Transcript Evidence-based Medicine Robert A. Harrington, MD, FACC

Evidence-based
Medicine
Robert A. Harrington, MD, FACC
Professor of Medicine
Division of Cardiology/Department of Medicine
Director, Cardiovascular Clinical Trials
Duke Clinical Research Institute
Duke University Medical Center
Evidence-based Medicine
and Drug Development: Outline
■ Background/philosophy
■ What is EBM?
■ Why does it matter?
■ Why is it important to drug development?
■ Basics of clinical research
■ Observational studies versus RCT
■ Types of trials
■ Concepts (randomization, sample sizes, etc.)
■ Overview of regulatory issues
Understanding the Need
for Evidence in Practice
Systematic Approach to Evaluating New Therapies
Half of what you learn in
your medical apprenticeship
(about therapy) will be correct…
you just don’t know which half.
Joe Greenfield, MD
Former Chair, Medicine
Duke University Medical Center
Global Burden of Disease: Top 10
Leading Causes of Death for All
Males and Females
United States: 2001
A Total CVD
B Cancer
C Accidents
D Chronic
Lower
Respiratory
Diseases
E Diabetes
Mellitus
F Alzheimer’s
Disease
Centers for Disease Control and Prevention/
National Center for Health Statistics
Percentage Breakdown of Deaths
from Cardiovascular Diseases
United States: 2001
Centers for Disease Control and Prevention/
National Center for Health Statistics
Estimated Direct and Indirect Costs (in Billions
of Dollars) of Cardiovascular Diseases and
Stroke United States: 2004
* Totals do not add up due to rounding and overlap.
All estimates prepared by Thomas Thom, NHLBI
The Graying of America
1995
Institute for the Future,
Health and Health Care 2010, 2000
2010
2030
Age-standardized Distribution of
Disability
IADL
Prevalence (/1000)
8
5–6 ADLs
Institutional
6
4
2
0
1982
Manton KG, Proc Natl Acad Sci, 2001
1989
1994
1999
U.S. Health Care Costs:
On The Rise Again!
Health Care Financing Administration
U.S. Health-related R&D Spending:
1986–2001
$ Billions
60
NIH
PhRMA
40
20
0
86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01
NIH Office of Extramural Research,
PhRMA Annual Survey, 2001
Medicare Growth 2002–2030
■ Medicare enrollment1
2002:
40 million
2030:
77 million
■ Worker: beneficiary ratio2
2002:
4.0 : 1
2030:
2.3 : 1
2070:
2.0 : 1
1Moon
M, N Engl J Med, 2001;
2www.whitehouse.gov
A New Era in Human Health and Medical Care:
Human Genome Project and Pharmacogenomics
Number of Drug Targets
New Drug Targets Expected:
Human Genome Project
12,000
5,000–10,000
10,000
8,000
6,000
3,000–5,000
4,000
2,000
~500
0
Known Drug
Targets
Drews J, Nat Biotechnol, 1996
Reiss T, Trends Biotechnol, 2001
1996
2002
New Targets Expected from
Human Genome Project
Cost of New Drug Development
Millions $
802
Analysis includes:
1. Discovery &
preclinical costs
2. Clinical costs
231
3. Capital costs
4. Project failures
1987
2000
http://csdd.tufts.edu/, accessed November 30, 2001
5. Impact of long
development
Evidence-based Medicine:
What’s the Problem?
“There is an unsettling, if little known,
truth about the practice of medicine…
Study after study shows that few
physicians systematically apply to
everyday treatment the scientific
evidence about what works best.”
Millenson, ML. Demanding Medical Excellence:
Doctors and Accountability in the Information Age, 1997
Why Do Clinical Trials?:
Lessons from Pediatrics
■ Top 10 prescription drugs
■ None approved by FDA for children
■ 900,000 children on SSRI antidepressants
■ Otitis media
■ Widespread antibiotics (?viral, ?long-term effects)
■ Asthma
■ More drugs, little to no data on long-term safety
■ Increasing mortality
Evidence-based Medicine
Why should we rely on evidence for
medical decision-making?
Because our intuition might be wrong!
