Purpose: The purpose of this study was to evaluate the psychometric properties (reliability, factor structure and validity) of the Problem Solving Inventory (PSI) among informal caregivers of community-dwelling persons with memory loss.
Methods: For this secondary analysis, data from the baseline and 2-month post-baseline assessments of a randomized controlled trial designed to improve informal caregivers’ problem solving skills related to medication management of community-dwelling persons with memory loss was used. The 32-item PSI which was originally tested among undergraduate college students with three subscales (problem solving confidence, approach-avoidance style, and personal control) was used to measure caregivers’ level of problem solving, with lower PSI total scores indicative of better problem solving. Reliability of the PSI was examined considering internal consistency using Cronbach’s alpha as well as test-retest reliability over a 2-month period for control group participants. Construct validity of the PSI was explored using principal component analysis for extraction with varimax rotation to explore the factor structure of the PSI. Pearson correlations between the PSI total score and scores on the Beck Depression Inventory II (BDI), Newest Vital Sign (NVS) for health literacy, as well as Self-Efficacy for Managing Chronic Disease (SEM) was tested at baseline for convergent validity. It was hypothesized that there would be at least a moderate positive association between PSI total score and BDI score; a moderate negative association between PSI total score and total scores of NVS as well as SEM.
Results: Participants (n=78 with complete data) were mainly white (87.2%), female (71.8%), and on average ± SD 66.5±12.3 years of age with 14.7±2.8 years of education. Estimates of internal consistency for the PSI using Cronbach’s alpha were 0.92 for the overall total score, and 0.90, 0.82, and 0.84 for the problem solving confidence, approach-avoidance style, and personal control subscales, respectively. Estimates of test-retest reliability over 2 months were 0.562 (p<0.001) for the total score and 0.435 (p=0.006), 0.587(p<0.001), and 0.559 (p<0.001) for the problem solving confidence, approach-avoidance style, and personal control subscales, respectively, for caregivers in the control group (n=39). Exploratory factor analysis using principal component analysis extraction revealed seven factors for the PSI. Two factors were dropped due to less than three items having loaded on those factors, and a total of 4 items were excluded following exclusion of the two factors as well as considering a factor loading threshold of ≥0.40. As a result, five factors were retained with a total of 28 items, where factors one and two were “personal control” (8 items) and “problem solving confidence” (9 items) as originally identified for the PSI. Most of the items for factor three “proactive approach” (3 items), factor four “avoidance style” (4 items) and factor five “decision processing” (4 items) were included in the approach-avoidance style in the original PSI. Pearson correlations showed a moderate positive relationship between PSI total and BDI total scores (r=0.372, p=0.001), and a moderate negative relationship between PSI total and SEM total scores (r=-0.408, p<0.001); however, the association between PSI total and NVS total scores (r=-0.096, p=0.411) was not significant.
Conclusion: The 32-item PSI was internally consistent and showed temporal stability over 2 months for measuring problem solving abilities in informal caregivers of persons with memory loss. The extracted factors for PSI (five factors with 28 items) within this context revealed a somewhat different factor structure compared to the original scale which was tested among undergraduate college students, with more specific factors of “proactive approach”, “avoidance style” and “decision processing” that could be helpful in explaining problem solving behaviors among caregivers of persons with memory loss. The correlation between PSI total score and total scores of BDI as well as SEM showed convergent validity, but not with the NVS total score.
Implication: The PSI has the potential to help healthcare providers more adequately evaluate problem solving behaviors among informal caregivers of persons with memory loss and ultimately aid in identifying interventions to improve clinical outcomes. Given the limitations of having a small and homogeneous sample in the current study, future research should include a confirmatory factor analysis of the identified structure in current study in a larger sample of informal caregivers, the psychometric assessment of the instrument in caregivers for different patient populations, and the examination of concurrent validity using another problem solving instrument.
See more of: Research Sessions: Oral Paper & Posters