Tailored treatment for depression with co-morbid dementia

Banerjee et al. (2011) have provided good news concerning treatment of depression when co-morbid with dementia. Their double-blind, placebo-controlled trial of sertraline and mirtazapine versus placebo showed a 38% reduction in mean score (from 13 to 8) on the Cornell Scale for Depression in Dementia (CSDD) (Alexopoulos et al., 1988) within 13 weeks, and the score was much the same at 39 weeks. However, the reductions were no greater in those treated with antidepressant than in the group given placebo. All participants had been referred to and treated by old age psychiatry teams. The authors concluded that antidepressants given with the usual care provided by such teams are no more clinically effective than placebo.

The study has been criticised. McFarlane et al. (2012, in this issue) note that two-thirds of those recruited (after referral to specialist services) had been depressed for more than 6 months. They therefore suspect that a treatment-resistant group of patients was selected for the trial. They suggest that referral to old age psychiatric teams usually occurs only after primary care doctors have prescribed first-line treatment for a time – and those referred are patients who have remained depressed in spite of treatment. These authors suggest that many of those referred may have had dysthymic disorder rather than major depression, because their mean CSDD score was only 13 at trial entry. It is also possible that teams and their patients were less likely to consider inclusion in a placebo-controlled trial if the depression was severe. McFarlane et al. (2012) suggest that participants given antidepressants were under-treated, and they question an absence of MMSE data for 23% of participants.

A major criticism of the trial’s methodology was the use of a screening tool (the CSDD) to rate whether participants were depressed. People scoring in the so-called depressed range on the CSDD may do so because of apathy, agitation or other features attributable to frontal or other brain damage rather than depression. The Geriatric Depression Scale (GDS) (Yesavage and Brink, 1983) appears to be better at screening for depression in those with MMSE scores of 15 or higher, but neither the GDS nor CSDD should be used for diagnosing depression. Structured clinical interviews, had they been conducted, would have allowed calculation of remission and response rates, and formulations about pathogenic factors and type of depression.

Banerjee et al. (2011) should also be criticised for (apparently) not giving attention to various factors that could affect outcome when treating depression in cases of dementia. They did not report data regarding proportions of early onset and late onset depression in the antidepressant and placebo groups, nor the proportions with no previous history of depression (all of whom, by definition, had late onset depression). Others have found more than two-thirds of late life depression (LLD) cases to be of late onset, many of them showing evidence of executive impairment and what has been called ‘vascular depression’ (Alexopoulos et al., 1997).

Further, in considering relevant factors, why did Banerjee et al. (2011) not report or control for use of other medications, especially those psychotropic drugs used to decrease behavioural and psychological symptoms of dementia?

When formulating plans for treatment of depression with co-morbid dementia, it is important to consider whether a person’s depression can be considered wholly or partly as an adjustment disorder with depressed mood (i.e. primarily psychological), depression attributable to neurobiological changes, depression primarily due to another medical condition, or what. Although given attention by McFarlane et al., the labels ‘dysthymia’ and ‘major depression’, if applied in cases of Alzheimer’s disease, are not particularly helpful, because they largely indicate severity rather than aetiological factors. Medical illness is a well-established risk factor for depression, and functional impairment due to disability has been reported as the most significant predictor of the development of pervasive depression (Prince et al., 1998). Antidepressants can be effective in a range of medical illnesses. Banerjee et al. (2011) gave scant details about the prevalence and type of medical diseases in their antidepressant and placebo groups. Nor did they provide details concerning whether people in the antidepressant and placebo groups might have been adjusting to worries, stresses or other depressing circumstances that might have led to elevated CSDD scores, whatever their degree of cognitive impairment.

Trials have shown antidepressants to be modestly effective in treatment of LLD. Nelson et al. (2008) reported pooled response rates of 44.4% (antidepressant) versus 34.7% (placebo). Successive augmentation strategies can lead to a cumulative response rate of more than 80% (Dew et al., 2007). There is as yet insufficient evidence to say whether LLD with co-morbid dementia can be helped by similarly persistent strategies to achieve comparable response rates. Banerjee et al.’s study, despite its faults, has shown that considerable reductions in CSDD scores can be achieved in such cases when a specialist service provides care – and because placebo appeared just as effective as antidepressants, it is reasonable to suggest that non-pharmacological elements (which lacked the risky side effects of antidepressants documented in their report) were responsible. There is good reason to follow advice from Lenze (2011), echoed by McFarlane et al. (2012), when treating co-morbid depression in cases of dementia. He recommended starting with watchful waiting rather than initiating antidepressant treatment, unless there are good reasons for expecting pharmacotherapy to be effective (e.g. a good response previously to such treatment, or depression that is severe and/or melancholic in type). Lenze recommended consideration of environmental changes, optimising pain relief and medical rehabilitation. Enhancement of sense of self-control may empower functionally challenged older people and help them manage the adversity of disability (Jang et al., 2011).

An attentive and enthusiastic service is probably the most important aspect of care needed to help depressed people with dementia to recover a sense of joy in life (see Banerjee et al., 2011). Evidence supports the provision of ‘tailored treatment’ when seeking to relieve depression in cases of dementia. A review of possibly contributory factors, and personalised attention from empathic health professionals, with a flexible approach, is needed.

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