Health Data for Linguistic Minority Group Research in Canada: Proof-of-Concept Centralized Health Care Metadata...

Area of Science:

• Health Services Research
• Data Science
• Public Health

Background:

• Language barriers between Canadian patients and healthcare providers are linked to adverse health outcomes, including compromised patient safety, misdiagnosis, and increased mortality.
• Limited research exists on language as a social determinant of health due to fragmented Canadian health data across multiple jurisdictions.
• Existing data infrastructure challenges hinder researchers' ability to identify, locate, and utilize relevant health data.

Purpose of the Study:

• To develop a proof-of-concept metadata repository (MDR) for Canadian health data at the variable level.
• To identify and label language-related variables within the Canadian health data MDR.
• To create an interactive, public-facing web application for searching the MDR.

Main Methods:

• Collected metadata from five Canadian health data sources (four provincial, one national survey).
• Developed a bottom-up approach to label language-related variables using search algorithms and consensus screening.
• Built an openly accessible web application using the Shiny framework in R (healthdatadictionary.ca) for MDR exploration.

Main Results:

• A total of 850,343 variables were aggregated into the repository.
• The majority of variables originated from Ontario (83.7%) and Manitoba (11.5%) provincial data holdings.
• Over a quarter of the variables (25.1%, n=213,696) were identified as language-related.

Conclusions:

• A national metadata repository offers a significant opportunity for Canadian researchers to utilize comprehensive health administrative data.
• This pilot study demonstrates the feasibility of a bottom-up approach to contribute to a national MDR.
• Collaboration between provincial and federal agencies is ideal for a comprehensive national MDR, but bottom-up efforts are valuable.