Introduction

Pediatric obsessive-compulsive disorder (OCD) is a chronic and disabling disorder and more common than previously thought [1]. Between one third and one half of adults with OCD report that the onset of their symptoms was in childhood [1]. In this paper, we describe the evidence-based treatments for treatment-naïve and treatment-resistant pediatric OCD. Agreed definitions of treatment-resistant and treatment-refractory OCD do not exist [2, 3]. As working definitions, we therefore define treatment-naïve patients as those that have not been exposed to any current first-line treatment (cognitive behavior therapy [CBT] and specific serotonin reuptake inhibitors [SSRIs]). Treatment-resistant patients are those that have failed one of the current first-line treatments [2]. Treatment-refractory patients are those that have failed both of the current first-line treatments (patients that have tried both an adequate trial of CBT and two trials of SSRIs) [2].

We conducted a PubMed search on April 1, 2015. We used the medical subject heading term “obsessive-compulsive disorder”, “child”, and “adolescent”. We also filtered in the terms of “randomized controlled trials” and “meta-analysis”. In addition, we read recent reviews and expert guidelines to find any relevant additional publications [1, 4–11].

We organize the results into the following three sections: (1) evidence-based treatments for treatment-naïve and treatment-resistant patients (CBT, SSRIs, and their combination); (2) evidence-based treatments for treatment-refractory patients; and (3) emerging treatments.

Evidence-Based Treatments in Treatment-Naïve and Treatment-Resistant Pediatric OCD

International expert guidelines diverge in terms of their recommendations for initial treatment. European “guidelines recommend CBT as the first-line treatment for children and adolescents with OCD” regardless of the initial severity and only switching to SSRIs when children decline or are unable to engage in CBT [12, 13]. An SSRI may also be added when the CBT response is inadequate [12]. However, American guidelines recommend CBT alone as the initial treatment for mild to moderate pediatric OCD (CY-BOCS <24) and the combined treatment of CBT and an SSRI for moderate to severe OCD (CY-BOCS > 15). In the following section, we present clinical evidence for these two established treatments and their combination both in treatment-naïve and treatment-resistant patients. We also describe the main results of the predictors of the treatment outcomes.

Treatment-Naïve Pediatric OCD

CBT

CBT is an action-oriented and short-term psychotherapy that focuses on teaching youth with OCD new strategies and responses to the symptoms of OCD. The main treatment components are psycho-education, cognitive training, and exposure and response prevention (E/RP). CBT has been studied extensively and shown to be superior to waitlist [14–19] (Hedge’s g, g = −1.42), pill placebo [20] (g = −0.96), psychotherapy placebo (attention-control) [21–23] (g = −0.74), and SRI [20, 24, 25] (g = −0.39) as reported in recent systematic reviews/meta-analyses [4, 5]. Furthermore, CBT has shown to be an effective treatment for a wide age group down to 3 years [18]. It has also been shown to be effective in a more parent-focused form [18, 22, 26].

Approximately 25 % of children with OCD may reject CBT. In addition, up to 30 % show little or no improvement [9, 20]. Factors have been found that can predict a poor outcome of CBT, such as higher initial OCD severity [27–29], more internalizing and externalizing symptoms/disorders [28–30], and family dysfunction or high family accommodation [27, 28, 31]. Interestingly, family accommodation does not seem to be associated with the clinical outcome when CBT follows a more intensive family approach [29]. Moreover, tic disorders do not seem to be a predictor of poor outcome either [27, 29, 30].

SRIs

The first-line drugs recommended for pediatric OCD [1, 12] are SRIs. This group includes clomipramine (tricyclic antidepressant), which is a non-specific SRI and SSRIs. Clomipramine was the first SRI to be approved by the U.S. Food and Drug Administration (FDA) for the treatment of OCD for children (age 10 years and above) [32]. It is often considered to be more effective than SSRIs (effect size g = –1.09 vs. pill placebo) [5]. However, this large effect size is derived from older studies that may have been less methodologically rigorous. In addition, because clomipramine was the only drug available in this setting, the study participants were usually SRI-naïve. Nowadays, clomipramine is rarely the first drug of choice for children due to its frequent adverse effects [5].

