1 / 72

Randomized Controlled Trials

Randomized Controlled Trials. Introduction to Sequence Generation. Critical Methodological Elements in RCTs. Randomization Avoiding and handling exclusions after trial entry Blinding. Excellent Familiarity with Blinding in the Scientific Community.

didier
Download Presentation

Randomized Controlled Trials

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Randomized Controlled Trials Introduction to Sequence Generation

  2. Critical Methodological Elementsin RCTs • Randomization • Avoiding and handling exclusions after trial entry • Blinding

  3. Excellent Familiarity with Blinding in the Scientific Community • Blinding embodies a rich history spanning several centuries • Most researchers worldwide understand blinding terminology, but . . . • confusion lurks beyond a general comprehension • terms such as “single-blind,” “double-blind,” and “triple-blind” mean different things to different people

  4. Inflating the Importance of Blinding • Some investigators, readers, and editors overstate the importance of blinding • Indeed, some consider RCTs as high quality if “double-blind,” i.e. sine qua non of an RCT • Unfortunately, scientific life is not that simple • A randomized trial can be methodologically sound and not be double-blind or, conversely, double-blind and not methodologically sound

  5. Inflating the Importance of Blinding Lasagna captured that notion long ago in 1955 "Let us examine the placebo somewhat more critically, however, since it and 'double blind' have reached the status of fetishes in our thinking and literature. The Automatic Aura of Respectability, Infallibility, and Scientific Savoir-faire which they possess for many can be easily shown to be undeserved in certain circumstances."

  6. Inflating the Importance of Blinding • “Mefloquine Prophlaxis Prevents Malaria during Pregnancy: A Double-Blind, Placebo-Controlled Study” • Nosten et al. JID 1994; 169: 595-603. • “We report here a double-blind, placebo-controlled trial of . . .” • Not only did the authors not describe how they randomized, but any mention of randomization at all is difficult to find

  7. Blinding Important, but Less Important than Widely Believed? • Unclear allocation concealment • ORs exaggerated by 33% (95% CI; 25%-40%) • Inadequate allocation concealment • ORs exaggerated by 41% (95% CI; 27%-52%) • Not double-blinding • ORs exaggerated by 17% (4%-29%) • Important, but also more difficult to successfully accomplish than allocation concealment • Side effects or even intervention effectiveness

  8. Confused Terminology of Single, Double, and Triple Blinding Permeates the Literature • Physicians, textbooks, and journal articles vary greatly in interpretations and definitions [Devereaux et al. JAMA 2001; 285: 2000-3] • Define “double-blind” inconsistently • Authors frequently fail to report their definitions clearly • When I use “double-blind”, participants, investigators, and assessors are blinded • In reporting RCTs, authors should explicitly state what steps were taken to keep whom blinded • Building upon once common double blindfold terminology (Lasagna) we further obfuscate . . .

  9. Double-Blinded Single-Blinded

  10. Double-BlindedTriple-Blinded

  11. Blinding or Masking • Different terms to describe the same procedures • “Masking” may be more appropriate semantics: • Participants with impaired vision • Less confusing with blindness an outcome • But more pc in all trials? • “Blinding” conveys strong bias prevention • Apparently, “blinding” terminology emerged when Benjamin Franklin et al. actually blindfolded participants when evaluating Mesmerism • Lasagna used the term “double blindfold” in 1955

  12. Blinded vs Masked

  13. Blinding or Masking • We prefer “blinding” because: • it rests on a long history • maintains worldwide recognition • If you use “masking” someone using PubMed in Asia or Africa may not know what you did • creates strong visual imagery • permeates the ICH guidelines

  14. Final Thoughts on “Double”-Blinding: Beyond Overvaluation • Overrated – cannot always double-blind • Usually reduces differential assessment • May improve compliance and retention • May reduce biased supplemental care or treatment (co-intervention) • Without it, better to at least blind outcome assessments • except with “hard” outcomes, such as death

  15. Randomization • Principal bias reducing technique • Success depends upon successful implementation • Chance rather than choice eliminates selection bias

  16. Assures Comparability • In observational studies, statistical methods allow investigators to control for confounding factors • Must be measurable • Must be measured © 1986 CCP

  17. Assures Comparability (Cont.) • No statistical method can achieve comparability on unknown or unmeasured factors in analysis phase • Random allocation is the only known method to assure comparability

  18. Hormone Replacement Therapy (HRT) and Coronary Heart Disease (CHD) • HRT decrease risk of CHD by 35% to 50%, according to 3 different meta-analysis of numerous observational studies • Especially strong for secondary prevention in women with CHD

