6.2.3.1. Clinical evidence statements⁸

6.2.4.1. Clinical evidence statements

6.2.5.1. Clinical evidence statements

6.2.6.1. Clinical evidence statements

6.2.7.1. Descriptive review

Five trials examined the continuation of treatment in patients with OCD (ANSSEAU; LOPEZ-IBOR1996; MONTGOMERY1993; BEASLEY1992; GREIST1995A).

In MONTGOMERY1993, treatment responders (N = 173) continued their blinded treatment for an additional 16 weeks (after an 8-week, double-blind, placebo-controlled trial of three fixed doses of fluoxetine; total N = 217).

Response was maintained during the extension phase. The mean improvement of Y-BOCS scores was fluoxetine 20 mg (−1.3, N = 18); fluoxetine 40 mg (−1.6, N = 20); fluoxetine 60 mg (−1.7, N = 23); placebo (−1.8, N = 12). However, there was no significant difference between fluoxetine and placebo on mean improvement on the Y-BOCS.

ANSSEAU was an unpublished 30-week extension phase trial of a 12-week acute-phase RCT (ZOHAR1996a) that compared the maintenance of efficacy in patients who had responded to paroxetine, clomipramine or placebo. Patients (N = 83) continued on the drug they received during the acute phase.

Ninety per cent of patients in the paroxetine group had maintained a response to treatment (defined as a reduction of at least 25% in the total Y-BOCS scores), compared with 89.5% of patients in the clomipramine group and 75% of patients in the placebo group. However, there was no significant difference between the continuation of paroxetine and placebo (N = 63; SMD = −0.51; 95% CI, −1.15 to 0.12) or between the continuation of paroxetine and clomipramine (N = 71; SMD = −0.01; 95% CI, −0.53 to 0.5) on reducing obsessive-compulsive symptoms as measured by the Y-BOCS. There was also no significant difference between treatment groups on the likelihood of reporting adverse events, paroxetine versus placebo (82% versus 66%, N = 63; RR = 1.24; 95% CI, 0.81 to 1.88), paroxetine versus clomipramine (82% versus 75%, N = 71; RR = 1.1; 95% CI, 0.83 to 1.46).

In a continuation study of BEASLEY1992 (Tollefson et al., 1994), treatment responders (N = 76) continued their blinded treatment for an additional 6 months after a 13-week, double-blind, placebo-controlled trial of three fixed doses of fluoxetine (N = 355).

Sixty-seven per cent of placebo-treated patients improved on the Y-BOCS at the end of the responder extension phase, as did 69.6% of fluoxetine 20 mg-treated patients, 76.2% of fluoxetine 40 mg-treated patients, and 76.9% of fluoxetine 60 mg-treated patients. Combining the fluoxetine-treated groups, 74.3% of patients experienced improvement beyond that seen during the 13-week acute phase. In terms of the number of patients who achieved ≥25% improvement in their Y-BOCS score, there was no difference between any dose of fluoxetine and placebo and between different doses of fluoxetine, 40 mg versus 20 mg (RR = 0.94; 95% CI; 0.57 to 1.54), 60 mg versus 20 mg (RR = 0.7; 95% CI, 0.4 to 1.21) and 60 mg versus 40 mg (RR = 0.74, 95% CI, 0.41 to 1.32).

In GREIST1995a, responders to a 12-week randomised trial of one of three fixed doses of sertraline (50, 100 or 200 mg) or placebo were continued on their treatment. Responders to the acute phase, defined as ‘marked’ or ‘moderate’ improvement on the CGI Efficacy Index were offered an additional 40 weeks of double-blind treatment at their assigned doses. Three hundred and twenty-five patients entered the acute phase, of which 125 patients were classified as responders. One-hundred-and -eighteen patients entered the continuation phase.

Overall, the pooled sertraline group exhibited greater improvement on measures of efficacy over the 48-week treatment period, Y-BOCS mean change scores (SDs): sertraline −5.7 (7.4) versus placebo −2.8 (5.8), F (1,289) = 7.06, p = 0.001. When changes from week 12 to endpoint were examined on efficacy, no differences were seen among any of the individual treatment groups or between the pooled sertraline group and placebo. Patients in the sertraline groups were more likely to report adverse events compared with patients in the placebo group, sertraline 50 mg 94%, sertraline 100 mg 91%, sertraline 200 mg 96%, placebo 79%. Two patients in the sertraline group and none in the placebo group attempted suicide.

LOPEZ-IBOR1996 continued patients for 12 weeks on the same double-blind fluoxetine versus clomipramine treatment received during an 8-week acute phase. Responders (fluoxetine, n = 11; clomipramine, n = 13) received a lower dose of the drug (20 mg fluoxetine and 100 mg clomipramine), while non-responders (fluoxetine, n = 14; clomipramine, n = 8) received a higher dose of the same acute-phase drug (60 mg fluoxetine and 200 mg clomipramine).

No formal statistical analysis took place due to the small size of the treatment arms. However, among responders to fluoxetine the mean endpoint Y-BOCS score was lower (8.8, SD 5.79) compared with baseline (9.6, SD 5.35), whereas among responders to clomipramine the mean endpoint score was higher (13.7, SD 11.79) compared with baseline (11.4, SD 6.13). Among non-responders, there was a decrease in the mean Y-BOCS score from baseline in patients receiving fluoxetine (mean baseline score 26.7, SD 6.38, mean endpoint score 24, SD 7.54), and similarly in patients receiving clomipramine (mean baseline score 21.5, SD 8.62, mean endpoint score 15, SD 9.32).

6.2.8.1. Descriptive review

Five trials were included in the review that examined relapse prevention in patients with OCD (ANSSEAU, BAILER, HOLLANDER2003; KORAN2002; ROMANO2001). A further trial (RAVIZZA1996A) was conducted open-label and therefore excluded from the review.

In BAILER, following a 6-month acute phase RCT of paroxetine versus placebo (BURNHAM) and a 6-month open label phase with flexible dosing of paroxetine (N = 154), non-responders (N = 44) were randomised to a 6-month double-blind relapse prevention phase of fixed dose paroxetine versus placebo.

For the percentage of patients with partial relapses (defined as a Y-BOCS score increase from the acute phase baseline score and an increase of one or more points on the CGI Severity of Illness subscale from the last open label score), the treatment effect was not significant (p = 0.22), although the percentage of partial relapses in the placebo group (63.6%) was greater than the percentage of partial relapses in the paroxetine group (42.1%). There was a significant difference favouring paroxetine over placebo on the Y-BOCS (N = 20; SMD = −1.17; 95% CI, −2.15 to −0.19).

In HOLLANDER2003, treatment responders (N = 105) were randomised to 6-month double-blind, fixed dose, parallel paroxetine/placebo treatment (after completing both a 12-week RCT of paroxetine/placebo [total N = 348] and 6 months of open-label paroxetine treatment [total N = 263].

The results indicated that more of the patients randomised to receive placebo (N = 52) experienced a relapse than those who continued to receive paroxetine (N = 53) when the criteria included an increase of one point or more on the CGI Severity of Illness subscale (38% versus 60%; RR = 0.63; 95% CI, 0.42 to 0.96). With a stricter criterion of a return to baseline of patients' Y-BOCS scores, the effect was stronger, but the difference was not statistically significant (9% versus 23%; RR = 0.41; 95% CI, 0.15 to 1.08). The mean time it took for patients continuing with paroxetine to relapse was 62.9 days compared with 28.5 days for those who relapsed after switching to placebo. Statistically significant differences favouring paroxetine over placebo on mean Y-BOCS scores were found as early as week 2 (F = 5.25; df = 1.68; p = 0.02) and up until 5 months. However, no difference was found between the groups at the end of 6 months. Significantly more patients receiving placebo left the study early because of adverse events compared with those receiving paroxetine (38% versus 5%; N = 105; RR = 0.15; 95% CI, 0.05 to 0.47).

In ANSSEAU, following an acute phase RCT of paroxetine versus clomipramine versus placebo (ZOHAR1996a) and an 8-week maintenance phase, patients were re-randomised within group to the drug or placebo (paroxetine/paroxetine versus paroxetine/placebo, clomipramine/clomipramine versus clomipramine/placebo, and placebo).

There was no significant difference between paroxetine and placebo on the rate of partial relapse (defined as an increase on the Y-BOCS from the baseline score or an increase in CGI severity by one or more points from the last observation), paroxetine 1/10 (10%) versus placebo 2/8 (25%). Based on a 25% response criterion on the Y-BOCS, the rate of response was lower in paroxetine (18.2%) compared with placebo (33.3%), but this was not statistically significant (p = 0.45). There was also no significant difference between paroxetine and placebo on the likelihood of reporting adverse effects, 57% versus 50%, respectively.

In ROMANO2001, treatment responders (N = 71) were randomised to 52 weeks double-blind, fixed dose fluoxetine or placebo treatment (after completing a 20-week single-blind treatment of fluoxetine [total N = 130]). People who continued to receive fluoxetine had a lower estimated 1-year relapse rate compared with those randomised to placebo, though this difference was not statistically different (20.6% versus 31.9%, p = 0.14). Among patients who responded to acute phase treatment with fluoxetine 60 mg/day, a subset of those who continued to receive fluoxetine, the 1-year estimated rate of relapse was lower compared with patients randomised to placebo (17.5% versus 38%, p = 0.04). This estimate was not significantly different for people who responded to acute phase treatment with fluoxetine 20 or 40 mg/day and who continued to receive fluoxetine for a year compared with patients randomised to placebo (28.6% versus 21.3%, p = 0.79). Change on the Y-BOCS from randomisation to endpoint was not statistically different between fluoxetine and placebo (p = 0.54).

In KORAN2002, treatment responders (N = 223) were randomised to 28 weeks double-blind sertraline or placebo treatment (after completing a 52-week single-blind treatment of sertraline [total N initially enrolled = 649]). Patients assigned to sertraline continued on the same dose of sertraline as during the 52-week phase, while patients assigned to placebo were blindly discontinued from sertraline. Mean Y-BOCS scores increased in both groups (mean change at end-point, −1.3 and −3.1 in sertraline and placebo groups respectively). However, patients continuing with sertraline had a lower mean score at endpoint than those receiving placebo (N = 221; SMD = −0.37; 95% CI, −0.63 to −0.10). When a strict criteria for relapse (increase in Y-BOCS >=5, Y-BOCS score >=20 and CGI Global Improvement >=1 continuing for a month) was applied, there was no difference between sertraline and placebo (3.6% versus 5.2%; N = 223; RR = 0.7; 95% CI, 0.2 to 2.4). However, more patients receiving placebo experienced a relapse or insufficient clinical response compared with sertraline (RR = 0.39; 95% CI, 0.20 to 0.76) and more patients experienced an acute exacerbation of OCD (RR = 0.34; 95% CI, 0.19 to 0.60).

6.3.3.1. Clinical evidence statements

6.3.4.1. Clinical evidence statements

6.3.4.2. Results summary from additional cross-over trials

In KHANNA1988, patients received double-blind clomipramine or nortriptyline in a randomised fashion for 6 weeks and then crossed over to the other drug after a 4-week drug-free interval. Out of the 18 patients who entered the trial, eight completed the full cross-over sequence.

There was no significant difference between pre- and post-trial scores on the Leyton Obsessional Inventory and the Maudsley Obsessive-Compulsive Inventory. There was also no difference between clomipramine and nortriptyline on obsessive-compulsive symptoms.

In ZOHAR1987a, after a 2–4 week placebo-phase, patients were randomly assigned to clomipramine or desipramine. After a 4-week placebo interval, patients were crossed-over to the other drug. The duration of each active treatment was 6 weeks. Fourteen patients entered the trial, of which 10 patients completed the entire cross-over sequence.

The reduction in obsessive-compulsive symptoms as measured by the Comprehensive Psychopathological Rating Scale – Obsessive-Compulsive subscale (CPRS-OC) and the NIMH-OC scale was greater in clomipramine than desipramine (CPRS-OC-5: F1,8 = 8.03, p = 0.02; NIMH-OC: F1,8 = 7, p = 0.03). Using the CPRS-OC subscale, the mean (±SD) improvement during the 6 weeks for clomipramine was 28.4% ± 20.1% compared with 4.2% ± 11.4% for desipramine.

