Steven Kerr, Chris Robertson, Cathie Sudlow, Aziz Sheikh
{"title":"Enabling health data analyses across multiple private datasets with no information sharing using secure multiparty computation.","authors":"Steven Kerr, Chris Robertson, Cathie Sudlow, Aziz Sheikh","doi":"10.1136/bmjhci-2024-101384","DOIUrl":"https://doi.org/10.1136/bmjhci-2024-101384","url":null,"abstract":"<p><p>The UK's health datasets are among the most comprehensive and inclusive globally, enabling groundbreaking research during the COVID-19 pandemic. However, restrictions on data sharing between secure data environments (SDEs) imposed limitations on the ability to carry out joint analyses across multiple separate datasets. There are currently significant efforts underway to enable such analyses using methods such as federated analytics (FA) and virtual SDEs. FA involves distributed data analysis without sharing raw data but does require sharing summary statistics. Virtual SDEs in principle allow researchers to access data across multiple SDEs, but in practice, data transfers may be restricted by information governance concerns.Secure multiparty computation (SMPC) is a cryptographic approach that allows multiple parties to perform joint analyses over private datasets with zero information sharing. SMPC may eliminate the need for data-sharing agreements and statistical disclosure control, offering a compelling alternative to FA and virtual SDEs. SMPC comes with a higher computational burden than traditional pooled analysis. However, efficient implementations of SMPC can enable a wide range of practical, secure analyses to be carried out.This perspective reviews the strengths and limitations of FA, virtual SDEs and SMPC as approaches to joint analyses across SDEs. We argue that while efforts to implement FA and virtual SDEs are ongoing in the UK, SMPC remains underexplored. Given its unique advantages, we propose that SMPC deserves greater attention as a transformative solution for enabling secure, cross-SDE analyses of private health data.</p>","PeriodicalId":9050,"journal":{"name":"BMJ Health & Care Informatics","volume":"32 1","pages":""},"PeriodicalIF":4.1,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144156818","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Rachel L Wasserman, Foster R Goss, Diane L Seger, Kimberly G Blumenthal, Ying-Chih Lo, Heba H Edrees, Sheril Varghese, Liqin Wang, Suzanne Blackley, David W Bates, Li Zhou
{"title":"Allergy alerting and overrides for opioid analogues across two health systems.","authors":"Rachel L Wasserman, Foster R Goss, Diane L Seger, Kimberly G Blumenthal, Ying-Chih Lo, Heba H Edrees, Sheril Varghese, Liqin Wang, Suzanne Blackley, David W Bates, Li Zhou","doi":"10.1136/bmjhci-2024-101259","DOIUrl":"10.1136/bmjhci-2024-101259","url":null,"abstract":"<p><strong>Objectives: </strong>Despite opioids comprising a majority of drug allergy alert (DAA) overrides, few studies have designed recommendations for clinical decision support (CDS) systems. We determined the frequency of opioid analogue DAA overrides and assessed DAAs overridden by patient and provider type, the most common allergic reactions documented in electronic health records (EHR), reaction severity, and associated hypersensitivity.</p><p><strong>Methods: </strong>We conducted an observational cross-sectional study of DAAs in two geographically remote health systems. Patients were included if they were 18 years or older and had an opioid DAA generated when a medication was ordered. Patient and provider demographics, drug allergies, medication ordered, alert overrides, drug allergy reactions and DAA history were collected. Opioid analogue allergies were analysed by reaction type documented in the EHR, reaction severities (high, medium or low) and hypersensitivity reaction. Based on these factors, alerts were recommended to be interruptive requiring a coded response or changed to be non-interruptive (informational).</p><p><strong>Results: </strong>There were 700 493 alerts concerning opioid analogues fired for 50 527 patients across both sites, and 71.8% of these alerts were overridden. Nearly three-quarters of overridden reactions had a low to medium severity level at both institutions. Only 29.3% of the overridden alerts were truly immune-mediated.</p><p><strong>Discussion: </strong>Our recommendations would reduce interruptive alerts in half by converting them to non-interruptive alerts (46.4%). The data suggest opportunities to improve opioid-related CDS systems.</p><p><strong>Conclusions: </strong>We evaluated overrides of opioids and used this data to suggest ways to redesign DAAs to decrease alert override rates, combat alert fatigue and improve patient safety.</p>","PeriodicalId":9050,"journal":{"name":"BMJ Health & Care Informatics","volume":"32 1","pages":""},"PeriodicalIF":4.1,"publicationDate":"2025-05-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144141342","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Yeunhyang Catherine Choi, Katrina Poppe, Vanessa Selak, Allan Ronald Moffitt, Claris Yee Seung Chung, Jane Ullmer, Sue Wells
