Dynamical characteristics of Covid-19 epidemic: Estimation from Colombia’s case

Diaz, Hernando; España, Guido; Castañeda, Nelson; Rodriguez, Laura; de la Hoz-Restrepo, Fernando

Objectives To characterize the dynamics of the epidemic, for modeling purposes. Methods Data from Colombian official case information were collated for a period of five months. Dynamical parameters of the disease spread were then estimated from the data. Probability distribution models were identified that represent the time from symptom onset to hospitalization, to ICU admission and to death. Kaplan Meier estimates were also computed for the probability of eventually requiring hospitalization, needing ICU attention and of dying from the disease (the CFR). Results Probability distributions of the times and probabilities were computed for the population and for groups based on age and sex. Our results show that for the times that characterize the course of the disease for a given patient (to hospitalization, to ICU or death), the variation from an age group to another is very small (around of 10% of fixed effect intercept) and the effect of sex is even weaker (around 1%). The course of the disease appears to be very similar for all patients. The probability that a patient advances from one stage of the disease to another (to hospitalization, to ICU, to death), on the other hand, is heavily influenced by sex and age. Relative risk of death for males (based on 22924 dead) is 1.7. Conclusions The times from one stage of the disease to another are almost independent of the major patient’s variables (sex, age). This is in stark contrast to the probabilities of going from one stage to another which show a strong dependence on age and sex. Data also show that the length of hospital and ICU stays are almost independent of sex and age. The only factor that affects this length was the eventual outcome (survival or death) of the disease. The time is significantly longer for surviving patients.