The association between anion gap and length of stay in patients undergoing hip fracture surgery: data from the MIMIC-IV database

Objective: To explore the relationship between anion gap (AG) and length of stay (LOS) in patients undergoing hip fracture surgery.

Methods: Clinical data of patients diagnosed with hip fracture and undergoing surgery were extracted from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. Logistic regression analysis by adjusting different covariables and threshold effect analysis were used to analyze the relationship between AG and LOS. Subgroup analysis and interaction test were also performed to detect their relationship. Receiver Operating Characteristic (ROC) analysis was performed to identify the prediction performance and cutoff value of AG.Kaplan-Meier (KM) survival analysis was used to explore the influence of AG on overall survival.

Results: A total of 1508 patients were enrolled and the median LOS was 4.9 days. The correlation between AG and LOS > 7 days was observed among 3 regression models when regarding AG as continuous variables (all OR > 1, all P < 0.05). After stratifying samples with AG quartiles, their relationship was only presented in the Q4 group both in model 1 and model 2 (all P < 0.001). The risk of LOS > 7 days gradually increased with increasing AG quartiles (all P for trend < 0.05). Further, threshold effect analysis found that their association was mainly observed when AG ≥ 14 mEq/L (OR = 1.122, P < 0.001). Subgroup analysis showed that their correlation was not influenced by sex, age, BMI, ethnicity, classification of fracture, therapeutic method, CHD, hypertension, osteoporosis, diabetes and admitted to the ICU (all P for interaction > 0.05). ROC analysis identified 14.5 as the cutoff value of AG for predicted LOS > 7 days. Survival analysis found that patients in the AG < 14.5 group had better overall survival.

Conclusion: In patients undergoing hip fracture surgery, the AG was positively correlated with LOS, and 14.5 mEq/L AG was the cutoff value for predicting LOS > 7 days. The cutoff value can favorably distinguish the survival difference of patients.