{"title":"Identifying long-term conditions in New Zealand general practice using structured and unstructured data: a cross-sectional study.","authors":"Yeunhyang Catherine Choi, Katrina Poppe, Vanessa Selak, Allan Ronald Moffitt, Claris Yee Seung Chung, Jane Ullmer, Sue Wells","doi":"10.1136/bmjhci-2024-101393","DOIUrl":"10.1136/bmjhci-2024-101393","url":null,"abstract":"<p><strong>Objectives: </strong>This study examined whether incorporating free-text entries into structured general practice records improves the detection of long-term conditions (LTCs) and multimorbidity (MM) in New Zealand (NZ) general practices.</p><p><strong>Methods: </strong>Data from 374 071 deidentified individuals in general practices were analysed to identify 61 LTCs. Structured data were extracted using Read codes from a national master list, and clinical raters independently identified condition-related free-text, including synonyms, negation terms and common misspellings in randomised samples. Keywords were categorised and refined through ten iterative tests. Programmatic text classification was developed and assessed against gold-standard clinician ratings, using sensitivity, specificity, positive predictive value (PPV) and F<sub>1</sub>-score.</p><p><strong>Results: </strong>A quarter of general practitioner classifications contained either unrecognised Read codes or consisted of free-text only. Clinician inter-rater reliability was high (kappa ≥0.9). Compared with clinical gold standard, text classification yielded an average sensitivity of 88%, specificity of 99% and PPV of 95%, with an F<sub>1</sub>-score range of 82%-95%. Incorporating free text increased LTC prevalence from 42.1% to 46.3%, reducing misclassification of MM diagnoses by identifying 12 626 additional patients with MM and 15 972 additional patients with at least one LTC.</p><p><strong>Discussion: </strong>In the course of workflow, general practitioners face barriers to accurate LTC coding or may simply annotate with text-based descriptions. Programmatic text classification has demonstrated high performance and identified many more patients receiving LTC care.</p><p><strong>Conclusions: </strong>Combining structured and unstructured data optimises MM detection in NZ general practices and has the potential to improve case management, follow-up care and allocation of healthcare resources.</p>","PeriodicalId":9050,"journal":{"name":"BMJ Health & Care Informatics","volume":"32 1","pages":""},"PeriodicalIF":4.1,"publicationDate":"2025-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144126533","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Francesca Gigola, Tommaso Amato, Marco Del Riccio, Alessandro Raffaele, Antonino Morabito, Riccardo Coletta
{"title":"Artificial intelligence in clinical practice: a cross-sectional survey of paediatric surgery residents' perspectives.","authors":"Francesca Gigola, Tommaso Amato, Marco Del Riccio, Alessandro Raffaele, Antonino Morabito, Riccardo Coletta","doi":"10.1136/bmjhci-2025-101456","DOIUrl":"10.1136/bmjhci-2025-101456","url":null,"abstract":"<p><strong>Objectives: </strong>The aim of this study was to compare the performances of residents and ChatGPT in answering validated questions and assess paediatric surgery residents' acceptance, perceptions and readiness to integrate artificial intelligence (AI) into clinical practice.</p><p><strong>Methods: </strong>We conducted a cross-sectional study using randomly selected questions and clinical cases on paediatric surgery topics. We examined residents' acceptance of AI before and after comparing their results to ChatGPT's results using the Unified Theory of Acceptance and Use of Technology 2 (UTAUT2) model. Data analysis was performed using Jamovi V.2.4.12.0.</p><p><strong>Results: </strong>30 residents participated. ChatGPT-4.0's median score was 13.75, while ChatGPT-3.5's was 8.75. The median score among residents was 8.13. Differences appeared statistically significant. ChatGPT outperformed residents specifically in definition questions (ChatGPT-4.0 vs residents, p<0.0001; ChatGPT-3.5 vs residents, p=0.03). In the UTAUT2 Questionnaire, respondents expressed a more positive evaluation of ChatGPT with higher mean values for each construct and lower fear of technology after learning about test scores.</p><p><strong>Discussion: </strong>ChatGPT performed better than residents in knowledge-based questions and simple clinical cases. The accuracy of ChatGPT declined when confronted with more complex questions. The UTAUT questionnaire results showed that learning about the potential of ChatGPT could lead to a shift in perception, resulting in a more positive attitude towards AI.