The impact of (some)
patient and site factors
on assay sensitivity
Robert H. Dworkin, PhD
Professor of Anesthesiology, Neurology, Oncology, and Psychiatry
Professor, Center for Human Experimental Therapeutics
University of Rochester School of Medicine and Dentistry
Patient factors
Pain duration: minimum
Pain duration: maximum
Baseline pain intensity: minimum
Baseline pain intensity: maximum
Baseline pain variability
Baseline pain consistency
Baseline diary compliance
Prior treatment failure
Psychopathology
Subject training
Considerations for improving
assay sensitivity
Study design factors
Research design
Number of treatment groups
Active comparator
Initial dose titration period
Flexible vs. fixed dosing
Rescue medication
Concomitant analgesics
Baseline period duration
Study duration
Considerations for improving
assay sensitivity
Study site factors
Number of sites
Staff training
Staff-patient interactions
Geographic location
Patient referral sources
Accelerated enrollment
Protocol concealment
Infrastructure
Considerations for improving
assay sensitivity
J Clin Psychopharmacol 2007;27:1-5
Effect size as a function of interview quality:
HAM-D ratings in major depressive disorder RCTs
mean difference vs pbo
SES
p
0.46*
1.33
0.61
0.02
Study 11
all raters
high quality
0.5
6.8
Study 22
all raters
high quality
2.14
5.99
0.20
0.01
*standardized effect size of SSRIs from meta-analysis.
1Kobak
KA, et al. Interview quality and signal detection in clinical trials. Am J
Psychiatry 2005;162:628.
2Cogger
KO. Rating rater improvement: a method for estimating increased effect
size and reduction of clinical trial costs. J Clin Psychopharm 2007;27:418-420.
Primary efficacy endpoint:
daily pain diary
Select the number that best describes your neuropathic pain
during the past 24 hours. (Circle one number only)
0
No
pain
1
2
3
4
5
6
7
8
9
10
Worst
possible
pain
Patient factors
Baseline pain
intensity: minimum
Study design factors
Baseline period duration
Study site factors
Staff training
Baseline pain
intensity: maximum
Staff patient interactions
Baseline pain
variability
Protocol concealment
Baseline pain
consistency
Baseline diary
compliance
Subject training
Can baseline pain diaries be used to
identify patients who will provide
increased assay sensitivity?
“Individuals with a greater pain variability index at baseline
were more likely to be responders ... to placebo ...
suggesting that a high pain variability may be a predictor of
a placebo response.”
Harris RE, et al. Arthritis Rheum 2005;52:3670-3674.
“Because of the larger placebo response for pain-related
endpoints … in the high variability patients, the apparent
treatment effect size of MLN was smaller in patients with
higher pain variability.”
Palmer RH, et al. International Conference on Accelerating the Development of Enhanced
Pain Treatments, Bermuda, March 2011.
“In duloxetine studies for 3 chronic pain conditions (DPNP,
CLBP, OA), patients with higher baseline pain variation had
a smaller effect size and/or lower treatment response
(relative to placebo) compared with patients with lower
baseline pain variation.
Zhang S, et al. International Conference on Accelerating the Development of Enhanced
Pain Treatments, Bermuda, March 2011.
Selecting subjects based on
characteristics of their baseline pain
ratings that might reflect rating
“diligence” or “competence” is likely to
be less efficient than providing training
before the baseline so that fewer
subjects will be excluded because of
their baseline ratings.
Patient selection
Patient training
Patient training:
1. Educating the patient that it is very important
that pain ratings be made in a conscientious and
thoughtful manner because this is necessary for
the study to succeed in determining whether or
not the treatment is effective.
2. Identification of personal anchors and use of
reminders for rating pain in a consistent manner
throughout the study.
3. Educational scripts and interviews
specifically designed to manage patient
expectations.
“Though it is natural to hope for a positive
outcome during the trial, we want you to be aware
that there is a good likelihood that you are on
placebo during this trial.
Though you are likely to appreciate the care that
you get from the study staff in this trial, it is very
important that you don’t tell us that you are better
if you’re really not, just because you are
appreciative of the study staff, and you feel that
you will be letting them down if you don’t improve.
We need you to report your condition as
accurately as you possibly can...”
—D.L. Zimbroff, 2001
“Both investigators and patients were
blinded to the following information: entry
criteria for patients’ pain intensity, baseline
pain intensity, definition of responder
groups, visit at which randomization
occurred, treatment during the withdrawal
phase, efficacy failure criteria, and
computation rules and time windows in the
IVRS system used to calculate the baseline
intensity and pain response.”
Hewitt DJ, et al. Pain 2011;152:514-521.
Baseline score inflation in psychiatry
“The HAM-D ratings completed by the on-site
evaluator were significantly higher than the
centralized raters’ scores at screening and baseline,
but not at endpoint.
At screening, 36% of patients who were judged to
be eligible for the study by the on-site evaluator (ie,
they scored at least 17 on the HAM-D, the study’s
minimum severity criterion) were rated as study
ineligible by a centralized rater….
…lower baseline scores are associated with greater
change with placebo.”
Patient education:
1. Missing data diminishes the scientific value of the study
and of the other subjects’ participation.
2. Emphasize importance of outcome assessments even if
subject decides to discontinue study treatment and visits,
although consent can, of course, be withdrawn at any time.
Internal validity vs. external validity
(i.e., generalizability)