with traditional attendants58. This could be associated with an increase in maternal and perinatal mortality and morbidity9,59. Poorer antenatal care follow-up could also lead to a higher number of pre-term births and stillbirths60,61.
Visits for children younger than 5 years of age with diarrhea and pneumonia declined in all countries reporting, which was also reported by others13,16. Part of these declines may be explained by a reduced incidence of diarrhea and pneumonia from social distancing, school and daycare closures, mask wearing and improved handwashing practices62. Some caregivers may have also opted to seek treatment from pharmacies, shops or the informal sector for their children’s illness rather than visit health facilities, which would not be reflected in our data.
Health system design and organization before the pandemic may be associated with health service resilience. For example, in Chile, maternal health services are provided exclusively by midwives who were not redeployed to COVID-19 care and were able to maintain regular service provision63. In South Korea, the number of hospital beds per capita is about three times higher than the Organization of Economic Cooperation and Development average64. Thus, the country may have been able to reallocate a large share of this capacity to COVID-19 care without substantial negative effects on other services. South Korea also benefited from prior investments and a stronger public health response system, given its experience handling the SARS outbreak of 2003, the novel influenza outbreak of 2009 and the MERS-CoV epidemic of 2015 (ref. 27). The private sector in Mexico is large and expanding, and private facilities were able to provide maternity care for a high percentage of public sector users while public hospitals were repurposed to COVID-19 care44.
Our analysis has several strengths. We estimated the effect of the COVID-19 pandemic on 31 health services using administrative and RHIS data that represented the complete, or nearly complete, census of all health facilities in the country (or province in the case of Kwa-Zulu Natal in South Africa). Unlike costly population health surveys, administrative and RHIS data can provide near real-time data on the performance of health systems. Our study also included countries from all income groups, which provides a more comprehensive picture of the effects of the pandemic. Nonetheless, our study has limitations. First, although we included a range of countries, our results cannot be generalized to their regions or to other parts of the world. Second, the number and type of indicators available in each country varied, including slight variations in definitions (Supplementary Table 1). Thus, cross-country comparisons should be made with care. Third, the exclusion of private providers in some countries in this analysis limits our ability to quantify the extent to which patients switched from the public sector to the private sector for healthcare during the pandemic. Similarly, the routine data systems generally did not include telemedicine consultations that were made during the pandemic. Fourth, disruptions were assessed only at the national level, and our estimates could hide disruptions that occurred in specific sub-national regions, cities, types of health facilities or population groups within a country. Fifth, it is possible that the pandemic affected the quality of reporting in administrative sources and RHIS. However, we used thorough data-cleaning procedures and only used data from facilities that continuously reported throughout the study period in the six countries with disaggregated data (Chile, Ethiopia, Haiti, Laos, Nepal and South Africa). Sixth, our main analysis covers only the first 9 months of the pandemic. However, data for the first 6 months of 2021 in a subset of countries reveal that service disruptions continued in many countries in 2021.
Our findings have implications for current health system planning and for the management of future pandemics. Despite the many efforts deployed to maintain the continuity of health services, we found considerable declines in healthcare use. Part of these declines may be linked to decreased healthcare needs during the pandemic from reductions in non-COVID-19 infectious illnesses and fewer injuries. Nonetheless, a larger share of these declines likely reflects a failure of health system resilience. Health systems must urgently resume essential care and plan to compensate for missed needed services. This includes catching up on missed preventive care (such as health screenings and immunizations) and identifying and addressing any adverse health consequences of missed services, such as trauma care, surgery, C-sections and chronic disease management. These can include physical sequelae (for example, obstetric fistulae), chronic disease complications and health-related suffering. Higher rates of uncontrolled hypertension and diabetes, for example, could lead to an increase in cardiac events and in complications of diabetes, such as blindness, kidney failure and lower limb amputations2. In Chile and Mexico, the decline in family planning could result in higher rates of unplanned pregnancies. Finally, the pandemic’s negative effects on mental health, combined with declines in in-person care, will lead to greater unmet needs for mental healthcare and a potential increase in suicides53. Increased investments in health systems are needed to address these consequences and the surge in pent-up demand as well as to prepare the health systems for more agile function in the future. Given limited resources in some countries that will be further strained by the global economic downturn due to the COVID-19 pandemic, priority should be given to health interventions that will have the greatest benefits on health65.
Further research is needed to understand the full indirect health effects of the pandemic and the factors responsible for service disruptions. For example, our analysis did not assess changes in the quality of healthcare, which was likely affected during the pandemic (including from poorer processes of care and shortages of medicines or supplies), resulting in poorer health outcomes even among those who received care7. Similarly, future research should monitor trends in mortality from non-COVID-19 conditions and assess whether certain population subgroups (such as ethnic minorities, teenagers or the poorest) were differentially impacted by health service disruptions. Finally, it is important to disentangle the factors responsible for disruptions in health services.
In 2021, nearly all countries included in this analysis experienced larger waves of SARS-CoV-2 infections and often re-implemented periodic lockdowns, including to prevent further spread of the Delta variant4. Given the widespread disruptions in health services demonstrated in this paper, many of which were unrelated to COVID-19 severity, our results call for rethinking pandemic preparedness and health system response. The unintended consequences of COVID-19 responses may have outweighed the loss of life from COVID-19 itself, particularly in LMICs66. Health system resilience must become a central component of national health plans. Given the likelihood of future pandemics and other major shocks, there is an urgent need to design more resilient health systems capable of addressing a crisis while maintaining essential functions.