A novel cohort analysis approach to determining the case fatality rate of COVID-19 and other infectious diseases

Charit S Narayanan

If the groups are homogenous and identical to each other in every way (have the 78 same mix), the cohorts would have the same CFR. This is because each cohort CFR 79 would change at the same rate, assuming they are the same age, have the similar 80 comorbidities (if any), etc. However, in reality this will likely not be the case since 81 groups will naturally have people with differing characteristics and their mix will affect 82 the CFR. In an idealistic sense, we would be able to have homogenous cohorts, filtered 83 by certain conditions (age group, smoking habits, etc.) and looking at the cohorts over 84 time, we would be able to establish a pattern, and obtain a reliable CFR [ [9] , [10] ]. 85 We would need to construct these triangle charts for groups that are as homogeneous 86 as possible to ensure that certain members do not skew the death rates. Age, high-risk 87 medical groups, and countries are the right granular segments to look at. Using these, 88 we can most accurately gauge the case fatality rate for each group. The case fatality rate as a function of time since infection can be modeled in multiple 91 different ways [11] . We use a logistical function to describe the exponential growth, and 92 subsequent flattening, of COVID-19 case fatality rate; it is the simplest form that 93 satisfies the S-shaped growth [12] . This can be expressed as: . CC-BY-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.