The principles of my COVID-19 virus analyses have been these:
1) Numbers per population are more important than total numbers. Total numbers dazzle but the population of a state is more significant in its capacity to deal with cases. Population correlates to numbers of hospitals, contact tracing, and tests. Some less populated states receive little publicity. This week North Dakota has had a rate of new cases per million population that is over 80% higher than New York state at its peak. It frustrates me to see depictions like this from NBC news (the map toward the bottom) that provide total numbers versus shouting out where the infection rates are at crisis level.
2) Numbers of new cases, tests and deaths per population are more important than total numbers. Total historical numbers from previous weeks and months will not change. The new cases need to be addressed.
3) Weekly numbers are better than daily numbers in determining trends. Illinois, for example, updates its probable cases and deaths on Fridays and will always have a jump on that day. Connecticut, Rhode Island, and Kansas don't report on weekends. Weekly numbers overcome such daily variances.
4) Numbers 1, 2, and 3 taken together mean that the number of new cases per population per week are the optimal form of reporting the current state of the infection.
5) The states, are in a sense, in competition with one another. State government decisions, in part, determine the number of infections. By ranking the states according to their weekly new cases, tests, and death rates, a picture forms of which states are doing better and which are doing worse.
Performance Beyond Replacement
Statistical geeks invented a metric for ranking baseball players that takes into account multiple aspects of player performance and compares those to how much better (or worse) that player would be compared to an average replacement performer. Finally, they extrapolate this into the number of likely additional wins that the player delivers. This is called Wins Above Replacement or WAR.
I propose looking at performance of states compared to the numbers from a median state for any given week, performance above (or below) replacement. For the week ending October 17, the number of new cases in the median state was in Colorado, 6481 cases for 1188 per million with 25 states (or DC) performing better and 25 performing worse.
Vermont performed best among states with 69 cases for a weekly new case rate of 111 per million. Vermont bested Colorado by 1077 cases per million population.
On the other extreme, North Dakota had 6184 new cases per million for the week ending October 17. North Dakota performed 4996 cases per million worse than Colorado.
A different way of looking at this is by extrapolating the numbers of infections for a given state to what those numbers would look like if the state had the population of California (2019 US Census estimate 39,551,223). If California had the rate of infections per million that North Dakota had this past week, California would be experiencing 34,954 new cases per day. (This past week California averaged 2,963 daily new cases. That may seem like a lot but, because of their population, they ranked 6th best in new infection rate per million people.)
Legend: Performance compared to median is derived by subtracting the weekly case rate for a given state from the rate for Colorado, the median state this past week. The numbers refer to the number of cases by which a state is doing worse (adjusted for population) than the median (negative numbers, the top portion of the list) or the number of cases by which a state is doing better (positive numbers, the bottom of the list). Cases per day if the state had a population of California is obtained by taking the case rate per million and multiplying it 39.551 for California's 39,551,000 people. I adjusted this to per average day even though most of my other figures are per week.
The Week Ending October 17.
I began this project in early May. Sometimes I feel as though I'm watching a zombie movie. The zombies keep marching on relentlessly and those places you once believed were safe, eventually fall prey. Very few states can say they have never had an out-of-control infection, most notably Vermont, New Hampshire and Maine. Continuous vigilance is necessary to prevent out-of-control numbers.
Several states that maintained very low levels for a long time are currently among the worse in infection rates including Montana and Wyoming. In May the number of new cases per week in Hawaii was below ten. In August, it peaked at 1783. This past week it registered 574.
The dominant story for the last three months has been the surge in cases in Republican-leaning states. Although states that Trump won in 2016 still account for the top 14 spots, they have been joined in the top 20 by resurgences in Democratic-leaning states, notably Rhode Island, Minnesota, and Illinois, with New Mexico at 21. New Mexico had maintained a reasonable rate of new infections even while bordering states surged, especially Arizona and Texas. New Mexico can no longer make that claim.
New Jersey, New York, and Massachusetts all went from very high levels to being among the best in the country. Although they still have among the lowest new infection rates in recent weeks those rates have more than doubled. Each was below 250 new cases per million per week; each is now above 500.
Overall the total number of states doing poorly has increased with the median rising to a new height. North and South Dakota have spent seven weeks in the top two positions with their numbers growing ever higher. Their numbers are likely even higher than what is presented. North Dakota has advised positive patients to do their own contact tracing. By my count, 40.9% of North Dakota's PCR tests were positive this past week.
No comments:
Post a Comment