Investigators conduct a case-control study of pancreatic cancer. Data were enter
ID: 3324379 • Letter: I
Question
Investigators conduct a case-control study of pancreatic cancer. Data were entered into STATA and analyzed via logistic regression, with outcome defined as case status (1 if subject has pancreatic cancer, 0 if not), and predictors defined as coffee drinker (1 for yes, 0 for no) and smoking status (1 for yes, 0 for no). Use the STATA outputs to answer the questions. In each response, state the output(s) you are using, unless it is already indicated in the problem.
OUTPUT A. logistic case coffee
Logistic regression Number of obs = 223
LR chi2(1) = 11.03
Prob > chi2 = 0.0009
Log likelihood = -148.87714 Pseudo R2 = 0.0357
------------------------------------------------------------------------------
case | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
coffee | 2.877778 .941664 3.23 0.001 1.515402 5.464956
_cons | .7142857 .1117801 -2.15 0.032 .5256124 .970685
------------------------------------------------------------------------------
OUTPUT B. logistic case smoking
Logistic regression Number of obs = 223
LR chi2(1) = 14.63
Prob > chi2 = 0.0001
Log likelihood = -147.07355 Pseudo R2 = 0.0474
------------------------------------------------------------------------------
case | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
smoking | 2.971429 .8637126 3.75 0.000 1.68091 5.252743
_cons | .625 .1074299 -2.73 0.006 .4462403 .8753692
------------------------------------------------------------------------------
OUTPUT C. logistic case coffee smoking
Logistic regression Number of obs = 223
LR chi2(2) = 19.01
Prob > chi2 = 0.0001
Log likelihood = -144.88428 Pseudo R2 = 0.0616
------------------------------------------------------------------------------
case | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
coffee | 2.071308 .7262949 2.08 0.038 1.041777 4.118269
smoking | 2.383469 .737036 2.81 0.005 1.300151 4.369433
_cons | .5678218 .1018896 -3.15 0.002 .3994597 .8071443
------------------------------------------------------------------------------
OUTPUT D. cc smoking coffee [freq=n]
Proportion
| Exposed Unexposed | Total Exposed
-----------------+------------------------+------------------------
Cases | 37 43 | 80 0.4625
Controls | 18 125 | 143 0.1259
-----------------+------------------------+------------------------
Total | 55 168 | 223 0.2466
| |
| Point estimate | [95% Conf. Interval]
|------------------------+------------------------
Odds ratio | 5.975452 | 2.941207 12.2855 (exact)
Attr. frac. ex. | .8326486 | .6600036 .9186032 (exact)
Attr. frac. pop | .3851 |
+-------------------------------------------------
chi2(1) = 31.29 Pr>chi2 = 0.0000
a)Is smoking status related to case/control status, ignoring coffee consumption? Answer this question by interpreting an appropriate association measure and by using its corresponding 95% confidence interval to assess the association. Is this relationship potentially causal? Briefly justify your response.
a.
b)Is smoking status related to coffee drinking, ignoring case/control status? Answer this question by interpreting an appropriate association measure and by using its corresponding 95% confidence interval to assess the association. Is this relationship potentially causal? Briefly justify your response.
c)Does smoking meet the 3 criteria for being a confounder of the relationship between pancreatic cancer and coffee drinking? Justify your answer (use calculations from the problems above to support your point).
Explanation / Answer
a)smokers case have 2.97 times higher risk of pancreatic cancer and it ranges from 1% to 4%. yes smoking status is realted since its p value is less than 0.05(standard tolerance limit).
controls have 0.625 times higher risk of pancreatic cancer and it ranges from 0.4 to 0.8
b) Yes, they are relate since odds ratio is 5.95 which is greater than 1. so it is highly correlated. it lies between 2% to 12%
Related Questions
drjack9650@gmail.com
Navigate
Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.