Background: Sickle cell disease (SCD) affects an estimated 300,000 infants born worldwide each year and has a total estimated prevalence of 100,000 in the United States alone (Piel 2017). Within this population, over 90% of healthcare encounters are due to painful vaso-occlusive episodes (VOE) (Kidwell 2021). VOE is often complicated by hypovolemia, making fluid administration a common intervention (Lovett 2017).

Lactated Ringer (LR) and 0.9% saline (NS) solutions are both isotonic crystalloids widely used for intravenous fluid resuscitation across many contexts and disease states (Myburgh 2013). NS contains a supra-physiologic concentration of chloride, which evidence suggests can lead to hyperchloremic metabolic acidosis and renal dysfunction when given in large volumes (Myburgh 2013). There has been ongoing study about the clinical significance of these effects, as well as whether and for whom LR might provide meaningfully better clinical outcomes (Myburgh 2013, Self 2018, Semler 2018, Self 2020).

Both these physiologic principles and recent preclinical evidence have called into question whether NS is the best resuscitation fluid for patients experiencing VOE due to concerns its tonicity and tendency to induce acidosis may promote sickling (Carden 2017a, Carden 2017b, Alwang 2024). To assess the clinical impact and relevance of these concerns, Alwang et al. performed the retrospective cohort study below (Alwang 2024).

Paper: Alwang AK, Law AC, Klings ES, Cohen RT, Bosch NA. Lactated Ringer vs Normal Saline Solution During Sickle Cell Vaso-Occlusive Episodes. JAMA Intern Med. 2024;184(11):1365-1372. PMID: 39250114

Clinical Question: In adult patients admitted for a vaso-occlusive episode, does early fluid resuscitation with Lactated Ringer solution compared to 0.9% saline affect hospital-free days by day 30?

What They Did:

• A multicenter, retrospective cohort study using data from the Premier PINC AI database, which includes enhanced claims and billing data, indexed by hospital day, for approximately 25% of US hospitalizations
• Applied a target trial emulation framework to simulate a randomized controlled trial (RCT) — which the authors deemed impractical due to the required enrollment size — comparing early LR and NS resuscitation for sickle cell patients admitted for a VOE
• Eligible hospital admission encounters were analyzed using targeted maximum likelihood estimation (TMLE), a machine learning methodology, to estimate the average treatment effects of LR vs. NS on the outcomes of interest. Cross-validation was performed.
• Analysis was controlled using covariables for demographics, measures of comorbidity, acute organ dysfunction on admission, hydroxyurea use on hospital day 1 (HD1), level of care on HD1, discharge year, hemoglobin SS genotype, and hospital/institution.
• The heterogeneity of treatment effects was also assessed among different hemoglobin SS genotypes, levels of care, volumes of fluid resuscitation, and diagnosis of acute kidney dysfunction on admission.

Population:

Inclusion Criteria:

• Adults (age 18 or older)
• Admitted to a hospital in the PINC AI database (1/1/2016 – 9/30/2022)
• Primary diagnosis of sickle cell disease with a vaso-occlusive episode, as determined by ICD-10 codes designated as present on admission
• Received LR or NS (based on charge codes for 500-1,000mL bags) on HD1

Exclusion Criteria:

• Received both LR and NS on HD1
• Patients transferred from outside hospitals
• Received renal replacement therapy on HD1
• Encounters from hospitals with less than 25 total encounters, due to concerns about model convergence

Intervention: Fluid resuscitation on hospital day 1 with lactated Ringer

Comparator: Fluid resuscitation on hospital day 1 with normal saline

Outcomes:

Primary Outcome: Hospital-free days by day 30 (HFD)

Secondary Outcomes:

• Intravenous opioid-free days by day 30
• Blood transfusion-free days by day 30
• Organ support-free days by day 30 (defined by kidney replacement therapy, invasive mechanical ventilation, or vasopressor use)
• Hospital mortality by day 30
• Intravenous diuretic use by day 30
• 30-day readmission
• Hospital length-of-stay by day 30

Results:

Enrollment:

• 127,262 encounters from the PINC AI dataset had primary diagnoses on admission of SCD and VOE
• 61,625 encounters received intravenous fluid resuscitation with LR or NS on HD1
• 52,079 encounters were eligible for inclusion
• Of these eligible encounters, 3,495 received LR, and 52,079 received NS

Outcomes:

Heterogeneity Analyses:

• The above improvement in hospital-free days was not observed for patients receiving less than 2L of fluid resuscitation on HD1 or for patients without the HgbSS genotype.

