Georgia College

NRSG 8410 Epidemiology

Activity 5 & 6

Confounding (1 point each)

Which method of control is being used to control for smoking in each of the studies below?

1. A case-control study of the effect of a new lipid-lowering medication on the incidence of

CHD that determines the relative odds (OR) of CHD separately in smokers and non-
smokers.

Method of control: Stratification.

2. A randomized controlled trial of the effect of a new lipid-lowering medication on the

incidence of CHD conducted in a population of smokers and non-smokers ≥50 years of
age.
Method of control: Randomization

3. A case-control study of the effect of a new lipid-lowering medication on the incidence of

CHD where male controls are selected in the same proportion as male cases, and female
controls are selected in the same proportion as female cases.

Method of control: Randomization

Consider each of the following scenarios and state whether the variable in question is a
confounder (circle or highlight the correct answer):

1. A study of the relationship between contact lens use and the risk of eye ulcers. The
crude relative risk is 3.0, and the age-adjusted relative risk is 1.5. Is age a confounder in
this study?
i. Yes or No
2. A case-control study of the relationship between cigarette smoking and pancreatic cancer.
In this study, coffee drinking is associated with smoking and is a risk factor for pancreatic
cancer among both smokers and non-smokers. Assume that coffee drinking is not on the
causal pathway between smoking and pancreatic cancer. Is coffee drinking a confounder in
this study?
i. Yes or No
3. A study of the relationship between exercise and heart attacks that is conducted among men
who do not smoke. Is gender a confounder in this study?
i. Yes or No
4. A cohort study of the risk of liver cirrhosis among female alcoholics. Incidence rates of
cirrhosis among alcoholic women are compared to those among non-alcoholic women.
 Non-alcoholics are individually matched to alcoholics on month and year of birth. Is age a
confounder in this study?
i. Yes or No

Assessment of Selection Bias and Information Bias

Is There Cancer in the Cup?

The story of coffee and pancreatic cancer

Description
Brian MacMahon (1923-2007) chaired the Epi Dept at Harvard from 1958 - 1988.
In 1981, he published results from a paper on coffee drinking and pancreatic cancer: MacMahon
et al. N Engl J Med. 1981; 304:630-3.
Concluded that "coffee use might account for a substantial proportion of the cases of this disease
in the United States."
Substantial protest from coffee drinkers and industry groups, with coverage in the New York
Times, Time magazine and Newsweek.

Study Methods

Hospital-based case-control study

Identified 587 cases with histologically confirmed pancreatic cancer diagnosed at 11 Boston and
Rhode Island hospitals during 1974 – 1979. Two hundred nine of the cases (35.6%) did not
participate because they refused, were too ill, or died, etc.
Identified 1,118 controls with other diseases who were under the case of the same physician who
had hospitalized the cases. Diseases in the control group included breast, colo-rectal and stomach
cancer; hernia, colitis, enteritis, gastritis, and bowel obstruction. Four hundred seventy-four of
the controls (42.4%) did not participate because they refused, were too ill, or died, etc.
Interviews with patients collected information on smoking; and tea, coffee, and alcoholic
beverage consumption. Asked about duration and intensity of smoking for cigarettes, cigars, and
pipes; frequency, duration and type of alcoholic beverage consumption before onset of illness;
“typical” consumption (# cups/day) of coffee and tea before current illness was evident.

Nothing stated about the location of interviews or whether interviewers were blinded.

Study Results

Weak association with cigarette smoking (Odds Ratios: 1.4-1.6 for > 1 pack/day). No
associations with use of cigars, pipes, alcoholic beverages, or tea.
Assessment of Information Bias (3 points each)

1. Describe possible sources.

The study faces potential biases that could affect the validity of its findings. Non-
participation rates were significant among both cases (35.6%) and controls (42.4%), often
due to refusal or health-related reasons, possibly skewing the sample's representativeness if
these reasons correlate with coffee consumption and pancreatic cancer risk. Additionally,
controls were drawn from hospital patients with other conditions, which might not mirror the
broader population. Selection bias would be introduced if these conditions relate differently
to coffee consumption than the overall public. Also, relying on patient interviews to gather
data on coffee intake raises concerns about recall bias, where cases might remember their
coffee habits differently from controls, especially if they are aware of the study's focus.
Furthermore, if interviewers were not blinded to the study hypothesis, interviewer bias could
occur, potentially influencing responses on coffee consumption between cases and controls.
These biases underscore the need for cautious interpretation of the study's conclusions
regarding coffee consumption and pancreatic cancer risk.

2. Describe likely impact on results (direction and magnitude of this bias).

Selection Bias
Direction: If the non-participants among cases and controls had different levels of coffee
consumption than participants, this could either exaggerate or underestimate the true
association. For instance, if sicker patients (who might consume less coffee) are less
likely to participate, the study might overestimate the association between coffee
consumption and pancreatic cancer.
Magnitude: The impact on the magnitude of the results is difficult to quantify without
more information, but substantial non-participation could significantly alter the observed
association.
Information Bias
Recall Bias Direction: If cases are more likely to recall and report higher coffee
consumption due to their diagnosis, this would likely overestimate the association
between coffee consumption and pancreatic cancer.
 Interviewer Bias Direction: If interviewers investigate cases more about coffee
consumption, this could also lead to an overestimation of the association.
Magnitude: The magnitude of the bias would depend on the degree to which recall and
interviewer bias are present. Given that these biases are likely to inflate the association,
the true association may be weaker than reported.

3. Describe possible solutions.

To lessen selection bias, it is essential to enhance participation rates among both cases and
controls, which may involve proactive follow-up with non-responders. Using controls from
the general population instead of hospital-based controls can also diminish selection bias by
ensuring a more representative comparison group.

Addressing information bias involves several strategies. Ensuring that interviewers are
unaware of the study hypothesis can reduce interviewer bias, preventing them from
inadvertently influencing responses. Standardizing interviews and providing thorough
training to interviewers can regulate data collection procedures across cases and controls,
thereby minimizing potential inconsistencies. Validation studies, to confirm self-reported
coffee consumption. and using objective measures like biomarkers of caffeine intake, can
additionally improve the accuracy of data and reduce recall bias.

By adopting these measures, researchers can enhance the strength of studies examining the
link between coffee consumption and pancreatic cancer, effectively easing both selection and
information biases.