Table 2.
Average Network Measures by Selected Hospital Characteristics
| n | % | Median Degree per 100 Shared Patients | PCP Relative Centrality* | |||||
|---|---|---|---|---|---|---|---|---|
| Mean | (SD) | p-value | Mean | (SD) | p-value | |||
| All | 528 | 100% | 208 | (107) | N/A | 0.90 | (0.68) | N/A |
| Number of Beds | ||||||||
| Smaller (<123) | 176 | 33% | 155 | (57) | <0.001 | 1.11 | (0.87) | <0.001 |
| Mid-size (123–265) | 176 | 33% | 187 | (86) | 0.79 | (0.54) | ||
| Large (>265) | 176 | 33% | 281 | (124) | 0.80 | (0.54) | ||
| Urban/Rural Location | ||||||||
| Urban | 332 | 63% | 253 | (110) | <0.001 | 0.78 | (0.49) | <0.001 |
| Suburban/Rural/Isolated | 196 | 37% | 131 | (36) | 1.11 | (0.88) | ||
| % Medicare Admissions | ||||||||
| Low (<42%) | 175 | 33% | 270 | (126) | <0.001 | 0.90 | (0.72) | 0.21 |
| Medium (42%–50%) | 183 | 35% | 190 | (88) | 0.83 | (0.57) | ||
| High (>50%) | 170 | 32% | 163 | (68) | 0.97 | (0.73) | ||
| % Medicaid Admissions | ||||||||
| Low (<15%) | 188 | 36% | 197 | (90) | 0.008 | 0.84 | (0.62) | 0.24 |
| Medium (15%–20%) | 156 | 30% | 197 | (103) | 0.89 | (0.63) | ||
| High (>20%) | 184 | 35% | 228 | (123) | 0.97 | (0.77) | ||
| Teaching Status | ||||||||
| None | 345 | 65% | 170 | (77) | <0.001 | 0.95 | (0.75) | 0.002 |
| Minor | 128 | 24% | 236 | (98) | 0.73 | (0.48) | ||
| Major | 55 | 10% | 378 | (106) | 0.97 | (0.56) | ||
| Profit Governance | ||||||||
| Not For Profit | 348 | 66% | 213 | (109) | 0.10 | 0.86 | (0.64) | 0.30 |
| Public | 90 | 17% | 186 | (108) | 0.98 | (0.74) | ||
| For Profit | 90 | 17% | 209 | (95) | 0.96 | (0.76) | ||
| US Region | ||||||||
| Northeast | 127 | 24% | 242 | (116) | <0.001 | 0.79 | (0.55) | 0.004 |
| South | 207 | 39% | 182 | (88) | 1.02 | (0.78) | ||
| Midwest | 101 | 19% | 219 | (119) | 0.93 | (0.78) | ||
| West | 93 | 18% | 206 | (106) | 0.77 | (0.41) | ||
p-values calculated using one-way analysis of variance.
*
For PCP relative centrality, outlier values were set equal to 1st and 99th percentile values. In addition, PCP centrality cannot be lower than 0, so the large SD also reflects the degree of positive skew in the data.