Hospital-Acquired Anemia: What Are the Implications?
ABSTRACT & COMMENTARY
Most hospital-based clinicians are aware that anemia develops in many hospitalized patients. This hospital-acquired anemia (HAA) has many etiologies including phlebotomy, hemodilution from intravenous fluid administration, procedural blood loss, and impaired erythropoiesis from acute illness. Transfusion thresholds in hospitalized patients have decreased over the last decade, but transfusion still carries risk. Given that the magnitude of the problem remains unclear, these investigators set out to study the frequency, severity, and implications of HAA.
This was a retrospective study of 417,301 hospitalizations in adult patients from January 2009 to September 2011 at the Cleveland Clinic Health System. Data were extracted from an administrative database (the University HealthSytem Consortium, or UHC, clinical database) consisting of demographics, baseline comorbidities, and outcomes and combined with hemoglobin data from the hospitals’ electronic medical record. Patients were excluded if they had anemia present on admission or if hemoglobin (Hgb) values were unavailable during the hospitalization. The final dataset consisted of 188,447 hospitalizations. HAA was categorized based on the lowest Hgb value during the hospitalization as no anemia, mild anemia (Hgb > 11 and < 12 g/dL in women, > 11 and < 13 g/dL in men), moderate anemia (Hgb > 9 and ≤ 11 g/dL), and severe anemia (Hgb ≤ 9 g/dL). Outcomes were all-cause in-hospital mortality, total hospital length of stay (LOS), and total hospital charges.
Among patients without anemia at the time of admission, the prevalence of hospital-acquired anemia was an astonishing 74%! Twenty-two percent (22%) developed mild anemia, 30% developed moderate anemia, and 22% developed severe anemia, while 26% did not develop anemia during their hospitalization. As might be expected, patients who developed HAA were older, had more comorbidities, and more commonly had surgical diagnoses than patients who did not develop HAA. Patients with mild HAA did not have higher risk-adjusted mortality than those without HAA, but as the anemia worsened, risk-adjusted hospital mortality increased in a dose-dependent manner compared to patients without HAA: mortality in moderate anemia was 1.51 times higher (95% CI: 1.33-1.71, P<0.001) and in severe anemia was 3.28 times higher (95% CI: 2.90-3.72, P<0.001). HAA was also associated with a significant increase in LOS and hospital charges even when adjusted for all comorbidities. Compared to patients without HAA, patients with severe HAA (Hgb ≤ 9 g/dL) had a mean relative increase in LOS that was nearly double at 1.88 (95% CI: 1.86-1.89, P<0.001) and total hospital charges for severe HAA that were also nearly double at 1.80 (95% CI: 1.79-1.82, P<0.001).
I believe that this is an important observational study that hopefully will stimulate more work on the prevention of hospital-acquired anemia. These authors identified a very high prevalence of HAA, even higher than I would have anticipated: 52% of patients developed either moderate or severe anemia after hospitalization. Patients who developed severe HAA had over a 3-fold increase in their risk of death even after adjusting for all measured comorbidities. This study suggests that HAA is an independent risk factor for in-hospital death, but causality can be difficult to ascertain in observational studies. Thus, it is not clear that the anemia itself caused the increased mortality. It could be that the anemia was simply a marker of increased risk caused by an unmeasured covariate or other unknown factors. Another concern I have is that the diagnosis of anemia at the time of admission was based on the use of the diagnostic code for Anemia Present on Admission. The accuracy of that code is not clear to me. Thus it is conceivable that this methodology underestimated the number of patients who had some degree of anemia on admission thus over-estimating the prevalence of HAA. Despite those concerns, the strengths of the study include its large size and the widespread acceptance of their particular administrative database for many other studies.
Even if the reported prevalence is an over-estimate, the association of HAA with an increased risk of death and increased LOS and hospital charges are highly significant and important observations. Given the caveats of an observational study, it is conceivable that either the anemia or the treatment thereof is one cause of the adverse outcomes observed in this study. While further studies are warranted, hospitalists are poised to initiate some fairly simple interventions targeting one of the major causes of HAA. By being meticulous in our ordering of laboratory testing, batching laboratory requests, using blood conservation devices in the intensive care units, and in advocating for our hospitals to move to the use of smaller-volume collection tubes (pediatric tubes) we can reduce the amount of blood loss from phlebotomy and laboratory testing. Whether this will translate into improved outcomes has yet to be shown, but this study highlights the importance of hospital-acquired anemia and its prevention.