Emergency department intravenous fluid resuscitation and renal outcomes among adults with sickle cell disease

Acidosis and increased tonicity in plasma can mediate pathologic changes in the membrane and cytoplasm of sickle red blood cells (sRBCs). These changes contribute to intravascular hemolysis, endothelial damage, and endothelial adhesion with propensity to microvascular occlusion, resulting in vaso-occlusive episodes (VOE) and end organ damage.¹ Kidney dysfunction, in particular, is common among adults with sickle cell disease (SCD) and is a major contributor to early mortality.^{1, 2} While adults with SCD seek emergency department (ED) care for VOE and other complications that may lead to hospitalization, selecting the most appropriate interventions, such as which crystalloid for hydration, remains a challenge and is still being studied.³ How these interventions impact end-organ injury also remain unknown. Evidence-based guidelines for optimal treatment of SCD in the ED do not recommend a specific fluid type,⁴ but recent pre-clinical models suggest increased extracellular fluid tonicity (i.e., higher sodium and chloride) can increase sRBC stiffness and endothelial adhesion, and subsequently higher risk of VOE under physiologic conditions.⁵ Further, normal saline (NS), which is the most commonly administered intravenous (IV) fluid in the ED, can cause hyperchloremic metabolic acidosis which may contribute to negative outcomes in patients with SCD.^{6, 7}

The seminal Saline Against Lactated Ringer's or Plasma-Lyte in the Emergency Department (SALT-ED) trial was a single-center, pragmatic, unblinded, multiple-crossover prospective randomized controlled trial (RCT) that enrolled non-critically ill adult patients treated in the Vanderbilt University Medical Center ED and randomized patients to receive NS or balanced crystalloids (BC), primarily lactated Ringer's (LR) based on calendar month (Figure S1).⁷ NS increased risk of major adverse kidney events and death at 30 days (i.e., MAKE30; Data S1) and in-hospital metabolic acidosis changes compared to more physiologic and balanced IV fluids like LR, which is hypotonic relative to plasma and has a more physiologic pH and osmolarity compared to NS (Table S1). In this intention-to-treat subgroup analysis of SCD patients enrolled in the study, our primary objective was to evaluate the same variables as SALT-ED specifically at the index visit (first visit). The primary aim was to determine if there were differences in length of stay (LoS) between those patients with SCD receiving NS versus BC, and secondary aims included evaluating the changes in kidney-related outcomes and suggestion of in-hospital metabolic acidosis by evaluating electrolyte status. We hypothesized that NS exposure would worsen all of these in those patients enrolled who had SCD. Exploratory aims included a per-protocol analysis of index visits and reviewing all ED visits among those with SCD and comparing all results to the overall SALT-ED cohort.

Comparisons between the overall SALT-ED cohort and those with SCD are outlined in Table S2. There were 247 ED visits among 97 unique adults with SCD during the trial period (~2.5 visits/patient), with N = 126 (51%) visits allocated to the BC group and N = 121 (49%) in the NS group (Table S3). Most documented chief complaints in the ED were VOE-related (Table S4). Patients with SCD were younger, more likely to identify as female and Black, and have a lower baseline serum creatinine (SCr) compared to the overall SALT-ED population (Tables S5 and S6). Total ED crystalloid volume and adherence to randomized IV fluid assignment were similar between all patients and those with SCD. Strikingly, patients with SCD enrolled in the SALT-ED trial had a 3.5 times higher odds of achieving a MAKE30 composite endpoint than the entire cohort. While no patients with SCD died during the study, there were 200 patient deaths in the overall study.

At the index visits, N = 58 adults with SCD were randomized to BC and N = 39 to NS (Table 1). Only one patient in the BC arm received Plasma-Lyte, and thus, we use the term LR for comparison. Groups were similar based on sex, race, total ED crystalloid volume, baseline and initial ED renal function (including SCr and eGFR), and baseline and initial serum electrolytes in the ED. Admission status and ICU transfers were similar between the two groups. No patients with SCD were on renal replacement therapy (RRT) for end-stage renal disease (ESRD). The 100% adherence to IV fluid assignment was lower in the LR group, suggesting a preference for NS in patients with SCD. Of note, prior to the SALT-ED trial, standard of care ED IV fluid at Vanderbilt was NS. Patients had evidence of hyperfiltration based on eGFR. While not statistically significant, compared to those with SCD randomized to LR, those patients who received NS had increased mean LoS (4.8 vs. 5.3 days) and more major adverse kidney events (i.e., MAKE30–17.9% vs. 12.1%). MAKE30 events were driven by the last SCr doubling from baseline and maximum in-hospital SCr at least double baseline SCr (20.5% NS vs. 13.8% LR). Maximum in-hospital SCr levels were significantly higher in patients randomized to NS in the ED compared to those receiving LR (p = .035). While last in-hospital SCr was higher in the NS group, it did not reach statistical significance. eGFR between the two groups at in-hospital maximum SCr and last SCr were similar. As was seen in the SALT-ED trial, mean electrolyte comparisons and post-hoc analyses suggested a trend toward increased in-hospital plasma BUN and sodium, as well as a trend toward metabolic acidosis in those receiving NS. Results were similar in the per-protocol analysis of the index visit and of the full data set (Tables S6 and S7).

