Junkfood Science: What are polls and surveys again? Marketing

October 12, 2008

What are polls and surveys again? Marketing

No, this post isn’t about those polls — we’ll leave those to the legal professionals already taking critical looks.

This is about polls masquerading as medical research. A new survey published in the current issue of the journal Obstetrics & Gynecology presented no actual science or new information that might benefit patients or their healthcare providers. Its purpose appeared to be to heighten alarm over obesity and cancer, and chastise fat women. See what you think…

Survey design

As has been exampled numerous times, polls and surveys are tried-and-true marketing techniques — ones that aren’t often appreciated and people can easily get swept up in them without even realizing they’re being manipulated. Most every public relations and advertising firm uses polls and surveys because they create the bandwagon effect and take advantage of our natural tendencies to get behind ideas seen as popular and what everyone else thinks. But, of course, surveys don’t measure opinions as much as they shape them. It’s enormously easy for a pollster to design polls to create any consensus the pollster wants to promote, simply by how the questions are worded, who is selected to be interviewed, how the questions are asked, and how the results are interpreted and reported (and what isn’t reported). Pollsters can define the issue simply by how they select and phrase the questions and restrict the choices.

The survey reported in this week’s news had examined “women’s lack of knowledge” about excess weight and risk for endometrial cancer. The researchers were led by Dr. Pamela T. Soliman, M.D., MPH, with the Division of Quantitative Sciences at M.D. Anderson Cancer Center and the University of Texas Health Sciences Center in Houston. They approached women at local health fairs and festivals, churches and other organizations. The survey was also made available in the waiting rooms of obesity, health and gynecology clinics around Houston, Texas, as well as on M.D. Anderson’s website. Between July 2005 and May 2005, a total of 1,545 women completed the survey. Nearly half (45%) of those surveyed were classified as ‘obese’ and another 24% were in the ‘overweight’ classification. At the completion of the survey, the women were given an educational pamphlet describing the risk of endometrial cancer associated with excess weight from M. D. Anderson Cancer Center.

The survey itself asked the women their age, ethnicity, education level, household income, insurance, height and weight. Concerning the study’s objective, it asked [click on image to enlarge]:

Topsy turvy

In interpreting the survey results, the authors said that women who responded that there were increased risks associated with obesity for cancers were considered knowledgeable of obesity’s cancer risks. The authors said that they then compared knowledge of obesity and risk for endometrial cancer risk to knowledge of obesity and risk for colon and breast cancer. Finally, they compared knowledge of obesity and cancer risks to the women’s BMI and demographic information.

The working assumptions of this paper, however, basically made the correct answers wrong, and the incorrect answers right, which only lends support of a focus on something other than an objective scientific investigation.

In the study introduction, the authors wrote:

It is well known that obesity increases risk for multiple medical problems... Recently, several studies have shown that obesity also increases risk for certain types of cancer….In addition, increased body weight has been shown to increase mortality due to multiple cancers including endometrial, colon, and breast cancer, with the highest RR for death associated with endometrial cancer.

Notice how associations with obesity (“risks”) were turned into active causations attributed to obesity? To support a link between obesity and cancer incidences and cancer mortality, they could only reference some of the weakest, most outdated and disputed studies in the scientific literature.

● One study was the first American Cancer Society study, which had simply looked for correlations between body weights and mortality among 750,00 adults selected from the population during 1959-1972. Not only are its figures outdated, but missed consideration of countless confounding factors.

● The second study had concluded that it provided evidence that obesity increases cancer deaths. While it's popularly referenced, this controversial study has not held up to peer review by the medical community. In this 2003 study by the American Cancer Society, volunteers around the country had enrolled nearly 1.2 million people to fill out questionnaires, with none of the data verified. The authors then excluded from their analysis 90% of the cancer deaths (including all those among women at the low end of the BMI scale). Their analysis, based on 11,648 cancer deaths in women between 1982-1998, had actually still been unable to find any tenable correlations between any of the body weight classifications and all cancer deaths over 16 years. This study reported higher relative risks of obesity among women associated with endometrial cancer (based on 704 deaths) and lower risk of lung cancer (based on 5,349 deaths, a far more prevalent and deadly cancer); higher relative risks associated with pancreatic cancer and lower risks associated with stomach cancer; higher relative risks associated with kidney cancer and significantly lower risks associated with brain cancer; and higher relative risks associated with cervical and lower relative risks associated with leukemias. They found no tenable correlations with any body weight category with any of the other cancers, including colorectal cancer. Again, this study found no tenable correlation between BMI and overall cancer deaths.

