Misclassification and the use of register-based indicators for depression.

Area of Science:

• Epidemiology
• Public Health
• Mental Health Research

Background:

• Register data are increasingly used as depression indicators.
• Potential biases in register-based data, such as differential misclassification, can impact study findings.
• Understanding these biases is crucial for accurate epidemiological research.

Purpose of the Study:

• To quantify differential misclassification in register-based depression indicators by gender, age, and social group.
• To assess the impact of misclassification on odds ratios derived from hospital and antidepressant treatment data.
• To provide insights for improving the accuracy of register-based mental health research.

Main Methods:

• Linked a cross-sectional survey of Danish adults (aged 40 and 50) with population-based registers.
• Analyzed misclassification using survey data compared to register data on major depression.
• Employed the Rothman and Greenland method for misclassification analysis.

Main Results:

• Found significant differential misclassification in depression indicators.
• Adjusting for misclassification substantially altered odds ratios: women's odds ratios for hospital treatment decreased from 2.18 to 1.00 and for antidepressants from 1.70 to 1.10.
• Lower social group odds ratios increased significantly, from 1.18 to 3.52 for hospital treatment and 1.35 to 2.32 for antidepressants.
• Age-related odds ratios remained largely unchanged.

Conclusions:

• Register-based hospital and antidepressant treatment data are subject to differential misclassification.
• These biases vary by gender and social group, necessitating careful consideration in research.
• Accurate interpretation of register-based depression indicators requires adjustment for potential misclassification.