Development and validation of a predictive model algorithm to identify anaphylaxis in adults with type 2 diabetes in<scp>U.S.</scp>administrative claims data

Beachler, Daniel C.; Taylor, Daniel J.; Anthony, Mary S.; Yin, Ruihua; Li, Ling; Saltus, Catherine W.; Li, Lin; Shaunik, Alka; Walsh, Kathleen E.; Rothman, Kenneth J.; Johannes, Catherine B.; Aroda, Vanita R.; Carr, Warner; Goldberg, Pinkus; Accardi, Andrew; O'Shura, John Shane; Sharma, Kristen; Juhaeri, Juhaeri; Lanes, Stephan; Wu, Chuntao

doi:10.1002/pds.5257

Cited by 7 publications

(11 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Censoring events included disenrollment from the health care system, discontinuation of the index drug (no prescription refill for >30 days after the end of the last prescription’s supply), switch to another GLP-1 RA (lixisenatide and insulin glargine/lixisenatide FRC were considered to be the same treatment, as were liraglutide and insulin degludec/liraglutide FRC), death, or end of the study period. This study was also purposed to develop and validate an algorithm to identify anaphylaxis cases in the setting of an administrative claims database ( 8 ).…”

Section: Methodsmentioning

confidence: 99%

“…The study outcome was the first anaphylaxis event identified using a predictive model algorithm developed as the initial stage of this study ( 8 ). The algorithm was based on ICD-9-CM and ICD-10-CM diagnosis codes, Current Procedural Terminology and Healthcare Common Procedure Coding System procedure codes, and National Drug Code and General Product Identifier medication codes.…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Risk of Anaphylaxis Among New Users of GLP-1 Receptor Agonists: A Cohort Study

Anthony,

Aroda,

Parlett

et al. 2024

Diabetes Care

Self Cite

View full text Add to dashboard Cite

OBJECTIVE To assess risk of anaphylaxis among patients with type 2 diabetes mellitus who are initiating therapy with a glucagon-like peptide-1 receptor agonist (GLP-1 RA), with a focus on those starting lixisenatide therapy. RESEARCH DESIGN AND METHODS A cohort study was conducted in three large, U.S. claims databases (2017–2021). Adult (aged ≥18 years) new users of a GLP-1 RA who had type 2 diabetes mellitus and ≥6 months enrollment in the database before GLP-1 RA initiation (start of follow-up) were included. GLP-1 RAs evaluated were lixisenatide, an insulin glargine/lixisenatide fixed-ratio combination (FRC), exenatide, liraglutide or insulin degludec/liraglutide FRC, dulaglutide, and semaglutide (injectable and oral). The first anaphylaxis event during follow-up was identified using a validated algorithm. Incidence rates (IRs) and 95% CIs were calculated within each medication cohort. The unadjusted IR ratio (IRR) comparing anaphylaxis rates in the lixisenatide cohort with all other GLP-1 RAs combined was analyzed post hoc. RESULTS There were 696,089 new users with 456,612 person-years of exposure to GLP-1 RAs. Baseline demographics, comorbidities, and use of other prescription medications in the 6 months before the index date were similar across medication cohorts. IRs (95% CIs) per 10,000 person-years were 1.0 (0.0–5.6) for lixisenatide, 6.0 (3.6–9.4) for exenatide, 5.1 (3.7–7.0) for liraglutide, 3.9 (3.1–4.8) for dulaglutide, and 3.6 (2.6–4.9) for semaglutide. The IRR (95% CI) for the anaphylaxis rate for the lixisenatide cohort compared with the pooled other GLP-1 RA cohort was 0.24 (0.01–1.35). CONCLUSIONS Anaphylaxis is rare with GLP-1 RAs. Lixisenatide is unlikely to confer higher risk of anaphylaxis than other GLP-1 RAs.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Risk of Anaphylaxis Among New Users of GLP-1 Receptor Agonists: A Cohort Study

Anthony,

Aroda,

Parlett

et al. 2024

Diabetes Care

Self Cite

View full text Add to dashboard Cite

show abstract

“…By excluding noncases using conditional sampling, however, it may be possible to target a sample that is enriched in people more likely to be cases 24 . For instance, one could design a “screening” algorithm aimed at capturing all (or nearly all) of the cases so that it has near‐perfect sensitivity, while still being specific enough to exclude many noncases 25 . Validating a random sample of people identified by such as screening algorithm could generate sufficient cases to be used to estimate sensitivity of algorithms designed to have high PPVs (Table 1).…”

