Designing patient-centered clinical trials: Results of the MDIC project to use patient preference information to design clinical trials

Transcription

1 1 Informative series of workshops featuring emerging trends in medical technology regulatory science, MDIC projects and subject matter experts sharing perspectives, progress and opportunities. Designing patient-centered clinical trials: Results of the MDIC project to use patient preference information to design clinical trials Speakers: Annie Saha, US FDA Center for Devices and Radiological Health Brett Hauber, PhD, RTI Health Solutions Shomesh Chaudhuri, PhD, MIT Laboratory for Financial Engineering The last 15 minutes are reserved for Q&A. Please enter your questions for the panel through the chat feature on Zoom

2 Incorporation of patientcentered outcomes research in clinical trial design Anindita Saha Director, External Expertise and Partnerships Office of the Center Director FDA Center for Devices and Radiological Health 2

3 Specific Aims 3

4 4 Aim 1: Identify Important Outcomes Review of literature and previous clinical studies Conversations with MJFF Patient Scientists and survey of Patient Council Coordination with FDA reviewers Output: Potential PD treatment benefits and risks to be considered when assessing patient preferences

5 5 Identifying Outcomes Activity Identify Refine Prioritize Define Participants Sponsors Patients Physicians Sponsors Patients Physicians Patients Physicians Sponsors

6 Identifying Outcomes 1 Identify and develop preliminary list of potential attributes through literature review Activity Identify Refine Prioritize Define Participants Sponsors Patients Physicians Sponsors Patients Physicians Patients Physicians Sponsors 6

7 7 Attribute Development Cognitive Impairment Memory Distress Slowed thinking Confusion Dementia Psychological Mood Depression Hallucinations Sensory Pain Fatigue Sensory abnormalities Loss of smell or taste Sleep Daytime sleepiness REM behavioral Insomnia Resting tremor Bradykinesia Rigidity Loss of postural reflexes Gross Motor Skills Impaired movement Freezing Gait disturbance Muscle spasms Movement speed Unwanted acceleration Slowness Stiffness Shuffling Falls Fine motor skills Autonomic dysfunction Swallowing Drooling Gastrointestinal Urinary dysfunction Nausea Hypertension Hypotension Sweating Speech No stimulation Lead migration and fracture Allergic or immune system response Erosion of device Intermittent stimulation Additional surgeries to maintain therapy Hardware revision required Battery failures Infection Confusion (temporary and long term) Seizures Cerebral infarction/stroke Brain hematoma Brain bleeding Coma Death Aborted procedures Muscle Spasms Gait disorders Short term rehabilitation required Long term nursing care required Worsening mobility Falling Disequilibrium

8 Refining Outcomes Refine and clarify attributes through 2 discussion and focus groups Activity Participants Identify Sponsors Refine Patients Physicians Sponsors 8

9 What s most important? What can we measure? How do patients feel? What s currently done? Refining Outcomes Brain bleed Infection Additional surgery No stimulation Serious adverse event Minimal adverse event Brain bleed List of Risks Commonalities/ Historical usage Clinical Endpoint 9

10 Prioritizing Outcomes Prioritize attributes by using ranking surveys and discussion 3 groups Activity Identify Refine Prioritize Participants Sponsors Patients Physicians Sponsors Patients Physicians 10

11 11 Survey Results Summary

12 Defining Outcomes Develop unbiased, patient-centered descriptions in discussion groups 4 Activity Identify Refine Prioritize Define Participants Sponsors Patients Physicians Sponsors Patients Physicians Patients Physicians Sponsors 12

13 13 Survey Results Summary

14 14 Resulting Benefits and Risks Benefits Increase in daily on time (X% decrease in off time ) X% decrease in motor symptoms Burdens Risk of (worsening) depression or anxiety Risk of serious adverse event (brain bleed) X% decrease in PD pain Increase in 1-year mortality risk X% decrease in cognitive impairment X% decrease in medication and side effect burden

15 15 15 August 2018 Patients Preferences for Parkinson s Devices MDIC Webinar Brett Hauber, PhD Senior Economist and Vice President Health Preference Assessment RTI Health Solutions Affiliate Associate Professor School of Pharmacy University of Washington

