Assessing Random Error, Confounding, and Effect Modification

Assessing Random Error, Confounding, and Effect Modification

Although researchers do their best to reduce error within every study, there will always be error. It is
important to identify and report any possible error within the research study in order to accurately interpret
the research study’s findings. In epidemiologic research, the focus is on assessing confounding and effect
modification along with normal statistical measures (p-values, confidence intervals, etc.).
For this Application Assignment, you will calculate and interpret the effects of confounding, random error,
and effect modification within epidemiologic research. Read each of the following questions and answer

them appropriately:

Introduction

Given the variables that normally affect research studies, health practitioners especially in
epidemiology normally come up with measures that would help minimize the errors. A
minimalistic error in any research tends to skew the findings more towards credibility if

2

WEEK6 EPI
appropriately applied to the findings. It is for these reasons that this paper tries to focus on
assessing confounding and effect modification along with the various statistical measures such as
p-values and confidence levels.

1. Case scenarios:
   i. Whether gender is a confounder.
   According to Elsevier (2007), ‘a confounder is a third variable that is associated with the
   exposure variable and the outcome variable in question.’ In this case, lots of exercises would
   improve the health of the women and thus reduce chances of heart attacks. To such, gender is not
   a confounder because the study is conducted among subjects of the same gender, women.
   This is because gender neither relates to exercises nor the outcome that is heart attacks.

ii. Whether smoking is a confounder.
In this case, alcohol consumption may result in liver cirrhosis. In addition, smoking is a
risk factor associated with liver cirrhosis among alcoholics and non-alcoholics. From this, it
would be prudent to deduct that smoking is a confounder in the relationship between alcoholism
and liver cirrhosis.
This is because smoking is related to the outcome, which is liver cirrhosis, as well as the
exposure which is alcoholism.

2. Interpreting result studies:

3

WEEK6 EPI
a) An odds ratio of 1.2 (95% confidence interval: 0.8-1.5) in the relationship between
low socioeconomic status and obesity. The confidence interval provides a range over
which a value is likely to occur. In this case, the odds of a case being socioeconomic
are 1.2 times the odds of a control being socioeconomic. It also means that low
socioeconomic status has 1.2 times the obesity rate among those of high
socioeconomic status.

b) A relative risk of 3.0 in the association between red meat and occurrence of colon
cancer, given the p-value of the association being 0.15. A relative risk is the ratio of
incidence rates and risks as shown below:

The P-Value of 0.15 estimates that those who consume red meat are 0.15 times most
likely to develop colon cancer compared to those who do not consume red meat.

c) An odds ratio of 7 (95% confidence interval: 3.0-11.4) for the association of smoking
and lung cancer. The odds of a case being a smoker are 7 times the odds of a control
being a non-smoker. This also means that non-smoking has a 7 times lung cancer risk
as the smokers.

3. Relationship between cigarette smoking and lung cancer. Cases 700, while the 425
   controls.

4

WEEK6 EPI

i. Crude odds ratio in heavy smoking; cases 450, controls 200, light smokers
cases 250, control 225.

This means that the light smokers are at 2.025 times lung cancer risk as their heavy smoking
counterparts.

ii. Design a 2×2 Table for the following. Relationship between alcohol use and
coronary heart diseases (CHD); Participants, 150 cases, 150 controls.
Alcoholics, cases 90, control 60. Non-alcoholics, cases 60, control 90.
Table 1.0 Coronary Heart Diseases (CHD) Cases and Controls According to Drinking Status

Case Control
Non-alcoholics 60 90
Alcoholics 90 60

150 150 N=300

This means that non-alcoholics are at 0.44 times risk of developing a CHD as
the alcoholics.

4. Smokers Non-Smokers

5

WEEK6 EPI
CHD No CHD CHD No CHD
Alcohol Use 80 40 10 20
No Alcohol Use 20 10 40 80

The odds ratio would be, OR= a/b
80/10=8.
Meaning that chances of developing CHD are at a rate of 8 whether the subject smoked
or not. Smoking is a confounder in this case because it associated with the outcome,
CHDs and the exposure, alcoholism.
After controlling for confounding, the relationship between alcoholism and CHDs is;
10/40=0.25. Meaning that the rate of developing CHD reduces with the control of confounding.

5. A) Association between taking a driver’s education class and risk of being in an
   automobile accident. 150 cases had been in an accident, 300 controls that had not. Of the
   cases, 70 took the driver’s education while of the controls 170 reported to have taken the
   training.

This result means that both those who are trained in a driver’s class and those who did not have a
risk factor of 0.67 likely to be involved in an accident.

6

WEEK6 EPI
B) Women Men
Accident No Accident Accident No Accident
Driver’s Ed 10 50 60 120
No Driver’s Ed 40 50 40 80
In women; OR=10/40
=0.25, while in men; it is 60/40=1.5
Effect modifier=1.5/0.25
=6
This is because an effect modifier is a variable that normally determines the outcome and is often
part of the causative network (Ann Aschengrau, 2013). As a result, we can conclude that gender
is an effect modifier in that men are found to be 6 times most likely to be involved in an accident
as compared to women.

Reference

7

WEEK6 EPI
Ann Aschengrau, G. R. (2013). Essentials of Epidemiology in Public Health. Jones & Bartlett
Publishers.
Elsevier, S. (2007). Epidemiology, Biostatistics, and Preventive Medicine. Philadelphia: Elsevier
Inc.