6.3.5.1. Clinical evidence statements

6.4.4.1. Clinical evidence statements

6.4.4.2. Results summary from additional cross-over trials

In KHANNA1988, patients received double-blind clomipramine or nortriptyline in a randomised fashion for 6 weeks and then crossed-over to the other drug after a 4-week drug-free interval. Out of the 18 patients who entered the trial, 8 completed the full cross-over sequence.

There was no significant difference between pre- and post-trial scores on the Leyton Obsessional Inventory and the Maudsley Obsessive-Compulsive Inventory. There was also no difference between clomipramine and nortriptyline on obsessive-compulsive symptoms.

In ZOHAR1987A, after a 2–4 week placebo-phase, patients were randomly assigned to clomipramine or desipramine. After a 4-week placebo interval, patients were crossed-over to the other drug. The duration of each active treatment was 6 weeks. Fourteen patients entered the trial, of which 10 patients completed the entire cross-over sequence.

The reduction in obsessive-compulsive symptoms as measured by the CPRS-OC subscale and the NIMH-OC scale was greater in clomipramine than desipramine (CPRS-OC-5: F1,8 = 8.03; p = 0.02; NIMH-OC: F1,8 = 7; p = 0.03). Using the CPRS-OC subscale, the mean (±SD) improvement during the 6 weeks for clomipramine was 28.4% ± 20.1% compared with 4.2% ± 11.4% for desipramine.

6.6.3.1. Clinical evidence statements

6.7.3.1. Clinical evidence statements

6.7.4.1. Clinical evidence statements

6.7.4.2. Results summary from additional cross-over trials

In INSEL1983B, 24 patients with OCD were randomised to placebo-clorgyline-placebo or placebo-clomipramine-placebo sequences and then crossed over to the other treatment over a 6-month period. The duration of each active treatment was 6 weeks. Ten patients completed the entire cross-over sequence.

Neither clorgyline nor clomipramine were effective in reducing symptom severity as measured by the Leyton Interference score, though clomipramine was effective in ameliorating scores on 13 of 19 measures while clorgyline showed no improvement on all 19 measures. When the 10 patients who completed the entire cross-over sequence were compared, improvement with clomipramine surpassed that seen with clorgyline on the CPRS-OC subscale, the Global Obsessive-Compulsive scale and the Obsessive-Compulsive rating scale. Only one patient displayed more improvement with clorgyline than with clomipramine on obsessive-compulsive symptoms. Side effects were prevalent on placebo, clorgyline and clomipramine.

6.8.4.1. Clinical evidence statements

6.8.5.1. Clinical evidence statements

6.8.5.2. Results summary from additional cross-over trial

In HEWLETT1992, 28 patients were initially randomly assigned to clomipramine, clonazepam, clonidine or diphenhydramine as control and then crossed-over to each of the other treatments. The study took place over 26 months, with 6 weeks in each treatment. Sixteen patients completed the entire cross-over sequence. Eight patients discontinued because of adverse side effects.

Clonazepam and clomipramine produced significantly lower Y-BOCS scores than did diphenhydramine and clonidine. In turn, Y-BOCS scores in diphenhydramine were significantly lower than baseline, whereas there was no difference between ratings for clonidine treatment and baseline. While the response rate for diphenhydramine and clonidine was 27% and 20% respectively, the response rates for clonazepam and clomipramine was higher at 48% and 54% respectively. Patients who responded to clonazepam tended to respond to clomipramine, the cross-response rate being 73% and patients who responded to clomipramine tended to respond to clonazepam, cross-response rate 80%.

6.9.3.1. Clinical evidence statements

6.10.4.1. Clinical evidence statements

6.10.5.1. Clinical evidence statements

6.11.3.1. Switching SRI

Given the limitations of data supporting alternative strategies, and the acceptability of switching from one SSRI to another, this remains the preferred option for many clinicians. Sometimes, however, it is appropriate to persist for longer with a particular SRI even in patients who show little sign of improvement, since a delayed response may occur after 6 months or more. A panel of international OCD experts made recommendations on switching (Expert Consensus Guideline, March et al., 1997) and they suggested changing the SRI after 8–12 weeks on the maximal dose if the clinical effect was incomplete. They proposed a 40% chance of responding to a second SRI, and a lesser response to a third and suggested switching to clomipramine after 2–3 failed trials on SSRIs. Since that time, a limited evidence-base supporting switching has accrued, but the data remains sparse and inconclusive. In a placebo-controlled sertraline trial, one third of patients benefited from a switch to a second SSRI (Koran et al., 2002). Denys and colleagues (2004) took 43 non-responders to 12 weeks treatment with either paroxetine or venlafaxine and switched them under double-blind conditions to the alternate treatment, whereupon 18 (43%) showed a clinical response. A single-blind study in 29 cases of SRI-resistant OCD showed encouraging results for venlafaxine (37.5–375 mg) as a monotherapy (Hollander et al., 2003b).

These results intimate that patients whose symptoms have failed to respond to an SSRI may benefit from switching agent, but fail to control for the effect of prolongation of treatment per se. RCTs comparing continuation of the original drug with switching are required to properly evaluate the role of switching in OCD.

6.11.3.2. Increasing dose

Results from uncontrolled case studies suggest that, for some patients, increasing SSRI doses above formulary limits may procure a better effect (Bejerot & Bodlund, 1998; Byerly et al., 1996). Although doses of clomipramine up to 300 mg have been systematically investigated and found to be acceptable, the risk of seizures and cardiotoxicity associated with this drug suggest that doses exceeding this should be generally avoided, and if attempted, ECG and clomipramine plasma-level monitoring should be considered.

6.11.3.3. Changing mode of drug-delivery

Changing mode of administration may be considered though this is not practical in many cases. Intravenous clomipramine has been shown to be more effective than placebo in a single double-blind study investigating refractory OCD, with 6 out of 29 patients randomised to clomipramine classed as responders after 14 daily infusions, compared with none receiving placebo infusions (Fallon et al., 1998). A positive open study showing benefit for 21 days intravenous citalopram has also been reported (Pallanti et al., 2002b) but requires confirmation using a controlled study design.

