Low hemoglobin levels and recurrent falls in U.S. men and women: prospective findings from the REasons for Geographic And Racial Differences in Stroke (REGARDS) cohort

Abstract

Background: There are few data available on low hemoglobin and incident falls in the general U.S. population.

Methods: Of 30,239 black and white U.S. adults ≥45 years in the population-based REasons for Geographic And Racial Differences in Stroke study, 16,782 had hemoglobin measured at baseline and follow-up data on falls. Hemoglobin was categorized by 1.0 g/dL increments relative to the World Health Organization anemia threshold (<13.0 g/dL for men, <12.0 g/dL for women). Recurrent falls (≥2 falls in the 6 months after baseline) were assessed during a telephone interview.

Results: Recurrent falls occurred in 3.9% of men and 4.8% of women. Compared with those with a hemoglobin level 1 to 2 g/dL above the anemia cut-off, multivariable adjusted odds ratios (95% confidence intervals) for recurrent falls associated with hemoglobin levels ≥3, 2 to <3 and 0 to 1 g/dL above the cut-off point, and 0 to <1 and ≥1 g/dL below the cut-off point were 0.73 (0.45-1.19), 0.84 (0.57-1.24), 1.29 (0.88-1.90), 1.32 (0.0.80-1.2.18) and 2.12 (1.23-3.63), respectively, among men (linear trend P < 0.001), and 1.59 (1.10-2.3), 1.24 (0.95-1.62), 1.42(1.11-1.81), 1.28 (0.91-1.80) and 1.76 (1.13-2.74), respectively, among women (linear trend P = 0.45; quadratic trend P = 0.016).

Conclusions: Among men, lower hemoglobin levels were associated with an increased risk for recurrent falls. Although our findings suggest an increased risk for recurrent falls at both lower and higher hemoglobin levels among women, these findings should be confirmed in subsequent studies.

Figure 1

Percentage of REGARDS participants with recurrent (2 or more) falls by hemoglobin category relative to World Health Organization anemia cut-point for (a) men and (b) women

Figure 2

Odds ratio (95% Confidence Intervals) for recurrent falls by level of hemoglobin relative to World Health Organization anemia cut-point for men (13.0 g/dL) and women (12.0 g/dL). Model 1 adjusted for age, race and geographic region. Model 2 adjusted for variables in Model 1, marital status, cigarette smoking, education, waist circumference, hypertension, diabetes, coronary artery disease, stroke, depressive symptoms, cognitive impairment, estimated glomerular filtration rate and albumin-to-creatinine ratio. Linear p, p-value for linear trend across hemoglobin categories. Quadratic p, p-value for quadratic trend across hemoglobin categories.

Figure 3

Odds ratio of recurrent falls (2 or more) by subgroups for men and women. Adjusted for age, race, gender, geographic region, marital status, education, waist circumference, hypertension, diabetes, coronary artery disease, stroke, depressive symptoms, cognitive impairment, estimated glomerular filtration rate (eGFR), albumin to creatinine ratio (ACR), Medical Outcomes Study Short-Form (SF-12) physical component score (PCS) < 50, and a fall in the year prior to baseline. Among women, analysis of hemoglobin as a continuous variable was limited to the linear range (hemoglobin 1 to <2 g/dL above the WHO anemia cut-point and below).