The association between anemia and depression in older adults and the role of treating anemia

Abstract Objectives To investigate the association between anemia and depression and whether the treatment of anemia modifies the effect of the association between anemia and depression. Methods This secondary data analysis is based on data from the Enquête sur la santé des aînés (ESA)‐Services study conducted in 2011–2013 on community‐dwelling older adults recruited in primary care and have given access to their medico‐administrative data (n = 1447). The presence of anemia was self‐reported, as was depression (major and minor) aligned with symptoms of the DSM‐5. Treated anemia was based on the presence of medications delivered to participants. Cross‐sectional associations were analyzed using multivariable logistic regression, controlling for confounders. Results The prevalence of self‐reported anemia in our sample was estimated at 6.7%. Self‐reported anemia was associated with increased odds of depression. Individuals with untreated anemia had a 2.6‐fold increased odds of depression compared to those with no anemia. In contrast, the odds of depression in individuals with treated anemia were not different from individuals with no anemia. Conclusion The findings underline the importance of treating anemia in older adults. Future longitudinal studies are needed to replicate the findings and further explore the role of treating anemia on symptoms of depression.

after iron deficiency anemia (Weiss & Goodnough, 2005). Other wellknown causes of anemia include chronic kidney disease, decreased erythropoietin secretion, reduced response to erythropoietin, inflammatory processes, bone marrow failure conditions, and polypharmacy (Bach et al., 2014;Smith, 2009). A Statistics Canada population survey showed that anemia among seniors was in part due to insufficient mean corpuscular volume and serum ferritin levels, as well as vitamin B12 deficiency but not red blood cell folate (Cooper et al., 2012).
Individuals reporting income in the lower quartile were also more likely to report anemia. Anemia is also associated with symptoms such as fatigue, weakness, dizziness, falls, depression, decreased quality of life, hospitalizations, and mortality (Ahmed & Vasiliadis, 2021;Corona et al., 2018;Culleton et al., 2006;den Elzen et al., 2009;Kikuchi et al., 2001).
Depression is a common mental illness worldwide, with an estimated 3.8% of the world's population affected and a prevalence of approximately 5.7% among adults aged 60 years and over (WHO, 2021). Chronic conditions are more prevalent in old age, such as cardiovascular diseases, diabetes, and polypharmacy; all considered risk factors for and complications of depression (Holt et al., 2014;Kilzieh et al., 2008;Minicuci et al., 2002;Spandel et al., 2016). Depression in old age constitutes a significant burden on health care systems with growing demand for mental health services (Liu et al., 2020).
A growing body of literature has investigated and shown a relationship between anemia and depression in older adults (Corona et al., 2014;Onder et al., 2005;Stewart & Hirani, 2012). Others also reported that older adults with anemia and poor cognition at baseline had a fivefold increased risk of depression after 4 years of follow-up (Ahmed & Vasiliadis, 2021). However, to date, none of these previously mentioned studies assessed the effect of treating anemia on the prevalence of depression in old age. This study extends the current body of literature by using data on anemia treatment to examine the association between anemia and depression. The objectives are to examine the association between the presence of symptoms of depression (minor and major depression) in older adults and anemia and receipt of treatment and examine whether the treatment of anemia modifies the effect of the association between anemia and depression.

Study population and procedures
The "Étude sur la Santé des Aînés" (ESA)-Services study consists of a convenience sample of 1811 French-speaking older adults aged ≥65 years. Participants were recruited in primary care practices between 2011 and 2013 while waiting for medical services in general practice clinics in one of the most populated administrative health regions of the province of Quebec in Canada and representing rural, semiurban, and urban areas. Study recruitment has been described elsewhere in detail (Préville et al., 2014).
Participants visiting one of the participating general practitioners (GPs) received a pamphlet describing the study objectives and an invi-tation to participate in a face-to-face interview with the study coordi- Written consent was also obtained to link participants' health surveys and medico-administrative data. This secondary study was approved by the ethics committee of the CIUSSS de l'Estrie-CHUS.
The current secondary analysis is based on an analytic sample of n = 1447 participants for whom complete data on anemia and depression were available and were covered under the RAMQ's public drug plan. The flow of participants through the ESA Services Study is presented in Figure 1.

