There is a consensus that antidepressant therapy is effective and should be recommended for the treatment of depression in the elderly. Because of the pharmacokinetic changes in the elderly, the tolerability of antidepressants, especially of the classic tricyclics, is reduced compared with younger populations. However, this problem can be managed by a careful dosing regime and a careful selection of the drug ().
Efficacy of antidepressant treatment in elderly depressed patients
There have been approximately 30 randomised placebo-controlled, parallel-group clinical trials involving elderly patients that support the efficacy of acute treatment (). Medications used have included nortriptyline, imipramine, doxepin, bupropion, desipramine, nomifensine, phenelzine, fluoxetine, fluvoxamine and mirtazapine.
The efficacy of antidepressants in the treatment of elderly depressed patients is also supported by a large amount of evidence from studies comparing different second-generation antidepressants with classic antidepressants, including mianserin, trazodone, fluoxetine, paroxetine, moclobemide, brofaromine and mirtazapine ().
To collect some more precise information about studies performed since 1980, my group performed a careful review of the literature (). Although we took into account only publications of an appropriate standard, we had to conclude that the methodological standard in this field is not comparable with the general standard (). As can be seen from Fig. 12.2, most compounds reduced the baseline total score on the Hamilton Rating Scale for Depression (HRSD) between 50% and 70%. Some of the studies are reported below.
Gerner et al (1980) showed a clear advantage of imipramine and trazodone over placebo. Considering the total number of side-effects, trazodone was comparable to placebo, while imipramine produced the most severe side-effects, which were mostly anticholinergic in nature. The total study duration of four weeks can be regarded as too short. Georgotas & McCue (1989), for example, reported that half of those elderly patients who did not respond to seven weeks of treatment with either phenelzine or nortriptyline responded during an additional two weeks. Furthermore, 26 dropouts (out of 60) seems high. In the imipramine group, as many as 60% of patients dropped out, which means that only eight out of 20 completed the trial in this group.
Scardigli & Jans (1982) compared mianserin with trazodone. The four-week period of the trial was too short. They postulated a quicker onset of action for mianserin, but due to the problematic use of multiple statistical testing without adjustment, this result does not seem to be acceptable. With mianserin, the total number of side-effects was considerably lower than with trazodone, which is mostly due to more anticholinergic side-effects in the trazodone group.
Gwirtsman et al (1983) compared maprotihne with doxepin in a six-week trial. They found maprotihne to be superior to doxepin, but this difference was apparent only after week 4 of the trial. Before this time, the two groups showed parallel improvement. The investigators undertook serum-level measurement but were not able to find a correlation between plasma level and recovery. There were no major differences regarding safety and tolerability. Since the authors state only the minimum dose given and the mean maximum daily dose, a mean daily dose is not clearly ascertainable.
Cohn et al (1984) were able to show the clear advantage of imipramine over placebo even with the small patient number in both groups (n = 21). In the imipramine group, a fairly high dose was used, which led to a side-effect ratio of 86% (19% in the placebo group). The most commonly reported side-effects were nervousness/restlessness, dry mouth and nausea/vomiting. This paper again shows the clear efficacy of this classic antidepressant drug, but also the poor tolerability of this compound in an elderly patient sample.
Merideth et al (1984) were able to show the superiority of imipramine over placebo in a small sample. The main side-effects in the imipramine group were dry mouth (36%), constipation (24%), drowsiness (18%), nervousness/restlessness (18%) and blurred vision (14%). The trial aimed primarily at the comparison of nomifensine and imipramine.
Eklund et al (1985) performed a comparison between mianserin and imipramine for four weeks, which seems too short. They found comparable efficacy. Because of a lower side-effect rate in the mianserin group, they concluded that mianserin is better tolerated. Since only completers were analysed, the fact that there were four dropouts with severe side-effects in the mianserin group (three with confusion and disorientation, one suicide) but none due to side-effects in the imipramine group is not represented in this analysis.
Ather et al (1985) report superiority of trazodone over amitriptyline in both efficacy and safety in a six-week trial. In the light of the fact that 76.5% of amitriptyline patients were on 50 mg a day during the whole trial period and that the “superior efficacy” was apparent at only one point of time (week 6) in one (current severity of patient’s illness) out of five measurements of efficacy, this does not seem to be a well founded conclusion. In their paper, it is not stated which parameters were prospectively defined as primary efficacy parameters. In this trial, 10 mg/day diazepam was chosen as reference, and no marked differences were detected between the tranquilliser and the two antidepressants. This raises the question of sample selection.
Feighner & Cohn (1985) report interesting results with a comparative trial of fluoxetine and doxepin. Both drugs were effective, with advantages in the safety parameters for fluoxetine. Doxepin caused, in particular, dry mouth, drowsiness/sedation, constipation, and dizziness/lightheadedness, while nervousness/anxiety and drowsiness/sedation were predominant with fluoxetine. What is striking in this trial is the high percentage of dropouts in both groups: only 53% of the fluoxetine group and 39% of the doxepin group completed the trial.
