Antidepressant treatment of depression in dementia

In the last 2 years, the two largest randomised controlled trials (RCTs) on the antidepressant treatment of depression in dementia have been published: the HTA-SADD study from the UK, with 326 participants, and DIADS-2 from the USA, with 131 participants (Banerjee et al., 2011; Rosenberg et al., 2010). Together these studies have more than trebled the number of participants enrolled in RCTs of depression in dementia. Each study was a large multicentre trial involving collaborations of the leading academic old age psychiatrists from their respective countries. Both studies showed no benefit of sertraline over placebo, while the HTA-SADD study also found no benefit of mirtazapine over placebo.

In their viewpoint, which only considers the HTA-SADD study, Macfarlane et al. (2012) provide a critique of the study methodology and make recommendations to clinicians and policy-makers. They caution readers about possible study limitations and that it is premature to conclude that antidepressants are ineffective in this population. While this is undoubtedly correct, it is just as important to examine the implications of these study findings for future research and current clinical practice.

Mounting RCTs of depression in dementia is a very difficult task. Despite the epidemiological data that suggest up to 20% of people with dementia suffer from depression (Draper, 1999), when it comes to recruitment into RCTs cases are hard to find. Indeed, the HTA-SADD study had to reduce their target sample size from 507 to 339 and was still unable to reach their target despite an extended period of recruitment (Banerjee et al., 2011). What is the best way to recruit subjects? Do you rely upon cases being treated in a specialist old age mental health service such as in the HTA-SADD study and then be criticised because you may be missing the typical case seen in primary care? Or do you take a case-finding approach such as we used in a RCT of treating depression and psychosis in dementia in nursing home residents, where subject recruitment was based on screening scores on depression scales undertaken in a facility-wide survey followed by a medical assessment of screen positives to ensure that subjects met study criteria (Brodaty et al., 2003)? In this latter approach the critique can be that because the subjects (or their carers) were not seeking treatment for their depression they might not be representative of people with depression in dementia that require treatment.

One of the fundamental dilemmas is case definition. Although there are diagnostic criteria for depression in dementia, it is still not clear whether they represent a distinct disorder. It has long been recognised that depression in dementia can have atypical presentations (Draper, 1999), hence the criteria are broad enough to allow for this, but the cost might be the inclusion of participants with questionable depression. The approach adopted in the HTA-SADD study was to use a cut-off score of 8+ on the Cornell Scale for Depression in Dementia (CSDD), rather than depression diagnostic criteria for study entry. Macfarlane et al., (2012) suggest that this raises the possibility that participants might not have had the type or severity of depression more likely to respond to antidepressants. DIADS-2 did use diagnostic criteria for depression in Alzheimer’s disease for study entry but allowed participants to have major depression, minor depression or Alzheimer’s associated affective disorder (Rosenberg et al., 2010). Outcome in both studies was not related to depression severity, type, or medication dose (Banerjee et al., 2011; Drye et al., 2011).

What might be getting lost in these study outcomes is the observation that despite the majority of the participants having suffered from depression for 6 months or more before study entry, each treatment group had a satisfactory treatment response that was not related to symptom duration. That is, if these were open label trials the reports might have suggested that the antidepressants seemed to be effective. So what is happening? In the HTA-SADD study all participants had specialist old age mental health team involvement in their care, while in DIADS-2 there was monthly attendance at a specialist clinic or outreach service and carers received a standardised psychosocial intervention that consisted of 20–30 minute counselling sessions at every study visit (with phone sessions between visits if needed), were provided educational materials, and had 24-hour access to crisis management.

Perhaps it is these interactions with specialist clinicians for both the carers (informal and professional) and participants that had an impact upon mood and outcome. If this is the case then specialist old age mental health services already have effective treatments available to them by simply providing specialist assessment and ongoing support of people with dementia and their carers augmented by the use of psychosocial interventions such as activity-based care (Macfarlane et al., 2012; Brooker et al., 2011). This has workforce implications as the staff required to deliver this level of service provision would be much higher than what is currently available in most parts of Australia and elsewhere (Draper and Anderson, 2010). Of course as many of the participants in these studies were in residential care, efforts to improve the quality of care in residential aged care facilities by adopting person-centred care as a standard of care might have the impact of reducing the rates of depression in these facilities and hence the potential demand on specialist services.

Many unanswered questions remain about the antidepressant treatment of depression in dementia. How effective are other types of antidepressants, including selective noradrenaline reuptake inhibitors and monoamine oxidase inhibitors (MAOI)? One large multicentre RCT that is usually excluded from meta-analyses due to missing data in the paper found that the reversible MAOI moclobemide was more effective than placebo (Roth et al., 1996). Does the type of dementia matter? Studies to date have either been on participants with Alzheimer’s disease or unspecified dementia. Do individuals with longstanding recurrent unipolar depression or bipolar depression have a different acute antidepressant response than individuals that have only developed depression during the evolution of their dementia? And how long should maintenance antidepressant treatment in these early-onset mood disorders be continued? There is already evidence that the introduction of cholinesterase inhibitors might increase the risk of depression relapse (Reynolds et al., 2011). Finally, what is the most effective treatment for very severe depression in dementia, cases usually excluded from RCTs due to risk, consent and urgency?

In conclusion, these studies should not be interpreted as evidence for clinicians to abandon their use of antidepressants in dementia, but rather for these drugs to be used as a second-line approach to specialist psychosocial interventions or first line when the severity of symptoms demands their early introduction.

See Viewpoint by McFarlane et al., 2012, 46(7): 595–597