Test of homogeneity of binary data with explanatory variables

Area of Science:

• Epidemiology
• Biostatistics
• Gerontology

Background:

• Disease risk can appear uneven across different populations.
• This apparent heterogeneity may be due to individual risk factors not equally present in all groups.
• Existing methods may not adequately adjust for these subject-specific factors.

Purpose of the Study:

• To develop and validate a score test for homogeneity of disease risk.
• To assess if apparent geographical heterogeneity of cognitive impairment in the elderly persists after adjusting for known risk factors.
• To explore the test's utility in studying familial aggregation of diseases.

Main Methods:

• Proposed a score test for homogeneity based on a random-effect logistic regression model.
• The test allows for adjustment for known disease risk factors.
• Employed simple computations, augmenting conventional logistic regression.

Main Results:

• Applied the score test to a study of 3,318 elderly subjects across 75 parishes regarding cognitive impairment.
• Demonstrated that apparent geographical heterogeneity in cognitive impairment diminished upon accounting for subject-specific risk factors.
• The proposed test requires minimal additional computation beyond standard logistic regression.

Conclusions:

• The developed score test effectively adjusts for subject-specific risk factors, clarifying disease heterogeneity.
• Apparent geographical variations in cognitive impairment among the elderly were explained by individual risk factors.
• The score test is a valuable tool for epidemiological studies, including those on disease aggregation within families.