Anemia and treatment
Self-reported anemia was assessed by asking participants if they were currently anemic or if a doctor told them they were currently anemic.
In addition, treatment options (generic and brand names) for anemia in Canada were identified (Lim, 2021). Anemia treatment was assessed by the presence of at least one oral or parenteral prescription for anemia treatment (such as oral iron salts, or polysaccharide-iron complex or heme iron polypeptides or parenteral iron or vitamin B12 or Folic Acid or erythropoiesis-stimulating agents) in the past 3 years in the pharmaceutical registry of the RAMQ. Participants were then classified into three groups: no self-reported anemia, self-reported anemia and treated, and self-reported anemia but not treated.

Depression
The presence of minor and major depression in the last 6 months was based on symptoms aligned with DSM-five criteria (Kocsis, 2013) with the exception of criteria regarding the duration of the disorder, social functioning and attribution to a drug or a physical disease. Participants were considered depressed if they had symptoms of major or minor depression. Otherwise, they were classified as not depressed. Participants were classified as having major depression if they showed the essential features of depression (either depressed mood or the loss of interest or pleasure in usual daily activities nearly every day and most of the day for at least two consecutive weeks) and reported at least five of the nine symptoms of depression where at least one of these symptoms is either a depressed mood or loss of interest or pleasure F I G U R E 1 The flow of participants through the ESA-Services study for the analytic sample.
in usual daily activities. Participants were classified as having minor depression if they showed one of the essential features of depression and reported between two and four of the nine symptoms of depression, where at least one of these symptoms is either a depressed mood or loss of interest or pleasure in usual daily activities. Additional analyses also considered the criteria related to the duration (lasting most or all of the day) and frequency (every day or the majority of the days) of symptoms of depression lasting more than two weeks (Kocsis, 2013).

Confounders
The following factors were considered potential confounders based on previous literature (

Statistical analysis
Descriptive characteristics were reported as mean (standard deviation) or frequency (percentage). Differences in baseline characteristics by self-reported anemia status (no vs. yes) were assessed using independent t-test for continuous variables, chi-square tests or Fisher's exact test as appropriate for categorical variables. Logistic regression was used to examine the cross-sectional associations between self-reported anemia and depression and the effect of treating anemia on these associations at baseline. Multivariable logistic regression analyses were undertaken using a two-stage, backward elimination procedure. First, backward elimination (a criterion to keep p ≤ .1) was used to improve model fit. Second, variables not significantly contributing to the fit were removed sequentially from the model. Initially, the set of potential confounders was assessed to determine those associated with depression with a p ≤ .1. All confounders met this criterion and were included in the subsequent model. Odds ratios (OR) and 95% confidence intervals reported in the table were estimated from the coefficients provided by the final reduced models. All analyses were performed using Stata version 17 (StataCorp, College Station, Texas), and a p value of less than .05 was considered statistically significant.

Characteristics of participants
The characteristics of the study population (n = 1447) are shown in

F I G U R E 2
The prevalence of major, minor, and total depression (n = 1447) by anemia status and treatment of anemia, ESA services study.
older participants being more likely to self-report anemia at baseline.
Around 57.2 % of participants were women, and the prevalence of selfreported anemia was 7.1% in men and 6.4% in women. The highest prevalence of self-reported anemia was observed among participants who reported a higher number of chronic conditions, were nonalcohol drinkers over the past 6 months (11.8%), were underweight (13.6%) or obese (8.1%), reported high functional limitations (11.2%), and depression (10.8%) at baseline (p < .05). The prevalence of self-reported anemia was not statistically different by sex, marital status, education, income, and smoking status (p > .05).
Depression varied significantly by anemia status and whether anemia was treated or not treated (p = .02). The highest prevalence of depression was reported among participants with untreated anemia (28.6%), while the lowest prevalence was observed among those who did not self-report anemia (14.4%). The prevalence of depression among those with treated anemia was 18.8%. A similar trend was observed for major and minor depression by anemia and treatment status ( Figure 2).

Cross-sectional analyses
Multivariable analyses of the association between self-reported anemia and depression are shown in Table 2. There was no evidence of a multiplicative interaction by sex in the fully adjusted models. Individuals reporting anemia had 82% (p = .03) higher odds of depression than those not reporting anemia. Similar estimates were observed for major and minor depression; however, these did not reach statistical significance in the model controlling for potential confounders.

