Research reportModeration of antidepressant and placebo outcomes by baseline severity in late-life depression: A systematic review and meta-analysis
Introduction
Although the placebo effect and its moderators have been examined extensively in adult populations with major depressive disorder (MDD) (Brunoni et al., 2009, Kirsch et al., 2008), comparable studies for late-life depression are scarce. There is no agreement upon definition of late-life depression; the term may be used to refer to patients with symptoms that fall on a continuum from sub-threshold to clinically significant, and a minimum age criterion in the range 55–65 years (Rodda et al., 2011). MDD is the most common psychiatric disorder in elderly people, showing a point prevalence of 4.6–9.3% (Meeks et al., 2011). In addition, subclinical symptoms such as minor depression and dysthymia are more common in old age, with a point prevalence of 10% (Pinquart et al., 2006). All of these forms of depression have been found to have a negative influence on the quality of life (Nelson et al., 2013). Late-life depressive disorders also increase disability (Nelson et al., 2013), are associated with poorer outcomes in clinically significant illnesses (Jiang et al., 2001), and a higher suicide rate (Conwell et al., 2002).
With regard to effective treatment of depression in elderly patients, practice guidelines identifies both antidepressants and psychotherapeutic interventions as a first line treatment for MDD, especially for mild to moderate depression, and a combination thereof or antidepressants alone for severe depression (American Psychiatric Association, 2010). Given that psychotherapy and pharmacotherapy did not show strong differences in effect sizes in elderly patients in a direct comparison (Pinquart et al., 2006), the authors recommend that treatment choice should be based on other criteria, such as contraindications, treatment access, or patient preferences. For neuropharmacological practice, selective serotonin reuptake inhibitors (SSRIs) and other second-generation antidepressants medications should be considered over monoamine oxidase inhibitors or tricyclic antidepressants (American Psychiatric Association, 20102015, Rodda et al., 2011). Moreover, antidepressant use in elderly people with depression increased over the last years, mainly due to a growing SSRI-use (Sonnenberg et al., 2008). SSRIs have been shown to be superior to a placebo pill in controlled clinical trials and meta-analyses investigating late-life depression (Kok et al., 2012, Mittmann et al., 1997, Nelson et al., 2008). However, overall drug effects in elderly patients with symptoms of depression are only modest, with an odds ratio (OR)=1.40 (95% CI: 1.24–1.57) for response (i.e., ≥50% improvement from baseline on mood scales), and OR=1.27 (95% CI: 1.12–1.44) for remission (i.e., no longer meeting diagnostic criteria) versus placebo in a meta-analysis of 10 trials (Nelson et al., 2008).
With regard to possible moderators of pharmacological and placebo outcomes in depression, mixed-age studies have repeatedly shown that the mean differences between groups treated with antidepressant medication and placebo become larger as baseline severity increases (Fournier et al., 2010, Khan et al., 2002, Kirsch et al., 2008). It is unclear whether the increasing benefits, as severity increases, of drug treatment over placebo treatment are due to a decrease in the response to placebo treatment or an increase in the response to pharmacological intervention. The data reported by Kirsch et al. (2008) indicated that the increased benefit of drug treatment for severely depressed patients is related to a decrease in responsiveness to placebos, with no change in responsiveness to the drug. However, two meta-analyses have shown that initial severity predicted symptom improvement in adult patients who took antidepressant medication (Fournier et al., 2010, Khan et al., 2002). In the Khan et al. (2002) analysis, improvement as a function of baseline severity increased in drug groups but decreased in placebo groups. In Fournier et al. (2010), improvement as a function of severity increased significantly in both drug and placebo groups (as would be predicted by regression toward the mean), but the increase was significantly larger in the drug group. It should be noted that a re-analysis of the Kirsch et al. (2008) data set, which controlled for the effect of structural coupling (this occurs when baseline values and change score are coupled algebraically, thus possibly leading to an inflated association between the variables; Tu et al., 2004) concluded that baseline severity did not influence treatment outcome (Fountoulakis et al., 2013).
Studies looking at predictors of treatment outcome in elderly patients with depression are limited and most studies in this field do not focus on baseline depression severity. To date, symptom severity at baseline has not been shown to be a moderator of outcome in depressed elderly people. A meta-analysis by Gibbons et al. (2012) found that in a geriatric subgroup, baseline severity was not related to a positive treatment outcome for fluoxetine compared with placebo. Another meta-analysis found an association between initial severity and drug over placebo efficacy in elderly patients who had suffered from depression for at least 10 years, but not in the majority of patients, who had a shorter disease history (Nelson et al., 2013). However, there are several limitations to the reported meta-analyses. First, they rely on a limited number of studies, thus Gibbons et al. (2012) included 4 geriatric studies, whereas Nelson et al. (2013) included 10 trials of second-generation antidepressants in patients with late-life depression. Second, the authors included only a restricted range of baseline severity scores as they focused on MDD. However, only a minority of significantly depressed elderly patients fulfill the diagnostic criteria for depression, yet the rate of sub-threshold late-life depression rises with age and is responsible for comparable disability and distress (Pinquart et al., 2006).
