Course: SPPH 512: Design and Analysis of Clinical Trials

Instructor: Dr. Joel Singer

Time: Tuesdays 9:30-12:30

Place: LPC B-108

Readings: At the suggestion of the School, readings will be distributed via

the web rather than on paper. This actually allows me to post additional
readings which may be of relevance to particular class members.

Contact hours: You can speak to me on Tuesdays after class or you can get
in touch with me at my office. I can be contacted either on CANVAS or via
email [email protected], and we can meet by phone or zoom.

Cell: 604-803-6949
Overview

This course covers areas of relevance to clinical trials, from their

organization, to their design, statistical evaluation, interpretation,
summarization and ethics.

Each session involves an interactive lecture with group problem solving to

test the students’ understanding of the issues being discussed.

Course evaluation involves a take home mid-term focused on issues

dealt with in class (25%), the design of a clinical trial in grant proposal
style (55%) and the critique of a classmate’s trial (20%).

Students must have an understanding of basic statistics and would benefit

from an introductory epidemiology course although it is not mandatory. The
handling of statistical issues in this course will be more from a conceptual
level and does not require manipulation of formula.
SPPH - 512

1. Prerequisites

It is assumed that you have a basic statistics course such as HSPPH 400. The analytical
sections of this course are heavier on understanding the methodological perspective of
different types of analyses and analytical issues than on the actual mechanics.
Nevertheless, it will be difficult for anyone not having a basic understanding of
elementary statistics.

In addition, a previous course in clinical epidemiology (SPPH 513) is recommended

though not necessary. An understanding of clinical epidemiology would enhance an
understanding of the advantages and disadvantages of randomized trials compared to
other designs (we will deal with this in at least some detail, also covered in SPPH 502) as
well as convey some of the measurement issues that must be taken into account when
designing a randomized trial (ie biases that can affect a trial, the limitations imposed by
diagnostic inaccuracy, the lack of clinical agreement among clinicians and the need for
objective outcomes that are reliable). All of these will be covered to some degree out of
necessity.

2. Basis for Evaluation

A. End of Term Assignment (55% of grade)

Your major assignment is to develop a full protocol that would be suitable for a funding
agency. It need not have a budget but it should include a general organizational structure
(how will the work be done, who will do it etc). This must be the type of study for
which a randomized design is appropriate. Preferably, this would be a study with
clinical endpoints although I have allowed students to work on more preliminary
studies with preclinical endpoints. A pilot study is not appropriate. It may be the
case that some relevant information is required from a pilot (eg the proportion of patients
achieving outcome in the control condition when a binary outcome is being used, or the
standard deviation when a continuous outcome is being used). You can go on the
assumption that in reality you would first do a pilot, but for your assignment make a best
estimate of the parameters you require from the existing literature.

You must get approval for your topic from me.

I have provided you with two examples of protocol outlines (see folder:
Reference Material), one which was used as a template by CIHR, and a second
guide developed by the Canadian HIV Trials Network for internal submissions.
Although the HIV Trials guidelines were developed specifically for clinical trials of
drugs, they can be generalized to evaluation of other types of manoeuvres (eg surgical,
psychosocial etc). Obviously, the sections devoted to "Drug Information" (section 3.2.3)
would not apply in non-drug trials. In fact, much of section 3.2.3. can be included in the
section on rationale. Note that CIHR has changed their template and has
decreased the allowable length of proposals, but we will stick with the longer
length to allow for a fuller explanation of methodological details. I have also
provided the SPIRIT publication, which goes through, in a more detailed fashion,
everything that should go into a protocol for running a study. This goes beyond
what one can usually include in a grant proposal. I have also included a recently
published paper on suggested Statistical Analysis Plan (SAP) guidelines. Again,
this goes into considerably more detail than you could ever do in a protocol, but
regulatory demands are increasing, and an SAP, even if not absolutely required
now, is highly advisable, both for RCT’s and for other types of studies.

The key is that these templates summarize the topics with which any protocol
must deal. From this perspective, they act as reminders about what should be
included in an RCT research proposal. Just make sure to cover the relevant
areas. Obviously, there may be questions that are not directly relevant to your
particular protocol (eg questions about drugs which would not apply to a non-
drug protocol), and in that case, you can omit certain sections.

Please also include as appendixes:

• a list of the personnel and their responsibilities,
• a study coordination plan including a schedule of events (ie once funding is
approved, how long will it take to hire staff, to print forms, to train study nurses,
as well as
• a schedule of patients visits and tests relative to patient's date of study entry).

You are NOT required to provide:

A budget
An informed consent letter

If possible, you should seek advice from an expert in the clinical evaluation area on
which you are focussing to ensure that your proposal is feasible. I would also suggest
that you begin thinking about the size of study sample you may require. In the past, a
number of students were forced to abandon projects when they realized the large number
of patients/subjects that would be required to evaluate their study hypothesis.