Patients Without Event (%)
CAST
Placebo
(n = 743)
100
Cardiac Arrhythmia
Suppression Trial
95
Encainide
or Flecainide
(n = 755)
90
85
P = 0.0004
80
0
91 182 273 364 455
Days After Randomization
Odds of death
2.64
1.6
-0.5
1
Echt, New Engl J Med, 1991
4.4
2
3
4
5
Preventable Deaths
Exposed by CAST
7 million pts with active CAD, 2 million with CHF
25% have significant ventricular ectopy
10% with ectopy rx with antiarrhythmics, for 10 yrs
5% annual mortality, doubled with antiarrhythmics
25,000 unnecessary deaths
Menopause and HRT Use in the U.S.
■ 50 million post-menopausal women in U.S.
■ 1.8 million reach menopause each year
■ ~38% of U.S. menopausal women use HRT
■ In 2000:
■ 46 million prescriptions for Premarin
— 2nd most frequently prescribed drug in US
■ 22 million prescriptions for Prempro
— 6 million users
— $900 million in sales
Women’s Health Initiative (HRT)
Post-menopausal Women
n = 27,347
With Uterus
n = 16,608
CEE + MPA
0.625 mg/day
+ 2.5 mg/day
No Uterus
n = 10,739
Placebo
CEE
0.625 mg/day
Primary prevention; 7% with baseline CAD
Primary endpoint: cardiovascular death, nonfatal MI
Co-primary endpoint: invasive breast cancer
WHI Investigators, JAMA, 2002
Evidence-based Medicine
Combining quantitative evidence
about medical practice with expert
judgment in an effort to ensure the
provision of medical care with
reproducible high quality
Adapted from D Sackett
Alternatives to Evidence-based Medicine
Basis for
Clinical
Decisions
Marker
Measuring Device
Unit of
Measurement
Evidence
Randomised controlled trial
Meta-analysis
Odds ratio
Eminence
Radiance of white hair
Luminometer
Optical density
Vehemence
Level of stridency
Audiometer
Decibels
Eloquence
(or elegance)
Smoothness of tongue
or nap of suit
Teflometer
Adhesin score
Providence
Level of religious fervour
Sextant to measure
angle of genuflection
International
units of piety
Diffidence
Level of gloom
Nihilometer
Sighs
Nervousness
Litigation phobia level
Every conceivable test
Bank balance
Confidence
Bravado
Sweat test
No sweat
Isaacs D, BMJ, 1999
Pay for Performance
■ Under the Premier demonstration, top performing
hospitals receive bonuses based on their performance
on evidence-based quality measures.
■ Heart attack
■ Heart failure
■ Pneumonia
■ Coronary artery bypass graft
■ Hip and knee replacements
■ Participation in the demonstration is voluntary and open
to hospitals in the Premier Perspective system as of
March 31, 2003.
Guidelines: Weighing the Evidence
Weight of evidence grades:
= Data from many randomized clinical trials
= Data from single randomized trial or
nonrandomized studies
= Expert consensus
Guidelines: Classes of
Recommendation
I IIa IIb III
Intervention is useful and effective
Evidence conflicts/opinions differ but lean
towards efficacy
Evidence conflicts/opinions differ but lean
against efficacy
Intervention is not useful/effective and
may be harmful
The Cycle of Research
Observation
and
Surrogates
Scientific
Discovery
RCT
Observation
and
Outcomes
Evidence-based Medicine:
Randomized Clinical Trials
“The true method of knowledge
is experiment.”
William Blake, 1788
Using Evidence for Clinical
Decision-making:
Role of the Randomized Clinical Trial
“Statistical methods may be no
substitute for common sense but
they are often a powerful aid to it.”
