It has been almost a year since COVID-19 was declared a pandemic, and since its early days, we have called attention to the importance of preserving essential health services and warned that COVID-19’s spread, and measures to slow its spread, may cause indirect health effects.
As response efforts continue and countries face new waves and variants, there is emerging evidence exploring some of these indirect health impacts, and new resources for how to track and address them. For example, last month, the World Health Organization (WHO) published its interim guidance on analysing and using routine data to monitor the effects of COVID-19 on essential health services. This guidance can be considered a companion to WHO’s interim guidance on maintaining essential health services published back in June.
In this blog, we describe some major global efforts that examine whether essential health services have been disrupted during the pandemic, summarize what they tell us, and highlight some of the remaining gaps in our understanding and knowledge. We recommend actions that global health organizations and their partners can take to start filling these gaps with the data and evidence needed to better understand the scope of disruptions and how to address them.
What we know based on real-world data and evidence
Were essential health services disrupted? The answer is an unequivocal yes. Last year, WHO published a “pulse” survey on continuity of essential health services during the COVID-19 pandemic, asking ministry of health officials between May and July 2020 to assess the impact of the COVID-19 pandemic on up to 25 essential health services. “Disruptions of essential health services were reported by nearly all countries, and more so in lower-income than higher-income countries,” WHO wrote. “The great majority of service disruptions were partial.” Importantly, WHO defined “partial disruption” as a change of 5–50 percent in service provision or use, while severe or complete disruptions were defined as a change of more than 50 percent in service provision or use.
In the same vein, results from the first wave of UNICEF’s pulse survey were published last June and provided more granular data with six categories of disruption: none, less than 10 percent, 10-25 percent, 25-50 percent, 50-75 percent, 75-100 percent, and “do not know.” Here, too, drops in coverage were reported across all 16 health services surveyed. However, for each service, a number of countries reported no drops in coverage (up to 40 percent of countries reported no drop for HIV treatment) and similarly, countries stated that they did not know (up to 60 percent of countries reported not knowing for noncommunicable disease—NCD—treatment).
The World Bank has been supporting the use of phone surveys to monitor the impacts of COVID-19 on households and individuals. The results of this high-frequency monitoring are now available and contain harmonized, comparable information on over 93 indicators across 14 key topics, including health, from 45 countries. These data show, for example, that significant percentages of households surveyed reported not receiving medical attention due to fear of catching COVID-19 or due to stay-at-home orders.
Routine data, including health management information systems (HMIS), is also available at the country level. HMIS are the systems whereby health data are recorded, stored, retrieved and processed. In theory, they can provide “real-time” data to track the magnitude, scale, and scope of disruptions to a range of health services. In practice, however, there are longstanding concerns regarding the quality, timeliness and completeness of the data, and these concerns have been exacerbated by the pandemic. Hence, significant data cleaning and processing is required to derive meaningful insights. Several development partners, including the Global Financing Facility (GFF), have been working with countries to analyze routine data. However, this work is not often published. Details from GFF have only been shared in a blog on 21 September 2020, at which time researchers had analyzed data reported by more than 63,000 facilities in 10 countries. The analysis of data through June 2020 included a finding that childhood vaccination was the most disrupted service in studied countries.
Similarly, researchers at Harvard University are working on an analysis in 13 countries, yet the only public description of this work appears to be on the Prince Mahidol Award Conference website, where preliminary findings were presented on a panel event last year. This work includes 13 countries (Chile, Ethiopia, Ghana, Haiti, Jamaica, Laos, Mexico, Nepal, South Africa, South Korea, Thailand, Zambia, and Vietnam), and preliminary data were shown for Ethiopia, Mexico, Haiti, South Africa (specifically Kwazulu-Natal), Mexico, Nepal, and Thailand. The “up to” date of the data analyzed varies in each country from June (Haiti) to October (Thailand). According to their presentation, all countries included in the study experienced a decline in the majority of services.