Subsequent drugs approved by the FDA and European Medicines Agency belong to the SSRI group: sertraline [20, 33], fluvoxamine [34], fluoxetine [35, 36], and paroxetine [37]. The FDA approves fluvoxamine for age 8 years and above (dose range 50–300 mg/day), fluoxetine for age 7 years and above (dose range 10–80 mg/day), and sertraline for age 6 years and above (dose range 50–200 mg/day). However, paroxetine does not have an indication in pediatric OCD [1, 38]. The studies on which approval has been based are methodologically rigorous, with an adequate duration (10–12 weeks), an adequate time (4–6 weeks), and potentially effective doses. The mean effect size of placebo-controlled trials of SSRIs for pediatric OCD is in the moderate range (g = −0.43) with no differences between them [5].

Clinical guidelines recommend continuous 6- to 12-month treatment after stabilization [1]. SSRIs are considered to be tolerated well and with clear benefits over adverse events [39]. Long-term (up to 1 year) treatment seems to be necessary [25, 40, 41]. SSRI treatment can be accompanied by both common adverse events such as gastrointestinal distress and rarer symptoms such as an initial increase in suicidal thinking or behavior [1]. However, Bridge [39] noted that suicidality is a relatively small problem with SSRI treatment compared with the benefits, particularly in OCD and anxiety disorders. Clinicians should also be aware of agitation (behavioral activation) in younger children and demotivation [42].

In the most recent review [27] of predictors of SSRI response, comorbid tics and externalizing disorders were associated with a poorer outcome. However, there was little evidence that gender, age, or duration of OCD predicted treatment response. Studies of adults indicate that the hoarding dimension of OCD predicts poor treatment outcome [43]. The only study of a pediatric sample confirming those results was a naturalistic follow-up study of 94 children treated with an SSRI, where children with hoarding symptoms reported worse outcomes than children with contamination obsessions and washing rituals [44].

CBT or SSRIs as the First Choice

Three studies have compared these treatments directly in treatment-naïve and shown the significant pooled superiority of CBT [20, 24, 25]. However, the difference is small and perhaps not clinically significant. In considering this result, treatment providers and patients should also reflect on other factors. First, therapist factors such as expertise may be important. In the largest head-to-head comparison of CBT and SSRIs [20], large and significant site differences were observed, showing that the site with considerable CBT expertise (Penn) had a large CBT effect, while the effect was only moderate at Duke. Second, long-term effects are important. Only one study compares the long-term stability of CBT with that of sertraline. No differences were observed between the groups after 12 weeks. At the 9-month follow-up, the CBT group showed significantly lower rates of relapse. However, the SSRI group discontinued sertraline after 12 weeks [25]. Third, one would need to take into consideration that the recommended minimum duration of SRI treatment is 1 year [41]. By contrast, CBT is usually more time-limited and has a durable effect after treatment termination [25, 45, 46].

Combined Treatment

Two studies have compared the combined treatment with CBT alone [20, 47]. Their combined effect size does not show the superiority of the combined treatment (g = −0.13) [5]. Adult OCD trials show that the combination of CBT and an SSRI generally does not confer additional benefits to CBT alone. The pooled results from a recent meta-analysis of pediatric and adult OCD trials concluded that the combined treatment does not provide an incremental benefit to CBT alone [48].

