A Brief Patient Self-Report Screening Measure of Cancer Treatment-Related Memory Problems: Latent Structure and Reliability Analysis.

Background: Memory can be deleteriously affected by cancer treatment. It is important to quantify these effects, to assess the effectiveness of interventions to ameliorate them, and to monitor changes in cancer patients' memories over time. However, brief and reliable screening measures of cancer treatment-related memory problems are lacking. We report on the latent structure and reliability analysis of a brief patient self-reported cancer-related memory problem screening measure (SRMP).

Methods: Participants (N=781) included in this analysis were part of a larger study on cancer-related fatigue. Patients completed the Fatigue Symptom Checklist (FSCL) at four time points. Five items from the FSCL that assess memory problems were aggregated into the SRMP. Suitability of the data for latent structure analysis was determined using the various criteria: Kaiser-Myer-Olkin, Bartlett's Test of sphericity, Kaiser's simplest criterion test of eigenvalue (λ) >1, and the presence of item-correlation coefficients of r ≥ 0.30. Principal components analysis (PCA) and Cronbach coefficient alphas (α_s) were calculated to determined validity and internal consistency of the SRMP.

Results: Reliability assessment of the SRMP revealed high internal consistency of the SRMP as indicated by α = 0.90. The PCA revealed a one-dimensional measure with a single component with λ > 1 that explained 71% of the variance. The KMO value was .87, and the Bartlett's Test of Sphericity was significant (p < 0.001). Subsequent reliability assessments of the SRMP revealed α_s of 0.90 and above, all with one component explaining 71% to 73% of the variances.

Conclusion: The results support the use of the SRMP as a one-dimensional brief screening measure of cancer treatment-related memory problems. The SRMP could be used as an initial indicator of underlying memory problems that may need further examination. Future studies need to establish the construct validity of the SRMP.