Background
Falls and fall-related injuries are considered a major public health problem in older people. Thirty percent of people aged 65 plus experience a fall at least once a year, and 15% at least twice per year [
1‐
3]. It has been well established knowledge that falls and fractures are among the principal causes of disability, admission to hospital care, and mortality in older people [
4‐
6]. Due to the inability to get up again without help, older people sometimes do experience a “long lie” [
7], thus further aggravation of injuries occurs. It has been shown that half of older people who did lie longer after having fallen, die within 6 months [
7,
8]. Falls are also associated with high costs. For Germany, Heinrich et al. estimated overall attributable costs between 2.1 and 3.8 billion Euros per year [
9]. Due to these serious health and economic consequences, the identification of treatable risk factors for falls of older people is important.
Anemia is common in old age, with a prevalence of approximately 11% in subjects aged 65 years and older as documented in the Third National Health and Nutrition Examination Survey (NHANES III) study. The prevalence of anemia increases rapidly after the age of 50, approaching a rate greater than 20% in those individuals aged 85 years or older [
10]. This also has been confirmed by the Leiden 85 plus study, in which the prevalence of anemia has been 26.7% for people aged 85 years or older [
11]. Regarding sex differences the results of the NHANES III study showed almost no difference in the group of 65 to 74 years (7.8% in men vs. 8.5% in women), yet with increasing age, the prevalence of anemia was much higher in men than in women (26.1% vs. 20.1%; aged 85+) [
10]. In the Leiden 85 plus study the difference was even higher: 35.5% anemic males vs. 22.3% anemic females [
11].
Recently, anemia in old age, defined as a hemoglobin level below 12 g/dL in women and below 13 g/dL in men [
12], has received increasing attention among researchers. Symptoms of anemia include low energy, fatigue, dizziness, and general weakness. In addition, late life anemia is associated with subsequent physical decline, as illustrated by increased disability, impaired performance, and muscle weakness [
13,
14]. With these consequences, anemia could result in an increased risk of subsequent falls, although research confirming such a potential link is limited.
So far, only a few publications exist investigating the relation of anemia and falls in older persons. Three studies found a significant increase in risk of falls with decreasing hemoglobin (Hb) levels [
15‐
17]. One other could not find a significant relation, but points out that anemic people fall more often than non-anemic people [
18]. Two of these publications included only people living in nursing homes [
15] or hospitalized for acute care [
17], and two others included community-dwelling people with conflicting results [
16,
18].
Therefore, the aim of this study was to analyze the association between anemia and falls in 967 community-dwelling men and women aged 65 years or older. Furthermore, we examined the combined relationship of anemia and disability as well as frailty with falls in older people.
Methods
Study design and participants
Data were collected in 2009 during the KORA (Cooperative Health Research in the Region of Augsburg)-Age study, a follow-up study of the four MONICA/KORA Augsburg Surveys [
19]. People aged 65 years and older were taken from a population based random sample (n = 5,991). In total, 4,127 people participated in a standardized telephone interview (response 67%). Out of this group a randomly drawn sample of 1,079 participants additionally underwent extensive physical examinations including the registration of drug intake, collection of blood samples, anthropometric examination, grip force measurement, gait analysis, and an additional interview amongst others. Blood samples and answers to fall questionnaires were available for 1,048 participants. After exclusion of 81 participants because of missing data the final data set for the present analysis consisted of 967 participants (477 women and 490 men) aged 65–91 years. The study was approved by the ethics committee of the “Bayerische Landesärztekammer”; all participants provided a written informed consent.
Outcome definition
History of falls was assessed with the question “Did you fall in the previous year?” with possible answers “yes, once”, “yes, more than once” and “no”. This variable was dichotomized in “at least one fall” and “no fall”. The question was derived from the National Health and Nutrition Examination Survey (NHANES)-questionnaire [
20].
A non-fasting venous blood sample was obtained from all study participants while sitting. Hematological parameters were determined from fresh venous EDTA blood samples using impedance measurements (Coulter® LH 780, Beckman Coulter GmbH, Krefeld, Germany). Anemia was defined according to WHO criteria as a hemoglobin (Hb)-level less than 12 g/dL in women and less than 13 g/dL in men [
12].
Covariables
Information on socio-demographic characteristics and lifestyle behavior was collected by an interview. Frailty was assessed according to the criteria of Fried et al. [
21]: weight loss in the last 6 months (short German SCREEN II questionnaire [
22]), fatigue (via interview questions), low physical activity (via interview questions [
23]), low walking speed (Timed Up and Go Test) and low grip force (measured via JAMAR, Saehan Corp.). Participants were classified as frail when they met 3 or more criteria, pre-frail when they met 1 or 2 criteria and non-frail when no criterion was met. The variable was dichotomized in “frail” (including pre-frail) and “non-frail”. The disability score was assessed via the Stanford Health Assessment Questionnaire [
24] and categorized as disability “yes” (score ≥0.5) or “no” (score <0.5) as suggested in literature [
25]. Multimorbidity was obtained with the Charlson-Comorbidity-Index from Chaudhry [
26]. Participants were classified as multimorbid, if they had ≥3 diseases. Drug intake during the last 7 days before the examination was recorded with the IDOM-Software [
27].
