Cross-cultural validation of the stroke riskometer using generalizability theory

Area of Science:

• Health Informatics
• Epidemiology
• Psychometrics

Background:

• The Stroke Riskometer app offers personalized stroke risk assessment globally.
• Its cross-cultural reliability has not been rigorously evaluated.
• Generalizability Theory (G-Theory) is ideal for assessing score reliability across diverse contexts.

Purpose of the Study:

• To evaluate the reliability and generalizability of the Stroke Riskometer's stroke risk assessment scores across 13 countries.
• To determine if the app's current form is optimal for international use.
• To assess the suitability of the Stroke Riskometer for cross-cultural comparisons.

Main Methods:

• Applied Generalizability Theory (G-Theory) to data from 1300 participants across 13 countries.
• Utilized a two-facet nested observational design (person by item nested in country).
• Conducted D-study analyses to assess reliability coefficients (G relative and absolute).

Main Results:

• The Stroke Riskometer demonstrated strong reliability across countries (G relative = 0.84, G absolute = 0.82).
• D-study analysis confirmed optimal reliability in the current app format, requiring no modifications.
• Scores are generalizable across populations and countries, supporting cross-cultural comparisons.

Conclusions:

• The Stroke Riskometer is a reliable tool for assessing stroke risk internationally.
• The app's scores are generalizable, enabling valid cross-cultural comparisons.
• Further longitudinal studies are recommended to assess temporal reliability.