AUTHORS
Mohan Jauhar, BA, Rahul Dadwani, MD, MSc2, Lillyan P. Mullin, BA3, Gabrielle Lapping-Carr, MD4, Elizabeth L. Tung, MD, MS5, Monica E. Peek, MD, MPH, MSc6, Austin Wesevich, MD, MPH, MS7
AFFILIATIONS
1 University of Chicago, Chicago, Illinois, United States
2 University of Chicago, Chicago, Illinois, United States
3 University of Minnesota, Minneapolis, Minnesota, United States
4 University of Chicago, Chicago, Illinois, United States
5 University of Chicago, Chicago, Illinois, United States
6 University of Chicago, Chicago, Illinois, United States
7 Ohio State University, Columbus, Ohio, United States
ABSTRACT
Introduction
Patients with sickle cell disease (SCD) are frequently hospitalized for vaso-occlusive episodes (VOE). Acute chest syndrome (ACS) can develop during VOE and may require opioids to control pain. While opioid dosage varies with clinical severity, clinician bias may also play a role in SCD pain management. One potential source of clinician bias is patient socioeconomic disadvantage, which can be proxied through neighborhood vulnerability. While prior studies describe disparities in opioid receipt for patients with SCD relative to other patients presenting with pain, differences in opioid receipt among patients with SCD are underexplored. We hypothesized that patients with SCD from more vulnerable neighborhoods would receive fewer opioids than those from less vulnerable neighborhoods, possibly due to clinician bias.
Methods
We included all patients with SCD (ages 12+) hospitalized for ACS and treated with opioids at the University of Chicago from 2017-2021. Via patient addresses, we used census tract to determine the 2022 Centers for Disease Control and Prevention Social Vulnerability Index (SVI) national percentile, a measure of 16 area-level socioeconomic and demographic characteristics, which was then categorized into quartiles. Higher SVI quartiles indicated greater neighborhood vulnerability. We converted all opioid administrations during the first 96 hours after ACS diagnosis to morphine milligram equivalents (MME) and calculated mean MME per day. We evaluated the association between MME per day and SVI quartile with a multilevel, multivariable linear regression model. The model accounted for hospitalizations clustered within patients and adjusted for treatment setting (pediatric versus adult hospital), history of ACS, history of VOE, mean pain score over the first 96 hours after ACS diagnosis, intensive care unit transfer, and 3 measures of ACS severity: diagnostic oxygen saturation to fraction of inspired oxygen ratio (SaO2/FiO2) < 310 (per our prior work), ≥3 opacified pulmonary lobes on chest radiography, and red cell exchange (RCE) receipt. Given differences by treatment setting, we then restricted models to adult versus pediatric hospitals.
Results
Of the 287 hospitalizations across 155 patients, 54% of hospitalizations were for female patients, the median age was 23.6 years (IQR 18.5-31.3), and all patients were non-Hispanic and Black. Most hospitalizations were for patients from more vulnerable neighborhoods: 24 (8%) in the first SVI quartile, 23 (8%) in the second, 102 (36%) in the third, and 138 (48%) in the fourth. Across all hospitalizations, a median of 148 MME per day (IQR 43-426) was administered, and the mean pain score was 5.4. After adjusting for clinical characteristics, fewer MME per day were administered for hospitalizations of patients in higher SVI quartiles than those in the first SVI quartile: second quartile (adjusted ß -303; 95% CI -580, -27), third quartile (adjusted ß -344; 95% CI -559, -130), and fourth quartile (adjusted ß -303; 95% CI -511, -95). Within this model, treatment in an adult hospital (adjusted ß 115; 95% CI 15, 214) and a higher mean pain score (adjusted ß 51; 95% CI 32, 69) were associated with receiving more MME per day.
When restricting the model to adult hospitals, the disparities in MME per day by SVI quartile worsened: second quartile (adjusted ß -430; 95% CI -840, -20), third quartile (adjusted ß -475; 95% CI -776, -174), and fourth quartile (adjusted ß -419; 95% CI -711, -126) relative to the first quartile. The only additional variable significantly associated with MME per day was mean pain score (adjusted ß 59; 95% CI 28, 91). When restricting the model to the pediatric hospital, there were no significant differences in MME per day by SVI quartile. MME per day was only significantly associated with mean pain score (adjusted ß 38; 95% CI 24, 52).
Conclusion
Our study demonstrates that clinicians manage SCD pain differently for patients from disadvantaged neighborhoods compared to those from more advantaged neighborhoods, despite treating all patients for the same life-threatening condition (i.e., acute chest syndrome). These healthcare disparities were absent in the pediatric hospital but present in adult hospitals, suggesting that clinician bias around patient-reported pain may play an important role in SCD treatment disparities.
DISCLOSURES
Mohan Jauhar, BA: No relevant conflicts of interest to declare.