I will also provide you with access to copies of protocols and critiques submitted by
former students so that you have a good feel for what is expected of you. Although the
guidelines are developed specifically for clinical trials of drugs, they can be generalized
to evaluation of other types of manoeuvres (eg surgical, psychosocial etc). Obviously,
the sections devoted to "Drug Information" (section 3.2.3) would not apply in non-drug
trials. In fact, much of section 3.2.3. can be included in the section on rationale.
Class time will best be spent discussing and applying the readings for that day to specific
problems in your own protocols. It is not mandatory but for your own benefit you should
submit (i.e. email) at least a general overview of your study by the middle of February-
this would include:

• The study hypothesis

• Brief rationale (why is the question important and why should this intervention be
addressed)
• The study population to whom the intervention will be applied
• The primary outcome that will serve as the basis of comparison for your study groups
• The expected duration of the study (eg if you expect that it would take 2 years to
recruit all patients and you want a minimum follow-up period of 1 year per patient, then
the duration would be 3 years). You may not be able to complete this section without
some knowledge of sample size calculation.

I am quite willing to give you feedback on your proposal as it develops, and students
who took advantage of this opportunity in previous years seemed to benefit in terms of
their study designs.

B. Study Critique (20%)

Each student will be responsible for doing a critique of the study proposal of one of their
classmates (I will assign the pairings). I will post copies of what I consider to be
excellent critiques both in terms of content and presentation. It is the responsibility of
each student to ensure that the person doing the critique has a copy of the protocol on the
appointed day (schedule found below).

C. Midterm Exam (25%)

There will be a take home midterm examination to be handed in the week after the winter
break. Basically, you will be asked to apply the concepts we have discussed in class to
various problems.

3. Class Structure
The structure of the classes will depend somewhat on the topic. For the more statistically
oriented sessions where there is a greater need for guidance, at least part of the session
will be "lecture-style" interspersed with class interaction. For other sections, we will
begin as a whole group to discuss some of the essential elements of the readings but may
split into smaller groups so that each of you can discuss particular aspects of your
protocol. We will then reconvene as a larger group to discuss some of the problems that
the small groups have confronted. The readings for each week are usually accompanied
by an orientation sheet which sets out the key issues for that set of readings. I will
attempt to summarize the issues and review where we have gotten at the beginning and
end of each class.

4. Course Objective:
To become aware of the major issues in the design and analysis of clinical trials and to be
able to apply them to your own project. Most clinical trials are collaborative efforts, and
it assumed that most trials would involve a methodologist/statistician, and clinical
experts. Thus, it is not expected that you would necessarily be able to do all analyses
yourself but you should understand the key issues in the analysis of trials and be able to
apply basic statistics.
SPPH 512 2022

Topics by week

Session 1 Jan. 11 Organizational Issues in Clinical Trials

Session 2 Jan. 18 Design Issues/Defining the Question
Session 3 Jan. 25 Population: Impact on design and interpretation/ Recruitment
Session 4 Feb. 1 Baseline Evaluation/Stratification/Blinding
Session 5 Feb 8 Sample Size/Equivalence (take home midterm to be
returned Mar 1, 930 pm)
Feb 15 reading week
Session 6 Feb 22 Outcomes

Session 7 March 1 Quality of Life

Session 8 Mar 8 Multiplicity
Session 9 Mar. 15 Interim Analysis
Session 10 Mar. 22 Crossover Design/Methods for reduction of variance
Session 11 Mar 29 Survival/Multivariate Models
Session 12 Apr 5 Ethics/ Cluster Randomized Trials

Protocol to be submitted to instructor and reviewer via email on April 12; critique
due via email on Apr. 19
Protocol Outlines

I have provided you with two examples of protocol outlines (see folder:
Reference Material), one which was used as a template by CIHR, and a second
guide developed by the Canadian HIV Trials Network for internal submissions.
Note that CIHR has changed their template and has decreased the allowable
length of proposals, but we will stick with the longer length to allow for a fuller
explanation of methodological details. I have also provided the SPIRIT
publication, which goes through, in a more detailed fashion, everything that
should go into a protocol for running a study. This goes beyond what one can
usually include in a grant proposal. I have also included a recently published
paper on suggested Statistical Analysis Plan (SAP) guidelines. Again, this goes
into considerably more detail than you could ever do in a protocol, but regulatory
demands are increasing, and an SAP, even if not absolutely required now, is
highly advisable, both for RCT’s and for other types of studies.

The key is that these templates summarize the topics with which any protocol
must deal. From this perspective, they act as reminders about what should be
included in an RCT research proposal. Just make sure to cover the relevant
areas. Obviously, there may be questions that are not directly relevant to your
particular protocol (eg questions about drugs which would not apply to a non-
drug protocol), and in that case, you can omit certain sections.