Associations between the stringency of COVID-19 containment policies and health service disruptions in 10 countries

In response to the COVID-19 pandemic and in an effort to reduce the spread of infections, countries imposed a series of non-pharmaceutical interventions, such as stay-at-home orders, curfews and quarantines. These measures and policies were known as “lockdowns”, or COVID-19 containment or closure policies. By the end of March 2020 more than 100 countries had instituted either a full or partial lockdown [1].

An increasing breadth of literature has shown disruptions in health services during the initial phase of the COVID-19 pandemic with heterogeneity in the degree and duration of these disruptions [2‐16]. Nonetheless, studies found no clear pattern in health service disruptions according to pandemic intensity (e.g., number of COVID cases or deaths) or by country characteristics, income, or health system characteristics [2, 13, 14]. Disruptions in health service utilization during the pandemic could be caused by a combination of factors including a decrease in health facility capacity (due to reallocation of staff or other resources to COVID-19 wards, staff getting infected, or staff burnout), patients choosing to defer medical care due to concerns about exposure to the virus, or people’s inability to pay for medical care from loss of income or employment. The barriers imposed by COVID-19 containment policies such as public transportation closures, stay-at-home requirements or restrictions on internal movement, could also result in declines in non-COVID health care utilization [17‐19].

Few studies have sought to understand the relation between the stringency of COVID-19 containment policies and the level of health service disruptions. In 18 low- and middle-income countries, a study found a significant relationship between the stringency of COVID-19 mobility restrictions and disruptions in outpatient consultations, child vaccinations, and the fourth antenatal care visit, whereby a standard deviation in the mobility restrictions stringency was associated with a 3.9 percent reduction in outpatient consultation volume on average [16]. In 11 sub-Saharan African countries, the stringency of COVID-19 containment policies was significantly associated with reductions in outpatient visits and inpatient admissions, where a 10 percentage points increase in the stringency index was associated with a 3.1–3.6 percentage points reduction in these two services [14]. Other studies have looked at country or health-service specific relationships between containment policies and service disruptions. In South Africa, significant reductions in child health visits and HIV services were observed during varying levels of lockdowns [10, 15]. In Nepal, significant increases in certain health services were seen after lockdowns were lifted at local levels [20]. While the aforementioned studies have sought to measure the relationship between the stringency of containment policies and changes in health services, they tend to be limited in terms of geographic areas or in the types of health services covered, focusing largely on maternal and child health services.

In this paper, we investigate associations between the stringency of COVID-19 containment policies and disruptions in 31 health services in 10 low- middle- and high-income countries. We further extend this analysis by examining the relative impact of 11 individual containment or closure policies on health service disruptions. This study was undertaken as a part of the QuEST Network, a global partnership of researchers and policy makers conducting health systems research. Participating countries were selected based on prior collaborations. Colleagues from 10 countries with different income levels, health system types, severity of COVID and government responses to COVID joined the effort.

Our analysis included data from ten countries including three in Latin America and Caribbean, three in Sub-Saharan Africa, three in East Asia and the Pacific and one in South Asia (Table 2). GDP per capita ranged from $ USD 2,297 in Ethiopia to $ USD 42,251 in South Korea. The average number of new diagnosed COVID-19 cases per million population per month in 2020 varied widely from 0.48 in Lao PDR to more than 3,500 in Chile (Table 2). Chile had the highest average monthly stringency index during the study period (79.70), while Lao PDR had the lowest (44.53).

Our final dataset included a total of 1,467 relative service volume observations estimated monthly for each health service from April to December 2020. Countries with the largest health service disruptions included Chile, Mexico, and South Africa. For example, in Chile, the number of consultations for chronic disease services from April to December 2020 was only 45% of that provided on average pre-COVID-19. Ethiopia and South Korea were among the least affected countries. Average monthly relative service volumes by categories of services during the COVID-19 period are provided in Table 2. Chronic disease care (including visits for hypertension and diabetes and cancer and tuberculosis screening) and child health services (e.g., visits for children with diarrhea or pneumonia) were among the most affected services. Reproductive, maternal, and newborn health services (e.g., delivery care and c-sections) were generally less affected. The magnitude of service disruptions estimated using an interrupted time series (ITS) analysis was described in a previous study [2].

The relationship between the stringency index and relative service volumes is shown in Fig. 1. Several health services were considerably lower than the average pre-COVID (where the blue line is below 100% in Fig. 1). The mean stringency index (shown as a red line in Fig. 1) was relatively constant from April to December 2020 in Chile, Mexico, and South Korea. In contrast, it varied considerably in the other countries, often starting with high stringency in April and May 2020 which was slowly reduced over time.

