How to access and process FDA drug approval packages for use in research
Based on: Turner, E. H. (2013). BMJ (Clinical Research Ed), 347(oct14), f5992–f5992. doi:10.1136/bmj.f5992
Video: http://www.bmj.com/content/347/bmj.f5992#F1
Erick Turner, M.D.Former FDA Medical Officer / reviewer
Staff Psychiatrist, Portland VA Medical CenterAssociate Professor, Oregon Health & Science University
Background on FDA workings Before the study
1. Sponsor decides to pursue marketing approval for a new drug-indication combination goals:• Get drug onto US market (1st indication)• Advertise new indication (print ads, TV)
2. Sponsor submits IND registers trial with FDA• Registering trials with FDA has been required for decades• IND is for a specific indication
3. FDA reviews protocol learns of trial’s......existence
• sponsor can’t later pretend it never happened
...methodological details • avoids post hoc “torturing of the data until it eventually confesses”
After the study (NDA stage)
• FDA receives clinical study reports (CSRs)– All studies– Summary statistics– Raw data
• FDA conducts review– Return to IND orig. protocol planned methods – Raw data + prespecified methods re-analyze– ? same results as sponsor?– Results of review written up
• FDA decision on drug-indication combination– Not approved “trade secret” results not made public– Approved review documents “Drug Approval Package”
• Drug Approval Packages posted on website “Drugs@FDA”Turner, E. H. (2004). A taxpayer-funded clinical trials registry and results database. PLoS Medicine, 1(3), e60. doi:10.1371/journal.pmed.0010060
Limitations of Drugs@FDA-1
• Dates: 1997 to present– Electronic Freedom of Information Act (eFOIA) of 1996– Approval ≤ 1996 can be “FOIA’d”
• Pre-marketing, not postmarketing• No off-label uses– Sponsor might have sought approval but failed
• First approved indication reliably available, less so for 2nd, 3rd, etc.
Limitations-2• Completeness (no patient-level data)– Reviews / summaries of CSRs, not CSRs themselves– Just summary statistics (like journal articles)
• May omit SDs, SEs (but can calculate ES from P values)
• User-unfriendly– Formatting issues
• Redaction of “confidential” info / “trade secrets”• Searchability
– Newer drugs = searchable PDFs– Older drugs = image files = not searchable (but you can apply OCR)
– Website cumbersome to navigate...
www.fda.gov
Before you dive in, avoid wasting time on blind alleys
• Are you sure the indication is FDA-approved and not off-label? Counter-examples:– Fluvoxamine (an SSRI, but not approved for depression) – Gabapentin
• Used for many types of pain• Approved only for postherpetic neuraligia (and szr d/o)
• Was this the first FDA-approved indication?– Yes job is easy – oldest item in approval history – No find approval date, then look in approval history
• Tip: often more efficient to use brand, not generic, name– Generics reviewed only for bioequivalence to innovator– No data on efficacy or safety
Example using generic name...
Generic name extraneous hits
Generic version. Not a new molecule or new formulation. Reviewed only for bioequivalence to brand.
No (re)review of or data on efficacy or safety
Metabolite of venlafaxine, but considered an NME. Approved in 2008 avail. online
“Brand generic”, same as Pristiq, approved 2013. Bioequivalence only, no efficacy.
Generic of Pristiq bioequivalence only, no efficacy or safety
If you use the brand name instead...
Brand name just 2 hits
New molecular entity (NME) approved in 1994, before eFOIA (so NOT available online)
New formulation of existing molecular entity (EME). Approved in 1997 available...
Approval history page
Scroll down to
earliest action date.
Earliest action date = original approval
Medical review: efficacy + safety
Statistical review: efficacy only
Inside the review
“The Registry”
Table of Studies, page 1
published
Use FDA results to verify results in published literature
Table of Studies, page 2
Not published
Study 367(unpublished)
Results on primary
outcome
NS x
2 doses
What if the indication is not the 1st one approved?(Warning: Less likely to find this online)
1- Be sure the indication was indeed approved by the FDA (not off-label use)• Product label
2- Find out when it was approved• Google
3- Find that date in approval history and see if there’s a “Review” link
Example
• Drug = aripiprazole (Abilify®)• Indication– Original FDA-approved indication = schizophrenia– Indication of interest = autistic disorder• Question: Is aripiprazole approved for autistic d/o?• Problem: Can any drug truly treat a developmental
disorder?• Indication needs to be more specific
Free online product label DailyMed
http://dailymed.nlm.nih.gov Collaboration between FDA and NLM
Indications & Usage section
Label List of FDA-approved indications
Locate date in the approval history see if there is an assoc’d review
The review document
Review not online? Consider FOIA request
http://www.accessdata.fda.gov/scripts/foi/FOIRequest/requestinfo.cfm
FOIA request, cont’d• Avoid “complex queue”. Make your request...
...small (1 drug, 1 indication)
...specific (see FDA’s approval letter to sponsor)• drug name, approval date, NDA #, supplement #
• Can take 1 month to ∞• May be “a bridge too far” given time constraints• If you get the data, consider sharing publicly for
future research– Supplemental data accompanying publication