Author: Jan van der Laan, Corine Witvliet-Penning, Suzanne Gerritsen, Agnes de Bruin
HSMR 2022 Methodological Report

5. Conclusions

Compared to the HSMR model of previous year the largest change in the present HSMR model of 2019-2022 is that the COVID-19 admissions have been included (for 2022). Smaller changes are the exclusion of admissions of healthy persons (e.g. healthy newborns and other healthy boarders), and the application of the new variable for socioeconomic status (SES) to all model years. In the model of previous year the new SES variable was applied to the 2021 data only.

Sections 4.4 and 4.5 evaluate the quality of the present HSMR model. The results are largely comparable to the models of previous years. Overall the C-statistics have changed little compared to the previous model. The order of the covariates with regard to explanatory power is mostly the same, although some minor changes have occurred in the relative importance of some of the variables. Year of discharge and month of admission have become somewhat more important, which is probably caused by larger swings in mortality caused by effects of the different waves of the COVID-19 pandemic over the years. The explanatory power of SES has also increased, which is most likely due to the implementation of the new SES variable.

The effect of including the COVID-19 admissions on the total HSMR of 2022 is relatively small – for most hospitals the effect is less than 2 points. The 2022 model for the COVID-19 admissions has somewhat more difficulty predicting mortality than the COVID-19 model of 2021 (which was not included in the HSMR 2021). This is probably due to the changes in patient and admission characteristics of the COVID-19 admissions in 2022 compared to 2021. The predictive power of the COVID-19 model of 2022 is however still good and comparable to the models for other diagnoses.