To investigate the relationship of maximum in-hospital SCr with randomized IV fluid in the ED during the index visit, relationship with other covariates and potential confounders were assessed. The variables age, sex, race, study month, study day, randomized ED fluid type (NS vs. LR), total crystalloids in ED, baseline SCr, and ED SCr were first assessed using univariate linear regression estimates with maximum in-hospital SCr as the outcome. Multivariate linear regression was then used and the model was iteratively reduced until the final model was derived (Table S8). The final model found that age, randomized ED IV fluid type, and ED SCr levels had the largest effect on maximum in-hospital SCr results. For patients receiving LR in the ED, their maximum in-hospital SCr was likely to be 0.086 mg/dL (95% CI 0.027–0.15; p = .005) less than those participants who were randomized to receive NS in the ED (Table 1).

Fluid replacement therapy in the ED for adults with SCD and its impact on clinical outcomes remains woefully understudied. Improving care and mitigating negative side effects of the care delivered in the ED are often cited as an area of great need by adult patients with SCD and their clinical practitioners.^{3, 4, 6} In a retrospective pediatric cohort study among patients 3–21 years old, we found IV fluid use varied among ED providers during VOE management but that NS was used 65% of the time and increased the likelihood of hospitalization, more time spent in the ED, and worse pain control.⁶ This SALT-ED sub-analysis is the first investigation of prospectively recruited adult patients with SCD randomized to NS or LR in the ED. The results from this large RCT sub-analysis further support recent evidence suggesting NS is not an ideal resuscitation IV fluid for adults or children with SCD presenting to the ED with pain and other complications. Further, a recent target trial emulation analysis comparing in-hospital use of NS to LR found that use of NS may increase hospital LoS during VOE.⁸ Of note, type of IV fluids administered in the ED was not included in the analysis.

In our analysis, patients with SCD had an approximate 3.5 times higher odds of suffering an acute kidney injury (AKI) event during hospitalization as compared to the entire SALT-ED cohort. This further supports other evidence that this patient population deserves special attention during ED care. In addition to in-hospital development of AKI, SCD patients may present with AKI, defined by a rise in serum SCr, or subclinical AKI, defined by a rise in other biomarkers of kidney injury. Further, the development of AKI is a well-established model for the progression of CKD in other populations as well as in SCD. Pre-clinical models simulating sRBC biomechanical interactions with the microvasculature have also demonstrated that low-sodium, hypotonic fluids can hydrate sRBCs and reduce endothelial adhesion, whereas fluids high in sodium chloride may do the opposite. This suggests IV fluids with higher sodium and chloride content may uniquely increase risk of vaso-occlusion under physiologic conditions.⁵

Results from this sub-analysis further support prior evidence suggesting NS is not an ideal resuscitation IV fluid for patients with SCD presenting to the ED with pain and other complications and that less acidotic, more balanced crystalloids like LR may be a better option for IV fluid resuscitation among adults with SCD presenting to the ED. Further research is needed to confirm these findings. To adequately determine the optimal IV fluid regimen for adults with SCD presenting to the ED with VOE or other complications, a well-controlled and powered RCT is needed.

M. A. C. designed the study, analyzed the data, and wrote the paper. W. R. assisted in statistical analysis. J. L., W. R., P. T., and V. D. contributed to the design of the study and analysis and critically reviewed and edited the manuscript.

M. A. C. is an employee at Cogent Biosciences, which did not sponsor, provide funding, or guide this work in any way. J. L. is a consultant for Agios and Novartis. P. T. is a consultant for CSL Behring. V. D. is a consultant for Novartis, Travere, Bayer, Forma Therapeutics (acquired by Novo Nordisk), Merck, Amgen, and iCell Gene Therapeutics.

This study was approved by the local ethics committees, and waiver of consent was approved (see Data S1).