Abstract
Recent work has hinted at the prognostic importance of anemia in the elderly and has tried to identify causes. In an effort to further explore this important clinical problem, we decided to determine the prevalence, etiology and prognostic implications of anemia in a large inner city cohort of elderly outpatients. A total of 40,417 individuals ≥65 years of age seen at Montefiore medical system outpatient clinics from January 1st 1997 to May 1st 2008 who underwent a complete blood count within 3 months of the visit were included in the cohort. Using gender specific WHO criteria, we determined 35.3% of men (out of 15,629) and 28.8% of women (out of 24,788) were anemic at the time of their first clinic visit. A follow up of their blood counts showed that ultimately 70.7% of men and 65.4% of women subsequently developed anemia, thus demonstrating a very high incidence in this population. Among anemic patients, 14.2% had microcytosis and 11.2% had macrocytosis. Of all microcytic anemic patients, 57.3% had both ferritin and iron saturation performed and 19.1% of those tested met criteria for iron deficiency anemia. Of all macrocytic anemic patients, 51.7% had folate and 53.2% had B12 levels checked. However, of those tested only 0.6% and 6.8% had folate and B12 deficiency respectively. This percentage of abnormal findings was similar to non-anemics (0.3% and 5.5% respectively). Evaluation for further etiologies of anemia revealed that 10.0% had thrombocytopenia, 13.5% had leukopenia, and 2.0% had neutropenia, raising the possibility of bone marrow failure and myelodysplastic syndrome (MDS) in these patients. Of the 3,471 anemic patients with other cytopenias, only 6.1% had a bone marrow examination (n=213). Subsequent diagnosis of MDS was made in a minority of these patients. A search for other chronic diseases among the anemic group revealed an elevated creatinine in 11.2%, a subsequently elevated TSH in 17.5%, transaminitis in 28.0% and hypoalbuminemia in 21.9%. Of anemic patients who had an ESR and an ANA assay performed, 69.7% and 23.1% had abnormal results respectively. Overall, a possible etiology based on the above laboratory findings was only identified in 61.9% of anemic patients. Of the small number of anemic patients who had every one of the above tests performed (n=662), 12.1% did not have any abnormal results.
To determine the functional importance of these findings, Kaplan-Meier curves and multivariate Cox regression analyses were performed to compare survival and mortality in this cohort. Anemic patients had reduced overall survival regardless of ethnicity (Figure 1). Interestingly, among the elderly, minorities appeared to have better survival than Caucasian whites regardless of anemia status. This is likely secondary to selection bias in this age group as minorities have higher mortality rates below age 65. Adjusting for age and gender, the overall hazard ratio (HR) comparing anemic patients with nonanemic patients was 2.15 (p<0.001). Repeating for each ethnicity subgroup resulted in adjusted HRs of 1.93 (p<0.001) in whites, 2.00 (p<0.001) in blacks, and 2.44 (p<0.001) in Hispanics. Multivariate analyses of various lab parameters demonstrated that an elevated RDW (>16.6%) was the strongest predictor of decreased survival in anemic patients with an age/gender/Hb-adjusted HR of 1.73 (p<0.001). (Figure 2) Thrombocytopenia (HR= 1.37, p<0.001), leukocytosis (HR=1.37, p<0.001), neutropenia (HR=1.23, p=0.032) and macrocytosis (HR=1.24, p<0.001) were additional poor prognostic factors, raising the possibility of MDS as an adverse diagnostic subcategory of anemia in the elderly.
In conclusion, we determined that anemia is very common among the elderly in the outpatient setting, and the majority of non-anemic patients ≥65 years of age ultimately develop this condition. Folate/B12 screening has minimal diagnostic utility in anemic patients. Anemia is associated with decreased survival in all racial groups, with larger effects in African Americans and Hispanics. An elevated RDW and the presence of other cytopenias are highly predictive of mortality in anemic patients even after controlling for age, gender and hemoglobin.
Disclosures: No relevant conflicts of interest to declare.
