09.08.2013 Views

Fundamentals of epidemiology - an evolving text - Are you looking ...

Fundamentals of epidemiology - an evolving text - Are you looking ...

Fundamentals of epidemiology - an evolving text - Are you looking ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Follow-up must be longer for <strong>an</strong>y given number <strong>of</strong> cases.<br />

6. The distribution <strong>of</strong> smoking habits varies considerably with age, as does c<strong>an</strong>cer incidence.<br />

Therefore, age adjustment or age-specific comparisons are needed.<br />

7. To demonstrate that disease diagnosis was not based not only on clinical judgment but<br />

documented by examination <strong>of</strong> tissue specimens. Particularly with a rare disease, like lung c<strong>an</strong>cer,<br />

false positives c<strong>an</strong> readily distort the observed association, leading to information<br />

(misclassification) bias.<br />

9. Yes: Prevalent cases may have failed to respond initially, <strong>an</strong>d therefore been deleted from the<br />

cohort; short period <strong>of</strong> follow-up <strong>of</strong> disease with long induction; doctors may receive better<br />

medical care, thereby reducing mortality.<br />

12. Ultimately, it is not possible to exclude all possible alternate expl<strong>an</strong>ations except by conducting<br />

<strong>an</strong> experiment in which smoking (or stopping smoking), c<strong>an</strong> be r<strong>an</strong>domly allocated in sufficiently<br />

large groups with sufficient adherence to the experimental regimen. For ethical <strong>an</strong>d practical<br />

reasons, such <strong>an</strong> experiment c<strong>an</strong>not be conducted with smoking onset as the experimental<br />

variable. Smoking cessation programs have not been sufficiently effective to serve as <strong>an</strong> alternate<br />

intervention. If they were, it is doubtful that they could be used in a lung c<strong>an</strong>cer trial, since the<br />

evidence for the harmful effects <strong>of</strong> smoking is so abund<strong>an</strong>t that ethical considerations would<br />

probably preclude such a trial.<br />

Though "absolute" pro<strong>of</strong> requires hum<strong>an</strong> experiments, the weight <strong>of</strong> observational <strong>an</strong>d <strong>an</strong>imal<br />

evidence is convincing. Any particular factor proposed as <strong>an</strong> alternate expl<strong>an</strong>ation (e.g., exercise,<br />

diet) c<strong>an</strong> be examined simult<strong>an</strong>eously with smoking, to see which factor is associated with lung<br />

c<strong>an</strong>cer when the other factor is controlled. There are no serious contenders. Traditional criteria<br />

for causal inference are well satisfied:<br />

a. strong association (RR about 8 for smokers overall);<br />

b. dose-response effect exists;<br />

c. replication in m<strong>an</strong>y studies;<br />

d. cohort studies demonstrate that exposure precedes disease;<br />

e. biologic expl<strong>an</strong>ation (<strong>an</strong>imal <strong>an</strong>d tissue culture models, autopsy studies);<br />

f. experimental confirmation available in <strong>an</strong>imal models;<br />

g. <strong>an</strong>alogy to other carcinogen-c<strong>an</strong>cer associations.<br />

Smoking <strong>an</strong>d lung c<strong>an</strong>cer is one <strong>of</strong> the strongest <strong>of</strong> epidemiologically-established relationships.<br />

Controversy remains, but that should not paralyze policy pl<strong>an</strong>ning (see M.A. Ibrahim, The<br />

cigarette smoking/ lung c<strong>an</strong>cer hypothesis. Editorial. Am J Public Health 1976; 66:131-132).<br />

Nevertheless, the debate has continued even into recent years (see references in Bibliography).<br />

_____________________________________________________________________________________________<br />

www.sph.unc.edu/EPID168/ © Victor J. Schoenbach 8. Analytic study designs - Assignment solutions - 268<br />

rev. 8/31/1998, 9/5/1999

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!