Severity-Adjusted ICU Mortality Only Tells Half the Truth—The Impact of Treatment Limitation in a Nationwide Database

Objectives: The influence of different forms of treatment limitation on mortality rate in the ICU is not known despite the common use of the latter as a quality indicator. The aim of the present study was to assess the prevalence of treatment limitation and its influence on ICU mortality rate. Primary outcomes were prevalence of treatment limitation and its influence on severity-adjusted ICU mortality rate. Secondary outcomes included the association of limitation with age, sex, type of admission, diagnostic group, treatment intensity, and length of ICU stay. Design: Retrospective, observational study. Setting: All Swiss adult ICUs. Interventions: None. Patients: A total of 166,764 patients were admitted to an ICU in 2016 and 2017. Of these, 9139 were excluded because of readmission or invalid coding. Measurements and Main Results: Of 157,625 ICU patients, 20,916 (13.3%) had a fully defined treatment limitation. Among this group, treatment limitation was defined upon ICU admission in 12,854 (61%), the decision to limit treatment was based on the patient’s advance directives in 9,951 (48%), and in 15,341 (73%), there was a decision to deliberately withhold certain treatment modalities. The mortality odds ratio for the group with a treatment limitation, considering relevant cofactors, was 18.1 (95% CI 16.8–19.4). Conclusions: Every seventh patient in a Swiss ICU has some kind of treatment limitation, and this most probably affects the severity-adjusted mortality rate. Thus, mortality data as a quality indicator or benchmark in intensive care can only meaningfully be interpreted if existence, grade, cause, and time of treatment limitation are taken into account.