Risk factors and consequences of depression in later life: Findings from the health in men study (HIMS)

Abstract

Background

The Health In Men Study (HIMS) has been collecting data on risk factors and health events for the past 25 years in a large community-representative sample of older men. This paper summarises key-findings of the study about depression in later life.

Methods

Narrative review of selected HIMS studies published over the past 15 years describing risk factors associated with prevalent and incident depression in older men, as well as clinical outcomes associated with depression.

Results

Data from HIMS showed that cardiovascular diseases and risk factors are associated with increased risk of depression, but this association is neither specific nor causative. Findings from HIMS are not supportive of the vascular hypothesis of depression in later life. Studies investigating lifestyle have generated risk tables capable of guiding risk reduction strategies. Other potentially modifiable risk factors associated with depression in the HIMS cohort included abnormal allostatic inflammatory response, high plasma homocysteine and low testosterone. The results from HIMS also showed that depression is most likely a prodromal manifestation of dementia rather that a true risk factor, but it increases frailty and mortality. The association between depression and suicide in older men is largely mediated by deteriorating health and increasing frailty.

Conclusion

HIMS has contributed to advance knowledge about risk factors associated with depression, as well as the health consequences of depression in older men. The study is ongoing and the investigators welcome the opportunity to share data with colleagues who are interested in the health of older people.

1. Introduction

The Health In Men Study (HIMS) started in 1996 as a population-based randomised controlled trial to investigate the effect on mortality of abdominal aortic aneurysm (AAA) screening. The study recruited men aged 65–84 years because the prevalence of AAA increases with age and AAA is much more frequent in men than women [1]. The total community-representative sample included over 40,000 participants, of whom 19,000 were invited to undergo ultrasound scanning for AAA – 12,203 completed the examination and have been followed by successive 3–5 yearly waves of clinical assessments and investigations [2]. The entire sample has been followed using electronic health record linkage.

Since its inception, over 250 papers have reported data arising from HIMS on topics such as cardiovascular diseases, endocrinology, neurology, frailty, injuries, biochemistry, genetic epidemiology, lifestyle, health economics and mental health. Among the latter are studies investigating the causes and consequences of substance use, neurodegenerative, psychotic, bipolar and depressive disorders. This paper will summarise key-findings on depressive disorders arising from HIMS.

2. Depression and its associated factors

The assessment of depression in HIMS relied on the use of the Geriatric Depression Scale (15 items) [3], the Patient Health Questionnaire (PHQ-9) [4] and the recorded diagnosis of a depressive episode during contact with general and mental health services of Western Australia [5]. The prevalence of depression among participants was about 5% [6], with the incidence of affective episodes increasing continually with increasing age [7].

2.1. Cardiovascular diseases

Alexopoulos and colleagues introduced the ‘vascular depression’ hypothesis in the mid 1990s [8]. They postulated that cerebrovascular disease contributed to cause or change the course of depression in later life, and suggested that older adults presenting with depression and concurrent cardiovascular risk factors or diseases had ‘vascular depression’. Many cross-sectional studies have reported that cardiovascular diseases and risk factors are more prevalent among older adults with than without depression [9], although this does not necessarily mean that this association is causal. We used data from HIMS to investigate whether the presence of coronary heart disease or cerebrovascular diseases (including stroke) affected the course of depression among older men, as would be expected according to the vascular depression hypothesis. We found no evidence that cardiovascular diseases altered the proportion of men with depressive disorders who experienced a recurrence of symptoms over the subsequent 6 years [10], nor did we find evidence that depression interacted with cardiovascular diseases to increase mortality [11]. We also found that low HDL cholesterol increased the risk of recurrent, but not of incident depressive episodes [12]. These findings, together with epidemiological evidence showing a dissociation between depression and the exponential increase in the prevalence of cardiovascular diseases with age, suggest that the vascular hypothesis of depression in later life has limited explanatory power when considered in isolation [13].

2.2. Lifestyle

Several lifestyle practices have been associated with increased risk of depression across the lifespan, including physical inactivity, smoking and harmful or hazardous alcohol use [14]. Whether such associations are causal, confounded or are due to reverse causality remains unclear. Data from HIMS have allowed us to address some of these associations. For example, we analysed prospective data over a period of 3–8 years to determine how physical activity, smoking, alcohol use and body mass index interacted to modulate the risk of incident depression in later life. As previously shown by other studies, we confirmed that physical inactivity, hazardous or harmful alcohol use, smoking and a BMI outside the normal range were each associated with incident depression [15]. More helpfully, we showed that these factors interact to modulate the probability of incident depression, allowing them to introduce a risk table to guide risk reduction strategies in clinical practice (Fig. 1) [15]. However, this risk table was inadequate to address the issue of causality.

