Abstract
This cross-sectional study evaluates the association between socioeconomic indicators and penicillin allergy delabeling prevalence in Boston, Massachusetts.
Introduction
Penicillin allergy labels in patient electronic health record (EHR) data often do not represent true or persistent allergies.1 Prescribing alternative (non–β-lactam) antibiotics increases antimicrobial resistance (AMR), adverse effects, health care costs, and mortality.1 Removing penicillin allergy mislabels, or delabeling, is an evidence-based strategy permitting β-lactam antibiotic prescribing.2 Clinicians delabel patients by evaluating allergy history, removing erroneous entries, and/or procedures like drug challenges. Given that AMR infections disproportionately impact economically disadvantaged populations,3 we assessed socioeconomic indicators’ association with penicillin allergy delabeling prevalence.
Methods
We used EHR data, including self-reported demographics and delabel status,4 to identify primary care patients with a penicillin allergy record from January 2019 to April 2022 at Mass General Brigham (MGB) and Tufts Medicine (TM). Merging patient zip code onto geospatial data, we calculated validated indicators: the social vulnerability index (SVI, total and subscales: socioeconomic status [SES], household characteristics, racial and ethnic minority status, and housing type and transportation) and American Community Survey uninsured and unemployment rates, as well as median household income (eTable in Supplement 1). This study was approved by MGB and TM institutional review boards with a waiver of informed consent because study design was retrospective and as such, research could not practically be carried out without the waiver. Reporting followed Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.
We assessed delabeling prevalence across the greater Boston area, comparing highest 10 vs lowest 10 per capita income zip codes using analysis of variance with Bonferroni adjustment for multiple comparisons. We computed penicillin allergy delabeling prevalence across each indicator’s quartile (Q) then compared the lowest Q with each subsequently higher Q using prevalence ratios with 95% CIs from generalized estimating equation regression models. Poisson distribution and log-link function accounted for patient-level clustering within zip codes. There were minimal missing data (eMethods in Supplement 1). Geospatial zip code analysis used R and RStudio version 4.4.2 (R Project for Statistical Computing). Statistical analyses were conducted in SAS version 9.4 (SAS Institute), with a 2-sided P < .05 considered significant.
Results
Of 76 709 patients (46 625 [61%] MGB, mean [SD] age 56 [18] years, 53 129 [69%] female, 3564 [4%] Asian, 4261 [6%] Black, 61 802 [81%] White, 7082 [9%] other [American Indian or Alaska Native or missing race], 5617 [7%] Hispanic ethnicity, 4021 [5%] non-English language preference), 5195 (6.8%) were delabeled. The mean delabeling prevalence difference between highest and lowest income areas was 6.4 percentage points (t1 = 3.2; P = .004) (Figure).Delabeling was less likely in those more socially vulnerable compared with less socially vulnerable areas, and in lower SES vs higher SES areas (P for trend all <.001) (Table). Delabeling prevalence was also lower in areas with higher uninsured and unemployment rates and lower median income (P for trend all <.001).
Figure. Geospatial Map of Penicillin Allergy Delabeling Prevalence .
Zip code areas for New England (upper left) and Greater Boston Area (right) with map of US for context (bottom left). Areas outlined in yellow represent the 10 highest income zip code areas in Massachusetts towns, including Weston (02493), Dover (02030), Wellesley (02482), Sherborn (01770), Lexington (02420 and 02421), Carlisle (01741), Lincoln (01773), Concord (01742), and Wayland (01778) with income per capita ranges from $150 253 to $354 387 in 2022 US dollars. Areas outlined in red represent the 10 lowest income zip code areas in Massachusetts towns including Lawrence (01840, 01841, and 01843), Chelsea (02150), Everett (02149), Brockton (02301 and 02302), and Lynn (01901, 01902, and 01904) with income per capita ranges from $17 984 to $23 099 in 2022 US dollars. The mean penicillin allergy delabeling prevalence was 6.4 percentage points higher, on average, in the 10 highest per capita income areas compared with the 10 lowest per capita income areas (P = .004). Only zip code areas with more than 3 observations are shown. Income per capita used the Massachusetts Department of Revenue for fiscal year 2022. Geospatial zip code tabulation areas are from the 2020 TIGER/Line Shapefiles: Zip Code Tabulation Areas from the US Census Bureau, Geography Division.