D. D. Reid, commenting on the work of
Austin Bradford Hill, father of the
randomized clinical trial
Measurement of Effect in Clinical Studies:
Randomized Clinical Trials versus
Observational Studies
■ Efficacy (RCTs)
■ Experimental setting
■ “Ideal” circumstances
■ Limited population
■ Optimal care
■ Effectiveness (observational)
■ Clinical practice setting
■ Broad range of patients/providers
■ Community standard of care
Ayanian JZ, Eur Heart J, 1999
Randomized Clinical Trials:
Basic Principles in Evaluating Therapies
■ Treatment effects usually modest
■ Need large sample sizes
■ Qualitative interactions uncommon
■ Simple studies reasonable
■ Quantitative interactions common
■ Biggest effect in sickest patients
■ Unintended targets common
■ Pathophysiological reasoning unreliable
■ Long-term vs. short-term effects may differ
■ Combinations are unpredictable
■ Class effect may not be valid
Drug Rx in the U.S. Prior to 1938:
The Wild, Wild West
■ Wild claims made for pills,
and drugs sold without
testing
■ Radam’s Microbe Killer
(99.9% water) advertised as
cure for measles  cancer
■ Remedy for teething baby
could include opium
■ Manufacturers not required
to list “secret ingredients”
Pure Food and Drug Act of 1906
■ Mostly focused on cleaning up interstate
commerce in food
■ Also required drug labels to be complete and
accurate
■ Did not regulate/require:
■ False claims of efficacy (snake oil)
■ Testing before marketing
■ Proof of safety
Sulfanilamide Antibiotics:
The First Modern Miracle Drug
■ President Calvin Coolidge’s son dies in 1924
of septicemia from a tennis blister
■ Sulfanilamide discovered in 1934 by G
Domagk
■ President Franklin Roosevelt’s son cured of
serious streptococcal infection
■ Intense competition among pharma
companies to sell the most sulfa pills
Tragedy Drives U.S. Health Policy:
The Case of the SE Massengill Co. of
Bristol, TN ■ Market opportunity: other companies making
sulfa pills, let’s make liquid sulfa
■ Developed 1 gallon bottles of Elixir
Sulfanilamide (drug dissolved in diethylene
glycol)
■ Before shipping, carefully tested for
appearance, flavor, and fragrance
■ Over 4 wks in 1937, 353 pts (many children)
took drug. Within 1 wk, 105 were dead
■ FDA confiscated supplies due to labeling
deficiency; fined company $26,000
■ Company denied any responsibility, but
responsible chemist committed suicide
Food, Drug, and Cosmetic Act
(FDCA) of 1938: Key Elements
■ Banned interstate commerce of harmful
substances
■ Required new drugs to be approved by
FDA via a New Drug Application (NDA)
■ Required scientific proof of safety for
drug approval
Tragedy Drives U.S. Health Policy
(Again): The Case of Thalidomide
■ Synthesized in West Germany in 1954 as
antihistamine for allergy
■ Found to be “wonder drug” for providing “safe,
sound sleep” and for relieving morning sickness of
pregnancy
■ Introduced to market in West Germany on Oct. 1,
1957; widely used outside U.S.
■ Animal testing showed it to be extremely safe; no
lethal dose ever found
Thalidomide Teaches World a Hard
Lesson About Drug Safety
■ Safety testing did not include pregnant
animals
■ Unappreciated that drug crossed placenta and
caused severe damage to fetus between 3 and
5 weeks post-conception
■ Caused fetal death or severe malformations of
limbs (phocomelia) and internal organs
The Thalidomide Tragedy:
The U.S. Dodges a Bullet
■ Company applied to FDA for approval in 1960
■ FDA administrator (Dr. Frances Kelsey)
delayed review, asked for more safety tests
■ By 1961,  reports of thalidomide-related birth
defects
■ Fewer than 2 dozen American children
affected (mothers overseas)
1962 Kefauver-Harris
Amendment to FDCA
■ Required extensive animal,
pharmacological, and toxicological
testing before initial testing in humans
■ These data must be submitted in form of
Investigational New Drug (IND)
application and approved by FDA
■ NDA must show “substantial evidence”
of drug’s efficacy (effectiveness) as well
as safety
Investigational New Drug (IND)
Application: Overview
■ Summarizes evidence that it is reasonable to
move from preclinical to RCTs
■ Provides exemption from federal statute that
prohibits interstate shipping of unapproved drugs
■ Three major components
■ Animal pharmacology and toxicology (safety)
■ Manufacturing
■ Initial clinical protocols
Prescription Drug User Fee Act
(PDUFA) of 1992
■ Response by Congress to concerns about
length of drug approval process in U.S.
