Etiology, interventions, and predictors of mortality in a tertiary malnutrition clinic: A retrospective cohort study.

BACKGROUND

Malnutrition worsens outcomes in chronic diseases, and delayed detection increases morbidity and mortality. This study described the characteristics of malnutrition in a tertiary referral clinic, summarized nutrition interventions, and identified predictors of mortality.

MATERIALS AND

METHODS

This retrospective cohort study included 431 adults referred to a tertiary malnutrition clinic between 2017 and 2023. Global Leadership Initiative on Malnutrition (GLIM) criteria were applied retrospectively to routinely documented clinical data to classify malnutrition severity. Logistic regression identified factors associated with severe malnutrition, and Cox proportional hazards models evaluated predictors of mortality.

RESULTS

Among 431 patients, 66.8% had severe malnutrition by GLIM criteria. Mean body mass index (BMI) was 20.9 ± 5.9 kg/m2, with weight loss of 19.0 ± 11.9% from usual weight. Leading etiologies were reduced oral intake (77.7%) and malabsorption (45.2%), including short bowel syndrome. After assessment, 37.6% received dietary counseling, 29.9% were recommended oral supplements, 13.9% initiated enteral feeding, and 18.6% received PN (11.6% newly initiated). Median clinical follow-up was 19 months; mortality ascertainment documented 105 deaths (24.4%). Low hemoglobin was associated with severe malnutrition (adjusted odds ratio, 2.49; 95% CI, 1.56-3.97). In the Cox model stratified by sex, higher Charlson Comorbidity Index (CCI) score (adjusted hazard ratio [aHR] per point, 1.20; 95% CI, 1.12-1.27), severe GLIM malnutrition (aHR, 1.76; 95% CI, 1.09-2.86), and low hemoglobin (aHR, 3.02; 95% CI, 1.99-4.56) were associated with mortality.

CONCLUSION

Severe malnutrition was common in this tertiary referral population. Greater comorbidity burden, severe GLIM malnutrition, and low hemoglobin were associated with higher mortality, supporting risk stratification and close clinical monitoring.