6.11.4.1. Clinical evidence statements

6.11.4.2. Combining two SRIs

If a patient has symptoms that fail to respond to successive SRI trials, augmented with CBT, the Expert Consensus Guideline (March et al., 1997) recommends adding another agent to the SRI. The evidence is acknowledged to be limited, based on the results of small RCTs and open case series.

Combining clomipramine with an SSRI has been proposed for adults or children unresponsive to, or intolerant of, SRI monotherapy. This strategy should be approached with caution since the pharmacokinetic interactions on the hepatic cytochrome P 450 isoenzymes may lead to a build-up of clomipramine that could be dangerous, and ECG and clomipramine plasma-level monitoring is advisable. Positive results have been reported from small, uncontrolled case series (Szegedi et al., 1996), although the fluoxetine–clomipramine combination resulted in ECG changes in some cases. Pallanti et al. (1999) compared 9 treatment-refractory patients given citalopram plus clomipramine with 7 given citalopram alone, in a randomised open-label trial. They reported a significantly larger improvement in Y-BOCS ratings for those given the combination, all of whom experienced decreases ≥35% from baseline. This combination is advantageous in not altering the metabolism of clomipramine, and was well tolerated. No controlled studies of the co-administration of different SSRIs have been published.

6.11.5.1. Studies considered

The review team conducted a new systematic search to identify studies that used augmentation strategies in the treatment of OCD. The search identified 38 studies that were of interest (Atmaca et al., 2002; Barr et al., 1997; Bejerot & Bodlund, 1998; Blier & Bergeron, 1996; Bogetto et al., 2000; Byerly et al., 1996; Crocq, 2002; D'Amico et al., 2003; Dannon et al., 2000; Denys et al., 2002a; Denys et al., 2004; Francobandiera, 2001; Fux et al., 1999; Grady et al., 1993; Hewlett et al., 1992; Hollander et al., 2003a; Koran et al., 2000; Koran et al., 2001; McDougle & Walsh, 2001; McDougle et al., 1990; McDougle et al., 1991; McDougle et al., 1993; McDougle et al., 1994; McDougle et al., 1995a; McDougle et al., 1995b; McDougle et al., 2000a; Maina et al., 2003; Metin et al., 2003; Mohr et al., 2002; Mundo et al., 1998; Noorbala et al., 1998; Pallanti et al., 1999; Pigott et al., 1991; Pigott et al., 1992a; Pigott et al., 1992b; Sevincok & Topuz, 2003; Shapira et al., 2004; Weiss et al., 1999).

6.11.5.2. Combining SRIs and drugs exerting antidepressant or anxiolytic properties

Controlled studies have found no evidence for the efficacy of lithium augmentation in OCD (McDougle et al., 1991; Pigott et al., 1991). Similarly, three double-blind placebo controlled studies have demonstrated that combining buspirone with an SRI is not helpful (Grady et al., 1993; McDougle et al., 1993; Pigott et al., 1992a). Clonazepam, a benzodiazepine with putative effects on serotonin neurotransmission, failed to impact on the core symptoms of OCD as a monotherapy in one study (Hollander et al., 2003c), but produced some evidence of anti-obsessional effect in another (Hewlett et al., 1992). Pigott and colleagues (1992b) reported limited efficacy for clonazepam given together with fluoxetine or clomipramine in a placebo-controlled study. Pindolol is a beta-blocker which also acts as an antagonist at presynaptic 5HT1A autoreceptors. Dannon and colleagues (2000) demonstrated efficacy for pindolol when combined with paroxetine in a double-blind placebo-controlled study of 14 treatment resistant cases, but another study combining it with fluvoxamine did not (Mundo et al., 1998). Blier and Bergeron (1996) found a beneficial effect for pindolol only when l-tryptophan was openly added to the combination.

The limitations of adding drugs acting on serotonin led investigators to re-examine the role of noradrenergic agents for OCD. While not evidently effective as a monotherapy, nortryptyline (50 mg) administered in combination with 150 mg clomipramine was found to be more effective than clomipramine plus placebo in a small 8-week study looking at 30 individuals with non-refractory OCD (Noorbala et al., 1998). However, Barr and colleagues (1997) investigated the addition of another noradrenergic antidepressant, desipramine, to 20 patients whose symptoms had failed to respond to SSRI monotherapy, in a double-blind placebo-controlled study, and found no added benefit.

6.11.5.3. Combining SRIs and drugs with antipsychotic properties

No positive studies of antipsychotic monotherapy in OCD meet today's standards, and OCD is not recognised as responding to these drugs individually. McDougle and colleagues (1990) reported a benefit from adding open-label pimozide (6.5 mg) in 17 patients unresponsive to fluvoxamine. Patients with comorbid chronic tics or schizotypal disorder were most responsive. A subsequent double-blind placebo-controlled study by the same author demonstrated a significant Y-BOCS improvement for low-dose haloperidol (6.2 mg) added to fluvoxamine. Eleven of 17 patients receiving the active drug achieved ‘responder’ status by as early as 4 weeks, compared with none on placebo. Again, a preferential response was seen in patients with comorbid tics (McDougle et al., 1994). Antipsychotics such as haloperidol and sulpiride are first-line treatments for Tourette's syndrome, so this finding supports a theoretical link between these disorders. This combination increases the side-effect burden, including extra-pyramidal effects. It is therefore recommended to start treatment with very low doses, and increase cautiously subject to tolerability (for example, 0.25–0.5 mg haloperidol, titrated slowly to 2–4 mg (McDougle & Walsh, 2001).

