IDinsight has partnered with the Global Health Group at the University of California, San Francisco and Evidence Action to form a Pandemic Response Initiative (PRI). In the Diagnostics workstream of this initiative, we have worked closely with the Clinton Health Access Initiative (CHAI) to develop structured, evidence-informed guidance for policymakers in low- and middle-income countries (LMICs) on strategic prioritization of limited COVID-19 test supplies. Our work has focused on the development of the PRIoritize_Dx Tool, a beta version of which will be publicly available in the coming weeks.
In this blog post, we recommend that COVID-19 tests in resource-limited settings be strategically allocated according to populations and priority groups based on three criteria: 1) tests are most likely to reduce new infections and support with epidemic management principles, 2) test results are likely to drive impactful action, and 3) the use of tests will reduce uncertainty about an individual’s or population’s COVID-19 status. In this way, LMIC policymakers can optimize the use of diagnostics even during periods of widespread test kit shortages.
The PRIoritize_Dx tool is now being rolled out and tested with CHAI , but we are seeking additional government partners interested in actively piloting this approach to testing and monitoring the results. For questions and comments, reach out our team at diagnostics@idinsight.org
Testing is one of our key tools for curbing the spread of the COVID-19 pandemic and is especially important as we wait for effective treatment options and the development and roll-out of a vaccine. Test results can lead to better decision-making by individuals, health facilities, and policymakers which can improve epidemic control and save lives. Tests clarify an individual’s COVID-19 status, which can encourage daily household decisions such as staying home from work or school, isolating from others, or seeking care. In health facilities, test results can also help providers make the best possible decisions about the care and treatment that patients require, as well as how to prevent transmission among health care workers. Population-level testing data can provide key inputs into resource allocation and lockdown decisions, including signals of emerging case clusters, insight on whether current prevention measures such as physical distancing are working, and indications of how widespread the virus has become in a community.
Testing capacity has expanded dramatically in many countries since the start of the epidemic. As of early August, about 8.7 million tests were conducted across all African countries, which is closely approaching the October target of 10 million tests set forth by the Africa CDC. Furthermore, a number of countries have implemented highly innovative strategies: Rwanda has a complex test pooling system and is using robots to monitor some COVID-19 patients. In Ghana, drones are now transporting some testing samples, an approach that is now being piloted in the United States. In India, the government has invested in a massive scale-up of rapid antigen testing, which has drastically expanded testing capacity.
There are signs, however, that these expansions in capacity and innovation are still insufficient to overcome undeniable shortages in testing capacity, including both the supply of test kits and the ability to efficiently process them, due to lack of human resources and lab capacity. The WHO offers a benchmark of a 5% positivity rate as an indicator of sufficient testing volume. While countries like Canada (0.8%) and Germany (0.9%) have been able to exceed this recommended test positivity rate considerably, many LMICs like Bangladesh (19.9%), Nigeria (16.5%), and Kenya (10.2%) are still struggling to push rates below the 5% benchmark. These high positivity rates indicate that expansions in testing are not keeping up with increased rates of transmission. Therefore, the official case counts alone in these countries are unlikely to reflect the true number of positive cases. In addition to low test volume, many health systems in LMICs as well as in high-income countries are also too slow in processing and communicating test results to meaningfully change behaviour. Even in South Africa, arguably the regional leader in testing in Africa, turnaround times have sometimes reached 12 days or more. In the Philippines, the Department of Health reported backlogs of over 12,000 tests in early July due to supply shortages.
When operating with insufficient testing capacity, it is critical to have a systematic way to allocate constrained COVID-19 testing resources to the highest impact uses for epidemic management. Effective allocation of limited testing supplies requires carefully considering and weighing trade-offs of different testing strategies. Over the past five months, our team has developed strategy guidance for policymakers to effectively allocate limited numbers of tests in their context to uses with the highest value.
We propose that a high-impact test meets the following three criteria:
These three criteria form the backbone of the PRIoritize_Dx tool. This tool provides a clear, actionable, and data-driven approach to support policymakers in effectively prioritizing their COVID-19 tests. The tool is interactive, allowing users to input key information about local conditions (such as isolation behaviour, test turnaround time, and degree of contact tracing in place) and receive tailored recommendations appropriate to their country context. We have designed this tool in close collaboration with our partners at UCSF and CHAI, and are currently piloting its use across a number of CHAI country offices across Sub-Saharan Africa and Asia.
These three criteria serve as a framework to prompt careful consideration of the highest-impact uses of a country’s limited tests for epidemic management. The first criterion measures how impactful the action based on this test could potentially be if all goes to plan. The second criterion notes how impactful we think this action will actually be, based on the reality on the ground for a given context. The third criterion prompts us to consider whether action can be taken presumptively (either a negative or positive COVID-19 status) even without a test.
How much could this test potentially contribute to reducing transmission, gaining information on the spread, determining treatment for patients, and preserving health system capacity?
