Quality Assurance and Quality Control in Clinical Research.pptx

Quality Assurance and Quality Control in Clinical Research

1. RESEARCH METHOTHODOLOGY COURSE

2. Quality Assurance and Quality Control in Clinical Research

3. Quality in Clinical Research What is quality Quality systems Quality Control Quality Assurance Responsibilities SOPs TMFs and site files Inspections

4. QUALITY in clinical research What is quality? Scientific integrity Robust protocol accurate, reliable data Clear specific aim / scientific question Well written, no ambiguity Comprehensive Meets legal / guideline /governance requirements Funded, resourced and insured Sensible recruitment

5. Quality in clinical research What is quality? Trustworthy procedures Standard operating procedures Reliable well maintained equipment No unnecessary risk to participants Safe procedures Medicine safety Reasonable burden Trained staff Informed participants

6. Quality Control QC definition – “The operational techniques and activities undertaken to verify that the requirements for the quality of the trial-related activities have been fulfilled” ICH GCP 1.47 QC in practice Monitoring Management oversight Steering committees Data / safety monitoring committees

7. Quality Assurance QA definition – “All those planned and systematic actions that are established to ensure that the trial is performed and the data are generated, documented and reported in compliance with Good Clinical Practice ICH GCP 1.46 and the applicable requirements” QA in practice Audits SOPs, policies, templates Inspections

8. Quality Assurance QA definition – “All those planned and systematic actions that are established to ensure that the trial is performed and the data are generated, documented and reported in compliance with Good Clinical Practice ICH GCP 1.46 and the applicable requirements” QA in practice Audits SOPs, policies, templates Inspections QA makes sure the QC is effective

9. Why do we need QA and QC? GCP requirement – “Systems with procedures that assure the quality of every aspect of the trial should be implemented.” ICH GCP Principle 2.13 Legal requirement for IMP trials, guidance for all other clinical research To minimise errors and detect mistakes

10. Why do we need QA and QC? A poor quality trial is A waste of time A waste of money A waste of resources An unnecessary burden to participants (unethical)

11. Responsibility for Quality ICH Section 5.1 – “The sponsor is responsible for implementing and maintaining quality assurance and quality control systems with written SOPs... “ Sponsors Should review systematically Will appoint monitors and auditors May appoint a QA Manager / Governance Lead Will delegate, therefore everyone involved in conduct of clinical trials must take responsibility for own work Imperial Clinical Trials Unit (ICTU)

12. Qua l ity Staff Recruitment of qualified/experienced staff Training ICH GCP on-the-job study / procedure specific Protocol Peer review/expert input – science, reports Statistician Imperial Clinical Trials Unit (ICTU)

13. Qua l ity Procedures SOPs, manuals, protocol, checklists, templates GLP/GCLP Equipment calibration / maintenance Appropriate suppliers Contracts and agreements High quality IMP GMP QP release Imperial Clinical Trials Unit (ICTU)

14. Standard Operating Procedures (SOPs) Detailed written instructions to achieve uniformity of the performance of a specific function (ICH GCP) A clearly written description of how a particular task is to be performed, who by, who is responsible, when and where needed.... Can be used to train staff Help to ensure consistency, compliance, accountability and efficiency

15. Standard Operating Procedures (SOPs) Typical sections Cover page with authorisations Introduction /Background Purpose (why) Scope (where and when) Responsibilities (who) Specific procedure (what) Associated documents – forms References Change history

16. Audit Defined in ICH-GCP – “ A systematic and independent examination of the trial to evaluate that activities were conducted, the data recorded, analysed, and accurately reported according to the protocol, sponsor’s SOPs, GCP and applicable regulatory requirements”

17. Audit QA Independent ‘snapshot’ examination by trained auditor (not by the monitor!) Processes, documentation, data Investigator sites Vendors (e.g. manufacturers of study drug) Laboratories Key documents Systems Imperial Clinical Trials Unit (ICTU)

18. Audit Audit is Systematic Independent Audit is against standards ICH GCP Protocol SOPs and policies Imperial Clinical Trials Unit (ICTU)

19. What is Monitoring? Part of QC The act of overseeing the progress of a clinical trial and ensuring that it is conducted, recorded and reported in accordance with the protocol, SOPs, GCP and regulatory requirements.

