Designing a Clinical Research Protocol

1
Designing a Clinical Research Protocol

• Matthew S. Mayo, Ph.D.
• Director, Center for Biostatistics and Advanced
  Informatics
• Professor, Preventive Medicine and Public Health

2
Organization and Planning

• An essential part of planning a clinical trial is
  writing a study protocol.
• A study protocol is a formal document that
  details how the trial is to be conducted.
• Three fundamental aspects of trial design which
  must be precisely defined early in the planning
  process
• Which patients are eligible
• Which treatments will be evaluated
• How will patient response be assessed and
  measured

3
Protocol

• The evolution of a protocol from its initial
  version to the final document creates a
  systematic approach to the development of a
  clinical trial that is acceptable on ethical,
  scientific, and organizational grounds.
• The final version of a protocol should serve two
  primary functions
• Provide detailed specifications of the trial
• procedure relating to each individual patient
• Describe the trials motivational background,
• specific aims and rationale for the design

4
Protocol

• Main Protocol Features
• Background and general aims
• Specific objectives
• Patient selection criteria
• Treatment schedules
• Methods of patient evaluation
• Trial design
• Registration and randomization of patients
• Patient consent
• Required size of study
• Monitoring of trial progress
• Forms and data handling
• Protocol deviations
• Plans for statistical analysis
• Administrative responsibilities

5
Protocol

• Background and general aims are useful to explain
  why the trial is considered worthwhile and how it
  builds upon previous research.
• Specific aims are concise hypotheses concerning
  treatment efficacy and safety that are to be
  examined by the trial.
• Expansive descriptions of patient selection
  criteria, treatment schedules, and methods of
  patient evaluation for the bulk of the protocol
  and ensure that the trial adheres to well-defined
  objectives.

6
Protocol

• Trial design should include choice of control
  group, the method of treatment allocation, and
  procedures of blinding participants and/or
  researchers.
• Registration and randomization procedures detail
  the sequence of events that each patient will
  undergo to enter the study and be randomized.
• Patient consent process should be detailed prior
  to receiving treatment.
• Size of the trial along with a statistical
  rationale behind the chosen number should be
  stated.

7
Protocol

• Monitoring of trial progress should be detailed
  especially if interim analyses or multiple stage
  designs are proposed.
• The process for completing forms and managing
  data should be stated and detailed.
• Procedure for handling protocol deviations such
  as compliance, dose modifications, early
  withdrawals and loss to follow-up should be
  specified in the protocol.
• A detailed statistical analysis plan that follows
  directly with the specific objectives of the
  study should be given.
• Administrative responsibilities should be clearly
  stated.

8
Administration, Staff and Finance

• Sources of Funding
• Pharmaceutical Companies
• National Health Organizations
• Local and Regional Organizations
• Coordination and Leadership
• Principle Investigator
• Coordinating Center
• Monitoring Committee

9
Administration, Staff and Finance

• Informed and Enthusiastic Participants
• Role of a Statistician
• Collaborating Scientist
• Study Design
• Study Monitoring
• Final Analyses

10
Selection of Patients

• Source of patient recruitment
• Disease state under investigation
• Specific Inclusion/Exclusion Criteria

11
Treatment Schedules

• Drug Formulation
• Route of Administration
• Amount of Each Dose
• Frequency of Dosage
• Duration of Therapy
• Side-effects, Dose Modifications and Withdrawals
• Patient Compliance with Therapy
• Ancillary Treatment and Patient Care
• Packaging and Distribution of Drugs
• Comparison of Treatment Policies

12
Evaluation of Patient Response

• Baseline Assessment
• Initial Clinical Condition
• Demographics
• Clinical History
• Principle Criteria for Response
• Survival
• Tumor Response
• Immune Response
• Cessation
• Quality of Life
• Toxicities
• Side-effects

13
Evaluation of Patient Response

• Patient Monitoring should be detailed within the
  protocol
• Accuracy of Evaluation Data
• Factual Information date of birth
• Measurements blood pressure
• Clinical Assessments depression scale
• Patient Opinion pain scores
• Blinded Evaluation
• Frequency of Evaluation
• Follow-up Studies

14
The Justification for Randomization

• The first randomized experiments were in
  agriculture where the experimental units were
  plots of land to which the treatments were
  assigned in a random manner.
• The purpose of randomization was
• To guard against any use of judgment or
  systematic arrangements leading to one treatment
  getting plots with poorer soil
• To provide a basis for the standard methods of
  statistical analysis, such as significance tests
• The idea of random patient allocation is not
  intuitively appealing to the medical profession
  or the layman.

15
Problems with Uncontrolled Trials

• When a new treatment comes to be, an adventurous
  and enthusiastic investigator may initially study
  this new treatment without any direct comparison
  to a similar group of patients on more standard
  therapy.
• Uncontrolled trials have the potential to provide
  a very distorted view of therapy. Example
  Laetrile. 70,000 patients had tried Laetrile of
  which 93 were submitted for evaluation and 6
  achieved a response.
• Uncontrolled phase II trials in oncology lead to
  varying reported response rates on nearly
  identically designed studies. Why?
• Uncontrolled studies are more likely to lead to
  enthusiastic recommendation of the treatment
  compared with properly controlled trials.