</p><p><strong>Conclusion: </strong>Our study reveals residents' positive receptivity towards AI, especially after being confronted with its efficacy. These results highlight the importance of integrating AI-related topics into medical curricula and residency to help future physicians and surgeons better understand the advantages and limitations of AI.</p>","PeriodicalId":9050,"journal":{"name":"BMJ Health & Care Informatics","volume":"32 1","pages":""},"PeriodicalIF":4.1,"publicationDate":"2025-05-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12097045/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144118606","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Assessing public awareness of myopia after the COVID-19 pandemic: an infodemiology study.","authors":"Junhan Chen, Deokho Lee, Shin-Ichi Ikeda, Yan Zhang, Kazuno Negishi, Kazuo Tsubota, Toshihide Kurihara","doi":"10.1136/bmjhci-2024-101156","DOIUrl":"10.1136/bmjhci-2024-101156","url":null,"abstract":"<p><strong>Objectives: </strong>Since the COVID-19 pandemic started, an increasing number of individuals use online resources to obtain health-related information, including myopia management. We aimed to obtain the search interest of global public health on myopia and prevention/treatment methods before and after the outbreak of COVID-19.</p><p><strong>Methods: </strong>Google Trends was used to assess the global search interest in myopia and related terms. Data spanning from 2019 to 2023 were collected and segmented into three periods: prepandemic (2019), pandemic year (2020) and postpandemic (2021-2023). Each average search volume index was used to examine public awareness with the Kruskal-Wallis test.</p><p><strong>Results: </strong>Among commonly used search keywords, Myopia is preferred over more colloquial terms (Nearsighted and Shortsighted). During the pandemic, searches for Myopia remained stable, while postpandemic, interest in Myopia surged (31.54%, p<0.0001). Myopia prevention/treatment awareness from 2019 to 2023 indicated notable shifts. In 2020, searches for Contact lenses and Spectacles decreased (-14.09% and -21.97%, respectively, both p<0.0001). These trends persisted postpandemic, with declining searches for Orthokeratology (-41.38%, p<0.01). Public interest for Light therapy (43.00%, p<0.0001) and Atropine (27.42%, p<0.0001) increased.</p><p><strong>Discussion: </strong>The research highlights significant increases in global online awareness of myopia during the COVID-19 pandemic. The extended period of remote work during the pandemic may alter lifestyle habits and affect public perception of treatment options. Those findings could offer valuable perspectives on global interest patterns related to myopia, which supports subsequent myopia studies and is further useful to develop relevant public health policies.</p>","PeriodicalId":9050,"journal":{"name":"BMJ Health & Care Informatics","volume":"32 1","pages":""},"PeriodicalIF":4.1,"publicationDate":"2025-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12086926/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144085825","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Ending nuclear weapons, before they end us.","authors":"Chris Zielinski","doi":"10.1136/bmjhci-2025-101572","DOIUrl":"10.1136/bmjhci-2025-101572","url":null,"abstract":"","PeriodicalId":9050,"journal":{"name":"BMJ Health & Care Informatics","volume":"32 1","pages":""},"PeriodicalIF":4.1,"publicationDate":"2025-05-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12083366/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144075672","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Evaluating machine learning algorithms for predicting HIV status among young Thai men who have sex with men.","authors":"Krittaka Soha, Sadiporn Phuthomdee, Thanapat Srichai, Lanchakorn Kittiratanawasin, Win Min Han, Sirinya Teeraananchai","doi":"10.1136/bmjhci-2024-101189","DOIUrl":"10.1136/bmjhci-2024-101189","url":null,"abstract":"<p><strong>Objective: </strong>This study aimed to develop machine learning (ML) models to predict HIV status and assessed the factors associated with HIV infection among young men who have sex with men (MSM) under the Universal Health Coverage (UHC) programme in Thailand.</p><p><strong>Methods: </strong>Young MSM aged 15-24 years who underwent HIV testing through the UHC programme from 2015 to 2022 were included. Data were divided into training (70%) and testing (30%) sets, with the Synthetic Minority Oversampling Technique (SMOTE) applied to address data set imbalance. ML models, including logistic regression, k-nearest neighbour (KNN), random forest, extreme gradient boosting (XGB) and AdaBoost, were used to predict HIV infection.</p><p><strong>Results: </strong>Among 146 813 young MSM, 11% were diagnosed with HIV. While KNN initially outperformed other ML models, the sensitivity of all models using the original data set was low due to imbalanced data. After applying SMOTE, the XGB model showed the best performance with an accuracy of 0.72, sensitivity of 0.73, specificity of 0.72 and the area under the curve of 0.72. The top predictors of HIV infection were the year of HIV testing (68%), age (55%) and targeted HIV testing (54%).