Strengths:

• The study addressed a common clinical scenario in which both compared interventions are widely available and, therefore, easy to implement into clinical practice.
• The enrollment was multicenter, large, and adequately powered for clinically meaningful changes in outcome.
• Supportive preclinical data, the adoption of a target trial emulation framework, and the heterogeneity analysis suggesting a dose-response relationship all help support a causal inference despite its cohort design
• Followed STROBE reporting guidelines, allowing for transparency and reproducibility of results
• PINC AI database is robust and has less than 0.01% missing data

Limitations:

• Generalizability: The study does not include pre-hospitalization data, notably excluding care in the emergency department (ED) limiting its generalizability to the ED setting.
  • This study only analyzed the impact of early fluid resuscitation occurring within 24 hours of admission; it is therefore difficult to generalize to patient care occurring after this first day. 
  • The enrolled population was young (median age 30; IGR 25-37); while representative of the sickle cell population, which is generally younger, the data may not be generalizable to older patients.
  • This study only included data from the United States; it may not be generalizable to other countries where clinical practice and patient populations differ.
• Outcomes: Primary and secondary outcomes were objective and patient-centered but not optimally measured.
  •  Reliance on a billing dataset, instead of EMR or prospective data, likely affected the quality of outcome measurement.
  • Hospital-free days did not account for hospital days during readmissions.
  • IV opioid-free days were assessed but lacked data on dosage or frequency of opioid use; morphine milligram equivalents (MME) would provide better insight.
• Cohort design: This was not an RCT, nor were interventions blinded. 
  • Conducting an appropriately powered RCT for the demonstrated effect size would require a sample size greater than 3,500, much larger than that of prior trials for SCD.
• Reliance on ICD-10 and billing codes: Because this study was performed retrospectively using a national database, data on eligibility, types and quantity of fluid received, and patient outcomes were all assessed using ICD-10 and billing codes. This introduces the risk that data may be less accurate than with prospective enrollment, access to medication administration records, etc.
• Focus on initial resuscitation: This study only analyzed the impact of early fluid resuscitation occurring within 24 hours of admission; it is therefore difficult to generalize to patient care occurring after this first day.
• Volume of initial Resuscitation: As acknowledged by the authors, this study does not inform the best volume of early fluid resuscitation to provide, nor should the heterogeneity analysis be interpreted to suggest that a higher volume of fluid resuscitation is better.

Discussion:

Target Trial Emulation, TMLE, and Confounders: As both methodologies are likely new to many readers, they are worth discussing briefly. Target trial emulation is a methodologic framework to improve observational studies that begins by creating a hypothetical protocol for the ideal RCT, and then designing your observational protocol to emulate it as closely as possible (Hernán 2022). Doing so can help avoid fundamental design errors that might otherwise limit or bias observational data (Hernán 2022). For example, in this study, patients were only enrolled if they received exclusively LR or NS, much as if they had been randomized to two treatment arms of an RCT. 

TMLE is a methodology designed to estimate causal effects from observational data while adjusting for confounders (Schuler 2017). It combines concepts from G-computation, which estimates the average treatment effect of an intervention by adjusting for effects of confounders on the outcome, with concepts from propensity matching, which estimates treatment effect by adjusting for effects of confounders on exposure (Schuler 2017). Having adjusted both exposures and outcomes for the confounders, it is said to be “doubly robust” — with the estimate, therefore, being accurate even if one accidentally omits important confounders for one of the two (Schuler 2017). The characteristics that were significantly different between the LR and NS exposure groups — race, organ dysfunction at presentation, ICU admission, hemoglobin SS genotype, discharge year, and hydroxyurea use — were appropriately included as confounders in the TMLE analysis. However, without more granular clinical data, it is unclear whether all relevant confounders were included. For example, were opioid-sparing pain treatments such as ketamine used differentially between groups?

Primary Outcome: A significant improvement in hospital-free days, even a modest one such as the 0.4 days observed here, is a notable finding for any study due to the innumerable factors that contribute to hospital length-of-stay and mortality. When considering the significant healthcare burden faced by patients with SCD, even a small benefit might be impactful. This is especially true as LR and NS are otherwise fairly interchangeable; they are widely available, similar in cost, require intravenous access for administration, and have similar risks (e.g. volume overload).

