Pragma&c Clinical Trials Susanne May, PhD Associate Professor Department of Biosta&s&cs, School of Public Health, University of Washington
Preliminaries PLEASE ask ques&ons at any &me REALLY good clinical trials web site: <www.rctdesign.org> Sept 19, 2014 Pragma&c Trials - Instructor: SMay 2
Outline Background Pragma&c clinical trials Example 1 efficacy vs effec&veness Example 2 cross- over Summary Sept 19, 2014 Pragma&c Trials - Instructor: SMay 3
Start at the end Where do we want to be? Find a new treatment that improves health of individuals Find a new treatment that improves health of the popula&on Treatments administered to a community Treatments tested on a popula&on Sept 19, 2014 Pragma&c Trials - Instructor: SMay 4
Start at the end Where do we want to be? Find a new treatment that improves health of individuals Find a new treatment that improves health of the popula&on Treatments administered to a community Treatments tested on a popula&on Write the manuscript? Sept 19, 2014 Pragma&c Trials - Instructor: SMay 5
Clinical Trials in Context Observa&onal epidemiology of disease, risk Preclinical experiments Laboratory, animal studies of mechanisms, toxicology Clinical trials Safety for further inves&ga&ons / dose Safety of therapy Measures of efficacy Confirma&on of efficacy / effec&veness Synthesis and quan&fica&on of evidence Adop&on of new treatment indica&on Sept 19, 2014 Pragma&c Trials - Instructor: SMay 6
Treatment Indica&on Disease Puta&ve cause vs signs / symptoms May involve method of diagnosis, response to therapies Popula&on Restrict by risk of AEs or actual prior experience Treatment or treatment strategy Formula&on, administra&on, dose, frequency, dura&on, ancillary therapies Outcome Clinical vs surrogate; &meframe; measurement Sept 19, 2014 Pragma&c Trials - Instructor: SMay 7
Level of Evidence U.S. Preven&ve Services Task Force Level I: At least one properly designed RCT Level II: II- 1: Well- designed, nonrandomized CT II- 2: Well- designed, mul¢er cohort/case- control II- 3: Mul&ple &me series with/without interven&on; Drama&c results from uncontrolled trial Level III: Opinions of respected authori&es = Eminence based (not their wording!) Sept 19, 2014 Pragma&c Trials - Instructor: SMay 8
Anecdotal observa&ons Case report Case series Hypothesis genera&on Types of Studies That's not an experiment you have there, that's an experience. Sir Ronald A. Fisher (1890-1962) The plural of anecdote is not data. Roger Brinner www.causeweb.org Sept 19, 2014 Pragma&c Trials - Instructor: SMay 9
Types of Studies Designed observa&onal study: Case - control Sample diseased and nondiseased Examine rates of exposures Efficient for rare diseases Can look at mul&ple risk factors Limita&on: Cannot infer cause and effect Correla&ons with other factors Protopathic associa&ons Sept 19, 2014 Pragma&c Trials - Instructor: SMay 10
Types of Studies Designed observa&onal study: Cohort study Sample exposed and nonexposed Examine rates of disease Efficient for common diseases Can look at mul&ple diseases Can iden&fy retrospec&ve cohort Limita&on: Cannot infer cause and effect Correla&ons with other factors Protopathic associa&ons Sept 19, 2014 Pragma&c Trials - Instructor: SMay 11
Types of Studies Designed interven&onal study: Clinical trial Assign subjects to treatments Examine outcomes Can look at mul&ple diseases Can infer cause and effect Sept 19, 2014 Pragma&c Trials - Instructor: SMay 12
Clinical Trials Experimenta&on in human volunteers Inves&gates a new treatment/preven&ve agent Safety: Do adverse effects outweigh poten&al benefit? Efficacy: Does treatment beneficially alter the disease process Effec&veness: Would adop&on of the treatment improve morbidity / mortality in the popula&on? Sept 19, 2014 Pragma&c Trials - Instructor: SMay 13