Statistical analyses
Besides descriptive analyses, different logistic regression models were computed to assess associations of anemia/hemoglobin values with falls, respectively. Hemoglobin levels were considered as continuous variable (decrease per SD) and anemia (yes/no) was defined via WHO criteria [
12].
Variables were considered as confounders in the multivariable regression model, if they were significantly (p <0.05) related to falls as well as Hb-level/anemia after adjustment for age and sex (for details see Table
1). However, the number of drugs was considered as confounder although it was not significantly related to falls, because the relation to anemia and Hb-level was significant and literature suggests it as one risk factor for falls [
28‐
30].
Table 1
Overview of p-values for considered confounders
Multimorbidity | 0.0049 | 0.0118 | 0.1334 |
Disability | 0.0023 | <.0001 | 0.0038 |
Diuretics | 0.0584 | 0.0318 | 0.6326 |
Frailty | <.0001 | <.0001 | 0.1866 |
Hypertension | 0.1297 | 0.8939 | 0.2585 |
Diabetes | 0.0004 | 0.0369 | 0.4392 |
Anticoagulants | 0.2342 | 0.0625 | 0.2856 |
Antihypertensives | 0.1361 | 0.0553 | 0.9232 |
Number of drugs | <.0001 | <.0001 | 0.3257 |
Nutrition score | 0.7864 | 0.1241 | 0.0044 |
Elevated alcohol intake*
| 0.0004 | 0.0702 | 0.4226 |
PASE†
| 0.7922 | 0.2476 | 0.7308 |
BMI | 0.0012 | 0.0689 | 0.0274 |
For the outcome “falls (yes/no)” the unadjusted model included hemoglobin level or anemia as independent variables. Model 1 additionally included age (continuous), and sex (in the unstratified analyses only) as confounders. In model 2, the number of drugs was added in addition to the confounders of model 1. In model 3 the variable “disability status (yes/no)” was added to the confounders of model 2.
We stratified all models by sex, as the normal Hb-level for men (≥13.0 g/dL) is higher than for women (≥12.0 g/dL) and due to the differences regarding the frequency of falls among men and women. We also stratified all models by age groups. The group of “young-old” included participants aged 65 to 74 years; the group of the “old-old” included all participants aged 75 years and older. Linearity was checked by including a quadratic term of the hemoglobin variable in the models (total study sample). As the p-values of the quadratic terms were not significant, linearity was assumed for the parameter hemoglobin.
To examine the joint effect of anemia and disability as well as frailty on the occurrence of falls, in a second step combined anemia and disability and anemia and frailty variables were created. The subjects were classified into four categories: 1. anemia “yes” and disability (frailty) “yes”, 2. anemia “yes” and disability (frailty) “no”, and 3. anemia “no” and disability (frailty) “yes”. Those with no anemia and no disability (frailty) were category 4 and thus the reference group. These analyses were performed in the total sample but not conducted separately for men and women or for the two age-groups due to low numbers in the different subgroups.
Results of the models were presented as Odds Ratios (ORs) and corresponding 95% confidence intervals (CIs). Significance tests were two tailed and p-values ≤0.05 were considered as statistically significant. All analyses were performed with SAS (version 9.2, SAS Institute Inc, Cary, NC, USA).
Discussion
In the present analysis including 967 community-dwelling elderly men and women no significant association between hemoglobin levels or anemia and falls could be found. Thus, the suggestion that anemia is an independent risk factor of falls in elderly people from the general population could not be supported by these results. However, we found an additive effect of anemia and disability on the occurrence of falls.
On the contrary to our study prior studies were mostly conducted in selected groups [
15‐
17] in women only [
32] or they used self-reported data regarding anemia status or information from chart reviews [
33]. Furthermore, most of the former analyses were based on smaller study samples [
15,
17,
18].
The group of Duh et al. employed a retrospective open-cohort design and analyzed data of 47,350 individuals regarding anemia and risk of injurious falls in community-dwelling elderly people. Results showed that anemia increased the risk of injurious falls by 1.47 times in multivariate analysis adjusting for age, gender, health plan, history of falls, co-morbidities, and concomitant medications [
16]. A 50% higher risk of fall injuries in community-dwelling adults aged 65 and older was also reported by Herndon et al. [
33]. However, contrary to the present study, in that case–control study (467 cases, 691 controls), anemia based on self-report was used. Both studies mentioned above concentrated on injurious falls whereas our database includes also falls which did not cause injuries.
