Our study was approved by the Ethics Committee of West China Hospital (No. 20191133). In this retrospective, single-center, cohort study, we compared long-term mortality in 12-months aSAH survivors who either had or did not have an episode of AKI during their hospital admission. Data were collected from the electronic health records of all patients admitted to the West China Hospital, Sichuan University between December 2013 and June 2019. Subsequent survival records were extracted from the Household Registration Administration System, a government-run provincial database in China which mandates updated mortality records on all patients, even those who have no follow-up after discharge from the hospital. The institutional review board of the ethics committee of West China Hospital approved the study and granted a waiver of informed consent. The study complied with the STROBE criteria (Strengthening the Reporting of Observational Studies in Epidemiology) (15).

All patients who were alive ≥12 months after the index aSAH were included. aSAH was confirmed by preoperative neuroimaging, cerebrospinal fluid analysis, or the patient's neurosurgeon during a surgical procedure to clip the aneurysm. Participants were excluded if their subarachnoid bleed was related to trauma, rupture of an arteriovenous malformations, detection of a fusiform aneurysm, or had a previous aneurysm treated before the index hospitalization. We also excluded patients whose serum creatinine was not obtained within 24 h of their admission and patients with a history of chronic kidney disease. We also excluded patients whose household registration was not in Sichuan province or whose personal identification number was not found in the electronic medical record system, because that number is the only way to study survival by searching the Household Registration Administration System database.

We adopted the definition of AKI as defined by The Kidney Disease Improving Global Outcomes (KDIGO) guidelines (16). Namely, the change in serum creatinine during hospitalization was compared with baseline serum creatinine: stage 1, >0.3 mg/dL (>26.4 μmol/L) or >1.5- to 2-fold increase; stage 2: >2- to 3-fold increase; and stage 3: >4.0 mg/dL (>353.6 μmol/L), >3-fold increase in serum creatinine, or a need for renal replacement therapy. Urinary output was not considered for the definition of the AKI in this study due to insufficient data. At the time of discharge renal failure recovery was evaluated by comparing the discharge serum creatinine to the baseline serum creatinine. In this study, renal recovery is defined as serum creatinine returning to 1.5-fold of baseline serum creatinine without the need for renal replacement therapy. If the discharge creatinine was >1.5-fold of baseline serum creatinine, persistent renal dysfunction was confirmed.

The primary outcome was long-term mortality, defined as all-cause mortality at the longest follow-up after 1 year. Secondary outcomes were in-hospital complications, including hydrocephalus, delayed cerebral ischemia, rebleeding, seizures, pneumonia, intracranial infection, urinary tract infection, and bloodstream infection.

Death records were extracted from the databases of the Household Registration Administration System with a censoring date of April 20, 2021. In China, every resident has a unique identification number. When a patient dies, the law requires that a death certificate must be entered in the household registration in the bureau of public security within 30 days. As the death certificate database is accurate and complete, the rate of loss to follow-up of this study was negligible, and event time or censoring time is known exactly.

All analyses were done with R version 4.0.3 (Foundation for Statistical Computing) and SPSS version 26 (SPSS Inc). Continuous variables were reported as means (with standard deviation), and categorical variables were reported as counts (frequencies). A 2-sided P < 0.05 was considered significant. We used multiple imputation to impute all missing values, the mean was imputed.

We examined unadjusted overall survival using Kaplan-Meier analyses and using the log-rank test to determine significant differences between groups. Multivariable survival analyses were done using Cox proportional hazards models to determine the independent effect of AKI on overall survival, with data presented as adjusted hazard ratios (HR) and 95% CI. All baseline variables with a value of p < 0.20 in univariate analyses were included into multivariate analyses. We selected the explanatory variables based on covariates examined in prior studies and clinical expertise. We assessed the degree of multicollinearity among the various variables using variance inflation factor (VIF) analysis with a VIF threshold of 10.