Consequently, to assess treatment effects in late-life depression, a meta-analysis including a broader range of studies and taking minor depression and dysthymia into account is of a high relevance. With this background, we undertook a systematic review and meta-analysis to test the assumption that mean differences between antidepressant and placebo interventions become larger as baseline severity increases in a geriatric population.
Section snippets
Search strategy and eligibility criteria
We performed searches in Cochrane, Embase, PsycINFO, PubMed, and Web of Science on studies published through September 30, 2014. Search terms were adapted to the electronic bibliographic databases and consisted of keyword combinations based on the inclusion criteria (for details see Appendix). In addition to the systematic search, the references of all included articles were reviewed.
We included peer-reviewed randomized, double-blind, placebo-controlled clinical trials reported in English or
Study selection and study characteristics
The study selection procedure is shown in Fig. 1. In total, 19 studies met inclusion criteria and provided relevant data for the meta-analysis. The trials included a total of 5737 elderly depressed patients, of whom 3226 received active drug and 2511 received placebo. Sample sizes of included studies were between N=16 and N=747. Publication year ranged from 1980 to 2014. Most trials were based on a parallel design, except one study, which used a crossover design (Brody et al., 2011). We
Discussion
The purpose of the present meta-analysis was to investigate the moderating effects of baseline severity on mean outcome measures in depressed elderly patients treated with antidepressants or placebos.
Concerning HDRS scores, we did find an increase in mean change in depressive symptoms with increasing baseline severity within antidepressant and placebo interventions. However, one must be careful in interpreting relations between baseline severity and within-group changes, as they can be strongly
Role of funding source
The authors received no specific funding for this study.
Conflict of interest
We wish to confirm that there are no known conflicts of interest associated with this publication and there has been no significant financial support for this work that could have influenced its outcome.
We confirm that the manuscript has been read and approved by all named authors and that there are no other persons who satisfied the criteria for authorship but are not listed. We further confirm that the order of authors listed in the manuscript has been approved by all of us.
We confirm that we
Acknowledgment
This study was supported by Grant project (P300P1_158427) awarded to Joe Kossowsky by the Swiss National Science Foundation. The authors thank Ted Kaptchuk for provided mentoring, which was supported by NIH Grant #2K24AT004095.
References (87)
- et al.
Placebo response and antidepressant response
Am. J. Geriatr. Psychiatry
(2007) - et al.
Executive dysfunction and the course of geriatric depression
Biol. Psychiatry
(2005) Selective serotonin reuptake inhibitors versus tricyclic antidepressants: a meta-analysis of efficacy and tolerability
J. Affect. Disord.
(2000)- et al.
Late-life depression: evidence-based treatment and promising new directions for research and clinical practice
Psychiatr. Clin. North Am.
(2011) - et al.
Loss of expectation-related mechanisms in Alzheimer׳s disease makes analgesic therapies less effective
Pain
(2006) - et al.
Escitalopram in the acute treatment of depressed patients aged 60 years or older
Am. J. Geriatr. Psychiatry
(2008) - et al.
Antidepressants in elderly: metaregression of double-blind, randomized clinical trials
J. Affect. Disord.
(2013) - et al.
Which patients improve: characteristics increasing sensitivity to a supportive patient–practitioner relationship
Soc. Sci. Med.
(2011) - et al.
Risk factors for suicide in later life
Biol. Psychiatry
(2002) - et al.
Randomized, double-blind, placebo-controlled trial of fluoxetine treatment for elderly patients with dysthymic disorder
Am. J. Geriatr. Psychiatry
(2005)
Biological, clinical, and ethical advances of placebo effects
Lancet
"Mini-mental state". A practical method for grading the cognitive state of patients for the clinician
J. Psychiatr. Res.
Relationship between the Hamilton depression rating scale and the Montgomery–Asberg depression rating scale in depressed elderly: a meta-analysis
Am. J. Geriatr. Psychiatry
Are drug and placebo effects in depression additive?
Biol. Psychiatry
Efficacy of treatment in older depressed patients: a systematic review and meta-analysis of double-blind randomized controlled trials with antidepressants
J. Affect. Disord.
A tune in “a minor” can “b major”: a review of epidemiology, illness course, and public health implications of subthreshold depression in older adults
J. Affect. Disord.
The efficacy, safety and tolerability of antidepressants in late life depression: a meta-analysis
J. Affect. Disord.
Efficacy of second generation antidepressants in late-life depression: a meta-analysis of the evidence
Am. J. Geriatr. Psychiatry
The placebo-effect in heart-failure
Am. Heart J.
