1. GCP Selskabet|Jan2017 Henny B. Jakobsen
Observational clinical studies – why, when and how? Henny Bang Jakobsen, LEO Pharma A/S Medlemsmøde i Dansk Selskab for GCP, januar 2017

2. Mere om NIS, mest om primære data
Hvorfor holder vi os ikke bare til ‘the Golden Standard’: RCTs?
Hvad driver valget mellem primære og sekundære data?
Eksempel
Hvordan komplementerer RCTs og observationelle studier hinanden?
Hvad betyder det i praksis, at vi ikke kan styre behandlingen?
Hvad gør vi, når vi ikke kan fastlægge visits?
Adverse event indsamling – GPV Modul VI krav vs real-life
GCP Selskabet|Jan2017 Henny B. Jakobsen

3. Why clinical trials are not enough
Safety data from clinical trials underestimates real-life risk
Efficacy data from clinical trials often do not translate to real-life effectiveness
Increasing demand from ‘Payers’ for real-life data, preferably comparative
Clinical Trials do not provide volume and coverage for pro-active post-launch risk management
Only retrospective data may provide timely results
Not all treatment environments can carry the burden of GCP but they still represent patient reality
Safety data from RCTs underestimates real-life risk – higher risk patients and off-label users are not represented.
Efficacy data from RCTs overestimates the effect due to highly selected population, much more patient attention, better treatment adherence etc.
Payers pay for effectiveness, not for efficacy, and want to know what they can expect to get for their money.
Very large patient numbers are needed to detect rare side effects, and side effects developing slowly take very long term follow-up.
If a safety signal evolves, we cannot wait for primary data collection but have to sort it out as quickly as possible
NIS can help push implementation of treatment guidelines, being used as a instrument for ‘treatment quality control’
GCP Selskabet|Jan2017 Henny B. Jakobsen

4. Primary or secondary data
Based on prospective data collected after study initiation and according to the protocol
Secondary data
Based on data already collected for a different purpose, e.g. medical charts, insurance claim databases, the cancer registry, the administrative hospital database (Landspatientregistret), the Danish prescription database, general population statistics (often a combination of databases)
But may also be
A combination
GCP Selskabet|Jan2017 Henny B. Jakobsen

5. When an observational study design is the only option
The way to show real-life effectiveness, while RCTs measure efficacy
With primary data collection:
Details of real-life drug use – compliance in prescribing and adherence based on patient reported data
If product is newly launched and data not yet available for secondary use
Use and safety of Over-The-Counter medication
With secondary use of data (administrative databases):
For urgent evaluation of a potential safety issue
To study off-label prescribing
To study associations that may lead to focused targeting of subpopulations
To study a very rare event
To study a very slowly developing event
Urgent evaluation: For quick insight into real life; for reference/comparative group levels
GCP Selskabet|Jan2017 Henny B. Jakobsen

6. GCP Selskabet|Jan2017 Henny B. Jakobsen
Examples NIS based on primary data collection Birth M, Figueras J, Bernardini S, Troen T, Günther K, Mirza D, Mortensen FV: Collagen fleece-bound fibrin sealant is not associated with an increased risk of thromboembolic events or major bleeding after its use for haemostasis in surgery: a prospective multicentre surveillance study. Patient Safety in Surgery 2009;3:13. Available on 222.pssjournal.com/content/3/1/13 Strandell B, Norgren-Holst E, Tran N, Jakobsen HB, Chen S. OTC use of a topical nasal spray solution containing xylometazoline plus ipratropium in patients with common cold. International Journal of Clinical Pharmacology and Therapeutics 2009;47:

7. ZyComb® - a Swedish pharmacy based NIS
ZyComb® - Xylomethazoline + Ipratropium:
relief for stuffed and runny noses
No Rx market for a common cold nasal spray
Sweden only country with direct OTC status
Usual requirement for shift from Rx to OTC:
Collect safety data from post-marketing use on prescription
Present as part of national application for Rx to OTC status shift
GCP Selskabet|Jan2017 Henny B. Jakobsen

8. ZyComb® NIS pharmacy study
Challenges
Very low exposure if Rx
Patients receiving prescriptions not representative
Design
Non-Interventional
Recruit customers buying ZyComb® in Swedish pharmacies
Collect safety data through patient questionnaires – no medical records
General Practitioner responsible and available to pharmacy staff
Study Nurses calling patients reporting potential ADRs to get full report
Reports validated and approved by GP
GCP Selskabet|Jan2017 Henny B. Jakobsen

9. ZyComb® NIS pharmacy study
Enrolled
Did not use;
N = 23
Questionnaire
N = 723
Received
Interviewed
by Study Nurse
N = 296
Lost to FU
N = 9
Completers, N = 1019
GCP Selskabet|Jan2017 Henny B. Jakobsen

10. Mere om NIS, mest om primære data
Hvorfor holder vi os ikke bare til ‘the Golden Standard’: RCTs?
Hvad driver valget mellem primære og sekundære data?
Eksempel
Hvordan komplementerer RCTs og observationelle studier hinanden?
Hvad betyder det i praksis, at vi ikke kan styre behandlingen?
Hvad gør vi, når vi ikke kan fastlægge visits?
Adverse event indsamling – GPV Modul VI krav vs real-life
GCP Selskabet|Jan2017 Henny B. Jakobsen

11. A brief overview of Drug Development
GCP Selskabet|Jan2017 Henny B. Jakobsen
Discovery
Proof of Concept
Full Development
LCM
Tech.
Feas-ibility
Lead
Discovery
Lead
Optimization
Pre-clini cal
Ph I
Ph IIa
Ph IIb
Ph III
Regist.
Launch
Target
Decision
Lead Selection
Targeted Clinical Profile
Enter Ph III
Launch
First
in Human
Submission
Continue
LCM 11 11 11

12. NIS based on secondary data in early drug development
Epidemiological studies can inform clinical development strategies by describing the population diagnosed with a given disease, e.g.
How does the condition evolve over time – in particular for rare conditions
How many are they and where?
What are their unmet medical needs?
What other diseases do they typically have?
GCP Selskabet|Jan2017 Henny B. Jakobsen

13. Secondary data can inform protocols
A sanity check in existing databases on the feasibility of trial protocols can save time and money, e.g. by
Understanding medication shift patterns on available regimens
Understanding where eligible patients can best be identified
Estimating the patient population meeting selected and alternative inclusion criteria – newly diagnosed, on-top-of/instead of
Estimating the population left after main exclusion criteria
Checking that relevant co-morbidities and co-medications have been taken into account in the protocol
GCP Selskabet|Jan2017 Henny B. Jakobsen

14. Data evaluation during trial conduct
Sponsor follows the risk/benefit ratio during the conduct, primarily based on reported Adverse Events/Adverse drug reactions, but is usually blinded to treatment
Existing databases can provide
A reference for the representativeness of the included population
A background level from a corresponding population for evaluation of a potential signal (beware of reporting bias!)
GCP Selskabet|Jan2017 Henny B. Jakobsen

15. The complementary beauty
Controlled Clinical Trials have high internal and low external validity:
Comparability of cohorts are secured through strict selection criteria and randomization
The treatment is controlled and compliance checked
Patients are carefully monitored and seen often
Collected data are scrutinized and queried if inconsistent or unexpected
Observational studies have lower internal and higher external validity:
Primary data collection studies:
All sites qualify
Limited training of site staff in correct and complete reporting
Loss-to-follow-up is a more common issue
Less data cleaning
No control on timing of data collection
Studies based on secondary data:
No risk for study interference with data
Often wanted data are not available and proxy data may be inaccurate
Coding and categorisation errors distort the data
Cohorts may not be stable enough for long-term studies
GCP Selskabet|Jan2017 Henny B. Jakobsen

16. It is not a walk in the park!
You are only in the introductory phase of the course! This is only an appetizer…
GCP Selskabet|Jan2017 Henny B. Jakobsen

17. Grey zone areas – NIS based on primary data
Examples:
Use of questionnaires not standard on site
Telephone contacts outside clinic routines
Extra analyses on blood samples that should anyways be taken
Contact to family doctor for follow-up information etc.
Interpretations vary by country;
may even depend on who and how you ask!
Clinical Studies
Market Reseach Non-Interventional Studies Clinical Trials
(incl. large, simple trials)
Purpose and type of data
Intervention, e.g. randomization
GCP Selskabet|Jan2017 Henny B. Jakobsen

18. What to do when visits and treatments vary!
GCP Selskabet|Jan2017 Henny B. Jakobsen

19. And the statistician?
GCP Selskabet|Jan2017 Henny B. Jakobsen

20. GVP Module VI -
GCP Selskabet|Jan2017 Henny B. Jakobsen
Management and reporting of adverse reactions to medicinal products (Rev 1)
VI.C Non-interventional post-authorisation studies with primary data collection
“Information on all adverse events should be collected from healthcare professionals or consumers in the course of the study unless the protocol provides differently with a due justification for not collecting certain events.”
How can we balance this with representativeness of data of interest?

21. Safety reporting in studies based on primary vs secondary data
ICSRs considered solicited.
Adverse events of special interest in study database.
All adverse drug reactions and adverse events with fatal
outcome must be collected. Protocol to define collection
and reporting of other AEs, and OEs
Primary
data
Secondary data
No ICSR reporting required.
Summary in Study Report, unless protocol provides differently.
Guideline on good pharmacovigilance practices (GVP) Module VI –
Management and reporting of adverse reactions to medicinal products (Rev 1) pages
GCP Selskabet|Jan2017 Henny B. Jakobsen

22. My main learnings - primary data collection
NISes take a different mindset, clinical trial glasses do not fit
Be inspired by GCP but always question the relevance from a NIS perspective; use common sense
GCP can be counterproductive, preventing collection of real-life data that can be extrapolated.
Large samples compensate for lower precision
Be very aware of bias and confounding!
GCP Selskabet|Jan2017 Henny B. Jakobsen

23. Håber, tågen er lettet lidt…
Spørgsmål?
Working our way out of the grey…
GCP Selskabet|Jan2017 Henny B. Jakobsen

24. GCP Selskabet|Jan2017 Henny B. Jakobsen
Back-up slides

25. Relevant international guidance documents
Pharmacovigilance Regulation (EU) No 1235/2010 Guideline Module VIII – Post-authorisation Safety Studies and Module VI – Management and reporting of adverse reactions to medicinal products EU Data Protection Directive 95/46/EC World Medical Association Declaration of Helsinki, Ethical Principles for Medical Research Involving Human Subjects, 1964, last amended in 2013, since October 2008 covers observational clinical studies EFPIA HCP Code (on the promotion of prescription-only medicines to, and interactions with healthcare professionals) – Article 15 Non-Interventional Studies Guidelines for Good Pharmacoepidemiology Practices (GPP), International Society for Pharmacoepidemiology (ISPE) The European Network of Centres for Pharmacoepidemiology and Pharmacovigilance (ENCePP) : Guide on Methodological Standards in Pharmacoepidemiology Code of Conduct Guidance for Industry and FDA Staff: Best practices for Conducting and Reporting Pharmacoepidemiologic Safety Studies Using Electronic Healthcare Data Sets Clinical Trial Directive 2001/20/EC – only for definition of Non-Interventional Drug studies and statement that Good Clinical Practice requirements are not valid
GCP Selskabet|Jan2017 Henny B. Jakobsen

26. The concept of non-intervention
”Do not intervene, just observe, be a fly on the wall in the doctors office” Or ”Do not interfere with the treatment decision” And Consider if participants are exposed to risks that those who say no to participation will not be exposed to
GCP Selskabet|Jan2017 Henny B. Jakobsen

27. The EU concept of non-intervention
Drug focused definitions:
Definitions of a ”non-interventional trial” (or better: ”study”) (Clinical Trial Directive 2001/20/EC):
”a study where the medicinal product(s) is(are) prescribed in the usual manner in accordance with the terms of the marketing authorisation. The assignment of the patient to a particular therapeutic strategy is not decided in advance by a trial protocol but falls within current practise and the prescription of the medicine is clearly separated from the decision to include the patient in the study. No additional diagnostic or monitoring procedures shall be applied to the patients and epidemiological methods shall be used for the analysis of the collected data”
Article 1: ”The Clinical Trial Directive does not apply to non-interventional trials”
Interestingly enough the first definition is from a directive which does not apply to these studies. The term ‘trial’ should be avoided when discussion non-interventional studies as it indicates that something is being ‘tried out’; so it gives associations to intervention. The ENCePP definition has stayed free of this term.
GCP Selskabet|Jan2017 Henny B. Jakobsen

28. What is non-intervention?
The European Convention on Human Rights and Biomedicine – 1997
The Oviedo Convention
Additional Protocol to the Convention on Human Rights and Biomedicine, concerning Biomedical Research - Strasbourg,
Article 2 – Scope
3. For the purposes of this Protocol, the term “intervention” includes :
i. a physical intervention, and
ii. any other intervention in so far as it involves a risk to the psychological health of the person concerned.
Non-intervention therefore excludes these.
GCP Selskabet|Jan2017 Henny B. Jakobsen

