How to address the placebo response in the prescription SSRIs and SNRIs in children and adolescents

ABSTRACT

SSRIs and SNRIs are prescribed as first-line pharmacological treatment for common mental disorders in children and adolescents. Despite their efficacy, they have a high risk for adverse events and exhibit a substantial placebo response. This editorial provides some background on the current evidence on the topic and suggests to carefully weigh the benefits of SSRIs and SNRIs against their potential harms. Therefore, the authors present two different set of conclusions – one for clinical practice, and one for future research designs.

KEYWORDS:

• Antidepressants
• children
• adolescents
• placebo response
• mental disorders
• efficacy
• safety

1. Introduction

Currently, more than half of the pharmacological treatments used for children are off-label or unlicensed drugs [1] and child health-care clinicians must often rely on evidence stemming from adult populations [2]. Ethical and methodological barriers in pediatric research such as difficulties enrolling children, small market share for active substances aimed at children, and low prevalence of many pediatric mental disorders [3] have led to the scarcity of high-quality pediatric randomized controlled trials (RCTs) [4]. This is especially problematic since both the safety and efficacy profiles of medications may be significantly different for children than for adults, due to divergences in developmental physiology, disease pathophysiology, or developmental pharmacokinetics and pharmacodynamics [5].

Selective serotonin reuptake inhibitors (SSRIs) are first-line pharmacological treatments, and serotonin-norepinephrine reuptake inhibitors (SNRIs) are considered second- or third-line treatments for common mental disorders in children and adolescents. These include depressive disorders (DDs), anxiety disorders (ADs), obsessive-compulsive disorder (OCD), and posttraumatic stress disorder. Among others, benefits of SSRIs and SNRIs include symptom reductions for cognitive-affective symptoms (e.g. attention, memory, depressed mood, and hopelessness), sleep symptoms (e.g. insomnia), somatic symptoms (e.g. pain), as well as anxiety symptoms [6]. Nevertheless, it should be noted that little research has been conducted on the efficacy of antidepressants on specific disorder-related symptoms [7].

The current meta-analytical evidence in the field of child and adolescent psychiatry is limited and includes a recent meta-analysis from our group, published in 2017 in JAMA Psychiatry [8] (see Table 1 for an overview). The results indicate that SSRIs and SNRIs are more effective than placebo in treating common psychiatric disorders in children and adolescents. In the DD group, SSRIs and SNRIs did not differ significantly; however, SSRIs were significantly more effective than SNRIs in the AD group. In general, the standardized drug-placebo difference in our analysis was found to be small and varied significantly by disorder, with effect sizes (Hedges g) ranging from 0.20 for DD, to 0.39 for OCD, and 0.56 for AD. This is in line with an earlier meta-analysis that found a mean improvement (Hedges g) of 0.20 in DD symptomatology, of 0.41 in OCD, and of 0.69 in non-OCD ADs (i.e. generalized AD, social AD, social phobia, separation/generalized AD) [9]. With regard to single drugs, a recent network meta-analysis found that only fluoxetine was better than placebo (standardized mean difference of 0.51) in DD [10]. However, meta-analytic evidence is always based on aggregated data and extrapolating these results to individuals can be inappropriate, while analyses of individual patient-level data in turn can help to explore the influence of individual characteristics on treatment effects [11].

Table 1. Drug-placebo differences in selected recent meta-analyses.

	Efficacy Hedges g (95% CI)			Safety TEAEs SAEs
First author (year)	DD	AD	OCD	DD	AD	OCD	DD	AD	OCD
Bridge (2007)	0.20^NR (0.12–0.29)	0.69 ^NR (0.47–0.91)	0.41^NR (0.26–0.56)	NR	NR	NR	RD (95% CI): 1%^a (−0.1%-2%)	RD (95% CI): 0.7%^a (−0.4%-2%)	RD (95% CI): 0.5%^a (−1%-2%)
Ciprani (2016)^a	0.51 ^NR (0.03 to 0.99) (Fluoxetine) 0.35 ^NR (−0.54 to 1.24) (Duloxetine)	NR	NR	NR	NR	NR	OR (95% CI): 0.90 ^NR (0.49–1.49) (Fluoxetine) OR (95% CI): 1.07 ^NR (0.51–2.01) (Duloxetine)	NR	NR
Locher & Koechlin (2017)	0.20*** (0.13–0.27)	0.56*** (0.40–0.72)	0.39*** (0.25–0.54)	RR (95% CI): 1.46* (1.03–2.07)	RR (95% CI): 1.40 (0.93–2.12)	RR (95% CI): 1.89** (1.23–2.88)	RR (95% CI): 1.99** (1.33–2.97)	RR (95% CI): 1.48 (0.77–2.83)	RR (95% CI): 1.35 (0.47–3.92)

^a No overall effect sizes, results are selected from the network meta-analysis. *<0.05 **<0.01 ***<0.001.