Newer second-generation antipsychotics that modulate serotonin and dopamine neurotransmission also offer promise and are associated with a lower risk of side effects, although they may cause dysregulation of endocrine and metabolic functions and weight gain. Positive reports from open case series were confirmed by McDougle and colleagues (2000a) in the first reported double-blind placebo-controlled study showing efficacy for risperidone augmentation in 36 patients unresponsive to 12 weeks on an SRI. Risperidone (2.2 mg) was superior to placebo in reducing Y-BOCS scores as well as anxiety and depression, was well-tolerated and there was no difference between those with and without comorbid tics or schizotypy. A smaller double-blind study by Hollander and colleagues (2003a) examined patients who had symptoms that were failing to respond to at least two trials of SRIs. Four of 16 patients randomised to risperidone (0.5–3 mg) turned out to be responders, defined as a CGI Global Improvement score of 1 or 2 and Y-BOCS decrease of >25% at the 8-week end-point, compared with none of the six patients randomised to placebo. However, the results, which were limited by the small sample size, did not reach statistical significance.

Quetiapine has also been the subject of recent controlled investigation. There have been contradictory results from open studies (Denys et al., 2002a; Mohr et al., 2002; Sevincok & Topuz, 2003). However, a placebo-controlled single-blind study looking at 27 cases of SRI-resistant OCD showed a significant advantage from adding quetiapine in doses up to 200 mg daily to ongoing SRI (Atmaca et al., 2002). Moreover, the recent double-blind placebo-controlled study by Denys and colleagues (2004) showed evidence of efficacy for 8 weeks quetiapine (<300 mg) augmentation in 20 patients whose symptoms had failed to respond to at least two SRIs, showing a mean decrease of 31% on baseline Y-BOCS, compared with 20 placebo-treated controls whose symptoms showed only 7% improvement (p < 0.001). Eight out of twenty (40%) patients responded to quetiapine therapy, compared with two out of 20 (10%) on placebo.

Encouraging results from a small number of open-label studies of olanzapine (Bogetto et al., 2000; Crocq, 2002; D'Amico et al., 2003; Francobandiera, 2001; Koran et al., 2000; Weiss et al., 1999) were not, however, supported by the double-blind placebo-controlled study by Shapira and colleagues (2004), which investigated 44 partial or non-responders to 8 weeks of fluoxetine. Both groups improved over the 6-week study period, with no additional advantage for the olanzapine-treated patients compared with extending the duration of fluoxetine monotherapy. Differences in entry criteria between studies may partly explain differences in the results.

There has been a positive open-label study of amisulpride (Metin et al., 2003).

Augmentation with clozapine has not been systematically investigated. The results for clozapine monotherapy in OCD have not been encouraging (McDougle et al., 1995a). Some authors have reported emergent obsessions during treatment with atypical antipsychotics, which may be related to their mixed receptor antagonist properties.

Altogether, these results favour the use of second generation antipsychotics such as risperidone and quetiapine as the first-line strategy for augmentation in resistant OCD. It remains uncertain as to how long patients need to remain on augmented treatment. A small retrospective study by Maina and colleagues (2003) showed that the vast majority of patients who had responded to the addition of an antipsychotic to their SRI, subsequently relapsed when the antipsychotic was withdrawn.

6.11.5.4. Other strategies for refractory OCD

Inositol (18 g/day) is an experimental compound that acts through intracellular messenger systems. It was thought to have mild anti-obsessional efficacy but results from a placebo-controlled augmentation study by Fux and colleagues (1999) refute this. Sumatriptan is a 5HT1D agonist used to treat migraine. A small open case series suggested improvement over 4 weeks of treatment but, in a double-blind placebo-controlled study of 10 patients, 5-day treatment was associated with a worsening of OCD (Koran et al., 2001).

6.12.3.1. Clinical evidence statements

6.12.4.1. Clinical evidence statements

6.12.4.2. Results summary from additional cross-over trials

In FLAMENT1985, following a 1-week evaluation period, patients were randomised to 5 weeks of clomipramine or placebo and then crossed-over to 5 weeks of the other treatment. Twenty-three patients entered the drug study and 19 completed the entire cross-over sequence.

Compared with placebo, clomipramine was significantly more effective in relieving obsessional symptoms, as measured on Leyton Obsessional Inventory – Child Version (LOI-CV) symptom (t = 2.19, p = 0.04), resistance (t = 2.12, p = 0.05) and interference scores (t = 2.24, p = v0.04), the OCR scale (t = 3.05, p = 0.007), and the NIMH-OC scale (t = 2.83, p = 0.02). Side-effects scores increased markedly during the first week with clomipramine and by week 5 they were significantly higher than with placebo (t = −2.52, p = 0.02). One patient experienced a grand mal seizure during clomipramine treatment.

In RAPOPORT1980, patients were randomised to 3–5 week consecutive periods of treatment with clomipramine, desipramine or placebo and then crossed-over to each of the other 2 treatments over 16 weeks. Nine patients entered the study and eight completed the entire cross-over sequence.

At the end of treatment, symptom severity as measured by the Leyton Interference and Resistance subscales was comparable across clomipramine, desipramine and placebo. Side effects did not differ between treatment periods.

6.12.5.1. Clinical evidence statements

6.12.5.2. Results summary from additional cross-over trials

In LEONARD1989a, following a 2-week single-blind placebo phase patients were randomised to 5-week consecutive periods of treatment with clomipramine or desipramine and then crossed-over to the other treatment. Forty-nine patients participated in the trial.

At the end of treatment, symptom severity was decreased to a greater extent by clomipramine than desipramine based on the NIMH-OC scale (F = 16.62; p = 0.002), Ward OCD Rating scale (F = 21.16; p = 0.00001) and CPRS-OC subscale (F = 10.34; p = 0.003). There was a significant effect of order of drug, whereby the rate of relapse was greater in the clomipramine-to-desipramine cross-over sequence than the desipramine-to-clomipramine cross-over sequence.

In RAPOPORT1980, patients were randomised to 3–5 week consecutive periods of treatment with clomipramine, desipramine or placebo and then crossed-over to each of the other 2 treatments over 16 weeks. Nine patients entered the study and eight completed the entire cross-over sequence.

At the end of treatment, symptom severity as measured by the Leyton Interference and Resistance subscales was comparable across clomipramine, desipramine and placebo. Side effects did not differ between treatment periods.

6.12.6.1. Descriptive review

Two studies examined the continuation of SSRIs in children, CARPENTER and LIEBOWITZ2002.

