Rapportcijfers voor de hartchirurg; validiteit van waargenomen verschillen in sterfte

To determine the possibility of comparing the mortality rates of patients operated by different heart surgeons with each other.

Design

Retrospective cohort study.

Setting

Academic Medical Centre, Amsterdam, the Netherlands.

Patients and methods

Clinical information, operation data and follow-up data on 783 patients who had undergone cardiac valve replacement, were collected from the clinical records. Aortic valve replacement had been performed in 446 patients (1979-1986) and mitral valve replacement in 337 patients (1980-1990).

Results

The one-year mortality rate was higher among patients operated on by heart surgeon A than among patients operated on by the other heart surgeons from the same team, viz. 16.4 and 9.5, respectively, an absolute difference of 6.9. The 95 confidence interval of the difference was 1.7-12.9. However, it was also found that the risk profiles of these patients of surgeon A differed from those of the other patients. After multivariate correction for this difference in risk profile, the difference in mortality was no longer statistically significant.

Conclusion

The differences in mortality observed in our study could not be attributed to difference in quality of the heart surgeons, but were related with the risk profiles of the patients operated by one of them. Thorough analysis with correction for risks is necessary for the assessment of the quality of care, if the conclusions are not to be misleading.