Wednesday 14 May 2008

LIES, DAMN LIES AND STATISTICS

Back in March, we brought to the attention of our French speaking readers the misinformation on obesity because of various statistical biases that cloud the issue. (Please refer to our article EST-CE QUE L'ALARME QU'ON SONNE EST VRAIMENT JUSTIFIÉ ? ).

Among other observations, we exposed Statistics Canada’s own words of caution on how to interpret their statistics when analyzing weight trends in Canadians. Here’s what they have to say about this particular bias right in their website: ‘’For more than a decade, information about the weight of Canadians has been based on self-reports, that is, survey respondents reported their own height and weight rather than being measured and weighed. However, such data are known to underestimate the prevalence of overweight and obesity.9, 10, 11, 12 The 2004 Canadian Community Health Survey: Nutrition (CCHS), which directly measured respondents’ height and weight, makes it possible to draw a more accurate picture (see Data sources and analytical techniques).’’

It is now Ian Janssen’s, Queen's University obesity expert, turn to tell Canada AM, that weight prevalence studies are unreliable. He claims that the bias tends to tilt the scale more to the light than the heavy side but we have of course learnt to expect such statements from ‘’obesity experts’’. The fact remains that comparisons cannot be drawn from these statistics as they are fatally flawed due to errors and biases much like many epidemiological or other statistical studies that they throw our way.

It appears from this article that these problems are very well known within the public health community yet they allow figures, estimates, cautions, fear mongering etc… to reach the mainstream media, alerting populations and the medical community, based on nothing more than flawed statistics. Some of these same public health experts dare call citizens who question their wisdom, ''denialists'' and wonder why the public is steadily losing faith in anything – potentially true or false -- that public health spews !

Canadians are skewing the national obesity stats

CTV.ca News Staff

Canadians are miscalculating their weight and height, resulting in faulty national statistics.

Two different studies looked at the effects of data collected through Statistics Canada's Canadian Community Health Survey (CCHS) and National Population Health Survey (NPHS). The studies relied heavily on self-reported weight and height information by respondents in order to estimate body mass index (BMI). The information was then used by StatsCan determine national obesity rates.

"People tend to over-report their height and under-estimate their weight and that results in huge problems," Ian Janssen, Queen's University obesity expert told Canada AM.

As people get older, the reporting errors tend to increase as well.

StatsCan discovered that men who responded to the 2005 CCHS survey tended to over-estimate their height by roughly one centimetre and women by about half that much.

Weight miscalculations were more pronounced. Women tended to under-report their weight by about 2.5 kilograms on average, and men by 1.8 kilograms. The study found the heavier the person, the more they under-reported.

In total, the study found 33 per cent of overweight individuals and 48 per cent of obese individuals (class I) were placed too low on the BMI index. Consequently, prevalence of obesity in Canada and its rate of change over time are unknown.

Furthermore, the relationship between obesity and disease such as diabetes, high blood pressure and heart disease may be distorted.

For example, among those classified as obese based on self-reported data in the 2005 CCHS, 360,000 people aged 40 years or older had diabetes. But among those classified as obese based on actual measured values, 30,000 people -- nearly 50 per cent more -- had diabetes.
But Janssen told CTV he is most concerned by trends he sees in children. Obesity rates among this age group have tripled in the past 25 years.

"Statistics suggest the burden of obesity will get worse in the next 20 or 30 years," said Janssen, making accurate surveillance of the problem more crucial than ever.

BMI is calculated by a person's weight in kilograms divided by their height in metres squared. The clinical definition "overweight" is a BMI reading of about 25. A BMI above 30 is considered "obese."

Statistics Canada has been using self-reported data since the 1990s and is currently working on methods of addressing the bias.

According to previous surveys, 5.5 million adults were obese in 2004 and 8.6 million were overweight.

No comments: