in Rare Patients " Some day risk will be zero My, oh my Some - PowerPoint PPT Presentation

6/22/2013 Ninewells Hospital & Medical School Some day drugs will be perfect If we try Some day drugs will be perfect " Assessing Therapies And no one will ever die in Rare Patients " Some day risk will be zero My, oh my Some day pills will be magic And they’ll taste of apple pie Tom MacDonald Hierarchy of Evidence How do we decide? 1a Systematic review of RCTs with homogeneity 1b Individual RCT with narrow confidence intervals 1c All or none (i.e. fatal before Rx or ‘Lazarus’) 2a Systematic review of cohorts with homogeneity 2b Individual cohorts (including low quality RCTs) 2c Outcomes research, ecological studies 3a Systematic review of case-control studies 3b Individual case-control studies 4 Case series 5 Expert opinion, bench research, first principles www.cebm.net/levels_of_evidence.asp 1

6/22/2013 Types of Rx to be evaluated • Rapid onset + rapid loss of effect – i.e. short acting symptomatic relief • Rapid onset with carry-over effects – i.e. short acting symptomatic relief with disease modifying effects • Slow onset + slow loss of effect – disease modifying drugs • Pre-licensing • Post-licensing More Tests = Mean of Multiple Measures the target Improved Precision Test Frequency Value Obtained 2

6/22/2013 SDD The Standard Deviation of the Differences between repeated measures in the same subjects 24h BP 5mmHg Lower Lumiracoxib v SDD values of 8.3 for ABPM and Ibuprofen 11.0 for Clinic SBP Lumiracoxib 100 mg qd MASBP (mmHg) Ibuprofen 600 mg tid 140 For a parallel trial to detect 135 ∆ of 10 mmHg SBP, 130 125 51 subjects using Clinic SBP 120 v 115 MacDonald et al J Hyp 2008;26:1695-1702 29 using ABPM 110 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Am J Hypertens 2002;15:101-104 Hour of the day using 24-hour clock 3

6/22/2013 Pros + Cons Outcome Measures • Symptom scores / QOL • Symptom scores – Pro: Easy • Exercise capacity – Con: Soft – 6 min walking • Exercise Tests – Treadmill, bicycle, etc (+/- VO 2 max) – Pro: Harder endpoint – Con: Tiring, training effect • Activity • Activity / Saturation Monitors – 24h activity monitors – Pro: Measure important variables – 24h O 2 saturation monitors – Con: Less ‘hard’ than testing • Something else? • Other endpoints – Pulmonary artery pressure? – Surrogates, ? relevance – Biomarkers? Cohort Studies • Sample size issues • Accurate baseline important • Serial measures of outcome (not randomisation) • Good for drugs with slow onset & disease modifying effects BMJ 2005;330;843 4

6/22/2013 Rx New Treatment Assess suitability & randomize Rx Placebo or Standard Rx Placebo Expensive placebo more effective than cheap placebo JAMA 2008;299:1016-7 5

6/22/2013 Repeated ‘measures’ tiring Where possible (pre licensing) for patients compare to placebo Need to embrace new technology On top of best Rx iPhone Accelerometer App Use Technology to capture large amounts of outcome Programmable to Buzz to data over time stimulate exercise! http://www.jawbone.com/up 6

6/22/2013 Phenotypes Patients are Precious Multiple Measurements improve precision, reduce variance and Severe disease: few Mild disease: many symptoms symptoms increase power Disease Severity Example It is hard to improve the nearly normal Study severe hypertension: big falls in BP Study mild hypertension: small falls in BP Study low (normal) BP no fall in BP 7

6/22/2013 ACCORD-BP: Not a Mysterious Result 154 RISK UKPDS 14mmHg Lower BP: 144 134 119 ACCORD No benefit on APTC events Reduced stroke N Engl J Med 2010;362:1575-85. BP Study More Severe Disease If Possible 8

6/22/2013 Given the between patient and within How many patients are needed to show the beneficial effects patient variability of a drug? How can we study the One effects of drug treatments? 9

