Introduction

Forty percent of adults with sickle cell disease (SCD) develop chronic kidney disease (CKD) and 15–30 % of deaths in patients with SCD are related to kidney disease [1–5]. Children with SCD develop hyposthenuria, hyperfiltration, and microalbuminuria [6–10]. Consequently, the early identification of risk factors and early initiation of therapies to mitigate disease progression could have significant long-term impact. A growing body of evidence shows that acute kidney injury (AKI) contributes to CKD in other patient populations [11]. Thus, an understanding of the prevalence of AKI during SCD events is paramount to improved care.

While the role of AKI in the development of CKD has become illuminated in other diseases, little research has focused on elucidating the role of AKI in SCD nephropathy [12–15]. To our knowledge, only one study to date has evaluated the incidence of AKI during SCD crisis, and based on the results, the authors suggested that 7–13 % of adult SCD patients hospitalized with acute chest syndrome (ACS) develop AKI [16]. However, this retrospective study is limited by the lack of an appropriate evaluation of serum creatinine (SCr) values, as the authors only looked at the day 1 SCr level (compared to prior baseline) to diagnose AKI [16]. In addition, patients in this adult cohort [16], as compared to a pediatric cohort, may have had underlying CKD which could influence the development of AKI during the acute setting. There are several potential risk factors for developing AKI during ACS. SCD is characterized by acute hemolytic/anemic events, ischemia/reperfusion episodes, hypoxia, and inflammation that may worsen during ACS. The risk of AKI may also be exacerbated by use of nephrotoxic agents (vancomycin and ketorolac) and concern for dehydration (poor oral intake during crisis and hyposthenuria).

To better understand the link between admission for ACS and AKI, we conducted a 6-year retrospective cohort study of pediatric patients admitted with ACS to our medical institution to determine the prevalence of AKI. We hypothesized that AKI is prevalent during ACS events and leads to prolonged hospital length of stay (LOS).

Methods

We conducted an institutional review board-approved 6-year retrospective cohort study of all pediatric SCD patients admitted to Children’s of Alabama with an ICD-9 code of ACS (517.3) as a primary or secondary diagnosis with the aims to determine the prevalence of AKI, identify potential risk factors for AKI at admission, and evaluate the association between AKI and hospital LOS. We excluded patients with an ICD-9 code of ACS (517.3) who did not have an identifiable pulmonary infiltrate on chest X-ray during their hospitalization or who were hospitalized for <24 h. AKI was defined by the Kidney Disease Improving Global Outcomes (KDIGO) definition of either a ≥0.3 mg/dL or 50 % increase in SCr from baseline [17]. Per institutional standard of care, patients with fever and pulmonary infiltrates (ACS) are treated with vancomycin, azithromycin, and ceftriaxone in the emergency room and can receive non-steroidal inflammatory medications (ibuprofen or ketorolac) as needed for pain or fevers [18]. After admission of the patient for ACS, individual attending physicians can determine whether they will discontinue vancomycin based on the initial clinical evaluation, wait for the 48-h cultures to be negative, or continue vancomycin until the patient is clinically stable. The type (simple or exchange) and timing of transfusion to the SCD patient is determined by the preferences of attending physician rather than by institutional protocol.

We conducted a chart review of patients diagnosed with ACS to determine the sickle cell genotype, age of patient, current SCD modifying therapy, results of the most recent outpatient (within 1 year of admission) complete blood count [baseline: hemoglobin (Hb) value, white blood cell count (WBC), and platelet count], and SCr level. We also collected the emergency room and daily inpatient complete blood counts and SCr values for all patients during their hospitalization. We calculated the difference in Hb and SCr from baseline (last Hb or SCr measurement prior to admission and within 1 year) to their admission for Hb and SCr values in ACS. We identified any transfusion intervention during the hospitalization period, transfer to pediatric intensive care unit (ICU), need for enhanced respiratory support [biphasic positive airway pressure (BIPAP) or mechanical ventilation] and length of hospital stay. Two categorical variables were created to reflect the clinical severity for patients based on SCD genotype: severe SCD (HbSS, SB0 thalassemia, and SD) and mild SCD (HbSC and SB + thalassemia). Among patients with multiple ACS admissions, we included only data from their most recent hospital admission in the analysis.

Characteristics of patients by AKI status were summarized using descriptive statistics. Comparisons of these characteristics were performed using the t test for continuous variables and Fisher’s exact test for categorical variables. To determine variables associated with AKI, we performed nominal logistic regression with baseline, admission, and change in the following variables: Hb, WBC and platelets (as explanatory variables). Similarly, we performed nominal logistic regression for type of transfusion support using laboratory variables, namely, Hb (baseline, admission, and change in Hb), WBC, and platelets. Kaplan–Meier curves and log rank tests were used to separately investigate the association of length of hospital stay with development of AKI, need for exchange transfusion, and need for increase in respiratory support. As intubation by itself can increase LOS, we stratified the association of LOS and AKI by need for increased respiratory support (BIPAP or intubation). All analyses were conducted using JMP Pro version 10 software (SAS Inc., Cary, NC).

