The covid-19 pandemic called for a massive ramp-up in the ability to conduct diagnostic tests. These tests needed to be able to accurately quantify the full range of the disease, from early to late stages, and to identify infected individuals who were asymptomatic. “Never before has there been a need for testing at this scale,” said Walt. “The need exceeded the entire lab testing capacity of the country—and the world, for that matter.”
Diagnostic tests for covid-19 look for three types of molecules: the genetic material of SARS-CoV-2, antibodies generated in hosts in response to infection, and antigens created by the virus such as the spike proteins on its surface.
The virus’s RNA and viral antigens are typically detectable before the appearance of symptoms, Walt explained. Usually within a week of infection, the body’s immune response begins to produce antibodies that bind to the antigens and lead to a reduction in the virus. However, viral RNA can usually be detected for weeks after the infection has cleared, likely due to residual RNA fragments that remain in the host after the virus has been eliminated.
Because covid-19 is primarily a respiratory disease, molecules from the virus can be detected in the nasal cavity using nasopharyngeal swabs that penetrate deep into the nasal cavity, anterior nasal swabs that take a shallower sample of the nasal cavity, and saliva. Blood can be collected by venipuncture or a finger prick. The first three sample types can contain molecules released from the virus such as RNA or proteins. Blood
usually does not contain the virus, but it contains antibodies produced by the host’s immune system.
Antibody tests measure whether someone has mounted an immune response to the virus. The tests typically occur too late to prevent infected individuals from infecting others, but they are useful in determining whether someone has been infected, how strong the host response to the virus has been, and community prevalence. They are being used in vaccine trials, for example, to determine whether trial subjects are developing robust immune responses and producing protective antibodies, and in communities to determine whether a population has reached a state of herd immunity.
Antigen tests that measure viral antigens such as the spike protein are rapid and can be conducted in a variety of locations. They have lower accuracy than the tests that measure viral RNA, but they are intended for routine use and cost much less—typically $5 to $10 apiece, as opposed to a viral RNA test that is typically reimbursed at over $100. Although typical antigen tests have an accuracy of around 70 percent, with repeated testing their accuracy can approach 100 percent. For example, sports teams were using such tests in preparation for going back to competition. “If they are done daily or every other day, you catch things early,” said Walt.
One innovative form of testing has been to use samples from wastewater treatment plants to detect viral RNA at the community, town, and city level. Such an approach could be especially useful in identifying hot spots before they get out of control, Walt observed.
Diagnostic testing is performed in three main places: central laboratories; point-of-care locations such as physicians’ offices or urgent care centers; and, through direct-to-consumer tests, homes, workplaces, or schools.
Molecular tests that detect RNA are conducted most often at central laboratories using techniques such as polymerase chain reaction (PCR). These processes have extensive personnel and handling requirements, though labs can also use high-throughput robots. Some countries, such as China, do almost all diagnostic testing in central laboratories; others,
such as the United States, have a much more distributed system, with public and private laboratories, different payers, and varying state and local implementation.
The logistics of getting samples to a central laboratory involve, first, the acquisition of a sample by a trained healthcare worker. When enough samples are collected, they are transported to the lab, where they are prepared and tested, after which the results are sent from the lab back to the physician who ordered the test. Locations closer to central laboratories might be able to get results in 24 to 36 hours. However, the greatly increased demand for tests during the pandemic introduced major delays in testing systems. Even the major national diagnostic chains, such as LabCorp and Quest Diagnostics, encountered capacity problems because of the tremendous demand for testing.
Modern high-throughput sequencers can be repurposed for massively parallel testing, where thousands of samples can be molecularly tagged, combined, and simultaneously tested to determine which samples contain virus and which do not. But this technique is cost effective only if these thousands of samples can be transported to a single site, which poses a logistical challenge. If that challenge can be addressed, this approach may eventually replace PCR as the per-sample cost is much lower and requires less labor.
Antigen tests typically are conducted in a point-of-care format, although they can also be done at home. The tests give an answer in 15 to 30 minutes. However, false positives or false negatives can be a problem, because these tests are less accurate than RNA tests.
Early in the pandemic, diagnostic testing encountered serious problems, Walt said. Initially, the Centers for Disease Control and Prevention developed a test and then approved only labs that used that test. Unfortunately, the CDC test had problems such as contamination of the reagents provided with the test. This created large market uncertainties, which delayed some of the bigger diagnostic companies from entering the market.
The Food and Drug Administration’s emergency use authorization (EUA) approval process also caused complications. To issue an EUA, a public health concern must be serious or life-threatening, sufficient evidence must exist that the product may be effective, the known and
potential benefits of the product must outweigh its known and potential risks, and no adequate approved alternatives to the product can be available. Though rapid, the use of the authorization imposed a very low bar on companies. Because of these low standards, over 1000 companies got into the covid-19 testing business, which created challenges both for the FDA and for product quality. Another problem was that validation of tests was done by the companies developing the tests, which resulted in problems such as corner-cutting in the numbers of validation samples used.
Finally, the large diagnostics companies kept a relatively low profile in the market before announcing the availability of tens of millions of tests per month. As a result, many smaller companies that took advantage of the EUA process and got their instruments on the market quickly encountered great challenges getting orders for their products.
