Multivariate Relationships Between Health Outcomes and Health System Performance Indicators: An Integrated Factor Analysis With Canonical Correlations

This study aimed to investigate the relationships between sets of variables related to health system performance indicators and health outcomes. The relationships between a set of health outcomes and a set of health system performance indicators of a developing country were examined using multivariate statistical analysis techniques. A combinative strategy of explanatory factor analysis and the canonical correlation coefficient was used to define linear structural relationships between study variables. Province-based data were gathered from2 official statistical records of the Turkish Statistical Institute for the year 2019. Life expectancy at birth, infant mortality rate, and crude death rate were accepted as health outcome indicators. The explanatory factor analysis indicated 2 independent variable groups, namely (1) health-related human resources and capacity and (2) health service utilization characteristics. The results of the canonical correlation analysis illustrated good performance to define sparse linear combinations of the 2 groups of variables. There existed strong positive correlations between health outcomes and health-related human resources and capacity indicators (rc = 0.83; P < .001) and health service utilization indicators (rc = 0.59; P < .001). The results of this study support the view that there is a linear and strong positive relationship between health outcomes and health-related human resources and capacity indicators. Further studies will combine big data analytics with multivariate statistical analysis techniques by studying large health system performance data sets. •In the existing literature, exploratory factor analysis, one of the multivariate techniques included in this article, is generally used in scale development and adaptation studies. In contrast, canonical correlation analysis is mostly used for primary data on health services.•Integrating exploratory factor analysis and canonical correlation analysis, which are separately used multivariate techniques, for previously published health services data will fill the gap in the literature.•The evaluation of secondary data obtained especially for health services performance and outcome indicators with multivariate statistical techniques will provide a road map for future policies.