States and municipalities recently began the phased reopening of their economies, and several news outlets subsequently reported a concerning divergence in daily new COVID-19 cases between the New York City metropolitan area and the rest of the country.

Indeed, excluding New York and New Jersey—two of the states that bore the brunt of cases in the beginning of the pandemic—cases are higher than ever, and rising still (Figure 1a). The pattern is similar outside of the New York/Newark metropolitan area (or the Census Bureau’s “New York-Newark, NY-NJ-CT-PA combined statistical area,”), as shown in Figure 1b.

Figure 1a. New Daily Positive Tests in New York and New Jersey vs. Rest of U.S., 7-Day Rolling Average

_{Source: The COVID Tracking Project at The Atlantic}

Figure 1b. New Daily Cases in New York City Metro Area vs. Rest of U.S., 7-Day Rolling Average

_{Source: COVID-19 Data Repository by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University}

However, important caveats apply to conclusions based on confirmed cases alone. For example, testing continues to increase, with daily test totals averaging nearly 500,000 nationally (Figure 2), and additional testing would lead to more official cases, all else equal, even if infection rates are flat. In fact, testing more people could even produce flat or rising cases when the true infection rate is falling.

Figure 2. Total Tests in New York and New Jersey vs. Rest of U.S., 7-Day Rolling Average

_{Source: The COVID Tracking Project at The Atlantic}

If more testing resulted in a higher share of tests yielding positive results, that would be evidence that the infection rate is rising, and suggest that more people are developing COVID-19-derived symptoms, leading them to get tested, and producing more cases among those who get tested.

Conversely, a flat or declining share testing positive could suggest that elevated cases simply reflect more testing. In that case, expanded testing would seem more like a reflection of greater concern about symptoms among both people who do and do not have COVID-19.

Outside New York and New Jersey, the share of tests coming back positive has been relatively stable over the past month, falling slightly (Figure 3), and this casts some doubt on whether the true infection rate outside the New York area remains at or above peak levels. (Note that in the past week, however, the share has ticked upward.)

Figure 3. Positive Tests as a Share of Total Tests, New York and New Jersey vs. Rest of U.S., 7-Day Rolling Average

_{Source: The COVID Tracking Project at The Atlantic}

COVID-19 deaths have been declining outside New York and New Jersey (and outside the New York metro area), which is another sign that infection rates may be falling (Figure 4a).

Figure 4a. Deaths in New York and New Jersey vs. Rest of U.S., 7-Day Rolling Average

_{Source: The COVID Tracking Project at The Atlantic}

Figure 4b. Deaths in New York City CSA vs. Rest of U.S., 7-Day Rolling Average

_{Source: COVID-19 Data Repository by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University}

It’s possible deaths from COVID-19 may be significantly underestimated, but looking at “excess deaths” yields the same conclusion. Excess deaths compare total deaths in a given week to the number of deaths during the same week in previous years, and the measure reflects deaths officially determined to be from COVID-19 but also COVID-19 deaths that are not recognized as such. (Note that the excess also reflects the net of increased deaths from causes other than COVID-19 that occurred because of the pandemic—such as heart attack victims who delay going to the hospital for fear of the coronavirus—and reduced deaths from modified behavior due to COVID-19—such as fewer traffic deaths.)  Figure 5 shows declines in excess mortality, which also suggest that infection rates are on the decline. However, because delayed reporting results in upward revisions of deaths, Figure 5 excludes the last three weeks of estimates.

Figure 5. Percent of Expected Deaths in Reported Week Compared to Deaths Across Same Week 2017-2019

_{Source: CDC, https://www.cdc.gov/nchs/nvss/vsrr/covid19/index.htm}

But there is one more crucial caveat before we conclude that infection rates are falling outside the New York area: there is often a significant lag between coronavirus infection and the onset of COVID-19 symptoms that would lead someone to be tested, and deaths also lag initial symptoms. The CDC reports that symptoms can show generally between 2 and 14 days after exposure, with WHO estimates placing the average incubation period at five days. Other studies indicate that some people experience longer periods of incubation, with one estimating durations as long as 24 days. Deaths typically occur between 2 and 3 weeks after initial COVID-19 symptoms occur.

That is to say, if the true infection rate outside the New York area were rising, we would not necessarily see the increase reflected in case trends for a week or two, and would not see it reflected in death trends for perhaps another two weeks after that. And because the death data are generally incompletely reported for three weeks, it would be even longer before the full nature of any increase would be apparent. If COVID-19 death trends are the most unambiguous indicator of changes in infection rates, then it is unclear that the current data would register any change even if infection rates have been rising since the end of May. Case trends would register an increase sooner, but any increase would be confounded by ongoing increases in testing.

In the meantime, there is plenty of evidence that individual states are seeing increases in infections—in some cases resulting in new highs. Our Reopen Readiness Metric Tracker now includes data on daily and cumulative deaths from COVID-19, as well as the latest available data on hospital capacity at the state level. The JEC Chairman’s office will continue to track these and other indicators to inform policymakers and the broader public.