Acute kidney injury (AKI) risk factors in hospitals have been extensively researched. However, the risk factors for identifying high-risk individuals for AKI in the outpatient context were uncertain and varied.   Observational electronic health record data were used to create a predictive model and externally validate it. Patients aged 18 to 90 years old with recurring primary care visits, a known baseline serum creatinine, and creatinine assessed over 18 months without documented advanced kidney disease were included in the study. New predictors and those that have been around for a while Inpatient AKI were studied using established factors. Hospitalization history, smoking, serum potassium levels, and past outpatient AKI were all potential new indicators. Outpatient AKI was defined as a 50% or more increase in creatinine level above a moving baseline of the most recent measurement(s) without the need for a hospital admission within 7 days. For backward stepwise covariate elimination, researchers employed logistic regression with bootstrap sampling. The model was then split into 2 binary tests, one to identify high-risk patients for the study and the other to identify patients who needed more clinical monitoring or intervention. In the development and validation cohorts, outpatient AKI was seen in 4,611 (3.0%) and 115,744 (2.4%) patients, respectively. In the development and validation cohorts, the model obtained C statistics of 0.717 (95% CI, 0.710-0.725) and 0.722 (95% CI, 0.720-0.723), respectively, using 18 variables and 3 interactions terms. The research test identified the validation cohort’s 5.2% most at-risk individuals, with a sensitivity of 0.210 (95% CI, 0.208-0.213) and a specificity of 0.952. (95% CI, 0.951-0.952). The clinical test exhibited a sensitivity of 0.494 (95% CI, 0.491-0.497) and specificity of 0.806 in identifying the 20% most at-risk patients (95% CI, 0.806-0.807). Only patients who survived and had their creatinine levels evaluated during the baseline and outcome periods were included in the study. The outpatient AKI risk prediction model worked well in both continuous and binary versions in the development and validation cohorts.


Source –