Mental disorders in the elderly represent a growing concern for public health. Psychiatric disorders are prevalent in elderly patients; major depression, in particular, has an estimated prevalence of 40 % among institutionalised patients aged 75 and over. Mental disorders constitute one of the first causes of morbidity and premature mortality in this population. In Europe, elderly people have the highest rate of suicide (in 2005, 16.4–22.9 per 100,000). Suicide in people aged 60 and over accounts for 30 % of the total number of suicide cases.

Even though the majority of patients with bipolar disorder are diagnosed before the age of 35, 15–20 % are diagnosed after the age of 55 (Almeida and Fenner 2002). The prevalence of bipolar disorder is lower in older than in younger people, estimated at 4 % in the general population (Hirschfeld et al. 2003) and ranging between 0.25 % and 1.03 % (Unutzer et al. 1998; Ritchie et al. 2004; Goldstein et al. 2006) in adults older than 65. The principal assumptions advanced to account for this low prevalence are the premature high death rate of individuals with a bipolar disorder, a certain degree of symptomatic amendment of the disorder with ageing and greater difficulty in diagnosing bipolar disorder in older adults. The latter arises in part from a degree of inadequacy of the diagnostic criteria used to evaluate adult subjects to sufficiently take account of evolutionary characteristics related to age. It should also be noted that the gender ratio among elderly patients with bipolar disorder suggests a female preponderance of 2:1 (Depp and Jeste 2004).

In a cohort of 8,012 elderly subjects aged 60 years and over, the 3-year incidence rate of bipolar I disorder and bipolar II disorder was 0.54 % (SE = 0.09) and 0.34 % (SE = 0.06), respectively (Chou et al. 2011).

The clinical characteristics of bipolar disorder differ in the elderly from those in younger patients, with more cases of rapid cycling and fewer suicide attempts (Oostervink et al. 2009). Several authors propose a phenotypical dismemberment of bipolar disorder according to age at onset of the disorder (Bellivier et al. 2003; Manchia et al. 2008).

Classically, we distinguished early-onset bipolar (EOB) disorder from late-onset bipolar (LOB) disorder; the age threshold retained being generally 50 years. However, this separation lacks consensus (Lala and Sajatovic 2012) even though the two ‘subtypes’ show distinct clinical and evolutionary characteristics. In 2004, a review of the literature found an equivalent sex ratio between EOB and LOB. According to the same study, the LOB was associated with a reduced family concentration of the disorder and with greater neurological comorbidities (Depp and Jeste 2004). Clinical specificities according to age at onset remain a point of controversy (Lala and Sajatovic 2012). For Sajatovic and colleagues, manic episodes may be less frequent and of lower severity among patients presenting a LOB compared to those with EOB, with, moreover, a mood that is more often irritable than inflamed (Sajatovic 2002). Some prior studies (Snowdon 1991), but not all (Schurhoff et al. 2000), suggest greater rates of mixed episodes among patients with LOB.

A recent exploratory study over 2 years compared patients presenting with a manic episode and showed that the LOB subjects reached remission and left hospital within shorter time periods than EOB patients (Oostervink et al. 2009). Lastly, Schurhoff and colleagues showed that late-onset episodes are associated with a better rate of response to lithium (Schurhoff et al. 2000).

Although lithium has long been regarded as the gold standard of long-term mood disorder treatment, prescription practice has gradually evolved in favour of valproic acid. In Ontario, Canada, between 1993 and 2001, the annual number of new lithium prescriptions for patients aged over 65 fell from 653 to 281, while prescriptions for valproic acid increased (Shulman et al. 2003). However, Van Melick and colleagues (2012) showed, over a period ranging from 1996 to 2008, that there was no higher rate of discontinuation or substitution of lithium among elderly patients on that treatment than for other age groups. Furthermore, current recommendations for good practice continue to recommend the use of lithium in elderly bipolar subjects (Llorca et al. 2010).

Reduction in the volume of distribution, concomitant with a reduction in the glomerular filtration rate and thus renal clearance, explains why therapeutic serum lithium concentrations are obtained in older patients with weaker posology than in younger adults (Sproule et al. 2000). In the same way, an increase in the half-life of elimination and possible neurotoxic effects with weaker doses than in younger adults have encouraged some authors to recommend lower therapeutic plasma concentrations in the elderly (0.4–0.7 mEq/L) (Foster 1992).

To date there has been no specific randomised controlled study of the effectiveness of lithium in elderly bipolar subjects for mania or for the prevention of relapse. Nevertheless, some retrospective uncontrolled studies, as well as post hoc analysis of data from a double-blind study, do indicate its effectiveness and tolerance among elderly bipolar subjects (Aziz et al. 2006; Department of Veterans Affairs – Department of Defense 2010). A higher incidence of undesirable effects, reported retrospectively, was found in elderly subjects compared to younger subjects (p < 0.02) (Smith and Helms 1982). Another retrospective study over 7 years showed that the most frequently reported undesirable effect among 114 elderly people presenting with a mood disorder treated with lithium was confusion (19.3 %) (Holroyd and Rabins 1994).