Effect of personal health events on prescribing volume and behaviour in 25815 Finnish doctors

Physicians’ personal health experiences (i.e. diseases or medications) may shape the care they deliver, yet this relationship has never been studied at scale. Here we construct longitudinal disease and medication histories for all eligible Finnish doctors (N = 25,815), from 1998 until 2023, and link them to patients (N = 6.5 million) through filled prescriptions. Using staggered difference-in-differences models, we examined two effects: A. the impact of physicians’ medical diagnoses on their prescription volume, and B. the impact of physicians’ personal use of medications on their prescription behaviour towards those same medications. Doctors were healthier than patients overall, with lower incidence rate of diagnoses (mean IR of 1.66 vs 2.10) and higher use of medications (mean IR 4.55 vs 3.43) compared to their patients. Out of 195 diagnoses tested, we found that childbirth, breast cancer, and severe depression substantially reduced total prescription volume within the event year (by 68%, 32%, and 27% respectively). Finally, examining a total of 310 medications, we found 9 for which doctors’ personal use was associated with increased prescriptions of that same medication to patients after the usage. Rosuvastatin showed the largest absolute change: starting from a baseline of 2.6/1000 prescriptions in controls, after personal use doctors prescribed an average 6.8/1000 of it (95% CI: 5.7–8.0), representing a 2.6-fold relative increase. Frovatriptan had the largest relative increase, with personal users prescribing it at 4.6 times the rate of controls (95% CI: 2.9–6.2). Overall, effects tended to be larger among physicians with a low number of prescriptions for these medications before the event (bottom vs. top 10%). Moreover, the effects were similar in magnitude among self-prescribers vs. non-self-prescribers. These findings identify personal health experiences as a previously unquantified determinant of prescribing behaviour and variability.

Key messages

• Staggered difference-in-differences models reveal that physicians’ own health events are a measurable, previously unquantified determinant of prescribing behaviour and variability.

• Linking 25815 Finnish physicians to 6.5million patients, we examined how personal diagnoses affect prescription volume and how personal medication use shapes prescribing behaviour of those medications.