Mortality risks of body mass index and energy intake trajectories in institutionalized elderly people: a retrospective cohort study

Abstract

Background Many factors can identify the mortality risks of institutionalized elderly people; among all such factors, body mass index (BMI) and energy intake (EI) can be employed as independent predictors. The objective of this study was to elucidate about the mortality risks and the trajectory of two parameters obtained from continuously monitored data. Methods This retrospective cohort study targeted 218 elderly people who were admitted to a nursing home for at least 6 months between 2007 and 2020 and passed away at the nursing home. BMI and EI (kcal) per body weight (BW) were continuously measured until death. Results BMI continued to decrease for 5 years until death. The rate of change of BMI significantly increased for 3 years before death (P = 0.004). In contrast, the rate of change of EI per BW significantly increased after 1 year before death (P < 0.001); in addition, 2 months before death, this rate of change significantly exceeded that of BMI (P = 0.007). In the four risk groups that were examined, a BMI of <18.0 + EI per BW of <29.2 and a BMI of <17.3 + EI per BW of <28.3 were significantly correlated with a high risk of death (log-rank test: P < 0.001, P = 0.002, respectively). There was no significant difference in the hazard ratio based on the age at the time of admission; however, when EI per BW was <23.8, the risk of death was significantly higher [hazard ratio = 4.36; 95% confidence interval: 2.31–8.24]. Conclusions Elderly people in the current study presented a tendency toward decreasing BMI starting 60 months prior to death even if EI per BW remained constant. In the 1 year before death, EI per BW rapidly decreased. When the rate of change of EI per BW exceeded the rate of decrease of BMI, it was considered to be the point of no return when death was imminent. Our study showed that identifying mortality risks from the relationship between the trajectories of the two parameters that were continuously measured for several months to years is possible.