An update on the state of covid-19 in the 50 US states

How are we doing with covid-19? Is the situation improving or worsening? How is it for different parts of the USA? How do we assess the situation?

The best measures are the number of covid-19 confirmed cases, covid-19 induced deaths, and testing statistics. Each measure has weaknesses. Many people think that confirmed cases vastly underestimate the true spread of the disease - because of limits in testing. However, deaths could be also both overcounted and undercounted.

Some analysts largely ignore the number of confirmed cases and focus solely on the number of deaths. Many such analysts estimate true spread of infection assuming a fixed death rate, which is not easy to determine. Rather small differences in the death rate could make a huge difference in the estimated spread of disease. Examples here and here.

While different methods have their merits, policy decisions should be based on all available information - regardless of the shortcomings of each measure. In addition to the number of testing, positive % rate is also useful - very high positive rate likely means a serious shortage in testing. Available hospitalization data also could be useful.

Whether the covid-19 virus situation is getting better or worse could be discussed using a simple measure of virus reproduction number. The basic reproduction number R(0) is the average number of infections from one sick person where everyone is susceptible.

R(0) = number of 'adequate' contacts per day * 1 / number of days when infectious

Vaccination efforts could reduce the spread of disease by reducing the number of susceptible people. Distancing and other mitigation measures could reduce the number of 'adequate' contacts. Accounting for these factors, we have effective reproduction number R(e) or reproduction number at a given time R(t) - or R. R is not a constant. Population density matters. More discussion here and here.

When R = 1, the number of new infected cases is expected to remain same indefinitely. When R > 1, we expect a growth in the number of new cases every day - with growth being exponential if R is sufficiently different from 1. When R < 1, the number of new cases is expected to go down daily - R value substantially smaller than 1 means the number of new cases goes down quickly while R of barely under 1 means that the number of new cases goes down slowly.

More than simply determining whether the spread of disease is getting better or worse, policy decisions should consider long-term goals. Should we try to reduce the number of deaths further? Is simply keeping the number of deaths same acceptable? What are acceptable costs in economic terms? Which policies should be kept? What more could we do?

With these in mind, let's look at the current situation in the 50 US states.

After partial reopening on May 1, the number of deaths in Alabama climbed for about 2 weeks before showing possible sign of plateau. There is no indication that R is below 1. Even if R is 1, are they okay with 15 deaths every day? IMHO, they should bring back large gathering ban at minimum. 

 Alaska is fine. I hope they are doing contact tracing, but I would say the same for ALL states.

With Arkansas, recent increases in daily cases is a cause for concern. Ditto recent increase in test positive rate. Hospitalization numbers are okay except for the possible recent blip. 

Even before the start of partial reopening, most indicators suggested that R was above 1 in Arizona. I don't see how the situation could improve with reopening. 

California is in a limbo with R very close to 1 with stay-at-home order still in effect. This is a very difficult situation, although they are doing better than average on per capita basis. 

A recent spike in deaths notwithstanding, Colorado seems to have a R value below 1 since the April 27 expiration of stay-at-home order. Test positive rate is on the high side. 

With stay-at-home order still in effect, Connecticut shows a R < 1 for now. It's a slow descent, however. The scale of contact tracing needed seems humongous. 

Plateau in disease spread is more obvious in DC now. 

The situation in Delaware looks more like Maryland and Virginia to south rather than New Jersey, New York, and Connecticutt to north. Plateau in disease spread looks real with R perhaps close to 1.

Compared to other states, there was a fairly long gap between peak death and peak new cases. Steady decline in new cases appears to have stopped completely with the end of stay-at-home order, with extension for a few populous counties such as Miami-Dade. 

After reopening, Georgia has showed a continued slow decline in disease spread, meaning R is still below 1. Needed scale of contact tracing is huge. Is daily deaths of 25 acceptable indefinitely as a cost of doing business?

Like the other minimally affected state of Alaska, Hawaii has mandatory quarantine for all visitors. 

Disease spread may have reached a plateau somehow in Iowa. 

The current R appears to be close to 1 in Idaho. While contact tracing should be manageable, complete lack of large gathering ban is a potential source of trouble. 

Illinois is one of the nightmare scenarios. Signs of plateau are barely visible by all indicators. The distancing measure looked very strong on paper, but it took a very long time to get to a plateau. The situation is so bleak here, and I am only glad that I'm not the one in charge here. 

Reopening has substantially slowed the decline in the rate of disease spread in Indiana. Some parts of the state is still under stay-at-home order.

Test positive rate is still on the high side in Kansas. While the number of new cases has been declining, this may be largely caused by reduced testing. My best guess is that R is close to 1. 

