Mayor de Blasio gets tested (photo: Ed Reed/Mayoral Photography)
As New York City public health officials try to get a handle on a series of COVID-19 “hot spots,” they lack the tools to know exactly how the voluntary testing data they collect skews the true prevalence of the virus, both in communities of concern and city-wide. In the absence of randomized testing, public health officials rely on the results of voluntary testing, an especially challenging process given new evidence some New Yorkers are organizing to game the system against that weakness.
New York's covid positivity rate has been ticking upward in recent weeks, driven largely by surging rates in so-called "hot spots," which has led Mayor Bill de Blasio and Governor Andrew Cuomo to shut down schools and potentially non-essential businesses in impacted parts of Brooklyn and Queens, many of which are home to Orthodox Jewish communities and where the reported rate of confirmed cases outpaces the citywide rate fourfold.
The neighborhoods home to covid uptick “clusters” cover nine zip codes across the two boroughs where the seven-day average Covid positivity rate has exceeded 3% for more than seven straight days, while another 13 zip codes are now labeled “areas of concern” by city health officials. Cuomo announced Monday the state is taking control over local enforcement of covid guidelines like mask-wearing and limited gatherings, and the governor promised hard conversations with religious leaders and the possibility of closure of religious institutions. On Tuesday, he released a plan for tiered, localized shutdowns with the greatest restrictions closer to cluster epicenters and easing as the radius widens.
Despite tens of thousands of tests every day, the city and state have no widely applied randomized testing program and few ways to gauge what members of the population they are missing in their survey of the virus. Health + Hospitals, the city's public hospital network that is in charge of New York's test and trace corps, says testing protocols do not include a questionnaire about motivation for seeking a test, which in lieu of a random sample, could itself hint at gaps in the data. Getting a percentage of those seeking tests who are symptomatic, may have been exposed to a covid-positive person in close proximity, or are just being careful and getting a test based on public health guidelines can reveal finer contours in the official picture of the viral spread.
While key decisions during the city's precarious reopening are pegged to the positivity rate, dictating the opening and closure of schools and businesses, scientists question how well the city and state data captures the spread of the virus among those who haven't been tested.
In the lead up to the announcement about closures in the cluster areas, some residents in Hasidic neighborhoods in New York City received messages not to have their children tested if they have symptoms or to get tested if they are believed to be negative in order to drive the positivity rates down, Gothamist reported. Heavily Orthodox Jewish areas are not the only places seeing covid increases -- the map of recent outbreaks overlaps with areas of the city home to greater numbers of Trump supporters, as reported by Bklyner, including many Orthodox neighborhoods.
The positivity rate is a ratio of the number of positive cases out of the number of people who choose to get tested. Assumptions about that population of test-takers factor into the official assessment and response. In reality much is unknown about who is getting tested -- and who is not -- a variable that can downplay the extent of the virus, which is believed not to manifest symptoms at all in as much as 40 percent of cases.
Some form of randomized sampling, which is being advocated by epidemiologists and set to soon be used on a limited scale in New York City schools, could get at these missing figures and provide a better overall assessment of the virus’ spread.
"Any community anywhere where we're looking at test positivity, you have to ask, 'well, who's determining who's getting tested,'" said Irwin Redlener, a health policy expert and director of Columbia University's Pandemic Resource and Response Initiative, on WNYC radio’s The Brian Lehrer Show on Friday. "Are they just sick people? Are we doing a scientifically randomized study where we're taking a number of people whether they are symptomatic or not and doing testing?"
"If you don't have a good basis for collecting the samples and doing the testing in a randomized, scientific way it's not unexpected that we're going to have problems interpreting the outcome and making policies [inaudible] determining the outcome," Redlener continued.
One of the graphs the mayor uses at his daily press briefings is a product of this. The graph of the citywide seven-day average positivity rate, considered the most stable existing metric for measuring the spread and one that potential closures are pegged to, looks like a backward ski jump, a relatively smooth slope that fell from a peak of 68% in early April to about 1% in June and has remained mostly flat since. But in April, far fewer people were being tested and the criteria limited testing to people who were likely to test positive.
"There is something deceptive, a bit, about any number that we're talking about," Redlener said, because of the lack of randomization.
On September 29, the city's positivity rate was over 3%, the highest it has been since June, but was back below 1% the next day. Officials viewed the one-day spike as an anomaly and cautioned members of the public to focus on the seven-day average, which rose to about 1.65% as of this Monday’s test results, but it spotlighted questions about how the city is accumulating data and assessing spread in the midst of a targeted testing "blitz."
"It is imperative that community leaders, community institutions, and everyday New Yorkers heed the call and go out and get tested and send the message of how important it is to get tested," de Blasio said at a press briefing on October 1.
De Blasio has for days discussed the possibility of local shutdowns in areas seeing upticks in positivity, especially those deemed clusters, and on Sunday called for the closures of schools and non-essential businesses in the nine zip codes. The city has ramped up testing, mask distribution, and public awareness efforts in parts of Brooklyn and Queens, particularly in Orthodox Jewish communities where mask-wearing has reportedly lagged and social distancing rules often ignored. Last week, the mayor announced six "block parties" for mass testing, which they hope will conduct 3,000 tests per day when fully operational, on top of 11 mobile testing sites in covid cluster sites.