16
Problems with Historical (Hysterical) Controls

• Most common way to avoid randomization is to
  compare future patients on new treatment to those
  previously studied under similar conditions.
• Flaws
• Patient selection
• Historical control group is less likely to have
  clearly defined criteria for patient inclusion
• Historical controls were recruited earlier and
  may come from a different population
• Investigator may be more restrictive in his
  choice of patient for new treatment

17
Problems with Historical (Hysterical) Controls
(Contd)

• Experimental Environment
• Quality of the recorded data is generally
  inferior for historical controls
• Criteria for response may differ between the two
  groups
• Ancillary patient care may improve on the new
  treatment
• Tendency to invalidate more patients on a new
  treatment than in historical controls
• Studies with historical controls tend to
  exaggerate the value of a new treatment

18
Problems with Historical (Hysterical) Controls
(Contd)

• To get around this, many try retrospective
  adjustment with prognostic factors for trials
  with historical controls, which in itself has
  issues
• Quality of historical data may be poor and
  reporting of prognostic factors may not be
  consistent
• What factors are actually prognostic may not be
  completely known
• Prognostic factors adjust for patient selection,
  however bias due to changes in experimental
  environment will remain

19
Problems with Historical (Hysterical) Controls
(Contd)

• Analysis techniques may be complex and difficult
  to explain and/or interpret clinically
• To propose that poor design can be corrected by
  subtle analysis techniques is contrary to good
  scientific thinking
• Many studies have limited number of patients that
  enhances a researchers desire to not randomize,
  however treatment allocation need not be 1 to 1
  if these concerns are great.

20
Problems with Concurrent Non-Randomized Controls

• Two possible approaches that are utilized for
  non-random allocation of controls are
• Systematic assignment
• Assign by date of birth, date of presentation,
  alternate assignment
• These methods do have a potential for bias that
  can be eliminated by randomization
• Judgment assignment
• Use of judgment is unacceptable, potential for
  bias too great

21
Is Randomization Feasible?

• First one must determine if it is ethical to
  randomize patients and then if sufficient
  investigators and patients will buy into this
  type of design.
• It has been argued by Hill (1963) that if
  superiority has not been established then it is
  unethical not to randomize.
• Investigators and layman must be able to draw a
  distinction between subject personal belief and
  objective scientific knowledge regarding safety
  and efficacy.

22
Blinding and Placebos

• If the patient or those responsible for
  administering the treatment and/or evaluating
  response no the treatment allocation there is a
  high potential for biased treatment comparisons.
• If it is feasible to conduct a double-blind trial
  then one should be conducted.

23
Justification for Double-Blind Trials

• The Patient
• Patient knowledge of being on treatment may be of
  psychological benefit
• Patient knowledge of being on placebo may result
  in unfavorable responses, such as non-compliance,
  withdrawal, etc.

24
Justification for Double-Blind Trials (Contd)

• The Treatment Team
• Knowledge of treating physician may alter course
  of therapy, including dose modifications,
  intensity of examination, other meds, etc.
• Ancillary care can be influenced if study nurses
  know treatment

25
Justification for Double-Blind Trials (Contd)

• The Evaluator
• Evaluators may err to more favorable evaluation
  for those on treatment versus placebo, or vice
  versa
• May need to blind person to other variables that
  may lead to knowledge of treatment regimen
• Knowledge by evaluators limits their ability to
  maintain objectivity

26
Justification for Double-Blind Trials (Contd)

• Value of Placebo Controls
• When no standard of care exists placebo controls
  help reduce bias
• Placebo-effect is eliminated
• Placebo versus active control, if active control
  has shown efficacy then many times it is
  unethical to use a placebo
• Placebos help make patients attitudes to the
  trial as similar as possible

27
The Conduct of Double-Blind Trials

• Matched Placebos
• Arrange for the manufacture of an oral placebo
  which is identical in every respect to the active
  medication (color, taste, texture, shape,
  packaging, etc.)
• Determine mode of therapy (capsule, liquid,
  tablet)

28
The Conduct of Double-Blind Trials (Contd)

• Coding and Randomization
• Each package must have a unique coding system
  that the statistician and pharmacist utilize
• That code is recorded in the patient forms

29
The Conduct of Double-Blind Trials (Contd)

• Breaking the Code
• Must be done prior to reporting results, may be
  done prior to analysis, not recommended
• Interim analysis and reports to DSMBs may
  require breaking of the code
• Severe adverse events will require breaking of
  the code
• Clinician determination for need of additional or
  alternative treatment requires breaking of the
  code

30
The Conduct of Double-Blind Trials (Contd)

• Other Types of Double-Blind Study
• Slow release versus twice daily, Group 1 Pill A
  Slow release, Pill B Placebo versus Group 2
  Pill A conventional tablet, Pill B
  Conventional tablet
• Saline IV versus IV chemotherapy
• Sham surgery, has been done but is rare

31
When is Blinding Feasible

• The need for blinding and randomization are two
  separate issues.
• Blinding should be based on the following
  considerations
• Ethics double-blind procedure should not result
  in any harm or undue risk
• Practicality some treatment may be impossible
  to arrange in a double-blind manner
• Avoidance of bias need to make an assessment of
  the potential bias
• Compromise partial blinding may be sufficient

32

• DOUBLE-BLIND TRIALS
• REQUIRE EXTENSIVE TIME
• AND EFFORT!!!