A Comparison of Different Intensive Care Unit Definitions Derived from the German Administrative Data Set: A...

Background: The intensive care unit (ICU) is a scarce resource in all health care systems, necessitating a well-defined utilization. Therefore, benchmarks are essential; and yet, they are limited due to heterogenous definitions of what an ICU is. This study analyzed the case distribution, patient characteristics, and hospital course and outcomes of 6,204,093 patients in the German Helios Hospital Group according to 10 derived ICU definitions. We aimed to set a baseline for the development of a nationwide, uniform ICU definition. Methods: We analyzed ten different ICU definitions: seven derived from the German administrative data set of claims data according to the German Hospital Remuneration Act, three definitions were taken from the Helios Hospital Group's own bed classification. For each ICU definition, the size of the respective ICU population was analyzed. Due to similar patient characteristics for all ten definitions, we selected three indicator definitions to additionally test statistically against IQM. Results: We analyzed a total of 5,980,702 completed hospital cases, out of which 913,402 referred to an ICU criterion (14.7% of all cases). A key finding is the significant variability in ICU population size, depending on definitions. The most restrictive definition of only mechanical ventilation (DOV definition) resulted in 111,966 (1.9%) cases; mechanical ventilation plus typical intensive care procedure codes (IQM definition) resulted in 210,147 (3.5%) cases; defining each single bed individually as ICU or IMC (ICUᴧIMC definition) resulted in 411,681 (6.9%) cases; and defining any coded length of stay at ICU (LOSi definition) resulted in 721,293 (12.1%) cases. Further testing results for indicator definitions are reported. Conclusions: The size of the population, utilization rates, outcomes, and capacity assumptions clearly depend on the definition of ICU. Therefore, the underlying ICU definition should be stated when making any comparisons. From previous studies, we anticipated that 25-30% of all ICU patients should be mechanically ventilated, and therefore, we conclude that the ICUᴧIMC definition is the most plausible approximation. We suggest a mandatory application of a clearly defined ICU term for all hospitals nationwide for improved benchmarking and data analysis.