STANDARDIZED MORTALITY FROM CERVICAL-CANCER - A MEASURE OF PERFORMANCE

A retrospective cohort analysis, using data extracted from clinical no tes, examined the validity of standardized mortality ratios (SMRs) for cervical cancer as an indicator of the quality of health services by exploring the mortality at five years of patients presenting with cerv ical cancer, identifying factors associated with survival, and studyin g the relationship of those factors to SMRs for the disease. The subje cts were 1038 women aged 15-64 registered at the Yorkshire Cancer Regi stry as having invasive cervical cancer between 1979 and 1983. It was found that independent risk factors for mortality within five years of presentation were stage at diagnosis and smoking habit. Method of tre atment had an effect that failed to reach statistical significance. In dependent risk factors for late stage at presentation were absence of a history of cervical smear and increasing age. Considerable variation remained unexplained in both models. District SMRs were not related t o five-year survival, stage at presentation or screening history. SMRs were related to proportions of smokers. It is concluded that SMRs are subject to considerable influence from a range of unidentified factor s, as well as identified risk factors. They are too broad an indicator to be a valid measure of health service performance. The health servi ce's ability to reduce case fatality does not outweigh the effects of incidence and severity. Better indicators would be survival to five ye ars of those presenting at stage II or less and distribution of stage at presentation.