Affiliations 

  • 1 Department of Anesthesiology, Faculty of Medicine, University of Malaya, Kuala Lumpur, Malaysia
  • 2 Department of Biomedical Imaging, Faculty of Medicine, University of Malaya, Kuala Lumpur, Malaysia
JPEN J Parenter Enteral Nutr, 2020 03;44(3):425-433.
PMID: 31173666 DOI: 10.1002/jpen.1666

Abstract

BACKGROUND: Low muscularity (LM) is associated with high mortality in the Caucasian critically ill population. Muscularity can be accurately measured by the skeletal muscle index (SMI; cm2 /m2 ) generated by computed tomography (CT). This study aimed to establish the overall and sex-specific cutoff values that predict hospital mortality in an Asian critically ill population.

METHODS: This single-center, retrospective, observational study included patients aged ≥18 years with an abdominal CT conducted within 72 hours of admission to the intensive care unit. SMI generated from CT images at the level of the mid-third lumbar vertebra were extracted from the medical records. Area under the receiver operating characteristic curves (AUC) was generated to determine the SMI cutoff values for hospital mortality. Association between LM (defined by SMI cutoff value) and hospital mortality was further evaluated by multivariable logistic regression.

RESULTS: In a sample of 228 patients, the overall SMI cutoff value (cm2 /m2 ) for hospital mortality was 42.0 (AUC: 0.637; sensitivity: 66.7%, specificity: 56.8%), whereas it was 46.5 in males and 35.3 in females. More males than females had LM (51.4% vs 37.5%), and >40% of overweight/obese patients had LM. Patients with LM were older and had a longer duration of mechanical ventilation and hospitalization. After adjusting for known confounders, LM independently predicted hospital mortality in the overall sample (adjusted odds ratio: 2.42; 95% CI 1.16-5.03; P = 0.003) and in both sexes.

CONCLUSION: This study established a set of SMI cutoff values that predict hospital mortality. LM is independently associated with hospital mortality.

* Title and MeSH Headings from MEDLINE®/PubMed®, a database of the U.S. National Library of Medicine.