States implemented various policies during the spring to stem the spread of the SARS-CoV-2 virus and COVID-19. Cases have risen sharply in recent weeks in some reopening states, demonstrating that the pandemic is not over. So where exactly do we stand?
A worst-case pandemic or herd immunity would result in or require between 60 and 80 percent of Americans getting COVID-19, albeit not necessarily with symptoms. This projects to 200 to 270 million cases. Through July 1, the COVID Tracker website reports 2.675 million cases nationwide, suggesting that we have had between 1 and 1.4 percent of the pandemic cases. This is depressing given the pain we’ve suffered to date.
Testing has lagged so millions of Americans have had COVID-19 without testing positive for it. We can instead use deaths, which are more accurately tallied, to infer the extent of spread. There are challenges with identifying true COVID-19 deaths. Some people who died of pneumonia in March or April were never tested for COVID-19, and there is debate about whether the illness could cause deaths through heart or kidney failure. I will leave these debates to the medical experts. Still, reported COVID-19 deaths are probably accurate to within +/-20%, versus a likely order of magnitude error with official cases.
Deaths to date relative to projected deaths in an unmitigated pandemic provide one measure of where we are at in the pandemic. In March the CDC projected 1.7 million deaths and Imperial College, London 2.2 million deaths in an unmitigated pandemic. These estimates were based on a higher infection fatality rate (IFR) than the CDC’s most recent estimate (0.8% and 0.9% respectively). The CDC recently estimated a 0.4% fatality rate among symptomatic cases with 35% of all infections being asymptomatic, for an IFR of about 0.27%. Adjusting the fatality estimates for this new IFR yields projections of 570,000 from the CDC and 660,000 from Imperial College. Through July 1, COVID Tracker reported 120,853 deaths, meaning we are at 18 to 21 percent of the unmitigated pandemic fatality total.
We can also project cases from deaths to date. A 0.4% case fatality rate implies that there must have been 250 cases per death, whether documented or not. The July 1 death toll implies 30.2 million cases. Deaths, however, are a lagging indicator of infection; COVID-19 deaths occur typically 14 to 28 days after infection, with an average of around 21 days. There must have been 30.2 million cases in the U.S. on or around June 10 to produce the July 1 death toll. Reported cases are also measure with a lag. People do not normally get tested until showing symptoms, which can take 5 days after infection. I’ll assume a seven day lag for cases. Infections on June 10 the appear as reported cases on June 17. Consequently, June 17’s total of 2.155 million cases reflects the inferred 30.2 million cases responsible for the July 1 deaths. Since June 17, COVID Tracker reports that there have been about 5 million additional cases, or a 24% increase. Applying this to the imputed 30.2 million cases gives us 37.5 million cases currently. These would be sympotmatic cases. If 35% of all infections are asymptomatic, as the CDC currently estimates, 20.2 million asymptomatic cases must be added for a total of 59 million infections through July 1. This puts us at 22 to 28 percent of the 200 to 270 million pandemic cases.
The accompanying table reports these COVID-19 case and infection rates by state, as a percentage of state population. Column A uses reported cases. Column B is cases inferred from deaths on July 1, followed by total inferred infections (the asymptomatic case adjustment) in column C, and column D adjusts infections for the percentage increase in reported state cases between June 17 and July 1. States are listed from highest to lowest current infection rate (column D). New Jersey has the highest estimated infection rate due to the addition of nearly 2,000 earlier deaths to its total on June 25. New Jersey’s imputed infection rate is 67% of its population. New York is at 50%, with Massachusetts and Connecticut between 45 and 50%. The range of estimates is wide, with Hawaii still below 1% of its population even with all these adjustments. Comparing columns C and D shows adjustment for recent cases, which have been greatest in Arizona and Florida.
One important caveat is in order. This imputation is based on deaths, and COVID-19 has hit nursing homes and long-term care facilities very hard. Several states unfortunately forced nursing homes to take in COVID-19 elderly not needing hospitalization. The death toll in these states could be considerably out of line with the population-wide infection rate. In other words, applying the CDC IFR to infer cases assumes that deaths have resulted from a similar exposure of both the vulnerable and healthy population. If the vulnerable population of a state has all been exposed, a state could reach its unmitigated pandemic death projection without the virus having spread through the healthy population. The percentages in column D might represent something closer to percent of a state’s unmitigated fatality count occurred to date. New Jersey may not be at 67% infections across the population. But if the infection rate is significantly lower than this among its healthy population, relatively fewer deaths should occur if or when COVID-19 spreads through the uninfected healthy population.