Efficacy: A Moving Target Defini&on of efficacy can vary widely according to choice of endpoint and magnitude of importance Basic science Does treatment have any effect on the pathway Clinical science Does treatment have a sufficiently large effect on a clinically relevant endpoint in some subpopula&on of the target popula&on Sept 19, 2014 Pragma&c Trials - Instructor: SMay 14
Effec&veness: A Moving Target A treatment is effec&ve if its introduc&on improves health in the popula&on Considers the net effect of safety and efficacy in the popula&on as a whole Takes into account such issues as Noncompliance Off- label use Sept 19, 2014 Pragma&c Trials - Instructor: SMay 15
Effec&veness vs Efficacy A treatment can be both efficacious and ineffec&ve depending on such factors as Target popula&on Restricted eligibility due to toxicity, compliance Interven&on Training, quality control, compliance Comparison treatment No treatment, ac&ve treatment, ancillary treatments Measurement of outcome(s) Clinical disease vs subclinical markers Summary measure of outcome distribu&on Effects on mean, median, outliers Sept 19, 2014 Pragma&c Trials - Instructor: SMay 16
Effec&veness vs Efficacy May differ with respect to Certainty of diagnosed disease Subgroups with more (less) severe disease Tolerance of treatment Willingness to comply with treatment Ancillary treatments Different risk factors Sept 19, 2014 Pragma&c Trials - Instructor: SMay 17
Outline Background Pragma&c clinical trials Example 1 efficacy vs effec&veness Example 2 cross- over Summary Sept 19, 2014 Pragma&c Trials - Instructor: SMay 18
Pragma&c vs Explanatory Thorpe et al define a pragma&c study as follows:... authors coined the term pragma&c to describe trials that help users choose between op&ons for care And explanatory to describe trials that test causal research hypotheses (i.e., that a given interven&on causes a par&cular benefit). Sept 19, 2014 Pragma&c Trials - Instructor: SMay 19
Pragma&c vs Explanatory Pragma&c trials are primarily designed to determine the effects of an interven&on under the usual condi&ons in which it will be applied whereas explanatory trials are primarily designed to determine the effects of an interven&on under ideal circumstances. Sept 19, 2014 Pragma&c Trials - Instructor: SMay 20
Consort Extension: Pragma&c From Table 1: Key differences between trials with explanatory and pragma&c aptudes, adapted from a table presented at the 2008 Society for Clinical Trials mee&ng by Marion Campbell, University of Aberdeen Sept 19, 2014 Pragma&c Trials - Instructor: SMay 21
Comparison Efficacy can the interven&on work? Sepng Well resourced, ideal sepng Par&cipants Highly selected. Poorly adherent par&cipants and those with condi&ons which might dilute the effect are oqen excluded Effec&veness does the interven&on work when used in normal prac&ce? Sepng Normal prac&ce Par&cipants Lirle or no selec&on beyond the clinical indica&on of interest Sept 19, 2014 Pragma&c Trials - Instructor: SMay 22
Comparison Efficacy Interven&on Strictly enforced and adherence is monitored closely Outcomes Oqen short term surrogates or process measures Effec&veness Interven&on Applied flexibly as it would be in normal prac&ce Outcomes Directly relevant to par&cipants, funders, communi&es, and healthcare prac&&oner Sept 19, 2014 Pragma&c Trials - Instructor: SMay 23
Comparison Efficacy Relevance to prac&ce Indirect lirle effort made to match design of trial to decision making needs of those in usual sepng in which interven&on will be implemented Effec&veness Relevance to prac&ce Direct trial is designed to meet needs of those making decisions about treatment op&ons in sepng in which interven&on will be implemented Sept 19, 2014 Pragma&c Trials - Instructor: SMay 24