As in other epidemics, the ultimate goal of any response effort to the COVID-19 pandemic is to reduce the overall morbidity and mortality in a population. Diagnostics are one key lever to achieving this goal, in addition to prevention and treatment. In particular, diagnostics, when acted upon, contribute to this goal through four pathways:
When laying out every possible use for a COVID-19 test, some uses are better positioned to contribute to these goals than others. Overall, we start by prioritizing the uses that are most obviously and directly linked to these four big-picture pathways for impact. These goals vary in importance depending on the stage of the outbreak. For example, in areas where there are no reported cases, the most important goal should be in assessing disease burden in order to rapidly contain cases if an outbreak is confirmed. The testing strategies most closely related to this goal, such as community or facility-based surveillance, are high-priority. However, in areas with large, ongoing outbreaks, the most important goal would be reducing transmission to get the rate of new infections down to a more manageable level. Targeted testing of people at a high risk of transmissions, such as frontline health care workers, essential workers, and recent travellers, most directly contribute to this goal.
Based on the context, how likely will this test actually lead to the desired change in behaviour?
The first criterion was about a test’s potential contribution to epidemic goals; the second focuses on whether, given the context, this test will actually contribute to these goals based on its likelihood of driving action. This criterion is the most crucial of the three because testing alone cannot achieve epidemic control if action does not follow. These actions may be at the individual level (e.g. someone’s ability or willingness to isolate or access to improved care or treatment) or the community level (e.g. localized lockdowns or changes in resource allocation). Ultimately, in resource-limited settings, we should prioritize tests for situations when the results are most likely to be directly used.
Imagine a hypothetical scenario where a health care worker has two options for whom to test. The first is a severe patient who has been recently admitted to the hospital. The second option is a symptomatic essential worker at a factory where his employer requires proof of a positive test result in order to use sick days. In both cases, it will take 3–4 days to receive a test result. For the severe patient, there are urgent actions which need to be taken to provide improved care and treatment. His doctors cannot wait this long for a test result before making these decisions, and so action will likely be driven by symptoms and exposure risk rather than the test itself. Due to long turnaround times, testing this patient may not be that valuable, and could actually stop them from receiving timely care. On the other hand, a test result for the essential worker will directly enable this person to isolate when he otherwise wouldn’t have been able to, due to restrictions from his employer. Therefore, testing the essential worker is the best use of a test.
Do we have a reasonable understanding of this person or this group’s COVID-19 status without needing a test?
Even if a test is highly relevant to epidemic management goals, and likely to trigger direct action, we must still ask ourselves whether this information must come from a test. While diagnostics are generally the most definitive information about someone’s COVID-19 status, a health care provider or policy-maker also considers the presence of symptoms, risk profile, likely exposure to confirmed cases, and the overall prevalence in the surrounding environment. In some scenarios, this other information provides a clear picture of someone’s likely status, meaning that a test result for this individual will not provide much new information. We should prioritize tests in contexts where our pre-existing level of uncertainty is the highest, and a test therefore has the greatest contribution to our knowledge base.
Consider a health facility in a capital city that is undergoing a large, uncontrolled outbreak. Imagine that, hypothetically, a health care worker must choose between testing two contacts of a confirmed case: either the index patient’s husband, who has a fever and shortness of breath, or a neighbour who works as a bus driver, and feels well at this point, but came into close contact with the index case. The household contact is most likely positive, based on both symptoms and likely close exposure from a known case. Does a test really add new information? We are likely more uncertain about the COVID-19 status of the neighbour, who seems to have been exposed but hasn’t experienced any symptoms yet. Therefore, given this uncertainty, the health care worker should prioritize testing the neighbour over the husband, because the test result will provide more new information that is difficult to determine otherwise.
COVID-19 testing resources are scarce, especially in LMICs. It is critical for decision-makers to carefully consider the optimal allocation of these resources in order to approach achieving epidemic control. These three criteria provide a framework to guide these challenging decisions. The PRIoritize_Dx tool supports LMIC policy-makers and public health advisors in applying this framework. A beta version of our tool is now available here. This tool is designed to weigh these three criteria along with elements of a user’s COVID-19 response context to give highly tailored recommendations. The trade-offs vary across countries given differences in factors such as lab system capacity, the ability of community members to self-isolate, and the operational feasibility of different testing approaches (e.g. contact tracing). But more broadly, we believe that these criteria are highly applicable for anyone involved in advising, implementing, or monitoring COVID-19 response in resource-limited settings. By ensuring that tests are prioritized for uses that are closely related to epidemic management goals, reduce uncertainty about someone’s COVID-19 status, and drive impactful action, policymakers can craft context-specific testing strategies that optimize the use of limited diagnostics.
We collaborate with government leaders to develop and roll out data-driven policy solutions aligned with their priorities and within their budgets.
10 August 2021
13 November 2020
1 July 2021
26 January 2021