1. GUIDELINES FOR THE DESIGN OF CLINICAL TRIALS INVESTIGATING THE EFFECT OF LIGHT TREATMENT IN DEPRESSION
Klaus Martiny
Psychiatric Research Unit, Frederiksborg General Hospital,Hillerød, Denmark, (Clinimetric Centre in Mental Health)
Light therapy is an non-patentable remedy for treating depression, and establishing evidence has been very difficult when compared to drugs. Studies in light therapy tend to be underpowered, leading to a disproportionately large number of statistically negative trials. Designs of light studies are very diverse and often their duration is too short to allow an antidepressive effect to build up. Placebo treatment is a problem shared with other physical therapies of depression and should be dealt with “to the best possible”. Before getting too distressed over the placebo problem it should be acknowledged that drugs trials by no means, due to side effects from the drugs under investigation are truly blinded either. This poster addresses these issues by suggesting the use of CONSORT statement (Moher et al., 2001) in a reverse manner. Thus, when designing a light therapy study, one should consider how the results of this study would be reported in a high impact journal and then “go backwards”, designing a state of the art light therapy study that will have the best chance of finding possible differences. The issue of power, i.e. the chance that a given true difference between groups is found is highlighted.
CONSORT statements*
Suggestions for the design light studies
Checklist of items to include when reporting a randomised trial
Item number
Description
Issues to remember when designing a clinical trial using light therapy
Title and abstract 1
How participants were allocated to interventions
A state-of-the-art study is: a randomised, double-blind, placebo- or active-controlled trial: 1. Design the randomisation procedure 2. Design the best possible blinding or active-control procedures 3. If placebo-controlled, pick a suitable placebo from the existing literature. 4. Secure that Good Clinical Practice guidelines is followed
(GCP, are followed.
Introduction
Background 2
Scientific background and explanation of rationale
Why is a new study needed? Good reasons for implementing a study: 1. Replication of findings if evidence is scarce 2. New ways of employing the treatment (colour, wavelength, intensity, timing of light, length of study) 3. Dose-response study comparing different light intensities 4. Active control group (drugs, ion-generator, psychotherapy exercise etc.) 5. Diagnostic groups (bulimia, sub-SAD, non-seasonal depression, sleep disorders, etc.)
Methods
Participants 3
Eligibility criteria for participants and the setting and locations where the data were collected
Can the study population be clearly defined? If not, generalisability is hampered. Can study participants be collected from one centre only, or are more centres necessary?
Interventions 4
Precise details of the interventions intended for each group and how and when they were actually administered
Is light treatment precisely defined with descriptions of: light intensity, treatment duration (min.), time of day, distance from light box screen, compliance measures (light treatment log book), written and oral instructions to participants on how to take light?
Objectives 5
Specific objectives and hypotheses
A study can rarely answer more than one research question. So don't try to pursue an array of interesting ideas!
Outcomes 6
Clearly defined primary and secondary outcome measures and, when applicable, methods used to enhance the quality of measurements (e.g.. multiple observations, training of assessors)
Use only well-validated outcome scales. Use both patient-reported outcome scales and clinician-rated scales and preferably a global measure such as the GAF and a quality of life measurement. Remember scales for side-effects and adverse effects (hypomania, irritability, etc.), (
Organise patient- or video-training sessions for the clinician-rated scales to secure inter-rater reliability .
Sample size 7
How sample size was determined and, when applicable, explanation of interim analyses and stopping rules
When using the Hamilton 17-item depression scale score, 3 points usually constitute a clinically relevant difference between two groups. With an expected standard deviation of around 4.5, the total number of participants (group A+ group B) would be 70 to achieve statistical significance at a 5% level with a power of 0.80 (2-sided test). Thus, studies with numbers below this might be over-relying on the efficacy of light and this might result in type-II errors.
Randomisation
Sequence generation
Allocation concealment
Implementation 8 9 10
Methods used to generate the random allocation sequence, including restriction details (e.g. blocking, stratification)
Methods used to implement the random allocation sequence (e.g. numbered containers or central telephone), clarifying whether the sequence was concealed until the interventions were assigned
Who generated the allocation sequence, who enrolled participants, and who assigned participants to groups?
Computer-generated random numbers should be obtained from an independent data manager, and the randomisation block size should not be revealed to the investigators.
Make firm efforts to secure the safe allocation to concealed numbers (envelopes) at inclusion, every day, all year round. Make written agreements stating who does what.
Blinding 11
Whether or not participants, those administering the intervention, and those assessing the outcomes were aware of group assignment. If not, how the success of masking were assessed.
Design procedures to create a clear separation between the investigators and those who handle the light equipment. Who will deliver the different light appliances? Is blinding of investigator secured? How to secure that patients don't mention their treatment to the investigator? Use expectancy ratings, just after inclusion, when patients are aware of their treatment group.
Statistical methods 12
Statistical methods used to compare groups for primary outcome(s); methods for additional analyses, such as subgroup analyses and adjusted analyses.
Who will perform the statistical analysis? Have power calculations been made to define an appropriate number of participants (see item no.7).
Results
Participant flow 13
Flow of participants trough each stage (a diagram is strongly recommended). Specifically, for each group, report the numbers of participants randomly assigned, receiving intended treatment, completing the study protocol, and analysed for the primary outcome. Describe protocol deviations from study as planned, together with reasons.
Study the CONSORT flow diagram very carefully * and think about if your procedures and Case Report Forms will give you all the data necessary to fill in the flow diagram. Journals assessors will ask for it!
In particular, focus on when participants drop-out of the study and why.
Recruitment 14
Dates defining the periods of recruitment and follow-up.
Use screening logs and inclusion logs.
Baseline data 15
Baseline demographic and clinical characteristics of each group
Will the necessary baseline characteristics be collected? Preferably, use multiaxial information as used in the DSM system (e.g. depression + evasive personality + hypertension + GAF + psychosocial stressors).
Numbers analysed 16
Number of participants in each group included in each analysis and whether the analysis was by “intention to treat”. State the results in absolute numbers when feasible (e.g. 10/20 ,not 50%)
Keep track of drop-outs and use intention-to-treat analysis for patients included and having started treatment.
Outcomes and estimation 17
For each primary and secondary outcome, a summary of results for each group, and the estimated effect size and its precision (e.g. 95% CL)
Choose outcome measures that are clinically meaningful and preferably commonly used, thus facilitating comparison with other studies. Avoid any measures not well validated. Go carefully through the literature and see what has been used in other light studies and in drug studies as well. Don't overload the study with scales!
Ancillary analyses 18
Address multiplicity by reporting any other analyses performed, including subgroup analyses and adjusted analyses indicating those prespecified and those exploratory.
Write a comprehensive protocol to start with and indicate if a subgroup analysis is envisaged. When writing a paper, the hypothesis put forward in the protocol should be tested and discussed.
Adverse events 19
All important adverse events or side-effects in each intervention group.
Side effects are often under-reported. Use standardized side effects scales ( with supplementary items for light-treatment side-effects (blurred vision, irritability, nausea, headache, migraine, etc).
Discussion
Interpretation 20
Interpretation of the results, taking into account study hypothesis, sources of potential bias or imprecision and the dangers associated with multiplicity of analyses and outcomes.
The better your knowledge of the existing literature, the better you will be able to have a relevant discussion.
Generalisability 21
Generalisability (external validity) of the trial findings.
When adequate data have been collected at baseline, generalisability should be quite easy to estimate.
Overall evidence 22
General interpretation of the results in the context of current evidence.
If the above suggestions are followed, you have come far in designing a study that will detect any clinically meaningful effect of light therapy.
*Moher D, Schultz KF, Altman DG and the CONSORT Group. The CONSORT statement: revised recommandations for improving the quality of reports of parallel-group randomised trials. The Lancet 357: