The mean age of the study participants was 72.8 (SD, 9.1); 70% of the patients were female, 44% were African-American, 39% had at least a high school education, 33% were living alone, and 40% were married. The mean administration time for PRISM-PC ranged from 20 to 26 minutes. The item language, mean score and the percentage of missing data are presented in . No new additional items were identified by the participants and approximately 95% of the participants felt comfortable answering the questions.
Item score, SD, and percentage of missing data
The VARIMAX-rotated factor loadings for the 8 screening acceptance items are presented in . The screen plot suggested the existence of one or two factors. Specifically, the eigenvalues of the first four factors were: 5.10, 1.03, 0.64, and 0.47, showing that the eigenvalues leveled off after two factors. Clinical judgment suggested that the two-factor solution was sensible. Specifically, we found that the 8 items on acceptance of screening loaded on two dimensions, accepting the screening to know the risk and accepting the actual testing for AD screening. We also found that two pairs of items were redundant with a high correlation of 0.81 for items 7 and 8 and correlation of 0.97 for items 12 and 13. Thus, we recommend selecting only items 7 and 13 in the final PRISM-PC questionnaire. These two factors explained 76.6% of the total variance and 90.7% of the shared variance in the eight items. Although the data suggest that scores from the acceptance scale be reported as two factors by summing the items from each subscale, it would also be appropriate to report a total score on this scale in addition to the two subscale scores given that the one-factor solution revealed loadings above .60 for all eight items.
Factor loadings for screening acceptance
The VARIMAX-rotated factor loadings for the 29 perceived benefits and harms items are presented in . Based on the scree plot (first four eigenvalues were 5.41, 4.42, 1.42, and 1.34) and clinical judgment, we obtained 4 major factors (see ). The first factor consisted of items related to benefits of dementia screening, while the second factor consisted of items related to the stigma of being diagnosed with AD. The third factor consisted of items related to insurance and loss of independence. The final factor consisted of items related to negative effects on family and emotion. One item did not load on any factor, item 16. In , the loading for each item is bolded under the factor loaded most highly. The exception was that item 39 was lumped under the Stigma factor (based on clinical judgment), despite the fact that this item loaded slightly higher (.38) on Benefits than Stigma (.37). These four factors explained 43.4% of the total variance and 78.8% of the shared variance in the eight items.
Factor loadings for benefits and harms of dementia screening
We report the internal consistency of all subscale scores and of the total acceptance score, using Cronbach’s alpha coefficient, in . The screening acceptance scales show good internal consistency with values of 0.77 or higher. All of the benefits and harms scales showed good internal consistency (values > 0.70) except the suffering/emotional scale.
Internal Consistency for the various scales of the PRISM-PC Questionnaire