Serotonin re‐uptake Inhibitors (SSRIs) versus placebo for obsessive compulsive disorders (OCD)
Abstract
This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:
1) To identify and systematically review evidence of effectiveness of SSRIs in obsessive compulsive disorder in patients of all age groups in comparison to placebo.
2) To estimate summary effect sizes (pooled odds ratios and weighted standardised mean differences as appropriate) of the treatments.
3) To carry out an assessment of relationship between effect sizes and methodological features of the studies and demographic and clinical features of patients and treatment characteristics.
Background
One of the most disabling of anxiety disorders is obsessive‐compulsive disorder (OCD). This disorder is suffered by approximately 2% of the adult population (Weissman 1994) and 0.5 ‐ 4.0% of adolescents (Flament 1989; Feehan 1994). Approximately 0.6% of the general population have severe OCD (NAMH 1993). As the onset of the disorder is often in adolescence or young adulthood, and most frequently follows an episodic or chronic course, it may impact on the individual for many years.
The disorder is characterised by obsessions and compulsions. Obsessions are recurrent intrusive, unwanted, nonsensical thoughts that the sufferer cannot dispel. Common themes of the obsessional thoughts include thoughts that the person may cause harm to others or that harm may befall others or thoughts that the person or others are contaminated. Other common themes of obsessional thoughts are centred on the need for order, symmetry or perfection. The obsessional thoughts are associated with negative affect, usually anxiety, but other emotions such as disgust, guilt or shame may also be experienced. As a response to the feelings generated by the obsessional thoughts the person may perform compulsions, and performance of the compulsions temporarily decreases the negative affect. The compulsions are stereotypic, ritualised behaviours that are usually observable but which may include covert mental rituals. Common overt rituals are repetitive checking, washing or cleaning, or repetitive rearranging and ordering of objects. Examples of covert rituals include repetitive counting, praying or thinking magical statements. The obsessions and compulsions may occupy hours of the person's day and cause severe disruption to the person's life (Rasmussen 1989; Rapoport 1990), as well as to that of their families (Chakrabarti 1992).
Over the last thirty years, a number of treatment strategies have been developed for most anxiety disorders, including OCD, and tested in randomised controlled trials (RCTs) (Noyes 1990). For OCD, the most effective psychological treatment strategy is behavioural therapy, which consists of exposure‐in‐vivo coupled with response prevention. Unfortunately about 25% of patients offered this form of treatment will refuse it, and of those who accept, 10‐20% make minimal gains.
Pharmacological interventions have also been shown to be effective. In three‐quarters of patients, clomipramine and other serotonin‐uptake blockers (fluoxetine, fluvoxamine, sertraline and paroxetine) reduce symptom levels by about 30‐40%. However, the medications have side‐effects and there are high relapse rates when they are discontinued (Pato 1990).
We have identified the following relevant systematic reviews of treatments for OCD (QAP 1985; Christensen 1987; Jenike 1990a; Jenike 1990b; Cox 1993; van Balkom 1994; Greist 1995; Piccinelli 1995; Stein 1995; Abramowitz 1997; Kobak 1998). When these reviews are appraised, using the guidelines for quality assessment proposed by Sacks 1987, they were all found to have significant deficiencies. Of particular concern are the poor descriptions of how individual trials were identified, included or excluded; absence of investigation of sources of heterogeneity; the lack of measures of methodological quality; absence of weighting for sample size; and lack of description of what and why a particular statistical model was used for pooling of the effect sizes.
In this proposed review, we plan to carry out a systematic review, of all the RCTs of SSRIs versus placebo in OCD, following the guidelines of the Cochrane Collaboration (Mulrow 1997).
Objectives
1) To identify and systematically review evidence of effectiveness of SSRIs in obsessive compulsive disorder in patients of all age groups in comparison to placebo.
2) To estimate summary effect sizes (pooled odds ratios and weighted standardised mean differences as appropriate) of the treatments.
3) To carry out an assessment of relationship between effect sizes and methodological features of the studies and demographic and clinical features of patients and treatment characteristics.
