• Edwin D Boudreaux, Sarah D'Autremont, Karen Wood, and Glenn N Jones.
• Cooper Hospital and UMDNJ-Robert Wood Johnson Medical School, Camden, NJ 08103, USA. boudreaux-edwin@cooperhealth.edu
• Acad Emerg Med. 2004 Jan 1;11(1):51-8.

ObjectivesThe contradictory findings reported in the emergency department (ED) patient satisfaction literature may be due to methodologic differences between studies, as well as actual differences in predictors. The authors examined the stability of predictors of ED patient satisfaction across multiple assessments over 17 months.MethodsAll patients who presented for emergency care to the authors' hospital during four designated time periods spanning 17 months were eligible. The participants were contacted by telephone and the following were assessed: demographics, visit characteristics, perceived waiting times, subjective quality of care indicators, and overall satisfaction. The authors computed logistic regressions to predict overall satisfaction for each of the four periods. They compared the results across the assessments, both visually and using an aggregated logistic regression, to determine the consistency of the final equations. Interpretations based on traditional p-value cut-offs and odds ratios (ORs) were compared.ResultsWhen using a p-value cut-off strategy of p < 0.05, notable discrepancies in the predictors of overall satisfaction were common. Six indicators, including age, perceived wait before bed placement, perceived wait before physician evaluation, physician care, discharge instructions, and waiting time satisfaction, were statistically associated with satisfaction for only one of the four assessments. In contrast, examining the size of the ORs associated with each predictor showed far fewer discrepancies. Only physician care appeared to have large differences in the strength of its relation to overall satisfaction. This trend was confirmed by the aggregated logistic regression analysis.ConclusionsUsing p-value cut-offs as the sole criterion for interpreting which variables are most important in determining ED patient satisfaction is ill-advised, and may lead to spurious conclusions of discrepant findings. Nevertheless, some determinants of ED satisfaction likely differ meaningfully based on the cohort that is being examined. Overgeneralizing conclusions derived from a single ED patient satisfaction study should be avoided, especially those studies that are cross-sectional and use a single site.

28 keywords...

hide…