Comparison of Two Caregiver Burden Scales in a Low-resource Primary Palliative Care Setting and Deriving a Three-item Caregiver Burden Inventory

Objectives: This study aimed to compare the Malayalam versions of the 9-item Achutha Menon Centre–Caregiver Burden Inventory (AMC-CBI) with the 22-item Zarit Burden Interview (ZBI-22), hypothesising that caregiver burden in palliative home care settings could be measured comparably using either tool. We also tried to identify a global question or brief screening items for caregiver burden based on the performance of the AMC-CBI items relative to the ZBI-22. In addition, we explored how participants understood and interpreted the ZBI-22 items. Materials and Methods: The study used a convergent mixed-methods design comprising a quantitative survey and qualitative interviews. The survey covered 60 adult primary caregivers of palliative care patients receiving home care in rural Kollam, Kerala, who were interviewed telephonically using a structured interview schedule with the ZBI-22 and the AMC-CBI scales and demographic, socioeconomic and caregiving-related details. Analysis was through descriptive statistics, internal consistency assessments using alpha and omega coefficients and validity assessments using agreement methods on z-transformed scores available in the SimplyAgree package for R and Jamovi, including average bias and limits of agreement and Receiver Operating Characteristic (ROC) analyses. Qualitative interviews were conducted telephonically, audio recorded and done by AMC-CBI. Cognitive interviews ( n = 7) were conducted with four experts and three caregivers, using a semi-structured guide, to explore item clarity, relevance, redundancy and comprehension of ZBI-22. In-depth interviews ( n = 4) were conducted with four additional caregivers (2 cancer, 2 non-cancer cases; one each with low and high burden) to explore caregiving experiences, and transcripts were analysed to compare narrative alignment with ZBI-22 and AMC-CBI items by two independent reviewers. Results: Internal consistency was high for both tools (α and ω > 0.8). Three AMC-CBI items (‘lost hope’, ‘feel sad’, and ‘financial burden’) were selected for a Brief AMC-CBI (ω = 0.787). Both AMC-CBI p = 0.747) and Brief AMC-CBI p = 0.731) strongly correlated with ZBI-22. Agreement analyses with z-scores (Bland-Altman, Deming regression) demonstrated good agreement, though one outlier was noted. ROC analysis showed AMC-CBI ≥12 and Brief AMC-CBI ≥3 as optimal cut-offs. Identity-related strain over a long caregiving duration might lead to discordant classification with ZBI-22 more likely to elicit burden. Qualitative interviews highlighted some item complexity and translation issues in ZBI-22, especially with the word ‘ duritham ’. Reviewers found the AMC-CBI easier to apply during transcript coding. Conclusion: Both tools, the AMC-CBI and Brief AMC-CBI correctly categorised caregiver burden levels in comparison to the ZBI-22, but AMC-CBI was more contextually appropriate. Selecting the full or the Brief AMC-CBI for caregiver burden assessments would be a thoughtful decision that considers the unique requirements and considerations present in palliative care settings. Further studies on the AMC-CBI or its short form for inter-rater reliability assessments and concordance with the ZBI-22 in larger samples would be useful for enhancing the applicability of these tools.