2000 Cause of death amongst frail and non-frail older people post-hospital admission in northern Tanzania

Abstract

Abstract Introduction Verbal autopsy (VA) is a tool used to determine cause of death (COD) in regions lacking routine medical certification. Automated algorithms are widely used to interpret VA data. This study aimed to investigate potential differences in COD between frail and non-frail older people in northern Tanzania. Method This work forms part of a longitudinal study investigating the clinical outcomes of 308 consecutive adults aged ≥60 years following admission to four hospitals in northern Tanzania. Frailty status was established on admission using the Clinical Frailty Scale (CFS) and dichotomised with CFS ≥5 indicating frailty. For participants who passed away in the 10-12 months following admission, VA data were collected through telephone interview using the 2022 World Health Organization (WHO) VA instrument. COD estimates were established using the SmartVA-Analyze program, implementing the Tariff 2.0 Method. Results After a mean follow-up period of 10.8 (±0.9) months, VA data were available for 69 participants. At admission screening, 51 (73.9%) were frail, while 18 (26.1%) were non-frail. SmartVA produced COD estimates for 42 (60.9%) participants, while 27 (39.1%) remained undetermined. Cause Specific Mortality Fractions (CSMFs) for non-communicable disease were higher for those with frailty than the non-frail group (70.6% vs. 54.0%, respectively). An undetermined COD was more likely in those with frailty. Of the 27 undetermined COD, 22 (81.5%) were attributed to frail individuals and 5 (18.5%) to non-frail. These undetermined COD represent 43.1% of frailty-related deaths and 27.8% of non-frailty-related deaths. Conclusions Older people with frailty living in Tanzania are more likely to die from non-communicable diseases compared to non-frail older people. Although employing SmartVA to analyse VA data from this population was feasible, it faced challenges ascribing COD to all participants. This limitation may be due to the multi-morbidity often present in older populations where multiple factors combine to cause mortality.