RESUMO
BACKGROUND AND HYPOTHESIS: Assess incidence of Acute Kidney Diseas and Disorders (AKD) and Acute Kidney Injury (AKI) episodes and impact on progression of renal dysfunction and risk of all-cause mortality in the community. METHODS: Community of 1 863 731 aged > 23 years with at least two serum creatinine measurements. eGFR was calculated using CKD-EPI formula. CKD, AKD and AKI were defined according to the harmonized KDIGO criteria (Lameire 2021). The sCr values and RIFLE scale was used to classify episodes. Progression of renal dysfunction and mortality were evaluated. RESULTS: 56 850 episodes of AKD in 47 972 patients in 4.8 years were identified. AKD incidence of AKD was 3.51 and 12.56/1000 patients/year in non-CKD and CKD, respectively. One AKD episode was observed in 87.3% patients, two in 9.3% and three or more in 3.4%. A second episode was less common in patients without CKD (10.3%) compared to those with CKD (18.4%). Among patients without CKD a total of 43.8% progressed to CKD, and those with previous CKD 63.1% had eGFR decline of > 50%. The risk of progression to CKD was higher in women, older, overweight-obesity and heart failure, as was the risk of eGFR decline > 50% in CKD patients, although the number of AKD episodes was also a risk factor. AKI episodes were observed in 5646 patients with or without CKD. Of these, 12.7% progressed to CKD and of those with pre-existing CKD, 43.2% had an eGFR decline of > 20%. In the toal population mortality within 3 months of detection of AKD episode occurred in 7% patients, and was even higher in patients with AKI, 30.1%. CONCLUSION: Acute elevations in serum creatinine in the community may pose a health risk and contribute to the development of CKD. Identification of therapeutic targets and provision of appropriate follow-up for those who survive an episode is warranted.
RESUMO
Artificial Intelligence is creating a paradigm shift in health care, with phenotyping patients through clustering techniques being one of the areas of interest. OBJECTIVE: To develop a predictive model to classify heart failure (HF) patients according to their left ventricular ejection fraction (LVEF), by using available data from Electronic Health Records (EHR). SUBJECTS AND METHODS: 2854 subjects over 25 years old with a diagnosis of HF and LVEF, measured by echocardiography, were selected to develop an algorithm to predict patients with reduced EF using supervised analysis. The performance of the developed algorithm was tested in heart failure patients from Primary Care. To select the most influentual variables, the LASSO algorithm setting was used, and to tackle the issue of one class exceeding the other one by a large amount, we used the Synthetic Minority Oversampling Technique (SMOTE). Finally, Random Forest (RF) and XGBoost models were constructed. RESULTS: The full XGBoost model obtained the maximum accuracy, a high negative predictive value, and the highest positive predictive value. Gender, age, unstable angina, atrial fibrillation and acute myocardial infarct are the variables that most influence EF value. Applied in the EHR dataset, with a total of 25,594 patients with an ICD-code of HF and no regular follow-up in cardiology clinics, 6170 (21.1%) were identified as pertaining to the reduced EF group. CONCLUSION: The obtained algorithm was able to identify a number of HF patients with reduced ejection fraction, who could benefit from a protocol with a strong possibility of success. Furthermore, the methodology can be used for studies using data extracted from the Electronic Health Records.