6/22/2013 N of One Crossover Studies Plac Rx A Preferred Rx - - - - - - 50 subjects to detect 30% + + + + + + preference with 85% Power 6 + v 6 - = P<0.05 even with a non-parametric test How many Crossovers of what duration? N of 1 studies carried out in • Effect size expected relatively small numbers can – Binary: 6 & 6 minimum be very informative – Small effects: many crossovers – Large effects: fewer crossovers Provides one solution to the • Pharmacodynamics determines pre-licensing assessment of Duration of treatments drugs in rare conditions – Short acting: weekly crossover – Medium acting: 1 month crossover 10

6/22/2013 How Drugs are Licensed New Rx Proportion of subjects Average of Placebo 8% benefit Disease Activity Benefit : Risk Analysis Individual Effects Benefit No Benefit 11

6/22/2013 Trial of Therapy Always Useful N-of-One Studies Lots of benefit: accept some risk Pick the Winners No benefit: accept no risk post licensing Streamlined Studies Treatments Mailed to Patients Research Pharmacy 12

6/22/2013 Only Significant Benefit Rx New Rx GP Rx n-of-one pack (free) Licensed Cost- effective Enters unique pack code threshold at and of crossover Patient enters Long term e-follow up data each week Washout Period Plac Rx A Downside or possible advantage Carry-over effects Helps reduce carry-over effects 13

6/22/2013 Washout Period (fixed or random) Vary block duration to measure with re-measured baseline carry over effects Plac Rx A 1 1 2 3 2 4 3 4 5 5 6 6 Plac Rx A ∆ ∆ ∆ ∆ ∆ ∆ ∆ ∆ ∆ ∆ ∆ ∆ Test: is ∆ Rx A > ∆ Plac ? Also: 6 time-dependent baselines following placebo and 6 following active treatment Random Variable Washout Periods Measure Carry-over 1 0 0 Plac Rx A 9 0 t 8 0 n e m 7 0 t a e 6 0 r t 5 0 o t 4 0 e s n e 3 0 p s e 2 0 Analyse the duration of R 1 0 0 carry over effects 0 7 1 4 2 1 2 8 3 5 4 2 4 9 5 6 D a y s 14

6/22/2013 Ethical Issues Carry-over • Everyone gets active Rx v placebo – Pre licensing • Or active Rx v alternative Rx effects help – Post licensing • Added to ‘best current Rx’ • Patients completing at least 1 Mitigate missed dose effects etc crossover informative • Multiple long term crossovers feasible N-of-one studies Catch-22 Pre & post licensing • Cannot get medicine used • Good for: until cost-effective & safe – Symptoms – Patient measured outcomes – ‘Activity’ / exercise capacity • Cannot get cost- – Measureable variables effectiveness (or safety) • Not so good for: data until medicine is used – Long term outcomes. 15

6/22/2013 Random introduction of ‘New Rx’ with Designs for licensed drugs ‘Designed delay’ in comparators 1 year? 50% Usual Care 50% 100% New New Rx Rx Designed Delay, Stepped Designed Delay, Stepped Wedge, or Similar description Wedge, or Similar description 100 100 Patients Not Rx Randomised Patients Not Rx Randomised if Possible if Possible 90% not Rx 50% not Rx 50% 10% Rx 50% Rx 0 0 0 100 0 100 Patients Treated/Time Patients Treated/Time 16

6/22/2013 Designed Delay, Stepped Wedge, or Similar description 100 Patients Not Rx Post Licensing 0% not Rx Stepped wedge / designed delay studies likely to be approved by ethics committees 100% Rx 0 0 100 Patients Treated/Time Cost and staff resources a major problem BMJ 2008;337:1085-7 17

6/22/2013 MEMO Sign up Patients To: Can we do internet only studies? Engage Patients Never or rarely see patients? 18

6/22/2013 Is this a crazy idea? eContact • Dedicated website https:// www.safetyswineflu.co.uk • Online / paper registration – Patient information sheets – Consent – Monthly follow-up • Automated email or SMS BJCP 2011 Nov 15. doi: 10.1111/j.1365-2125.2011.04142.x. [Epub ahead of print ] 19

6/22/2013 20

6/22/2013 MEMO MEMO “Investigators should develop and improve methods to help decision makers appraise the evidence” Harveian oration at the Royal College of Physicians, London www.rcplondon.ac.uk/pubs/brochure.aspx?e=262 21

6/22/2013 Balancing Benefit & Risks My Advisors 22

6/22/2013 23