Catrin Tudur Smith
Carrol Gamble
North West Hub for Trials Methodology Research
Outline Purpose of monitoring
Risk proportionate approach
Empirical example from cancer
Survey of current practice
Future work
Purpose of monitoring
Verify that :
(a) The rights and well-being of human subjects are protected.
(b) The reported trial data are accurate, complete, and
verifiable from source documents.
(c) The conduct of the trial is in compliance with the currently
approved protocol / amendment(s), with GCP and with the
applicable regulatory requirement(s).
(ICH GCP 5.18)
Approaches to trial monitoring
Oversight of the quality of the trial
o Central monitoring - use of centralised procedures for
quality control of trial dataIncludes checks for missing data across sites, exploring
unusual patterns in the data, checking accuracy of data with
external sources
o On-site monitoring – use of procedures for quality
control of trials data undertaken during on-site visits Includes source data verification: procedure used to check
that the data contained in the Case Report Form match the
primary source (e.g. medical record)
ICH GCP‘.... In general there is a need for on-site monitoring
before, during and after the trial; however in exceptional
circumstances the sponsor may determine that central
monitoring in conjunction with procedures such as
investigators’ training and meetings … can assure
appropriate
conduct of the trial in accordance with GCP’
(ICH GCP 5.18.3)
Risk assessment Process of identifying potential hazards, their
likelihood and harms
Risks associated with IMP
Safety risks from clinical procedures in the protocol
Risks related to participant rights
Risks to the reliability of trial results
Plans to mitigate or manage risks
trial management and monitoring plans
Determine the intensity and focus of monitoring
Risk proportionate approaches Joint working group (http://www.mhra.gov.uk/home/groups/l-ctu/documents/websiteresources/con111784.pdf)
Develop a risk assessment that gives practical guidance on risk adaptations.
Considered in relation to what is known about the IMP and standard medical care (Brosteanu et al. Clinical Trials 2009:585-596)
Identify lower risk trials where simplification is possible for regulatory approvals and trial conduct
IMP risk category and safety monitoring plan be submitted to the MHRA with the CTA
Type A: no higher than that of standard medical care
• Medicinal products licensed in any EU Member State if they relate to the licensed range of indications, dosage and form
• Off-label use if this off-label use is established practice and supported by published evidence/guidelines
Type B: somewhat higher than that of standard medical care Medicinal products licensed in any EU Member State
if : used for a new indication* (patient pop/disease group)• Substantial dosage modifications for licensed indication Used in combinations for which interactions are suspected
Medicinal products NOT licensed in any EU Member State if :The active substance is part of a medicinal product licensed in the EU
*Grading of Type A may be justified if there is extensive clinical experience with the product and no reason to suspect a different safety profile
Type C: markedly higher than that of standard medical care
Trials involving a medicinal product not licensed in any EU Member State
A grading other than TYPE C may be justified if there is extensive class data or pre clinical and clinical evidence
Guidance on focus and intensity of monitoringConcerns identified in the risk assessment with the design, methods or conduct of the trial (excl IMP) which remain after mitigations are in place
Risk assoc with IMP
No Yes
Type A Low intensityCentral monitoringNo requirement site visits
Low+Plus appropriate monitoring to address specific vulnerabilities
Type B Moderate intensityCentral monitoringTriggered visits
Moderate +Plus appropriate monitoring to address specific vulnerabilities
Type C Higher IntensityMore intense monitoring to include effective site visits
Higher+Plus appropriate monitoring to address specific vulnerabilities
PILOT:
http://www.netscc.org.uk/news/PDFs/MHRAPilotRept_FINALPUBL
ICATION.PDF
Empirical example from cancer*
Non-commercial cancer trial designed and initiated pre-2004 UK regulations
Parallel, open-label, multicentre (UK), phase III, superiority RCT comparing control chemotherapy with experimental chemotherapy
At the close of recruitment the drug company initiated 100% SDV of all data collected
All source verified data entered onto a ‘new’ database
Comparison of original data and source verified data
*Catrin Tudur Smith, Deborah Stocken, Paula Ghaneh, Trevor Cox, David Cunningham, John Neoptolemos
Strengths/limitations Strengths
Independent review of data
Independent database
Rare for 100% SDV to be performed in non-commercial trials
Limitations
Original ‘un-monitored’ data may not represent current practice
SDV may have changed trial conduct towards end of the trial
Outcomes Primary outcome
Overall Survival (OS)
Secondary outcomes
Progression Free Survival (PFS)
Objective Response
Serious Adverse Events
Overall survival
VariableDiscrepancies n (%)
Control Experimental Total(n=266) (n=267) (n=533)
Date of death all 21 (7.9) 22 (8.2) 43 (8.1)
Date of death excluding missing
5 (1.9) 8 (3.0) 13 (2.4)
Death status 15 (5.6)* 14 (5.2) 29 (5.4)
*1 additional patient had missing date of death in original data but date of death in SVD but death status agreed
Overall survivalNon-
monitored
data
Source
verified data
HR (95% CI)* 1.19
(0.99 to 1.42)
1.18
(0.99 to 1.41)
Number of
patients
533 533
Deaths 469 498
Log-rank
statistic
3.33 3.44
Log-rank p-
value
0.068 0.064
*HR>1 indicates benefit to E
Central monitoring for OS Discrepancies in death data – not clear whether SDV
accurate
Central collection of death data from ONS
Provides a ‘third’ data set for comparison
Central monitoring for OS 56 (11%) discrepancies between SDV and ONS date of
death
2 patients still alive in SDV but dead in ONS
2 dates were discrepant by 1 year
52 dates were discrepant by a few days
Central monitoring for OSNon-
monitored
data
Source verified
data
Central
monitored
data
HR (95% CI)* 1.19
(0.99 to 1.42)
1.18
(0.99 to 1.41)
1.18
(0.99 to 1.40)
Number of
patients
533 533 533
Deaths 469 498 499
Log-rank
statistic
3.33 3.44 3.22
Log-rank p-
value
0.068 0.064 0.073
*HR>1 indicates benefit to E
Overall survival
0. 00
0. 25
0. 50
0. 75
1. 00
t i me
0 200 400 600 800 1000 1200
STRATA: dat a=ONS Censor ed dat a=ONS
dat a=SDV Censor ed dat a=SDV
dat a=un- moni t or ed Censor ed dat a=un- moni t or ed
*HR>1 indicates benefit to E
Empirical example conclusions
SDV identified errors, BUT
Errors did not impact analysis of primary outcome OS
Central monitoring suggested possible errors in source verified data
Central monitoring (via ONS) more efficient and accurate in this example
Empirical example conclusions
SDV did impact RECIST response data and SAE data
Data collection difficult/subjective for these outcomes
More scope to introduce systematic error
Suggests a need to focus training research staff
Further investigation required
SDV resource intensive and may not necessarily provide error free data
End of trial ‘checklist’ of critical data to site staff may be an alternative more efficient approach for some data?
Current practice - CTTI
Electronic survey of over 200 organizations in academia, government, industry, and clinical research – 65 (30%) responders
• Clinical Trials Transformation Initiative (CTTI)
• On-site monitoring is routinely performed by industry and CROs but less frequently
and less extensively by academic coordinating center/cooperative group/government
organizations.
• The scope of on-site monitoring visits varies by organization
-some procedures, source documents and regulatory documents reviewed more often by CROs and
industry sponsors than by academic/cooperative group/government organizations.
Current practice - CTTI The rationale for using a specific monitoring approach does not appear
to be based on empirical evidence
little empirical evidence to determine which, if any, onsite monitoring practices lead to improved patient safety and data quality.
substantial lack of information to guide the planning of trials that include this commonly used and expensive quality assurance procedure.
... more research is needed to assess the potential impact of the variations in monitoring practices we observed.
Current practice – CTU survey
Norah Mawoko and Catrin Tudur Smith 2011
On-line survey of 48 UKCRC registered Clinical Trials Units
Only 19 (40%) complete responses
All 19 CTUs involved with CTIMPs
Monitoring plan: 37% all trials, 32% most trials, 32% some trials
Risk assessment to determine level of monitoring: 53% all trials, 16% most trials, 31% some trials
All CTUs used some level of central monitoring, either with or without on-site monitoring
Current practice – CTU survey
Current practice – CTU survey
Final remarks
Variation in monitoring practice across organisations
Evidence based guidelines for monitoring not available
Resource intensive approaches such as source data verification are often inefficient and unnecessary
Further empirical evidence required to evaluate alternative methods
Approaches to monitoring should be flexible and risk proportionate
Training in monitoring needs to reflect this
Monitoring workshop Clinical trial monitoring workshop is being developed
cross-hubs initiative (North West (lead hub), London, Bristol, Oxford, Leeds CTU)
collate and present a practical summary of current practice through consultation with UKCRC registered CTUs
focus on risk-based and central approaches to monitoring
include MHRA perspective
share best practice
establish a network of methodologists from hubs and CTUs with interest in monitoring