Table 2.
Risk ratio results of multivariable regression models of bloodstream infection rates in New England hemodialysis facilities throughout 2015 and 2016
| Patients with any access | Patients with Catheters | Patients without Catheters | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Hemodialysis Facility Subgroupsa,b | No. of BSIs | Patient-Months | Risk Ratiosa (95% CI) | No. of BSIs | Patient-Months | Risk Ratiosa (95% CI) | No. of BSIs | Patient-Months | Risk Ratiosa (95% CI) |
| Batch-submitting organizationc | P=0.02 | P≤0.001 | P=0.98 | ||||||
| Provider 1 | 1.23 (0.87 to 1.73) | 1.47 (1.05 to 2.05) | 0.98 (0.50 to 1.92) | ||||||
| Provider 2 | 1.23 (0.97 to 1.56) | 1.44 (1.07 to 1.93) | 1.01 (0.76 to 1.36) | ||||||
| Provider 3 | 1.00 (Referent) | 1.00 (Referent) | 1.00 (Referent) | ||||||
| Provider 4 | 1.41 (1.08 to 1.86) | 1.61 (1.18 to 2.18) | 0.87 (0.54 to 1.39) | ||||||
| Provider 5 | 1.52 (1.15 to 2.02) | 1.91 (1.39 to 2.63) | 1.02 (0.70 to 1.48) | ||||||
| Hospital-based versus nonhospital-based facilities | P=0.28 | P=0.42 | P=0.10 | ||||||
| Hospital-based | 68 | 7647 | 1.21 (0.86 to 1.71) | 44 | 1427 | 1.23 (0.74 to 2.05) | 24 | 6220 | 1.46 (0.93 to 2.29) |
| Nonhospital-based | 1448 | 290,390 | 1.00 (Referent) | 881 | 48,640 | 1.00 (Referent) | 567 | 241,750 | 1.00 (Referent) |
| Percentage of patients with catheter per facility | P=0.68 | P≤0.001 | P=0.30 | ||||||
| <10% | 275 | 47,552 | 0.94 (0.74 to 1.18) | 139 | 3597 | 2.03 (1.42 to 2.91) | 136 | 43,955 | 1.27 (0.90 to 1.77) |
| 10%–14.9% | 357 | 77,992 | 0.93 (0.79 to 1.10) | 195 | 9759 | 1.41 (1.14 to 1.75) | 162 | 68,233 | 1.13 (0.86 to 1.46) |
| 15%–19.9% | 400 | 81,288 | 1.02 (0.87 to 1.19) | 261 | 14,131 | 1.36 (1.13 to 1.63) | 139 | 67,157 | 0.97 (0.74 to 1.28) |
| ≥20% | 484 | 91,205 | 1.00 (Referent) | 330 | 22,580 | 1.00 (Referent) | 154 | 68,625 | 1.00 (Referent) |
| Total no. of patients per facility | P=0.20 | P=0.24 | P=0.48 | ||||||
| <25 | 43 | 7305 | 1.02 (0.71 to 1.47) | 33 | 1676 | 1.00 (0.64 to 1.56) | 10 | 5629 | 0.70 (0.35 to 1.41) |
| 25–49 | 186 | 35,373 | 0.87 (0.68 to 1.13) | 125 | 6681 | 0.93 (0.67 to 1.28) | 61 | 28,692 | 0.79 (0.55 to 1.14) |
| 50–99 | 685 | 145,005 | 0.82 (0.67 to 1.00) | 410 | 24,329 | 0.79 (0.60 to 1.03) | 275 | 120,676 | 0.87 (0.68 to 1.11) |
| ≥100 | 602 | 110,354 | 1.00 (Referent) | 357 | 17,381 | 1.00 (Referent) | 245 | 92,973 | 1.00 (Referent) |
| Season of the year | P=0.50 | P=0.77 | P=0.36 | ||||||
| Spring | 359 | 74,329 | 0.92 (0.80 to 1.07) | 222 | 12,648 | 0.96 (0.79 to 1.17) | 137 | 61,681 | 0.85 (0.69 to 1.05) |
| Summer | 401 | 75,242 | 1.02 (0.88 to 1.17) | 244 | 12,337 | 1.07 (0.88 to 1.29) | 157 | 62,905 | 0.95 (0.77 to 1.16) |
| Fall | 373 | 75,466 | 0.94 (0.81 to 1.10) | 233 | 12,497 | 1.01 (0.83 to 1.23) | 140 | 62,969 | 0.84 (0.67 to 1.06) |
| Winter | 383 | 73,000 | 1.00 (Referent) | 226 | 12,585 | 1.00 (Referent) | 157 | 60,415 | 1.00 (Referent) |
BSI, bloodstream infection; 95% CI, 95% confidence interval.
Each column represents a separate log-binomial regression model that was fit using generalized estimating equation methods. Each model was adjusted for the following variables: facility catheter percentage, patients per facility, batch-submitting organizations, hospital- or nonhospital-based outpatient facility, and season of the year. When the model is calculated with the hospital-based outpatient facilities removed, the results are essentially the same, with only small changes in the second decimal place (P remains <0.001).
Number of BSIs, total patient-months, and adjusted risk ratio comparisons are shown for hemodialysis facility subgroups for all patients with any vascular access, patients with catheters, and patients without catheters.
For the three large dialysis organizations and the two independently owned facilities, we have maintained the same randomized order as Table 1 and omitted the number of bloodstream infections and patient-months to preserve confidentiality of the provider organizations. Among these groups of providers, the number of bloodstream infections ranged from 71 to 508 overall and from 43 to 287 in the patients with catheters.