29. Myndighedsanmeldelser og godkendelser i DK
Ikke-interventionsstudier i Danmark
skal ikke anmeldes til Lægemiddelstyrelsen, med mindre der er tale om bestemte typer non-interventions PASS studier, MEN Lif anbefaler konsultation i tvivlstilfælde
skal ifølge Helsinki deklarationen anmeldes til uafhængig videnskabsetisk komité, MEN de danske videnskabsetiske komiteer behandler ikke i øjeblikket NIS anmeldelser
skal godkendes af Datatilsynet, såfremt der behandles (direkte eller indirekte) personhenførbare, følsomme data, såsom helbredsoplysninger
I øvrigt:
Datatilsynet kan dispensere fra informeret samtykke til brug af data, hvis databehandlingen alene sker med henblik på at udføre statistiske eller videnskabelige undersøgelser af væsentlig samfundsmæssig betydning.
GCP Selskabet|Jan2017 Henny B. Jakobsen

30. Relevant Danish requirements for Non-Interventional Studies
Sundhedsstyrelsen (Competent Authority): Agreed between Lif, Sundhedsstyrelsen and Ethics Committees that the industry submits study to the MPA for the MPA to confirm the categorization as non-interventional and confirm that the study does not violate rules on promotion
Ethics Committees: No notification or approval but this is a special case; usually IEC/IRB should be at least notified
Datatilsynet: If Data Holder is Danish, approval is needed unless the study is only conducted in Denmark; a general approval can be obtained for holding such data.
GCP Selskabet|Jan2017 Henny B. Jakobsen

31. My main learnings - primary data collection
NISes take a different mindset, clinical trial glasses do not fit
Be inspired by GCP but always question the relevance from a NIS perspective; use common sense
GCP can be counterproductive, preventing collection of real-life data that can be extrapolated.
Large samples compensate for lower precision
Be very aware of bias and confounding
GCP Selskabet|Jan2017 Henny B. Jakobsen

32. My main learnings - continued
Make the study as easy as possible for the sites
Avoid elaborate site agreements
Keep ICF strictly to ‘use of existing data’, plus any extra questionnaires
Use an intuitive eCRF
Collect only data that you need and that you expect will be available in the standard medical records
Program electronic queries on key data
Avoid text fields and manual queries
Plan for deletion of true outliers (or skip reporting of ranges)
Share the data with sites as the study progresses to keep them interested
Communicate (on relevant subjects) to the sites to keep them alert
GCP Selskabet|Jan2017 Henny B. Jakobsen

33. My main learnings – use of existing data
Select validated databases, if possible
Check if the covered populations in the databases are stable and representative for the target population
Make sure the key data are available for the vast majority of the patients
Take into account any lag-time in data availability
For prospective studies, set up a step-wise approach so that the elaborate data extraction does not start until there are enough patients available
Select valid proxies for variables not directly available
Plan relevant sensitivity analysis
Partner with an expert who knows the databases to be used
Be aware that quality control is not standard in Academia
GCP Selskabet|Jan2017 Henny B. Jakobsen

34. TachoSil®- a thrombin/fibrinogen sealant
TachoSil® - Medicated sponge
A collagen carrier coated on one side with two-component human origin fibrin “glue” activated by fluids such as blood or saline
Indication:
To use during surgery for ‘improvement of haemostasis, to promote tissue sealing, and for suture support in vascular surgery where standard techniques are insufficient’
Updated from TachoComb®, based on bovine thrombin and fibrinogen, marketed since early 1990s, tested in over 2500 patients
Approved in EU June 2004, based on TachoComb® experience and clinical trials with TachoSil® in a limited patient population
GCP Selskabet|Jan2017 Henny B. Jakobsen

35. - an EMEA post-marketing safety commitment
As a condition for the Marketing Authorization for TachoSil, the EMEA requested
A 3000 patient surveillance study with 6 month Follow-Up
Focus on
Immunological Events
Thromboembolic Events incl. drug-interactions
Major Bleeding Events incl. drug-interactions
Focus events were collected in Case Report Forms
Other Adverse Events were not collected
Non-focus Adverse Drug Reactions were reported via usual post-authorisation surveillance process
GCP Selskabet|Jan2017 Henny B. Jakobsen

36. Design and challenges Design
Non-interventional, prospective, single cohort, 12 European countries
Main challenges:
Selection bias
Collection of 6 month data
Under-reporting of events
Interpretation of results – the hen and the egg
Lack of control group
GCP Selskabet|Jan2017 Henny B. Jakobsen

37. Consent and risk for selection bias?
Written consent to use of data
If before surgery:
Risk of interference with decision to use TachoSil®
Need to “consent” large numbers of patients not receiving TachoSil®
If after surgery:
Death during surgery?
Serious surgical complications?
GCP Selskabet|Jan2017 Henny B. Jakobsen