Junkfood Science: A cure is rarely found inside a computer

June 21, 2007

A cure is rarely found inside a computer

The news reported that pesticides may raise the risk for Parkinson’s disease. It is curious why the news chose to headline that correlation — when the study couldn’t find a link to pesticides that was greater than chance, a fluke or statistical error — and chose to ignore the single only tenable correlation.

According to Reuters Health:

Pesticides may raise Parkinson's risk

...It's likely that both genetic and environmental factors lead to Parkinson's disease, Dr. Finlay B. Dick from Aberdeen University, UK, told Reuters Health. “While pesticides have been shown to be a risk factor for PD, the risk of the disease is only modestly increased," he noted. Dick and colleagues investigated possible associations between Parkinson's disease and environmental factors in five European countries. Three factors emerged as having a significant impact on the risk for Parkinson's disease, the investigators report in the journal Occupational and Environmental Medicine.

First, low exposure to pesticides increased the risk for Parkinson's disease by 13 percent, and high exposure increased the risk 41 percent.

Second, having been knocked unconscious conferred an even higher risk, the researchers report, with a 35 percent increase for having been knocked unconscious once and a 153 percent increase for having been knocked unconscious more than once.

Third, tobacco use appeared to protect against developing Parkinson's disease, the report indicates, reducing the odds by half.

“This study has provided important evidence of the increased risk of Parkinson's disease in relation to exposure to pesticides," the team concludes. The fact that the risk increased with greater exposure “suggests that pesticide exposure may be a causative and potentially modifiable risk factor."

Junkfood Science readers probably already caught a number of the flaws in this conclusion and story. But we’ll fill in a few additional blanks the media didn’t report.

These researchers compiled a list of Parkinson patients from five regional centers in Scotland, Sweden, northern Italy, eastern Romania and Malta. The diagnoses were only confirmed in two of the centers, however, and they used hospital records for the rest. The researchers noted that studies of environmental factors to date have been contradictory, inconclusive, plagued by small sample sizes and methodological problems, and have failed to demonstrate a solid link. For this study, lifetime exposures to pesticides and other chemicals were not measured in any of the people, but were estimated using their occupations and then modified by “subjective exposure models that estimate likely workplace exposures” that used factors such as likely ventilation and protective measures employed. In other words, with these exposures, they were looking at correlations of correlations. There could be countless other factors among different occupational and socioeconomic groups they didn’t include or consider.

The data on the Parkinson patients were compiled by administering a questionnaire asking them to recall their lifetime employment and hobbies; history of smoking and alcohol consumption; episodes of private water supply use by both duration and geographical location; education; drugs taken for anxiety, depression or sleep (although they didn’t ask if such use predated diagnosis or followed it); history of having been knocked unconscious (although they didn't ask if this was before diagnosis or the result of a fall after developement of PD), and family history of Parkinson’s disease. They got 64% of the questionnaires back for a total of 959 cases. All of this information was put into a computer and their model estimated odds ratios for correlations in the PD patients as compared to a control group.

We know that in epidemiological studies such as this, most scientists do not view as tenable any relative risk of less than 3. In other words, correlations must be at least 200% greater than baseline before they’re considered sufficiently valid to even be reported. And the weaker the data used, the still stronger any correlations must be to be taken seriously. And, no matter how strong any correlation, it can never show causation.

So, they came up dry on virtually everything — they were nonfindings. It’s valid to question why they chose to report pesticides and in fact made it their first “Main Message” point in their study. Instead, they could have reported any number of other nonfindings, such as 33% odds ratio for sleeping pills, 95% for anti-anxiety medications or 92% for antidepressants...or that they derived a 12% lower risk associated with the highest exposures to solvents and 6-8% lower risk with exposure to heavy metals, manganese and copper...or that nonsmokers had double the risk. Or, they could have made this study part of an anti-boxing scare story, if they'd wanted.

There was only one reportable correlation in this study, yet it was the one not reported!

Having a family history of Parkinson’s disease was associated with a 4.85 odds ratio — a 385% greater risk.

This study didn’t add any new finding to help reach a treatment or cure for Parkinson’s disease. Scientists don’t yet know what causes PD, according to the Parkinson’s Disease Foundation. It’s one of those tragic neurodegenerative diseases that afflicts between 0.1% and 0.3% of the population, mostly men over age 60. But studies have consistently shown that having an immediate family member with it increases one’s risk by two to three-fold. Beyond that, it’s all unsupported speculations.

But if you or a loved one has PD, you’re likely hoping that researchers are pursuing concrete correlations to track down a cause and cure, rather than chasing after nonfindings because they may be more politically popular to believe.

Bookmark and Share