160 Chapter 7 Now taking probability limits and using Equation 21, Equation 22, as well as we obtain This is the desired consistency result. We calculated effect sizes with protocolised S/F94 as outcome measure, with estimated from the ISARIC4C data. We took our estimate of to be equal to our estimate of and were estimated using data from the RECOVERY trial. WHO ordinal scale We used a proportional odds model to estimate the relationship between day 28 WHO ordinal scale category and individual characteristics, Here, pij is the probability that the WHO ordinal scale category of individual i at day 28 is are a set of constant terms for each WHO ordinal scale category, β is an n-dimensional vector of coefficients, and is an n-dimensional vector of covariates. j = 10 is the highest WHO ordinal scale category and corresponds to death. Then qi10 := 1- pi10 is the probability of day 28 mortality for patient i. We consider an individualspecific treatment effect given by
RkJQdWJsaXNoZXIy MTk4NDMw