Acute and long-term treatment of late-life major depressive disorder: duloxetine versus placebo
Am. J. Geriatr. Psychiatry
A double-blind, placebo-controlled study of venlafaxine and fluoxetine in geriatric outpatients with major depression
Am. J. Geriatr. Psychiatry
Trends in antidepressant use in the older population: results from the LASA-study over a period of 10 years
J. Affect. Disord.
Mathematical coupling can undermine the statistical assessment of clinical research: illustration from the treatment of guided tissue regeneration
J. Dent.
A comparison of placebo effects in clinical analgesic trials versus studies of placebo analgesia
Pain
Development and validation of a geriatric depression screening scale: a preliminary report
J. Psychiatr. Res.
The cost of dichotomising continuous variables
Br. Med. J.
Beck depression inventory (BDI)
Arch. Gen. Psychiatry
Operating characteristics of a rank correlation test for publication bias
Biometrics
Mechanisms of placebo and placebo-related effects across diseases and treatments
Annu. Rev. Pharmacol. Toxicol.
Depression and neurological disorders
Curr. Opin. Psychiatry
Mechanisms and clinical implications of the placebo effect: is there a potential for the elderly? A mini-review
Gerontology
Sustained cognitive impairments after clinical recovery of severe depression
J. Nerv. Ment. Dis.
Introduction to Meta-Analysis
Psychosocial and clinical predictors of unipolar depression outcome in older adults
Int. J. Geriatr. Psychiatry
Treatment of depression associated with age-related macular degeneration: a double-blind, randomized, controlled study
Ann. Clin. Psychiatry
Placebo response of non-pharmacological and pharmacological trials in major depression: a systematic review and meta-analysis
PLOS One
Severity of depression and risk for subsequent dementia: cohort studies in China and the UK
Br. J. Psychiatry
Clinical features and efficacy of escitalopram treatment for geriatric depression
J. Int. Med. Res.
The combination of estimates from different experiments
Biometrics
The Hopkins Symptom Checklist (HSCL): a self-report symptom inventory
Behav. Sci.
Influence of context effects on health outcomes: a systematic review
Lancet
A nonparametric "trim and fill" method of accounting for publication bias in meta-analysis
J. Am. Stat. Assoc.
Cited by (23)
Efficacy of light therapy for a college student sample with non-seasonal subthreshold depression: An RCT study
2020, Journal of Affective DisordersCitation Excerpt :Importantly, there was a significant correlation between the baseline expectancy of improvement and the improvement in depressive symptoms over the course of the study (Rutherford et al., 2013). Consistent with this finding, several clinical trials reported strong placebo effects on reducing depressive symptoms when patients were aware that they were receiving medications (Locher et al., 2015; Rutherford et al., 2009; Zilcha-Mano et al., 2019). Nevertheless, we feel that not all of the improvement recorded for the low-intensity group was due to a placebo response.
Optimal doses of antidepressants in dependence on age: Combined covariate actions in Bayesian network meta-analysis
2020, EClinicalMedicineCitation Excerpt :The current recommendation is therefore that the lower range (20–40 mg/dayFE) of the licensed dose (20–80 mg/dayFE) probably achieves an optimal balance between efficacy and tolerability for the majority of patients receiving antidepressants in MDD [1]. The meta-analysis by Furukawa et al. [1] however did not adjust dosing for age, another important demographical covariate in antidepressant treatment [4–10]. Antidepressant use in the elderly ( > 60 years) is associated with an increased risk of potentially clinical significant adverse events (AEs).
The placebo and its effects: A psychoneuroendocrinological perspective
2019, PsychoneuroendocrinologyCitation Excerpt :Also, using a placebo version of social-evaluative stress, i.e. omitting this characteristic constituent of stress, resulted in the expected lack of the cortisol stress response in healthy subjects (Het et al., 2009). Matters become considerably more complicated when addressing the concundrum of specific actions of antidepressants on central glucocorticoid receptor-mediated processes (Pariante et al., 2012) in face of large to very large placebo effects in depressive disorders across all age groups (Khan and Brown, 2015; Locher et al., 2015, 2017a, 2017b; Rutherford et al., 2017), which are most likely mediated through opioidergic action (Peciña et al., 2015). Likewise, it would be of interest to scrutinize why and how placebos are clinically effective in disorders known to be marked by a hypofunctional hypothalamic pituitary adrenal (HPA) axis, such as irritable bowel syndrome (Kaptchuk et al., 2008, 2010; Videlock et al., 2016), chronic low back pain (Carvalho et al., 2016; Sudhaus et al., 2009) as well as chronic fatigue syndrome and fibromyalgia (Davidson et al., 2011; Tak et al., 2011) and whether and how these inert, but otherwise effective treatments effect HPA axis functioning in these syndromes.
