Preoperative Anemia and Postoperative Outcomes in the Elderly

Abstract & Commentary

By David J. Pierson, MD, Editor, Professor, Pulmonary and Critical Care Medicine, Harborview Medical Center, University of Washington, Seattle, is Editor for Critical Care Alert.

Synopsis: In this study of 310,311 veterans aged 65 or older who underwent major noncardiac surgical procedures, 30-day mortality increased 1.6% for every percentage-point decrease in preoperative hematocrit below 39%.

Source: Wu WC, et al. JAMA. 2007:297(22):2481-2488.

Using data from 132 US veterans hospitals collected in the National Surgical Quality Improvement Program, Wu et al performed a retrospective cohort study of patients aged 65 or older who underwent major non-cardiac surgery between 1997 and 2004. Major surgery included all procedures, elective and emergency, performed in the operating room under general, spinal, or epidural anesthesia. Postoperative mortality and cardiac events (cardiac arrest or Q-wave myocardial infarction) were correlated with preoperative hematocrit values. The latter were divided into 14 groups, from < 18% to ≥ 54%, and referenced to a normal range of 39% to 53.9%.

The study cohort was comprised of 310,311 veterans (98% male, 80% white). Of these, 42.8% had preoperative anemia, defined as hematocrit < 39%, and 0.2% had polycythemia (hematocrit ≥ 54%). Anemic patients had significantly (p < 0.001) more diabetes, cardiac disease, neurologic disorders, renal disease, long-term corticosteroid use, and cancer than non-anemic patients. They also tended to be older, inpatients rather than outpatients before surgery, and non-independent in functional status, and to have higher American Association of Anesthesiologists class.

For the entire study population, 30-day postoperative mortality was 3.9% and the cardiac event rate was 1.8%. Both of these outcomes rose monotonically for patients with progressively lower and higher hematocrit levels than normal. For example, the mortality rates were 1.5% for patients with hematocrits of 45.0 to 47.9%, 5.8% with values of 33.0 to 35.9%, 14.9% with values of 24.0 to 26.9%, and 35.4% with values < 18%. There was a 1.6% (95% confidence interval, 1.1%- 2.2%) increase in 30-day postoperative mortality associated with every percentage-point decrease in hematocrit level from the normal range. This increase was not just observed in patients with very low hematocrits, and began when the level fell below 39%.


In this study, elderly patients with preoperative hematocrits below the lower limit of normal had worse outcomes, and mortality increased progressively with lower and lower hematocrits. This does not necessarily mean that the anemia was the cause of the worse outcomes, or, that raising the hematocrit into the normal range in these patients would have improved their outcomes. The authors acknowledge both of these points. In fact, the anemic patients were substantially different from their non-anemic counterparts in a lot of ways that would be expected to push the findings in the direction observed, including being older, having more comorbidities, and being less functional prior to surgery. This suggests that preoperative anemia is a marker for, rather than an independent cause of, worse postoperative outcomes.

This was not an ICU study, and the proportions of patients in the cohort who were critically ill prior to surgery or managed in the ICU postoperatively are not given. Only about 8% of the operations were classified as emergent. The authors found essentially no relationship between hematocrit and outcomes in this subset of patients. This may have been because of the overriding effects of comorbidities in these patients, or because the preoperative hematocrit values in the database did not reflect the patients' status at the time of surgery: 21% of the total patient population had no hematocrit recorded within 30 days of the operation. Another interesting finding was that patients who received more than 4 units of transfused blood preoperatively had a reduced rate of postoperative death: the odds ratio was 0.88, with a 95% confidence interval of 0.79-0.98. This supports the notion that recognition and management of serious anemia prior to a scheduled major operation is a good idea.

What should ICU clinicians make of this study's findings in the context of the recent attention focused on hematocrit levels and outcomes in critically ill patients? Studies using large databases have shown associations between lower hematocrits and increased mortality in patients with acute myocardial infarction. However, prospective studies randomizing ICU patients to lower vs higher transfusion thresholds have found that a more liberal transfusion strategy does not improve outcome. It may be that, in general, the presence and severity of anemia are markers for poor prognosis from a multiplicity of disease processes, and that treating the marker per se may not affect the factors responsible for that poorer prognosis. Nonetheless, the present study may prove useful to clinicians in helping to identify patients at increased risk for unfavorable outcomes after elective major noncardiac surgery, so that such patients can be monitored closely if they are in the ICU.