In Reply: Dr Bellati and colleagues raise the important point of considering hemoglobin variation between the preoperative and postoperative periods, in addition to absolute values, to predict postoperative outcomes. They speculate that this may partially explain the higher postoperative event rates of patients with polycythemia, whose decrease in red blood cells may have been larger than that of the remaining study population. This is possible because hematocrit levels often drop after surgery because of blood loss, intravascular volume shifts, or both. We found that patients with polycythemia had greater mean hematocrit count decreases (16.55% [SD, 9.13%]) than patients with hematocrit values in the normal range (9.81% [SD, 5.86%]) or in the anemic range (5.80 [SD, 4.68%]) (1-way analysis of variance, P < .001). We agree that an account of these hematocrit variations is potentially useful to better predict postoperative outcomes in statistical models, as a postoperative decrease in red blood cell count has been associated with higher postoperative event rates.1 However, this measure may not be clinically useful in the setting of preoperative risk evaluation, which cannot anticipate the patient's postoperative values. This is our rationale for using only the preoperative hematocrit values in the study.
Wu W, Schifftner TL, Henderson WG. Preoperative Hematocrit Levels and Outcomes After Noncardiac Surgery—Reply. JAMA. 2007;298(13):1512-1514. doi:10.1001/jama.298.13.1513