CARPENTER was an unpublished 32-week two-phase trial of paroxetine. In the first phase, patients received open-label paroxetine for 16 weeks. In the second phase, patients who responded to the first phase were either continued with paroxetine at the final daily dose achieved during the first phase or discontinued from paroxetine onto placebo in a double-blind randomised fashion. The duration of the second phase was 16 weeks. Three hundred and thirty-five patients participated in the phase I trial, of which 194 entered the second phase.

The results indicated that the rate of relapse, defined as (a) an increase in CGI Global Improvement score by 1 point for 2 consecutive visits, (b) an increase in CGI Global Improvement score by 2 or more points at any single visit, and (c) a CGI Global Improvement score of 5 or more points at any time, was not significantly different between patients continuing with paroxetine (34.7%) and patients discontinuing from paroxetine (43.9%). Comparing the response rates between the two groups based on a 25% or greater reduction of CY-BOCS from baseline favoured paroxetine continuation over discontinuation (25% versus 13.3%; RR = 0.86; 95% CI, 0.75 to 0.99). Symptom severity increased in both groups during the continuation phase as measured by the CY-BOCS, though this increase was lower in the paroxetine continuation group than the discontinuation group (SMD = −3.3; 95% CI, −5.77 to −0.83).

In LIEBOWITZ2002, patients who responded to an 8-week acute phase of fluoxetine or placebo were entered into an 8-week maintenance phase. Response was defined as ‘much improved’ or ‘very much improved’ on the CGI Global Improvement scale. Patients continued to receive the same drug as during the acute phase and were maintained on the final dose achieved during the acute phase.

Of the 43 patients who entered the acute phase, 18 entered the maintenance phase. At the end of the 8-week maintenance phase, fluoxetine patients had lower CY-BOCS scores than did placebo patients, mean (SD) was 5.64 (5.41) and 11.43 (11.12) respectively, F (1,14) = 9.22; p = 0.009. None of the fluoxetine patients relapsed, while one placebo patient relapsed. There were significantly more fluoxetine than placebo responders (57% versus 27%; χ² = 3.94; p = 0.05). Twenty-seven percent of the fluoxetine patients reported at least one adverse event, compared with none among the placebo patients.

6.13.6.1. Antidepressants

SRIs other than fluoxetine or clomipramine may be of benefit both theoretically and from the evidence of four case series. There are similar modest benefits from four open label trials. There are two open label case series of fluvoxamine (Perugi et al., 1996; Phillips et al., 1998), one of citalopram (Phillips & Najjar, 2003) and one case series of clomipramine (Hollander et al., 1989).

Phillips and colleagues (1998) entered 30 participants with BDD who received fluvoxamine over 16 weeks. The average dose was 238.8 mg and the range was 50 to 300 mg. The Y-BOCS modified for BDD decreased from 31.1 (SD 5.4) at baseline to 16.9 (SD 11.8) at 16 weeks. This represents a 45% reduction on the main outcome measure. Sixty-three per cent of participants responded based on a criterion of a 30% decrease or more on the Y-BOCS modified for BDD. Fluvoxamine was as effective in participants with an additional diagnosis of delusional disorder as without.

Perugi and colleagues (1996) entered 15 participants with BDD in an open label trial. The duration was for 10 weeks. The average dose was 208 mg and the range of dose was 100 to 300 mg. They did not use the modified Y-BOCS as an outcome measure but reported a 60% reduction over 12 weeks on symptom scores and 10 out of the 15 participants responding on the CGI.

Phillips and Najjar (2003) entered 15 participants with BDD (including those with beliefs of delusional intensity) in an open label study of citalopram over 12 weeks. The average dose was 51.3 mg and the range was 10 to 60 mg. The Y-BOCS modified for BDD decreased from 30.7 (S.D 4.9) at baseline to 15.3 (SD 10.6) at 12 weeks. This represents a 50% reduction on the main outcome measure. 73.3% of participants responded defined as a 30% decrease or more on the Y-BOCS modified for BDD. Citalopram was as effective in participants with an additional diagnosis of delusional disorder as without.

Hollander and colleagues (1989) reported on a case series of five participants with BDD who all responded to either clomipramine or fluoxetine. Four of the five patients' symptoms had failed to respond previously to drugs that had some serotonergic action including tricyclics, trazodone and lithium.

Phillips (1996b) conducted a retrospective case review of 130 patients who had 316 treatment trials of which 42% of 65 SRI trials had led to an improvement on the CGI, compared with 30% of 23 trials with a monoamine-oxidase inhibitor and 15% of 48 trials with a non-SRI tricyclic drug.

No continuation, maintenance or discontinuation studies of an SRI have been reported. Expert opinion and clinical experience suggest that, like OCD, there may be small further gains with an SRI after 12–16 weeks treatment. Furthermore, like OCD, there is a high rate of relapse on discontinuation on a SRI (Phillips et al., 2001).

Lastly there are case reports on the benefit of sertraline (El-Khatib & Dickey, 1995) and two cases with intravenous clomipramine (Pallanti & Koran, 1996). There are two case reports of the benefit of an MAOI, tranlcypromine (Jenike, 1984) and one with a combination of phenelzine, trimipramine and perphenazine (Phillips, 1991).

6.13.6.2. Antipsychotic drugs as a monotherapy

Antipsychotic drugs were prescribed for BDD following a case series describing the benefit of pimozide in individuals with delusional disorder (Riding & Munro, 1975). However the case series included cases of delusions of infestation, delusions of body odour and dysmorphic delusions.

Phillips and colleagues (1996b), in a retrospective survey of medication trials, reported that only 3% of 83 trials of an antipsychotic were of any benefit in BDD. Grant (2001) has described one case report of olanzapine for BDD without delusional disorder. The individual however fulfilled diagnostic criteria for BDD, alcohol dependence and bipolar II disorder. At the end of 3 weeks, she reported no preoccupation with her appearance and no longer met criteria for BDD.

6.13.6.3. SSRI switching or augmentation studies

In individuals with BDD whose symptoms have failed to respond to an SSRI, or who have a partial response to an SSRI, then switching to another SSRI or augmentation with another drug has been tried.