20. Monitoring Purpose – To ensure that the clinical trial results are reliable, i.e. treatment effects are real and unbiased Further information and detail to be covered in a later module

21. GCP Inspections Thorough review of the documents, facilities, and resources May inspect The sponsor and/or host organisation A trial A service provider e.g. Labs, CRO Inspections may be Routine Triggered MA driven

22. Critical and Major Inspection Findings (MHRA 2016-2017) Common Findings in Non-Commercial Studies

23. What is GCP? Good Clinical Practice is “an international ethical and scientific quality standard for designing, conducting, recording and reporting trials that involve the participation of human subjects” ICH Provides Assurance that the data and reported results are credible and accurate Assurance that the rights and confidentiality of trial subjects are protected

24. What is Clinical Research? Any research involving human subjects Includes Clinical trials using a medicinal product Questionnaire/interview studies Research on biological samples/tissues Healthy volunteers & patients NHS and non NHS

25. Research Governance “systems or regulations that ensure good practice and high quality at all stages of the research process” (RGF 2001,2005) Ensuring that researchers do the right things, in the right ways, at the right times. Not new legislation but brings existing regulations and guidance on clinical research into one document UK Policy Framework for Health and Social Care Research (October 2017)

26. NHS Research Governance Purpose To ensure the dignity, rights and wellbeing of the public To improve research quality To prevent poor performance and adverse events To help prevent misconduct and fraud Covers the entire research process design, management, undertaking, funding, hosting or participating in research

27. Annual Reports Two types of reports need to be submitted during the trial Annual Progress Report Development Safety Update Report At the end of the trial End of Study Notification Final report within 12 months

28. Archiving SOP on archiving Keep a log of all documents and where they will be located Ensure that the data and any software is retrievable Label medical notes to be retained Electronic records (and software/readers) need careful archiving

29. Archiving How long for? Local requirements e.g. 10 years after the end of the trial At least 3 years after the last marketing authorisation (if the trial contributes to that authorisation If a trial involves children until youngest participant reaches the age of 21 At least 30 years for Advanced Therapy Trials e.g. Gene Therapy, Cell Therapy WILL BE 25 YEARS when new regulations in force HTA = indefinitely

30. Adverse Events (AEs) Any untoward medical occurrence in a patient/subject administered a medicinal product, which does not necessarily have to have a causal relationship with this treatment. – Can be any unfavourable sign, symptom or disease SOP on JRCO website Record on AE form or log and the medical notes

31. Adverse Event Collection Questioning participants at each visit Abnormal lab results, ECGs Medical notes – any admissions or visits to other clinics, anything unexpected Concomitant medications Diary entries

32. Adverse Reactions (ARs) All untoward and unintended responses to a medicinal product related to any dose administered. There is a reasonable possibility the event is related; i.e. it cannot be ruled out.

33. Serious Adverse Events (SAEs) Any adverse event that: Results in death Is life-threatening 1 Requires inpatient hospitalisation or prolongation of existing hospitalisation Results in persistent or significant disability/incapacity Is a congenital anomaly/birth defect Medically important event 2 1 Participant was actually risk of death at the time of the event; not an event which hypothetically might have caused death if it were more severe 2 According to medical and scientific judgment

34. Serious ≠ Severe Severity Describes intensity e.g. mild, moderate, or severe e.g. severe headache Seriousness Serves as a guide for reporting (6 scenarios on previous slide)

35. Serious Adverse Reactions (SARs) An SAE where a causal relationship between a medicinal product and the SAE is at least a reasonable possibility ; i.e. the relationship cannot be ruled out. Causality must be assessed by a medically qualified investigator

36. Foreseeable SAE/SARs Often certain events can be predicted Can define in protocol as not requiring immediate reporting Disease related / pre-existing condition / population related e.g. elderly or serious medical condition (predictable SAEs) Drug related events (predictable SARs)

37. Serious Breach A ‘serious breach’, is defined as a breach which is likely to affect to a significant degree the safety or physical or mental integrity of the subjects of the trial; or the scientific value of the trial Medicines for Human Use (Clinical Trials) Amendment Regulations 2006

