# Covid Continuity Prevails

The last two posts looked at case counts and percent test-positives, evaluating their statistical strength as proxies for infection rate and as correlates/predictors of covid-related deaths. In running those analyses I was surprised that the lagged correlations didn’t differ much from the contemporaneous correlations. Infection precedes death by two or three weeks, yet case counts correlated strongly with body counts even within the same three-week observation window. Maybe in the epidemic continuity is a stronger force than change.

I ran a set of correlations looking at the same variable measured across consecutive intervals in time. Here are the results:

• Case count, 5/26 to 6/15 compared with 6/15 to 7/4:  r = +.50
• Case count, 6/15 to 7/4 compared with 7/4 to 8/5:  r = +.90
• Percent test-positive, 5/26 to 6/15 compared with 6/15 to 7/4:  r = +.82
• Percent test-positive, 6/15 to 7/4 compared with 7/4 to 8/5:  r = +.90
• Deaths, 5/26 to 6/15 compared with 6/15 to 7/4:  r = +.84
• Deaths, 6/15 to 7/4 compared with 7/4 to 8/5:  r = +.39

The past is a good predictor of the future. States that were high in case counts, test-positive percentages, and deaths during one interval were still high in the next interval as well. The weakest within-variable correlation is for death counts during the second and third intervals. Per this earlier post, case counts from the second interval were a stronger predictor of deaths during the third interval (r = +.90).

Covid continuity makes sense. An epidemic is self-perpetuating; contagion breeds more contagion. Once the ball gets rolling it’s hard to slow it down.