Definition

1
To randomise or not to randomise a matter of
perspective?
Mark Rodgers Duncan Chambers Nerys Woolacott
2

• This research was commissioned and funded by the
  NIHR Health Technology Assessment Programme
  (Project number 06/13/01) and will be published
  in full in Health Technology Assessment. See the
  HTA Programme website for further project
  information http//www.hta.ac.uk
• The views expressed in this presentation are
  those of the authors and not necessarily those of
  the NHS, NIHR or the Department of Health.

3
Introduction

• Randomised controlled trial (RCT) gold standard
  for deriving unbiased estimate of efficacy
• Where randomisation not possible, other forms of
  controlled evidence may be relied upon
• Evidence from uncontrolled case series generally
  considered inadequate to establish efficacy
• Situations where case series evidence jeopardises
  equipoise required to undertake higher quality
  studies?
• Case study of systematic review of an emerging
  technology

4
Background

• Systematic review of radiofrequency catheter
  ablation for typical atrial flutter
• Typical atrial flutter
• Distinct atrial arrhythmia diagnosed on ECG
• Heart rate abnormally fast
• Palpitations, shortness of breath, dizziness,
  nausea

5
(No Transcript)
6
(No Transcript)
7
Alternative treatments

• Rhythm control convert flutter to a normal heart
  (sinus) rhythm
• Cardioversion (electrical or pharmacologic)
  with/without anti-arrhythmic drug (AAD)
  maintenance
• Rate control slow the ventricular rate to normal
  (60-100 bpm)
• Drugs (beta-blockers, cardiac glycosides, calcium
  channel blockers, some AADs)

8
Inclusion criteria

• Participants Adults with established typical
  atrial flutter
• Intervention Radiofrequency catheter ablation
• Comparisons Rate or rhythm control
• Outcomes Freedom from atrial flutter, occurrence
  of other arrhythmias, symptoms, QoL,
  complications/adverse events
• Study designs RCTs (ngt20), non-randomised
  controlled studies (ngt100)

9
Uncontrolled case series

• Literature dominated by uncontrolled case series
• Reviewers primarily considered only larger series
  to be of value for rarer complications/adverse
  events
• Clinical advisors convinced of near 100
  effectiveness of RFCA on basis of small number of
  influential case series
• Findings of any review excluding these case
  series would not be taken seriously by
  clinicians
• Restricted inclusion to larger (ngt100) case
  series reporting consecutive patients

10
Results (1)

• 2 RCTs (n164), 23 case series (n4,238)
• Case series
• 19/23 rated poor quality
• Freedom from flutter at follow-up 68-98
• Rate of post procedural atrial fibrillation (AF)
  9-53
• Complications and adverse events uncommon
• Inconsistency in post-procedural drug treatment,
  where reported
• Potential double counting of patients could not
  be excluded definitively

11
Results (2)

• RCTs freedom from flutter
• Da Costa (n103) 1.36 (95 CI 1.13, 1.64)
• Natale (n61) 14.03 (95 CI 3.67, 53.7)
• Both favoured RFCA, but very different estimates
• Different populations, comparison groups
• Unusually strict monitoring of arrhythmia in
  Natale trial
• Inconsistent findings on occurrence of AF

12
Conclusions

• Very small amount of randomised evidence suggests
  higher proportion of patients undergoing RFCA are
  free from atrial flutter during follow-up in
  medium-term, relative to AAD based treatment
• Several outstanding uncertainties
• Relative impact on symptoms and QoL
• Conversion of flutter to AF
• Need for repeat procedures
• Influence of concomitant heart disease, age,
  duration of flutter
• Effectiveness in non-pioneering settings

13
Recommendations

• Prospective national registry of RFCA procedures,
  with set of standards for data collection.
• May be justification for a realistic
  multicentre RCT
• Representative population
• Against current best practice
• Standardised measurement of arrhythmia
• Patient-centred outcomes
• Typical treatment centres

14
Justification for future RCT

• Existing randomised evidence extremely limited
  several important outstanding uncertainties
• Case series evidence is partial and biased
• Publication bias
• Pioneer bias
• Represents the best that can be achieved, rather
  than what is likely
• Subsequent clinical and cost-effectiveness
  estimates likely to be overly optimistic

15
Arguments against future RCT

• Special case RFCA is curative, alternatives
  are not
• Associations seen in case series imply causation
  (Bradford-Hill critieria)
• Strength
• Consistency
• Specificity
• Temporal relationship
• Biological plausibility
• Coherence
• Point of equipoise required for ethical
  randomisation already passed

16
Role of case series evidence in HTA

• If case series predominant and influential, then
  need to be acknowledged
• if only to make explicit their limitations
• Biases publication bias, pioneer bias
• Poor reporting relevant clinical and
  methodological details frequently absent
• But practicalities must be considered
• Time required to screen, extract etc.
• Validity assessment
• Decisions about inclusion thresholds
• When to stop adding weight or diminishing
  returns?

17
Thank you