Using machine-learning methods to predict in-hospital mortality through the Elixhauser index: A Medicare data analysis.
Jianfang Liu, Sherry Glied, Olga Yakusheva and 5 others
PMID 37221452WHAT IT FOUND
The Elixhauser comorbidity index predicts in-hospital death well in Medicare patients.
Adding patient demographics and admission source did not improve prediction. Machine learning models identified weight loss, neurological disorders, and metastatic cancer as the highest risk factors.
Key findings
01The Elixhauser comorbidity index showed good to strong discrimination for predicting in-hospital mortality across all tested statistical models.
02Adding demographic information and admission source to the comorbidity measures did not improve the models' predictive power or accuracy.
03Weight loss, neurological disorders other than paralysis, and metastatic solid tumors were associated with the highest increased risk of in-hospital death.
STILL TO COME
How it was doneWhat they foundWhat it means for RNs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The study relied entirely on administrative claims data, meaning it could not account for clinical variables like functional capacity or laboratory results that heavily influence mortality. The findings are only applicable to adult Medicare populations and cannot be generalized to pediatric patients. The negative associations for conditions like uncomplicated hypertension and obesity may reflect coding biases where sicker patients have their less acute diagnoses omitted from the record. The artificial neural network operates as a black box, meaning it cannot explain how individual patient variables contribute to its final risk score.
Declared interests
The authors declare no conflicts of interest. The research was supported by an NIH grant.
The easy way to misread this
Do not assume that demographic factors like age or race improve mortality prediction models. The study explicitly found that adding these factors to comorbidity measures did not improve predictive power or accuracy.