OBJECTIVE: To develop measures of hospital performance over time with particular reference to maternal and neonatal care by controlling for case mix. DESIGN: Analysis of computerised records of births. SETTING: Scotland, 1980-7. SUBJECTS: Over half a million singleton live births and stillbirths. MAIN OUTCOME MEASURES: Numbers of perinatal deaths and caesarean sections. RESULTS: Scottish maternity hospitals perform more or less equally with regard to perinatal mortality. When caesarean sections are considered, there is evidence that hospitals differ in their treatment of different groups of women; in two examples one hospital had an increased rate among women of parity 2 or more and another had a reduced rate of repeat caesarean section. CONCLUSIONS: Developing measures of performance over time by controlling for case mix is a valid system for monitoring hospital outcomes and activity, and allows comparison either between hospitals or with data for all Scottish maternity hospitals. Hospital profiles permit identification of differences for particular patient groups after allowance is made for other case mix variables.
OBJECTIVE: To develop measures of hospital performance over time with particular reference to maternal and neonatal care by controlling for case mix. DESIGN: Analysis of computerised records of births. SETTING: Scotland, 1980-7. SUBJECTS: Over half a million singleton live births and stillbirths. MAIN OUTCOME MEASURES: Numbers of perinatal deaths and caesarean sections. RESULTS: Scottish maternity hospitals perform more or less equally with regard to perinatal mortality. When caesarean sections are considered, there is evidence that hospitals differ in their treatment of different groups of women; in two examples one hospital had an increased rate among women of parity 2 or more and another had a reduced rate of repeat caesarean section. CONCLUSIONS: Developing measures of performance over time by controlling for case mix is a valid system for monitoring hospital outcomes and activity, and allows comparison either between hospitals or with data for all Scottish maternity hospitals. Hospital profiles permit identification of differences for particular patient groups after allowance is made for other case mix variables.