38. Serious Breaches Safety breaches Breach of GCP or the protocol leading to death, hospitalisation or permanent disability Eligibility violation IMP dosing error (if a risk to health) Failure to report SUSARs, SAEs etc. Urgent safety measure not implemented at all sites Fraud Persistent use of wrong Patient Info Sheet Systematic error in data recording / entry / processing

39. Urgent Safety Measure Can be implemented immediately without waiting for an amendment to be approved Inform the MHRA by phone Follow up in writing within 3 days What measures have been taken Reasons why Amendment form + covering letter

40. E xpermental design and informed consent

41. What is a clinical trial? Prospective assignment is the key issue (note also “one or more INTERVENTIONS”

42. Classification of Clinical Trials

43. Phase I, II, III, IV Clinical Trials – simple definition Human pharmacology Gross safety, Pharmacokinetics, possible Pharmacodynamics Therapeutic exploratory More safety evidence, early evidence of clinical efficacy, dose finding Therapeutic confirmatory Substantial evidence of efficacy and safety (risk- benefit) Therapeutic use Safety and/or efficacy in clinical practice

44. Why are we undertaking clinical trials? Costs of healthcare provision and medical services are high - need to ensure that the available resources are used wisely and efficiently This has led to development of the notion of “evidence- based” medicine: the selection of treatments and interventions based on the scientific evidence of the efficacy and safety of the treatment

45. Randomized, controlled trials – highest quality evidence The Cochrane Collaboration considers RCTs to be of most value in provision of evidence:

46. International Conference on Harmonisation Many of the ideas and principles we will discuss are contained in three important documents published by the International Conference on Harmonisation (ICH). ICH publishes guidelines for the design, conduct, analysis and reporting of clinical trials that will be used to support formal registration of a new drug treatment. Although not all clinical trials fall into this category, the ICH Guidelines are a good source of reference for all who work on clinical trials

47. BIAS in clinical trials A key objective of clinical trial design and analysis is to avoid inherent bias in the trial. bias is any systematic tendency to distort or exaggerate a treatment effect We want to avoid making erroneous conclusions about the effectiveness of a treatment and so we need reassurance that trials are unlikely to have any inherent bias either in the way we collect data, the way we organize the trial, and the way in which we analyze and report data

48. Key steps in trial design/analysis Identify the patient population of interest When we run a clinical trial, we are hoping to make an inference, based on the small sample of subjects, about a bigger broader POPULATION of subjects who might have been treated according to the trial protocol.

49. Key steps in trial design/analysis Select a sample of subjects from the POPULATION of interest We need to select a sample of subjects that is representative of the subject POPULATION

50. Key steps in trial design/analysis Choose an appropriate control group When we investigate an intervention of some kind, we will usually need to make a comparison of outcomes seen with this intervention with those see (or expected) under some appropriate control condition

51. Key steps in trial design/analysis Choose an appropriate trial design We will need to choose a trial design that meets the objectives of the trial, that fits in with any logistical or practical constraints, that avoids bias as much as possible, and that gives efficient estimates of the effects in which we are interested.

52. Key steps in trial design/analysis Record appropriate and relevant data We need to ensure that we record all relevant treatment outcome data, and also that we record all other data that might have an impact on the analysis and interpretation of the outcome data

53. Key steps in trial design/analysis Identify relevant hypotheses and statistical methods We need to choose appropriate and clinically relevant hypotheses, and choose appropriate statistical methods with which to investigate these hypotheses.

54. Key steps in trial design/analysis Analyze and report results We need to produce all relevant data summaries and analyses and report these in a comprehensible way according to standard reporting guidelines.

55. In a clinical trial of a new medical treatment, we are trying to draw valid and reasonable conclusions about what would happen if the new treatment were widely used in a large patient population But limitations of resource and practicality mean that we can only study and collect data from a small sample of patients, drawn from this larger population SAMPLES AND POPULATIONS

56. INFERENCE FROM SAMPLE TO POPULATION

57. We have two distinct requirements: To ensure that our small sample of data is genuinely “representative” of the larger patient population To ensure that we use the sample in a way that avoids error and misinterpretation HOW WE SELECT THE SAMPLE, HOW WE USE THE SAMPLE

58. The first requirement in avoiding bias is to try to ensure that we have a sample of patients which properly represents the wider patient population that we are interested in. Clearly if the sample is not representative of patients as a whole, the trial will be biased in respect of inferences about that wider population. “Representative Sample” From ‘Statistics in Drug Research: Methodologies and Recent Developments, by SC Chow and J Shao’ “For a valid statistical assessment of the efficacy and safety of a study drug, it is important that a representative sample of qualified patients be selected from the targeted patient population.”