  19. Experimentation Trumps Observation* • RCT of hormone-therapy for secondary presentation of CHD • Relative hazard = 0.99; 95% CI 0.80 - 1.22 • No effect of hormone therapy • Recent Women’s Health Initiative RCT in healthy women (JAMA 2002; 288: 321-333) • CHD: HR=1.29; 95% CI 1.02-1.63) *Petitti DB. JAMA 1998; 280:650-2

  20. Randomized Controlled Trials • The methodologic standard of excellence for medical research • RCTs: gold standard in theory • Frequently a bronze standard in practice

  21. Randomized Trials Require Methodological Rigor • Improperly conducted RCTs yield biased results • Researchers must devote assiduous attention to design and conduct of RCTs • Only properly conducted RCTs will fulfill their promise of minimizing bias

  22. Randomization Depends Upon Two Interrelated Processes • Sequence generation • An unpredictable allocation sequence must be generated based on a random procedure • Allocation concealment • Strict implementation of that schedule must be secured through an assignment mechanism that prevents foreknowledge of treatment assignment

  23. “Randomization”(Cont.) • Crucially, allocation concealment shields those who admit participants to a trial from knowing upcoming assignments • Accept or reject decisions and informed consent obtained without foreknowledge

  24. Quality of Reports of Randomized Trials • Reviewed 2000 randomized trials of all treatments for schizophrenia • Only 4% (n=80) of the trials clearly described the methods of allocation Thornley B, Adams CE. Content and quality of 2000 controlled trials in schizophrenia over 50 years. British Medical Journal1998;317:1181-1184.

  25. Reporting of Method of Randomization • Review of 122 RCTs of selective serotonin uptake inhibitors in patients with depression (Hotopf et al. 1997) • Only 1 trial report included details of the method of randomization Hotopf M, Lewis G, Normand C. Putting trials on trial - the costs and consequences of small trials in depression: a systematic review of methodology. Journal of Epidemiology and Community Health1997;51:354-358.

  26. Method of Allocation of Treatment in 208 Controlled Trials in Head Injury Method of allocation No. of trials Adequate Centralised randomisation by telephone 1 Numbered/coded identical containers administered 11 sequentially Randomisation scheme controlled by pharmacy 8 Sequentially numbered, sealed, opaque envelopes 2 Not adequate Other 18 Date of birth 1 Day of week 3 Alternation 3 Not stated 161 Dickinson K, et al., BMJ 2000;320:1308-1311.

  27. Allocation Methods • Whim • Judgement • Chance • The quasi- simulated “Randomized” approach

  28. Allocation Methods (Cont.) ABABABAB... • Alternate assignment • Chart number • Date of birth Smith01-1956-M 12_02_1956 Random?

  29. A Controlled Trial of Povidone-Iodine as Prophylaxis Against Ophthalmia Neonatorum AbstractBackground. Neonatal conjunctivitis Ophthalmic neonatorum) continues to cause blindness, because the agents used prophylactically to prevent this condition are not completely effective and are not widely available in many parts of the world. Povidone--iodine ophthalmic solution is an effective antibacterial agent with broad antibacterial and antiviral activity to which no bacteria are known to be resistant, and it is “Randomization was achieved by rotating the three medications after each was used for a week.” NEJM 1995;332:562-6

  30. Stopping Smoking in Pregnancy: Effect of a Self-help manual in Controlled TrialSummary. For medical reasons, encouraging women to stop smoking during pregnancy and post partum has high priority. Many smokers want to stop smoking but decline clinical treatment when it is offered. The aim of this study was to find a method which was accepted by a large number of smokers, had a high success rate and, at the same time, “Women were randomized . . . born on days 1-10 of every month formed the control group (n=231), and those born on days 11-31 formed the treatment group (n=492). ” British Journal of Obstetrics and Gynaecology

  31. Nifedipine in the Treatment of Severe PreeclampsiaWe conducted a randomized clinical trial in which patients with severe preeclampsia between 26-36 weeks of gestation receive either nifedipine (10-30 mg sublingually, then 40-120 mg/day orally; N= 24) or hydralazine (6.25-12.5 mg intravenously, then 80-120 mg/day orally; N= 25) “We conducted a randomized controlled trial. . . ” “Subjects were assigned to the nifedipine or hydralazine group according to the week of the month.” Obstetrics and Gynecology

  32. The use of Histoacryl for Episiotomy RepairSummary. Histoacryl-tissue adhesive (B. Braun Melsungen AG W. Germany) was used in place of skin sutures (2/0 chromic catgut, Ethicon Ltd, Edinburgh, Scotland) for episiotomy repair in a group of 20 women. This group was compared with two groups of women undergoing first and repeat episiotomy. Variables analysed included pain in the episiotomy site, pain while walking, sitting, sleeping, lying down, breast-feeding, micturating and defaecating. The Histoacryl group was superior with regard to all the variables. This simple, “Groups 1 and 3 (first episiotomy repair) were selected randomly, by registration number; group 1 odd and group 3 even numbers.” British Journal of Obstetrics and Gynaecology ADONIS 0306545910019N

  33. Two Problems • Non-random • Difficult to conceal anyone assigning or referring patients know in advance the next assignment • Decide eligibility • Time their referral Eva Canoutas/FHI

  34. Sequence Generation • Flipping a coin? • Rolling dice? • Shuffling cards? • Random but tempt investigators toward non-randomness (independent events–Jack B) • Adequate methods but not optimal • Cannot be checked – no audit trail • More difficult • Not recommended

  35. Sequence Generation(Cont.) • Table of random numbers? preferable & recommended • Random • Reproducible - can be checked • Easier (not widely recognized) • Same can be said for most computer random number generators

  36. Types of Sequence Generation • Simple randomization • Replacement randomization • Blocked (random permuted blocks) randomization • Small or large block sizes • Randomly varied block sizes • Restricted shuffled randomization • Stratified randomization

  37. Type of Sequence Generation (Cont.) • Simple • Perfectly unpredictable • Large sample sizes, generate groups of relatively similar sizes • Small sample sizes, will sometimes generate groups that differ quite substantially in sizes (5% will result in an imbalance in size greater than 1.96 √ ; N = total study size) • Probably used less frequently than should be N

  38. The Difference (T-C) and Total Size for96 Unrestricted Trials 95% Prediction interval boundary 50% Prediction interval boundary * * * * * * * * * * * * * * * Treatment minus control * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 50% Prediction interval boundary * * 95% Prediction interval boundary Total Trial Size

  39. Power reduction as the proportion on new treatment is increased 1 Percentage of Patients on the New Treatment

  40. Restricted Randomization • Biased–coin – alters probability of assignment • Restricted shuffled – essentially one large block the size of the study (random allocation rule) • Shuffling and envelopes not ideal, though acceptable • Later assignments obvious • Minimization – several prognostic factors • Blocked (balanced) • Probably the most common restricted approach

  41. Blocked Randomization • Every ?B? number of participants, the group sizes are the same • At least two good properties • Similar group sizes at the end of the trial • Similar group sizes throughout the study, which reduces variability due to a time trend in the prognosis of participants and is helpful for interim analyses

  42. Blocked Randomization Caution • If the block size is unraveled, and if treatment assignments become known after allocation, a sequence can be discerned from the pattern of past assignments • Some future assignments could then be accurately anticipated and bias introduced, regardless of the effectiveness of allocation concealment

  43. Lessening the Chances of Deciphering with Blocked Designs • Long block lengths are more difficult to decipher than short block lengths (12 better than 4) • Randomly vary the block lengths throughout the trial • Consider the importance of blocking when your trial is not double-blinded

  44. END

  45. Cohort ExposureOutcome Case-Control OutcomeExposure RCT Randomized Controlled (Clinical) Trials (RCT) Epidemiologic Study Designs

  46. RCT Paradigm Randomize Population of Interest PTX Placebo © 1999 Luke Mwanza/CCP Sample (Subset) Child < 5 year presenting at hospital with severe malaria Outcome Assessment Death within 7 days

  47. COHORT TROHOC RANDOMIZED DEZIMODNAR

  48. Biocompatible membranes in acute renal failure: prospective case-controlled study. Schiffl H, Lang SM, Konig A, Strasser T, Haider MC, Held E. “The mortality of critically ill patients with acute renal failure has been halved through intervention by haemodialysis. However, several reports suggest that the course of the disorder may be prolonged by this procedure. Our prospective randomised study was done …” Lancet 1994; 344: 570-2

  49. A. Bradford Hill’s Early Attempts at “Randomization” “The reason . . . the allocation of alternate cases to the treated and untreated groups is often satisfactory, is because no conscious or unconscious bias can enter in, as it may in any selection of cases, and because in the long run we can fairly rely upon this random allotment of the patients to equalise in the two groups the distribution of other characteristics that may be important.” A. Bradford Hill. Principles of medical statistics. Lancet 1937, pp. 41-43.

More Related