Download presentation
Presentation is loading. Please wait.
Published byDarleen Farmer Modified over 8 years ago
1
Comments on FDA Concept Paper Sidney N. Kahn, MD, PhD President Pharmacovigilance & Risk Management, Inc. www.pvrm.com Risk Assessment of Observational Data
2
Conceptual omission - Therapeutic use (phase IV) studies and risk assessment Most sponsors conduct extensive therapeutic use (TU) studies of newly-approved products TU studies can provide additional characterization of known signals and identify potential new signals TU studies similar to Large Simple Safety Studies (Premarketing Risk Assessment Concept Paper) Assessment and reporting relatively simple and not excessively burdensome Results may allow more targeted approaches to other activities, e.g. pharmacoepidemiology studies, active surveillance programs –improved utility of results, less time and cost
3
Case follow up vs. new labeling proposal FDA proposed label includes only adverse “reactions” –omit events occurring in study population at a similar frequency absent drug therapy Pattern of spontaneous adverse events very similar to totality of adverse events in clinical trials Effect – marked increase in number of “unexpected” events, requiring expedited reporting and/or aggressive follow up; resource implications, increased “noise” vs. signal
4
Alternative label proposals Continue to include totality of adverse events with comparative frequencies Create additional label section containing common events considered “expected” for regulatory reporting purposes Omit common events from prescribing information; agree at NDA approval and in ongoing periodic reports what common events considered expected and not subject to aggressive follow up
5
Population background rates Knowledge of background occurrence rates, especially for relatively uncommon events, in a population directly comparable with that exposed to the product of concern, is essential for interpreting adverse event reporting rates and estimated incidence rates in the treated population FDA uniquely placed to obtain such data from control populations in archived NDA and BLA datasets CDER JANUS initiative should facilitate this in future Sponsors should allow their proprietary submission data to be used to help develop these assessments
6
Under-reporting -I High reporting rate with a high population background rate uninterpretable Event inclusion in or exclusion from labeling could affect reporting rate FDA public position: spontaneous reporting rate <10% of actual incidence rate Actual example –FDA assessed incidence rate of a severe AE at 20x population background rate based on spontaneous reports –company-sponsored epidemiology studies could not confirm or refute –independent expert opinion; 80%+ of cases reported –company received multiple reports of same case from different reporters –no increase in reporting rate after publications, label black box, Dear Dr. letter
7
Under-reporting -II Hypothesis: the more severe the event / outcome, and the more obviously drug-related (i.e. the less common it is in the treated population and without alternative explanation), the more closely the reporting rate will approach the incidence rate FDA should consider comparing the incidences in relevant databases of known adverse reactions of varying degrees of severity with the corresponding number of reports in AERS
8
Improving spontaneous report quality Environment in which health care professionals –understand the public health importance of reporting SADRs –have time and commitment to provide full data on SADRs to manufacturers or FDA –do not fear penalties for reporting medical mishaps or errors Education of physicians of the importance of reporting SADRs (medical curriculum, company sales reps.) Company sales representative encouragement of physicians to report SADRs Company or FDA feedback to physicians who report SADRs, e.g. notifying them of report disposition, label change, etc. Modest expense reimbursement to physicians and/or CME credit for time taken to provide follow-up information
9
Data mining - I No prespecified hypothesis Various statistical models designed to detect apparently disproportionate frequencies of product/event associations Many limitations – data quality, completeness, follow up capability Prospective value vs. passive surveillance not established False positive signals common, often reflecting only population comorbidity
10
Data mining - II More appropriate for regulatory authorities than pharmaceutical companies Unnecessary for strong AE signals Potentially useful for weak signals, but subject to confounding and false positives Unknown effects of different AE codification standards Continue exploratory use but establish no position without consensus criteria for signal validity
11
ICSR causality assessment Requires accurate, complete clinical and diagnostic data No universally accepted criteria –existence of multiple algorithms suggests none is definitive –all require subjective interpretation –most give indeterminate results (e.g. “possible”) An art, not a science Legal liability in USA
12
Use of pharmacoepidemiology studies Different uses vs stage / extent of clinical use Early clinical use –number of patients increasing but insufficient number in any databases to draw conclusions regarding uncommon SADRs –establish background rate of events of concern in relevant reference population (i.e. one that matches demographic and clinical profile of the treated population) to provide clinical context for spontaneously reported events Later in life cycle –hundreds of thousands or millions treated –databases permit analysis of signals identified by other methods –quantitation of comparative risks, cf. spontaneous reports
13
Thank you
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.