Table. Socioeconomic Geospatial Indicators and Association With Penicillin Allergy Delabeling Prevalence.
| Indicator and Q-based categoriesa | No. total | No. delabeled | Delabeling prevalence (95% CI) | Prevalence ratio (95% CI) | P value for trendc | |
|---|---|---|---|---|---|---|
| Unadjusted | Adjustedb | |||||
| Social Vulnerability Index (SVI), rankings | ||||||
| SVI total | ||||||
| Q1 | 18 901 | 1344 | 7.11 (6.74-7.48) | 1 [Reference] | 1 [Reference] | <.001 |
| Q2 | 19 466 | 1337 | 6.87 (6.51-7.22) | 0.97 (0.90-1.04) | 1.00 (0.93-1.08) | |
| Q3 | 19 023 | 1471 | 7.73 (7.35-8.11) | 1.09 (1.01-1.17) | 0.94 (0.87-1.01) | |
| Q4 | 19 319 | 1043 | 5.40 (5.08-5.72) | 0.76 (0.70-0.82) | 0.85 (0.79-0.93) | |
| SVI SES | ||||||
| Q1 | 19 135 | 1560 | 8.15 (7.76-8.54) | 1 [Reference] | 1 [Reference] | <.001 |
| Q2 | 19 221 | 1351 | 7.03 (6.67-7.39) | 0.86 (0.80-0.92) | 0.93 (0.87-0.99) | |
| Q3 | 19 059 | 1239 | 6.50 (6.15-6.85) | 0.80 (0.74-0.86) | 0.84 (0.78-0.90) | |
| Q4 | 19 294 | 1045 | 5.42 (5.10-5.74) | 0.66 (0.62-0.72) | 0.80 (0.74-0.87) | |
| SVI household characteristics | ||||||
| Q1 | 19 141 | 1343 | 7.02 (6.65-7.38) | 1 [Reference] | 1 [Reference] | .02 |
| Q2 | 19 213 | 1405 | 7.31 (6.94-7.68) | 1.04 (0.97-1.12) | 1.01 (0.94-1.08) | |
| Q3 | 19 178 | 1313 | 6.85 (6.49-7.20) | 0.98 (0.91-1.05) | 0.98 (0.91-1.06) | |
| Q4 | 19 177 | 1134 | 5.91 (5.58-6.25) | 0.84 (0.78-0.91) | 0.91 (0.84-0.98) | |
| SVI racial and ethnic minority status | ||||||
| Q1 | 19 178 | 1235 | 6.44 (6.09-6.79) | 1 [Reference] | 1.03 (0.96-1.12) | .55 |
| Q2 | 19 176 | 1276 | 6.65 (6.30-7.01) | 1.03 (0.96-1.11) | 1.08 (1.00-1.16) | |
| Q3 | 19 135 | 1570 | 8.20 (7.82-8.59) | 1.27 (1.19-1.37) | 0.95 (0.87-1.03) | |
| Q4 | 19 220 | 1114 | 5.80 (5.47-6.13) | 0.90 (0.83-0.97) | 1.03 (0.96-1.12) | |
| SVI housing type and transportation | ||||||
| Q1 | 19 172 | 1243 | 6.48 (6.13-6.83) | 1 [Reference] | 0.97 (0.90-1.05) | .60 |
| Q2 | 19 183 | 1255 | 6.54 (6.19-6.89) | 1.01 (0.94-1.09) | 0.99 (0.92-1.07) | |
| Q3 | 18 724 | 1335 | 7.13 (6.76-7.50) | 1.10 (1.02-1.18) | 0.97 (0.90-1.05) | |
| Q4 | 19 630 | 1362 | 6.94 (6.58-7.29) | 1.07 (0.99-1.15) | 0.97 (0.90-1.05) | |
| American Community Survey Measures | ||||||
| Uninsured rate | ||||||
| Q1 | 18 965 | 1517 | 8.00 (7.61-8.39) | 1 [Reference] | 1 [Reference] | <.001 |
| Q2 | 19 475 | 1449 | 7.44 (7.07-7.81) | 0.93 (0.87-1.00) | 0.99 (0.93-1.06) | |
| Q3 | 19 319 | 1186 | 6.14 (5.80-6.48) | 0.77 (0.71-0.83) | 0.89 (0.82-0.96) | |
| Q4 | 18 950 | 1043 | 5.50 (5.18-5.83) | 0.69 (0.64-0.74) | 0.79 (0.73-0.85) | |
| Unemployment rate | ||||||
| Q1 | 19 098 | 1568 | 8.21 (7.82-8.60) | 1 [Reference] | 1 [Reference] | <.001 |
| Q2 | 19 249 | 1295 | 6.73 (6.37-7.08) | 0.82 (0.76-0.88) | 0.97 (0.90-1.04) | |
| Q3 | 19 236 | 1186 | 6.17 (5.83-6.51) | 0.75 (0.70-0.81) | 0.85 (0.79-0.92) | |
| Q4 | 19 126 | 1146 | 5.99 (5.66-6.33) | 0.73 (0.68-0.79) | 0.83 (0.77-0.90) | |
| Median family income | ||||||
| Q1 | 18 781 | 1146 | 6.10 (5.76-6.44) | 1 [Reference] | 1 [Reference] | <.001 |
| Q2 | 19 569 | 1248 | 6.38 (6.04-6.72) | 1.05 (0.97-1.13) | 1.05 (0.97-1.14) | |
| Q3 | 19 205 | 1241 | 6.46 (6.11-6.81) | 1.06 (0.98-1.14) | 1.11 (1.02-1.20) | |
| Q4 | 19 154 | 1560 | 8.14 (7.76-8.53) | 1.33 (1.24-1.44) | 1.30 (1.20-1.40) | |
Abbreviations: Q, quartile; SES, socioeconomic status; SVI, social vulnerability index.
