Show simple item record

dc.contributor.advisorHeather, Haugen A.
dc.contributor.advisorMichael, Wechsler E.
dc.contributor.authorBeuther, David
dc.contributor.committeememberLisa, Cicutto C.
dc.contributor.committeememberDouglas, Everett C.
dc.contributor.committeememberElizabeth, Kern
dc.contributor.committeememberLori, Silveira J.
dc.date.accessioned2020-01-14T15:44:31Z
dc.date.available2020-01-14T15:44:31Z
dc.date.submitted2019
dc.descriptionIncludes bibliographical references.
dc.descriptionFall
dc.description.abstractBackground Clinical trial recruitment is inefficient. New, more efficient and phenotype-targeted recruitment strategies require analysis of electronic health record data to automate severe asthma case finding. Objective To determine whether a rules-based, electronic health record-driven algorithm can reliably identify severe asthma cases among adults with asthma. Methods Three adult pulmonologists with expertise in asthma performed independent chart reviews on 148 adult asthmatics seen by subspecialists at National Jewish Health between 2013 and 2015. At least two experts independently rated confidence that the diagnosis was severe asthma and how they came to their determination. Disagreements on severe asthma status were adjudicated by an independent third reviewer. A guideline-based, electronic, severe asthma case finding algorithm was created and compared to expert assessment. Discordant cases were examined in more detail for insights on algorithm performance. A logistic regression model was created to identify novel severe asthma predictive variables that were used to improve algorithm performance. Results Expert agreement on severe asthma status was substantial (89% of 148 cases, Fleiss’ kappa = 0.71). Severe asthma prevalence was 24% (95% CI, 17%-31%). The best initial algorithm identified all 36 cases of severe asthma, but at a high false positive rate (PPV 42%, 95% CI, 31%-53%). A three-variable logistic regression model demonstrated association between severe asthma and younger age, use of chronic daily oral corticosteroids, and a sleep apnea diagnosis (likelihood ratio χ2 = 34, p < 0.001). Patients 65 or older had only a 9% probability of severe asthma in the model (95% CI 5%-19%, p<0.001). Reclassifying initially algorithm positive cases with age ≥65 and initially negative cases with sleep apnea using chronic OCS improved overall algorithm performance, with a sensitivity of 86% (95% CI 70%-95%), specificity of 77% (95% CI 68%-84%), positive predictive value of 54% (41%-67%), and a negative predictive value of 95% (95% CI 87%-98%). Conclusion A guideline-based severe asthma algorithm can automate the efficient identification of expert-confirmed cases of severe asthma from an electronic health record. This algorithm could improve severe asthma care and discovery by reducing the time and cost of severe asthma clinical trial recruitment.
dc.identifierBeuther_ucdenveramc_1639D_10694.pdf
dc.identifier.urihttps://hdl.handle.net/10968/4776
dc.languageEnglish
dc.publisherUniversity of Colorado at Denver, Anschutz Medical Campus. Health Sciences Library
dc.rightsCopyright of the original work is retained by the author.
dc.subjectclinical trials
dc.subjectsevere asthma
dc.subject.meshElectronic Health Records
dc.subject.meshAlgorithms
dc.subject.meshAsthma
dc.subject.meshPhenotype
dc.subject.meshPatient Selection
dc.titlePerformance of a rules-based, electronic health record-driven severe asthma case finding algorithm for clinical trial recruitment
dc.typeThesis
thesis.degree.disciplineClinical Sciences
thesis.degree.grantorUniversity of Colorado at Denver, Anschutz Medical Campus
thesis.degree.levelDoctoral
thesis.degree.nameDoctor of Philosophy (Ph.D.)


Files in this item

Thumbnail
Thumbnail

This item appears in the following Collection(s)

Show simple item record