Georgotas et al (1986) undertook a very interesting trial comparing nortriptyline, phenelzine and placebo. First of all, the duration of seven weeks is sufficient, which is also proved by the fact that it took five weeks to observe a significant antidepressant response. Secondly, the dose was optimised, either on the basis of plasma levels of nortriptyline or of the amount of monoamine oxidase (MAO) inhibition as measured by platelets (in the phenelzine group). Therefore the fairly high response rate of approximately 60% in both groups is not astonishing. The trial must have been supervised very well, as evidenced by the low dropout rate, which was highest in the placebo group. There was an overall good tolerability of the trial drugs, with certain advantages for phenelzine concerning anticholinergic side-effects, and certain advantages for nortriptyline concerning nasal congestion, urinary symptoms and dermatological problems. The most astonishing result therefore is that classical MAO inhibitors also seem to be safe and effective in this special patient population.
Wakelin (1986) reports the results of a pooled analysis from three independent trials comparing fluvoxamine, imipramine and placebo. The simultaneous use of retrospective data pooling and statistical hypothesis testing for the whole sample is a critical issue. The stated earlier onset of efficacy of the fluvoxamine group was not a prospectively defined aim of the trial, and it was found by multiple retrospective testing of subfactors of the HRSD. The claim that fluvoxamine is better tolerated than imipramine was proved only for orthostatic blood pressure regulation, but the number of adverse experiences did not differ significantly between the two active-treatment groups. The treatment duration of four weeks can be regarded as too short.
Siegfried & O’Connolly (1986) found comparable efficacy of mianserin and maprotiline. The trial duration of four weeks seems too short for such a conclusion. In the maprotiline group, considerably more side-effects occurred, but this could be due to the low dose of mianserin. The trial seems to have been performed on a more representative group of elderly depressed patients than in other studies because nearly all patients had concomitant somatic diseases. Although such inclusion criteria make the primary diagnosis slightly more uncertain, this approach increases the representativeness and the generalisability of the results. In this trial, a cognitive test battery was used to measure the cognitive effects of the drugs. In the critical flicker fusion frequency test and in the choice reaction task it was claimed that mianserin is superior to maprotiline.
Altamura et al (1989) found similar antidepressant effects for amitriptyline, mianserin and trazodone. The doses in the control groups (amitriptyline and mianserin) were rather low, so the result may be biased in favour of trazodone. Trazodone shows an advantage concerning the number of side-effects, especially anticholinergic and cardiovascular side-effects, which may be especially useful in patients with cerebral or cardiac diseases.
Fairbairn et al (1989) reported equal efficacy of lofepramine and dothiepin but better tolerability of lofepramine. These advantages were reported only for dry mouth, daytime drowsiness and blurred vision, and the results of the overall analysis of the total number of adverse events are not given.
De Vanna et al (1990) found no differences in the efficacy and tolerability of moclobemide and mianserin. The latter is true only for the overall evaluation of tolerability, because the frequency of adverse events was a disadvantage in the moclobemide group. The treatment period (four weeks) seems too short.
Hutchinson et al (1991) reported the results of a six-week trial with 90 patients to test paroxetine and amitriptyline (in a 2:1 randomisation) . The statistical analysis is very good (intent-to-treat principle, power calculation), and the adverse events and patient withdrawals are also reported comprehensively. They state a comparable efficacy of the two treatments, but with a quicker onset in the paroxetine group. The latter statement is based on the fact that more patients in the paroxetine group showed a 50% reduction in HRSD score over weeks 1 and 2 (11% v. 3% and 27% v. 17%, respectively). This difference was not tested statistically and, taking the same figures for week 6 into consideration (76% v. 80%), one might even speculate that amitriptyline is more effective. The major side-effects in the paroxetine group were nausea, vomiting and dizziness, and in the amitriptyline group dry mouth, nausea, dizziness, somnolence and asthenia.
Dunner et al (1992) pooled the data of two independent trials of paroxetine versus doxepin, without reporting the results of the individual trials. They were able to show a similar efficacy to doxepin and even superiority in certain items (e.g. HRSD score, depressed mood or the CGIC severity of illness scale), although the dose of the comparator, as they state themselves, was fairly low. There were safety advantages of paroxetine, especially concerning anticholinergic side-effects, but in the paroxetine group more patients stopped the trial due to drug-related side-effects (19% v. 15%).
Moller & Volz (1993) reported on a total sample of 189 elderly patients who were randomised in a 2:1 ratio to either brofaromine or to imipramine. Brofaromine and imipramine were found to be similarly effective, but the lack of a placebo group makes it difficult to draw any final conclusion about overall efficacy. In terms of tolerability, brofaromine was superior to imipramine (not, however, in the intent-to-treat analysis), but there was no major difference concerning the relative number of adverse events (31.8% v. 35.9%). The major side-effects in the brofaromine group were restlessness, vertigo, nausea and headache, while in the imipramine group dryness of mouth, impaired vision, vertigo and headache were observed. Their paper discusses the problem of low dosing in the imipramine group (87 mg/day). A direct comparison with a result in a normal-aged group is possible, since the same design was used in another large trial ().
