RETROSPECTIVE ONLY VERSUS A RETROSPECTIVE-PROSPECTIVE METHOD TO IDENTIFY ACUTE KIDNEY INJURY: AN EPIDEMIOLOGICAL STUDY EXAMINING INCIDENCE, PREVALENCE AND ADVERSE EVENTS.

Acute Kidney Injury (AKI) is a common complication among hospitalised patients. However, its diagnosis in retrospective studies rely on the availability of a baseline creatinine measurement. In many cases, such measurements are not always available, leading to under reporting of AKI prevalence. For patients with no baseline creatinine measurements, prospective creatinine values can be used to determine recovery from AKI, thereby identifying episodes of AKI that might otherwise go undiagnosed.  In this study, we aimed to compare two clinical approaches to diagnosing AKI, the retrospective only approach (AKI-RO) and the combined retrospective and prospective approach (AKI-RP). Our goals were to estimate the incidence and prevalence of AKI using each approach, to evaluate the associated clinical outcomes and to characterise individuals with a missed AKI diagnosis (those with an AKI-RP but no AKI-RO diagnosis). 

We conducted a retrospective cohort study of adult patients admitted to an Australian Local Health District (2008- 2017). AKI was diagnosed based on serum creatinine and classified using two approaches: the retrospective only method (AKI-RO), which relied solely on retrospective creatinine measurements, and the retrospective-prospective method (AKI-RP), which incorporated both retrospective and prospective measurements. We examined the incidence, prevalence and risk of mortality and kidney replacement therapy (KRT) using Cox proportional hazard regression to determine hazard ratios (HR) and 95% confidence intervals (CI). A propensity matched analysis of AKI-RO with missed AKI diagnoses was also undertaken to better characterise the missed AKI cohort. 

Of the 81,017 patients included in the study cohort, AKI was diagnosed in 16,551 patients (20%) using the AKI-RP method and in 15,479 patients (19%) using the AKI-RO method. A total of 1,072 patients were identified as having a missed AKI diagnosis, diagnosed by AKI-RP but not AKI-RO. This group had a higher proportion of younger patients with a lower comorbidity prevalence. The incidence of AKI using the AKI-RP was 43.7 per 1,000 patient years (95%CI 43.1- 44.4) compared to 41.6 (95%CI 40.9- 42.2) using AKI-RO. Irrespective of the diagnostic method used, AKI was found to be an independent risk factor for both mortality and commencement of (KRT). In the propensity-matched analysis, there was a significantly increased risk of mortality in the AKI-RO group when compared to the missed AKI group, even after adjustment (HR 0.56, 95%CI 0.48- 0.66, P<0.001), but no difference in the risk of KRT (HR 0.33, 95%CI 0.10- 1.09, P= 0.07). 

Our study highlights that AKI incidence and prevalence could be underestimated using the AKI-RO method compared to the AKI-RP approach. Although overall risks of mortality and KRT appear similar between methods, relying solely on the AKI-RO method risks overlooking younger patients with fewer comorbidities. These patients, identified only through AKI-RP, exhibit a lower risk of mortality even after adjusting for confounders. We therefore recommend that retrospective studies incorporate prospective creatinine measurements to improve AKI detection and ensure more accurate characterisations of affected patients.