Methods
Criteria for considering studies for this review
Types of studies
All trials described as randomised controlled trials (RCTs) of SSRIs (excluding Clomipramine) versus placebo in obsessive compulsive disorder. Cross‐over trials will be included depending upon availability of the relevant data up to the point of the cross‐over. RCTs in which SSRIs are used in combination with some other treatment, and which do not allow direct comparison of SSRIs with placebo (i.e. confounded RCTs) will not be included. Both published and unpublished studies will be included as there is evidence of publication bias in favour of studies with positive results (Easterbrook 1991; Dickersin 1992; Scherer 1994). The studies will not be excluded on the basis of sample size or duration of follow up, but these characteristics will be used as independent variables within sensitivity analysis to see what effect they have on dependent variable effect size.
If a study meets the inclusion criteria but does not present the data necessary for estimating effect size and such data are not available from the authors, then that study will excluded from the analysis, but will be commented upon critically and will be listed as eligible but dropout from meta‐analysis (Petitti 1994).
Types of participants
Participants of all the age groups (including children and adolescents) and both sexes with obsessive compulsive disorder according to the criteria of ICD9 or 10 or DSMIII, III‐R or IV or some other accepted /standardised criteria will be included. Studies with other DSM Axis I or similar co‐morbid disorders (but not other DSM disordrs or equivalant) will be excluded.
Types of interventions
SSRIs at any dose and regimen versus placebo. The SSRIs include fluoxetine, fluvoxamine, setraline, paroxetine and citalopram.
Types of outcome measures
Outcomes will be considered in terms of change differences, however where these are not available then end point differences will be considered.
Outcomes of primary interest:
1) Symptomatic improvement, using standardised / objective measures, (the usual instruments used for obsessive compulsive symptom severity are the Maudsley Obsessive‐Compulsive Checklist (Hodgson 1977) and the Yale‐Brown Obsessive‐Compulsive Scale (Goodman 1989) and also the Clinical Global Impression of Severity; for depressive symptoms are the Hamilton Depression Scale (Hamilton 1967; Hamilton 1969) and the Beck Depression Inventory (Beck 1961); and for anxiety symptoms are the Hamilton Anxiety Scale (Hamilton 1959), the Strait‐Trait Anxiety Inventory (Spielberger 1983) and the Beck Anxiety Inventory (Beck 1988)).
2) Acceptability of treatment measured as a) number of post‐randomisation exclusions and number of subsequent dropouts, and and patient satistifation with the treatment; and b) number of patients with adverse events and deaths including suicides.
Outcomes of secondary interest:
3) Social and occupational function improvement, using standardised / objective measures.
4) Quality of life measures.
5) Relapses if possible
6) Cost‐effectiveness if possible
Duration of outcomes will also be considered.
Search methods for identification of studies
A definite and explicit search strategy will be used following the guidelines of the Cochrane Collaboration (Depression Anxiety Neurosis (DAN) group module strategy) for various databases. The search strategy will be presented explicity.
The following sources will be searched.
1) Computer Assisted Searches (using terms 'obsessi*' , 'compulsi*' and and the relevant search terms for SSRIs)
The register of trials maintained by Depression, Anxiety and Neurosis Group of Cochrane Collaboration
Cochrane Controlled Trials Register,
MEDLINE database,
EMBASE database,
Science Citation Index / Scisearch
Psyclit
LILACS
2) The Obsessive Compulsive Disorder Database maintained by the Dean Foundation for Health, Research and Education, Madison, Wisconsin, USA.
3) Dissertation abstracts and conference abstracts
4) Handsearches of the Specialist Journals and other relevant journals (this data will be obtained from Cochrane Collaboration where available). Also handsearches of the references cited in the studies obtained by the other search strategies.
5) Personal Communication; a) The active researchers in the area of OCD from last 5 years will be contacted by letter for information on the relevant clinical trials (the information about the appropriate researchers will be identified from the studies obtained from the literature search). b) Major Pharmaceutical Companies will be contacted by letter for information on the relevant clinical trials. c) Information for more non‐English language literature will be obtained through contacts with appropriate members of Depression, Anxiety and Neurosis Group of Chocrane Collaboration; and this will be in addition to non‐English language literature obtained through the electronic searches.
Data collection and analysis
Selection of studies:
The titles and abstracts of studies obtained by the search will be scrutinised by the two reviewers to see whether they are potentially / possibly eligible for selection and thus for full scrutiny (by reading full articles).
Eligibility for selection after scrutiny of the full articles (and the additional information obtained from the authors where relevant) will be decided on the basis of the inclusion / selection criteria:
1) that they are RCTs
2) that they are comparisons of SSRIs and placebo
3) that the population studied is patients with OCD
Data collection:
The specially designed pilot tested forms / coding sheet will be used by the reviewers to collect data on methods, participants, intervention and outcome measurements and other relevant results of the studies. Any disagreements will be resolved through discussion and obtaining more information where relevant. Missing data will be obtained from the investigators where possible using a form designed for the purpose to minimise possible bias.
When information about the missing patients / lost to follow up patients is not available from the articles or the authors, it will be assumed that they dropped out because of adverse effects and / or treatment failure.
The choice of summary statistic:
For categorical data, Peto odds ratio will be calculated and for continuous data standardised mean difference (SMD) will be calculated; and these will be pooled as summary effect sizes, i.e. pooled Peto odds ratio and weighted standardised mean difference respectively (Petitti 1994). 95% confidence intervals will also be calculated.
Assessment of publication bias:
Relevant data from all the included trials will be entered into funnel plot (trial effect size plotted against trial precision / sample size) to detect any publication bias (Eggar 1997a).
Assessment of heterogeneity:
Heterogeneity will be assessed from clinical perspective (and if the studies are found to be clinically too heterogenous then quanititative sysnthesis of results will not be carried out). It will also be assessed quantitively / statistically by graphical presentation and tests of heterogeneity. Tests of heterogeneity will be estimated for odds ratios and SMD of the studies. If there is evidence of statistical heterogeneity, further analysis (using subgroup analysis and random effects meta‐regression) will be carried out to identify the sources in terms of patient, treatment or study design characteristics (Thompson 1994).
The choice of statistical method for meta‐analysis:
To estimate the summary effect sizes both fixed effects model and random effects model will be used with both odds ratios and SMD. In absence of heterogeneity random effects model will serve to check robustness of fixed effects model; and in presence of heterogeneity it will serve as the method choice for pooling the effect sizes ( as in this latter situation fixed effect method will not be appropriate for summarising the studies) (Petitti 1994; Eggar 1997b).
Cumulative meta‐analysis:
This will be carried out to investigate at what stage good evidence for efficacy of SSRI in OCD was availabe.
Investigation of relation between effect size and study methodolgical, patient and treatment characteristics:
Sensitivity analysis‐
1) date of publication (beacause the patients used in later trials may have been relatively more treatment resistent)
2) sample size (to investigate at what sample size important results start to appear)
3) duration of follow up (at what stage various outcomes start to appear)
4) study quality particularly the three main quality characteristics; a) concealment of allocation, b) blinding and c) assumption about dropouts ( to investigate their effect on rebustness of results). However sensitivity analysis for dropouts would only be possible for dichotomous data and not for interval data. The information about reasons for dropout will be collected and commented upon.
Subgroup analyis / meta‐regression / exploration of heterogeneity‐
1) age of the participants (children and adolescent will be analysed as a separate subgroup)
2) severity of OCD
3) obsessions and compulsions
4) presence of significant depression and anxiety
5) different drugs
6) different dosage levels
These will be considered to investigate whether age, severity of illness, obsessions and compulsions separately, presence of depression or anxiety, and different drugs and dosage levels have different effects.
Exploration of heterogeneity will also include methodological features of studies as listed above under sensitivity analysis.
Presentation of tables and figures of results:
The following tables will be presented‐
1) data / characteristics of the studies used
2) list of comparisons made
3) analyses of data for each comparison made
4) list of excluded trials with reasons for exclusions
The following figures will be presented‐
1) individual study and pooled odds ratios and with respective CI
2) for continuous / interval outcome measures, differences between individual and summary effect sizes between treated and control groups will be plotted; and in case of studies where different scales are used then they will be plotted in standard deviations and with the length of plot bar showing CI.
