Adjustment for missing confounders in studies based on observational databases: 2-stage calibration combining...

Area of Science:

• Epidemiology
• Biostatistics

Background:

• Observational studies face confounding bias due to missing data on factors like smoking and alcohol consumption.
• Administrative databases often lack detailed information on crucial confounders.

Purpose of the Study:

• To propose and validate a 2-stage calibration (TSC) method for adjusting confounding bias from missing data in observational studies.
• To assess the association between chronic obstructive pulmonary disease (COPD) and herpes zoster (HZ) while accounting for smoking and alcohol consumption.

Main Methods:

• Developed a 2-stage calibration (TSC) method integrating external validation data with main study data.
• Utilized propensity scores to summarize confounding information from both datasets.
• The TSC method's validity does not depend on specific measurement error models.

Main Results:

• Applied the TSC method to a study on COPD and HZ using administrative and National Health Interview Survey data.
• The adjusted odds ratio for HZ associated with COPD was 1.91 (95% CI: 1.62, 2.26).
• This indicates a significant association after accounting for smoking and alcohol consumption.

Conclusions:

• The 2-stage calibration (TSC) method effectively adjusts for confounding bias caused by missing confounders.
• The findings highlight a statistically significant association between COPD and HZ, emphasizing the importance of accounting for lifestyle factors.