Note. CI: confidence interval; NR: not reported; OR: odds ratios for suicide-related outcome; RD: pooled risk difference of suicidal ideation/suicide attempt; RR: risk ratio for mean number of TEAEs and SAEs per patient across symptoms; SAEs: serious adverse events; TEAEs: treatment emergent adverse events

Besides the disorder-specific efficacy of antidepressants over placebo, two further findings should be emphasized. First, it is important to consider the increased risk for adverse events, especially in the pediatric population. On the one hand, this includes treatment-emergent adverse events (TEAEs), most prominently headache, nausea, insomnia, abdominal pain, agitation, and diarrhea. On the other hand, severe adverse events (SAEs), such as an increased risk of suicidal thoughts and behavior, can occur. The latter led to the implementation of a ‘black box’ warning on the labels of all antidepressants for pediatric use by the FDA in 2004. However, it remains controversial due to conflicting findings of reanalyses of the data [12]. In our analysis, we found that rates of TEAEs and SAEs are significantly higher in antidepressant groups compared to placebo [8]: when examining the percentage of patients with TEAEs in the antidepressant groups compared to placebo, the risk ratio (RR) was 1.07 (CI 1.01–1.12), indicating a small yet significantly higher risk for TEASs in the antidepressant group when compared to the placebo group. When evaluating the mean number of TEAEs per patient across symptoms, the RR was 1.49 (CI 1.22–1.82). Further, when looking at SAEs, the RR was 1.76 (CI 1.34–2.32), which is in line with other meta-analytical findings [9,13] that also found significantly more SAEs in the antidepressant group compared to the placebo group.

Second, the substantial placebo response that is present in antidepressant trials warrants consideration. Here, it is important to distinguish between placebo effect and placebo response. While the latter encompasses the observable response to the administration of a placebo pill as well as natural course of the given disorder, the former can only be estimated by the subtraction of the placebo response from the natural course [14]. Due to ethical concerns of including a nonintervention group in children and adolescents in RCTs, no conclusions regarding the placebo effect in this age group can be drawn to date.

From studies in healthy children, we know that the placebo effect arises from learning processes that involve conditioning, formation of expectation, and social learning mechanisms [15] and it can be elicited by factors such as the patient-clinician relationship and patients’ expectations of improvement [13]. However, as justifiably noted in a recent meta-analysis [16], factors associated with the placebo response, which are both clinician specific (e.g. experience with the disorder under investigation) and patient specific (e.g. treatment expectation), are difficult, if not impossible, to consider in meta-analytical analyses. Still, a previous meta-analysis found that expectancy of receiving a drug rather than placebo seems to be less important in children and adolescents [17] than in adults. Instead, they found a significant interaction between participants’ age and the amount of contact with the research staff, in that older participants showed a higher placebo response with increasing number of study visits compared to younger participants. The placebo response may in fact explain much of the variability in pediatric antidepressant trials, especially in trials of pediatric depression, where response to placebo is higher compared to trials of OCD and non-OCD ADs [18]. Indeed, in our meta-analysis, we found a large placebo response in pediatric DD (within-group effect size: g = 1.57) and AD (g = 1.03), and a moderate response in OCD (g = 0.63).

2. Expert opinion

We advocate to carefully weigh the benefits of SSRIs and SNRIs against their potential harms. If antidepressants are prescribed, we suggest to communicate and explain all components of the clinical response, thus also address the substantial placebo response and explain possible underlying mechanisms. Here, two main questions should be considered.

First, how can aspects and processes which underlie these substantial placebo responses best be applied in clinical practice within an ethical framework? This essentially involves the following principles: (I) to speak honestly and openly about the mechanisms underlying the pharmacological treatment, the empirical status of the treatment, including placebo responses and their underlying processes [19]. Also, the associated side effects and their likelihood of occurrence need to be described. This will address the ethical key principles of patient autonomy and transparency. (II) Given the importance of the role of the clinician providing the treatment [20], prescribers should be open about their own personal expectations and experiences with the given pharmacological treatment and ask about the patient’s idiosyncratic expectations in order to increase the treatment’s credibility and plausibility. (III) Since the placebo response is most likely related to the patient’s feeling of being understood and cared for, an emotionally warm and empathic style in order to obtain a good patient-clinician relationship should be prioritized [21].

Second, how can future studies advance our knowledge with regard to the safety of SSRIs and SNRIs? One option might be to evaluate placebo pills as dose extenders. Given a drug’s dose-dependent efficacy and side effect profile, this might allow us to reduce the dose of the active drug to a level at which side effects are better tolerable for patients, while still being effective in treating the disorder. To our knowledge, only one study has looked at the potential of placebo to serve as a dose extender of pharmacotherapy in children [22], focusing on children receiving amphetamine salts for ADHD. In this study, after a conditioning phase of pairing the most effective dose with placebo, the dose of the drug was reduced by half and supplemented with an openly administered placebo pill. Interestingly, this experimental group showed a similar pattern of symptom severity and a decrease in reported side effect severity compared with a full dose group. These very preliminary results point in an innovative direction for future studies, as well as potentially important clinical implications. However, studies in depression are needed to test these assumptions. A different innovative study design that has been shown to be effective in several conditions [23,24] in adults is called open-label placebo. This entails openly prescribed placebo and a scientific rationale and discussion with the patient why and how placebo works. However, the only study in depression has been conducted in adults and has shown limited evidence for symptom improvement [25]. In order to test this in a pediatric sample, the design would need to be adapted carefully: originally, the open-label placebo rationale was designed and applied for several pain conditions such as irritable bowel syndrome [26]. We propose that the open-label placebo rationale would need to be modified in accordance with the specific needs of depressed patients in order to gain credibility (e.g. focus on fostering the feeling of hope). Moreover, the rationale would need to be comprehensible for children as done in a study with pediatric patients suffering from ADHD [27].

Taken together, we suggest carefully weighing the benefits of SSRIs and SNRIs against potential harms. Additionally, placebo mechanisms can and should be used clinically to boost treatment effects while possibly aiding in obtaining favorable treatment results with minimal doses of antidepressants in children and adolescents.

Declaration of interest

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties. Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Additional information

Funding

This paper has not been funded.