PSYCHOMETRIC EVALUATION OF A PRE-DEATH GRIEF (PDG) SCALE FOR DEMENTIA FAMILY CAREGIVERS

Abstract

Caregiver support in dementia care traditionally focuses on the emotion of stress, with emphasis on improving coping skills and reducing burden. However, the emotional experience of family caregivers can be more complex, and may be better understood with the concept of pre-death grief(PDG). PDG includes a range of emotional responses as caregivers mourn for the psychologically absent patient and anticipate impending losses. It is prevalent among dementia caregivers and is associated with negative consequences. The 50-item Marwit-Meuser Caregiver Grief Inventory(MM-CGI) is a useful measure of PDG. It has three subscales representing different losses, with the Heartfelt Sadness and Longing(HSL) subscale specifically measuring the core of PDG (that is, one’s intrapersonal reactions to lost relationship). Because MM-CGI was never validated outside of USA, we aimed to evaluate its psychometric properties in a multi-ethnic Asian population. Alternatively, we also examined whether the 15-item HSL subscale alone is sufficient as a reliable, valid and unidimensional measurement. Spouses or children of non-nursing home dementia patients were recruited from the clinics of two tertiary hospitals. They completed questionnaires containing the following scales: MM-CGI, Zarit-Burden-Interview(ZBI) and Centre-for-Epidemiologic-Studies-Depression(CES-D). 60% of participants repeated the questionnaires one week later for test-retest reliability. Internal consistency was assessed by Cronbach’s alpha; test-retest reliability by intra-class correlation(ICC); construct validity by correlation with ZBI and CES-D; and confirmatory factor analysis(CFA) by Chi-Square, Standardized-Root-Mean-Square-Residual(SRMR<0.08), Comparative-Fit-Index(CFI>0.90) and Tucker-Lewis-Index(TLI>0.90). Table 1 showed the demographic information. Cronbach’s alpha was 0.97 for MM-CGI and 0.93 for HSL. ICC was 0.94 for MM-CGI and 0.93 for HSL. MM-CGI showed convergent validity with ZBI(Pearson’s r=0.77) and CES-D(r=0.77); and divergent validity with Finance subscale of ZBI(r=0.47) and positive affect subscale of CES-D(r=0.40). HSL showed convergent validity with ZBI(r=0.64) and CES-D(r=0.69); and divergent validity with Finances subscale of ZBI(r=0.37) and Positive Affect subscale of CES-D(r=0.36). In CFA, MM-CGI showed Chi-Square p<0.001, SRMR 0.059, CFI 0.81 and TLI 0.80. HSL had better model-fit, with Chi-Square p<0.001, SRMR 0.046, CFI 0.90 and TLI 0.89. Both MM-CGI and HSL demonstrated reliability and construct validity. However, MM-CGI may require re-exploration of its factor structure. The briefer HSL subscale offers a viable alternative to measuring PDG.