REASSESSING POSITIVE RESULTS: DETECTION BIAS IN EPIDEMIOLOGIC DEMENTIA STUDIES OF ELECTRONIC HEALTH RECORD DATA
2024

Detection Bias in Dementia Studies Using Health Records

Sample size: 404620 publication Evidence: moderate

Author Information

Author(s): Wang Jingxuan, Choi Minhyuk, Zimmerman Scott, Chen Ruijia, Schaefer Cathy, Blacker Deborah, Glymour M Maria

Primary Institution: University of California San Francisco

Hypothesis

Are electronic health record data in dementia studies affected by detection bias?

Conclusion

The study found that certain health conditions are associated with a temporary increase in dementia diagnoses, likely due to detection bias.

Supporting Evidence

  • 298,784 participants from the UK Biobank and 105,836 from All of Us were included in the study.
  • Three health conditions were evaluated for their association with dementia diagnoses.
  • The association with dementia was strongest in the first year after exposure to the health conditions.

Takeaway

If you go to the doctor for something like a kidney stone, you might get diagnosed with dementia more often right after, but that's probably because doctors are paying more attention to you.

Methodology

The study used Cox models to evaluate associations of three negative control exposures with dementia in participants aged 55 and older.

Potential Biases

Detection bias may affect the results due to increased healthcare contact.

Limitations

The influence of the studied conditions on dementia pathology is likely to be small or null.

Participant Demographics

Participants were aged 55 and older without baseline dementia.

Digital Object Identifier (DOI)

10.1093/geroni/igae098.0530

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication