Skip to Main Content

Despite three weeks of assurances, most Americans who have symptoms of Covid-19 still can’t get tested for SARS-CoV-2, the coronavirus that causes the disease. Last Friday, Sen. Brian Schatz (D-Hawaii) tweeted succinctly, “Hey where are the tests.”

The hardest-hit states are rationing tests to health workers and those who have Covid-19 symptoms. The Trump administration, meanwhile, continuously reassures us that we are getting more and more testing out there.

While increasing the number of tests is absolutely necessary, it is not sufficient. If we want to stop Covid-19, we must also focus on gathering other essential information as these tests are administered. There’s little value to getting a bunch of positive test results without being able to understand the clinical conditions of the individuals they came from, how they were exposed to SARS-CoV-2, and, perhaps most importantly, how they compare to people who test negative.


Even South Korea, which has tested more than 250,000 people, has failed to do this effectively.

Based simply on the number of positive cases, young adults would appear to be at highest risk of Covid-19 in South Korea. But is that due to higher testing rates among this age group? We don’t know.


We need more than data — we need insights. That’s why we need “smart testing” to address four fundamental public health questions.

The first two questions, which will be key to guiding policymakers on whether the extreme measures taken to suppress the outbreak are working, are these: Is the Covid-19 outbreak in a city or state getting better or worse? And how fast?

The answers to these questions cannot be determined just by looking at how many people test positive, especially as more testing capacity comes online. What we really need to look at is the percentage of people in the population who are testing positive each day. This requires knowing not just the number of positive tests, but how many tests of people in the overall population were performed — the numerator and denominator.

Epidemiology is a bit like baseball. Knowing that a ball player has gotten 134 hits isn’t that informative. What is informative is knowing that those 134 hits were made during 335 at-bats, which translates into a batting average of .400. But we can only know the batting average if we know the player’s total number of at bats and hits. It’s the same thing for the coronavirus: We need to know the number of all tests in in each age group and each locale, as well as the number of positive ones.

To know if Covid-19 is getting better or worse, we need to know how the percentage of positive cases — not the number — changes day by day, accounting for delays in testing and reporting and how the percentages change in response to public health measures such as sheltering-in-place or suppression. This will tell us how effective these measures are in curtailing the spread of SARS-CoV-2 and thus whether when social restrictions could be relaxed or additional policies might need to be implemented.

The other two questions that need answers are essential for guiding doctors in their medical decision making: Who is getting Covid-19? And what are their outcomes?

Consider the CDC’s latest study on the first 4,226 coronavirus cases in the United States. For one-third of these cases, no information was available about whether the patient had to be hospitalized. More than half lacked information on whether the patient required treatment in an intensive care unit. For just under half, it was unclear if the patient survived the infection or died. And, vitally important for a disease that has such disparate impacts on different segments of the population, there wasn’t even the most basic demographic data — age, sex, race— for 10% of the cases.

If we are going to get a handle on this fearsome disease, we need to stop emphasizing the sheer number of tests and whether any citizen can walk up and get tested. Instead, we need to insist on four basic groups of information gathering as we expand testing.

First, as a condition of receiving approval and test kits, laboratories should be required to submit basic information like age, county of residence, and testing site on every person tested — not just the positives.

Second, we need a sero survey of multiple communities. Such studies test blood samples from randomly chosen individuals in a defined population. This is the way to assess the real percentage of people in a community who test positive for recent coronavirus infection. This gives a picture that is wider than just the individuals who are bringing themselves in for testing. Right now we simply have no idea how many Americans are infected with the coronavirus. This will be a key input to models trying to predict when herd immunity can begin to blunt the outbreak.

That means we need to do random testing of people in communities like New Rochelle or Seattle that have been hard-hit by the outbreak. When one of us (F.M.) worked in New York City leading the fight against West Nile virus, health workers went door to door, testing and surveying households in the epicenter of the outbreak to understand the true case-fatality rate and the true infection rate of the virus.

Third, the CDC needs to rapidly help state and local public health agencies set up what is known as sentinel testing for Covid-19. This means they need money and technical support to collect comprehensive clinical and exposure information on a systematic sample of patients. That could be through existing sentinel clinics set up for influenza surveillance, drive-through coronavirus testing sites, or having select hospitals volunteer to systematically collect key information and test results for a sample of patients with cough and fever or severe acute respiratory illness.

