Associations Between Predialysis Creatinine, SOFA Score, and Mortality in Acute Kidney Injury Patients Requiring Dialysis

Background: Creatinine is widely used to estimate renal function, but this is not practical in critical illness. Low creatinine has been associated with mortality in many clinical settings. However, the associations between predialysis creatinine level, Sepsis-related Organ Failure Assessment (SOFA) score, and mortality in acute kidney injury patients receiving dialysis therapy (AKI-D) has not been fully addressed. Methods: We extracted data for AKI-D patients in the eICU (n = 1,992) and MIMIC (n = 1,001) databases. We conducted a retrospective observational cohort study using the eICU dataset. The study cohort was divided into the high-creatine group and the low-creatinine group by the median value (4 mg/dL). The baseline patient information included demographic data, laboratory parameters, medications, and comorbid conditions. The independent association of creatinine level with mortality was examined using multivariate logistic regression analysis. We also carried out a sensitivity analysis using the MIMIC dataset. Results: In all 1,992 eICU participants, the 30-day intensive care unit mortality rate was 32.2%. The crude overall mortality rate in the low-creatinine group (43.7%) was signicantly higher than that in the high-creatinine group (20.6%; P < 0.001). In the fully adjusted models, the high-creatinine group was associated with a lower risk of all-cause mortality (odds ratio, 0.56; 95% condence interval, 0.42–0.75; P < 0.001) compared with the low-creatinine group. The nonrenal SOFA score was higher in the low-creatinine group. The results were consistent when the MIMIC dataset was used as an external validation dataset. Conclusions: AKI-D patients with a low predialysis creatinine value had a signicantly higher risk of mortality that might be associated with more organ dysfunctions. Moreover, SOFA and nonrenal SOFA scores did not suciently reect the severity of illness without considering the impact of the creatinine value in AKI-D patients.


Introduction
Acute kidney injury (AKI) is a common and signi cant problem in the intensive care unit (ICU), and about 25% of patients with AKI require renal replacement therapy (RRT) 1,2 . However, a high mortality rate of 30%-50% is noted 3,4 . Although many severity of illness scoring systems have been developed for mortality prediction, they either performed well in a single-center study without external validation or had limited results during external validation [5][6][7][8] . In terms of the Sepsis-related Organ Failure Assessment (SOFA) score, it doesn't have good discrimination in this patient group 5,9 . Creatinine is a metabolite of creatine and creatine phosphate, which are in the highest concentration in skeletal muscle, and is mainly eliminated via the kidney 10,11 . Therefore, serum creatinine is used to not only estimate renal function but also to re ect muscle mass. Low serum creatinine is also a marker of malnutrition 11 . However, it is also related to sex, age, diet, and uid status 12,13 . Because creatinine is affected by many factors, it usually overestimates renal function in critically ill patients 14 . Studies have shown that low creatinine was associated with high mortality in the ICU and an increased mechanical ventilation use rate and was also a risk marker of mortality in hemodialysis (HD) patients 11,12,15 . Only a few studies with small sample sizes have addressed mortality and the creatinine level in patients with AKI 5,8,13,16 .
For this reason, we conducted this retrospective study using two public datasets to explore the associations between predialysis creatinine, SOFA, and mortality among patients with AKI who were receiving dialysis (AKI-D) in the ICU.