Results from multi-level linear regression models are presented in Table 3. After adjusting for COVID cases per million, health service type and GDP per capita, a 10% increase in the mean stringency index was associated with a 3.3 percentage point (95% CI -3.9; -2.7) reduction in relative service volume. COVID-19 cases per million were not associated with relative service volumes. A higher GDP per capita was also associated with lower relative service volumes. We also found relatively high within-country correlation evidenced by a covariate-adjusted intracluster correlation coefficient (ICC) of 0.17 (0.07–0.34).

In the adjusted model, two health service types (service use overall and injuries and child health services) were associated with significantly lower relative service volumes (worst disruptions) compared to reproductive and maternal services. Associations between the stringency index and relative service volumes was robust to alternate specifications including using the monthly median stringency index and the monthly maximum stringency index instead of the mean (see supplemental Tables 3 and 4).

Associations between each of the 11 individual COVID-19 containment or closure policies and relative service volumes were assessed by fitting separate multilevel regression models, all of which included a country random effect and the following independent variables: COVID-19 cases per million, health service type and GDP per capita. We found that each containment policy was negatively associated with relative service volumes. Coefficients ranged from a 4.2 percentage-points decline in relative service volumes for public transport closures to a 14.1 percentage-point decline for curfews. Curfews, restrictions on gatherings and presence of a national state of emergency had the largest coefficients (Fig. 2). These three policies were associated with a 10-percentage point or greater reduction in relative service volumes.

In this paper, we used RHIS and administrative data to estimate the associations between the stringency of COVID-19 containment policies and health service disruptions in 10 countries in 2020. We found that a 10% increase in the mean stringency index was associated with a 3.3 percentage-point reduction in relative service volumes after adjusting for COVID caseload, health service type and country GDP. We also found that nearly all individual containment policies were also associated with significantly lower essential health service use, with curfews and presence of a state of emergency being associated with the greatest disruption in service utilization. In contrast, COVID-19 cases were not associated with health service use. Although the overall effect size we find is small (3.3 percentage-point reduction with 10% increase in stringency), this finding has policy relevance and demonstrates a lack of health system resilience. Any reduction in essential health service use could have important consequences on health, particularly in lower income countries where health care utilization had low baseline levels. In future pandemics, policy makers must ensure that the population can continue accessing essential care while policies are in place to contain outbreaks.

Our study highlights that utilization of essential health services during a pandemic is affected by the types of policies implemented by governments to respond to the crisis. There are many mechanisms by which COVID-19 containment and closure policies could affect people’s willingness or ability to seek health services. Although essential services were permitted during stay-at-home orders, people might still be deterred from seeking care if restrictions were rigidly enforced. In South Africa, people reported fearing being arrested or fined as a reason for not visiting healthcare facilities in April 2020 [28]. In Nepal, strict lockdowns were enforced by the police and citizens were punished if they defied restrictions [29]. In addition, the majority of individuals in the countries studied rely on public transport systems. In 2020, public transports were either fully interrupted, passenger capacity was limited, or they were restricted to short timeframes for transport of essential workers. In some countries, the cost of public transport also increased in 2020 [30‐32]. Stringent restrictions might also increase people’s fear of infection in general and when visiting health facilities [19, 33, 34]. Furthermore, the economic effects of COVID lockdowns were distributed inequitably [11, 35, 36]. These lockdowns and containment policies had significant economic impacts on many people, especially those working in informal sectors [37]. Periods of reduced or no income may be associated with lower health care seeking in countries with high out-of-pocket costs. Nonetheless, restrictions may have also led to a reduced need for certain health services including fewer road traffic injuries from reduced mobility and fewer communicable diseases from social distancing, handwashing and mask wearing.

It is important to note that COVID caseloads during this period were not associated with health service disruptions. A study in 18 low- and middle-income countries also found no significant relationship between COVID-19-related deaths and the magnitude of change in any health service [16]. During the first wave of the pandemic, many countries put in place stringent restrictions even when caseloads were low as understandably, much was still unknown about COVID-19 and its effects [37]. For example, in Lao PDR, the worst health service disruptions took place in April 2020, when less than 20 total COVID-19 cases had been recorded in the country.

It is apparent that COVID-19 containment policies have had important indirect effects beyond infection control. COVID-19 restrictions had deleterious economic effects on people globally, especially women [35, 36]. Reports of violence against women and children also increased during COVID-19 lockdowns [38, 39]. Studies have also found that lockdowns may have worsened chronic disease and mental health outcomes. In Nepal, worsening health outcomes among older adults with pre-existing conditions have been reported in part due to significant difficulties in obtaining medications during lockdowns [40]. In Chile, important indirect effects on cardiovascular mortality were reported and were found to be inequitably distributed between women and men [41]. In a study across 15 countries, more stringent COVID containment policies were associated with more psychological distress [42]. COVID-19 restrictions have also resulted in a loss of quality of life. A study using online quality of life surveys estimated that 3,259 million quality-adjusted life-years have been lost globally due to COVID restrictions [43].

