Transcript Differences in Drugs vs. Devices in Designing Clinical Trials

Drugs vs. Devices
Jeng Mah & Gosford A Sawyerr
Sept 16, 2005
Outline
•
•
•
•
•
•
Design Challenges
Analysis Issues: Device Database Challenges
Challenges with Diagnostics
Post-marketing surveillance
Working with Clinical Investigators
Regulatory Landscape
Design Issues
•
•
•
•
Double-blind device trials not common with device trials
Single blind possible within device type
Difficulty in going head to head with competitor devices
Pre-market trials tend to be simpler to demonstrate safety with
regard to intended use
• Compliance is easier to measure in device trials than in drug
trials
• In medical devices, single arm trials with Objective
Performance Criteria are used to determine safety of leads
• Single arm pharma trials may appear in early Phase II where
the objective may be to rule out drugs with little possibility of
clinical benefit (e.g., two-stage designs; ref: Simon)
Design Issues (2)
• Dose of drug may be an issue; size of device may be
an issue
• Implanting carries greater risk than prescribing a drug
• Device trials offer more opportunity to be Bayesian –
lots of prior information re: pacemakers and
defibrillators
• Device implant studies drive revenue much more than
prescription drug trials
• Post-market device trials are important in providing
clinical evidence for physicians, payors and patients
Design Issues (3)
• More precise endpoint determination due to
electronic information storage on device
• Drug trial endpoints subjective at times
• Consideration of recurrent episodes may provide
increased statistical power and hence fewer patients
• Some endpoint definitions in device trials may vary
from device to device (e.g., AF burden calculation)
• Memory capacity may lead to missing episodes
• Device date stamp resets may lead to inaccurate
information re: endpoints
Analysis Issues
• Analysis cohort may differ from full randomized set:
e.g., analysis of proportion of VT/VF episodes
appropriately treated may include only patients with
episodes
• Device storage limitations may miss episodes
• Heterogeneity in occurrence of episodes between
patients needs to be considered in choice of statistical
method so that few patients do not drive the results
• Sensitivity and specificity defined only relative to
what the device records and hence are biased
estimates
Challenges with Diagnostics
• Thresholds (cutpoints) which determine diagnostic
action need to be established a priori (in drug trials
retrospective analysis may be possible to determine
“sensitivity” to various definitions)
• Co-development of drug with diagnostic difficult
(studying two drugs simultaneously not necessarily at
same level of difficulty)
– see FDA Concept Paper:
• http://www.fda.gov/cder/genomics/pharmacoconceptfn.pdf
• Sensitivity and specificity issues critical
• Clinical utility of test – drug only trials do not have
this challenge
Post-marketing Surveillance
• In both device and drug trials, registry/observational
studies can play a key role in better understanding of
endpoints
• Possible to trace all patients with device: problematic
in drugs
• Data from return products, MDRs and registry studies
can assist in better assessment of system longevity
and adverse device effects
– Extent of exposure difficult to measure for all patients who
may have taken a drug
• MDR analysis may help trend analysis
Working with Clinical Investigators
• Strong partnership between device company
personnel and implanting physicians (EP,
Cardiologist, etc.)
– Development of study design highly collaborative with PI
– Delivery of device at implant and device programming
requires participation by company personnel
– Steering committee actively involved in assessment of
study and publication
– Events adjudication by clinicians on periodic basis
Regulatory Process
• Drugs:
– INDs, NDAs, sNDAs etc.
• Devices/Diagnostics
– IDEs, 510(k)s, PMAs, PMA-Ss etc.