Phillips and colleagues (2001) reported that in those subjects whose symptoms failed to respond to an adequate SRI trial, 42.9% (n = 6) responded to at least one subsequent SRI trial and 43.5% (n = 10) of subsequent SRI trials received by these subjects were effective.

Phillips and colleagues (2001) reported on an open label series of buspirone (extended from Phillips, 1996a) in a chart review of patients whose symptoms failed to respond to an SRI alone or have had only a partial response. In 12 participants, buspirone was added after an adequate does of an SRI; 33.3% of trials were successful. The mean dose was 56.5 mg (range 30–80 mg daily) and was as effective in delusional as non-delusional cases.

6.13.6.4. Young people with BDD

Albertini and Phillips (1999) reported on 33 children and young people with BDD of whom 10 out 19 (53%) treated with an SRI improved on the CGI. No non-SRI medications in 8 trials were effective.

There is one case report on clomipramine in an adolescent with BDD with delusional disorder (Sondheimer, 1988) and one case report of doxepine and behaviour therapy in an adolescent with BDD (Sobanski & Schmidt, 2000).

Children and young people

6.14.1.1.

If psychological treatment is declined by children or young people with OCD or BDD and their families or carers, or they are unable to engage in treatment, an SSRI may be considered with specific arrangements for careful monitoring for adverse events. [B]

Starting the treatment

6.14.2.1.

Common concerns about taking medication for OCD or BDD should be addressed. Patients should be advised, both verbally and with written material, that:

• craving and tolerance do not occur [C]
• there is a risk of discontinuation/withdrawal symptoms on stopping, missing doses, or reducing the dose [C]
• there is a range of potential side effects, including worsening anxiety, suicidal thoughts and self-harm, which need to be carefully monitored, especially in the first few weeks of treatment [C]
• there is commonly a delay in the onset of effect of up to 12 weeks, although depressive symptoms improve more quickly [C]
• taking medication should not be seen as a weakness. [GPP]

Monitoring risk

6.14.2.2.

Adults with OCD or BDD started on SSRIs who are not considered to be at increased risk of suicide or self-harm should be monitored closely and seen on an appropriate and regular basis. The arrangements for monitoring should be agreed by the patient and the healthcare professional, and recorded in the notes. [GPP]

6.14.2.3.

Because of the potential increased risk of suicidal thoughts and self-harm associated with the early stages of SSRI treatment, younger adults (younger than age 30 years) with OCD or BDD, or people with OCD or BDD with comorbid depression, or who are considered to be at an increased risk of suicide, should be carefully and frequently monitored by healthcare professionals. Where appropriate, other carers – as agreed by the patient and the healthcare professional – may also contribute to the monitoring until the risk is no longer considered significant. The arrangements for monitoring should be agreed by the patient and the healthcare professional, and recorded in the notes. [C]

6.14.2.4.

For adults with OCD or BDD at a high risk of suicide, a limited quantity of medication should be prescribed. [C]

6.14.2.5.

When adults with OCD or BDD, especially with comorbid depression, are assessed to be at a high risk of suicide, the use of additional support such as more frequent direct contacts with primary care staff or telephone contacts should be considered, particularly during the first weeks of treatment. [C]

6.14.2.6.

For adults with OCD or BDD, particularly in the initial stages of SSRI treatment, healthcare professionals should actively seek out signs of akathisia or restlessness, suicidal ideation and increased anxiety and agitation. They should also advise patients to seek help promptly if symptoms are at all distressing. [C]

6.14.2.7.

Adults with OCD or BDD should be monitored around the time of dose changes for any new symptoms or worsening of their condition. [C]

Selective serotonin reuptake inhibitors (SSRIs)

6.14.2.8.

For adults with OCD, the initial pharmacological treatment should be one of the following SSRIs: fluoxetine, fluvoxamine, paroxetine, sertraline, or citalopram. [A]

6.14.2.9.

For adults with BDD (including those with beliefs of delusional intensity), the initial pharmacological treatment should be fluoxetine because there is more evidence for its effectiveness in BDD than there is for other SSRIs. [B]

6.14.2.10.

In the event that an adult with OCD or BDD develops marked and/or prolonged akathisia, restlessness or agitation while taking an SSRI, the use of the drug should be reviewed. If the patient prefers, the drug should be changed to a different SSRI. [C]

6.14.2.11.

Healthcare professionals should be aware of the increased risk of drug interactions when prescribing an SSRI to adults with OCD or BDD who are taking other medications. [GPP]

6.14.2.12.

For adults with OCD or BDD, if there has been no response to a full course of treatment with an SSRI, healthcare professionals should check that the patient has taken the drug regularly and in the prescribed dose and that there is no interference from alcohol or substance use. [GPP]

6.14.2.13.

For adults with OCD or BDD, if there has not been an adequate response to a standard dose of an SSRI, and there are no significant side effects after 4–6 weeks, a gradual increase in dose should be considered in line with the schedule suggested by the Summary of Product Characteristics. [C]

6.14.2.14.

For people with OCD or BDD, the rate at which the dose of an SSRI should be increased should take into account therapeutic response, adverse effects and patient preference. Patients should be warned about, and monitored for, the emergence of side effects during dose increases. [GPP]

6.14.2.15.

If treatment for OCD or BDD with an SSRI is effective, it should be continued for at least 12 months to prevent relapse and allow for further improvements. [C]

6.14.2.16.

When an adult with OCD or BDD has taken an SSRI for 12 months after remission (symptoms are not clinically significant and the person is fully functioning for at least 12 weeks), healthcare professionals should review with the patient the need for continued treatment. This review should consider the severity and duration of the initial illness, the number of previous episodes, the presence of residual symptoms, and concurrent psychosocial difficulties. [GPP]

6.14.2.17.

If treatment for OCD or BDD with an SSRI is continued for an extended period beyond 12 months after remission (symptoms are not clinically significant and the person is fully functioning for at least 12 weeks), the need for continuation should be reviewed at regular intervals, agreed between the patient and the prescriber, and written in the notes. [GPP]

6.14.2.18.

