There are a number of issues to be considered before starting specific pharmacological treatment of mood disorders in the elderly⁴. With advancing age, there are important and clinically significant changes in distribution, metabolism and elimination of these drugs. The age- related increase in volume of distribution results in a longer half-life for all psychotropic drugs. Hepatic drug metabolism decreases with age, also resulting in a prolonged half-life, which may be two to three times longer than in younger patients, and the decrease in renal function with advancing age is particularly relevant with regard to lithium, which also results in two to three times higher plasma levels than those in younger patients on the same daily dose. A recent review⁵ emphasized decreases in passive drug absorption, changes in the proportion of body fat (increased) and lean muscle mass (decreased) and the resultant increase in the elimination half-life of (fat-soluble) antidepressants. In addition, changes in binding proteins result in lower free levels of tricyclics but increased free levels of selective serotonin re-uptake inhibitors (SSRIs). Of particular importance, however, has been the study of the inhibitory effects of the SSRIs on the cytochrome p450 enzymes; these pharmacodynamic actions have pharmacokinetic consequences for co-administered drugs, such as tricyclic antidepressants (TCAs), which are dependent on these enzymes for biotransformation. Patients on anticoagulants in particular need close monitoring. The potential of the various SSRIs to cause significant interactions varies considerably; citalopram, escitalopram and sertraline may carry relatively low risk.

To determine the best pharmacological treatment options for individual patients requires careful consideration of a number of clinical factors, which include the following: type of mood disorder; degree of urgency for treatment; previous response to treatment; concurrent medical problems; concurrent drug therapy; risk of overdose; reasonable half-life; dosing flexibility; and affordability⁶.

Older patients with depression can be successfully treated with conventional TCAs, monoamine oxidase inhibitors (MAOIs), SSRIs and atypical antidepressants. TCAs have the major limitations of anti- cholinergic effects, postural hypotension, excessive daytime sedation and cardiotoxicity in overdose. Their advantages, however, are their established efficacy and low cost. Among the TCAs, the secondary amines desipramine and nortriptyline have more favourable side- effect profiles, with desipramine having the fewest anticholinergic effects and nortriptyline causing the least postural hypotension⁶. The secondary amines also have the advantage of therapeutic drug monitoring, with an established therapeutic window for nortriptyline and a therapeutic plasma level of desipramine. Therapeutic drug monitoring enables the clinician to determine the minimal therapeutic dose and to monitor compliance. Nortriptyline has been well investigated for use in elderly patients, including patients older than 80 years⁷ and patients who have had strokes⁸.

Among the MAOIs, phenelzine has been shown to be effective in treating elderly patients with depression⁹‘¹⁰, particularly for patients who could not tolerate TCAs and those with resistant depression. Their main limitation is their interaction with tyramine-rich foods, causing hypertensive crisis. For this, they are superseded by moclobe- mide, a reversible and selective inhibitor of monoamine oxidase type A. Comparative trials in the elderly have established its efficacy compared with imipramine¹¹, nortriptyline¹² and placebo¹³. Moclobe- mide also showed benefits on cognitive function in patients who had dementia and co-morbid depressive symptoms¹⁴.

The SSRIs, however, have provided the most important practical advance in the successful and safe management of depression in late life. This group of drugs includes fluoxetine, fluvoxamine, sertra- line, paroxetine and citalopram. The SSRIs have replaced tricyclic antidepressants as first-line antidepressants for older people as well as for younger adults. More recently, several other antidepressants have become available including bupropion, mirtazapine, another SSRI (escitalopram), and the serotonin and noradrenaline re-uptake inhibitors (SNRIs) venlafaxine and duloxetine¹⁴. The advantages of SSRIs over TCAs are the absence of anticholinergic effects, ortho- static hypotension and arrhythmia in the side-effect profiles, and their safety in overdose⁷. Comparison of fluoxetine with nortriptyline in in-patients with severe depression and heart disease showed that nor- triptyline may be more effective in this population¹⁵. A meta-analysis of the comparative efficacy and safety of SSRIs and TCAs in elderly patients¹⁶ showed no differences in safety and dropout rates.

A recent meta-analysis¹⁷ of the efficacy of second-generation antidepressants (SSRIs, SNRIs bupropion and mirtazapine) showed response rates of 44% for active drug and 35% for placebo, suggesting only modest efficacy (number needed to treat = 11). A similarly small difference in favour of active drug was found for remission rates (33% vs. 26%). There authors reported marked heterogeneity between studies, with superiority over placebo found for fluoxetine in one of three studies, for paroxetine in two studies, for sertraline and for duloxetine but not for escitalopram or venlafaxine. Results for bupropion showed borderline superiority over placebo in terms of response rate but not in terms of remission. Studies lasting 10 weeks or more were associated with greater superiority for active drug over placebo than studies lasting 6-8 weeks, supporting the traditional view that antidepressant response may take longer to emerge in older people¹⁴.

