Quality of care and its impact on population health: A cross-sectional study from Macedonia

Abstract

Research has revealed significant variation in both the quality of clinical care and the health status of populations. We conducted a study to determine if variations in the quality of clinical care can be quantitatively linked to variations in health status, at the patient and the population level. This study, conducted at health facilities in four municipalities in Macedonia, collected cross-sectional data on (1) structural measures (such as infrastructure, facilities, equipment and costs) and the quality of clinical care provided by physicians (as measured by clinical vignettes); (2) detailed health and socioeconomic status information on patients using the facilities; and (3) nearly the same information on a random sample of adults in each municipality.

Data were collected from a total of 57 facilities, 273 physicians, 1451 patients, and 1627 adults from the general population. The main outcome measure was health status, based on self-reported health surveys. Objective health measures were obtained to control for preexisting conditions. The main explanatory variable was quality of clinical care, based on physicians’ clinical vignette scores. Structural measures were included in our model but had a more distal relationship to health status.

We found that quality of care strongly predicted self-reported health status of patients using the facilities even after controlling for other factors ($p<.05$). Quality of care was also associated with higher health status for the population living in the surrounding community, regardless of utilization ($p<.05$).

This linkage between quality of clinical care and health suggests that policies that improve clinical practice have the potential to improve population health more rapidly than other interventions.

Introduction

Nearly 30 years ago, population-level surveys revealed substantial variation in health status among populations ranging from villages to countries (Glass et al., 1982; Lightbourne, Singh, & Green, 1982). Subsequent research showed that this variation remains even after correcting for socioeconomic determinants (Murray & Frenk, 2000; Schieber & Maeda, 1997).

A parallel stream of research began to document substantial variation in the quality of clinical care (Wennberg & Gittelsohn, 1973). Researchers found that variation in the process of care—what physicians actually do when they see patients—is more closely linked to health status than are structural elements of care (such as infrastructure, facilities, equipment, and costs) (Donabedian, 1980). Moreover, this variation in process exists not only among individual physicians, but also across geographic areas and systems of care, and appears to be independent of the cost of care (Skinner, Fisher, & Wennberg, 2001).

A confluence of these two research streams led a growing number of researchers to postulate that variations in population health status may be closely related to variations in the quality of clinical care the population receives (Donabedian, 1980; Peabody et al., 1999). However, this postulated linkage has been difficult to quantify.

We hypothesized that the better quality of clinical care a population receives, the better its health status. Further, if quality affected the health status of those receiving care, we hypothesized it would have a similar—albeit attenuated—effect on the health of the population of the surrounding community. To test these hypotheses we used a unique, validated tool, clinical vignettes, to measure the quality of care (Peabody, Tozija, Munoz, Nordyke, & Luck, 2004). Simultaneously, we measured the health status of those receiving care and of the general population. We did this in a setting where we could observe the link between quality measures and health status (Peabody & Gertler, 1997).