59. The ideal technique for drawing a representative sample from a large population is to draw a random sample In random sampling, every patient in the population has an equal chance of being selected for the sample, and the sample is selected using a random process RANDOM SAMPLES

60. But doing this is logistically impossible in most clinical trials Organisationally we have to restrict ourselves to patients in certain centres, and to those patients who happen to present for treatment during the course of the trial So a key technique for ensuring representative samples is lost to us in clinical trials – IN PRACTICE WE CAN RARELY DRAW A RANDOM SAMPLE IN A CLINICAL TRIAL RANDOM SAMPLES – NOT USUALLY POSSIBLE

61. Two practical options: “sample frame” with random selection “Quota” sampling What other sampling methods are available?

62. If a restricted set of subjects is available – e.g. at certain centres only – consider constructing a list of all possible available patients and selecting randomly from all patients on that list e.g. use a list of all patients’ hospital numbers, and select randomly from that list This is better than just accepting the next available patients who attend the centre Sample Frame and Random Selection

63. Sample Frame and Random Selection

64. Sample Frame and Random Selection To generate a sampling frame of households with appropriate aged children for recruitment and gather basic demographic information about the target population, a census and brief survey was conducted from July to October 2004. From a database of all households enumerated in the census, a random list of households with at least one child less than 10yrs of age was generated.

65. If even a restricted random sample such as above cannot be selected, then consider potentially important subject characteristics e.g. age, sex, disease severity and ensure that the sample selected has representatives from each of the subject groups Even though not random, this goes some way towards giving a “representative” sample Quota sample

66. Quota sample

67. In practice, patients recruited into most clinical trials generally constitute a “convenience sample”: we use those subjects who happen to be available at a convenient place and convenient time Typically this means that we select the next group of patients that attend the center for a consultation This is not a random method, but has the practical advantage of simplicity and convenience BUT we usually use a “Convenience Sample”

68. The reason that random sampling is considered to be the ideal is that random sampling from a population guarantees that “on average”, the characteristics of the sample will match those of the broader population. In other words, random sampling gives us some reassurance about the representativeness of the sample – which other methods do not. Why is “random sampling” better?

69. But since random sampling isn’t usually possible, by convention - however we recruit the sample – we assume that the subjects are “representative” and we draw inferences from them as though they were a random sample we must always try to demonstrate that they are ”representative” – by good, clear and comprehensive documentation of the relevant characteristics of those subjects who are in the sample DOCUMENTING THE CHARACTERISTICS OF THE SAMPLE

70. the second possible source of bias in our sample – the way in which we use the sample in order to draw conclusions about the wider population We need to use our small sample (however it was selected) in a scientific and unbiased way…. AVOIDING BIAS WHEN WE USE THE SAMPLE

71. (ICH E9 Guideline) “the most important techniques for avoiding bias in clinical trials are blinding and randomization …these should be normal features of controlled clinical trials” Two Principles: Blinding and Randomization

72. (ICH E9 Guideline) “randomization introduces a deliberate element of chance into the assignment of treatments to subjects - it provides a sound statistical basis for evaluation of evidence and tends to produce treatment groups in which prognostic factors are balanced ” Randomization (of treatment allocation)

73. Randomized Study – balance of baseline characteristics Also tends to balance unknown/unrecorded characteristics

74. In it’s simplest form, this is just a process of choosing at random which treatment each patient should be assigned to; in practice, this is done by software. Can be done by generating a list of patient numbers, and a parallel list of randomly chosen treatment codes. As each patient is entered into the study, they are given the next available patient number on the list, and given the treatment specified on the list for that patient number. How do we randomize?