Kentucky never declared stay-at-home order - instead they had something called 'healthy at home', which is likely weaker. Disease spread according to death has been pretty flat for about a month.

Before stay-at-home order was lifted a few days ago, the disease spread was slowing in Louisiana with deaths and hospitalization at about half the peak. 

Massachusetts had numbers that looked somewhat different when I used data from Johns Hopkins (below) instead of the covid tracking project (above). With Hopkins data, decline in disease spread look more reliable.

Nothing better than a plateau in disease spread was evident in Maryland prior to reopening of most of the state - exceptions are Baltimore, Montgomery, and Prince Georges counties. 

The numbers for Maine are small and noisy. Contact tracing could be manageable with more testing.

Outside of the tri-state of New York, New Jersey, and Connecticutt, Michigan is the only state with sustained reduction of disease spread - about one third of the peak. 

A possible plateau finally appearing in Minnesota. Minnesota never had a large gathering ban, which is one of the most effective way of making R smaller. 

Mixed signals for Missouri. My best guess is that R is under 1 without stay-at-home order. Despite a somewhat high death rate, Missouri might be in a slightly better shape than some of smaller neighbors such as Kansas and Iowa.

Even with early end of stay-at-home order, Mississippi has managed to keep R close to 1. 

There is a very small rebound in new cases in Montana. 

A plateau has been reached with stay-at-home order in North Carolina. Not great.

No large gathering ban in North Carolina. The disease spread appears to be increasing slowly. 

The number of new cases in early May was quite big for a sparsely populated Nebraska. The death numbers suggest that such level of new infections may have been present in early April - although low test positive numbers does not support this notion. I don't know what's going on here. 

Barely a sign of recent plateau in New Hampshire. 

One of few states with real signs of decline in disease spread. Recent boost in testing numbers is welcome.

Nary a sign of plateau in New Mexico. Another nightmare scenario alongside Illinois despite strong distancing measures on paper. 

No early signs of disease spread becoming worse after the end of stay-at-home order in Nevada. Death rate has been treading water. 

Another state with sustained reduction of disease spread. A few upstate counties were reopened in New York on May 15. 

Ohio appears to have an R close to 1 after downgrading stay-at-home order. Is 40 deaths every day indefinitely good enough?

Oklahoma reached a plateau without a stay-at-home order. The number of new cases is still big enough to pose a big challenge for contact tracing. 

Oregon looks manageable. Stay-at-home order remains in effect for Portland.

Pennsylvania has been showing some signs of sustained reduction of disease spread - although not to the degree of New York, New Jersey, Connecticutt, and Michigan.

Disease spread has continued to slow after the end of stay-at-home order in tiny Rhode Island.

I am going to guess that R is close to 1 in South Carolina. 

I am going to guess that sporadic large outbreaks of disease are responsible for the sudden peaks in daily cases in South Dakota.

Deaths and new cases are both trending up slightly in Tennesse, indicative of R > 1 possibly.

Texas appears to have R > 1 with the end of stay-at-home order. 

Utah is treading water at best. 

Like the neighboring Maryland, populous counties next to DC in Virginia remain under stay-at-home order. Plateau seemed real enough prior to partial reopening.

Disease spread had come down to near zero prior to reopening in Vermont. 

After the early peak and subsequent reduction in disease spread, Washington has been treading water. The data from Johns Hopkins looks nicer (see below).

Stay-at-home order in Wisconsin was ended by the state supreme court. R was close to 1 at the time.

West Virginia was less affected by covid-19 compared to other states east of Mississippi river. 

Wyoming appeared to have had a cluster of cases near the start of May but no sustained spread thereafter. 

Summarizing good news, six heavily affected states (New York, New Jersey, Connecticut, Michigan, Pennsylvania, and Lousiana) have shown a sustained reduction in disease spread for a while. Three states (Alaska, Hawaii, and Montana) were minimally affected and are in good shape. Three more states (Oregon, Vermont, and Wyoming) are also in pretty good shape. 

Summarizing bad news, Illinois and New Mexico are really struggling despite strong measures on paper. Iowa and Minnesota are also worrisome. Texas should be watched carefully. Also, many other states are merely treading water with disease spread. With further reopenings, the risk of disease spread becoming worse could only worse. 

Looking at all 50 states at a glance - this time ranked by the number of deaths per million.

The data used in this post were taken from the COVID tracking project as well as Johns Hopkins University for an alternative source of total cases and deaths. I used the COVID tracking project mainly because of faster updates and additional information on hospitalization and testing. Additional information was obtained from the US census bureau for population and land area, and Kaiser Family Foundation for the summary of various state covid-19 policies.