Trends in General Internal Medicine Patient and Care Complexity from 2017 to 2022 Across 21 Hospitals

There is ongoing uncertainty about whether hospital inpatients have become more complex over time, and how this might contribute to strained hospital resources. We examined measures of patient and care complexity for General Internal Medicine (GIM) inpatients over a 7-year period.

This retrospective cohort study used individual-level administrative and clinical data for GIM hospitalizations at 21 Ontario hospitals between April 2015 and June 2022. Complexity measures included patient characteristics (e.g., advanced age), patient outcomes (e.g., in-hospital mortality), and resource use (e.g., imaging). Hospitalizations were categorized into complex and non-complex categories and generalized linear mixed-effects models were used to estimate adjusted rate ratios, representing the relative change in the number of monthly admissions in each year for each category compared to 2015-2016.

The cohort included 687,512 hospitalizations (median [IQR] age, 73 [57-84] years; 50% female). The volume of admissions grew by 17%, from 103,851 in 2015-2016 to 121,103 in 2021-2022. There was a consistent increase in the use of radiology tests, particularly driven by greater use of CT scans (49% of hospitalizations in 2015-2016, 61% in 2021-2022, adjusted rate ratio 1.45 [CI, 1.40-1.50]) and MRI scans (11% in 2015-2016, 14% in 2021-2022, adjusted rate ratio 1.46 [CI, 1.39-1.54]). Other patient characteristics, processes of care, and outcomes did not consistently change across the study period. The COVID-19 pandemic period (April 2020-June 2022) was associated with increased complexity with respect to hospital length of stay, in-hospital mortality, laboratory testing, and costs of care.

GIM admission volume increased over the study period, but patient complexity as measured by the metrics considered here did not show consistent increases. The rising intensity of imaging use without evident changes in patient characteristics or short-term outcomes warrants further investigation.