Governor Cuomo has attributed the upticks to a lack of compliance with social distancing protocols and has blamed local authorities for failing to enforce his ordinances. On Friday, he issued a directive to local governments to step up enforcement or face penalties of up to $10,000 per day in an effort to tamp down the outbreaks. On Monday, he said the state would take over enforcement of the relevant local law enforcement, though he said he would need local personnel.
The state is also deploying rapid testing in hot spots and coordinating resources with local health departments, and in some cases providing staff, according to Gareth Rhodes, an attorney at the state Department of Financial Services and a member of Cuomo's COVID-19 Response Task Force. The state conducts about 500,000 tests per week, a large portion of which are rapid, where the results are back within a day, allowing officials to monitor the virus in closer to real time. Of the 120,000 tests reported Friday, roughly one-third were completed within 24 hours, Rhodes told Gotham Gazette.
New York's approach to mapping the spread has been to provide opportunities for and in some places heavily encourage testing in the hopes of casting as wide a net as possible, and with it a more representative sampling. Short of a complete census, all testing models simply capture a snapshot that public health officials try to extrapolate to say something about the entire population and set policies to address it.
Officials are looking out for places where testing may be too low. "What is the testing per capita in that zip code?," Rhodes said by phone Friday. "Is it where we think it should be?"
But some testing strategies can have the effect of distorting the positivity rate, even as they are intended to more closely monitor key groups. At various moments during the pandemic, for example, the state has required certain populations to get tested, like health care workers, nursing home residents, or salon employees.
At this point, the best example of randomized testing could be the Department of Education's plan to randomly test 10 to 20 percent of the in-person student body at each school on a monthly basis. A team of NYU researchers recently released a report showing that the city may need many times that rate of testing, a sentiment Cuomo echoed Monday.
"I understand daily, I understand weekly, I don't understand random monthly," he said in response to questions about whether he wanted to see a more robust testing strategy in schools. "If you're saying it could possibly be a month between tests, that doesn't work for me. I would not accept that plan for my school."
The state has been testing wastewater in some state university (SUNY) dorms for COVID-19, as one way to get a random sample, and officials say they are working to expand that testing to other areas.
Randomized testing is difficult to perform. Even with the city's school testing program, parents must consent to their children being tested at random once a month. The children of parents who miss the notice among the flood of novel back-to-school instructions or who object on philosophical grounds will be left in the shadows.
Rhodes says the ideal testing environment would be one of rapid, on-the-spot results, that are affordable, widely available, and accurate.
"Obviously we are not there yet," he said. "I think the scientists and the people on the innovation side will say that's where we will get to eventually, but the question is are we all vaccinated before then?"
In July, Cuomo penned an op-ed to the New York Times, calling for federal funding for "broad community screening" to account for asymptomatic carriers. Rhodes echoed that call in a recent post to Twitter, responding to Gotham Gazette’s reporting about the city’s and state’s lack of randomized testing.
The state is looking at one testing model -- the BinaxNow, by the multinational medical device company Abbott -- that checks some of those boxes: it is quick, relatively cheap (about $5 per test), and does not need to be sent to a lab for results. Rhodes said state officials expect the Trump administration to distribute the new tests in the near future. But the BinaxNow is an antigen test, which is less effective for testing in people who are asymptomatic. The current cost of a rapid PCR test, which is better at detecting the virus in asymptomatic people, is $100 per test, which is "prohibitively expensive," Rhodes said.
Another metric officials rely on to track the spread of COVID-19 independent of testing is the hospitalization rate.
"If covid is spreading, it will result in people going to the hospital from that community," said Rhodes. "I'm looking at what addresses are hospitalizations coming from...It very closely tracks where we are seeing new cases."
If hospitalizations in a region are up, but the positivity rate doesn't reflect an uptick, the state would set up rapid testing there, Rhodes said.
"It obviously is not ideal because you want to catch this long before people go to the hospital," he added.
But even this measurement isn't really randomized. Access (and admission) to an emergency room is not uniform across the population, and multiple factors influence whether a person receives treatment, especially during the pandemic's so-called "apex" in New York, and especially in predominantly Black and Latino communities.
The city's covid testing does not include a questionnaire about why a person is seeking testing -- because of potential exposure, proximity to testing sites, or being referred by a doctor or an advertisement -- questions that could give officials a sense of the profile of individuals who are getting tested and reveal potential blindspots in virus surveillance.
"We do not ask why they are getting a test," wrote Karla Griffith, a spokesperson for Health + Hospitals, the city’s public hospital system, in an email.
Individuals are asked some questions before receiving a test, like whether they've had COVID-19 previously, if they have symptoms, if they have upcoming surgery, if they are pregnant, and where they work. "These questions are for reporting purpose to the NYS Department of Health and to assist with contact tracing efforts should someone test positive," Griffith wrote.
In Borough Park and Williamsburg, two areas with large Hasidic populations, parents received robocalls urging them not to test their children and to remove them from school without getting tested if they are showing symptoms, as Gothamist reported. One religious leader urged members of his congregation who have already gotten the virus, and are believed to be less likely to still have it, to get tested in order to drive the official positive rate down. And some medical centers in the communities have refused to share test results with officials, according to Gothamist.
The city's Department of Health and Mental Hygiene did not say how or whether it was assessing the impact of this behavior.
Asked how the state accounts for potential distortions like the ones reportedly being machinated in certain Hasidic clusters, Rhodes said, "Let's say you don't test in certain areas, or test less. If covid is spreading, you're going to see an uptick in hospitalizations."
"In these areas we haven't seen what I would consider to be a testing desert," he added. "The message we've been saying is urging everyone to get tested."