The relationship between patient satisfaction and dissatisfaction with nursing care in hospitalized postpartum and medical-surgical patients

The purpose of this study was to describe the relationship of satisfaction to dissatisfaction with nursing care among homogeneous patients, represented by postpartum patients, and heterogeneous patients, represented by medical surgical patients of diverse diagnoses, in hospital settings. Data derived from these two samples were used to determine if satisfaction/dissatisfaction is a bipolar concept, or if these two concepts are orthogonal, through confirmatory factor analyses (LISREL). The Revised Risser Patient Satisfaction Scale (54 items, 6 subscales) was distributed to convenience samples of 281 postpartum patients and 263 medical-surgical patients, who had received dismissal orders from their physicians. Data from both samples indicated Cronbach's alpha values ranged from.83 to.96 for the six subscales. Subscale-subscale correlations ranged from.40 to.87. The lower subscale-subscale correlations involved the education subscales. Principle components analysis with varimax rotation indicated a similar three-factor solution for both samples. Factor 1 was composed negative items, and was labeled "dissatisfaction." Factor 2 was composed of positive items, and was labeled "satisfaction." Factor 3 was composed of a mixture of items, and was labeled "mixed behaviors." When principle components analysis with oblique rotation was repeated using data sets which deleted marginal items, Factor 3 disappeared from the homogeneous sample. With the heterogeneous sample, Factor 3 was retained, and consisted of items from the education positive subscale and the trust positive subscale. Multivariate analysis of variance comparing the two samples revealed no statistically significant differences in subscale values. The differences between the education subscales produced the largest F value ($F=2.642,p=.10$). Confirmatory factor analyses, with sequential model modifications, were performed on data from the two samples. A model of the orthogonal relationship of satisfaction to dissatisfaction (with the technical-professional and trust domains and correlated technical-professional error terms) perfectly fit data obtained from both samples (chi-square = 0, goodness of fit = 1). Confirmatory factor analysis did not support inclusion of the education subscales as a conceptual component of satisfaction or dissatisfaction.

The two samples differed in their response to the education subscales. Deletion of the education subscale substantially improved the goodness of fit of the heterogeneous sample model. Correlation of the error terms of the technical-professional subscales substantially improved the goodness of fit of the homogeneous sample model.