Data dredge of the day: exercise and breast cancer
On Friday, news began reporting that postmenopausal women can cut their risks for breast cancer by 30% with vigorous exercise, and that even moderate physical activity won’t cut it. This was shown, we heard, in a huge study of over 32,000 women followed for 11 years, conducted by scientists at the National Cancer Institute at the National Institutes of Health.
We can be certain that no reporter or editor read the study or, more importantly, understood what it had shown. Or, did they really believe it provided evidence that elderly women need to do 23 hours every week of strenuous exercise — like chopping wood, running, competitive tennis or biking uphill — to reduce their risks for breast cancers?
This is a scenario we’ve heard countless times before: A press release went out on Halloween Eve about a study: "Vigorous activity protects against breast cancer." The following morning when the embargo was lifted, at the stroke of midnight no less, news around the world began reporting the story, which continues to make news headlines today.
The study was a data dredge that had been unable to find a tenable link between physical activity and breast cancers. We could stop there, point to this and say “step and repeat.” But I’d get letters… and readers would miss out on some of the most interesting findings.
Let’s start with a brief overview
This was an observational study by NCI-NIH researchers, led by Dr. Michael F. Leiltzmann, currently at the Institute of Epidemiology and Preventive Medicine at University Hospital Regensburg in Germany. It was published in Breast Cancer Research, a free-access online, peer-reviewed journal of BioMed Central, the newest addition to Springer publishing.
The authors used the database called the Breast Cancer Detection Demonstration Project (BCDDP) Follow-up Study. The BCDDP was originally a project of the National Cancer Institute and the American Cancer Society to look at mammography screening. It had enrolled 283,222 women across the country who had breast exams between 1973 and 1980. The BCDDP Follow-up Study arm was established in 1979 to follow 64,182 of the original women — it included 4,275 women who had been diagnosed with breast cancer; 25,114 women who had false positives and been found to have benign breast conditions confirmed by biopsy and no malignancy; 9,628 women with positive screenings and BCDDP had recommended surgical intervention but who had no follow-up procedure; and a random sampling of another 25,165 women. This BCDDP Follow-up Study database was created from telephone interviews of the women, and up to six annual phone interviews through 1986. Thereafter, the women were mailed self-administered questionnaires intermittently through 1998.
This just-published secondary analysis set out to specifically examine the relationship between the intensity and duration of physical activity of all types (job-related, recreational and household) and postmenopausal breast cancers. For this paper, the NCI-NIH authors used data from the first questionnaire that asked the women about their physical activity. Of the 81% (51,696 women) who returned this questionnaire, these authors excluded those with breast cancer diagnoses, women with previous cancers, those who were premenopausal, and those with missing information; leaving 32,269 women who were monitored for breast cancer diagnoses from state cancer registries, National Death Index and self reported cases. The actual number of identified cases of postmenopausal breast cancer used in this paper’s analysis was 1,506.
Physical activity had been assessed by asking the women to estimate how many hours per typical weekday and weekend over the previous year that they spent doing both moderate and vigorous physical activities. The questionnaire included an extensive list of examples from housework to fast jogging to help their recall. Each activity was assigned a MET (metabolic equivalent task) score. While the authors said that the data on physical activity hadn’t been validated, their method of assessing physical activity “is similar to that employed in the Framingham Heart Study, which demonstrated a correlation between questionnaire-based and indirect calorimetry-based physical activity of 0.43. In addition, our physical activity measure includes selected components of the College Alumnus Physical Activity Questionnaire, an assessment tool that is positively correlated with maximum oxygen uptake, percentage body fat, high-density lipoprotein levels and body mass index.”
The authors divided the women’s total amounts of vigorous and moderate physical activity into quintiles, with #1 being the least active (<1 hour/week of vigorous activity and 14.8 hours/week of moderate-intensity activity) and #5 being the most active (22.7 hours of vigorous activity per week and 59.7 hours/week of moderate-intensity activity). They then threw all of this data into their computer and used Cox proportional hazards regression modeling to estimate the relative risk of postmenopausal breast cancers associated with physical activity.
Findings — Here’s how we know that no reporter or editor read the study:
1. A few interesting observations were found among the 32,269 women, divided by levels of physical activity at baseline. Their average caloric intakes were nearly identical (within 20 kcal/day) and their average BMIs varied by less than one unit (25.4 to 24.6). Based on their heights, their weights differed by about a mere 4 pounds — 148 to 144 pounds — even though the amounts of strenuous exercise varied by 22 hours a week.
