Table 2.
Physician and Patient Panel Characteristics Associated with Physician-Physician Relationships
| Unadjusted proportion of Dyads/Mean Difference with ties | Unadjusted proportion of Dyads/Mean Difference without ties | Unadjusted Rate Ratio+ | Adjusted Rate Ratio*+ | |
|---|---|---|---|---|
| Physician Characteristics (Differences) | 92.2 | 7.8 | -- | -- |
|
| ||||
| Sex | ||||
| Male-male | 65.1 | 54.6 | 1.68 (1.68, 1.69) | 1.32 (1.32, 1.32) |
| Female-Female | 3.8 | 6.5 | .72 (0.71,0.72) | .79 (.78, .79) |
| Male-Female | 29.1 | 36.8 | 1.00 (ref) | 1.00 (ref) |
| Difference in Age (mean) Specialty | 11.5 | 12.5 | .80 (.80, .80) | .88 (.88, .88) |
| PCP-PCP | 10.1 | 15.8 | .77 (.76, .77) | .62 (.62, .62) |
| PCP-Medical | 28.0 | 27.1 | 1.00 (ref) | 1.00 (ref) |
| PCP-Surgical | 17.5 | 20.3 | .72 (.72, .72) | .65 (.65, .65) |
| Medical-Medical | 16.9 | 12.1 | 1.52 (1.52, 1.53) | 1.36 (1.36, 1.36) |
| Medical-Surgical | 20.7 | 17.9 | .94 (.94, .95) | .90 (.89, .90) |
| Surgical-surgical | 7.0 | 6.7 | .76 (.76, .77) | .66 (.66, .66) |
| Distance (Mean) | 13.2 | 24.2 | .99 (.99, .99) | .98 (.98, .98) |
| Different Hospital (%) | 69.2 | 96.0 | .07 (.07, .07) | .12 (.12, .12) |
| Completed medical school at different medical school | 6.1 | 3.6 | .53 (.53, .53) | .99 (.99, .99) |
| Completed residency at different institution | 5.3 | 2.7 | .55 (.54, .55) | .88 (.88, .89) |
| Practice Style (ETG Intensity) | .29 | .31 | .91 (.91, .91) | .93 (.92, .93) |
|
| ||||
| Patient Panel Characteristics+ | ||||
|
| ||||
| Difference in % White | 11.5 | 20.2 | .72 (.72, .72) | .89 (.89, .89) |
| Difference in % Black | 8.8 | 14.0 | .75 (.75, .75) | .92 (.92, .92) |
| Difference in % Hispanic | 2.9 | 5.3 | .59 (.59, .59) | .75 (.75, .76) |
| Difference in %Female | 13.0 | 15.6 | .80 (.80, .81) | .86 (.86, .86) |
| Difference in % Medicaid | 15.3 | 24.4 | .69 (.69, .69) | .86 (.86, .86) |
| Difference in Age (mean) | 4.1 | 5.4 | .42 (.42, .42) | .75 (.75, .75) |
| HCC Score (mean) | 1.0 | 1.1 | .78 (.78, .78) | .93 (.93, .93) |
*
Adjusted rate ratios and p-values were calculated using a negative binomial regression model, adjusting for all variables in the table. Rate ratios are used because the outcome (number of shared patients) is a count rather than binary. Results are similar when a binary outcome variable was analyzed using logistic regression. All p-values are less than .001
+
Rate ratios reflect the increase in the expected number of shared patients (and thus likelihood of a true information sharing relationship) for every 10% point difference in patient panel characteristics (not applicable to HCC score). Ref signifies reference category