16 16

17 17 Aim 2 Objectives 1. Estimate Parkinson s patients benefit-risk preferences for potential device treatments Benefits and risks identified in Aim 1 2. Provide inputs to Aim 3 model Preference weights for benefits and risks will define the loss function Time tradeoffs will inform the discount rate 3. Have an enduring survey as part of the Fox Insight platform Add data from additional respondents as they enter Fox Insight Collect preference data longitudinally among participating patients

18 18 Aim 2 Process Started with the patient- and reviewer-relevant benefits and risks identified in Aim 1 Developed definitions for each attribute Review and discussion with members of the MJFF Patient Council Review and input from FDA reviewers and staff Drafted survey with input from project team and FDA reviewers Conducted 20 telephone pretest interviews with people with Parkinson s Administered the online survey to Fox Insight participants

19 19 Aim 2 Survey

20 20 Final Set of Benefits - Levels Benefits Increase in daily on time Decrease in motor symptoms Decrease in PD pain Decrease in cognitive impairment Decrease in medication and side effect burden Magnitude of Improvement 50% decrease in daily off time from self-reported baseline (hours) 50% decrease in severity of motor symptoms from self-reported baseline (on an 11-point scale) 50% decrease in severity of PD pain from selfreported baseline (on an 11-point scale) 50% decrease in difficulty thinking clearly from self-reported baseline (on an 11-point scale) 50% reduction in self-reported number of daily pills and associated side effects

21 21 Final Set of Burdens Levels Risk Risk of (worsening) depression or anxiety Risk of serious brain bleed Increase in 1-year mortality risk Increase in wait time (discounting) Starting Level 40% annual risk (based on clinical input) 4% annual risk (based on clinical input) 2% annual risk (based on clinical input) 3 years (based on expected lead time and patient input)

22 22 Threshold Technique Respondent presented with 2 alternatives: Reference Treatment Typically standard of care or Current treatment Target Treatment Hypothetical or real-world alternative to the existing state of the world Respondent asked to choose between reference and target treatment The level of one attribute of the target alternative is varied systematically until the respondent switches his or her preferred alternative Target made systematically better (more attractive) if reference treatment chosen first Made systematically worse (less attractive) if target is chosen first

23 23 Threshold Technique Question 5 treatment benefit outcomes New device (target) defined by 50% improvement in selfreported level of one benefit outcome 3 treatment risks Risk varied systematically to determine maximum threshold

24 24 Time Tradeoff Question Device A Device B Severity of movement symptoms MS1-1 (on a scale from 0 to 10) MS1 2 (on a scale from 0 to 10) Time until you get the device Now 3 years Which option would you choose? Choice between a very small improvement today and a larger improvement in the future

25 25 Aim 2 Results

26 26 Demographic Characteristics Demographics n=2,740 Mean Age (SD) 65.4 (9.01) Female, n (%) 1,279 (46.7%) Employed outside the home, n (%) 680 (24.8%) Caucasian, n (%) 2,593 (94.6%) 4-year college degree or higher, n (%) 1,912 (69.7%)

27 27 Parkinson s Symptoms Symptom Number reporting symptom (%) Symptom Level Mean (SD) Average hours of on time* 1,677 (61.2%) 10.8 (3.78) Severity of movement symptoms ** 2,649 (96.7%) 4.3 (2.06) Severity of pain** 1,348 (49.2%) 4.5 (2.23) Severity of cognitive symptoms** 1,217 (44.4%) 4.4 (2.17) *symptom was off time; respondents reporting off time were asked how many hours of on time they had in a 16 waking hours each day **symptoms rated on a 10-point scale in which 0 indicated no symptoms and 10 indicated very severe symptoms

28 28 Other Characteristics Parkinson s Related Characteristic n=2,740 Number of daily pills mean (SD) 7.6 (5.39) Years since diagnosis mean (SD) 5.3 (4.91) Prior deep brain stimulation n (%) 219 (8%) Biological relative with PD n (%) 569 (20.8%) Experience with Risk Outcomes n mean (SD) Severity of current depression or anxiety** 1, (2.09) Prior brain bleed 67 (2.4%) Known someone who died after an operation 922 (33.6%) **symptoms rated on a 10-point scale in which 0 indicated no symptoms and 10 indicated very severe symptoms

29 29 Maximum Acceptable Risk for Increases in Daily On Time 25.0% Maximum acceptable risk 20.0% 15.0% 10.0% 5.0% Depression Brain bleed Death 0.0% Increase in on time (hours per day)

30 30 Maximum Acceptable Risk of Each Risk for Each Benefit for Respondents with Median Sample Age (66 years) 18% 16% Maximum acceptable risk 14% 12% 10% 8% 6% 4% Depression Brain bleed Death 2% 0% On time Movement Pain Cognition Pills Benefit The benefit levels were set as the midpoint of the benefits offered in the threshold questions. They are an improvement of hours of on time, a reduction in movement, pain, or cognition of 3.25, and a reduction of 4.5 pills per day.

31 31 Maximum Acceptable Mortality Risk by Subgroup 3.5% 3.0% Maximum acceptable mortality risk 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% On time Movement Pain Cognition Pills Benefit No covariates Non-ambulatory Cognition problems DBS The MAR of death for an improvement in cognition is similar to the MAR of death for all other benefits The benefit levels were set as the midpoint of the benefits offered in the threshold questions. They are an improvement of hours of on time, a reduction in movement, pain, or cognition of 3.25, and a reduction of 4.5 pills per day. Cognition was not included as a covariate in any of the models in which the benefit was an improvement in cognition, because all respondents who saw those versions of the threshold exercise reported cognition problems.

32 32 Maximum Acceptable Mortality Risk for Pain Improvement by Age Group and Covariates 3.5% Maximum acceptable mortality risk 3.0% 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% 60 years old years old years old > 71 years old Age group Assumes a benefit of a reduction in pain of 3.25 points on an 11-point scale. No covariates Non-ambulatory Cognition problems Covariates for ambulation, cognition, and DBS experience were only included in the individual age group models if more than 15 observations had the characteristic expressed by the covariate. DBS People aged who are less ambulatory have the highest MAR of death to reduce pain

33 33 Aim 2 Key Findings People with PD are willing to accept device-related risks to improve traditional outcomes: On-time Movement symptoms People with PD are also willing to accept comparable increases in device-related risks to improve other outcomes: Pain Difficulty thinking clearly People who have already had DBS are, on average, slightly more willing to accept device-related risks than those who have not Risk tolerance varies systematically with patient age

34 Patient-Centered Randomized Clinical Trials for Parkinson s Disease Shomesh E. Chaudhuri, PhD August 15, 2018 MIT Laboratory for Financial Engineering

35 Statistical Inference Involves Trade-offs Approve Reject Effective Type II error Ineffective Type I error Standard approach sets Type I error = 5% What if patients prefer higher Type I error in exchange for smaller Type II error? 8/15/2018 Slide 35

36 Statistical Inference Involves Trade-offs 8/15/2018 Slide 36

37 Review of Clinical Trial Statistics Balanced two-arm randomized clinical trial Patient value model based on survey results Maximize value (minimize harm) using BDA framework λ = 1.65 β = 0.36 α = 0.05 Z-Statistic 8/15/2018 Slide 37

38 Bayesian Decision Analysis BDA-optimal decision maximizes expected value to patients 8/15/2018 Slide 38

39 Patient -Values for Neurostimulator % 9% 240 8% Sample size (2n) % 6% 5% 4% 3% Significance level (α) 60 2% 1% Mild symptoms, No DBS Subgroups ordered from low to high risk tolerance Non-DBS DBS 0% Severe symptoms, DBS 8/15/2018 Slide 39

40 Key Findings A fixed significance level of 5% does not maximize patient value For risk-tolerant subpopulations (DBS, severe cognitive and motor function impairment), lengthy clinical trials are overprotective of the type I error rate For patients with less severe symptoms, traditional thresholds of 5% may be too permissive 8/15/2018 Slide 40

41 Thank You!

42 42 Questions and Discussion Submit your questions via the Chat Box Next MDICx webinar Sept. 12