Construction of hospital management indices using principal component analysis

OBJECTIVE To construct useful indices for hospital management, based on descriptive multivariate techniques. MATERIAL AND METHODS Data were collected during 1999 and 2000, on hospital admissions occurring during 1997-1998 at Hospital General de Algeciras, part of Servicio Andaluz de Salud (SAS) of Sistema Nacional de Salud Espanol (Spanish National Health Service). The following variables routinely monitored by health authorities were analyzed: number of admissions, mortality, number of re-admissions, number of outpatient consultations, case-mix index, number of stays, and functional index. Variables were measured in a total of 22486 admissions. We applied the Principal Components Analysis (PCA) method using the R correlation matrix. RESULTS The first two components were selected, accounting cumulatively for 62.67% of the variability in the data. CONCLUSIONS The first component represents a new index representing the number of attended persons, which we have termed Case Load. The second PC represents or explains the difficulty of the attended cases, which we have termed Case Complexity. These two indices are useful to classify hospital services.