Estimation Results for the Share of Provider Effects in Antibiotic Prescribing
| Number of Prescriptions, Two-Way Fixed Effects Estimationa | ||||
|---|---|---|---|---|
| All Antibiotics | Penicillins | Second-Line | Other | |
| 0.494*** | 0.435*** | 0.828*** | 0.485*** |
| (0.033) | (0.034) | (0.036) | (0.072) | |
| Event dummiesb | Yes | Yes | Yes | Yes |
| Observations | 7,789,908 | 7,789,908 | 7,789,908 | 7,789,908 |
| Groups (patients) | 1,371,604 | 1,371,604 | 1,371,604 | 1,371,604 |
Notes: This table reports the average share of antibiotic prescribing differences between clinics that is attributable to provider practice style differences, the coefficient of
. The term Δi denotes the difference in mean antibiotic prescribing between patient i’s assigned sets of physicians, and is estimated by , the leave-one-out estimator of antibiotic prescribing based on untreated patients. The term Dit represents the post-treatment indicator. Standard errors are calculated using a bootstrap with 50 repetitions at the patient level. ***p < 0.01.↵aTwo-way fixed effects estimation with calendar year fixed effects and patient fixed effects.
↵bEvent dummies include an indicator for treatment onset (relative year 0) and post-treatment.
