| HOME | HELP | CONTACT US | SUBSCRIPTIONS | ARCHIVE | SEARCH |
|
|
|
|||||||
|
|||||||||
Rapid Response to:
|
|
Responses published:
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif){kind=icon}
Bias in observational studies
|
|
|||
|
Peter S Millard, Family Medicine Residency Faculty Eastern Maine Medical Center
Send response to journal:
pmillard{at}emh.org Peter S Millard
|
Dr. Hall bases his analysis on observational data that is of questionable validity. The fact that churchgoers live longer than people who do not attend church may very well have nothing with churchgoing, but may result from uncontrolled confounding. Observational studies which showed a benefit of exogenous estrogens in post-menopausal women were de-bunked by the Women's Health Initiative randomized controlled trial. The accepted explanation for the discrepancy between the observational findings and the experimental findings is that women who took estrogens were systematically different than non-users in ways which resulted in improved outcomes (e.g. reduced coronary disease). Barrett-Connor referred to this as the 'healthy user effect' (1). The inability to control for the 'healthy user effect' resulted in the biased findings of many observational studies of estrogen use. Similarly, churchgoers are systematically different than non- churchgoers in ways that are difficult to measure but are likely result in improved health outcomes that may have nothing to do with churchgoing. Churchgoers are more likely to be employed, have intact families, and are less likely to be homebound by illness or disability. Until the 'healthy attender effect' can be controlled for, it is unwise to attempt to draw any causal effects between churchgoing and health. 1. Barrett-Connor E, Bush TL. Estrogen and coronary heart disease in women. JAMA 1991; 265: 1861-1867. |
|||
| HOME | HELP | CONTACT US | SUBSCRIPTIONS | ARCHIVE | SEARCH |
Rapid Response: