In order to gain a license for treatment in a specific mental disorder a drug must demonstrate its superiority against placebo in the form of a randomised control trial. Comparisons versus other drugs allow establishment of treatment hierarchy but also require larger clinical trials to detect differences between agents.
Clinical trials are complex affairs, commonly bedevilled by a range of problems; not least the loss of participants to follow-up. When participants do not complete the trial a range of measures have been proposed to compensate for their being missing in the final analysis.
For example, the ‘Last outcome carried forward’ approach uses the last recorded data for the missing participants within the final analysis. The impact of missing data on analysis is clearly hard to predict and problematic, particularly in trials where the number of missing participants can approach half of those originally taking part, meaning that more than half of the final data gathered may be based on estimates from the data.
In order to explore the potential impact of missing data on clinical trials, researchers from around the United Kingdom have reported the findings from a meta-analysis into the efficacy of Quetiapine in the treatment of schizophrenia. Quetiapine is available in two formulations: Immediate (IR) and sustained release. For this project the Quetiapine IR formulation was considered.
The researchers defined two aims for their review:
- To assess the efficacy and adverse effects of Quetiapine IR in the treatment of schizophrenia in comparison with placebo
- To explore the potential impact of missing data on research findings
The team searched the published academic literature in order to identify studies in which participants were randomised in a blind fashion to receive either Quetiapine IR or placebo.
Data was extracted from the identified studies to address defined primary and secondary outcome measures:
- Primary measures:
- Average reduction of severity rating scale (PANSS) at the end of the trial
- Number of people achieving a clinically important response to medication (more than 50% reduction in PANSS)
- The authors assessed for clinically significant findings by considering a threshold reduction in PANSS necessary to produce a change perceived as beneficial by patients or providers. In this case, the minimal necessary change was identified as 11 points on the PANSS
- Secondary measures:
- Evidence of relapse
- Positive symptoms
- Negative symptoms
- Quality of life
- Need for additional antipsychotic medication
- The number of participants leaving studies was also identified
- Adverse effects were noted
- Movement disorders (noted through prescription of anti-Parkinsonian medication)
- Withdrawal from the trial due to sedation
- Number of drug attributable adverse events
- Weight gain
- Weight loss
- Results from identified trials were combined through meta-analysis
- The effects of missing data were assessed against the hypothesis that: trials with more than half of participants lost to follow up would have smaller drug-placebo differences than other trials (i.e. loss to follow up of participants masks drug efficacy)
- 15 trials were included in the analysis
- 11 assessed only short-term efficacy
- Trials were assessed for a risk of bias; the authors identified the following domains as being of concern:
- Selective reporting of secondary outcome measures and adverse effects
- Risk of attrition bias from loss to follow up
- Risk of unblinding due to sedation side-effects
Primary outcome measures
- Quetiapine outperformed placebo over the range of 2-12 weeks assessed
- The effect size was small and confidence intervals did not include the level identified as being the minimum for clinical significance
- Longer trial duration was associated with a more positive outcome
- The authors found no evidence that loss of participants to follow-up may have masked treatment efficacy
- Quetiapine was associated with a slight reduction in need for hospital care over 2-6 weeks
- Small effects were observed for change in positive and negative symptom subclass
- A slight reduction in need for additional sedative medication was noted
- No change in quality of life was observed
- Quetiapine was associated with a slight increased risk of minor, but not major, adverse events
- Quetiapine was associated with increased weight gain
The authors concluded:
Quetiapine IR has a small beneficial effect on overall psychotic symptoms over 2-12 weeks, but also leads to weight gain and sedation.
- In this review, the authors have sought to re-appraise the efficacy of Quetiapine in the light of an increased volume of data; they have also considered the impact of missing data from clinical trials; particularly the common claim that loss of participants to follow-up may mask true drug efficacy.
- Their choice to use a measure of clinical efficacy as a yard-stick could be considered controversial by some, although they do take steps to justify their choice of measure, notably comparison with previous pragmatic clinical trials and pre-stated predicted values of efficacy from the included studies in their review.
- That Quetiapine IR fails to meet this specified standard of clinical efficacy is hardly surprising as very few current psychotropic or psychological treatments do meet these standards.
The study can be criticised in that the authors have deviated in a number of methods from their pre-report study protocol, which was registered within a database of systematic reviews. The authors justify this deviation as being an adaptation in the face of newly available statistical techniques, and also in the light of peer review comments. Indeed, given the language employed throughout the paper it would seem apparent that this review has been through a fairly bruising, and arduous, peer review process before this final report was produced.
Loss of participants to follow-up in many of the trials included in this review was marked, and the authors call for steps to be taken to ensure that these participants can be included in analysis wherever possible since no amount of statistical modelling, or corrective technique, can fully account for the potential impact of participant loss.
What is particularly important about this review however is the steps the authors have taken to obtain previously unpublished data from trial coordinators. This is an essential step in such a reappraisal, but depressingly draws comment from the authors with regard to the difficulty of obtaining this data. The authors end their report with a well justified plea that all clinical trial data be made readily and transparently available through resources such as the All Trials campaign.
Hutton P, Taylor PJ, Mulligan L, Tully S, Moncrieff J. (2015). Quetiapine immediate release v. placebo for schizophrenia: systematic review, meta-analysis and reappraisal. The British Journal of Psychiatry, 206(5), 360–370. http://doi.org/10.1192/bjp.bp.114.154377 [Royal College of Psychiatry paywall :( ]