The POTS [20] trial showed that the combined treatment of CBT and SSRIs resulted in significantly lower total CY-BOCS scores compared with CBT or an SSRI alone at posttreatment. However, the site differences were large, as the CBT group showed a large effect size at one site (Penn) similar to the combined treatment, but a moderate and significantly lower effect size at Duke. Interestingly and perhaps consequently, baseline OCD severity did not turn out to be a moderator of treatment outcomes, indicating that all treatments (CBT alone, an SSRI alone, and the combined treatment) were effective regardless of OCD severity at baseline [28]. These CBT site differences may be explained by various factors such as therapist ability and experience [20], and it is important that such effects are studied. One could speculate that when CBT expertise is available, the addition of an SSRI does not confer an additional benefit, while the lack of CBT expertise requires the addition of an SSRI to obtain the same benefit as when CBT expertise is available. It is necessary to study these effects before recommendations for the combined treatment are given. In conclusion, there is little evidence supporting the additional effect of an SSRI in combination with CBT. Nevertheless, the combined treatment seems to be significantly more effective than an SSRI alone [20, 49] with a pooled effect size of g = −0.59 [5]. The pooled results from a recent meta-analysis of pediatric and adult OCD trials also showed the superiority of the combined treatment over SRI alone [48].

Treatment-Resistant Pediatric OCD

CBT non-Responders

If standard CBT is unsuccessful, the clinician may choose from three different strategies: (1) continue with CBT, (2) continue with CBT and augment with an SSRI, or (3) terminate CBT and switch to an SSRI. International guidelines recommend that SSRIs should be added to unsuccessful CBT [1, 12]. However, the sequence of treatment (i.e., start with CBT and if the patient does not respond, augment with an SSRI) is only based on expert consensus, not empirical evidence. Only one study has compared different strategies for CBT non-responders. Children and adolescents that still had moderate to severe OCD (CY-BOCS >15) after 14 weekly sessions of CBT were randomized to continue with their CBT for ten sessions or switched to an SSRI (sertraline) for 16 weeks. No significant differences between the groups were found for the primary outcomes. However, both groups seemed to benefit from the treatment [50]. Preliminarily, clinicians should thus evaluate whether switching to an SSRI could be a better choice for pediatric OCD CBT non-responders with comorbid tic disorder as shown in a post hoc analysis within the same study [51].

Beyond the available evidence, the clinician must consider a number of factors with CBT non-responders. For instance, partial response to CBT may indicate that continued CBT alone should be tried first. However, when patients do not comply with treatment despite the therapist’s efforts to increase compliance, a natural step would be to switch to an SSRI as guidelines recommend [12]. An additional strategy for CBT or SSRI non-responders is to provide more concentrated CBT. One pediatric OCD open trial with non- or partial responders to SSRIs showed that the majority benefitted by receiving one CBT session per day for 3 weeks [52].

Switching to Another SSRI or Clomipramine

If a child does not respond to an SSRI, the next step could be to switch to another SSRI [1, 12]. However, minimal data exist that can direct the choice [1]. By analogy, switching fluvoxamine non-responders to fluoxetine was effective in the Research Unit on Pediatric Psychopharmacology Anxiety Study [53].

Another strategy recommended by expert guidelines is to add clomipramine to an SSRI [1, 12]. The rationale stated is to combine the serotonergic effects of each medication but to minimize adverse events across different drug classes. Fluvoxamine may have the greatest synergistic effect when added to clomipramine because it is able to inhibit the conversion of clomipramine to desmethylclomipramine and to increase it to a ratio that is in favor of the serotonergic parent compound. One needs to monitor the EKG indices when clomipramine is used, but when combined with an SSRI, such as fluvoxamine, or CYP-450 2D6 inhibitors, such as fluoxetine or paroxetine, additional care must be taken because of the potentially toxic increases in serum clomipramine levels that need to be monitored. Higher dose of clomipramine and lower dose of fluvoxamine (as a metabolic blocker mainly) may work best. Additionally, the risk should be weighed against the harm to the patient from the OCD [1, 54]. The evidence for this combination is limited. Only two case series exist in pediatric OCD showing beneficial effects on OCD symptoms as well as adverse events in the form of [55, 56]. One double-blind RCT in adult OCD showed the superiority of clomipramine + fluoxetine compared to quetiapine + fluoxetine in SSRI non-responders. However, no difference was detected between clomipramine + fluoxetine and pill placebo + fluoxetine indicating that the period of SSRI alone should be extended (optimally with maximum dose) before a combination with clomipramine is applied [57].

It is also important to note that the dissemination of CBT is a large and universal problem; thus, in many areas of the world, children and adolescents with OCD are often treated exclusively with SSRIs [58]. Despite that combining CBT and SSRIs may be necessary and in fact much more effective than continuing with an SSRI alone, even for those that do not respond to SSRIs [59].

Evidence-Based Treatments for Treatment-Refractory Pediatric OCD

This section discusses evidence-based treatments for patients that still have significant and functionally impairing symptoms despite one standard CBT trial and two SSRI trials (or one SSRI trial and a clomipramine add-on) [1, 3]. Expert guidelines recommend pharmacological augmentation or more intensive CBT [1]. Augmentation is the procedure off adding drugs with different mechanisms of action compared with the first-line SSRI in order to aid the therapeutic effects. Research indicates that in addition to serotonin, other neurotransmitters may be involved in the pathophysiology of OCD such as glutamate [10] and dopamine [11]. However, the evidence-based literature—at least for pediatric OCD—is lacking. Thus, pharmacologists always find themselves in off-label settings when treating this group. When appropriate evidence is lacking, we will introduce studies of adult OCD.

Atypical Antipsychotics (Neuroleptics)

The most prominent augmentation strategy in OCD is the use of atypical antipsychotics, also called second generation antipsychotics (SGAs). These work mainly by blocking the dopamine pathways within the cortex. In adults with OCD, randomized controlled trials (RCTs) have been conducted to show the effects of risperidone (e.g., [60]), aripiprazole, quetiapine, olanzapine, and haloperidol, which is a typical or first generation antipsychotic [11, 61]. The effect seems to be greater in patients with comorbid OCD and tic disorder [61]. However, no RCT exists in pediatric OCD, and inferring that children and adolescents should benefit in the same way as adults are fraught with difficulties and depend on the presupposition that adult and pediatric OCD are equivalent, even though the transmitter systems in young people are not mature as those in adults [62]. However, naturalistic open trials using risperidone and aripiprazole as an augmentation to an SSRI in pediatric OCD indicate that the combination may be beneficial [44, 63, 64]. In a naturalistic comparative study, children with tic-related OCD that were non-responders after an SSRI trial (N = 69) showed equal benefits from risperidone and aripiprazole. However, risperidone was associated with serious metabolic adverse events such as weight gain and sedation and aripiprazole to mild/moderate agitation [63]. Similarly, case studies show that risperidone might be beneficial [65–67]. However, many of the patients included were CBT-naïve before using SGAs. A study of adult OCD shows that SSRI non-responders that have not received CBT previously fare better if randomized to CBT in contrast to risperidone [68]. Thus, by taking the effects and adverse events into account, we argue that using SGAs is not a viable option unless adequate trials of CBT and SSRI have been conducted. The risk of untreated OCD should be compared with potential SGA-adverse events, and the pharmacotherapist should follow the advice of Mayan and Correll [69].

Benzodiazepines

Using benzodiazepines such as clonazepam is also off-label [1]. Only one case study exists in pediatric OCD, reporting considerable improvement after a combination treatment of fluoxetine and clonazepam [70]. Two double-blind placebo control studies in adult OCD showed no effect of clonazepam when added to an SSRI in SSRI non-responders [71, 72]. Thus, in view of the current state of knowledge, this strategy is not recommended.

Other Possible Medications

Other potential strategies for treatment-refractory patients are stimulants, gabapentin, sumatriptan, pindolol, morphine, tramadol, anfranal, inositol, opiates, St. John’s wort, N-acetylcysteine, and glutamate antagonists (e.g., memantine and riluzole) [1]. In addition, it has been suggested that switching to serotonin-norepinephrine reuptake inhibitors such as venlafaxine and duloxetine might be beneficial. However, RCTs among adult OCD do not suggest that venlafaxine is more effective than the SSRI paroxetine [73]. In a subsequent study of the non-responders of these drugs, switching from venlafaxine to paroxetine was found to be significantly more effective than vice versa [74]. However, serotonin-related genes may predict which patients respond to SSRIs and which patients respond to serotonin-norepinephrine reuptake inhibitors [75]. Furthermore, increasing SSRI doses have been suggested to be an effective strategy in adults with OCD. One RCT in SSRI non-responders showed the benefits of 400 mg of sertraline compared to 200 mg daily [76]. Indeed, as stated in the most recent expert guidelines, these strategies are off-label, and with the exception of riluzole, they are at most based on a limited number of case reports [1]. The other strategy is augmenting an SSRI with a serotonin 5-HT3 receptor antagonist [77]. One double-blind RCT has shown the benefits of augmenting an SSRI with granisetron in adult OCD [78], and two double-blind RCTs have shown the benefits of ondansetron in adult OCD [79, 80]. These treatments are clearly to be used by experts in true treatment-refractory OCD.