75. Patient 1 2 3 4 5 6 7 8 Treatment A B B A B B B A …etc. Randomization list

76. But if this process is done entirely at random, then there is a possibility that the numbers of patients assigned to the two treatment groups may become imbalanced more patients may be assigned to A than B (as in the above example), and in small trials especially this imbalance may make analysis difficult to interpret Randomization – may not work well in small trials

77. (ICH E9 Guideline) “there are advantages to be gained by randomising in blocks…block lengths should be short enough to limit possible imbalance, but long enough to avoid predictability” Randomization: blocking

78. In blocked randomization, random assignment of treatment is done within specified “blocks” or groups of patients such that within each block equal numbers of each treatment have been assigned, but in random order within the block Randomization: blocking

79. (ICH E9 Guideline) “details of block lengths should not be in the protocol” This limits the possibility of the investigator being able to predict the next treatment assignment from knowledge of previous assignments Block lengths may be varied within a given list as an additional safeguard Randomization: block length

80. This general principle of “structuring” randomization lists in order to avoid possible imbalances can be applied for other reasons For example, in multi- centre trials it is desirable that the numbers allocated to each treatment group should be roughly equal within each centre ….but simple randomization may not achieve this Randomization: structured lists

81. (ICH E9 Guideline) “stratify by centre … especially in small trials” i.e. generate a separate - blocked - randomization list for each centre, so that balance within each centre is maintained at least approximately Randomization: stratification

82. The same principle can also be applied to ensure that there is reasonable balance between treatment groups in respect of important prognostic characteristics of the patients Randomisation: balance of prognostic variables

83. (ICH E9 Guideline) “stratify by centre and by important prognostic variables … especially in small trials” This will often result in several levels of stratification within a trial: Randomization: stratification by prognostic variables

84. Generally not a good idea to try to stratify using too many factors – this may result in overall poor treatment group balance - especially in small trials No more than 3 or 4 factors - choose the ones that are of most importance in terms of possible influence on the outcome of treatment Randomization: stratification – do not over-stratify

85. Randomization: alternative technique - minimisation Minimisation is based on a different principle from randomisation. The first participant is allocated a treatment at random. For each subsequent participant we determine which treatment would lead to better balance between the groups in the variables of interest. Treatment allocation by minimisation. Altman DG, Bland JM. BMJ. 2005 Apr 9;330(7495):843.

86. Imbalance: different measures are possible Need to choose a method of measuring imbalance between groups, and various options are possible – some simple, some based on more complex statistical criteria (rule above probably too simplistic for use in practice) Protocol should specify method of measuring imbalance Element of chance should still be included e.g. assign next subject to preferred group with probability > 0.5, but not 1.0.

87. (ICH E9 Guideline) “A procedure by which one or more parties to the trial are kept unaware of the true treatment identities” “Blinding is intended to limit the occurrence of conscious or unconscious bias in conduct and interpretation of a clinical trial” Blinding (Masking) treatment allocation

88. Knowledge of treatment identity can influence either the patient’s or the investigator ‘s response to treatment, or their assessment of the response The three options are: neither patient nor investigator knows which treatment has been assigned the investigator knows but the patient does not both investigator and patient know the treatment identity Blinding - options

89. (ICH E9 Guideline) “In a double- blind study both the subjects and anyone involved in the treatment or clinical evaluation of a subject…remain unaware of the true treatment identity” “Double- blinding is the optimal approach” Double- blind studies

90. Double- blinding requires standardization of appearance of treatments and details of administration of treatments. This can sometimes be complicated and may in some instances be logistically or practically impossible. If so, then a single- blind study is the next best option. Double- blind studies – may not be possible

91. (ICH E6 Guideline) “double- blind studies are not always feasible….if not, then consider a single- blind study in which the subject is unaware of the true treatment identity” “even in a single- blind study it is important to try to avoid bias in treatment assignment….use a central randomisation service and ensure that the decision to enter a subject precedes knowledge of the treatment identity to which that subject will be randomised” Blinding: single- blind studies

92. If the investigator is unblinded, try to have the primary endpoint assessed by an independent, blinded, assessor The patient can often still be blinded provided that their knowledge of the treatment characteristics is limited Single- blind studies: blinded-assessor

93. Both the patient and investigator are aware of which treatment has been assigned This may influence both the patient’s response and the investigator’s assessment of the response If possible, use a completely objective measure of response Non-blinded study or “open- label” study