Outline Background Pragma&c clinical trials Example 1 efficacy vs effec&veness Example 2 cross- over Summary Sept 19, 2014 Pragma&c Trials - Instructor: SMay 25
Example 1 ROC PRIMED Lancet Sept 19, 2014 Pragma&c Trials - Instructor: SMay 26
Example Example Design/Sta&s&cs Double- blind placebo controlled cross- over trial Completed Needed help with sta&s&cal analysis and write- up Sept 19, 2014 Pragma&c Trials - Instructor: SMay 27
Outline Background Pragma&c clinical trials Example 1 efficacy vs effec&veness Example 2 cross- over Summary Sept 19, 2014 Pragma&c Trials - Instructor: SMay 28
Example: Cross- over trial Typical design for two treatments A B Wash-out period B A Outcome before treatment Outcome after end of treatment Outcome before treatment Outcome after end of treatment Sept 19, 2014 Pragma&c Trials - Instructor: SMay 29
Example: Cross- over trial Design for this study Primary outcome Hemoglobin level Secondary outcome Frequency and intensity of nose bleeds Time points of evalua&on Before treatment ini&a&on 3 months 6 months Sept 19, 2014 Pragma&c Trials - Instructor: SMay 30
Example: Cross- over trial Design for this study A B No Wash-out period B A Outcome before treatment Outcome Outcome after end of = = before treatment treatment Outcome after end of treatment Sept 19, 2014 Pragma&c Trials - Instructor: SMay 31
Example: Cross- over trial Poten&al issues Last observa&on from first phase = first observa&ons last phase No wash- out period jus&fiable? Sept 19, 2014 Pragma&c Trials - Instructor: SMay 32
Example: Cross- over trial What if there is an effect under A? Increase by 5. Calcula&ons Change under A AB sequence hba [t=0] hba [t=3] BA sequence hba [t=3] hba [t=6] Sept 19, 2014 Pragma&c Trials - Instructor: SMay 33
Example: Cross- over trial What if there is an effect under A? Increase by 5. Calcula&ons Change under A Change under B AB sequence hba [t=0] hba [t=3] ( hbb [t=3] hbb [t=6] ) BA sequence hba [t=3] hba [t=6] ( hbb [t=0] hbb [t=3] ) Under this calcula&on they should be the same!!! Sept 19, 2014 Pragma&c Trials - Instructor: SMay 34
What if there is an effect under A? Increase by 5. Calcula&ons Example: Cross- over trial Change under A Change under B AB sequence hba [t=0] hba [t=3] ( hbb [t=3] hbb [t=6] ) 9 14 ( 14 9 ) = 10 increase by 5 return to BL Sept 19, 2014 Pragma&c Trials - Instructor: SMay 35
Example: Cross- over trial What if there is an effect under A? Increase by 5. Calcula&ons Change under A Change under B AB sequence hba [t=0] hba [t=3] ( hbb [t=3] hbb [t=6] ) 9 14 ( 14 9 ) = 10 BA sequence hba [t=3] hba [t=6] ( hbb [t=0] hbb [t=3] ) 9 14 ( 9 9 ) = 5 increase by 5 return to/stay at BL Under this calcula&on they should be the same!!! Sept 19, 2014 Pragma&c Trials - Instructor: SMay 36
Example: Cross- over trial Absence of carry- over effect violated Basic premise underlying the design Misunderstanding wash- out period Wash- out of drug in the system Sustained effect on hemoglobin Sept 19, 2014 Pragma&c Trials - Instructor: SMay 37
Example: Cross- over trial N = 106 pa&ents interested and contacted N = 34 examined N = 22 started drug intake N = 2 non- compliant 9 Female, 11 Male 52 mean age (range 34 72) Sept 19, 2014 Pragma&c Trials - Instructor: SMay 38
Example: Cross- over trial Borom line? Sta&s&cian NEEDS to understand concepts of the applica&on Inves&gator NEEDS to understand concepts of the sta&s&cal analysis Sept 19, 2014 Pragma&c Trials - Instructor: SMay 39
Outline Background Pragma&c clinical trials Example 1 efficacy vs effec&veness Example 2 cross- over Summary Sept 19, 2014 Pragma&c Trials - Instructor: SMay 40
Summary Interplay science ó sta&s&cs Efficacy / explanatory Effec&veness / Pragma&c Start at the end. Sept 19, 2014 Pragma&c Trials - Instructor: SMay 41