Factors associated with placebo response in depression trials: A systematic review of published meta-analyses (1990–2017)
2018, Neurology Psychiatry and Brain ResearchCitation Excerpt :Ten studies were excluded from further consideration because of not fulfilling the selection criteria for pooled analysis and/or quality of underlying study results. The examination of the reference lists of the remaining 18 studies yielded an additional 40 studies, resulting in a total of 58 studies included in the current systematic review (Fig. 1) (Bridge, Birmaher, Iyengar, Barbe, & Brent, 2009; Brown, Johnson, & Chen, 1992; Brunoni, Lopes, Kaptchuk, & Fregni, 2009; Dodd, Berk, Kelin, Mancini, & Schacht, 2013; Dunlop et al., 2012; Entsuah & Vinall, 2007; Entsuah, Shaffer, & Zhang, 2002; Evans, Sills, Wunderlich, & McDonald, 2004; Faries et al., 2000; Fountoulakis & Möller, 2011; Fountoulakis, Veroniki, Siamouli, & Moller, 2013; Fournier et al., 2010; Furukawa et al., 2016; Gibbons, Hur, Brown, Davis, & Mann, 2012; Greenberg, Fisher, & Riter, 1995; Hunter, Cook, & Leuchter, 2010; Iovieno & Papakostas, 2012; Iovieno, Tedeschini, Ameral, Rigatelli, & Papakostas, 2011; Iovieno, Tedeschini, Levkovitz, Ameral, & Papakostas, 2012; Khan, Bhat, Faucett, Kolts, & Brown, 2011; Khan et al., 2010; Khan, Brodhead, Kolts, & Brown, 2005; Khan, Faucett, & Brown, 2014a; Khan, Faucett, & Brown, 2014b; Khan, Khan, Walens, Kolts, & Giller, 2003; Khan et al., 2004; Khan, Leventhal, Khan, & Brown, 2002; Khan, Schwartz, Kolts, Ridgway, & Lineberry, 2007; Khin, Chen, Yang, Yang, & Laughren, 2011; Kirsch et al., 2008; Klemp, Tvete, Gåsemyr, Natvig, & Aursnes, 2011; Lam & Andersen, 2006; Lee, Walker, Jakul, & Sexton, 2004; Locher et al., 2015; Mallinckrodt, Meyers, Prakash, Faries, & Detke, 2007; Mancini et al., 2014; Melander, Salmonson, Abadie, & van Zwieten-Boot, 2008; Montgomery & Kasper, 2007; Nelson, Delucchi, & Schneider, 2013; Nelson et al., 2012; Papakostas, Fan, & Tedeschini, 2012; Papakostas & Fava, 2009; Posternak & Zimmerman, 2007; Quitkin et al., 2002; Rabinowitz et al., 2016; Rief et al., 2009; Rutherford et al., 2013; Rutherford et al., 2011; Rutherford, Tandler, Brown, Sneed, & Roose, 2014; Sinyor et al., 2010; Stein, Baldwin, Dolberg, Despiegel, & Bandelow, 2006; Stolk, ten Berg, Hemels, & Einarson, 2003; Tedeschini et al., 2011; Thase et al., 2007; Trivedi & Rush, 1994; Turner, Matthews, Linardatos, Tell, & Rosenthal, 2008; Undurraga & Baldessarini, 2012; Walsh, Seidman, Sysko, & Gould, 2002). The vast majority of identified meta-analyses (75%) were published between 2005 and 2015.
Predictors of treatment outcome in depression in later life: A systematic review and meta-analysis
2018, Journal of Affective DisordersCitation Excerpt :Our findings were different from other meta-analysis studies because we looked at the predictor's effect on overall outcome, whereas others focused on the predictor's effect on the differences between intervention and comparison groups. Locher et al. (2015) did not find a relationship between baseline severity and change in symptoms in either antidepressant or placebo group in their meta-analysis. However, the study did not consider treatment duration in their analysis and limit to antidepressant studies, which may have produced the non-significant result.
Initial symptom severity of bipolar I disorder and the efficacy of olanzapine: a meta-analysis of individual participant data from five placebo-controlled studies
2017, The Lancet PsychiatryCitation Excerpt :Moreover, baseline symptom severity seems to have an effect on drug–placebo difference in symptom improvement in bipolar I disorder, as was shown to be the case in schizophrenia and autism.7,8 Concerning major depressive disorder, the relationship between baseline symptom severity and drug–placebo difference remains complex; findings of different studies provide inconsistent results that preclude any straightforward answer.2–5,42–46 In part, this inconsistency might be explained by limitations of aggregated data meta-analysis, which use mean values of studies instead of individual patient data.
- 1
Contributed equally to the article and should both be considered first authors.