Our study shows that the stringency of containment policies was also linked to a reduction in health service utilization. Our findings are consistent with those of other studies that have assessed lockdowns and health service utilization [10, 15]. To our knowledge, only two other multi-country studies have looked at associations between the stringency of COVID-19 restrictions and health service disruptions [14, 16]. Our study contributes to this literature and expands the range of health services and countries studied thus far. We included a broad range of health services and data from 10 countries with varying income levels, COVID caseloads and types of pandemic response.

Nonetheless, our study has limitations. Although a range of countries were included, the association between the stringency of COVID containment policies and service volumes might not be generalizable to all countries or contexts. Second, the associations we describe cannot be interpreted as causal. Models were adjusted for GDP per capita, monthly number of new COVID-19 cases, and type of health service. Nonetheless, it is possible that other variables ( e.g., quality of governance or population age or health needs) may confound the association between restrictions and health services. Third, the administrative datasets used in Chile, Mexico, South Korea and Thailand may differ from the DHIS2 data used in the other six countries, However, co-authors ensured that indicator definitions were comparable across countries. In addition, in some countries, routine health data excluded private facilities, telemedicine visits or some regions in the country [2]. It is also possible that the reporting of routine health data was affected by the pandemic, but rigorous cleaning and verification methods were used and only facilities continuously reporting throughout the study period were included [2]. In addition, COVID cases depend on countries’ surveillance systems and testing capacities and may have been under-counted due to limited testing capacity in the early months of the pandemic, which might bias our results. Although the containment policy tracking followed rigorous validation methods, it is possible that the policies and policy dates could be misclassified [1]. Furthermore, while the association between the stringency of containment policies and disruptions in specific health types (e.g., antiretroviral therapy) may be of interest, our study did not have sufficient power to conduct these stratified analyses by service type. Finally, our study includes data until December 2020. Associations between the stringency of policies and health service utilization has likely evolved in later months of the pandemic.

Our study contributes to understanding the effect of pandemic responses on health service utilization [14, 16]. Although COVID containment policies were crucial to reducing COVID-19 mortality in many contexts, it is important to consider the indirect effects of these restrictions. These indirect effects will likely have long-term population health consequences, as many of the countries studied were already struggling to meet healthcare needs before the pandemic. Containment policies or closures might be necessary to reduce infection, but they should be combined with other policies to promote essential health service utilization. Policies to promote essential health services during times of crisis should be expansive and extend beyond just maternal and child health [44]. Several service adaptations to maintain essential non-COVID healthcare have been implemented in the studied countries and have included the use of telemedicine, community outreach and innovations for medicine delivery. In South Africa, the central Chronic Medicines Dispensing and Distribution program was expanded for community delivery of drugs for chronic diseases [45]. Similarly, in Thailand, service adaptions included telemedicine, shipping of medicine for chronic disease patients and home delivery of medicine by village health volunteers [46‐48]. In Chile, an online platform was developed by the Ministry of Health to provide mental health care [49]. In Ethiopia, adaptations to RMNCH services included an increase in phone based follow-up care as well as provision of multiple months of contraceptives at a single visit [50]. Multi-sectoral collaborations and private partnerships were also designed to support the Ethiopian health system [51]. Multi-month dispensing of antiretroviral therapy (ART) guidelines were also distributed by The United States President's Emergency Plan for AIDS Relief (PEPFAR) in many countries [52]. In Nepal, various hospitals and the Nepal Medical Association initiated telemedicine services and deferred elective surgeries and procedures so that resources could be reallocated to other services [53]. The Ministry of Health in Mexico launched home visits to detect COVID-19 cases while also delivering health promotion and prevention activities, including continuity of care for patients with uncontrolled chronic diseases [54]. These adaptations should be standardized and expanded in times of crisis while ensuring quality of care when services are adapted.

Recently, the WHO revealed that an estimated 14.9 million excess deaths occurred globally between January 2020 and December 2021 of which only 5.42 million were reported COVID deaths, calling for more resilient health systems globally [55]. Health service disruptions likely contributed to this increase in indirect mortality. Plans for pandemic response should consider health system resilience in order to minimize the indirect health consequences of pandemics. As policy makers weigh re-implementing containment policies to reduce infection spread during future waves of COVID or future pandemics, it is important to take these indirect effects into consideration. This study reiterates the importance of resilience in health systems and the need for greater adaptive and transformative capacity, which has the potential to lessen the adverse impact of restrictions imposed during future crises [56].