Somatic Treatments

Among somatic treatments is ablative neurosurgery, which has been used for many decades for treatment-refractory adult OCD only [81]. In addition, the emerging treatment of deep brain stimulation involves the surgical implantation of electrodes and targeted electrical stimulations. The advantage compared with neurosurgery is that this procedure is reversible. A review of 90 cases of treatment-refractory adult OCD shows some indications of benefits [82]. Another treatment option is repetitive transcranial magnetic stimulation, which is noninvasive. Here, one uses the magnetic field to apply electrical activity within the cortex. Currently, no evidence supports the use of repetitive transcranial magnetic stimulation in OCD [83].

Emerging Treatments

In this section, we describe novel and experimental treatments. Some of these treatments can be considered for treatment-naïve and treatment-refractory patients alike, while others should be reserved for treatment-refractory patients due to cost or adverse events.

CBT Combination

d-cycloserine and CBT

d-cycloserine is believed to enhance the extinction of learned fear [84]. The theoretical mechanism is that the N-methyl-d-aspartate (NMDA) receptor is involved in fear extinction and that d-cycloserine as an NMDA partial agonist is able to enhance this extinction. As the mechanism of CBT for OCD is to extinguish the relationship between an anxiety-producing stimulus and obsessive fear by exposing patients to these stimuli, one could potentially enhance this extinction by taking d-cycloserine approximately 1 h before CBT sessions. Two RCTs have confirmed that d-cycloserine has an additive effect when combined with CBT, as it was proven to be superior to CBT + pill placebo [85, 86]. In line with this synthesized theory, d-cycloserine does not seem to work when administered after E/RP sessions [87]. Pending further studies, the augmentation of d-cycloserine may become a viable and safe option in the future to reduce OCD symptoms quickly and effectively.

CBT with other Combinations

CBT is ineffective for some patients as they engage poorly with E/RP or other components of the treatment. Adding motivational interviewing may thus be helpful. One preliminary RCT showed that adding motivational interviewing to intensive CBT for 3 weeks was more effective than adding extra psycho-education sessions [88].

Attention modification programs (AMPs) have also been suggested as an add-on to CBT/SSRI. AMPs are used to disengage attention from cues that are considered to be threatening. One preliminary RCT with severe or complicated cases of anxiety disorders exists (the majority of children had OCD as their primary diagnosis). The participants added either AMPs or attention control conditions to their CBT, and the results showed clear benefits to the AMP group [89].

Modified CBT

Modifying CBT may enhance response or reduce cost and thus the availability of CBT. By comparing these with established standard CBT, one can examine whether they outperform CBT in terms of mean effect, the effect of a subset of patients, or cost reduction. International expert guidelines recommend intensive CBT as a strategy for treatment-resistant or treatment-refractory patients [1, 12]. However, research on the effectiveness of intensive CBT is limited. One RCT shows that daily CBT for 3 weeks is as effective as weekly sessions over 14 weeks. Additional well-powered studies are thus needed. Two open trials exist in treatment-resistant patients showing the benefits of intensive CBT [52, 90].

Group-based CBT seems to be equally effective as individual CBT [14, 45]. Group CBT has been studied more extensively in adults with OCD, and the results from well-designed studies consistently show no differences between CBT and group CBT [91–93]. However, group CBT is dependent on a steady flow of patients [94], meaning that it may be more feasible in large rather than small clinics. Such a group setting also provides less therapist contact than an individual setting.

