Abstract
To the Editor,
We read with interest the recent article by Kang et al. 1 determining the risk factors associated with acute kidney injury (AKI) after hip fracture surgery in elderly patients. By the multivariate logistic regression analysis and receiver operating characteristic (ROC) curve analysis, they showed that important risk factors for postoperative AKI were blood loss and postoperative serum albumin levels. Because postoperative AKI has been significantly associated with a prolonged hospital stay, increased short- and long-term complications, and increased in-hospital and long-term mortality following hip fracture surgery in elderly patients, 2 their findings have potential clinical implications. However, there are several issues in this study that are not well addressed.
First, in the methods, the authors described that chronic kidney disease (CKD) was defined by the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines based solely on estimated glomerular filtration rate (eGFR), that is, an eGFR of less than 60 mL/min/1.73 m2 was regarded to have CKD. In the results, the authors only provided mean preoperative eGFR in patients with and without AKI. We would like to know why this eGFR standard was not used to determine if patients had CKD before surgery. It must be noted that preoperative CKD is common among elderly patients with hip fracture, with an incidence of 40% on admission according to the diagnostic criteria of eGFR. 3 The available evidence indicates that CKD is one of the most established preoperative predisposing factors for AKI after noncardiac surgery. 4
Second, it is generally believed that time window of AKI assessment after surgery is important for diagnosis and staging of AKI. 4 In this study, the occurrence and severity of postoperative AKI were defined by AKI Network classification system. In fact, this classification system requires a time window of 48 h after surgery. In the methods, the readers were not provided with exact observed time of AKI in this study. In the results, moreover, we noted that mean onset of postoperative AKI in this study was 8.0 days and recovered after 7.0 days following the occurrence of AKI. Perhaps, this study design should include the consensus definitions of AKI with a longer time window, such as the KDIGO criteria and Risk/Injury/Failure/Loss/End-stage classification system.
Third, the readers were not provided with the details of intraoperative hemodynamic changes, especially for the occurrence of intraoperative hypotension. The available literature shows that a short duration of severe intraoperative hypotension can result in an increased risk of postoperative AKI, with an independent graded relationship between duration of intraoperative hypotension and postoperative AKI. 4 Recently, the association of intraoperative hypotension with postoperative AKI has also been demonstrated in elderly patients undergoing hip fracture surgery. 5 Most important, mean onset of postoperative AKI in this study was 8.0 days, but early postoperative complications were not observed. It has been shown that sepsis, acute myocardial infarction, infection, and anemia in early postoperative period are independently associated with an increased risk of AKI after total arthroplasty. 6 We argue that not taking these intraoperative and postoperative risk factors into the model would have tampered with the inferences of multivariate logistic regression analysis for odds ratios of risk factors for postoperative AKI.
Fourth, to evaluate the predicting ability of estimated blood loss and postoperative albumin level for postoperative AKI, the authors performed the ROC curve analysis and provided the areas under the ROC curve, cutoff values, sensitivities, and specificities of two risk factors for postoperative AKI. However, it was unclear why the authors did not provide their positive and negative predictive values and Youden indexes at the cutoff values. The Youden index is important not only because it can provide the cutoff value but also because it is a direct measure of diagnostic accuracy at the cutoff value, that is, the maximal overall correct classification rate that a predictor can achieve. 7 A predictor with a large area under the ROC curve may have an unsatisfactory overall correct classification rate at the cutoff value, and vice versa. As the positive and negative predictive values and Youden index at the cutoff values of two risk factors were not provided in this study, we cannot determine whether they are the valuable predictors for the occurrence of AKI after hip fracture surgery.
Footnotes
Author contributions
All authors had carefully read the manuscript of