Fig. 1.

Probability of incident depression among older men according to different lifestyle exposures (modified from Almeida et al. Preventive Medicine 2013;57:878–82).

HIMS investigators then used genetic epidemiology to determine if alcohol use if truly causally related to depression. We completed a Mendelian randomisation study investigating the association between depression and a genetic polymorphism linked to alcohol abuse and dependence (ADH1B rs1229984 G > A). The study replicated previous findings showing that none of the participants with the AA genotype used alcohol either harmfully or hazardously, with those with GA genotype forming an intermediate alcohol consumption group between GG and AA participants [16]. However, there was no evidence that the rs1229984 G > A polymorphism was associated with depression, which indicates that the link between alcohol use and depression is unlikely to be truly causal [16].

2.3. Inflammation

Several research groups around the world have explored, and continue to explore, the association between inflammation and depression. In HIMS, we used a common marker of inflammation, C-reactive protein (CRP), to investigate whether inflammation increases the risk of depression in later life. We found that there was a graded association between the serum concentration of CRP and depression, with a 50% increase in the odds of depression among older men with CRP greater than 3 mg/L [17]. However, the association between CRP and depression became less pronounced once concurrent health morbidities were taken into account in the analyses, raising the possibility that CRP (and other inflammatory measures) is simply a marker of physiological stress rather than a cause of depression [17]. Once more, we used Mendelian randomisation to explore the nature of the link between inflammation and depression.

Two polymorphisms of the CRP gene have been associated with high (rs1130864 C > T) and low (rs1205 G > A) basal serum concentrations of CRP. Hence, if CRP concentration is causally related to depression, these polymorphisms should be associated with high and low risk. We found that the rs1130864 C > T polymorphism was associated with higher serum concentration of CRP, as expected, but not with depression [18]. This result showed that high CRP is unlikely to be a cause of depression in later life. In contrast, we found that the rs1205 G > A polymorphism decreases the serum concentration of CRP, as expected, and increased the risk of depression by over 60% [18]. We also found an inverse association between a marker of poor physical health and the serum concentration of CRP among people with depression, suggesting that carriers of the rs1205 G > A polymorphism are less competent at mounting an adequate allostatic response when physically unwell [18]. Thus, raised CRP among older people with depression seems to represent an adaptive response to physiological stress – it is the failure to unleash such an adaptive response that renders older people more vulnerable to experiencing symptoms of depression. Other inflammatory markers, such as cortisol (which is a potent anti-inflammatory hormone), may also play a role in modulating certain aspects of ageing, including vulnerability to depression and cognitive decline [19], [20].

2.4. Homocysteine

Homocysteine is an amino acid derived from the demethylation of the essential amino acid methionine. Re-methylation of homocysteine to methionine takes place in the 5-methyltetrahydro-folate pathway, which relies on the action of the methyltetrahydrofolate reductase enzyme, which in turns uses vitamins B9 (folate) and B12 as co-factors. Methionine is the immediate precursor of S-adenosylmethionine, which is the methyl donor of many methylation reactions in the brain, and 5-methyltetrahydrofolate is involved in the synthesis of the neurotransmitters serotonin, noradrenaline and dopamine [21].

We used data from HIMS to investigate the association between plasma homocysteine and depression and found that the risk of depression in our sample increased 4% for every unit increase in plasma homocysteine, and that older men with total plasma homocysteine ≥ 15 mmol/L were twice as likely as their counterparts to experience clinically significant symptoms of depression [22]. As the plasma concentration of homocysteine can be affected by lifestyle practices, medications and physical morbidity, residual confounding could conceivably explain these results. In order to address this issue, we examined the MHTFR C677t C > T polymorphism, which is associated with increased basal plasma homocysteine. We found that adults with TT genotype have a 20% higher risk of depression than CC carriers [22].

This led us to run a double-blind, placebo-controlled randomised trial to examine the effects of supplementation with vitamins B9 (folate), B12 and B6 on depression outcomes. We found that treatment with vitamins did not increase the efficacy of antidepressant treatment, but decreased relapse of symptoms over the subsequent year [23]. These results suggest that these vitamins may have a role to play in the management of depression, with findings from a separate trial of vitamin B supplementation for the management of stroke survivors lending further support to this conclusion [24].

Of note, the health benefits of vitamin B supplementation may be more noticeable in countries where the enrichment of flour with folate is not mandated by law [25].