Participants and Measurements
This retrospective, observational cohort study was performed using two publicly available ICU datasets, the MIMIC-III 29 and the eICU 30 Collaborative Research Database (eICU-CRD). The MIMIC-III database was released in 2016 by the Massachusetts Institute of Technology Laboratory for Computational Physiology (MIT-LCP) and contained data from a single tertiary care hospital (Beth Israel Deaconess Medical Center).
The eICU-CRD is a multicenter critical care database containing data from rural/nonacademic hospitals across the US and was made available in 2018 by Philips Healthcare with the help of researchers from MIT-LCP. There is no overlap in the patients included in these two databases.
We included adult patients 18 years of age or older who received RRT (either intermittent HD or continuous RRT [CRRT]) in the ICU for AKI. The AKI in this study was de ned according to the Kidney Disease Improving Global Outcomes clinical practice guidelines 31 and diagnosis codes.. We only used the creatinine criteria because of unreliable urine data in the retrospective databases. For patients who did not have more than one creatinine value to make a comparison, but who had RRT records, we included patients who were diagnosed as having AKI based on their ICD-9 diagnosis codes (Table S1). If a patient had been admitted to the ICU multiple times in one hospitalization course, data from the ICU admission that included the initial dialysis treatment was extracted for the study. Patients with a history of end-stage kidney disease who underwent chronic peritoneal dialysis or HD (Table S1) were excluded from the study. We also excluded patients who had chronic kidney disease (CKD) stage 4 and 5 based on ICD-9 codes (Table S1), because we were interested in patients who did not have advanced CKD at baseline. Patients with a history of any organ transplant were also excluded as they may have other confounding risk variables that affect mortality. We excluded patients who did not have complete records of vital signs and creatinine data one day before RRT start.
The variables collected consisted of demographics, medical history, mechanical ventilation usage, AKI stage, vital signs, laboratory test results, dialysis modality (intermittent HD or CRRT), and medications (diuretics and vasopressors, see Table S2). The time window of mechanical ventilation, vital signs, laboratory tests, and medications were recorded one day before RRT initiation. Past medical history was extracted from database records using ICD-9 codes (Table S1). Relevant past medical history included in the study were diabetes mellitus (DM), CKD, hypertension (HTN), congestive heart failure (CHF), liver cirrhosis (LC), and cancer. Vital signs in this study included the mean values of the following variables: shock index (SI), Glasgow Coma Scale (GCS), mean arterial pressure (MAP), respiratory rate (RR), and heart rate (HR). The mean SI was calculated by the formula: SI = mean HR/mean systolic blood pressure. For laboratory tests, we used the mean value of all variables recorded one day before the date of the rst dialysis therapy because some laboratory data values would have been in uenced by dialysis. We excluded the variables with >25% missing values, except for albumin level, because we thought that albumin was an important factor for mortality prediction. Table S3 reveals the percentages of missing data in the laboratory tests. Multiple imputation by chained equations (MICE) with ve imputed datasets was used to impute the missing values of the laboratory tests and vital signs and the results were pooled using the MICE package 32 .
We modi ed the codes from https://github.com/nus-mornin-lab/oxygenation_kc and https://github.com/MIT-LCP/mimic-code/tree/master/concepts/severityscores to calculate the SOFA score using variables collected one day before RRT start in the eICU and MIMIC datasets based on methods used in the original study 33 . For patients with missing variables, the The primary aims of the investigation were to assess whether the predialysis creatinine level was associated with 30-day mortality independent of other risk factors and to explore the association between the predialysis creatinine level and the SOFA score.

Statistical Analyses
The study cohort was strati ed into two groups according to the median creatinine value. Categorical variables were presented as counts, proportions, and frequencies; continuous variables were expressed as mean with standard deviation. Numeric variables of clinical characteristics between the two groups were compared using the Student's t test. The chi-square test was used to compare the differences of the categorical variables. Logistic regression was used to calculate odds ratios (ORs) and 95% con dence intervals (95% CIs) for the analyses of predictors of mortality. The comparison of survival status between the two groups was done using the Kaplan-Meier curve with signi cance levels determined by the log rank test. We implemented four models for the adjustments of the covariates: model 1, adjusted for age, sex, and ethnicity; model 2, adjusted for all variables in model 1 plus DM, HTN, CKD, malignancy, and LC; model 3, adjusted for all variables in model 2 plus GCS, HR, MAP, RR, SI, ICU days before dialysis, CRRT, diuretics, vasopressors, and mechanical ventilation; model 4, adjusted for all variables in model 3 plus laboratory parameters. We used the Kruskal-Wallis test to compare the SOFA score difference between the groups.
Each analysis was repeated using the MIMIC dataset to explore the heterogeneity using sensitivity analysis. Analyses were performed using R version 3.6.1 (R Foundation for Statistical Computing).

