An Ontology to Improve Transparency in Case Definition and Increase Case Finding of Infectious Intestinal Disease: Database Study in English General Practice
Simon de Lusignan,Stacy Shinneman,Ivelina Yonova,Jeremy van Vlymen,Alex J Elliot,Frederick Bolton,Gillian E Smith,Sarah O'Brien
DOI: https://doi.org/10.2196/medinform.7641
2017-09-28
Abstract:Background: Infectious intestinal disease (IID) has considerable health impact; there are 2 billion cases worldwide resulting in 1 million deaths and 78.7 million disability-adjusted life years lost. Reported IID incidence rates vary and this is partly because terms such as "diarrheal disease" and "acute infectious gastroenteritis" are used interchangeably. Ontologies provide a method of transparently comparing case definitions and disease incidence rates. Objective: This study sought to show how differences in case definition in part account for variation in incidence estimates for IID and how an ontological approach provides greater transparency to IID case finding. Methods: We compared three IID case definitions: (1) Royal College of General Practitioners Research and Surveillance Centre (RCGP RSC) definition based on mapping to the Ninth International Classification of Disease (ICD-9), (2) newer ICD-10 definition, and (3) ontological case definition. We calculated incidence rates and examined the contribution of four supporting concepts related to IID: symptoms, investigations, process of care (eg, notification to public health authorities), and therapies. We created a formal ontology using ontology Web language. Results: The ontological approach identified 5712 more cases of IID than the ICD-10 definition and 4482 more than the RCGP RSC definition from an initial cohort of 1,120,490. Weekly incidence using the ontological definition was 17.93/100,000 (95% CI 15.63-20.41), whereas for the ICD-10 definition the rate was 8.13/100,000 (95% CI 6.70-9.87), and for the RSC definition the rate was 10.24/100,000 (95% CI 8.55-12.12). Codes from the four supporting concepts were generally consistent across our three IID case definitions: 37.38% (3905/10,448) (95% CI 36.16-38.5) for the ontological definition, 38.33% (2287/5966) (95% CI 36.79-39.93) for the RSC definition, and 40.82% (1933/4736) (95% CI 39.03-42.66) for the ICD-10 definition. The proportion of laboratory results associated with a positive test result was 19.68% (546/2775). Conclusions: The standard RCGP RSC definition of IID, and its mapping to ICD-10, underestimates disease incidence. The ontological approach identified a larger proportion of new IID cases; the ontology divides contributory elements and enables transparency and comparison of rates. Results illustrate how improved diagnostic coding of IID combined with an ontological approach to case definition would provide a clearer picture of IID in the community, better inform GPs and public health services about circulating disease, and empower them to respond. We need to improve the Pathology Bounded Code List (PBCL) currently used by laboratories to electronically report results. Given advances in stool microbiology testing with a move to nonculture, PCR-based methods, the way microbiology results are reported and coded via PBCL needs to be reviewed and modernized.