|
medpundit |
||
 |
 |
|
Wednesday, July 17, 2002Lacey’s research involved 44,241 post-menopausal women whose health histories were tracked for about 20 years as part of a major breast cancer study. Among those women, 329 developed ovarian cancer. The researchers found that compared to similar women not on hormones, those taking estrogen therapy had a 60 percent greater risk of developing ovarian cancer. The risk increased proportionately with longer duration of hormone use; those who used estrogen therapy for 20 or more years were approximately three times more likely to develop ovarian cancer. However, women who used an estrogen-progestin combination did not appear to have a significantly increased risk of ovarian cancer. “The main finding of our study was that post-menopausal women who used estrogen replacement therapy for 10 or more years were at significantly higher risk of developing ovarian cancer than women who never used hormone replacement therapy,” Lacey said in a statement released by the cancer institute. Saying that women on estrogen are 60% more likely to develop ovarian cancer may be a statistically true statement, but it leads one to believe that the risks are greater than they are. Most people reading that statement would think that 60 out of 100 women who take estrogen develop ovarian cancer. Nothing could be further from the truth. Unfortunately, the original paper is as clear as mud when it comes to assessing individual risks. The study tracked the medical histories of 44,241 postmenopausal women for anywhere from one month to 20 years. (The mean follow-up was 13 years.) Nowhere in the paper do the authors tell us how many of those were taking estrogen, and how many weren’t. They identified 329 cases of ovarian cancer, but they don’t tell us how many of those were taking estrogen or for how many years they took it before diagnosis. Instead, they express their data in terms of “person-years” and “rate ratios”. The highest rate ratio of ovarian cancer that they came up with for estrogen users was 1.6. This is where the figure of “60% more likely to develop ovarian cancer” came from. The non-users of estrogen were given the value of 1.0 for their rate ratio, since that is the standard to which the users of estrogen were being compared. What does this all mean? Who knows? The paper is written in such a haze of statistical analysis that it’s impossible for a non-mathematician to cut through it. (Looks like a case for Numberwatch) What I do know, is that the numbers of ovarian cancers detected were very small in proportion to the number of women followed. That, combined with all the stastitical mumbo jumbo in the paper, makes me doubt the clinical significance of the findings. Even the accompanying editorial in JAMA admits that the data “do not establish causality.” This is one report that should be taken with a very large grain of salt. posted by Sydney on 7/17/2002 09:37:00 AM 0 comments 0 Comments: |
|