● Finally, they cited a methodological weak case-control study done 16 years ago in which the authors had selected 405 cases of endometrial cancer and compared selected risk factors among them to those in among a group of 297 cancer-free women.

Why did the M.D. Anderson authors not do a more rigorous search of the medical literature and include larger, stronger and more recent research, all of which counter their working assumptions? For example, 2,471 studies on 17 cancers were reviewed by the World Cancer Research Fund and American Institute for Cancer Research last fall, and found no tenable associations between cancer incidents or deaths and BMIs.

The authors didn’t include the Million Women Study which had examined the BMIs of 1.22 million women and the incidences of 17 cancers and deaths between 1996 and 2001. This study did adjust for known confounding factors and found no tenable associations between BMI and incidences of all cancers. Some cancers risks fell on one side of null, some on the others, but none of the relative risks were tenable or exceeded random coincidence or statistical error.

Nor did the authors mention the largest study from CDC senior research scientists at the National Center for Health Statistics, which had examined the associations between body weights and deaths. This study used National Health and Nutrition Examination Surveys (NHANES) done from 1971 to 1994, which had involved in-person health physical examinations and body measurements done on a nationally representative sample of the U.S. population, and U.S. vital statistics deaths. Even over more than 30 years, they were unable to find any tenable associations between overall cancer deaths and BMI — some associations with obesity fell slightly on one side of null, some on the other, leaving no increased risks overall. The CDC scientists concluded: “Our results showed little or no association of excess all-cancer mortality with any of the BMI categories. None of the estimates of excess deaths was statistically significantly different [from null].”

Survey says

Dr. Soliman and colleagues reported that they found no association between ‘knowledge’ of obesity-related endometrial cancer risk and the women’s education, age, racial/ethnic group, household income, insurance, or weight. Women with higher levels of education and insurance were more likely associated with beliefs of an increased risk for breast cancer and colon cancer associated with obesity, but even this modest correlation was untenable.

“There was no evidence of an association between racial/ethnic group and ‘knowledge’ of increased risk associated with obesity for endometrial, colon, or breast cancer,” they wrote. However, “for each cancer, black women were the most likely to answer that they did not know about the increased risk for cancer associated with obesity.” [Secure your neck brace here.] “These finding suggest that black and/or hispanic women, who may be at higher risk for endometrial cancer due to the increase prevalence of obesity, could be potential target groups for educational programs.”

In other words — while the strongest body of evidence continues to show no relationship between body weight and overall cancer risks, and these authors found no correlation between race/ethnicity and women who said they believed obesity was related to increased risks for cancers — the authors conclude that minority women are promising targets for programs to convince them of a risk!

Their conclusion stated:

This study provides continued evidence of the gap in knowledge within the general population regarding the health risks, and in particular the cancer risks, associated with obesity. In 2001 the U.S. Department of Health and Human Services and the Surgeon General made a call to action to prevent and decrease overweight and obesity… efforts were made to target lower socioeconomic and minority population groups who were thought to be at highest risk…. Based on our findings, there is a significant lack of awareness of the relationship between obesity and cancer risk and, particularly, endometrial cancer risk. Although an effective intervention still needs to be developed, obstetrician– gynecologists should start by educating their patients about the risk of cancer associated with obesity.

According to the authors, counseling to support improvements in diet and physical activity is considered a first-line intervention and should be initiated by the obstetrician–gynecologist. They added:

While our study findings are compelling, this is only the first step for improving public education regarding this important issue. Once educational interventions are developed, it will be important to evaluate their effectiveness on both awareness of obesity-related cancer risk and how obese women in the population will respond to this information. Will knowledge of this increase in cancer risk among obese women change their behavior?

