The Patient Satisfaction Survey System
Twenty direct care mental health professionals were then asked to edit the original list of patient concerns so that a final list of 90 items resulted. The list was then formatted with 5-point Likert scales and administered to 75 patients as they were being discharged from an acute treatment unit. An item analysis was done, and all items which had an item-total correlation of less than .30 were excluded from the scale.
Nunnally (1978)argues for a .20 cutoff for item-total correlations, but the authors used the more stringent .30 in order to reduce the total number of items. Further elimination of items which overlapped in meaning with other items resulted in a scale of 41 items.
The scale was then given to all patients that were admitted and discharged from five state hospitals in the state of Missouri for a two-month period of time. Forms from this sample (n=366) were then factor analyzed using the principle component method. A Scree Test was applied to determine that five factors should be retained for varimax rotation. To verify the construct validity of this factor structure, the form was again given to all patients being admitted to and discharged from these same acute inpatient treatment units for another two-month period of time (n=390). Principle component factor analysis was again used with varimax rotation, retaining five factors. These five factors, named Treatment Effectiveness, Trust of Staff, Hospital Environment, Staff Competence, and Patient Involvement, were compared to the results of the factor analysis of the first sample using the Coefficient of Congruence and the Cosine Method (Cattell, 1978).
Three of the factors in the first sample were replicated in the second sample. These were Treatment Effectiveness, Trust of Staff, and Hospital Environment. Both samples were then combined (n=756) and re-factor analyzed with varimax rotation and restricting the analysis to three factors. Three items were eliminated because they did not load on the three factors. An internal reliability test was done with Coefficient Alphas obtained on the three factors of .92, .80, and .70. The factors were correlated with each other to determine shared variance. The first factor correlated with Factor II, .37 and with Factor III, .22 and, therefore, seems to be independent. Factor II correlated .60 with Factor III and, therefore, shows some shared variance (36%).
The face validity of the scale was gained because the original content of the items came from patients and then was edited by professionals. However, criterion validity of the scale was indicated by the fact that patients were able to clearly distinguish different areas of satisfaction. Percentage of satisfaction calculated by adding all the 4 and 5 responses together, and percentage of dissatisfaction was calculated by adding all the 1 and 2 responses. Sixty-nine percent (69%) of all patients expressed an overall satisfaction on the first factor of Treatment Effectiveness with 12.8 percent expressing dissatisfaction. Sixty-four percent (64.2%) expressed satisfaction on Trust of Staff, with only 48.7 percent expressing satisfaction on Hospital Environment (Holcomb, Adams, Ponder, & Reitz, 1989).
Because the authors felt that the strongly worded negative items may be biasing the respondents' answers, it was decided to attempt to rewrite the negatively worded items into positively worded items. For example, the item, "At times I was not treated with respect," became, "I was treated with respect while in the hospital." Also, since the involvement of families has come to be understood as an important factor in treatment gains, family members in the Alliance for the Mentally Ill organization were asked for their comments on the questionnaire; and as a result, four items were added. A few items that were felt to be redundant in meaning were dropped, and some grammatical editing was done. Finally, a scale of 29 items and three factors remained.
This edited Patient Satisfaction Questionnaire was then administered to 160 adults being discharged from a private psychiatric hospital, 88 adolescents being discharged from a private psychiatric hospital, and 3,456 adults being discharged from state operated inpatient psychiatric units. This total sample was factor analyzed using principle component analysis followed by varimax rotation. A Scree Test was applied, and again three factors emerged with replication of the original three factors with one additional item added by family members of patients loading on the Hospital Environment scale, one item loading on Staff Competence, and the other two family items loading on the Treatment Effectiveness scale.
All items obtained item-total correlations of .40 or greater. The Coefficient Alphas for the three scales with the new sample were .81, .90, and .79 respectfully.
Because of the success of the development of the adult version of PSAT, a state-operated CMHC that was associated with a major Midwestern university requested that we modify the questions for a child interview and parent form. The forms were intended for children, ages 5 to 12, being admitted to a hospital for psychiatric problems and their parents.
The PSAT questionnaire was also modified for use with adults and adolescents being admitted for alcohol and drug abuse problems. Norm statistics were gathered from a state-operated drug/alcohol abuse program (28-day program).
There is extensive literature on measuring consumer satisfaction with outpatient services. Some authors have chosen to emphasize one overall dimension of client satisfaction (Attkisson, 1982). However, other authors, including the present authors, feel that research supports multidimensional measurement of outpatient satisfaction (Love, et al., 1979, Holcomb, et al., 1989). As a result, the authors have developed a PSAT form for outpatients that stresses the three key dimensions: Satisfaction with Treatment, Staff Competence, and Environment/Atmosphere. The form has also been adapted for clients receiving alcohol/drug abuse outpatient services and medical/surgical services.
Behavioral Health Concepts (BHC, Inc.), 2716 Forum Blvd., Suite 4, Columbia, MO 65203. Tel: 573-446-0405; Fax: 573-446-1816.