Osler Journal Club
Dan Munoz and Adnan Malik
September 13, 2006
Background
• Population-based case-control study
• Investigating association between ACE
inhibitors and abdominal aortic aneurysm
(AAA) rupture
• Based on animal data suggesting that
ACE inhibition may prevent expansion and
rupture of AAA’s
• AAA’s are common (4-8% of men > 50 yrs)
• Mortality of AAA rupture is 80-90%
Study Outline
• 15,326 consecutive patients, greater than 65 in
age, admitted with primary diagnosis of ruptured
or intact AAA over a 10 year period (1992-2002)
• Data derived from four linked administrative
databases in Ontario, Canada:
1. Ontario Drug Benefit database
2. Canadian Institute for Health Information Discharge Abstract
database
3. Ontario Health Insurance Plan database
4. Ontario Registered Persons database
Study Outline
• Analysis looks at association between
ACE inhibitor use prior to admission and
odds of admission for unruptured vs.
ruptured aneurysm (primary diagnosis
responsible for admission)
• Primary outcome = AAA rupture
• Multivariable analysis performed to adjust
for demographics, risk factors for
aneurysm rupture, comorbidities, health
care access and screening
Results
• Table 1. Baseline characteristics of
patients.
• Cases=patients with AAA rupture
• Controls=patients with unruptured AAA
• Notable for higher rate of imaging,
electrocardiography, and cholesterol
testing in control group
• Other health measures, comorbidities
were similar
Results
• Patients receiving ACE-I before admission
significantly less likely to present with
ruptured aneurysm, OR 0.82 (CI 0.740.90)
• Dosing and specific type of ACE inhibitor
did not affect association
• Table 2. Other antihypertensives were not
associated with lower incidence of AAA
rupture
Implications
• Association between ACE-I use and
reduced risk of aortic rupture in elderly
population with underlying AA Dx
• Potentially unique effect of ACE-I among
anti-hypertensives in imparting protection
against rupture
• Animal models and this study suggest
current role for ACE-I Rx in selected subpopulations, while we await more definitive
clinical investigation
Strengths of Approach
•
•
•
•
Rarity of condition (M 4-8%, F 0.5-1%, age >50) makes case-control design
useful/practical
Canadian HC system imparting uniformity of access to hospital care,
physician services Rx drug coverage
– Use of national utilization databases
– Further attempts to control for case-mix: no obvious inherent health
status differences between groups
Selection of controls
– (+) Demonstrated unruptured aneurysms
– (-) Question of how coming to attention
Findings not overstated
– Authors caution: association vs. causation
– Direction: non-surgical candidates, trials
Weaknesses of Approach
• Inherent to case-control design
– Missed populations
• Ruptures not making it to medical attention
• Silent, undiagnosed aortic aneurysms
• Absence of detailed info on smoking status
– Major risk factor for AA enlargement
• 10 year experience
– Uniformity of intervention/technique quality?
• Definition of ACE-I therapy (Rx record)
– Adherence/administration difficult to verify
Discussion Points
• How might this study have been better designed? Do
you agree with the case-control model here?
• What other primary endpoints would be clinically relevant
in examining ACE-I benefits/effects?
– Rate of enlargement, threshold size, etc.
• What would be next investigational step?
– How might a RCT be feasibly designed?
References
• Hackam D, et al., “Angiotensin-converting
enzyme inhibitors and aortic rupture: a
population-based case-control study,”
Lancet 2006, 368: 659-665
• Dr. Jeanne Clark, Division of GIM
Taxonomy of Clinical Research
Designs
Experimental
• Controlled Trials
– Non-Randomized
– Randomized
• Other Intervention
Studies
– Before vs After
– External Controls
– Historical Controls
Observational
• Prospective Cohort
Study
– Concurrent
– Non-Concurrent
• Case-Control Study
• Cross-Sectional Study
• Other / Mixed / ?
Comparison of 4 Major Study Designs
Goal or Feature
RCT
PCS
CC
X-S
• Estimate Prevalence
• Estimate Incidence
• Prove Causality
F
BA+
F
A+
B+
F
F
B-
A+
F
D
• Generalizability
• Feasibility
BC
AB
AA
A
A
RCT=Randomized, controlled trial;
CC=Case-control study;
PCS=Prospective cohort study;
X-S=Cross-sectional study
Case Control Studies
• Ppts selected on disease
– Present = cases; Absent = controls
– Look back at exposures
• Etiology
– Is head injury more common in pts with Alzheimers?
– Is MI more common in women on HRT?
• Outcomes
– Is history of FOB testing less common in people with
colon CA?
Examples of Exposures &
Outcomes in Clinical Research
Exposures
•
•
•
•
•
•
•
•
Traits & Behaviors
Genetic Variants
Infectious Agents
Environmental Toxins
Medications
Surgical Procedures
Behavior Modification
Screening Programs
Outcomes
• Death
• Disease
• Sub-clinical Indicators of
Disease
• Health-Related Traits
• Quality of Life
• Physical Function
• Costs