Validation of a short form of the practice environment scale of the nursing work index: The PES‐5

Abstract

AbstractThe Practice Environment Scale of the Nursing Work Index (PES‐NWI) has been utilized for two decades globally to measure nurse work environments. Its 31 items in five domains present a substantial respondent burden, threatening survey response rates. The purpose of this study was to derive and validate a short form: the PES‐5. We conducted a cross‐sectional, secondary analysis of survey data from nurses in 760 hospitals in six U.S. states in 2016 or 2019. One representative item per subscale was selected by highest item‐to‐subscale R2 from the original PES‐NWI publication. Five psychometric properties of the PES‐5 were evaluated. The reproduced structure of the full form was confirmed in the 2016 data by the highest R2 for the selected items. The unidimensional structure of the PES‐5 was confirmed through confirmatory factor analysis. The correlation between the composite values of the 28‐item and 5‐item versions was 0.94. The Cronbach's alpha reliability of the PES‐5 was >0.80. The intraclass correlation coefficient (ICC 1, k), which evaluates the stability of aggregated values when data are clustered, i.e., nurses are nested within hospitals, was >0.80 in both datasets, demonstrating satisfactory aggregate properties. Construct validity was supported by the selected items being ranked highly in their respective subscales by an expert panel. Criterion validity was supported by an analysis of variance of the PES‐5 mean value across responses to a single‐item work environment measure. Similar patterns of relationships with other key variables were identified by statistically significant odds ratios in regression models predicting patient mortality from the PES‐5. The classification accuracy of the PES‐5 was high, with 88% of hospitals classified identically by both versions. The PES‐5 shows promise for measurement of nurses’ work environments while maximizing response rate by reducing participant burden.