Transcript April 2010 Who Was Studied? - HIV Research Catalyst Forum

WHAT IS THIS
TELLING ME?
Robert J. Munk, PhD
AIDS InfoNet / ATAC
http://www.aidsinfonet.org
A project of the
New Mexico AIDS Education and Training Center at the
University of New Mexico Health Sciences Center
April 2010
Too Much Information!
Google search results 4/7/2010:
AIDS: 95 million results
HIV: 50 million results
April 2010
Too Many Sources!
• Health web sites
• AIDS web sites
• Ads on search pages
• Facebook
• Twitter
April 2010
Conflicting Information
• Antioxidants and the Heart: good or
bad?
• Start treatment early! Or - delay for
reasons of side effects, cost?
• d4T: “least” toxic nuke? Or the
worst?
April 2010
Research Papers
• Less hype
• Better info
– Author affiliations, contact info
– Financial support
– Conflicts of Interest
• Journal reputation / peer review
April 2010
Abstract
•
•
•
•
Background
Methods
Results
Conclusions
Abstract: a “teaser” for full article or
study. Do you want to read it all?
April 2010
Abstract 124, CROI 2010
Rates of Cardiovascular Disease
following Smoking Cessation in
Patients with HIV Infection: Results
from the D:A:D Study
• Author list (10 people, “and the
D:A:D Study Group”)
• Author affiliations
April 2010
Background
• Why should I care?
“Smoking rates are high in many
populations. Most smokers can’t stop. In
HIV negative people, substantial drops in
heart disease within 1 to 2 years of
stopping smoking. This study estimated
CVD event rates after HIV+ stopped
smoking.”
April 2010
Methods
• How was the study done? Who was
studied?
“Pts. included if in D:A:D study, no CVD
before enrollment. Noted current
smoking, how long since stopping, and
heart disease events. Calculated rates
for never smoked, stopped before study,
stopped during study, or still smoke.”
April 2010
Methods (Continued)
• How was the study done? Who was
studied?
“Incidence rates calculated. Adjusted for
age, sex, cohort, calendar year, ARV
treatment, diabetes, family history of
CVD, lipids, blood pressure.”
April 2010
Results
• Findings of the study
“27,586 patients had smoking status
reported. There were 432 MI (myocardial
infarctions = heart attacks), 600 CHD
(coronary heart disease), 746 CVD
(cardiovascular disease) and 1902
deaths.”
April 2010
Results (Continued)
• More findings
“After adjustment for other covariates,
the IRR (incidence rate ratios) of
CVD in patients who stopped
decreased from 2.32 within 1st yr of
stopping to 1.49 after 3+ years.”
April 2010
Results (Continued)
• More findings
“Similar trends were observed for the
MI and CHD endpoints, but
reductions in risk were less
pronounced for all cause mortality.”
April 2010
Do I Care?
You should have enough
information from the abstract
to know if you want to read the
whole article!
April 2010
Consider the Source
• University researcher
• National Institutes of Health
• Drug company staff
• Who would pay to study this
question?
• Alternative therapies?
April 2010
Who Was Studied?
“Inclusion Criteria”
•
•
•
•
Age
Gender
Ethnicity
Geographic location
April 2010
Who Was Studied?
• Drug approval: pharma wants
“clean” population, good adherence
• HIV/Hepatitis C coinfection can be
difficult for docs to deal with –
confuses the data!
Generalizability? Is the study
population “real world”?
April 2010
Who Wasn’t Studied?
“Exclusion criteria”
• Who would be harmed by study?
– If drug has liver side effects?
– If drug is injected?
– Has psych side effects?
– Causes birth defects?
April 2010
Statistics
• Lots of jargon! (p values, two-tailed,
IRR, statistical significance)
• The bottom line: Could the results
be just by chance?
Skills Building Workshops II: 2:45pm ~ 4:00pm
Statistics: The Basics: Tim Horn, AIDSMeds.com
April 2010
Design Issues
How confident are we in the results?
– How many were studied? The “N”?
– Control group?
– Randomization?
– Blinding?
Skills Building Workshops III: 4:15pm ~ 5:30pm
From SMART to START: Understanding
Randomized Controlled Trials: Paul Dalton, AIDS
Treatment Activist Coalition (ATAC)
April 2010
Data Analysis
Was analysis “pre-specified”?
“Torture the data until it confesses”
April 2010
What’s the Trial Design?
“People who switched to Geico
Insurance saved an average of
$350.”
August 2003
What’s the Trial Design?
“85% of people who used our herbal
ointment had their herpes lesions
clear up within 4 weeks.”
August 2003
http://www.aidsmeds.com
• Patient oriented
• “Lesson” format
• Expert interviews from key
conferences
April 2010
http://www.natap.org
• Comprehensive information
organized by topic
• Many original articles
• Expert articles on key issues,
conferences
• Fairly technical content
April 2010
http://clinicaloptions.com
• “Professional” orientation
• Extensive conference coverage
• Many CME offerings for MD, RN,
RPH and others
• Expert “discussion” of key studies
and conferences
April 2010
You Be the Judge
Authors are trying to prove
something!
Are you convinced?
Researchers want a solid answer.
Will they get one with this study?
April 2010
Correlation is not
Causation!
Ice cream causes drowning!!!
Spearman
Correlation
Coefficient =
.86
Ice Cream
Sales
Drowning
Deaths
April 2010
Customers Who Switched to
GEICO Saved an Average of
$350
• Inclusion criteria?
• Control group?
• Who DIDN’T Save?
August 2003
Herbs Effective Against
Herpes
We gave an herbal supplement to 86
people with genital herpes
outbreaks.
After 4 weeks, 86% of them had no
pain, itching, or visible herpes
sores.
April 2010