What does comparing two time windows in the same person add to a fracture study?
Each person supplies the comparison
The authors used Taiwanese insurance records to examine fractures among 6,010 older adults with zolpidem or benzodiazepine use. In their case-crossover design, exposure close to a fracture was compared with exposure in earlier windows for the same person. That approach holds stable personal characteristics constant.
Things that change over time still matter
Recent illness, a change in mobility or another medication can differ between the windows. The overall association for zolpidem or the benzodiazepine class did not remain statistically significant after adjustment for measured medications and comorbidities. The result depends on the window and adjustment choices.
Use the design to ask better timing questions
For a particular fracture, reconstruct when the person took the medication, what changed recently and what was documented about the injury mechanism. Claims can suggest when a drug was available. They leave uncertainty about the timing of an actual dose.
Source paper
Tang et al., PLOS ONE (2015) · Full text