</p><p><strong>Discussion: </strong>This study demonstrates the potential of ML models, particularly XGB, in predicting HIV infection among young MSM in Thailand under the UHC programme. The application of SMOTE improved model sensitivity, addressing data imbalance and enhancing predictive accuracy.</p><p><strong>Conclusions: </strong>ML models have the potential to enhance HIV risk assessment and inform targeted prevention strategies for high-risk populations.</p>","PeriodicalId":9050,"journal":{"name":"BMJ Health & Care Informatics","volume":"32 1","pages":""},"PeriodicalIF":4.1,"publicationDate":"2025-05-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12083282/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144085604","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Cary J G Oberije, Rachel Currie, Alice Leaver, Alan Redman, William Teh, Nisha Sharma, Georgia Fox, Ben Glocker, Galvin Khara, Jonathan Nash, Annie Y Ng, Peter D Kecskemethy
{"title":"Assessing artificial intelligence in breast screening with stratified results on 306 839 mammograms across geographic regions, age, breast density and ethnicity: A Retrospective Investigation Evaluating Screening (ARIES) study.","authors":"Cary J G Oberije, Rachel Currie, Alice Leaver, Alan Redman, William Teh, Nisha Sharma, Georgia Fox, Ben Glocker, Galvin Khara, Jonathan Nash, Annie Y Ng, Peter D Kecskemethy","doi":"10.1136/bmjhci-2024-101318","DOIUrl":"10.1136/bmjhci-2024-101318","url":null,"abstract":"<p><strong>Objectives: </strong>Evaluate an Artificial Intelligence (AI) system in breast screening through stratified results across age, breast density, ethnicity and screening centres, from different UK regions.</p><p><strong>Methods: </strong>A large-scale retrospective study evaluating two variations of using AI as an independent second reader in double reading was executed. Stratifications were conducted for clinical and operational metrics. Data from 306 839 mammography cases screened between 2017 and 2021 were used and included three different UK regions.The impact on safety and effectiveness was assessed using clinical metrics: cancer detection rate and positive predictive value, stratified according to age, breast density and ethnicity. Operational impact was assessed through reading workload and recall rate, measured overall and per centre.Non-inferiority was tested for AI workflows compared with human double reading, and when passed, superiority was tested. AI interval cancer (IC) flag rate was assessed to estimate additional cancer detection opportunity with AI that cannot be assessed retrospectively.</p><p><strong>Results: </strong>The AI workflows passed non-inferiority or superiority tests for every metric across all subgroups, with workload savings between 38.3% and 43.7%. The AI standalone flagged 41.2% of ICs overall, ranging between 33.3% and 46.8% across subgroups, with the highest detection rate for dense breasts.</p><p><strong>Discussion: </strong>Human double reading and AI workflows showed the same performance disparities across subgroups. The AI integrations maintained or improved performance at all metrics for all subgroups while achieving significant workload reduction. Moreover, complementing these integrations with AI as an additional reader can improve cancer detection.</p><p><strong>Conclusion: </strong>The granularity of assessment showed that screening with the AI-system integrations was as safe as standard double reading across heterogeneous populations.</p>","PeriodicalId":9050,"journal":{"name":"BMJ Health & Care Informatics","volume":"32 1","pages":""},"PeriodicalIF":4.1,"publicationDate":"2025-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12083354/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144075670","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Claire Tochel, Miguel O Bernabeu, Alice McTrusty, Andrew J Tatham, Emma Pead, Fiona Buckmaster, Jonathan Penny, Tom MacGillivray, Malihe Javidi, Heather Anderson, Ana Paula Rubio, Robert Wallace, Jamie B R Kidd, Ruairidh MacLeod, Niall Strang, Baljean Dhillon
{"title":"SCONe: a community-acquired retinal image repository enabling ocular, cardiovascular and neurodegenerative disease prediction.","authors":"Claire Tochel, Miguel O Bernabeu, Alice McTrusty, Andrew J Tatham, Emma Pead, Fiona Buckmaster, Jonathan Penny, Tom MacGillivray, Malihe Javidi, Heather Anderson, Ana Paula Rubio, Robert Wallace, Jamie B R Kidd, Ruairidh MacLeod, Niall Strang, Baljean Dhillon","doi":"10.1136/bmjhci-2024-101236","DOIUrl":"10.1136/bmjhci-2024-101236","url":null,"abstract":"<p><strong>Objectives: </strong>To safeguard Scotland's community-acquired retinal images (colour fundus photographs) in a secure, centrally held repository and support a variety of research including ocular, neurodegenerative and systemic disease prediction.</p><p><strong>Design: </strong>Retinal images captured in optometry practices linked to national, routinely collected, longitudinal healthcare data.</p><p><strong>Setting: </strong>Community optometry and the Public Health Scotland National Safe Haven.</p><p><strong>Participants: </strong>Adults (mostly aged 60+) who have attended their optometrist since 2006 for an eye examination during which a retinal image was captured.</p><p><strong>Main outcome measures: </strong>Successful retrieval of linkable colour fundus photographs from systems in use in practice and delivery to the Safe Haven for linkage and secure storage.</p><p><strong>Results: </strong>Scottish Collaborative Optometry-Ophthalmology Network e-research (SCONe) currently contains over 367 000 retinal images matched to over 36 000 patients. Healthcare data (hospital inpatient and outpatient, general ophthalmic, death and prescribing) records were retrieved for patients with one or more images, providing demographic and healthcare information for 95% of the cohort. The linked data allow the application of condition labels or phenotypes at specific points in time, facilitating research into retinal manifestations of vascular and neural diseases. The cohort is representative of the Scottish 60+ population in terms of sex (54% female), and there is a slight over-representation of people of black, Asian and minority ethnic groups (2% vs 1%) and those living in areas of lower deprivation (30% vs 16% in lowest two categories). Early research work has begun and is focusing on ocular and neurodegenerative disease prediction.</p><p><strong>Conclusions: </strong>The SCONe retinal image repository has been successfully established. We believe it offers enormous potential to support research into earlier detection of disease.</p>","PeriodicalId":9050,"journal":{"name":"BMJ Health & Care Informatics","volume":"32 1","pages":""},"PeriodicalIF":4.1,"publicationDate":"2025-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12083377/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144075674","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Jie Pan, Seungwon Lee, Cheligeer Cheligeer, Bing Li, Guosong Wu, Catherine A Eastwood, Yuan Xu, Hude Quan
{"title":"Assessing the validity of ICD-10 administrative data in coding comorbidities.","authors":"Jie Pan, Seungwon Lee, Cheligeer Cheligeer, Bing Li, Guosong Wu, Catherine A Eastwood, Yuan Xu, Hude Quan","doi":"10.1136/bmjhci-2024-101381","DOIUrl":"10.1136/bmjhci-2024-101381","url":null,"abstract":"<p><strong>Objectives: </strong>Administrative data are commonly used to inform chronic disease prevalence and support health informatic research. This study assessed the validity of coding comorbidities in the International Classification of Diseases, 10th Revision (ICD-10) administrative data.</p><p><strong>Methods: </strong>We analysed three chart review cohorts (4008 patients in 2003, 3045 in 2015 and 9024 in 2022) in Alberta, Canada. Nurse reviewers assessed the presence of 17 clinical conditions using a consistent protocol. The reviews were linked with administrative data using unique patient identifiers. We compared the accuracy in coding comorbidity by ICD-10, using chart review data as the reference standard.</p><p><strong>Results: </strong>Our findings showed that the mean difference in prevalence between chart reviews and ICD-10 for these 17 conditions was 2.1% in 2003, 7.6% in 2015 and 6.3% in 2022. Some conditions were relatively stable, such as diabetes (1.9%, 2.1% and 1.1%) and metastatic cancer (0.3%, 1.1% and 0.4%). For these 17 conditions, the sensitivity ranged from 39.6-85.1% in 2003, 1.3%-85.2% in 2015 and 3.0-89.7% in 2022. The C-statistics for predicting in-hospital mortality using comorbidities by ICD-10 were 0.84 in 2003, 0.81 in 2015 and 0.78 in 2022.</p><p><strong>Discussion: </strong>The undercoding could be primarily due to the increase in hospital patient volumes and the limited time allocated to coding specialists. There is the potential to develop artificial intelligence methods based on electronic health records to support coding practices and improve data quality.</p><p><strong>Conclusion: </strong>Comorbidities were increasingly undercoded over 20 years. The validity of ICD-10 decreased but remained relatively stable for certain conditions mandated for coding. The undercoding exerted minimal impact on in-hospital mortality prediction.</p>","PeriodicalId":9050,"journal":{"name":"BMJ Health & Care Informatics","volume":"32 1","pages":""},"PeriodicalIF":4.1,"publicationDate":"2025-05-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12083369/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143972136","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}