Link to the Practice Exercises
(a) Research question: Does a monastic life-style decrease the risk of
morbidity and disability?
(b) Target population: Dutch men.
(c) Exposure: Monastic life-style as practiced by Trappist and Benedictine Dutch monks.
(Lifestyle includes vegetarianism, prayer, meditation, contemplation,
long periods of silence, etc.)
(d) Disease: Morbidity and disability, self-reported, standardized scales.
(e) Study design: Observational, individual-level observations, cross-sectional
(f) Measures of Association: SMR for morbidity; SMR for disability,
each adjusted for age and
education.
(g) Results:
SMR(morbidity) = 1.07 (95% confidence interval: 0.89, 1.26), suggesting no
association between the Monastic lifestyle and morbidity.
SMR(disability) = 2.21 (95%
confidence interval: 1.44, 3.32), more than
doubling of risk (120% increase in risk)
(h) Study limitations (speculative): Information bias--data are self-reported. Possible
"cart before the horse bias"--monks may be self-selected for disability (i.e., persons with
disabilities may be more likely to become monks).
(i) Bottom line: Data suggest Monks have similar morbidity rates but twice the expected disability
rates.
(a) Research question: Do low-estrogen oral contraceptive cause thromboembolism at the same
rate as intermediate- and high-dose formulations?
(b) Target population: Michigan Medicaid women of child-bearing agre receiving oral
contraceptives through Medicaid reimbursement.
(c) Exposure: Three levels of exposure: E0 = low-dose oral contraceptives (<50 mcg estrogen); E1
= intermediate-dose oral contraceptives (50 mcg estrogen); E2 = high-dose oral contraceptives
(>50 mcg of estrogen). The low-dose group serves as the "unexposed" group.
(d) Disease: Deep venous thromboembolism (blood clot of the lower extremity traveling to lungs,
very dangerous)
(e) Study design: Cohort with incidence density data.
(f) Measures of Association: Relative risks.
(g) Results: RR(intermediate-dose vs. low-dose) = 1.5 (95% confidence interval 1.02,
2.1); about a 50% increase in risk.
RR(high-dose vs. low-dose) = 1.7 (95% confidence interval 0.9, 3.0);
about a 70% increase in risk.
Possible dose-response relation.
(h) Study limitations (speculative): First author is a
curmudgeon.
(i) Bottom line: Data suggest risk increases with estrogen dose.
(a) Research question: Is baldness related to myocardial infarction?
(b) Target population: men admitted to Massachusetts and Rhode Island hospitals.
(c) Exposure: E0 = no hair loss; E1 = frontal baldness; E2 = vertex baldness; E3 = severe vertex baldness.
(d) Disease: non-fatal heart attacks.
(e) Study design: Case-control.
(f) Measures of Association: Author refers to these as "relative risks," however, they must be
odds ratios since this is a case-control study.
(g) Key results:
RR(frontal baldness) = 0.9 (95% confidence interval 0.6, 1.3); no association.
RR(vertex baldness) = 1.4 (95% confidence interval 1.2, 1.9); positive
association (40% increase in risk).
RR(severe vertex baldness) = 3.4 (95% confidence interval 1.7, 7.0);
more than tripling of risk.
(h) Study limitations (speculative): Is baldness associated with other diseases (confounding?);
is it a risk indicator and not a true risk factor; reverse-causality bias?
Bottom line: Vertex baldness is associated with coronary artery disease in men.
(a) Research question: Is extent or progression of baldness associated with coronary heart disease
risk.
(b) Population: Framingham Massachusetts men, 1956 - 1986
(c) Exposure levels:
The extent of baldness was classified as follows: E0 = no baldness; E1 = one area bald; E2 = two
areas bald; E3 = all areas bald.
The progression of baldness was classified as: E0 = mild or no progression, no baldness; E1 =
moderate progression, E2 = rapid progression
(d) Disease: New occurrences of the following outcomes:
D0 = no cardiovascular disease; D1 = coronary heart disease incidence; D2 = coronary heart
disease death; D3 = cardiovascular disease; D4 = death due to any cause
(e) Study design: 30 years follow-up (cohort) study
(f) Measures of Association: Relative risks adjusted for age and other known cardiovascular
disease risk factors.
(g) Key results:
No association between extent of baldness and any of the outcomes.
Rapid progression of baldness associated with the following outcomes:
RR(coronary heart disease incidence) = 2.4 (95% confidence interval: 1.3 - 4.4)
RR(coronary heart disease incidence) = 3.8 (95% confidence interval: 1.9 - 7.7)
RR(all cause mortality) = 2.4 (95% confidence interval: 1.5 - 3.8)
(h) Study limitations (speculative):
(i) Bottom line: Rapid progression of baldness may be a risk indicator or risk factor for coronary
heart disease.
(a) Research question: What factors are predictors of breast and cervical cancer screening in
Vietnamese women in California?
(b) Population: Vietnamese women in four California counties.
(c) Exposure levels: Multiple exposures were studied, including age, number of years in the US,
ever married, having health insurance, having a Vietnamese doctor, being unemployed, and being
of Chinese-Vietnamese background.
(d) Disease: 3 separate screening outcomes: clinical breast exam; mammogram; Pap test
(e) Study design: Cross-sectional telephone survey
(f) Measures of Association: Odds ratios adjusted by logistic regression
(g) Key results: Positive associations with age, number of years in the US, ever married, having
health insurance; negative associations with having a Vietnamese doctor, being unemployed, and
being of Chinese-Vietnamese background.
(h) Study limitations (speculative): Further information is needed
(working from abstract only).
(i) Bottom line: Data suggests target populations for screening addressing both Vietnamese
consumers and physicians.