How to design Homeopathy clinical randomised trials that

Transcription

1 How to design Homeopathy clinical randomised trials that work Dr Clare Relton BA, MSc, PhD, FSHom Research Fellow School of Health & Related Research (ScHARR)

2 Homeopathy & trials Myths Only evidence required is trials Trials show that homeopathy works homeopathy does not work Cannot design trials of homeopathy that work Reality Lots of trials asking different questions using different (in/appropriate) methods Providing a variety of answers

3 Background Q: What type of evidence is needed? A: Many different types Clarity Stakeholders Questions ( it, work ) Ensure the evidence matches question! Prioritise Acceptable? Safe? Cost effective?

4 Types of clinical evidence Remedies Homeopaths Principles Acceptable I ; II Safe Cost effective Combinable Effective vs placebo? Effective vs comparators? ~ Effective for condition a,b,c...

5 Cost effective Acceptable + Safe + Effective (vs nothing & vs other treatments) = Cost effective Calculated by using information derived from observational and experimental (randomised) studies

6 EBM evidence hierarchy EBM hierarchy Premise: practice based on sound research evidence Supremacy of double blind placebo RCT No acknowledgement of what questions being addressed

7 What is a randomised controlled trial (RCT)? Routine healthcare doctors and patients try treatments everyday... Interpret and act on results A Trial is a formal: set of tools address research question rigorous and replicable Controlled trials Randomised trials Randomised Controlled Trials analyse results using probabilistic statistics

8 Gold standard dbp RCT Referred patients Eligible patients Obtain multiple consents No Yes Trial treatment Randomised Placebo Trial treatment Placebo Outcomes Outcomes

9 Problems with trials (Relton, 2010) Fail to recruit 71% of 114 publicly funded RCTs fail to recruit sufficient numbers (Campbell, 2007) _ Unrepresentative Ethical issues Lack of adequate comparators

10 Problems: multiple types of consent (I) be treated (II)...particip ate in research (III) be observed (IV)..have a doctor who does not know which treatment is best for me (V) have no control over which treatment I get I consent to (VI)..the possibility that I will get a dummy treatment and I wont know if it is or not.and neither will my Low UNCERTAINTY High

11 Less than 5% of trials funded by UK Dept of Health (HTA/MRC funded) are double blind placebo randomised controlled trials Majority of HTA/MRC funded trials are pragmatic randomised controlled trials

12 Trial designs Placebo & sham Explanatory Pragmatic Superiority Equivalence Non-inferiority Cross over Cluster Randomised consent designs Cohort multiple design Stepped wedge design Comprehensive cohort study Patient preference trial Two stage trial Clinician preferred trial

13 Pragmatic RCT (FMS) Referred patients N= 72 Eligible patients Obtain multiple consents Not interested n=18 Ineligible n=7 No Yes Standard care Randomised Homeopath care N=24 N= 47 N= 23 Standard treatment New treatment Completed Outcomes N= 16 Outcomes not completed N= 8 Completed Outcomes N= 20 Outcomes not completed N= 3

14 Problems: multiple types of consent (I) be treated (II)...particip ate in research (III) be observed (IV)..have a doctor who does not know which treatment is best for me (V) have no control over which treatment I get I consent to Low UNCERTAINTY High

15 Cohort multiple RCT design Relton, C. et al. BMJ 2010;340:c1066 Copyright 2010 BMJ Publishing Group Ltd.

16 Cohort multiple RCT design (I) Recruitment of a large observational cohort of patients with the condition of interest (II) Regular outcome measurement for the whole cohort (III) Capacity for multiple RCTs over time (IV) For each RCT, all eligible patients in the cohort are identified (NA) (V) Random selection of some eligible patients (na), and trial treatment offered (VI) Outcomes of randomly selected patients (na) are compared i.e. the outcomes of those eligible patients not randomly selected (NA na) (VII) Patient centred informed consent i.e. patient information and consent aim to replicate that of real world routine healthcare

17 Cohort multiple RCT design Eligible patients (consent: contact & provide data) Standard Random selection of some Offer of new tx Obtain consent to tx Standard treatment No Standard treatment Yes New treatment Outcomes Outcomes Outcomes

18 Types of consent (II)...particip ate in research (III) be observed (I) be treated I consent to Low UNCERTAINTY High

19 Stepped wedge design Participati ion/clusters Time periods Shaded cells represent intervention periods Blank cells represent control periods Each cell represents a data collection point

20 Types of consent (I) be treated I consent to Low UNCERTAINTY High

21 Stepped wedge design Variety of reasons for use Experimentally staged introduction of intervention Used to evaluate interventions in developing countries (especially HIV) Description of methods used & data analysis require more consistency

22 To obtain Conclusions rigorous information on acceptability, safety and cost effectiveness that is... generalisable/ relevant to routine healthcare... then need pragmatic RCT designs cohort multiple RCT design ready made cohorts/populations e.g. Disease registers stepped wedge RCT design introduce homeopathy in rural or regional clinics by stages/steps

23 References Campbell, M. (2007), Recruitment to randomised trials: Strategies for trial enrolment & participation study (STEPS) Relton et al (2010) Rethinking pragmatic RCT design: the cohort multiple RCT design. doi: /bmj.c1066 Brown C, Lilford R, The stepped wedge trial design, a systematic review, BMC Medical Research Methodology 2006, 6:54