Administrative Data Should Be Questioned

Hospital discharge data (=administrative data) have been used for some time in order to measure the quality of hospital treatments. In this context, Mansky in a recently published review article pointed out the good validity of administrative data. By contrast, a systematic qualitative review found that the sensitivity of quality indicators—drawn from administrative data—is usually poor (1). This is also the case for risk scores, which take into account a number of comorbidities.

On this background, Nimptsch and Mansky (2) write that national hospital discharge data in Germany provide an opportunity—because of their completeness— for analyzing even rare events, such as deaths after cholecystectomies and herniotomies. The authors conclude that death rates after such procedures could be reduced.

The clinical epidemiologist Richard Lilford was one of the first to point out how poorly suited deaths are as a measure for assessing the quality of hospitals. Hogan et al. concluded (3) that “any metric based on mortality is unlikely to reflect the quality of a hospital.” This prompted The BMJ to publish an editorial alongside the research, entitled: “The death of death rates?” In our opinion, the list of suggestions made by Nimptsch and Mansky is lacking one crucial item: adequate staffing levels (4).