What’s Wrong With The Golf Course Parkinson’s Study

I searched as diligently as I could, which of course means there might be somebody who could search better, but I found no evidence professional golfers develop Parkinson’s Diseases (PD) at rates higher than non-golfers. One pro, John Senden, has it, and continues to play.

This is important because if “exposure” to the treatments used on the grass at golf courses causes PD, then we’d expect the people with the highest “exposure”, i.e. golfers and groundskeepers, would have the highest rates of the disease. Lack of any reporting on golf celebrities and PD, especially in our celebrity-obsessed culture, may be taken as reasonable evidence, though obviously not proof, that no signal exists.

And that’s of interest because of lurid headlines like this: “OUT OF BOUNDS  Disturbing link between Parkinson’s disease and living beside a golf club revealed by new study“. That is one report of many on the peer-reviewed paper “Proximity to Golf Courses and Risk of Parkinson Disease” by Brittany Krzyzanowski and a slew of others in JAMA Network Open. (Many sent me this study asking me to look into it.)

I put scare quotes around exposure to remind us exposure does not equate to dosage. We have no idea how much of any chemical used to treat grounds made its way into any body. Hence “links” with “exposure” and any outcomes must be classed as correlations, and not causations. But you knew that.

This paper, as we’ll see, is yet another example of what I call the Epidemiologist Fallacy. This occurs when researchers claim or imply X causes Y, but never measure X, and the causal claim is backed with wee Ps.

Here the X is some vague list of pesticides used in groundskeeping at golf course; Y is of course PD.

Exposure evidence is weak here. Here’s how researchers classified “exposure”:

The distance in miles to the edge of the nearest golf course specific to each PD patient and control was identified based on the latitude and longitude of their home address of residency (eFigure 1 in Supplement 1). The home address 2 or 3 years prior to PD symptom onset (for individuals with PD) or index date (controls) was used to calculate distance to allow for a delay between potential environmental exposure and the development of PD motor symptoms.

They grouped people into Disease and Control buckets. How many of the Controls golfed and for what duration we don’t know (you don’t have to live next to a course to golf). Same for the Diseased. So we never can know what dosages anybody got of any pesticide.

The Disease group were 419 PD patients gathered from quite a long period, 25 years from 1991 to 2015. That’s about 17 per year, though this includes some who moved in to the area over that time. This is not a high rate. PD patients were all living in Olmsted County, Minnesota, which had a population of 109,000 (or so) in 1991, and about 165,000 now (from Google). Whatever is causing PD here is not a large problem.

Although the patients all had to be from one single county, the “Controls were identified from the 27-county REP study region in Minnesota and Wisconsin.” That area had 139 golf courses. Controls were people without the disease, of course.

Maybe you can see it already. If not, think a moment before reading further.

How many people had PD in those 26 other counties and how far from golf courses did they live? We don’t know. Researchers only took the healthy people from these other counties (where by “healthy” I mean no PD).

I don’t know what the average distance to a golf course is in Olmstead County, but it contains the city of Rochester. A check of the map indicates 9 courses (but somebody check me). Mower county (bordering to the southwest), about the same size in land as Olmstead, has 3 courses. Fillmore county (bordering southeast, also about the same size) has 3, too. And the same sort of counts can be found for other counties in that region, with more near the Mississippi. Rochester is the largest city until you reach Minneapolis, which wasn’t included in the study. It’s easier to find courses where more people live.

In other words, it’s much easier to live near a golf course in a city in Olmstead county, which is the only place they gathered PD patients. The Control patients, who did not have the disease, on average lived farther away from courses, and mostly outside larger cities.

That means, even if they got the age and sex and race matching on the Controls right, which seems to be roughly the case (see their Table 1), there will be a distinct signal “linking” distance from golf course and PD. Even if pesticides do not cause PD in any way.

I saw other criticisms of this study, waving it off saying older people tend to live nearer golf courses and older folks tend to get PD more. Which is true, but they did the age matching, which weakens this criticism.

Incidentally, I did find one more Abstract (which is a brief report given at medical conferences, often not accompanied by a paper, as it appears this one was not) on the subject: “An Apparent Cluster of Parkinson’s Disease (PD) in a Golf Community” by J Nutt and others. Their conclusion:

Compared to a meta-analysis estimating PD prevalence in North America to be 572/100,000 in persons aged 45 or greater (Marras et al. 2018), the estimate for this community was much higher, although the sample is small and confidence intervals are wide and include the meta-analysis results.

In other words, there was no signal.

Video

Subscribe or donate to support this site and its wholly independent host using credit card click here. Or use the paid subscription at Substack. Cash App: \$WilliamMBriggs. For Zelle, use my email: matt@wmbriggs.com, and please include yours so I know who to thank. BUY ME A COFFEE.