Comparison of Patient Panel and Area-Level Characteristics between Prescribers with High and Low PIM Prescription Rates
| Low Prescribers (Q1–Q3)a | High Prescribers (Q4)b | Difference P-Value | |||
|---|---|---|---|---|---|
| Mean | SD | Mean | SD | ||
| Patient panel characteristics | |||||
| Patient average age | 71.5 | 4.2 | 70.9 | 4.0 | .000 |
| Proportion female | 0.611 | 0.099 | 0.610 | 0.099 | .034 |
| Proportion of patients < age 65 y | 0.185 | 0.130 | 0.191 | 0.133 | .000 |
| High proportion (> 0.74) of white | 0.668 | 0.471 | 0.774 | 0.418 | .000 |
| High proportion (> 0.35) of duals | 0.349 | 0.477 | 0.259 | 0.438 | .000 |
| Patient average HCC risk score | 1.305 | 0.406 | 1.223 | 0.327 | .000 |
| Area-level characteristics | |||||
| PCP-to-specialist ratio per 100,000 residents | 0.575 | 0.070 | 0.566 | 0.066 | .000 |
| PCP per 100,000 residents | 74.5 | 12.2 | 72.8 | 11.8 | .000 |
| Average Medicare Spending | 9,520 | 1189 | 9,755 | 1197 | .000 |
| Number of PCP-year observations | 209,901 | 69,957 | |||
| Number of PCPs* | 95,179 | 40,339 | |||
PCP, primary care physician; PIM, potentially inappropriate medication; Q, quartile; HCC, hierarchical condition category; SD, standard deviation.
Data obtained from 2013–2015 Medicare Part D PUF and 2015 American Medical Association Masterfile.
↵a Low prescribers included PCPs with PIM prescription rates in Q1–Q3.
↵b High prescribers were PCPs with PIM prescriptions rates ranked in the highest quartile, Q4.
↵* The total number of PCPs (111,461) in the sample is less than the total number of PCPs in the low and high prescriber groups (135,518). This is due to some PCPs being in different groups across multiple years.