Previous osteoporotic fracture

Never 165

Ever 82

709 165

CI, confidence interval.

Melhus et al., 1998.

n inactivity, and use of hormone replacement therapy were verified, providing a general indication that the selection of subjects was successful and thereby increasing confidence that the patterns of association for uncertain risk factors is more likely to be valid as well.

Conventional wisdom can be incorrect, of course, with supposedly known associations subject to error in previous studies. Furthermore, the prevalence of important cofactors may vary across populations, so that an exposure that operates as a risk factor in one population may truly not operate in that manner in another. As discussed above, the potential for selection bias will vary across risk factors, with a given study able to generate valid measures of association for one exposure yet incapable of doing so for others. Even if valid for confirming known associations, the study may not be valid for addressing the association of primary interest. Finally, there is the ever-present random error to consider, a par ticular concern in attempting to corroborate well-accepted associations that are modest in magnitude. Failure to confirm an association between a history of heavy cigarette smoking and lung cancer would raise serious doubts about the validity of other measures of association with lung cancer identified in a case-control study, whereas many of the established risk factors for diseases such as breast cancer are so modest in magnitude that random error alone could well yield a spurious absence of measurable association. For example, although a positive association between alcohol intake and breast cancer is generally accepted (Longnecker, 1994), the magnitude is modest. Thus, studies that fail to find it, a non-trivial proportion of all studies (Singletary & Gapstur, 2001), are not by any means rendered invalid as a result. Like the other strategies, the examination of known and strongly suspected associations helps to direct the effort to scrutinize potential selection bias without providing conclusions in isolation from other considerations.

Beating The Butt On Your Own

Beating The Butt On Your Own

Need To Stop Smoking? Are You Willing To Follow My Powerful Strategies To Stop Smoking And Vividly Transform Your Life Today? Proven Tips, Tools and Tactics To Stop Smoking And Live An Awesome Life You Always Wanted.

Get My Free Ebook


Post a comment