We also performed a propensity score-matched analysis (4, 17). A propensity score was calculated based on the probability of a patient having AKI. This score was calculated using a logistic regression model with AKI as the outcome and patient factors as determinants (age, sex, hypertension, diabetes mellitus, smoking, alcohol use, Hunt and Hess grade, Fisher grade, aneurysm location, aneurysm size, external ventricular drain, and aneurysm treatment). The matching variables were based on covariates examined in prior studies and clinical expertise (10, 12, 18–20). This predicted probability was used as a score for one to four matching cases with the nearest neighbor matching with calipers of 0.20 standard deviation; a difference more than 0.10 is considered meaningful. Overall survival was compared between matched groups.

We conducted a sensitivity analysis to assess the association between AKI and in-hospital complications, including AKI, hydrocephalus, delayed cerebral ischemia, rebleeding, seizures, and hospital infection.

We used the E-value to measure the robustness of the association between AKI and long-term mortality for unmeasured or unadjusted confounding (21). E-values were computed with an online E-value calculator (https://www.evalue-calculator.com/) (22). Model assessment was conducted using C statistic and Hosmer-Lemeshow goodness-of-fit statistics.

To our knowledge, the association between AKI and long-term outcomes in patients with aSAH has not been studied previously; however, we are aware of several studies assessing short-term outcomes, which suggested increased mortality among patients with AKI (Supplementary Table 6). In a study of 787 patients with aSAH, patients with risk for renal failure as defined by RIFLE(Risk, Injury, Failure, Loss, and End stage) criteria, had worse 3-month outcomes and higher mortality rates compared to those not at risk (P < 0.0001) (12). In a claims database analysis of hospitalizations for aSAH in the United States, the incidence of acute renal failure (based on validated ICD-9-CM codes) in patients hospitalized for aSAH was 4.0%, and patients with acute renal failure had an increased likelihood of in-hospital death (OR, 2.14; 95% CI, 2.03–2.26) (10). In a recent study of 243 patients with SAH, AKI (defined by the AKIN [acute kidney injury network] criteria) was more frequent in ICU non-survivors but was not an independent predictor of ICU mortality in multivariable analysis (9).

The underlying mechanisms of the relation between AKI and long-term mortality are still under investigation (8). AKI can be a reflection of the general insult of critical illness and may be a marker of systemic illness. Moreover, AKI can exhibit important effects on outcomes that extend well after hospital discharge. Several studies have identified that an episode of AKI may herald eventual long-term renal disease, such as recurrent AKI and development of chronic kidney disease (23, 24). Chronic kidney disease has been reported as 17.76 cases per 100 person-years following AKI (23). AKI also is associated with poorer long-term cardiovascular health. A systematic review of 25 studies involving 254,408 patients found that AKI is associated with a 58% increased risk of eventual heart failure and a 40% increased risk of acute myocardial infarction (25). Patients with chronic kidney disease and cardiovascular disease after AKI have higher mortality rates (8, 26).

This study has several strengths. The large cohort size allows adjustment for potential confounders and results in an enhanced dose-response relationship. We determined long-term mortality based on the household registration system that is considered accurate, up to date, and without lost follow-up.

This study has some limitations. First, potential weaknesses include the retrospective, observational analysis making causal inference difficult and subject to bias from unmeasured factors. Our hospital records could not systematically capture data on daily urine output values, which may lead to underestimation of the incidence of AKI. Second, this study did not contain information on either co-morbid diseases during the follow-up period or actual cause of death. Future studies should examine the association with cause-specific mortality. Third, our study is a single-center study. Whether our findings could be applied to other hospitals needed more research to confirm. Fourth, we cannot collect all confounders of the association between AKI and mortality, and thus other factors may affect the results. However, the E-value of the association suggested that unmeasured confounders were unlikely to explain the entirety of the effect.

This study found that AKI during hospitalization for aSAH was associated with long-term mortality, even for patients with renal recovery at the time of discharge (13, 27). The current KDIGO guidelines recommend patients should be followed by a nephrologist for more than 3 months after an AKI episode in order to estimate kidney recovery and/or progression to chronic kidney disease (28). The present study indicates that patients should have long-term post-discharge monitoring of renal function, regardless of renal recovery at the time of discharge. AKI may result in ongoing progressive renal damage beyond the acute episode, despite marginal increases in serum creatinine. This is an important insight for all physicians who care for patients with aSAH, and future studies will need to determine the optimal timing for follow-up of renal function and develop biomarkers of the transition from AKI to chronic kidney disease for patients with AKI during hospital.