What we still don’t know
While we have some data on the depth of essential health service disruption, it remains unclear how long these disruptions linger and their consequences, and how these impacts are distributed among subsets of the population and those comparatively disadvantaged or marginalized. That’s in part because we’re still very much in the midst of the pandemic, with countries now tackling a second, and in some cases a third, wave.
We also need more timely and granular data. For instance, while useful as a call to arms, the pulse surveys from the WHO and UNICEF need to be supplemented with real data. To pick one example, WHO’s definition of partial disruption includes disruptions from 5-50 percent, a range that would clearly have very different consequences and implications for health policy. Moreover, we have not seen much by way of sub-national reporting on disruptions, or disruptions by gender or income quintile. Aggregate indicators may mask differential impacts across different population subgroups and health services.
Many studies have focused on maternal and child health or communicable diseases, with little coverage on some indicators that are not typically well captured through routine systems. This may include coverage of preventative services, and diagnosis and management of NCDs, in particular for conditions that may have deteriorated rapidly during the pandemic such as diabetes and mental illnesses. Finally, efforts have been concentrated in some countries. In WHO’s pulse survey 105 out of 159 countries replied. The first wave of UNICEF’s dashboard showed data from 85 countries, while wave two showed data from 159 countries. The World Bank’s dashboard has data from households in 45 countries.
1. Coordinate and harmonize efforts.
In a recent report, PATH analyzed available data on disruptions to essential health services in Burkina Faso, Ethiopia, India, Kenya, Nigeria, and Pakistan. The authors noted that difficulties in the design and inquiries of dozens of different survey efforts became limiting, and advised further efforts to coordinate and harmonize future survey efforts. We welcome WHO’s recent interim guidance on essential health services, and we hope that moving forward it will contribute to greater harmonization of efforts like PATH suggests.
2. Fill data gaps.
There are some countries where we know very little and others where we know relatively more. Based on today’s evidence-base, generalizing results is difficult because there are differential impacts across services, countries, population groups and across different phases of the outbreak. Recognizing that data cannot always be shared, what needs to be systematically surfaced from the efforts to date is where do we not have any data, for which services, which population groups? Researchers need to collect new data or process data from existing HMIS to ensure that we get insights from, and to, more countries. We need data on a more comprehensive coverage of health services (beyond those traditionally supported by development partners), and the data should be disaggregated by gender and location.
3. Triangulate estimates.
The global health community needs WUENIC (WHO/UNICEF estimates of national immunization coverage)-like estimates of country and program-specific estimates of disruption. We need triangulated “consensus” estimates from the many sources of data on this topic out there, including UNICEF’s, WHO’s, the Bank’s mobile phone surveys, data from routine data systems, and the plethora of program-specific surveys in countries, crowdsourced data, and others.
4. Create collaborative efforts to improve the speed at which data are analyzed and insights are generated.
There are obvious challenges with sharing of health data; for instance, patient privacy needs to be respected when sharing electronic medical records or any forms of data that may include patient identifiers. However, given the need for timely information, the world needs innovative approaches to fostering collaboration. One example that gains momentum in the health community is the establishment of safe data havens. Safe data havens are a means to release potentially sensitive data in an effective, trustworthy, and safe manner, through de-identification and with good governance (with the establishment of rules on use and reporting) that then potentially fosters collaboration on data cleaning, preparation or analysis.
5. Evaluate service adaptations and strategies to mitigate indirect impacts.
With more timely and better data on essential health service disruptions (ideally including health impacts), as well as data on epidemiological trends and response measures, we can start to generate much better evidence regarding what service adaptions work to minimize the detrimental impacts of COVID-19 on other key health areas, while mounting an appropriate response to slow the spread of—or even eliminate—the virus.
6. Invest in resilient data systems.
Not only have services been disrupted, but also data systems, further impeding our understanding of what’s happening on the ground. We have become over-reliant on models, to the detriment of data systems, even though the former would clearly benefit from stronger data systems. In the absence of strong routine data systems in most low- and middle-income countries, the global health community remains largely reliant on quarterly, educated guesses from the field.