stats and epi 210911

Upload: bshinkins

Post on 07-Apr-2018

220 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/4/2019 Stats and Epi 210911

    1/30

    Rethinking pragmatic randomised

    controlled trials: cohort multiple

    randomised controlled trial design

    Dr Clare ReltonSheffield School of Health and Related Research (ScHARR)Faculty of Medicine

    University of Sheffield

    [email protected]

  • 8/4/2019 Stats and Epi 210911

    2/30

    Pragmatic trials

    The term pragmatic was first applied to

    clinical trials by Schwartz & Lellouch(1967) whose made the distinctionbetween explanatory trials (which aim to

    further knowledge as to how and why) andpragmatic/practical trials (which aim toinform healthcare decisions within routinepractice).

  • 8/4/2019 Stats and Epi 210911

    3/30

    21/04/2012

    Problems with standard designs Recruitment: 70.6% of 114 multi centre

    MRC & HTA funded trials failed to recruit

    required number within time originallyspecified (Campbell M., 2007)

    Is this important?: Failure to recruit has

    implications for the cost, and thevalidity/reliability/comparability of theresults of the RCT.

  • 8/4/2019 Stats and Epi 210911

    4/30

    Problems with standard designs

    Poor recruitment rates PLUS informed

    consent a barrier to recruitment (Ross,1999), patient and clinician treatmentexperiences altered, lack of long term

    outcomes, unrepresentative recruitedpopulation, poor external validity....

  • 8/4/2019 Stats and Epi 210911

    5/30

    cohort multiple RCT design Recruit observational cohort of patients with condition of

    interest

    Regular outcome measurement for the whole cohort Capacity for multiple RCTs over time

    For each RCT, eligible patients identified from which somerandomly selected to be offered the intervention

    Outcomes of randomly selected patients compared to notrandomly selected.

    Patient information and consent replicate real world routinehealthcare i.e. patients are not told about treatment that theymight not receive

  • 8/4/2019 Stats and Epi 210911

    6/30

    Copyright 2010 BMJ Publishing Group Ltd.

    Relton, Torgerson, OCathain & Nicholl.

    BMJ 2010;340:c1066

    Cohort multiple RCT design

  • 8/4/2019 Stats and Epi 210911

    7/3021/04/2012

    Based on (Zelen) randomisedconsent design

    Used to avoid Hawthorne, resentful

    demoralisation, avoidance of contamination,simpler IC & recruitment, avoid unnecessarydistress & confusion

    Reviews: Schellings 2006, Adamson 2006

    Similar to cluster RCT design

  • 8/4/2019 Stats and Epi 210911

    8/30

    Differences: cmRCT & Zelen Large cohort of patients with condition of

    interest with capacity for multiple trials

    Regular outcome measurement of wholecohort

    Rethinking of..... Randomisation

    Informed consent procedures

  • 8/4/2019 Stats and Epi 210911

    9/30

    Randomisation

    Random allocation of all

    Vs

    Random selection of some

  • 8/4/2019 Stats and Epi 210911

    10/30

    Patient centredinformation and consent

    Replicates information and consent

    procedures in routine healthcare wherepossible

    Patients not told about treatments that

    they are not then offered

    Patients not told prospectively that theirtreatment will be chosen at random

  • 8/4/2019 Stats and Epi 210911

    11/30

    Rethinking informed consent

    Two trials

    CFS (Weatherley-Jones, 2004 )

    FMS (Relton , 2009)

    Patients behave differently

    Ethical issues

    Addressed - patient centred approach

  • 8/4/2019 Stats and Epi 210911

    12/30

    Standard pragmatic RCT

    Issues with the type, timing and audience

    of information provided Combine multiple types of information

    All at one single time point

    For all trial subgroups (regardless oftreatment eventually offered)

  • 8/4/2019 Stats and Epi 210911

    13/3021/04/2012

    (II)and

    there isresearchongoing

    (III) and wewant to observeyou.

    (IV)andwe are notsure whichtreatment

    is best.

    (V)andwe aregoing toplay a

    game ofchance

    (I) There isa treatment

    and thebenefitsare the

    risks are

    Type and timing ofinformation

    (standard RCT design)

  • 8/4/2019 Stats and Epi 210911

    14/30

    Patient centred approach

    Type, timing and audience of information

    provided mimics routine healthcare Separate research information from

    intervention information

    Differentiate between treatment groups andnon treatment groups - adjust informationaccordingly........ Patients not told abouttreatments they wont receive

  • 8/4/2019 Stats and Epi 210911

    15/3021/04/2012

    (II)there isresearchongoing

    (III) andwe want toobserve

    you.

    (IV)andwe are notsure whichtreatmentis best.

    (V revised)

    and youhave been

    selected atrandom to tryit....

    (I) There isa treatmentand thebenefitsare the

    risks are

    Type and timing of information(patient centred used in cmRCT design)

  • 8/4/2019 Stats and Epi 210911

    16/30

    Information mismatches

    (between pragmatic open trials and routine healthcare)

    Chance

    Trials: Patients are informed that their treatment will be decided by chance

    Routine healthcare: this is not the case.

    Unavailable treatments

    Trials: Some patients provided with information about treatments not

    available to them

    Routine healthcare: this rarely happens.

    Timing

    Trials: Multiple types of information (I-VI) at single time point (prior to

    randomisation)

    Routine healthcare: flow of information is staged and shaped by the patients

    needs at any particular time.

  • 8/4/2019 Stats and Epi 210911

    17/30

    Examples of cmRCT

    Menopausal hot flush (short term

    condition) South Yorkshire Cohort (population based)

    Rotherham Institute for Obesity Research

    Facility (long term condition and clinicalpopulation)

  • 8/4/2019 Stats and Epi 210911

    18/30

    South Yorkshire Cohort:a cohort

    trials facility study of health and weight

  • 8/4/2019 Stats and Epi 210911

    19/30

    NIHR CLAHRC funded

    Obesity theme

    Population based

    Research Facility

    http://www.biomedcentral.com/1471-2458/11/640

    South Yorkshire Cohort

  • 8/4/2019 Stats and Epi 210911

    20/30

    South Yorkshire Cohort Recruitment 40 GP practices, 180,000

    patients mailed postal health questionnaires

    NHS number

    Consent to

    Contact you again

    Use the information you provide to look at thebenefit of health treatments

    Look at your health records

  • 8/4/2019 Stats and Epi 210911

    21/30

    South Yorkshire Cohort

    10,000 patients recruited

    Full recruitment (20,000) Oct 2012

    Two studies embedded (DaWM, IQuEST)

    Further studies sought

    Qualitative

    Quantitative including pragmatic cmRCTs

  • 8/4/2019 Stats and Epi 210911

    22/30

    South Yorkshire Cohort (June 2011)

  • 8/4/2019 Stats and Epi 210911

    23/30

    South Yorkshire Cohort (June 2011)BMI category BMI

    Range

    Count Percent

    Underweight < 17.5 56

  • 8/4/2019 Stats and Epi 210911

    24/30

    Self reported Long-standing conditions (n=9,532)*P-value

  • 8/4/2019 Stats and Epi 210911

    25/30

    Rotherham Institute of Obesity

    Clinical condition based cohort

  • 8/4/2019 Stats and Epi 210911

    26/30

    Benefits of cohortmultiple RCT approach

    Recruitment improved quantity and morerepresentative sample

    a multiple RCT facility

    long term outcomes as standard

    ongoing information as to the natural history of thecondition and treatment as usual

    increased comparability between each trial conductedwithin the cohort

    increased efficiency, particularly for expensive or highrisk interventions (unequal randomisation)

  • 8/4/2019 Stats and Epi 210911

    27/30

    Most suited to.. Open trials with treatment as usual as comparator

    Research questions with easily measured & collectedoutcomes

    Clinical conditions where many clinical trials will be

    conducted

    Chronic conditions

    Highly desired treatments or expensive treatments

  • 8/4/2019 Stats and Epi 210911

    28/30

    Least suited to..

    Closed trial designs with masking or placebo arms

    Research questions with hard to measure and hard to

    collect outcomes

    Acute or short term conditions

  • 8/4/2019 Stats and Epi 210911

    29/30

    Thank you

    Acknowledgements

    Professor Jon NichollProfessor Alicia OCathainProfessor David Torgerson

    Department of Health FellowshipUniversity of SheffieldSouth Yorkshire CLAHRC

  • 8/4/2019 Stats and Epi 210911

    30/30

    References Adamson, J., Cockayne, S., Puffer, S. et al. (2006), Review of

    randomised trials using the post-randomised consent (Zelens) design,Contemporary Clinical Trials, 4, 305-319.

    Hewitt, C.E., Torgerson, D.J., Miles, J.V.N., (2006), Is there anotherway to take account of noncompliance in randomized controlled trials?

    Canadian Medical Association Journal, 175, 347. Schellings, R., Kessels, A.G., ter Riet, G. et al. (2006), Randomized

    consent designs in randomized controlled trials: Systematic literaturesearch, Contemporary Clinical Trials, 27, 4, 330-332.

    Campbell, M. (2007), Recruitment to randomised trials: Strategies fortrial enrolment & participation study (STEPS)

    http://www.hta.ac.uk/fullmono/mon1148.pdf Relton et al (2010) Rethinking pragmatic RCT design: the cohort

    multiple RCT design. doi: 10.1136/bmj.c1066http://www.bmj.com/cgi/content/full/bmj.c1066

    http://www.hta.ac.uk/fullmono/mon1148.pdfhttp://www.hta.ac.uk/fullmono/mon1148.pdf