For adults with OCD or BDD, to minimise discontinuation/withdrawal symptoms when reducing or stopping SSRIs, the dose should be tapered gradually over several weeks according to the person's need. The rate of reduction should take into account the starting dose, the drug half-life and particular profiles of adverse effects. [C]

6.14.2.19.

Healthcare professionals should encourage adults with OCD or BDD who are discontinuing SSRI treatment to seek advice if they experience significant discontinuation/withdrawal symptoms. [C]

Other drugs

With the exception of clomipramine, other antidepressants should not normally be used in the treatment of OCD or BDD. Most other drugs have limited or no use in this context.

6.14.2.20.

The following drugs should not normally be used to treat OCD or BDD without comorbidity:

• Tricyclic antidepressants other than clomipramine
• Tricyclic-related antidepressants
• Serotonin and noradrenaline reuptake inhibitors (SNRIs), including venlafaxine
• Monoamine-oxidase inhibitors (MAOIs)
• Anxiolytics (except cautiously for short periods to counter the early activation of SSRIs). [C]

6.14.2.21.

Antipsychotics as a monotherapy should not normally be used for treating OCD. [C]

6.14.2.22.

Antipsychotics as a monotherapy should not normally be used for treating BDD (including beliefs of delusional intensity). [C]

How to use SSRIs in children and young people

6.14.4.1.

When an SSRI is prescribed to children and young people with OCD or BDD, it should be in combination with concurrent CBT (including ERP). If children and young people are unable to engage with concurrent CBT, specific arrangements should be made for careful monitoring of adverse events and these arrangements should be recorded in the notes. [C]

6.14.4.2.

Children and young people with OCD or BDD starting treatment with SSRIs should be carefully and frequently monitored and seen at an appropriate and regular basis. This should be agreed by the patient, his or her family or carers and the healthcare professional, and recorded in the notes. [GPP]

6.14.4.3.

An SSRI should only be prescribed to children and young people with OCD or BDD following assessment and diagnosis by a child and adolescent psychiatrist who should also be involved in decisions about dose changes and discontinuation. [GPP]

6.14.4.4.

A licensed medication (sertraline or fluvoxamine) should be used when an SSRI is prescribed to children and young people with OCD, except in patients with significant comorbid depression when fluoxetine should be used, because of current regulatory requirements. [A]

6.14.4.5.

Fluoxetine should be used when an SSRI is prescribed to children or young people with BDD. [C]

6.14.4.6.

For children and young people with OCD or BDD who also have significant depression, the NICE recommendations for the treatment of childhood depression¹² should be followed and there should be specific monitoring for suicidal thoughts or behaviours. [GPP]

6.14.4.7.

Children and young people with OCD or BDD starting treatment with SSRIs should be informed about the rationale for the drug treatment, the delay in onset of therapeutic response (up to 12 weeks), the time course of treatment, the possible side effects and the need to take the medication as prescribed. Discussion of these issues should be supplemented by written information appropriate to the needs of the child or young person and their family or carers. [GPP]

6.14.4.8.

The starting dose of medication for children and young people with OCD or BDD should be low, especially in younger children. A half or quarter of the normal starting dose may be considered for the first week. [C]

6.14.4.9.

If a lower dose of medication for children and young people with OCD or BDD is ineffective, the dose should be increased until a therapeutic response is obtained, with careful and close monitoring for adverse events. The rate of increase should be gradual and should take into account the delay in therapeutic response (up to 12 weeks) and the age of the patient. Maximum recommended doses for children and young people should not be exceeded. [C]

6.14.4.10.

Children and young people prescribed an SSRI and their families or carers, should be informed by the prescribing doctor about the possible appearance of suicidal behaviour, self-harm or hostility, particularly at the beginning of treatment. They should be advised that if there is any sign of new symptoms of these kinds, they should make urgent contact with their medical practitioner. [GPP]

6.14.4.11.

Where children or young people with OCD or BDD respond to treatment with an SSRI, medication should be continued for at least 6 months post-remission (i.e. symptoms are not clinically significant and the child or young person is fully functioning for at least 12 weeks). [C]

How to use clomipramine in young people

6.14.4.12.

Children and young people with OCD or BDD and their families or carers should be advised about the possible side effects of clomipramine, including toxicity in overdose. [C]

6.14.4.13.

Before starting treatment with clomipramine in children and young people with OCD or BDD, an ECG should be carried out to exclude cardiac conduction abnormalities. [C]

6.14.4.14.

For a child or young person with OCD or BDD, if there has not been an adequate response to the standard dose of clomipramine, and there are no significant side effects, a gradual increase in dose may be cautiously considered. [C]

6.14.4.15.

Treatment of a child or young person with OCD or BDD with clomipramine should be continued for at least 6 months if the treatment appears to be effective, because there may be further improvement in symptoms. [B]

Stopping or reducing SSRIs and clomipramine in children and young people

6.14.4.16.

In children and young people with OCD or BDD, an attempt should be made to withdraw medication if remission has been achieved (i.e. symptoms are no longer clinically significant and the child or young person is fully functioning) and maintained for at least 6 months, and if this is their wish. Patients and their family or carers should be warned that relapse and/or discontinuation/withdrawal symptoms may occur. They should be advised to contact their medical practitioner should symptoms of discontinuation/withdrawal arise. [C]

6.14.4.17.

For children and young people with OCD or BDD, to minimise discontinuation/withdrawal reactions on reducing or stopping antidepressants, particularly SSRIs, the dose should be tapered gradually over several weeks according to the individual's need. The rate of reduction should take into account the starting dose, the drug half-life and particular profiles of adverse effects. [C]

6.14.4.18.

Children and young people with OCD or BDD should continue with psychological treatment throughout the period of drug discontinuation because this may reduce the risk of relapse. [C]

Other drugs

6.14.4.19.

Tricyclic antidepressants other than clomipramine should not be used to treat OCD or BDD in children and young people. [C]

6.14.4.20.

Other antidepressants (MAOIs, SNRIs) should not be used to treat OCD or BDD in children and young people. [C]

6.14.4.21.

Antipsychotics should not be used alone in the routine treatment of OCD or BDD in children or young people, but may be considered as an augmentation strategy. [C]