For easing newcomers along , please consult the glossary of terms that I’ve written to make the terminology very understandable. Also, consult the post that explains the essential background
Title: Reproductive risk factors for endometrial cancer among Polish women.
Authors: LA Brinton*,1, LC Sakoda1, J Lissowska2, ME Sherman1, N Chatterjee3, B Peplonska4, N Szeszenia-Dabrowska4, W Zatonski2 and M Garcia-Closas1
1 Hormonal and Reproductive Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, 6120 Executive Blvd., Suite 550, Rockville, MD 20852-7234, USA; 2 Department of Cancer Epidemiology and Prevention, Cancer Center and M. Sklodowska-Curie Institute of Oncology, W.K. Roentgena 5, Warsaw 02-781, Poland; 3Biostatistics Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Rockville, MD 20852, USA; 4Department of Occupational and Environmental Epidemiology, Nofer Institute of Occupational Medicine, Lodz, Poland.
Journal: British Journal of Cancer (2007) 96, 1450 – 1456
In this study, conducted in Poland, the authors consider reproductive risk factors associated with endometrial cancer. It is included as a validation of the biological model associated with breast cancer and the effects of estrogens. It is also included to shed some light on the games Dr. Brinton plays with her characterization of statistics.
The study is a case-control study involving 551 women with endometrial cancer (Cases) and 1925 healthy women (Controls).
Results are reported with 95% CI relative risks are reported as odds ratios (OR’s), which for our purposes are essentially the same.
Risk Factors:
Ever had a Full Term Birth
No OR=1.00 (This is the control/reference value)
Yes OR=0.53 (A 47% reduction is risk)
Number of Full-term births
0 OR=1.00
1 OR=0.63 (37% reduction in risk)
2 OR=0.54 (46% reduction in risk)
3+ OR=0.31 (70% reduction in risk)
Number of Induced Abortions
0 OR=1.00
1 OR=1.34 (34% increased risk)
2 OR=1.32 (32 % increased risk)
3+ OR=1.33 (33% increased risk)
Ever problem with Infertility
No OR=1.00
Yes OR=1.33 (33% increased risk)
Of the results, the authors have the following to say:
“A significantly decreased risk was associated with parity (OR adjusted
for parous vs nulliparous ¼ 0.51, 95% CI 0.4 – 0.7) (Table 2).”
{So a 47% decrease in risk with endometrial cancer associated with parity is a significant reduction in risk (As opposed to increased risks in breast cancer ranging from 30%-90% which are discounted as “statistically insignificant”).}
“Similar to previous investigations (Brinton et al, 1992; Albrektsen et al, 1995; Hinkula et al, 2002), we found a substantially reduced risk of endometrial cancer associated with parity, with women having three or more full-term births being at a 70% lower risk than nulliparous women.”
{Here 70% becomes substantial}
Later in the paper, Brinton et al. dismiss the induced abortion data as being possibly contaminated by recall bias, yet they offer no discussion of the biological effects of estrogens in inducing cancer. They make this omission in spite of estrogen being a WHO group 1 carcinogen.
So in this paper, we see the protective effect of full term pregnancy in preventing endometrial cancer.
We see an associated increased risk with induced abortion as well.
Most importantly, we see that when the discussion involves other cancers, Brinton regards 47% as significant and 70% as substantial.
Important data as we move the discussion along.
Let me get this right: Brinton reported the correct OR in the body of the report, but then in her conclusion, the OR’s for induced abortions were rendered not significant?
Where do you get the 30-90% range of increased risk? The highest OR I see is 1.34 for Induced Ab.
As you know, “statistically insignificant” is a strict criteria based on number of data points, CI, etc. This is different than saying something is substantial based on an opinion. By criteria, are Brinton’s claims to “statistically insignificant” at all valid? IN other words, are they truly statistically insignificant, or just qualitatively “insignificant”?
snaul,
I am using this study in juxtaposition to Brinton’s characterization of what 30%-90% means in relation to risks associated with breast cancer and induced abortion in the other studies. I thought I was clear about this, but perhaps was not. Sorry for the confusion.