In Reply:—
We wish to thank the correspondents for their interest in our review1and for correcting the typographical errors. Although a small typographical error in the sickle genetic code has extensive consequences, the equivalent inaccuracies in the review fortunately do not have similarly far-reaching effects on our conclusions. Regarding the work by Vichinsky et al. ,2–4the fact that this was analysis of subpopulations of the same patient group was explicitly stated in the relevant studies. Our motivation for reviewing these new studies in detail was the additional information provided, not ignorance of the database from which it originated. The data from Haberkern et al. 2(Complications, page 1534) are in fact correctly reported in table 4 of the original article,1which is repeated now as table 1in this reply. The correspondents are confusing sickle-specific events with overall complications, rates that variously included fever, transfusion reactions, and postoperative surgical problems.2These general outcomes are not the appropriate endpoints by which to assess the need for or effect of transfusion.
With respect to the work of Griffin and Buchanan,5the correspondents seem to have formed their opinions from the abstract summary. We refer the correspondents to the tables, text, and prominent citations6,7discussed by Griffin and Buchanan.5The complications comprising the rates cited by the authors were largely atelectasis and transient, uncomplicated postoperative pyrexia, which the investigators5emphasized were not specific to sickle cell disease. Again, it is misleading to cite the incidence of a heterogeneous group of complications in a heterogeneous population as motivation for an intervention against the lower occurrence of a specific subset of complications in a specific subpopulation. The incidence of complications specific to sickle cell disease in the relevant minor procedure subpopulation was, as reported,12% (n = 1 in 46).5The sole patient with the sickle complication of acute chest syndrome had significant preexisting cardiopulmonary dysfunction. The researchers5cited these facts in their extensive discussion of the limits of any potential role for transfusion. They concluded, “our data support the concept that preoperative blood transfusions may be unnecessary for children with sickle cell disease … undergoing most minor operations (who) might therefore be spared the cost, inconvenience, and risks of infection, alloimmunization and transfusion reactions inherent in RBC transfusions” (page 685).5
In response to the suggestion that table 5, Guidelines for the Use of Perioperative Prophylactic Erythrocyte Transfusion, from the original article is not evidence based, we refer the correspondents to the text of the review and relevant selected references.1Some of the rationale and evidence that simple transfusion be avoided in low-risk situations has been reemphasized above. As the correspondents' call for prospective randomized trials acknowledges, the evidence for the efficacy of simple transfusion in intermediate-risk situations is incomplete, and we cannot therefore say conclusively that transfusion is or is not indicated. Our guidelines for high-risk cases were based on a systematic review of the primary neuroanesthetic and cardiothoracic literature of the preceding four decades, although we limited our citations of this fragmentary evidence to a short selection at the request of the Anesthesiology reviewers. We concede that our guidelines for transfusion in uncomplicated pain crises are not evidence based—simply because we are unaware of conclusive evidence to support or refute the practice of transfusion. We cannot advocate an intervention in the absence of supportive evidence and concur with authoritative peer opinion that transfusion is not indicated.8A detailed critique of practice based on the largely uncontrolled data on transfusion for acute chest syndrome was beyond the scope of an already lengthy review.1The guidelines on acute chest syndrome are consequently limited to the well-documented evidence that transfusion can improve arterial hemoglobin oxygenation, a predictable and possibly nonspecific physiologic consequence of increasing mixed venous saturation and pulmonary capillary transit time by correction of anemia in the face of pulmonary shunting and impaired gas exchange. The guidelines are therefore based on what data are available. Because this evidence is incomplete, we simply provided guidelines, rather than making more proscriptive recommendations.
The correspondents state that we are compelled to prove that no causal relation exists between hypoxia, dehydration, and hypothermia and acute perioperative complications before abandoning practices associated with decreased perioperative morbidity and mortality. The only definitive way to do this, subjecting patients to these injuries in a well-constructed study, is practically difficult and ethically impossible. The studies cited, including exposure to inhalational and hypobaric hypoxia, the use of occlusive arterial tourniquets, and the coexistence of sickle cell disease and cyanotic heart disease, strongly suggest that hypoxia-induced sickling (and by extension acute cellular dehydration) are not triggers of acute complications.1The relevant practices that are associated with an apparent decrease in morbidity and mortality are probably the overall general improvements in basic anesthetic and perioperative care, not specific unproven deviations from standard practice. We do not advocate therapeutic nihilism, as the correspondents aver, but rather, we encourage adherence to these basic standards of anesthetic care.
We are surprised by the correspondents' continued enthusiasm for transfusion, given their familiarity with the relevant literature.2–4The previously widespread adoption of a practice, exchange transfusion, in the absence of controlled studies, was no guarantee of efficacy or lack of harm.2–4Although the liberal use of transfusion may be well established in the correspondents' practice, this is similarly not proof of efficacy or freedom from injury. For considerations of effect, we direct them to long-established alternative approaches of avoiding perioperative transfusion.6,7For evidence of harm, consider a recent study of 150 multiply transfused American patients that found an incidence of hepatitis C infection of 35.3%, a sobering demonstration of iatrogenic inury.9By contrast, in Jamaica, where transfusion practices are far more conservative, a study of 250 patients documented an infection rate of 2%.10Rather than preventing acute sickle problems, transfusion can actually precipitate acute pain11and pulmonary complications.12We do not suggest that our review is the last word on management or even that the model we outline explains all aspects of sickle cell disease. We therefore strongly support the call for prospective randomized studies of anesthetic practice for sickle cell disease. Given the proven potential for iatrogenic injury, we urge the correspondents to consider the cautious methods of others1,5–8,12and, in the absence of conclusive evidence, not to abandon a well-established practice: First do no harm.
* Medical College of Georgia, Augusta, Georgia. ahead@mcg.edu
