02.11.2011Measuring instrument for comparing mortality rates in different hospitals still has many contingencies

On Wednesday 2 November 2011, Wim van den Bosch, sr. quality and responsibility advisor at the St. Antonius Hospital, obtained his doctorate with a thesis entitled: De HSMR beproefd: aard en invloed van meetfouten bij het bepalen van het gestandaardiseerde ziekenhuissterftecijfer (The HMSR tested: nature and influence of measurement errors in determining the standardised mortality ratio). In his thesis, he concludes that the current HSMR (Hospital Standardised Mortality Ratio), the hospital mortality rate adjusted for patient characteristics, still has too many contingencies, which hampers the mutual comparison of hospitals. Van den Bosch carried out his study within the Santeon hospitals and in cooperation with NIVEL. He took his degree at the VU Medisch Centrum.

Santeon decided to use the HSMR as a measuring instrument in a joint attempt to improve the quality of healthcare. Gradually, doubts arose. Didn’t measuring errors influence the HSMR scores more than the quality of care? This was why doctoral student Wim van den Bosch examined the nature and influence of measuring errors. He compared data from the databases of the Santeon hospitals and national medical registration (LMR) data, and spoke with doctors and staff members of the Santeon hospitals to reveal the mechanisms behind the statistics.

His study shows the HSMR is not adjusted for a number of differences. It turns out that considerable differences in HSMR scores are caused by variations in hospital registrations, re-admissions, the gravity of the disorder and risky operations that are only performed in a limited number of hospitals. Like open-heart surgery. Factors the HSMR does not adjust for and that involve other issues than quality.
For example, one hospital admits the same patient twice, for diagnosis and therapy, while another hospital combines both in a single admission. What is more, a patient in one hospital doesn’t always get the same code – the HSMR works with codes – as in another hospital. Van den Bosch was also struck by the dissimilarity between heart centres and non-heart centres; this distinction is not reflected in the HSMR. And the substantial difference between acute and non-acute patients is not registered unambiguously either. Factors like these affect the HSMR, but involve other issues than the healthcare quality a hospital provides.

Van den Bosch’s conclusion: the current HSMR does not adjust adequately; some corrections are required before the HSMR instrument can be effectively implemented to compare hospitals and improve the quality of healthcare. So, he thinks that the standardised mortality rates that the NVZ will publish for the first time in late November still contain a great deal of contingency.