Identification of Heterogenous Long-Term Kidney Function Trajectories After AKI

Background: The aim of this study was to assess heterogeneity in eGFR trajectories after an acute kidney injury (AKI) episode, and examining the differences in patient-characteristic across these distinct trajectory profiles. With this, we take the first step towards identifying at-risk patients earlier using routine data, which might enable the optimization of kidney protection at an earlier stage. Methods: We linked tertiary care serum creatinine measurements of 567,527 individuals between 1993-2024 with primary care measurements, encompassing >5.5 million measurements. Patients aged ≥18 years with AKI (KDIGO) with ≥365 days follow-up, were included. Distinct eGFR trajectory profiles were identified using unsupervised latent class mixed model (LCMM) analysis, starting from seven days post-AKI onset up to ten years. The characteristics age, sex, baseline eGFR, AKI duration (<3/≥7 days), AKI stage, and ICU admission were compared between the identified trajectory profiles. Ten year survival probability for each trajectory profile was assessed using Cox proportional hazards analysis. Results: In total, 30,150 patients were diagnosed with AKI, 20,119 had ≥365 days follow-up. Mean age was 62.0y, 46% was female. AKI stage 1/2/3 were present in 80/14/6%. Transient AKI (<3 days) occurred in 47%. Eight long-term eGFR trajectory profiles were distinguished: High stable; Low stable; Rapidly increasing; Moderately increasing; Slowly increasing; Rapidly decreasing; Moderately decreasing; Slowly decreasing. Substantial differences in the characteristics were noted across the eGFR trajectories, as females and younger patients were overrepresented in profiles with an increasing eGFR after AKI. Patients admitted to the ICU had higher chances of a decreasing eGFR trajectory. The profiles with the highest risks of death were: rapidly increasing, rapidly decreasing and moderately decreasing eGFR. Conclusion: We showed wide heterogeneity in kidney function trajectories after AKI, with differing risks of all-cause mortality. Where some patients may benefit from intensive follow-up, others might have lower risk of long-term kidney dysfunction. Insight in these distinct trajectory profiles might provide valuable information for the development of tailored treatment and follow-up strategies for patients with AKI.