See eTable in Supplement 1. Q4 indicates greater area deprivation for all indicators except median family income, where Q4 represents less area deprivation.
Adjusted model controlled for patient age, sex, race, ethnicity, and hospital system (Mass General Brigham vs Tufts).
P value for trend is derived from the significance of a single continuous variable with values set to the within quartile median in a separate adjusted model that changes the functional form of each geospatial variable from categorical to continuous.
Discussion
We found reduced penicillin allergy delabeling in lower income areas; in adjusted analyses, reduced delabeling was associated with higher social vulnerability, uninsured and unemployment rates, and lower SES and median household income. These data suggest delabeling access or uptake is not equitable across SES, even among demographically similar patients at the same primary care institutions.
Limited observational data suggest White, non-Hispanic individuals of higher SES more likely carry penicillin allergy labels.2 This study demonstrates lower rates of delabeling in vulnerable populations among those eligible for delabeling while controlling for individual demographics. Given antibiotic prescribing disparities5 and AMR infection disparities,3 improving equitable access to penicillin allergy delabeling might effectively improve infectious diseases outcomes.
We used data from 2 health care systems in greater Boston, thus limiting generalizability to areas with different population characteristics and underrepresenting rural settings. Also, geospatial indicators are highly correlated, so we only adjusted for individual patient confounders. Ideal penicillin allergy label assessment would consider the temporal relationship between allergic reaction, label placement, and its subsequent removal, but EHR allergy data are cross-sectional, and label entry does not correspond to reaction date.
Across a large Boston-based multisite primary care cohort, penicillin allergy delabeling prevalence was lowest in areas of social vulnerability and economic deprivation, suggesting inequitable delabeling for those economically disadvantaged. Future research should identify barriers and facilitators to delabeling and evaluate access expansion strategies for vulnerable communities.
eTable. Geospatial Economic Indicators
eMethods. Additional Information on Assessment of Missing Zip Code Data
eReferences.
Data Sharing Statement
References
- 1.Shenoy ES, Macy E, Rowe T, Blumenthal KG. Evaluation and management of penicillin allergy: a review. JAMA. 2019;321(2):188-199. doi: 10.1001/jama.2018.19283 [DOI] [PubMed] [Google Scholar]
- 2.Arasaratnam RJ, Chow TG, Liu AY, Khan DA, Blumenthal KG, Wurcel AG. Penicillin allergy evaluation and health equity: a call to action. J Allergy Clin Immunol Pract. 2023;11(2):422-428. doi: 10.1016/j.jaip.2022.12.001 [DOI] [PubMed] [Google Scholar]
- 3.Casale E, Harvey EJ, De Brún C, Finistrella V, Ashiru-Oredope D. Association between health inequalities, antibiotic use and antibiotic-resistant infections in high-income countries: a scoping review. Int J Pharm Pract. 2022;30(suppl 2):ii10-ii11. doi: 10.1093/ijpp/riac089.011 [DOI] [Google Scholar]
- 4.Blumenthal KG, Jiang B, King AJ, Mann JT, Zhang Y, Wurcel AG. Derivation and validation of an electronic health record penicillin allergy de-labeling prevalence measure. Clin Infect Dis. 2024;80(4):723-726. doi: 10.1093/cid/ciae641 [DOI] [PubMed] [Google Scholar]
- 5.Kim C, Kabbani S, Dube WC, et al. Health equity and antibiotic prescribing in the United States: a systematic scoping review. Open Forum Infect Dis. 2023;10(9):ofad440. doi: 10.1093/ofid/ofad440 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable. Geospatial Economic Indicators
eMethods. Additional Information on Assessment of Missing Zip Code Data
eReferences.
Data Sharing Statement
