Translation and validation of the Dutch Pittsburgh Fatigability Scale for older adults

Background The original Pittsburgh Fatigability Scale (PFS) was developed to assess perceived fatigability in older adults. The objective of this study was to translate the PFS into Dutch and investigate its validity and reliability among hospitalized older adults aged ≥70 years. Methods The PFS was translated into Dutch and pretested for comprehensibility by the Three-Step Test Interview method. The factor structure underlying the final version was evaluated by confirmatory factor analysis (CFA) and exploratory factor analyses (EFA). Internal consistency of the identified subscales was evaluated by Cronbach’s alpha. Construct validity was evaluated by hypothesis testing. Test-retest reliability was evaluated using intraclass correlation coefficients (ICC) and Bland Altman plots. Results The validation sample included 233 patients. CFA of the original factor structure resulted in poor model fit in our Dutch sample. EFA of PFS physical and mental subscales resulted in a two-factor solution underlying the data with good internal consistency of the identified subscales (Cronbach’s alpha: 0.80–0.92). Five out of six hypotheses were confirmed, indicating good construct validity. Retest assessments were performed among 50 patients and showed good reliability for both the physical (ICC: 0.80, 95%CI: 0.68; 0.88) and mental subscale (ICC: 0.81, 95%CI: 0.68; 0.89). Conclusion The Dutch PFS is a valid and reliable instrument to assess fatigability in older hospitalized patients.

Results of the two-factor solution of the PFS mental subscale are presented in Table S3 -4.
Among elective patients, a crossloadings exists for item I (hosting a social event for one hour). Among acutely admitted patients, a crossloading existed for item C (light household activity for one hour). Cronbach's alphas indicated good internal consistency of the two factors for both subgroups.     Results of the two-factor solutions for both subgroups are presented in Table S3 -7. Among both non-surgical and surgical patients the 2-factor solution resulted in a clear factor structure with all items loadings > 0.5 on one factor, except for the item household activities (item C), which show a crossloading. Theoretically we can distinguish the first factor that includes items that required physical effort to perform (A, B, C, D, G, and J). The second factor included items that require less physical activity (E, F, H, and I). An explanation why household activities load on both factors may be that depending on what type of household activity is thought of by answering; cooking can be considered as a predominantly mental activity, whereas vacuuming and cleaning windows requires physical effort to perform.

EFA PFS physical subscale
Cronbach's alphas including household activity in the first factor indicated good internal consistency of the two factors for both subgroups.

EFA PFS mental subscale
KMO of the PFS mental subscale were 0.87 and 0.85 for non-surgical and surgical patients respectively, and Bartlett's test of sphericity was >0.001 for both subgroups, indicating that EFA could be applied.
The scree test and eigenvalues above 1 opt for a one factor solution underlying the PFS physical subscale data among non-surgical patients. Corresponding eigenvalue (6.31) explained 63% of the variance. Surgical patient data opt for a two-factor solution based on eigenvalues above 1 and scree test. Eigen values (% explained variance) were 5.31 (53%) plus 1.26 (13%) for surgical patients. Results of the factor solutions of the PFS mental subscale of non-surgical and surgical patients are presented in Table S3 -8. All factor loadings of nonsurgical patients were higher than 0.5 indicating a clean factor structure of the single factor solution. Exploring the two-factor solution for non-surgical patients resulted in an unexplainable cross-loading of item G (moderate to high intensity strength training). Among surgical patients, the two-factor solution resulted in a clean factor structure with six items (A, B, C, D, G, J) loading to the first factor and four items (E, F, H, I) loading to the second factor. All factor loadings were higher than 0.5 and no cross loadings indicating a clean factor structure. Cronbach's alphas indicated good internal consistency of the single factor solution for non-surgical patients and the two factor solution for surgical patients.    Bland Altman plots showed good agreement for the PFS physical subscale (mean difference -  Abbreviations: PFS, Pittsburgh Fatigability Scale; ICC, intraclass correlation coefficient for agreement using a 2 way mixed effect model; SEM, standard error of measurement; SDC, smallest detectable change; MIC, minimal important change. a. SEM (%) and SDC (%) are expressed in percentages of the scale range (0-50); percentages are rounded off.