Health service research definition builder: An R Shiny application for exploring diagnosis codes associated with services reported in routinely collected health data.
PLoS One
; 18(1): e0266154, 2023.
Article
em En
| MEDLINE
| ID: mdl-36634112
ABSTRACT
Many administrative health data-based studies define patient cohorts using procedure and diagnosis codes. The impact these criteria have on a study's final cohort is not always transparent to co-investigators or other audiences if access to the research data is restricted. We developed a SAS and R Shiny interactive research support tool which generates and displays the diagnosis code summaries associated with a selected medical service or procedure. This allows non-analyst users to interrogate claims data and groupings of reported diagnosis codes. The SAS program uses a tree classifier to find associated diagnosis codes with the service claims compared against a matched, random sample of claims without the service. Claims are grouped based on the overlap of these associated diagnosis codes. The Health Services Research (HSR) Definition Builder Shiny application uses this input to create interactive table and graphics, which updates estimated claim counts of the selected service as users select inclusion and exclusion criteria. This tool can help researchers develop preliminary and shareable definitions for cohorts for administrative health data research. It allows an additional validation step of examining frequency of all diagnosis codes associated with a service, reducing the risk of incorrect included or omitted codes from the final definition. In our results, we explore use of the application on three example services in 2016 US Medicare claims for patients aged over 65 knee arthroscopy, spinal fusion procedures and urinalysis. Readers can access the application at https//kelsey209.shinyapps.io/hsrdefbuilder/ and the code at https//github.com/kelsey209/hsrdefbuilder.
Texto completo:
1
Base de dados:
MEDLINE
Assunto principal:
Medicare
/
Dados de Saúde Coletados Rotineiramente
Tipo de estudo:
Diagnostic_studies
/
Risk_factors_studies
Limite:
Aged
/
Humans
País/Região como assunto:
America do norte
Idioma:
En
Revista:
PLoS One
Assunto da revista:
CIENCIA
/
MEDICINA
Ano de publicação:
2023
Tipo de documento:
Article
País de afiliação:
Estados Unidos