Pain Control: That length-of-stay and intravenous opioid-free days experience a similar magnitude of benefit (0.4 days) as hospital-free days, without an observed effect on mortality or invasive support, suggests that perhaps the means by which LR resuscitation improves hospital-free days in VOE is improved pain control. Unfortunately, due to the nature of the dataset, patients’ reported pain was not directly assessed by this study. Additionally, as discussed above, the frequency and dose of IV opioids — and other analgesia — are not assessed by any of the measured outcomes. An outcome such as total MME or MME per day might be beneficial in assessing this hypothesis.

Applicability to Emergency Medicine: For emergency medicine providers, the greatest challenge with this study is that it did not include or account for ED care. Only fluids administered after admission were considered and only in the subset of patients whose pain could not be controlled adequately in the ED to discharge without an admission. Future research that either focuses on or includes care in the ED is needed, as this is where most VOE acute care begins and may end for patients discharged home.

Author’s Conclusion: “Using a large, multicenter database to compare the effectiveness of LR vs NS resuscitation in patients hospitalized with SCD VOEs, we found that LR use was associated with small but significant improvements in several outcomes including HFD [hospital free days] and hospital readmission. In addition, fluid volume and possibly HbSS genotype modified the effects of LR vs NS.”

Our Conclusion:

This study suggests early LR resuscitation may improve hospital-free days in VOE but is limited by data constraints, its observational design, and the exclusion of ED care. Future research should address these gaps. Given that LR and NS are otherwise clinically interchangeable, resuscitationists should consider using LR for fluid resuscitation in SCD patients with VOE.

Clinical Bottom Line:

Choose LR for fluid resuscitation in SCD patients with VOE.

Citations:

1. Piel FB, Steinberg MH, Rees DC. Sickle Cell Disease. N Engl J Med. 2017;376(16):1561-1573. PMID: 28423290
2. Kidwell K, Albo C, Pope M, et al. Characteristics of sickle cell patients with frequent ED visits and hospitalizations. PLoS One. 2021;16(2):e0247324. Published 2021 Feb 22. PMID: 33617594
3. Lovett PB, Sule HP, Lopez BL. Sickle Cell Disease in the ED. Hematol Oncol Clin North Am. 2017;31(6):1061-1079. PMID: 29078924
4. Myburgh JA, Mythen MG. Resuscitation fluids. N Engl J Med. 2013;369(13):1243-1251. PMID: 24066745
5. Self WH, Semler MW, Wanderer JP, et al. Balanced Crystalloids versus Saline in Noncritically Ill Adults. N Engl J Med. 2018;378(9):819-828. PMID: 29485926
6. Semler MW, Self WH, Wanderer JP, et al. Balanced Crystalloids versus Saline in Critically Ill Adults. N Engl J Med. 2018;378(9):829-839. PMID: 29485925
7. Self WH, Evans CS, Jenkins CA, et al. Clinical Effects of Balanced Crystalloids vs Saline in Adults With Diabetic Ketoacidosis: A Subgroup Analysis of Cluster Randomized Clinical Trials. JAMA Netw Open. 2020;3(11):e2024596. Published 2020 Nov 2. PMID: 33196806
8. Carden MA, Fay ME, Lu X, et al. Extracellular fluid tonicity impacts sickle red blood cell deformability and adhesion. Blood. 2017;130(24):2654-2663. PMID: 28978568
9. Carden MA, Fay M, Sakurai Y, et al. Normal saline is associated with increased sickle red cell stiffness and prolonged transit times in a microfluidic model of the capillary system. Microcirculation. 2017;24(5):10.1111/micc.12353. PMID: 28106307
10. Alwang AK, Law AC, Klings ES, Cohen RT, Bosch NA. Lactated Ringer vs Normal Saline Solution During Sickle Cell Vaso-Occlusive Episodes. JAMA Intern Med. 2024;184(11):1365-1372. PMID: 39250114
11. Hernán MA, Wang W, Leaf DE. Target Trial Emulation: A Framework for Causal Inference From Observational Data. JAMA. 2022;328(24):2446-2447. PMID: 36508210
12. Schuler MS, Rose S. Targeted Maximum Likelihood Estimation for Causal Inference in Observational Studies. Am J Epidemiol. 2017;185(1):65-73. PMID: 27941068

Post Peer Reviewed By: Anand Swaminathan, MD (Twitter/X: @EMSwami)