Baseline Characteristics of the Study Cohort
The cohort from the eICU database included 8,201 patients who required dialysis therapy. Of those patients, 1,992 patients met the inclusion and exclusion criteria for the study. The cohort from the MIMIC database included 3,357 patients who required dialysis therapy. Of those patients, 1,093 patients met the criteria for inclusion in the study (Figure 1).
The median creatinine level among all 1,992 eICU participants was 3.96 mg/dL. Their mean age was 62.8 ± 14.7 years, and 1,167 (58.6%) patients were men. The overall mortality rate was 32.2%. The cohort was divided into low-and high-creatinine groups according to the median creatinine value (4 mg/dL); the patient characteristics are presented in Table 1. The high-creatinine group had a lower mean age and a greater proportion of patients who were men and patients who were black than the low-creatinine group.
The high-creatinine group had higher mean levels of blood urea nitrogen, hemoglobin, albumin, anion gap, platelet count, and potassium than the low-creatinine group. Comorbid conditions of malignancy and LC were more common in the low-creatinine group, whereas the prevalence of DM, HTN, and CKD was higher in the high-creatinine group. The high-creatinine group had a lower proportion of prescriptions for diuretics, vasopressors, CRRT, and mechanical ventilation than the low-creatinine group. In terms of vital signs, the low-creatinine group had lower mean values for GCS and MAP but higher mean values for RR and HR.

Creatinine and All-Cause Mortality
The crude mortality rate was 20.6% (n = 204) for the high-creatinine group and 43.7% (n = 437) for the low-creatinine group (P < 0.001). The Kaplan-Meier analysis revealed that the patient survival was signi cantly better for the high-creatinine group than for the low-creatinine group (P < 0.0001) (Figure 2).
The unadjusted and adjusted ORs are presented in Table 2. Compared with the low-creatinine group, the OR for the high-creatinine group was 0.33 (95% CI: 0.27 to 0.41) for all-cause mortality in the unadjusted model. In the fully adjusted model (model 4), the risk of mortality in the high-creatinine group was 44% lower (OR, 0.56; 95% CI: 0.42 to 0.75). Although CKD was signi cantly associated with mortality in the unadjusted analysis, it was no longer signi cant in the fully adjusted model. Figure 3 shows that there was a signi cant interaction between creatinine and age. That is, the OR was 0.46 for 30-day mortality in subjects aged ≥65 in the high-creatinine group, whereas the OR was 0.77 in subjects with a younger age (P for the interaction 0.04). We also noted a signi cant interaction (P = 0.02) between creatinine and albumin with regard to 30-day mortality. In patients with CKD, high creatinine was associated with an OR of 1.34 (P for the interaction 0.03).