However, some therapeutic benefits might be unique to group CBT such as the modeling of CBT exercises, support from peers, reinforcement of homework exercises, normalization and de-stigmatization of the symptoms of OCD, and acquisition of knowledge on other OCD symptoms [94]. Hence, there might be an added benefit to combine a group setting with one-to-one contact between a therapist and a patient. One RCT in adults with OCD shows the benefits of an individually delivered intensive E/RP in a group setting [95].

Another cost-effective approach might be to minimize the initial therapeutic resources while maximizing the effects as much as possible (e.g., by providing bibliotherapy as the first step [96]). While no studies of the benefits of bibliotherapy in pediatric OCD exist, one open trial showed the effects of computerized bibliotherapy with minimal therapeutic contact [97]. Optimally, this treatment form should be compared with standard CBT in a stepped care study in order to evaluate both effects and cost [96].

Another treatment form is to reduce the number of sessions. One brief format of CBT that consisted of five sessions was compared with standard CBT (12 sessions) and waitlist. While both CBT forms showed superiority over waitlist, there was no difference between the two [15].

Remote CBT is another viable option where CBT is delivered through telephone or videoconferencing. One RCT showed that CBT delivered through Skype was superior to waitlist [16]. At least one uncontrolled feasibility study also exists [98]. A non-inferiority RCT comparing telephone-delivered CBT with standard in-office CBT showed that CBT could be effectively delivered over the telephone [99].

Medications

Recent evidence from a number of directions (genetic, neuroimaging, animal models, and treatment studies) implies that an abnormal concentration of glutamate, the main neurotransmitter within cortico-striatal-thalamic circuits, may be associated with OCD [10, 100]. Thus, researchers have shown increasing interest in the potential efficacy of glutamate-targeting agents for OCD treatment. One of these is riluzole, which has been approved by the FDA for treating amyotrophic lateral sclerosis. Although uncontrolled studies had previously shown effects in children [101] and adults [102, 103], the first RCT trial conducted among (S)SRI non-responders did not show any effect over pill placebo. However, 28 % of participants had concomitant autism disorder [104].

Other glutamatergic drugs are ketamine, memantine, topiramate, and lamotrigine [10]. No studies in the pediatric OCD population exist [6]. N-acetylcysteine, a naturally occurring amino acid, is also a glutamate-modulating drug with a beneficial safety profile that may be a good alternative to children that have failed both CBT and two SSRI trials.

Conclusion

What are the implications of the results of current studies for clinical guidelines? First, when CBT expertise is available, one can successfully treat pediatric OCD patients with CBT and continue even when the patient does not respond after a standard duration of treatment. Second, when CBT expertise is absent, treatment-naïve patients can confidently be treated with an SSRI. However, if they do not respond, previous studies indicate that the addition of full CBT may be necessary to obtain a response [59].

Moreover, American clinical guidelines recommend the combined treatment of CBT and SSRIs for patients with moderate to severe OCD [1]. The pooled results of the combined treatment of CBT and SSRIs were not found to be superior to CBT alone [20, 47] with no evidence of different effects based on symptom severity [28]. Thus, the data support the European guidelines [12]. One could hypothesize that CBT alone is as good as the combined treatment when delivered by experts [20, 47]. When the results of CBT are attenuated for some reason, the addition of medication may be important [20]. However, continuing CBT beyond a standard trial of 14 sessions seems to be a viable option [50].

The evidence beyond first-line treatments for pediatric OCD is limited. More studies are needed in CBT non- or partial responders. Moreover, we need new compounds that have higher efficacy than the (S)SRI class of drugs, based on the improved understanding of OCD neurobiology and neurochemistry that is emerging. Some forms of augmentation such as SGAs have limited evidence for their effects and may be dangerous with regard to adverse events; hence, they should only be used by experts. Long-term follow-up studies of patients recruited to RCTs are also needed to increase our understanding of the duration and robustness of the treatment effects.