■ “User fees” for NDAs used to hire > 600
drug reviewers and support staff
■ By 1997, added $84 million/year to FDA
budget
■ Goal: standard application review 12 mos,
priority application review 6 mos
Food and Drug Administration
Modernization Act (FDAMA) of 1997
■ Reauthorizes Prescription Drug User Fee Act of ‘92
for 5 more years
■ Creates “fast track” review of rxs for serious/life
threatening disorders
■ Allows drug companies to disseminate info about
off label use (must file suppl. application)
■ Preserves the general assumption that 2 adequate
and well-controlled studies are needed to prove
safety and effectiveness
Safety Evaluation of Marketed Drugs:
U.S. FDA Perspective
■ Clinical testing
■ 1994 drug safety standard: 1500 patients exposed,
with 600 exposed for 6 mos and 300 for 1 yr
■ Adequate to detect 1/300–500 AE
■ Recognized limitations
— “Clinical trials are not real life”
■ FDA review
■ Toxicology, clinical studies
■ Review of proposed label, promotions
■ Postmarketing surveillance
■ MedWatch system
Friedman MA, JAMA, 1999
RCT Drug Exposure versus Actual
Use: Recently Withdrawn Drugs
Clinical Trials
(N)
Prewithdrawal
(N)
Terfenadine
5000
7,500,000
Fenfluramine
340
6,900,000
Dexfenfluramine
1200
2,300,000
Mibefradil
3400
600,000
Bromfenac
2400
2,500,000
Drug
Friedman MA, JAMA, 1999
Thalidomide Rises From the Ashes
■ Early use suggested thalidomide had some
anti-inflammatory properties
■ In 1964, MD in Jerusalem used some
remaining stock of drug in leprosy pt with
severe painful skin lesions
■ Within a few days, pt’s fever  and skin
lesions disappeared
■ In 1998, Calgene received FDA approval to
market thalidomide for leprosy
Recent Major RCTS and Registries in
NSTE ACS: Enrolling > 1000 patients
(Total n > 200,000 patients)
Antiplatelets
Strategies/
Registries
GUSTO II
PRISM/PRISM +
TACTICS
OASIS 2
PURSUIT
RITA-3
PARAGON A & B
GUARDIAN
TIMI 11
GUSTO-IV
NRMI
FRAXIS
OPUS
CRUSADE
SYMPHONY 1 & 2
GRACE
Antithrombins
ESSENCE
FRIC
FRISC 1&2
CURE
ACC/AHA NSTE ACS Guidelines:
Hospital Care
Initial Antithrombotic Therapy
I IIa IIb III
Immediate aspirin; clopidogrel if no ASA
Aspirin + clopidogrel, for up to 1 month*#
Heparin (IV UFH, LMW)
Any GP IIb/IIIa inhibitor + ASA/heparin for
all patients, if cath/PCI planned+
* ASA/clopidogrel to 9 months is 1B recommendation
# Withhold clopidogrel 5–7 days for CABG (1B)
+Cath is 1A recommendation for  risk NSTE ACS patients
Link Between Overall Guidelines
Adherence and Mortality
7
6
5.95
6.33
Adjusted
5.16 5.07
5
4.97
Unadjusted
4.63
4.16 4.17
4
3
2
1
0
Every 10%  in guidelines adherence results in an
11%  in mortality (OR = 0.89, 95% CI: 0.81–0.98)
 25%
Peterson E, ACC, March 2004
25–50%
50–75%
 75%
Clinical Research:
Basis of Evidence for Clinical
Practice
■ We learn what is effective and safe by evaluating
therapies in the clinical context
■ Increasingly, this effort will require comparisons
of active treatments and strategies, raising new
challenges/complexities
■ Answers to these questions cannot and should
not come from extrapolations or thought
exercises
“There are those who wander
around on the wards and those
who are doctors. The difference is
in having the data.”
EA Stead Jr.
Former Chair, DOM
Founder, Duke CV Databank
Founder, PA Profession