Table 3.
Multivariate Cox regression analysis assessing the influence of perioperative allogeneic blood transfusion and the number of units transfused on survival
| Blood transfusion (exposure/no exposure)* | Blood transfusion (per unit transfused)* | |||||
|---|---|---|---|---|---|---|
| Standard error | Hazard ratio (95% CI) | p value | Standard error | Hazard ratio (95% CI) | p value | |
| Age (years) | 0.004 | 1.016 (1.009–1.023)§ | < 0.001§ | 0.004 | 1.015 (1.008–1.022)§ | < 0.001§ |
| Modified Charlson Comorbidity Index | 0.024 | 1.054 (1.009–1.102)§ | 0.019§ | 0.024 | 1.057 (1.011–1.105)§ | 0.014§ |
| Preoperative hemoglobin level (g/dL†) | 0.028 | 0.937 (0.884–0.994)§ | 0.031§ | 0.027 | 0.950 (0.898–1.004) | 0.073 |
| Total estimated blood loss during surgery (mL†) | < 0.001 | 1.000 (1.000–1.000) | 0.521 | <0.001 | 1.000 (0.999–1.000) | 0.175 |
| Anesthesia time (minutes†) | < 0.001 | 1.002 (1.000–1.003) | 0.063 | <0.001 | 1.001 (1.000–1.003) | 0.074 |
| Duration of hospital admission (days) | 0.007 | 1.021 (1.007–1.036)§ | 0.003§ | 0.008 | 1.017 (1.003–1.032)§ | 0.021§ |
| BMI (kg/m2†) | 0.009 | 0.993 (0.976–1.010) | 0.421 | 0.009 | 0.994 (0.977–1.011) | 0.462 |
| Male sex | 0.089 | 1.035 (0.874–1.225) | 0.690 | 0.088 | 1.044 (0.884–1.232) | 0.612 |
| Poor prognosis tumor type‡ | 0.147 | 1.694 (1.429–2.008)§ | < 0.001§ | 0.148 | 1.713 (1.445–2.029)§ | < 0.001§ |
| Pathological fracture | 0.095 | 1.006 (0.836–1.210) | 0.950 | 0.093 | 0.983 (0.818–1.183) | 0.859 |
| Lower extremity | 0.108 | 0.991 (0.800–1.227) | 0.933 | 0.104 | 0.967 (0.783–1.195) | 0.758 |
| Other bone metastatic lesions | 0.123 | 1.205 (0.987–1.471) | 0.067 | 0.120 | 1.167 (0.955–1.427) | 0.132 |
| Visceral metastases | 0.138 | 1.584 (1.334–1.880)§ | < 0.001§ | 0.138 | 1.586 (1.337–1.882)§ | < 0.001§ |
| Previous local radiotherapy | 0.114 | 1.094 (0.892–1.341) | 0.389 | 0.114 | 1.097 (0.895–1.345) | 0.371 |
| Previous systemic therapy | 0.099 | 1.055 (0.877–1.226) | 0.571 | 0.098 | 1.042 (0.867–1.252) | 0.661 |
| Preoperative embolization of tumor | 0.301 | 1.153 (0.692–1.922) | 0.584 | 0.280 | 1.068 (0.639–1.786) | 0.801 |
| Operative treatment type | ||||||
| Intramedullary nailing | Reference value | Reference value | Reference value | Reference value | Reference value | Reference value |
| Endoprosthetic reconstruction | 0.124 | 1.126 (0.907–1.397) | 0.283 | 0.124 | 1.135 (0.915–1.408) | 0.251 |
| Plate–screw fixation | 0.145 | 1.029 (0.780–1.358) | 0.838 | 0.142 | 1.010 (0.766–1.331) | 0.946 |
| Dynamic hip screw | 0.267 | 0.788 (0.406–1.533) | 0.483 | 0.283 | 0.834 (0.429–1.620) | 0.592 |
| Year of surgery | 0.010 | 1.011 (0.991–1.031) | 0.295 | 0.010 | 1.014 (0.993–1.034) | 0.187 |
| Hospital 1# | 0.083 | 0.859 (0.711–1.038) | 0.115 | 0.085 | 0.881 (0.729–1.064) | 0.187 |
| Blood transfusion* (exposure versus no exposure) | 0.109 | 1.061 (0.867–1.298) | 0.565 | – | – | – |
| Blood transfusion* (per unit transfused) | – | – | – | 0.026§ | 1.069 (1.020–1.121)§ | 0.005§ |
§p < 0.05; *allogeneic blood transfusion within 7 days before until 7 days after surgery; †preoperative hemoglobin level was available in 740 patients, total estimated blood loss in 672 patients, anesthesia time in 704 patients, and BMI in 610 patients. Missing values were imputed using multiple linear imputation including all variables (the number of imputations was set to 40); ‡primary tumor type with good prognosis includes: breast, kidney, prostate, thyroid, myeloma and lymphoma; #see text for transfusion guidelines per hospital (Hospital 2 is the reference group).