Measuring health care performance : variations in care process, resource use and outcomes in childbirth care

Abstract

Introduction: Measurement of health care performance is important for quality assurance and improvement of health services. With regular monitoring and follow-up, unwarranted variations in performance can be reduced. For performance measurement to have the desired effects, meaningful indicators of performance need to be selected, methods that ensure comparability between hospitals need to be employed, and information must adequately be fed back to providers and professionals. Birth care is one of the most common causes of hospitalisation and large variations in practice have been observed which highlights the relevance of performance measurement in this area. More knowledge is needed on how to use indicators and appropriate methods for measurement of performance and how this information can be used to improve clinical practice.

Aim: The overall aim of this thesis is to show how routinely collected data can be used to measure health care performance in the area of birth care and to assess how such measurement can support quality improvement.

Method: Three of the four studies are based on quantitative analyses of a research database with extensive information on patient characteristics, care process, resource use, and health outcomes for almost 140 000 women giving birth. Regression analyses were employed to investigate the importance of case mix adjustment and the variations in performance between Swedish birth clinics. The fourth study is based on interviews with managers and staff in a hospital department to understand how they perceived the use of technology for feedback of performance data in improvement efforts.

Findings: Patient characteristics have a significant effect on birth care performance indicators and adjustment for differences in patient populations is a prerequisite for meaningful performance measurement. There are large variations in case mix adjusted performance between birth clinics in Sweden in terms of both the care process and the health outcomes achieved. If all clinics performed as the top 20%, around 2200 caesarean sections would be avoided annually in the regions studied. Similarly, almost 900 perineal tears of grade 3 or 4 and 1500 post-partum infections would be avoided. There are a number of factors that facilitate or hinder the adoption of technology for timely feedback of relevant performance data. Managers and staff perceive that such feedback of data supports quality improvement.

Conclusions: Adjustment for patient characteristics is a prerequisite for meaningful comparisons of performance between hospitals and can be used to analyse unwarranted variations. Analysis of case mix adjusted variations in performance between Swedish birth clinics reveals significant potential for improvement of outcomes and reduced costs. Continuous use of performance data can support quality improvement and lead to reduced variations in performance.