Currently, few data are available to indicate for how long patients should take an antidepressant. Practice guidelines, however, suggest continuing the dose for six to eight weeks after the therapeutic dose has been satisfactorily established to determine an initial response. For the patient who has responded to the drug, it is recommended to continue for at least six months, to make certain that remission can be stabilised. Patients with histories of recurrent depression probably should be maintained on the antidepressant indefinitely to prevent recurrence. For bipolar patients, lithium should be the first choice for long-term treatment. With respect to elderly patients there are insufficient empirical data in this field, especially concerning long-term prophylactic treatment ().
In cases of therapy resistance, strategies similar to those used for younger populations are suggested, among others augmentation therapy with lithium or, in cases of delusional depressions, combination with a neuroleptic. However, the empirical base is weaker than for younger populations. In cases of severe non-response, electroconvulsive therapy is indicated ().
Safety of antidepressant treatments for elderly patients
The safety of antidepressants is a very important issue, especially in the elderly. When choosing a tricyclic antidepressant to treat major depression in elderly patients, safety aspects influence the decision more than efficacy aspects.
Different pharmacological groups can be differentiated with respect to safety, as is the case with the antidepressant treatment of younger populations. The main groups are covered briefly below.
Tricyclic antidepressants. They have a well known complex side-effects profile, resulting from their interaction with several central nervous receptor sites. Blockade of muscarine receptors can cause dry mouth, tachycardia, sweating, urinary retention and confusion. Lack of tolerability can make it difficult to achieve the appropriate dose for an elderly patient. Postural hypotension and confusion, associated with tricyclics, can lead to falls, with devastating consequences in the elderly, such as hip fracture. Cognitive dysfunctions can be observed in the elderly treated with anticholinergic tricyclics. Because suicide is a potential risk in elderly depressed patients, the danger of overdosing must be recognised. A dose of more than 1500 mg of imipramine equivalents can be lethal. The secondary amine tricyclics nortriptyline () and desimipramine are considered safer than the tertiary amine tricyclics imipramine, amitriptyline and doxepine, and are preferred among the tricyclics for the treatment of the elderly depressed.
Selective serotonine reuptake inhibitors. Selective serotonin reuptake inhibitors (SSRIs) like fluoxetine, paroxetine, citalopram, fluvoxamine and sertraline may be especially beneficial for the elderly because of their selective mode of action (). Their selectivity for serotonin receptors results in relatively few, mostly gastrointestinal side-effects, such as nausea. The SSRIs are usually free of cardiovascular effects; furthermore, they produce negligible, if any, anticholinergic, antihistaminic or alpha-adrenergic reactions. They do not interfere with cognitive function. There are some data showing that in a group of patients with depressive illness, treatment with paroxetine produces more improvement of cognitive function than does fluoxetine (). In contrast, significant impairment of cognitive function was observed with tricyclics (). Safety in overdose might be of particular importance in the treatment of elderly patients.
Monoamine oxidase inhibitors. These drugs interact with thiamine-rich food, thus limiting their use in elderly patients (problem of hypertensive crisis). However, the selective MAO-A inhibitor moclobemide does not induce such problems and is generally well tolerated by elderly patients.
Coming back to our own review of the studies published since 1980, Fig. 12.3 gives an overall impression of safety and tolerability. This can only be an outline, since the ways in which the side-effects were ascertained differed significantly from trial to trial. The tricyclics show a clear predominance of antichohnergic side-effects, above all dry mouth and sedating effects, whereas the SSRIs show a predominance of nausea and vomiting, although dry mouth is also one of the most important side-effects in this group. This was also the most common symptom in the trials with MAO inhibitors, followed by syncopes and dizziness in the phenelzine trial and by restlessness in the brofaromine trial. Compared with classic antidepressants, there appear to be safety and tolerability advantages for SSRIs or MAO inhibitors, and this should have an effect on clinical practice since non-compliance – a practical problem in elderly patients – is often caused by side-effects. The new generation of antidepressants seems to possess a certain advantage in this field, so that their use in this patient group might especially increase compliance (). However, some authors assume that the efficacy of an SSRI will be inferior to that of tricyclics in the treatment of severe depression ().
Selections from the book: “Late-Onset Mental Disorders: rates of antipsychotic drugs depend on the elderly patient’s characteristics. Medical comorbidity should be taken into consideration as well as concomitant medication. The starting dose of antipsychotic drugs in older patients should be in the region of 25-50% of that recommended for younger patients ().
Selections from the book: “Late-Onset Mental Disorders: The Potsdam Conference” (1999)