Fourth, we need to look deeply into the information we already have. We should investigate data being collected daily in nearly every state on what are known as syndromic surveillance clusters. One of us helped design and build such a system for monitoring emergency room visits in New York City, which are now showing unprecedented increases in respiratory and flu-like complaints since March 1. By last Thursday, there had been 4,663 emergency visits with these complaints in 2020, while last year there were only 1,603. We need to look at the data of these patients to understand how much of this increase in emergency room visits is actually related to Covid-19 and how much is due to worried people imagining they are infected. This could be a key way to understand just how much we have undercounted the severity of the spread of Covid-19, and a tool that other communities can use to detect when the outbreak is spreading rapidly.

As our nation wrestles with the challenge of Covid-19, we should take heart from history. This is not the first major public health challenge our nation has faced, and we will overcome it. But we need to remember, as Tom Frieden, former director of the Centers for Disease Control and Prevention, recently said, “testing is not a panacea” but intelligent application of testing will provide vital public health information needed to effectively fight this pandemic.

Farzad Mostashari, M.D., is the CEO of Aledade Inc., the former assistant commissioner of the New York City Department of Public Health, and a former senior official in the U.S. Department of Health and Human Services. Ezekiel J. Emanuel, M.D., is the chairman of the department of medical ethics and health policy at the University of Pennsylvania and serves on former Vice President Joe Biden’s public health advisory committee addressing the coronavirus outbreak.

  • Correct. Random sampling (not sampling from symptomatic cases) is the way to go if you want to conclude anything about the prevalence of the virus in the overall population.
    Knowing the number (or even the percentage) of positive cases from symptomatic cases says nothing about the population parameters. Why ? Because testing based on non-random samples does not extrapolate to the population.
    Statistics 101.

  • Such a poorly-informed article….. that only shows how un-“smart” both authors are. And in the end, they self-servingly insert a blurb of their own product development!

    I can only say that the authors seem to have forgotten what they were doing (or supervising/managing) in their previous jobs…… All serious countries that do testing will of course be analyzing / crunching the data to spot target areas + target demographics and assist in contact tracing.

    Do you not think that countries like South Korea would take into account that there are more young people being tested positive (naturally as there is a greater population of under-40s)? Since when did you ‘discover’ the value of per capita or geographical/local data? Where were you weeks ago when various media outlets were talking about the smart phone apps the far-east asians used to track confirmed cases?

    ….. Stat really needs to up their game in commissioning “smarter” writers……

  • Don’t say anything positive. What r u doing to increase stats. Sick of your negative. Sick of Congress w hidden agenda. Sick sick and embarrassed of all of the media who can’t get behind and help. Just the way it has become. Easy to criticize. Different if you are in a moths blow your horn position

  • More testing will be efficient way for spreading of corona virus. So how can be increase this testing in primary stages. In my opinion, use some simple smart devices like smartphone with all protective materials for detection of primary stage cases for the corona virus and only focused on those persons who have abnormal for the further testing procedure. Now lot of smartphone facilities for testing blood and other kind of health measures.why we can use this idea for the primary testing

  • This is exactly what I’ve been thinking. I’ve been looking at the percentages of positives using the data from The Covid Tracking Project, and what is interesting to me is the states like NM (6,842 tests with 100 positive which is 1.5%) and NC (8,539 tests with 398 positive which is 4.7%). Now look at OH (704 tests with 564 positive which is 80%) and DE (140 tests with 104 positive which is 74%). These numbers are all over the place and I’m trying to figure out how to interpret them. Anyone have any thoughts?

  • if we could just test 100 random persons, we could establish the percent who have Coronain the whole population . It is very important that it is completly random. Antibody tests can show how many have had it. 100 random of these would be fantastic. Maybe authorities are afraid of doing this. It could make them look really stupid if we see that over half of the population has antibodies, and possibly show that it has been around much longer than we think.

  • What a load of rubbish. We need smart testing not just more. The issue here is we have been listening to so called politicians self proclaimed medical experts and also doctors who have too much of a god status. We need to test everyone!. Why? Because we need to assume it is already spreading out of control and we already know many will be asymptomattic but contagious. Test every single citizen. Put those positive for the virus in quarantine. There is no such thing as smart testing anymore becasue we haven’t been smart about what we have been doing from the start. Why? Becasue we have been listening to incompetent leaders who make poor judgement and poor decisions and really shouldn’t be leading or making decisions that impact others,.

  • Here is a couple of problems that you are missing.