DISCUSSION
The current study contributes to present literature by being the first, to our knowledge, to report on the association between the presence of anemia and its treatment on depression in primary care community-  (Canada, 2012;Cooper et al., 2006). The prevalence of self-reported anemia in the current sample was estimated to be 6.7%, consistent with previous data in older adults (Cooper et al., 2012). According to the WHO reports on the public health significance of anemia in different countries, the estimates of anemia prevalence in our Quebec sample suggest that anemia represents a TA B L E 2 Unadjusted and Multivariable-adjusted Logistic regression models: associations between anemia and depression, baseline data: The current findings suggested that self-reported anemia was associated with depression and untreated anemia was associated with close to a threefold increased odds of depression compared to participants with no anemia. Similar findings were observed for minor and major depression among participants with untreated anemia. In contrast, those with treated anemia did not have an increased odds of depression compared to those with no anemia after controlling for potential confounders.
The relationship between anemia and depression in old age has been explored in cross-sectional (Corona et al., 2014;Onder et al., 2005;Stewart & Hirani, 2012) and longitudinal studies (Ahmed & Vasiliadis, 2021;Steptoe et al., 2012). The current findings are consistent with the results of earlier studies suggesting a link between anemia and depression in old age. In addition, an accumulating body of evidence indicates a plausible biological pathway between anemia and depression. Anemia contributes to altered brain neurotransmitter homeostasis through poor brain myelination and impaired monoamine metabolism. These alterations in the brain's homeostatic mechanisms can lead to emotional and psychological problems predisposing to depression diagnosis (Beard & Connor, 2003;Kim, 2014). Our findings did not show a significant difference between individuals without anemia and those treated for anemia in the odds of reporting depression, while there was a significant increase in the odds of depression among those with untreated anemia. These results are consistent and support findings from previous studies suggesting that improving brain oxygenation, particularly to frontotemporal and subcortical brain structures, which are involved in the depression pathway through the treatment of anemia, may improve brain function and reduce the risk of depression (Kusunoki et al., 1981;Li et al., 2018;Oda et al., 2003).

Strengths and limitations
There are a number of limitations that need to be considered when interpreting the results with respect to selection and information bias. First, comparisons were carried out between those included in the analysis, that is, those with complete RAMQ medication data (n = 1447) and those not included in the analysis but completed the questionnaire at baseline (n = 318). Participants not included in the current analyses were not different from those included with respect to age, sex, education, income, marital status, number of chronic conditions, alcohol consumption, and cigarette smoking (p > .05). However, those included were more likely obese than those not included (p = .03).
Comparisons with respect to anemia and depression did not show any significant difference and therefore minimizing the risk of a selection bias. Second, anemia was self-reported and may be subject to recall bias. Published research using self-reported measures of anemia showed that close to 10% of participants had not declared that they were anemic, although they were after blood tests (Hisa et al., 2019).
As we do not expect this bias to be differential between individuals with and without depression, the associations reported in the current study may be underestimates.
Third, we have not explored causes of anemia in our study, and the treatment was assessed by the presence of at least one oral or parenteral prescription for anemia treatment according to the Canadian guidelines. Our findings are only applicable to anemias responsive to pharmacologic therapy.
Fourth, the assessment of anemia and depression was done at the same time point. The cross-sectional design does not allow to conclude the temporality of events and whether improvement in depression symptoms could also contribute to improvement of anemia. Finally, the results of the current study can be generalized to similar populations that include French-speaking community-living older adults, the majority-white, who were covered under a public medical plan for their medical consultations and public drug plan for their medications.
Future research should focus on replicating these findings in other, more diverse data sets using longitudinal data.

CONCLUSIONS AND IMPLICATIONS
Anemia was associated with higher odds of depression, and this association was similar in older adult men and women. The prevalence of depression was lowest among participants with treated anemia compared to those untreated. A higher odds of depression was also observed among participants with untreated anemia as compared to those without a history of anemia. Participants with treated anemia were not different from those with no history of anemia in their odds of depression after adjustment for potential confounders. Anemia is a treatable condition in old age (Weiss, 2015), and the biological pathway between anemia and depression should be further studied to elucidate the possible effect of screening and treatment of anemia in older adults. Future longitudinal studies should focus on better understanding the etiologic pathway between anemia and depression and confirm the role of treating anemia on symptoms of depression.

AUTHOR CONTRIBUTIONS
T.A, C.L.L, D.B., and HM.V: study concept and design. T.A: statistical analysis. T.A: drafting of the manuscript. All authors: interpretation of data, preparation, and critical revision of the manuscript.

DATA AVAILABILITY STATEMENT
For data requests, please contact Dr. Helen Maria Vasiliadis. Data sharing is subjected to rules governed by the ethics committee of the University of Sherbrooke.