2.5. Endocrine associations

Data from HIMS, like that of several other observational studies, have shown that depression is more prevalent among older adults with (<30 nmol/L) than without (≥50 nmol/L) vitamin D deficiency [26]. (25OHD is a hormone rather than a vitamin) However, we found no evidence that vitamin D increased the risk of incident depression during the subsequent 6 years [26], raising concerns that the cross-sectional association between vitamin D and depression could be due to confounding or reverse causality. Indeed, the results of a recent trial of vitamin D supplementation that randomised 9181 adults aged 50 years or over to treatment with vitamin D and 9172 to placebo for 5 years showed that treatment with vitamin D does not decrease the incidence or recurrence of clinically significant symptoms of depression [27].

Diabetes is another common endocrinological disorder that is frequently associated with adverse health outcomes, including depression. As in previous studies, data from HIMS showed that diabetes is more frequent among older adults with than without depression (past or current) [28]. A particularly interesting finding of the study was the observation that the risk of depression was modulated by the duration of diabetes in a ‘J-shaped’ like curve, suggesting that symptoms of depression may arise initially as part of the adjustment process associated with the diagnosis and management of diabetes and, later, as a result of its health complications [28].

HIMS was also used to investigate the association between thyroid dysfunction and depression in later life. Such studies were partly driven by our desire to generate data to support existing guidelines that recommend the assessment of thyroid function among adults presenting with clinically significant symptoms of depression. There was no obvious correlation between thyroid hormonal levels and depression, nor was there evidence that subclinical hypothyroidism or subclinical hyperthyroidism were associated with increased risk of depression [29]. A subsequent large meta-analysis of cohort studies using individual participant data (including data from HIMS) confirmed that subclinical thyroid dysfunction is not associated with depression [30].

We also used data from HIMS to investigate the association between testosterone and depression, given that several other observational studies had suggested the existence of an association between hypogonadism in men and depression [31]. We found that the serum concentration of testosterone was lower in older men with than without depression, although the strength of this association decreased after statistical adjustments were made for other health morbidities [32]. Men with primary hypogonadism had an 86% greater risk of experiencing a depressive episode during the subsequent 9 years than men with normal testosterone serum concentrations [33]. These findings suggested that testosterone replacement may be effective at lessening the severity of depressive symptoms in men with hypogonadism, with existing evidence from randomised controlled trials providing tentative but inconclusive support to this hypothesis [34].

Table 1 summarises evidence from HIMS showing the effect of several exposures on the risk of depression in older men.

Table 1.

Factors associated with increased risk of depression in older men.

	Findings in HIMS	Implications for practice
Cardiovascular diseases	Cardiovascular diseases are more prevalent among older men with than without depression, but this does not change the course of illness nor does it predict death.	The association between cardiovascular diseases and depression is unlikely to be specific or causal. (Chronic diseases and frailty are associated with increased risk of depression.)
Lifestyle	Physical inactivity, smoking, body mass index in the obese range and hazardous or harmful alcohol use are more prevalent in older men with than without depression. These factors also increase the risk of incident depression. Mendelian randomisation suggests that the association with alcohol is unlikely to be causal.	The association between certain aspects of lifestyle and depression have been replicated by several other observational studies. These findings have been supported by trials of physical activity. People with or at risk of depression should be encourage to adopt healthy lifestyle practices.
Inflammation	A higher proportion of older men with than without depression have elevated circulating inflammatory markers.	Several studies have confirmed that inflammatory markers are elevated in people with depression. This link seems complex and may be partly driven by the effectiveness of allostatic mechanisms in addressing physiological challenges.
Homocysteine	High plasma homocysteine is associated with increased risk of depression. Mendelian randomisation studies suggest that this link may be causal.	These findings have been replicated by several studies. Evidence from randomised controlled trials is consistent with a causal link, but results are not yet compelling.
Vitamin D	A higher proportion of older men with than without vitamin D deficiency have depression.	The association between low vitamin D and depression is most likely confounded by other factors. The results of a large randomised controlled trial showed that vitamin D deficiency is not causally linked to depression.
Testosterone	Low serum testosterone is more prevalent among older men with than without depression.	Several observational studies have replicated these findings, but persuasive supportive evidence from randomised controlled trials is lacking.
Thyroid hormones	There is no association between subclinical hypo or hyperthyroidism and depression.	A large meta-analysis of individual participant data from observational studies confirmed that mild thyroid dysfunction does not increase the risk of depression. In the absence of symptoms of thyroid dysfunction, depression should not trigger the assessment of thyroid function.
Diabetes	Diabetes increased the risk of depression, with time from diagnosis modulating risk.	There is a ‘J-shaped’ association between time lived with diabetes and probability of depression. Adaptation to a chronic disease may contribute to initial risk and health complications and frailty to later increased risk. There is no compelling evidence that the association is direct. Decreasing the complications of diabetes may also reduce the burden of depression.