Section: Validation To Inform Quantitative Bias Analysismentioning

confidence: 99%

Validation to correct for outcome misclassification bias

Lanes

Beachler

2023

Pharmacoepidemiology and Drug

Self Cite

View full text Add to dashboard Cite

1. Outcome validation is often requested by regulators to address misclassification bias in database studies of drug safety and comparative effectiveness.2. Validation studies commonly report only one positive predictive value (PPV) estimate.3. Since a high value of PPV does not imply misclassification bias is negligible, and a low value of PPV does not imply misclassification bias is important, this approach does not adequately address outcome misclassification bias.4. Validation should be designed to inform quantitative bias analysis that corrects results for misclassification bias. 5. To correct for misclassification bias, quantitative bias analysis requires parameters for false positive errors and false negative errors in each comparison group.

show abstract

“…To date, most attempts to accurately identify anaphylaxis events using automated algorithms applied to structured medical claims data have had limited success ( 20 ). This is in part due to the challenges of diagnosing a condition with diverse clinical presentation ( 19 ), reliance on structured medical claims data ( 21 , 22 ), and the practice of “rule-out” coding ( 23 ). An algorithm published by Walsh et al in 2013 ( 22 ) intended to improve identification of anaphylaxis in FDA safety surveillance studies using structured diagnosis, procedure, and encounter data.…”

Section: Abbreviationsmentioning

confidence: 99%

“…Ball et al ( 34 ) and Yu et al ( 35 ) have found that information extracted from EHRs about exposures, comorbidities, presenting symptoms, treatments, and disease severity appearing in chart notes may be useful for identifying actual anaphylaxis. Recent studies have used natural language processing (NLP) and machine-learning methods to improve the identification of anaphylaxis events but did not include external populations to validate their findings ( 21 , 36 ).…”

Section: Abbreviationsmentioning

confidence: 99%

Improving Methods of Identifying Anaphylaxis for Medical Product Safety Surveillance Using Natural Language Processing and Machine Learning

Carrell¹,

Gruber²,

Floyd³

et al. 2022

American Journal of Epidemiology

View full text Add to dashboard Cite

We sought to determine whether machine learning and natural language processing (NLP) applied to electronic medical records could improve performance of automated healthcare claims-based algorithms to identify anaphylaxis events using data on 516 patients with outpatient, emergency department, or inpatient anaphylaxis diagnosis codes during 2015-2019 in two integrated healthcare institutions in the Northwest United States. We used one site’s manually reviewed gold standard outcomes data for model development and the other’s for external validation based on cross-validated (cv) area under the receiver operating characteristic curve (cv AUC), positive predictive value (PPV), and sensitivity. In the development site 154 (64%) of 239 potential events met adjudication criteria for anaphylaxis compared to 180 (65%) of 277 in the validation site. Logistic regression models using only structured claims data achieved a cv-AUC of 0.58 (95% CI: 0.54, 0.63). Machine learning improved cv-AUC to 0.62 (0.58, 0.66); incorporating NLP-derived covariates further increased cv AUCs to 0.70 (0.66, 0.75) in development and 0.67 (0.63, 0.71) in external validation data. A classification threshold with cv-PPV of 79% and cv-sensitivity of 66% in development data had cv-PPV of 78% and cv-sensitivity of 56% in external data. Machine learning and NLP-derived data improved identification of validated anaphylaxis events.

show abstract

Development and validation of a predictive model algorithm to identify anaphylaxis in adults with type 2 diabetes inU.S.administrative claims data

Cited by 7 publications

References 17 publications

Risk of Anaphylaxis Among New Users of GLP-1 Receptor Agonists: A Cohort Study

Risk of Anaphylaxis Among New Users of GLP-1 Receptor Agonists: A Cohort Study

Validation to correct for outcome misclassification bias

Improving Methods of Identifying Anaphylaxis for Medical Product Safety Surveillance Using Natural Language Processing and Machine Learning

Contact Info

Product

Resources

About