Results

We identified 236 episodes with ACS codes in 163 patients admitted for hospitalization during a 6-year period. As only the most recent episode of ACS was included in the analysis, we excluded 73 episodes of ACS as they were repeat episodes of ACS in the same patient. Of the 163 unique patients with ACS, we excluded 14 patients due to an inability to determine AKI during admission because these patients did not have baseline SCr data for the preceding year. Among the 149 eligible patients, the mean SCr at baseline was 0.4 mg/dL [standard deviation (SD) 0.1] and the mean maximum serum creatinine during hospitalization was 0.5 mg/dL (SD 0.1). The mean age of patients admitted with ACS was 9.5 years (SD 4.8), and 64 % were male. Genotypes included 119 patients (80 %) with HbSS, seven patients (5 %) with HbSB0 thalassemia, eight patients (5 %) with HbSB + thalassemia, 14 patients (9 %) with HbSC, and one patient (1 %) with HbSD. Four patients (2 %) were on chronic transfusion at the time of ACS, 50 patients (32 %) were on hydroxyurea, and 95 patients (66 %) received no disease-modifying therapy. Twenty eight patients were admitted or transferred to the ICU of which 22 required >1 day in the ICU. Eleven patients (7 %) required mechanical ventilation and two were escalated to BIPAP (1 %) but not intubated. Thirty-three patients (22 %) required red cell exchange therapy, 102 patients (69 %) required simple transfusion, and 14 patients (9 %) were not transfused.

Table 1 summarizes the characteristics of the patients by AKI status as defined by KDIGO. AKI was identified in 12 of 149 (8 %) episodes of ACS over the 6-year study period. Compared to children without AKI, those patients with AKI had a larger drop in hemoglobin (2.7 vs. 1.4 g/dL; p = 0.003), lower Hb at admission (6.4 vs. 7.5 g/dL; p = 0.03), and higher WBC count at admission (33.1 vs. 19.8 × 103/L; p < 0.0001). The groups were similar for age, gender, and platelet count. We identified ten episodes of AKI in 127 cases of ACS for patients with severe SCD phenotypes and two episodes of AKI in 22 cases of ACS for patients with milder SCD phenotypes. No significant differences in the development of AKI were identified by genotype (p = 0.8). Among the 149 defined cases, two of the 50 patient cases (4 %) on hydroxyurea developed AKI, none of the four (0 %) on transfusion developed AKI, and ten of the 95 (11 %) receiving no disease-modifying therapy developed AKI. No differences in the development of AKI were identified by current SCD therapy (p = 0.3).

Table 1 Baseline characteristics of patients admitted for acute chest syndrom
Full size table

Nominal logistic regression revealed that compared to baseline values, a drop in Hb and WBC count at admission and in (g/dL; ×103/L) were associated with AKI (p values = 0.01 and 0.0007, respectively). The estimated odds ratios were 1.81 [95 % confidence interval (CI) 1.17–2.92] for change in Hb (g/dL) at admission and 1.10 (95 % CI 1.05–1.18) for WBC count (×103/L) at admission. The majority of patients developed AKI within 2 days of admission, with resolution at a median of 2 (range 1–7) days. The median hospitalization LOS for all patients was 4 (range 2–35) days, and the median and mean hospitalization LOS was significantly longer among patients who developed AKI (p < 0.0001) (Fig. 1a). In addition, the median and mean hospitalization LOS was significantly longer among patients who required exchange transfusion (p < 0.0001) (Fig. 1b) and in patients who required increased respiratory support (either intubation or BIPAP; p < 0.0001) (Fig. 1c). As increased respiratory support (intubation) was an expected strong predictor of LOS, additional analysis on AKI and LOS was performed with stratification by need for increased respiratory support. Among those who were intubated or required BIPAP, the development of AKI was still significantly associated with LOS (p = 0.02). (Fig. 1d)

Fig. 1
figure 1

Length of hospital stay for patients admitted with acute chest syndrome. a. Total length of hospital stay (LOS) by acute kidney injury (AKI), b total LOS by type of transfusion [no transfusion (None), simple red blood cell (RBC) transfusion (Simple), RBC pheresis (Pheresis)], c total LOS by type of respiratory support [intubation or biphasic positive airway pressure (BIPAP)] vs. no increase in respiratory support, d total LOS by acute kidney injury (AKI) among patients who required intubation or BIPAP. CI Confidence interval

Full size image

Discussion

This research shows that a subset of pediatric patients with SCD who were admitted to our medical institution for ACS developed AKI during hospitalization. These results suggest that clinicians should monitor for this complication with daily SCr monitoring during admission for ACS as it could alter management strategies. At our institution, nephrotoxic agents are often administered to patients with severe ACS and fever (vancomycin and ibuprofen) or pain (ketorolac) that could potentiate kidney injury [18]. As protection against pneumococcus from vaccination improves and resistance to third-generation cephalosporin stays limited, institutional policies and individual clinicians should consider the option of withholding vancomycin or of discontinuing therapy with this agent early when treating presumed resistant pneumococcus as a rare pathogen for ACS and, alternatively, consider less aggressive strategies to manage fever with ibuprofen or ketorolac in patients that develop AKI [19].