Walt drew several lessons from experiences with diagnostics during the pandemic. First, federal coordination is essential. Leaving testing to states is not the right approach, he said. States were vying for tests, and those willing to pay more got them. But when supplies are finite, they must be deployed where they are needed rather than to whoever can pay. “This point is important to ensure that diagnostic testing is available to disadvantaged communities,” said Walt, and “the need for nimble deployment applies both nationally and globally.”
Second, testing should be distributed. While other countries have been able to rely on centralized testing, the United States cannot rely on the logistics of sending tests to a few central laboratories. “In the future, we need to have public health labs in each state or region capable of filling the testing needs until direct-to-consumer and point-of-care tests become available.”
Third, independent validation must be done by accredited laboratories and not left to the companies that are making these products.
Finally, innovation is important, but scale is even more important. The big diagnostics companies had the necessary scale to cope with a global challenge.
During the forum, Sabina Alkire, associate professor of development studies at the University of Oxford and director of the Oxford Poverty and Human Development Initiative (OPHI), discussed a different metric of relevance to the pandemic: trends in acute multidimensional poverty. OPHI defines multidimensional poverty as the distribution of deprivations that strike the same person at the same time. “It’s more effective to fight poverty in an integrated way, not going at deprivations one by one,” said Alkire.
With the United Nations Development Program, OPHI has computed a measure of acute multidimensional poverty over time for 107 developing countries that are home to 5.9 billion people. The Multidimensional Poverty Index (MPI) incorporates three equally weighted dimensions: health, education, and living standards.
The health dimension considers nutrition and child mortality; the education dimension, years of schooling and school attendance; and the living standards dimension, cooking fuel, sanitation, drinking water, electricity, housing, and assets. Deprivation in nutrition is assessed by determining whether anyone in a household is undernourished. The child mortality indicator is whether any child has perished in the past five years. In years of schooling, the indicator is whether no one has finished sixth grade in a household. In school attendance, the measure is whether any child is not attending school through 8th grade. A household is deprived if it does not have cooking fuel, adequate sanitation, safe drinking water, electricity, or sturdy housing. Another measure of deprivation is whether a household does not own one of the following: a radio, television, telephone, computer, animal cart, bicycle, motorcycle, car, or refrigerator.
People who are deprived in one-third or more of the weighted indicators are identified as poor. The data are disaggregated by age and subnational regions, broken down by indicator, and presented with confidence intervals, with robustness tests displaying the sensitivity of the results to a range of plausible parameters.
Before the pandemic, of the 5.9 billion people covered by the data, 1.3 billion, or 22 percent, were multidimensionally poor, Alkire said.
Half of these were children. Two-thirds lived in middle-, not low-, income countries; 84 percent lived in South Asia and sub-Saharan Africa, with roughly equal numbers in each. Of these 1.3 billion people, 99 percent had at least three dimensions of deprivation, and 82 percent had at least five.
Prior to the pandemic, poverty trends had been changing. For example, between 2005–06 and 2015–16, over 270 million people left poverty in India, and the country’s poorest groups—children, Muslims, scheduled castes and tribes—and poorest states reduced poverty the fastest. For the 75 countries and 5 billion people for which OPHI has traced trends over time, MPI had been reduced by a statistically significant amount.
Then the covid-19 pandemic hit. For many, it affected mobility, health, mental well-being, and livelihoods. But for the poor, covid-19 was, as Alkire said, “one more burden that they now carry in addition to the deprivation load they already bore.” When OPHI simulated scenarios based on the United Nations’ predicted increases in nutrition and education across 70 countries, it found that covid-19 set back the global trends by three to ten years.
The comorbidities caused by poverty made the virus more deadly. People who are undernourished or drink dirty water may have a depressed immune response. Those who cook with wood, dung, or charcoal may be more prone to acute respiratory infections. Of the 5.9 billion people covered by the indicators, 3.6 billion face at least one of those deprivations, and 435 million have all three at the same time. “Covid strikes them differently, not to mention if they lost their jobs, have no place to wash their hands, or live in overcrowded or violent surrounds.”
“What does this mean for engineers?” asked Alkire. “That’s a question you are far better placed to answer than I.” But she pointed to several key opportunities. Big data, machine learning, and artificial intelligence could be used to predict levels of poverty in the world. Although the pandemic interrupted many forms of data collection, new sources of data could track the deprivation bundles poor people face and the best ways to counter those deprivations.
Such data can motivate and inform integrated responses, Alkire explained. As an example, she cited the chief executive officer of the biggest bank in Costa Rica who was shocked at the poverty he discovered in a slum that he visited. Subsequently, using an online platform for interviewing staff, the bank was astonished to find that 12 percent of its own employees were poor by the country’s national MPI. “This led to a surge of solidarity and creativity to redress the challenges one’s own colleagues confront,” Alkire said.
An appalling number of people still lack food, water, electricity, and schools and live on a dirt floor with a bamboo roof that leaks, Alkire concluded. “A question is whether the pandemic and the economic recession can jumpstart not only research on the vaccine, as vital as that is, but also a space where engineers could create a true inflection point in poverty by untangling interlinked deprivation bundles that deeply afflict billions of our companions on the planet.”