3. The consequences of depression in later life

Depression negatively affects the life of older adults directly and indirectly [35]. There is evidence that depression increases the risk of health morbidities and decreases survival, with data from HIMS contributing to explore possible mechanisms associated with such undesirable outcomes.

3.1. Dementia

The Lancet Commission has recently reported that depression is a potentially modifiable risk factor for dementia, accounting for about 4% of all cases in the community [36]. Such conclusions were derived from observational data and raise questions about the attributed causal inferences of the analyses, particularly because depressive symptoms may represent a prodromal manifestation of dementia. Data from HIMS allowed us to explore the link between depression and dementia in greater detail.

Like previous studies, we found that depression is more frequent among older men with than without cognitive impairment, but we also observed that the association between cognitive impairment and depression did not hold for men with past history of depression who were not depressed at the start of the follow up period of 5 years [37]. We then investigated whether the time between the episode of depression and the diagnosis of dementia influenced this association. We noticed that the association between depression and dementia was only apparent during the first 5 years after the diagnosis of depression, but not for longer periods [38]. We concluded from the above evidence that depression is unlikely to be a cause of dementia – instead, depressive symptoms seem to represent an early clinical expression of dementia in later life. At this stage, available evidence about the nature of the relationship between depression and cognitive impairment or dementia is inconclusive.

3.2. Fractures

Many chronic health conditions have been associated with depression [14], although the direction of this association is not always clear. Data from a large study of primary care patients showed that depression is associated with increased risk of falls in later life, with the strength of the association increasing with increasing age and with the use of antidepressants [39]. Falls are the most important cause of fractures in later life. In HIMS we found that both the history of past depression and depression at the start of the follow up period of 17 years increased the risk of fractures, even when history of past fractures were taken into account in the analyses [40]. The physiological mechanisms supporting this association are not entirely clear, but data from randomised controlled trials of adults exposed to fluoxetine (a selective serotonin reuptake inhibitor antidepressant) showed that treatment increases the risk of fractures [41], possibly by enhancing osteoclastic activity [42] and compromising balance.

3.3. Frailty and death

Many studies have shown that depression decreases life expectancy, but the pathway leading to increased mortality is not clear [6]. A small proportion of these early deaths are due to suicide and, in such cases, depression seems to be an important contributing factor [43]. But even among older people who complete suicide, deteriorating physical health and frailty have been shown to be robust predictors of death by self-harm [43]. Deteriorating physical health and function are important driving factors behind the high rates of suicide among older men.

Data from HIMS have also shown that older men with history of depression are more likely to experience a decline in physical function than their peers without depression, particularly if depressive symptoms are present at the start of follow up [44]. The consistency of the association between depression and frailty has been reported by various independent studies [45] and is consistent with the view that depression may not only increase the risk of frailty, but may be one of the clinical expressions of frailty itself. (See Morley et al. for a detailed explanation about the concept of frailty [46].) Longitudinal data from HIMS are consistent with this view. We examined the mortality of older men over a period of 5 years: men without depression and those with history of past depression had similar mortality hazard, but the men who were depressed at the start of the follow up period were more than 4 times as likely to die [47]. Interestingly, this excess mortality all but disappeared once frailty was included in the model, which suggests that the increased risk of death among older people with depression is strongly associated with deteriorating physical health.

4. Summary

HIMS is an ongoing cohort study of older men that has been following participants for the past 25 years. During this time, the investigators have collected detailed clinical information from participants in 7 different waves of assessments, and this has included the retrieval and storage of serum, plasma, DNA, as well as physical and cognitive measures. Health events and health service use are monitored on an ongoing basis through electronic health data linkage.

Depression has been but one of the health outcomes and exposures investigated in HIMS, with other areas of interest including endocrinology, stroke, ischaemic heart disease, peripheral arteria disease, vascular diseases, cancer, dementia, bipolar disorder, frailty and Parkinson’s disease, among others. The insights gained from investigating systematically the effects of ageing on the health outcomes of well characterised cohorts are clear and can guide the assessment, management and prevention of common health conditions of later life, including dementia [36]. HIMS has contributed data to several multinational collaborations and the investigators welcome the opportunity to engage with novel ideas and colleagues interested in pursuing ageing-related research.

Funding

The work described in this paper has been supported by competitive project grants from the National Health and Medical Research Council of Australia. The funder had no role in the design, collection, analysis and interpretation of the data; nor did it have a role in the writing of the report or in the decision to submit the article for publication.

Declaration of Competing Interest

The author declares that he has no competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Biography

Osvaldo P. Almeida is the Professor of Geriatric Psychiatry of the Medical School of the University of Western Australia. His research interests include clinical trials and cross-sectional and longitudinal studies of older people, with a particular focus on affective disorders.