healthmedicinet

We conducted a retrospective cohort study to assess the effect of using different baseline SCr methods, SCr_GFR-75 and SCr_ADM, on the incidence of AKI as well as prognostication performance of the KDIGO definition. We demonstrated that using SCr_GFR-75 as a baseline SCr for AKI diagnosis was associated with the larger number of classified AKI cases. However, using SCr_GFR-75 resulted in the misclassification of non-AKI cases into AKI, as AKI cases only by SCr_GFR-75, but not SCr_ADM, had a no significant increase in 60-day mortality risk. This finding is consistent with SCr_GFR-75 as a more sensitive and SCr_ADM as a more specific surrogate for AKI diagnosis when compared with outpatient SCr as the reference standard.

Ideally, outpatient SCr values, which are reflective of patient pre-morbid kidney function, should be used as the baseline SCr for AKI diagnosis [25]. However, the lack of baseline outpatient SCr level is a common problem encountered in clinical practice [25, 26]. Following the ADQI statement, the use of SCr_GFR-75 as a baseline SCr has been shown to lead to misclassification of AKI in ICU patients [16, 17] and following cardiac surgery [27]. The European Renal Best Practice (ERBP) has recommended the use of SCr_ADM, rather than SCr_GFR-75, when baseline outpatient SCr measurements are not available [18]. In our study, we demonstrated that using either SCr_GFR-75 or SCr_ADM led to misclassification of AKI diagnosis and staging. Use of SCr_GFR-75 has been shown to inflate AKI incidence, whereas use of SCr_ADM underestimates AKI incidence [14, 16, 17]. The inflation of AKI diagnosis bySCr_GFR-75as surrogates for baseline kidney function was due to inaccurate baseline SCr estimation in patients with chronic kidney disease [14, 28]. In contrast, the underestimation of AKI diagnosis when using SCr_ADM was due to the under-diagnosis of community-acquired AKI [14]. We found most of AKI cases missed by SCr_ADM occurred within the first 24 hours of ICU admission.

The analysis in a simulated cohort of patients with baseline outpatient SCr showed the higher sensitivity and lower specificity of AKI diagnosis bySCr_GFR-75 in comparison with SCr_ADM. The decision to use SCr_GFR-75 or SCr_ADM for AKI diagnosis and classification depends on the purpose of the AKI definition. In clinical practice, AKI prevention and prompt treatment might improve patient outcomes. Thus, for risk stratification purposes in clinical practice, we encourage the use of SCr_GFR-75 for AKI diagnosis, as it can conceivably identify more AKI cases. Conversely, for research studies that enroll patients with AKI for invasive tests or treatments, using SCr_ADM may be more suited, as it would be more likely to enroll patients who are going to benefit from the intervention.

The use of estimated SCr by back-calculation with the MDRD formula can allow investigators or clinicians to be more flexible regarding the sensitivity and specificity of the AKI definition. As shown in Table 4 and Additional file 1: Table S3, when using different assumed GFR for SCr estimation, it provides different levels of sensitivity and specificity that we may individualize in each patient in different encounters. Since GFR decreases with age, the use of SCr_GFR-75 might result in over-AKI classification in the elderly [29], therefore use of a different assumed GFR for SCr estimation could be considered in different age groups. For example, using SCr_GFR-70 in elderly but SCr_GFR-100 in younger adult would yield the highest sum of sensitivity and specificity. In addition, abnormal renal function at presentation may be due to pre-existing chronic kidney disease or due to acute elevation of SCr from community-acquired acute kidney injury. Therefore, estimated SCr based on lower assumed GFR might be more suitable for patients without baseline outpatient SCr, who present initially with an abnormal SCr. In contrast, using SCr_ADM in patients with abnormal renal function based on initial SCr value in hospital would result in lower sensitivity in the diagnosis of AKI.

This study has several limitations. (1) This is a single-center retrospective study. (2) We did not include the urine output criterion for AKI diagnosis since an indwelling urinary catheter was not used to obtain accurate hourly urine output data for all critically ill patients. (3) Our study compared only two common surrogates for baseline SCr and did not include other advanced methods to estimate the baseline SCr. For example, Siew and colleagues recently demonstrated that a multiple imputation method can improve accuracy in estimating missing baseline SCr and reduce misclassification of AKI [30]. However, the use of this technique in clinical practice is limited and still requires further validation. A multi-center prospective study is ultimately required to address some of these limitations.