94. (ICH E9 Guideline) “The blind should only be broken when data are cleaned to an acceptable level of quality…any intentional or unintentional breaking of the blind prior to this should be reported” Important to implement procedures which will clearly reveal any early breaking of the blind If blind is broken at a centre, consider closing further recruitment at that centre to avoid completely any possibility of bias Breaking the blind

95. HIERARCHY OF CLINICAL STUDY DESIGNS RANDOMISED CONTROLLED TRIALS Other types of clinical trial Studies that are not clinical trials Within these categories, there are various possible study types, with no clear consensus on order in the hierarchy

96. STUDIES THAT ARE NOT CLINICAL TRIALS Cross- section surveys of subject disease states Qualitative descriptions of subject outcome Cohort Study Case-control study These are typically OBSERVATIONAL (i.e. we observe what happens, rather than prospectively assigning an intervention), often HISTORICAL (i.e. no prospective aspect), sometimes only qualitative (i.e. NO MEASURED OUTCOME).

97. EXAMPLES OF STUDIES THAT ARE NOT CLINICAL TRIALS

98. EXAMPLES OF STUDIES THAT ARE NOT CLINICAL TRIALS

99. EXAMPLES OF STUDIES THAT ARE CLINICAL TRIALS BUT NOT RCTs NON-RANDOMISED TRIALS: subjects are prospectively assigned to interventions, BUT NOT USING RANDOM ALLOCATION SINGLE ARM TRIALS: subjects are all prospectively assigned to a single intervention BIOMARKER STUDIES: prospective random allocation to one or more interventions, but the outcome measurement is not (demonstrably) a health outcome – e.g. measurement of a biomarker values of some kind where the biomarker has not been proven to be related to health outcome

100. RANDOMIZED, CONTROLLED TRIALS – THE GOLD STANDARD Randomised Controlled Clinical trials are NOT the only means of collecting such evidence but have become recognised as the “gold-standard”

101. (ICH E9 Guideline): “The most common design for a confirmatory trial is the parallel group design, in which subjects are randomised to one of two or more groups” RCT: PARALLEL GROUPS

102. Each group receives just one of the specified treatment regimes, and subjects are recruited in parallel into the various groups Trt A ------------- - Trt B ------------- - ------- > (time) (recording baseline measurements is a good idea) This type of study is the simplest to analyze and interpret, and is less likely to suffer from problems which may render the results unusable or difficult to interpret: it should be the preferred option for most trials EACH GROUP RECEIVES JUST ONE “REGIME”

103. Parallel trials can sometimes require large numbers of patients, and it may be important or essential in some cases to use an alternative which may reduce required patient numbers (ICH E6 Guideline) “cross- over studies usually reduce the numbers of subjects needed” because each subject acts as his/her own control and hence natural variability in responses is reduced for purposes of treatment comparison RCT: CROSS- OVER: REDUCES PATIENTS NUMBERS

104. (ICH E9 Guideline) “in a cross- over trial, each subject is randomized to a sequence of two (or more) treatments” (washout) Treatment A ---------- - Treatment B --------- - ------ > (time) CROSS- OVER: WASHOUT BETWEEN TREATMENTS

105. (ICH E6 Guideline) “cross- over studies have a number of problems that can invalidate results…the chief difficulty is carry- over of treatment effects from one study period to the next” CROSS- OVER – POSSIBLE PROBLEMS – CARRY-OVER EFFECTS this problem is especially acute in two-period cross- over designs, because carry-over effects cannot be distinguished statistically, and because tests for carry- over generally have low power

106. (ICH E6 Guideline) “avoid carry- over procedurally - ensure washout period is long-enough in relation to half-life of drug” washout periods should be long in relation to the half- life – how long depends on the drug and the indication Do not forget about duration of pharmacodynamic effects CARRY-OVER EFFECTS: TRY TO AVOID BY USING LONG WASHOUT

107. Cross- over trials can be extended in various ways: e.g. 3 period/3 treatment design: Period: 1 (wo) 2 (wo) 3 Group 1: A B C Group 2: B C A Group 3: C A B Baseline measurements prior to each treatment are recommended FOR ALL CROSSOVER DESIGNS CROSS- OVER: 3 TREATMENTS, 3 PERIODS

108. Or designs in which each patient does not receive all treatments e.g. Period: 1 2 Patient 1: A B Patient 2: B C Patient 3: C A Etc… CROSS- OVER: 3 TREATMENTS, 2 TREATMENT PERIODS

