Table 3. Hazard Ratios of Death in Relation to the Number of Transfused Units From Female, Previously Pregnant, and Sex-Discordant Donorsa.
| Outcomes | HR per Unit Transfused (95% CI)b | Units of Each Exposure Categoryc | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 1-2 | 3-4 | 5-6 | ≥7 | ||||||||||||
| Deaths | Person-Years | HR (95% CI) | Deaths | Person-Years | HR (95% CI) | Deaths | Person-Years | HR (95% CI) | Deaths | Person-Years | HR (95% CI) | Deaths | Person-Years | HR (95% CI) | ||
| KPNC (n = 34 662) | ||||||||||||||||
| Female | 0.99 (0.98-1.01) | 2932 | 18 431 | 1 [Reference] |
5359 | 32 878 | 1.02 (0.97-1.07) | 1453 | 6377 | 1.01 (0.93-1.10) | 546 | 1723 | 1.02 (0.90-1.16) | 652 | 1511 | 0.92 (0.78-1.07) |
| No. of transfusions from previously pregnant female donorsd | 1.01 (1.00-1.01) | 8281 | 48 857 | 1 [Reference] |
2215 | 10 671 | 1.10 (0.95-1.28) | 302 | 973 | 1.16 (0.92-1.46) | 84 | 230 | 1.02 (0.66-1.55) | 60 | 189 | 1.28 (0.91-1.81) |
| No. of sex-discordant transfusions | 1.01 (1.00-1.02) | 2532 | 16 480 | 1 [Reference] |
5497 | 33 563 | 1.08 (1.03-1.14) | 1544 | 7056 | 1.05 (0.97-1.14) | 607 | 2032 | 1.14 (1.01-1.29) | 762 | 1790 | 1.02 (0.88-1.19) |
| SCANDAT (n = 918 996) | ||||||||||||||||
| No. of transfusions from female donors | 1.00 (0.99-1.00) | 89 197 | 771 834 | 1 [Reference] |
214 660 | 1 596 273 | 0.99 (0.99-1.00) | 81 949 | 415 685 | 0.99 (0.97-1.00) | 35 928 | 138 516 | 0.99 (0.97-1.01) | 53 515 | 156 369 | 0.98 (0.95-1.01) |
| No. of transfusions from parous female donors | 1.00 (1.00-1.01) | 159 509 | 1 316 102 | 1 [Reference] |
210 706 | 1 386 682 | 1.00 (0.99-1.01) | 58 258 | 243 202 | 1.00 (0.98-1.01) | 21 611 | 69 294 | 1.03 (1.00-1.06) | 25 165 | 63 397 | 1.00 (0.95-1.04) |
| No. of sex-discordant transfusions | 1.00 (1.00-1.01) | 59 604 | 484 952 | 1 [Reference] |
210 369 | 658 873 | 1.00 (0.99-1.01) | 93 278 | 533 747 | 1.00 (0.99-1.02) | 43 509 | 186 742 | 1.02 (1.00-1.04) | 68 489 | 214 363 | 1.00 (0.98-1.03) |
Abbreviations: HR, hazard ratio; KPNC, Kaiser Permanente Northern California; REDS-III, Recipient Epidemiology and Donor Evaluation Study-III; SCANDAT, Scandinavian Donations and Transfusions.
All analyses were based on the full cohorts of patients who had all received at least 1 red blood cell transfusion. Each row in the table represents a separate statistical model. Patients who received no red blood cell units from female donors thus implicitly received at least 1 red blood cell unit from a male donor; patients who received at least 1 unit from a female donor may also have received red blood cell units from a male donor. Absolute rates for categorized estimates are available in eTable 2 in the Supplement.
The HRs per unit transfused were computed by fitting models where each risk variable was included as a log-linear term.
For each exposure variable, comparisons should be done across so that, for example, in the KPNC cohort, the risk of death increased with 0.99 for each additional unit from a female donor, and that recipients of ≥7 units from a female donor had an HR of 0.92 (95% CI, 0.78-1.07) compared with to recipients of no units from female donors.
Due to missing data on donor pregnancy in the KPNC and REDS-III cohorts, analyses for the association between the number of units from previously pregnant female donors and risk of death were conducted using multiple imputation. Multiple imputation was used to account for 16.2% of transfusions from female donors in whom the pregnancy status was unknown in the KPNC cohort and 13.6% in the REDS-III cohort.