One variable identified as a strong risk factor for developing AKI was an acute drop in Hb. For every 1 unit drop in Hb from baseline to admission, the odds of AKI is estimated to increase by about 81 % While prior research has identified severe anemia at baseline as a risk factor for overall disease severity in pediatric SCD, clinicians should also consider an acute drop in Hb from baseline as a risk factor for kidney injury during hospitalization [20, 21]. Patients with an acute and sustained drop in Hb during hospitalization are at risk for complicated ICU admissions, stroke, and ischemic brain injury, and now AKI [22–26]. Potential mechanisms linking an acute anemic event and AKI include renal ischemia from reduced oxygen carriage or acute increases in plasma free heme/Hb from hemolysis, producing direct kidney injury [27, 28]. Future prospective studies should evaluate the relationship between anemia and kidney injury in patients with SCD.

Prior studies of hydroxyurea in pediatrics have suggested a protective effect on progressive sickle cell nephropathy in children with SCD, yet the role of hydroxyurea in preventing AKI has not been explored [6, 29]. While hydroxyurea can prevent admissions for ACS, we were unable to demonstrate that patients prescribed hydroxyurea had statistically lower rates of AKI (2 of 50 cases of AKI) as compared to patients not on disease-modifying therapy (10 of 95 cases of AKI) [8]. One prior study evaluating the role of hydroxyurea in preventing kidney injury suggests that at steady state, hydroxyurea reduced biomarkers of inflammation and oxidative stress in the kidney [30]. The lack of a statistical advantage for hydroxyurea against AKI may reflect an underpowered retrospective study, poor adherence to hydroxyurea among those patients admitted with ACS, issues inherent in defining AKI by SCr in SCD, or lack of efficacy of hydroxyurea to prevent AKI.

An additional finding suggests that the development of AKI may contribute to possible prolonged hospitalization. The length of hospitalization for ACS frequently results from the time required to discontinue supplemental oxygenation. Recent research in bi-directional organ cross-talk during critical illness offers insight into the association between AKI and acute lung injury. Hypoxemia and acute lung injury due to respiratory failure lead to a decrease in renal blood flow and is a risk factor for AKI [31, 32]. Understanding the pathophysiology of ACS in the context of AKI could explain part of the severity of this disease and the prolonged need for respiratory support.

The strengths of this study are the use of the most contemporary definition of AKI, clear SCD phenotypes, including standard therapy, and the availability of baseline SCr values for most of the children. Despite these strengths, we acknowledge several limitations to this study by its nature of being a retrospective study. As some patients in this study were seen at satellite sickle cell clinics, a baseline SCr was not performed on all patients admitted to our hospital. Similarly, all laboratory data and clinical records were reviewed from Children’s of Alabama electronic medical records; therefore, patients admitted and transferred from an outside emergency room or hospital may have had additional SCr values that we did not evaluate. As some patients in this study did not have daily SCr measurements, it is possible that a few cases of AKI may not have been appropriately captured. Strict urine output was not calculated in most patients, and our definition only uses SCr-based criteria. An inherent concern in AKI reporting in SCD is that SCr may not accurately reflect the estimated glomerular filtration rate. Despite the concern that SCD patients may have a lower SCr value, this study utilized the contemporary definition of AKI which incorporates a change in SCr from baseline. Large multi-center studies will be needed to determine if this current definition of AKI predicts short- and long-term outcomes. . A limitation affecting the generalizability of these findings to other institutions is the inclusion of vancomycin at diagnosis of ACS. The only adult study for which data are available identified a similar incidence of AKI (7 % in moderate ACS and 14 % in severe ACS), but the authors did not report their antibiotic regimen for ACS. Finally, as we evaluated the most recent admission for patients with ACS in a single center, we identified 12 cases of AKI among 149 unique patients admitted over a 6-year period. As not all patients with HbSS and SB0 thalassemia develop ACS, this study is limited by a small sample size of patients that developed AKI. Future studies that could better delineate risk factors and outcomes of AKI require a larger, multi-center study.

In conclusion, AKI occurs during hospitalization for ACS, and the role of AKI in progression to SCD nephropathy needs further evaluation. We suggest that clinicians monitor for the risk factors of AKI that we identified, namely, acute decline in Hb concentration and leukocytosis, and consider modifying empiric therapy for ACS by limiting nephrotoxic agents and optimizing fluid balance. We are currently conducting a long-term prospective study for all SCD patients admitted with ACS to better define AKI and evaluate potential mechanistic links between ACS, AKI, and progression to CKD. With early detection of AKI, better methods for identifying AKI, and modifications to present-day therapy for patients with AKI, we hope to improve outcomes in sickle cell nephropathy.