In other words, these authors appear to be promoting beliefs that larger body sizes among women are due to bad behavior and that scaring women about cancer might incentivize them to lose weight. These opinions, however, were unsupported by any evidence in their study or in the body of the medical literature.

● There is no evidence that naturally larger women who lose weight thereby change their risks associated with cancers.

● Nor is there evidence of a safe and effective way for women to permanently change their body weight classifications — through good behavior, diet or exercise.

● There is no evidence that using scares is a healthful, effective or compassionate way to improve anyone’s health. In fact, just the opposite. Negative attitudes about fat women among healthcare professionals have also been documented for decades and shown to not only adversely affect the medical care they receive, but that weight stigma felt by fat women results in them delaying seeking medical care, thereby jeopardizing their health.

And when scares are not evidence-based, fear mongering is incompatible with the trusted ethical role of medical professionals to provide helpful and sound information for all patients.

Reading the media coverage of this survey, it appears its only purpose was to pile on with more cruel and damaging scares about obesity. It heightens blame on fat women that if they get cancer, it’s their own fault.

For women and their loved ones — news you can use

The factual information about endometrial cancer offers positive, empowering and reassuring information that all women can use to better protect themselves.

According to the National Cancer Institute, endometrial cancer is nearly always found early as it causes symptoms at an early stage, when there is a good chance of recovery. All women can help to protect themselves simply by knowing the symptoms and seeking medical care from a licensed gynecologist should they experience them.

This is a helpful health message.

The symptoms are: any unusual bleeding, spotting or other discharge that lasts more than two weeks.

“About 90% of patients diagnosed with endometrial cancer have abnormal vaginal bleeding such as bleeding between periods or after menopause,” according to the American Cancer Society. The other 10% have unusual discharge, especially after menopause. Pain, palpable masses and weight loss are late stage symptoms.

More than 95% of endometrial cancers occur in older women, age 40 and older, according to patient literature from M.D. Anderson Cancer Center. Women with a personal or family history of Lynch syndrome (hereditary non-polyposis colorectal cancer) and certain ovarian diseases, or are on tamoxifan, which increase estrogen are also advised to be especially attentive to abnormal bleeding.

There is no benefit in scaring asymptomatic women into worrying about endometrial cancer. There is no known way to prevent it or recommended screening test for endometrial cancer. No medical body, including the American Cancer Society, recommends screening. A recent international review of the medical literature on the effectiveness of various gynecological screening methods, published in the European Journal of Surgical Oncology, concluded that “presently, screening for endometrial, vaginal and vulval cancers is not justified.”

“Currently, screening for endometrial cancer is not advocated as most women present with symptoms in early disease with good survival outcomes,” the authors stated.

This was the same conclusions of the National Cancer Institute Summary of Evidence on Screening for endometrial cancer for health professionals: “Routine screening of women for endometrial cancer is not of any proven benefit.” Any recommendations for screening of certain groups of women, it stated, “are based on opinion regarding presumptive benefit” rather than clinical evidence.

This Summary of Evidence also provided population data which counters obesity scares. It stated that, as obesity rates purportedly rose: “age-adjusted endometrial cancer incidence in the United States declined steadily between 1978 and 1988, and has remained fairly constant since 1988.” Endometrial cancer deaths have also steadily declined from 1974 to present, it stated, with a 26% drop in deaths from this cancer.

While there is no known way to prevent endometrial cancer for most women, women who take hormone replacement therapy are advised to use an estrogen with progesterone, which has been shown to help reduce abnormal growth of the uterine lining. There was a blip of an increase in incidences of endometrial cancer in the mid-1970s with hormone replacement therapy before the risks of unopposed estrogen were recognized, according to the NCI Summary of Evidence.

Wouldn't a more valuable survey question to have asked women - of benefit to them and healthcare professionals - is how many knew the symptoms of endometrial cancer and when to seek medical care?

Bottom line, surveys and polls are nearly always marketing, regardless of how credible their source might seem. They are done to manipulate you into believing or doing something someone else believes, but opinions never make scientific facts.

© 2008 Sandy Szwarc

Bookmark and Share