    1: In your own publication the article, “A new self-triage tool can help you decide if you need medical care for Covid-19 By Michael Hochman, Michael Wang, and Katy Butler /March 20, 2020” stated, ‘…will suffer symptoms that are no more serious than a bad cold or a mild flu and will be better within two weeks. There is no need to seek testing or go to a doctor’s office.’

    So, if I feel like I have a cold or the flu, but actually have COVID-19, stay home, have some chicken soup, watch TV, and nap a whole lot and get better, well, the data that you are trying to collect and analyze is going to be fairly skewed.

    2: I am all for an individual’s privacy, in fact I am an Intelligence Oversight Officer in the Department of Defense, so I am keenly aware of the importance of protecting certain personal information. The adherence to what is currently our privacy laws just WILL NOT WORK in a pandemic. Now I am not saying, I want to have everyone’s personal information posted on Facebook, but we are going to have to name names here. We can not have the politicians dicker about this, we cannot have the lawyers out in salivating packs, we cannot worry about little Timmy’s name being published because it may cause him some kind of “trauma” that will ruin his entire life.

    A scenario for you. Little Timmy has COVID-19, but does not know it, for whatever reason. He is contagious though. Now, little Timmy is playing with his friend little Bobby from school that lives a couple streets over. They are doing what 10 year old boys do and however it happens, little Timmy has given little Bobby COVID-19.

    Now, the street lights are coming on and it is time to go home. Little Timmy gets sick a few days later, it is diagnosed COVID-19. Little Timmy gets the care he needs. Thankfully it is a mild case and he gets to hang out at home. Little Timmy is now a statistic for the news and doctors to track and yadda, yadda, yadda.

    Back to little Bobby. The day after little Timmy gets diagnosed with COVID-19, little Bobby and his family went to visit his grandmother. She lives in a retirement community, has lots of friends, and loves to show off how “big” little Bobby is getting. Little Bobby stays a week. He becomes infectious during his last two days there. Little Bobby does not become symptomatic until a day after he gets home. To late for the 30+ people that little Bobby just spent two days infecting.

    But, everyone rest easy. Little Timmy’s name was not publicized. Thank the Maker that he won’t suffer any “trauma”. Just little Bobby will when his grandmother dies a week or so later.

    One of the most important actions needed to help “flatten the curve” during this pandemic is to know who actually has it.

    Back to the article I mentioned at the start of this. The “self-triage tool” asks, ‘Have you been in contact with someone diagnosed with COVID-19 either while that person was sick or within a week of their developing symptoms?’

    How the heck would I know? Unless it is a friend who tells me they have COVID-19, I have no idea. Now, if I knew it was little Timmy, because I saw him on a list and I tutor him in the evenings… well now I can answer the question, (so the “self-triage test”) would actually we a helpful tool. Then, with the information of who specifically has it, we can begin to back-track it, get the right people tested, the right areas sterilized, and inform which medical centers to prepare for a possible influx of COVID-19 patients. All of which would allow for a much better distribution of scarce resources.

    TL;DR, you have COVID-19, your name, where you have been, people you have been in close contact with, and other vital pandemic related information; goes into an open access national database that is actually easy to use (if the average 10 year old child cannot use it, you made it too complicated). That way, before little Bobby went to kill, sorry, I mean, visit his grandmother, his parents could have looked at the list, seen little Timmy was sick, then not traveled. Instead, they could begin to prepare for the possibility that little Bobby might be sick and the local health officials et. all could begin getting actual COVID-19 infected people off the streets and into their beds.

    You could even just make it a voluntary thing. If you have COVID-19 and are willing to share, please post the following information so that others might be safe. It needs to be nation wide or its value goes way down, but it would be better than nothing.

    Sorry for the long rant, but not actually knowing who I have been in contact with that has COVID-19 is beyond stressful. Not knowing if today is the day I come home from work and infect my children.

    If any or all of this is OBE, well, oops. I can only read so much during the day.

  • What the hell has happened to the CDC? They are a complete good in this entire response.

    In the massive void, state health departments need to take over the process of allocating rationed access to testing and prioritize critical public health protocols to control testing including random population sampling tracking.

    The states entire plan was to defer to the CDC. Well that failed, we need plan ‘B’ ASAP!

    • So when you retire from the FDA, you can just do a 180 and promote an unproven treatment, and you think it gives you more clout to say you were from the FDA? You also completely missed the point about getting significant data from testing that will lead to an actionable response. Stay retired.

Comments are closed.