A Patient Satisfaction Instrument: precision by replication.

Abstract

The Patient Satisfaction Instrument (PSI) was developed over a series of five clinical and administrative studies during a period of eight years, with a total of 600 patients, primarily medical-surgical inpatients and outpatients. The process illustrates measurement precision by replication. The PSI is a Likert-type summated rating scale with three dimensions of patient satisfaction: technical-professional care, trust, patient education. It was adapted for use with inpatients from Risser's out-patient instrument. Internal consistency estimates appear satisfactory and stable across the various studies; for example, alpha coefficients for the Technical-Professional subscale average .79, Education coefficients average .78, and Trust coefficients average .88. Interitem, item-subscale, and interscale correlations corroborate the alphas. Construct validity estimates were made via convergent/discriminant technique, discriminance, and predictive modeling. Empirical correlations moderately substantiated the multiple, convergent/discriminant predictions. Discriminance was strongly documented for all but the Education subscale, which had modest support. Predictive modeling produced moderate to strong validity estimates. Overall, the PSI has acceptable levels of validity and reliability with refinements indicated.