The study authors attributed the stronger correlation between vigorous exercise among lean women to speculate that heavier women may “misreport as vigorous” light activities and not really “exercise as intensely as lean women.” [It seems that everybody “knows” that fat women can’t possibly exercise vigorously, let alone teach back-to-back aerobics classes or be Olympic athletes.]
2. The study found no correlation between levels of moderate, nonvigorous physical activity and postmenopausal breast cancers, even though the various women were doing about 15 hours to 60 hours a week! So much for beliefs that moderate levels of physical activity, no matter how many hours we clock in, have special abilities to reduce risks for breast cancer.
3. The authors reported: “We found that a higher amount of vigorous physical activity was associated with a small, statistically non-significant decrease in postmenopausal breast cancer risk in our cohort as a whole.”
None of the relative risks associated with breast cancer at the various levels of physical activity were statistically significant (p = 0.21) even when adjusted for confounding factors of age, family history of breast cancer, benign breast disease, breast cancer screening, BMI, age at menarche and menopause, age at first live birth, oral contraceptive use, hormone therapy, education, smoking, dietary fat and alcohol. It goes without saying that the relative risks were also untenable, ranging from 0 to 13%.
They also found no correlation at all between total physical activity or vigorous activity and breast cancers among fat women (BMI≥25).
This study was a null study — it found no link between the amount or intensity of physical activity and postmenopausal breast cancers.
3. The authors reported: “The apparent inverse association between total physical activity and postmenopausal breast cancer was virtually entirely contributed by vigorous physical activity.”
They are referring to the finding that the largest relative risk their computer found was at the highest quintile of (both vigorous or total) exercise when compared to the lowest quintile, an untenable 13% lower relative risk for breast cancer. This is precisely the meaningless statistic that Dr. John P. Ioannidis, M.D. described in his investigative report, “Why most published research findings are false,” as working in a null field. [Follow the links from here for more information.]
4. But here is the money shot. This untenable, happenstance correlation was not only spurious, but this fact confirms that this study was unable to find that physical activity has anything to do with breast cancers.
There was no dose response! No linear relationship. The risks of breast cancer didn’t steadily drop with increasingly greater total amounts of physical activity. The relative risks yo-yoed up and down with each quintile.
In other words, for example, women in quintile 4, working out strenuously for 10.3 hours/week, could “lower” their risks of breast cancer by cutting their vigorous activity to 3.6 hours/week.
And the women averaging 22.7 hours each week of the most strenuous exercise — that’s 3 hours 25 minutes of vigorous exercise every day — had nearly the identical risks as women doing just 3.6 hours a week.
The actual incidences of breast cancers among the 32,269 women at all the quintiles of physical activity varied by only 0.84%. But, again, these actual incidences of breast cancers had no relationship to the amounts of physical activity and jumped around between quintiles 1 -5 with no rhyme or reason: 4.96%, 4.64%, 4.78%, 4.82%, and 4.12%.
This study epitomized an epidemiological study that came up empty. It was a null study and found no viable link between physical activity and post menopausal breast cancers.
As we know, epidemiological studies are the first step towards narrowing down potential factors for diseases and when a strong link with a biological plausibility is found, then an hypothesis is tested in a series of clinical intervention trials to see if there is a causal role, and to learn if an intervention is safe and effective. Without even a link, scientists move on look somewhere else for a possible cause. That’s why these null studies can give us one less thing to worry about. There is no support that older women need to worry that they are putting themselves at risk for breast cancer if they don’t do strenuous exercise, let alone for at least 3.25 hours a day!
There was one aspect of this study that defies an explanation, however. The National Cancer Institute authors’ conclusion bore no resemblance to their findings. The authors wrote:
In conclusion, our results support the hypothesis of an inverse association between physical activity and postmenopausal breast cancer. Risk reduction appeared to be limited to vigorous forms of activity… Future studies designed to evaluate in detail the relations of individual components of physical activity, including specific vigorous and non-vigorous activities throughout the life course in relation to risk of breast cancer overall and by hormone receptor phenotype will allow further insights into possible biological mechanisms of breast carcinogenesis.
Explaining this is above my pay grade, I’m afraid.
© 2008 Sandy Szwarc