A similar review of single versus dual action antidepressants in older people included head-to-head comparator trials as well as those with a placebo arm, and continuation as well as acute trials¹⁸. They concluded that overall, dual action agents ‘do not appear to confer any additional benefits in efficacy over single action agents (SSRIs)’. Unlike the meta-analysis reviewed above¹⁷, the primary analysis in this study was by class and they did not examine whether there was heterogeneity between individual studies.

One striking feature of these trials is the relatively low overall response and remission rate for active antidepressant medication. Sneed et al.¹⁹ have carried out a systematic review comparing reported response rates in placebo-controlled and comparator trials of antidepressants in older people. They identified nine comparator trials and seven placebo-controlled trials published since 1985 that met their entry criteria. Response rates were considerably higher (60%) than in the placebo-controlled trials (46%); the odds of being a responder were 1.82 times higher in the comparator trials. Type of trial (placebo-controlled vs. comparator) accounted for 27% of the variation in responder rate. The authors conclude that expectation is an important contributor to outcome in such trials and that concern about possibly being on placebo may lead to a suppression of the antidepressant response¹⁴.

The side-effects profile of SSRIs includes nausea, diarrhoea, insomnia, headaches, agitation, anxiety and sexual dysfunction. The SSRIs also seem to worsen Parkinsonism. As with other antidepressants²⁰, there have been case reports of hyponatraemia, hypomania and seizures⁷. An important aspect of their use is their drug-drug interactions, and their inhibitory effects on hepatic cytochrome p450 isoenzymes, the route through which many drugs commonly prescribed for elderly people are metabolized²¹. Paroxetine, norfluoxetine and sertraline have clinically important inhibitory effects (in vivo) on cytochrome P2D6, resulting in increased plasma concentrations of co-administered TCAs, such as desipramine, and antipsychotics, such as haloperidol²². Sertraline, however, had a modest effect on plasma nortriptyline levels in depressed elderly patients²³. Fluoxetine increases plasma levels of co-administered carbamazepine; alprazolam and fluvoxamine increase plasma concentration of co-administered TCAs and antipsy- chotics by inhibiting cytochrome P1A29. An extensive list of drugs metabolized by various p450 isoenzyme types is provided in the reviews by Catterson et al.⁴ and by Rivard⁶.

Despite these potential problems, two recent reviews of (SSRIs) in older people^24,25 have found that they are generally well tolerated but that gastrointestinal (GI) problems including nausea are relatively common. Increased risk of GI bleeding is potentially important because older people are in any case at greater risk of such bleeding. SSRI-induced restlessness, sedation and extra-pyramidal movement disorders may also be particularly disabling in older people. SSRI- induced hyponatraemia is increasingly recognized to be common in older people. Older age, female gender, low body weight and co-administration of diuretics are indicators of increased risk. Inappropriate secretion of antidiuretic hormone (mediated by activation of 5HT2 and 5HT1C serotonergic receptors) is a likely mechanism. Serum sodium should be checked frequently in older patients on SSRIs, particularly in the first few months of treatment. Drug interactions involving the cytochrome P450 enzyme system may be clinically relevant in older people. Patients on anticoagulants in particular need close monitoring. The potential of the various SSRIs to cause significant interactions varies considerably; citalopram, escitalopram and sertraline may carry relatively low risk.

In the light of this it is perhaps surprising that a recent Australian study²⁶ found that although antidepressant use had increased in all age groups and most markedly in the 65+ age group, older people remained the highest users of tricyclic antidepressants whose side effects are well documented as being particularly hazardous in older people. This may, however, reflect the tendency to reuse a previously effective antidepressant for a recurrent episode of depression, and the still frequent use of tricyclics in low dose as hypnotics and analgesics in an older population.

Early studies indicated a poor outcome of late-life depression²⁷, with high relapse, recurrence and chronicity. This view was based on naturalistic observation without monitoring of compliance, adequate dosage and duration of treatment, including prophylactic treatment, and was therefore challenged²⁸. Controlled studies of maintenance antidepressant medication, however, showed a relatively good outcome for late-life depression²⁹. The Pittsburgh group³⁰