Poor state reporting hampers pandemic fight

Short of a cure or vaccine, our most powerful weapon against the COVID-19 pandemic is the information: The right data can tell us the severity of the problems we face and whether social distancing is working. Despite considerable handwringing by public officials about what we don’t know about this disease, state health departments already have most or all of the relevant data. Unfortunately, they don’t make it available in ways that provide a clear picture of what’s happening. 

This problem can be solved at little cost, with the stroke of a pen. The Centers for Disease Control and Prevention (CDC) or the White House Coronavirus Task Force should issue guidance about how to report information that is already known to most state and local health departments but currently presented in ways that make it difficult or impossible to track the daily course of the epidemic. 

Let’s first stop paying so much attention to cases, which most departments focus on. These make for frightening click-bait headlines, but right now are not useful. We cannot tell whether an increase in cases is due to the daily increases in testing, more testing of symptomatic patients, or a true change in the infection rate. So-called case maps are closer to testing maps.  

Case counts will be useful only when we move to scientific surveillance of population samples to properly estimate prevalence. They currently contribute to three dangerous illusions: they underestimate true prevalence because they don’t count infections scientifically, they exaggerate the increase in infections as we test more people, and they inflate the apparent mortality rate per infection because we’re underestimating the number of people infected. This is a formula for transforming rational concern into a panic.

What numbers can we rely on now to know whether the current lockdowns are working? We should be looking at trends over time in three measures: new hospitalizations, new ICU admissions and the daily number of deaths. These indicators reflect the true prevalence of infection, not just of our ability to find them through testing, and they also are measures of the burden on the healthcare system. Trends in those indicators tell us directly whether interventions are working for the outcomes that matter.

Not one of the 11 states with the highest number of COVID-19 deaths provides daily trends in even two of those numbers while reporting on cases in copious detail. Washington State comes closest by reporting daily deaths and weekly (not daily) hospitalizations on graphs that show how the state is doing overtime. None of the other 10 (New York, New Jersey, Michigan, Louisiana, California, Illinois, Massachusetts, Georgia, Connecticut and Florida) provides or displays trend data in either daily admissions or deaths. All report deaths as cumulative totals. Similarly, the five that report any data on hospitalizations (California, Louisiana, Florida, Connecticut and Georgia), provide only the current hospital census or cumulative totals, not the trend in daily admissions. How can we know if we are flattening the curve without seeing curves?   

The CDC website provides little help. On April 8th, the main page reported 12,754 total U.S. deaths, but for more detail one is directed to a page with weeks-old data that reports 3307 confirmed U.S. deaths with no trend data or curves. Rates of new hospitalizations are displayed for 14 states, but only from selected counties, the interactive data are hard to find, are 1-2 weeks old, and are averaged weekly, not daily. When delays by decision-makers of even 3-5 days make huge differences, this is too slow.  

The best health department reporting comes not from a state but from New York City, whose website clearly displays daily hospitalizations and deaths, making apparent improving trends this week that made headlines. 

Because of this information vacuum, enterprising scientists and media outlets have had to step in, asking questions about why data is not easily accessible, or re-displaying data obtained from the health departments themselves in interpretable form. Newspapers should not have to perform basic health department functions.

It is urgent for citizens, journalists, policymakers, and researchers to see as quickly as possible how a particular state and subregion is faring under social distancing policies, and soon, under their relaxation. Without that information we’re not just flying blind, we are willfully donning a blindfold. 

In the short term, the Federal Government should issue formal guidance about the content and format of coronavirus data reporting from all state and local departments of health. The CDC last week released detailed guidance on the content of death certificates for those who died from COVID-19; why not release similar guidance to improve reports of data aimed at preventing those deaths?

These reports should include, at a minimum, daily (not total) counts of new cases (which hopefully will soon reflect proper population surveillance), daily hospital admissions, ICU admissions and deaths, with historical data and trend lines for all of these indicators. These should also be augmented with breakdowns by sex, age and race, combined with underlying conditions. 

In the longer term, we must dramatically increase investments in public health data infrastructure and coordination at every level of government, from counties to states to the federal level. This crisis has exposed that information infrastructure to be inexcusably inadequate, while also showing that a strong public health system is a critical piece of our national defense. We now understand the steep price of disinvestment.

We need to use all our assets to fight COVID-19. Right now, one of our most important and least costly assets is the information we already have. It’s beyond time to show us that data. 

Steven Goodman, M.D., Ph.D., is an associate dean of clinical and translational research, and professor of epidemiology and population health and of medicine, at the Stanford University School of Medicine.

Nigam Shah, MBBS, Ph.D. is an associate professor of medicine (Biomedical Informatics) and of Biomedical Data Science at the Stanford University School of Medicine.