Shift of the Serum Erythropoietin-vs.-Hemoglobin Curve with Frailty Status in Community-Dwelling Older Adults. Results from the Women’s Health and Aging Studies (WHAS) I and II.

Introduction: The criteria currently used to define normal hemoglobin concentration (Hb) in older adults do not take into consideration the heterogeneity of health status. To gain insight into the potential relevance of such a consideration, we examined whether the relationship between Hb and serum erythropoietin (EPO) - a surrogate marker for tissue hypoxemia - would be modified by frailty status in community-dwelling older women. Methods: Cross-sectional analysis of data from WHAS I and II, two complementary population-based studies investigating the epidemiology of disability progression and onset (WHAS-II, 1994–1996). Baseline WHAS I (1992–1995) and II (1994–1996) data from women 70–80 yrs with known frailty status, EPO, and Hb that was moderate-to-mildly low and normal Hb (i.e., 10–16g/dL) were pooled. The final sample size was 642, after excluding subjects with outlier EPO values. Serum EPO was measured in samples stored at −70C by enzyme immunoassay; within- and between-essays coefficients of variability were 13.3% and 11.6%, respectively, indicating acceptable reproducibility. Frailty was defined according to a previously-validated, Geriatrics-accepted classification; i.e., considered present if 3 or more of the following were present: slowness (walking speed less than bottom 20th percentile), unintentional weight loss >10%, low energy expenditure (bottom 20th percentile of calculated Kcal using the Specific Activity Scale), weakness (grip strength <20th percentile), and self-reported low energy. A generalized linear model with a log-link and gamma distribution was used to estimate mean EPO, the dependent variable, as a function of Hb (continuous) and frailty status, while controlling for age (continuous) and tertiles of Cockcroft-calculated creatinine clearance. Results: Relationships between EPO and Hb in both frail and non-frail subjects were non-linear, with lowest EPO around mid-normal Hb concentrations. However, the curve in frail subjects was shifted to the right and upwards, so that mean EPO for the same Hb was on average 1.59 (1.22 – 2.06) times greater in frail than in non-frail subjects, according to a model that controlled for age and calculated creatinine clearance. Additional analyses were conducted to determine the difference in mean EPO across different Hb concentrations and between the frail and non-frail groups. For example, we estimated that when compared to the predicted mean EPO value for a Hb of 14 g/dL in frail subjects, mean EPO values linked to Hb concentrations greater or equal to 12.3 g/dL in non-frail were statistically-significant lower (p<.043). Conversely, there was no statistically-significant difference in mean EPO linked to a Hb of 14 g/dL in frail subjects vs. mean EPO linked to Hb concentrations between 12–10 g/dL in the non-frail group (p>.100). Conclusion: The shift of the EPO vs. Hb curve by frailty status observed in this population-based study of older women suggests that, as compared to their non-frail counterparts, frail subjects might require significantly higher Hb levels to achieve similar tissue oxygenation levels, as surrogate-measured by EPO levels. These findings warrant further exploration of how best take into account health status heterogeneity issues for the development of improved anemia-related clinical decision-making in older adults.