Confidence Intervals, Confounding

Question #1

How does sample size affect a confidence interval?  You need to list all the ways the interval is affected.  If you did the exact same study twice, but had a sample of n = 500 for the first study, and a sample of 1,000 for the second study, which one do you think would be more useful in determining a potential association and why? (Hint: some of this is answered by your response to the first question!)

Also, will an increase in sample size always be an “improvement”?  Why or why not?  How is power involved with both the sample size and confidence interval?

Question #2

A case-control study was conducted to investigate the hypothesis that adult alcohol consumption is associated with the risk of liver cancer.  For the study, 6000 patients under 55 years of age with incident liver cancer, diagnosed between 2007 and 2010, were identified from the Detroit SEER registry records.  A random sample of 1,200 individuals with a similar age, sex, and race distribution as the cases was also identified.  All of the study participants were interviewed to assess average adult alcohol consumption and other lifestyle factors, including smoking and exercise.  Average adult consumption was determined as drinking at least 12 ounces of alcohol per day.  The data are as follows.a)  Calculate and interpret the Odds Ratio.  Then, calculate and interpret the 95% confidence interval for this odds ratio.

Research has shown an association between the amount of alcohol consumed and smoking, so the data from the case-control study were stratified by smoking, as shown below.

b)  Calculate and interpret the OR for the association between average alcohol consumption and liver cancer in never smokers. Then do the same calculation for smokers.

c)  Calculate the 99% confidence interval for the OR you calculated in Part a above and explain why the interval is narrower or wider than the 95% confidence interval you calculated in Part a.

d) Here, we have stratified on the potential confounder of smoking.  Based on what you are reading, what variable seems to have the most influence on having liver cancer, alcohol consumption or smoking?  Why

Question #3

Briefly describe the main similarities and differences between each of the following:

a)Prevalence and incidence

b)Incidence rate and cumulative incidence

Question #4

What information does a p-value provide and why is a p-value less informative than a confidence interval?  (4 points)

Question #5

In 2017, there were 4,500 new adult cases of osteoporosis diagnosed out of a population of 70,000 adults at risk in Wellsville.  In 2018, 2,000 new cases of osteoporosis were diagnosed and the adult population increased by 10,000.

Assuming none of the osteoporosis cases died, what is the prevalence of osteoporosis on December 31, 2018? (3 points)

Next, what was the cumulative incidence (risk) of developing osteoporosis among the adult population at risk during the year 2018?  Remember the numerator and denominator (think carefully).

Question #6

Explain why the z-scores increase for increasing percentages of the confidence intervals.  Cut-and-paste a normal curve from one of many websites and indicate where the mean is located (if everything is labeled, this is ok!).  Then, indicate where the z-scores would be located on this curve (with text, no need to draw!).  Remember that standard deviation is on both sides of the mean (this helps make the confidence interval!).

This assignment partially fulfills the following Course Objectives for this NUR 735 course:

1.Describe concepts of epidemiology and its impact on population health.

2.Interpret biostatistical methods in epidemiology relevant for the development, implementation, and evaluation of health services.