Discussion
In this study of patients with AKI-D, we identi ed that a low creatinine level was independently associated with all-cause mortality. A high creatinine (≥4 mg/dL) was associated with a 44% lower risk of mortality in these patients. The association between high creatinine and lower 30-day mortality was constant in the ICU. Moreover, most of the subgroup analyses further con rmed our results. The SOFA and nonrenal SOFA scores in the low-creatinine group were higher than those in the high-creatinine group, indicating that AKI-D patients with low predialysis creatinine values have more organ dysfunctions.
A low creatinine level can relate to an increase in excretion or a decrease in generation. Real kidney function improvement was uncommon in critically ill patients 17 , whereas those patients usually have more complicated underlying problems that affect creatinine generation, such as sepsis 18 , uid accumulation 13 , poor nutrition status and low muscle mass 11,19 , liver failure, and older age 19 . The relationship between low creatinine at the start of RRT and mortality in AKI-D patients was demonstrated in previous reports 5,8 . Our results showed that the low-creatinine group had a higher mean age, a lower mean albumin value, and a higher proportion of patients with LC, as well as increased mortality. In addition, in the interaction analysis, we found that lower creatinine increased the mortality risk in all subgroups, except for the CKD subgroup. The high creatinine group seemed to have better survival in the older age (≥65 years) and high albumin (≥3.5 g/dL) subgroups. Implicitly, nutritional status is not the only reason for the relation of mortality to low creatinine, as oliguria and uid accumulation may also be reasons 8,13 . Oliguric AKI was an independent factor for mortality and would lead to low creatinine caused by uid overload (FO) 20,21 . FO can result in tissue edema and organ dysfunctions and increased risk of mortality 13,20,22,23 . The possible reason that the effect of lower creatinine was not prominent in CKD patients may be that elevated levels of creatinine were associated with increased risk of myocardial infarction, cancer, and infection [24][25][26] .
In addition to causing organ dysfunctions, oliguria and FO are also the results of organ failure. In other words, there may be other causes than oliguria and FO in relation to low creatinine associated with high mortality. The hypothetical reason is that the more severely and rapidly critical illnesses develop, the lower the predialysis creatinine is in AKI-D patients. Creatinine needs time to achieve a steady state in AKI patients 27 . However, for severe and critically ill patients, there would not be enough time for the creatinine to reach a steady state when RRT starts. In our study, the low-creatinine group had higher SOFA and nonrenal SOFA scores. Compared with the high-creatinine group, the nonrenal SOFA score was more representative than the SOFA score in the low-creatinine group. Therefore, AKI patients with low predialysis creatinine but who require dialysis therapy implicitly have more organ dysfunctions. Because it does not consider the effect of creatinine, the SOFA score does not perform well for mortality prediction in AKI-D patients 5,9 . The nonrenal SOFA score is slightly better at mortality prediction than the SOFA score, but it still does not take into account the effect of a low creatinine level. The impact of creatinine should be considered in scoring systems used for AKI-D patients, as with the Acute Physiology And Chronic Health Evaluation (APACHE) score 28 , the HEpatic failure, LactatE, NorepInephrine, medical Condition, and Creatinine (HELENICC) score 8 , and the ATN score 5 .

Strengths and Limitations
Our study has several advantages. The rst is that the eICU and MIMIC datasets are from distinct regions and hospitals across the US. Also, our ndings were consistent in these two datasets. However, there were several limitations to our study. First, as previously mentioned , some data were missing and the urine output might not be reliable in the retrospective databases. Second, other potential confounding data, like FO and body mass, were not available in these datasets. Third, the association between CKD and predialysis creatinine in AKI-D patients needs further study. Fourth, the underlying mechanism of the creatinine impact on mortality remains unknown, and further studies are warranted.

Conclusion
The low-creatinine group had a signi cantly higher risk of mortality compared with that of the highcreatinine group. The higher mortality risk of low creatinine was consistent across patient subgroups, except for CKD. Moreover, we should be more cautious regarding AKI-D patients with lower creatinine because implicitly they had more organ dysfunctions. The mechanism of how creatinine impacts mortality remains unclear and needs further studies for clari cation.   Abbreviations: OR, odds ration; 95% CI, 95% con dence interval; ICU, intensive care unit; CRRT, continuous renal replacement therapy. Figure 1 Participant ow diagram. n is patient unit encounter. Abbreviations: RRT, renal replacement therapy; ESKD, end-stage kidney disease; CKD, chronic kidney disease; ICU, intensive care unit; AKI, acute kidney injury.

Figure 2
Kaplan-Meier curve of mortality according to creatinine category. The high creatinine (Cr ≥ 4 mg/dL) group was associated with better survival than the low creatinine (Cr < 4 mg/dL) group (log rank test, P < 0.0001).

Figure 3
Subgroup analysis of associations of high creatinine (Cr ≥ 4 mg/dL) group with the mortality in patients with acute kidney injury undergoing dialysis. Abbreviations: OR, odds ratio; 95% CI, 95% con dence Page 16/17 interval; CKD, chronic kidney disease. Sepsis-related Organ Failure Assessment (SOFA) and nonrenal SOFA scores in patients with low creatinine (Cr < 4 mg/dL) were higher than scores in patients with high creatinine (Cr ≥ 4 mg/dL). Boxplot showing the SOFA and nonrenal SOFA score between the groups. ****P value < 0.0001 Supplementary Files