Pandya et al. found even more than twice the risk of falling for anemic participants when retrospectively analyzing the relationship between anemia and falls in 564 nursing home residents. However, in that study, only a selection of relevant covariables was available and collected by chart review [
15]. Dharmarajan et al. investigated in a case–control study the relationship between the presence of anemia and falls during hospitalization in 362 ambulatory adults aged 59–104 years from long-term care and community settings. They found a 22% decreased risk of falls for every 1.0 g/dL increase in Hb-concentration and a 1.9 times increased risk of falls in anemic hospitalized patients [
17].
In another study, Lawlor et al. analyzed data of 4,050 female participants of the population-based British Women's Heart and Health Study aged 60 to 79 and found that women reporting at least one fall per year had lower mean Hb-concentrations than women who did not report a fall; the inverse association remained even after adjustment for social class, BMI, chronic diseases, and each class of drug used (OR of any falls for a 1 SD increase of Hb was 0.90, 95% CI 0.81-0.99) [
32]. In contrast to these findings we could not find significantly lower Hb-concentrations for women who reported a fall. Additionally we found no significant relationship of Hb-concentrations and falls in women. Contrary to the other publications the observational study of Penninx et al. including 394 community-dwelling older persons from The Netherlands could not prove that anemia directly affects risk of falls although longitudinal data were used and disease and functioning status were adjusted for. The authors hypothesized that anemia could still be only a manifestation of disease severity which would be the true link between anemia and falls [
18].
So far, no data are available showing the joint association of anemia and disability with the occurrence of falls in the elderly general population. The results of our study extend the understanding that different factors in elderly people can have an additive effect on the occurrence of falls. Although no independent association between anemia and falls was found in the present sample, it could be shown, that persons with disability and concomitant present anemia represent a high risk group for falls. The findings from the present data therefore suggest that in elderly, mainly multimorbid persons single risk factors should not be considered isolated in relation to certain outcomes, but rather in the context of other relevant disorders and restrictions.
We recognize that our study has limitations. First, participants were asked for falls which happened in the past year on the day the blood sample was taken. Therefore, the anemia status of a participant could have been different at the time when the fall occurred. However, it could be assumed that in the present sample almost all persons with anemia suffered from chronic and not from acute anemia and thus were well adapted to the low hemoglobin levels. Hence, it is most likely that the status of anemia was similar at both points of time. Second, the interview-based assessment of falls could be imprecise, as the interrogated person could have forgotten these events (consistency bias), or did not want to report them in order to not appear frail. We did not find an investigation on the reliability and validity of the NHANES questionnaire [
20] used to assess falls, but the short and easy way to ask people about their history of falls indicates reliable results. Third, the exclusion of participants who could not come to the study center because of severe illnesses could have influenced the results as they would most likely have lower hemoglobin levels. This could possibly have led to an underestimation of the true association between anemia and falls. On the other hand it was necessary to exclude people who could not walk anymore as they would not experience falls. Fourth, the cross-sectional design of the study represents a limitation, implicating that cause and effect relationships cannot be discerned. Fifth we included the intermediary variable disability as confounder in the logistic regression model which could result in a decreased association between anemia and falls. But as the results of the logistic regression models were already not significant before including these variable, this should not have influenced our results.
Finally, another issue worth considering would be the number of falls, as it was analyzed by Penninx et al., who found a 1.91 times greater risk of recurrent falls in elderly people with anemia (n = 394) [
18]. But in our study group only 92 participants fell once, and 51 fell more than once. When looking at the anemic participants only 22 people fell once and only 9 fell more than once. We considered these groups as too small for meaningful analyses. The strength of the study is the inclusion of a relative large number of individuals randomly drawn from the general population, and the availability of data on lifestyle, medication, and multiple metabolic risk factors.
Acknowledgements
The authors would like to thank all study participants, the KORA Augsburg field staff and all members of the Institute of Epidemiology II from the Helmholtz Zentrum München who were involved in planning and/or conducting the study.
This research was supported by the Graduate School of Information Science in Health (GSISH) and the TUM Graduate School.
Trial registration/Funding
The KORA-Age study was part of the program “Health in Old Age”, which was financed by the German Federal Ministry of Education and Research (BMBF FKZ 01ET0713). The KORA-Age project was conducted on the KORA platform (KORA, Cooperative Health Research in the Region of Augsburg), which was initiated and financed by the Helmholtz Zentrum München, German Research Center for Environmental Health, Neuherberg, and funded by the German Federal Ministry of Education, Science, Research and Technology as well as the State of Bavaria. The study “Functioning and Disability in Old Age” was funded by the German Research Foundation (Deutsche Forschungsgemeinschaft) (GR 3608/1-1). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests
The authors report no potential conflicts of interest.
Authors’ contributions
KTK did the statistical analysis, generated all tables and drafted the manuscript. All calculations were checked and recalculated by CM to guarantee correctness. AD, AH and CM gave major suggestion on how to analyze and how to interpret the data and on the manuscript draft. All authors contributed to the design of the study and data collection. They also reviewed and edited the manuscript. All authors approved the final version of the manuscript.