109. Or designs in which each patient receives a treatment more than once: e.g. Period: 1 2 3 Patient 1: A B A Patient 2: B A A Patient 3: B A B Etc… CROSS- OVER: 2 TREATMENTS, 3 TREATMENT PERIODS

110. And designs in which each patient receives the same sequence of treatments: e.g. Period: 1 2 Patient 1: A B Patient 2: A B Patient 3: A B Etc… CROSS- OVER WITH FIXED- SEQUENCE Commonly used for drug interaction studies

111. Prior to Treatment On Treatment Patient 1: X A Patient 2: X A Patient 3: Etc... X A “PAIRED” DATA ESSENTIALLY HAVE THE SAME DESIGN i.e. a simple study of prospective assignment of subjects to a single intervention, with comparisons of the health- outcome made before/after treatment, is essentially a type of “degenerate” crossover design

112. The more complex cross- over designs can offer substantial advantages in terms of the different effects that can be estimated, but they need expert statistical input to design and analyse, as some allocations of treatments sequences are more efficient than others CROSS- OVER: COMPLEX DESIGNS NEED CAREFUL PLANNING

113. Second issue with cross- over trials is their suitability only for certain types of indication: (ICH E6 Guideline) “ensure that the disease being studied is stable, chronic, and returns to baseline state during washout - ” (i.e. no long term trends over time in disease severity) CROSS- OVER – POSSIBLE PROBLEMS - TIME EFFECTS

114. Further issue is that cross- over trials often have to last longer than parallel trials, which increases the possibility that subjects will drop out of the trial before completion. The need to replace withdrawn subjects may diminish the main advantage of the crossover – the need for fewer subjects Longer trials may also impose logistical problems CROSS- OVER – POSSIBLE PROBLEMS – DROP- OUT

115. Choosing a control group (and some notes on endpoints)

116. CONTROL GROUP: AFFECTS ETHICS AND SCIENTIFIC CREDIBILITY

117. MAIN OPTIONS FOR CONTROL GROUPS

118. “NO-TREATMENT” CONTROL:

119. PLACEBO CONTROL i.e. use of PLACEBO is primarily a matter of blinding treatment allocation

120. DOSE- RESPONSE CONTROL: USING MORE THAN ONE DOSE Either the lowest dose functions as a control, or the demonstration of a dose- response relationship provides evidence of efficacy

121. ACTIVE CONTROL: WHEN A KNOWN ACTIVE TREATMENT EXISTS Ethically necessary if an effective treatment exists for a serious disease: may occasionally be ethically acceptable to omit for short periods in mild disease

122. EXTERNAL CONTROL: PATIENTS FROM OUTSIDE THE TRIAL

123. MULTIPLE CONTROLS: ESPECIALLY PLACEBO AND ACTIVE Common in equivalence studies: placebo included so that we can demonstrate that new treatment is BETTER than placebo, and active control included so that we can demonstrate that (a) it is better than placebo and (b) the active control and the test are «equivalent»

124. SOME NOTES ON ENDPOINTS Choice of endpoint in a clinical trial should be based on clinical relevance Primary and secondary endpoints should be specified in clinical trial protocols: some flexibility is permissible in exploratory trials, but pre-specification is essential in confirmatory trials Measure the endpoints at baseline if possible: and consider taking several such measurements to get a good average value Take baseline measurements in crossover studies prior to each treatment period

125. SOME NOTES ON ENDPOINTS Record values of relevant prognostic factors at baseline to allow assessment of baseline comparability and possible influence on outcome. Do not use statistical significance tests to compare patients groups at baseline with respect to prognostic factors: this is meaningless in RCTs and misleading in non-randomised trials. Effects of baseline prognostic factors must be evaluated by including these factors in statistical models at analysis.

126. SOME NOTES ON ENDPOINTS When planning repeated measurements of an endpoint during the course of a trial, consult a statistician – the optimum number and timing of repeated endpoint measurements depends on the expected trend over time. Note that repeated measurements of an endpoint does not necessarily give the trial more power or make the trial more informative – endpoints measured at a single, clinically-relevant timepoint may work just as well.

127. Questions? Thank you For more information Dt- Raneem @ Raneem_AM