Vitamin D and diabetes: An over-simplified solution to a complex problem
"Pluralitas non est ponenda sine necessitate," I say.
No, no, don't flee in fear of Latin, I am merely referring here to that well-known notion called Occam's razor. Its guiding principle is often summarized as: "All other things being equal, the simplest solution is the best."
What I want to show you in this column is the confusion that comes about in science and the reporting of science when you try to find a simple solution to what increasingly looks like a complex problem.
Think of it as being sliced and diced — and then fricasseed — by Occam's razor.
That issue at hand is the attempt to narrow down the cause of Type 1 diabetes — formerly known as juvenile diabetes — which has a variety of incidence rates around the world. Unlike Type 2 diabetes, whose cause is strongly associated with excess weight, up until now scientists have not been sure what causes Type 1.
Then in late June what was described as a "new" study came out that linked increasing rates of the disease in children to the amount of sunshine they are exposed to. Ergo, the closer to the equator you lived, the less likely you were to get the disease.
The paper then suggested that this could be linked to lower levels of vitamin D, which sunlight creates in the body, and accordingly vitamin D deficiency is the likely cause of Type 1 diabetes.
And this meant a potential cure was at hand.
"The research suggests that Type 1 diabetes may be preventable with a modest intake of vitamin D3 for children, ideally with five to 10 minutes of sunlight around noontime when good weather allows," said Cedric Garland, University of California at San Diego professor of family and preventative medicine, and one of the authors of the report. And accordingly, doctors and nutritionists should prescribe vitamin D to all kids living in higher latitudes "without further delay."
U-shaped curve, a 'smile'
To help make the argument, the study had a graphic showing a U-shaped curve, which one reporter described as a "smile" graph. It was alleged to show low diabetes Type 1 levels in kids living near the equator and higher levels as you moved toward the less sunny poles.
So what's the problem?
The first thing to know is that the American authors themselves hadn't generated any of their own data, but had taken incidence levels collected around the world and first published by Finnish researchers in a 2000 paper.
Only, when you looked carefully at that data you found yourself scratching not just your head, but every scratchable body part.
It turns out that the area on the planet with the highest level of Type 1 diabetes is the sunny, sunny Italian island of Sardinia. Novosibirsk in Siberia (think a locale situated north of Flin Flon), had an incidence rate less than one-third that of Kuwait, two-and-a-half times lower than Puerto Rico, and below that of Sao Paolo and parts of Tunisia.
Huge disparities existed between what amounted to next-door neighbours. The rate in Puerto Rico was about 17.4 new cases yearly per 100,000 kids, but in Cuba the rate was 2.9 new cases.
Ethnicity made a monumental difference. No place in China had a high rate, even when you travelled to Harbin — which is north of most of Mongolia and on a line with Ottawa. Nonetheless, Harbin's rate was less than a quarter that of Hong Kong, which is situated more than 2,800 kilometres to the south.
Not to mention that the estimated rates in the Finnish paper for Sao Paolo were roughly the same as in Tierra del Fuego, some 5,000 kilometres to the south.
With this jumble of data in hand, the Finnish researchers rather humbly wrote: "The earlier reported polar-equatorial gradient in the incidence does not seem as strong as previously assumed, but the variant seems to follow ethnic and racial distribution in the world population."
In other words, while there is some general increase as you move toward the poles, it now looks as if something genetic is the chief cause of this disease.
Doing the math
Finally, the American paper had an odd bit of statistical jiggling. The Finnish paper reports on boys and girls, and then gives a combined average yearly incidence number for both. The California paper only produced the boys' data.
What's going on here? What explanation did the San Diego group have for the high rates in Sardinia, the low rates in Novosibirsk, and the Puerto Rico-Cuba paradox, the Sao Paolo and Tierra del Fuego disconnect?
I e-mailed Garland and asked, and he replied that the data was confusing. "We are extremely interested in the high rates of Type 1 diabetes in Sardinia. It is anyone's guess about the reasons. Our guess is that it could be that some residents of the island have an unusual dietary intake pattern, or a higher-than-usual prevalence of an inborn defect in metabolism of vitamin D that interferes with absorption, metabolism or action of vitamin D."
As to Novosibirsk: "Here we also must guess, as there is no international database on vitamin D supplements. Low rates at high latitudes are likely to be due to vitamin D supplements or cod liver oil that are given to children to prevent rickets."
Puerto Rico and Cuba? Again a guess: "Puerto Rico, as a U.S. territory, has a modern development level. Air conditioning is available in the cities, and many people work indoors and are free from most required outdoor work that involves exposure to the noonday sun," he replied.
"Cuba is less developed from an economic perspective. Many Cubans must work outdoors in agrarian activities, such as sugar cane culture and cutting, which requires them to be outdoors in the sun near noon. These factors could increase the dose of UVB for Cubans per unit of solar UVB available compared to citizens of Puerto Rico."
The paper itself tried to explain the high rates in Kuwait as a function of people — most notably women — covering up and not getting high doses of sun exposure.
Clearly, nothing seems simple or obvious. So why the two different interpretations of the data?
"Different people sometimes see different patterns. We recognized a signature pattern of data points by latitude that is in general associated with vitamin D deficiency diseases," Garland wrote in an e-mail. There was no explanation of why the Finnish interpretation hadn't been mentioned.
But the most interesting answer related to why the boys were used. "Because Type 1 diabetes occurs equally in males and females, we chose to display data only on males in order to save space," he said.
That really doesn't make any sense, because all you have to do is put in a single combined data point. But I have another explanation — Tierra del Fuego. A tiny survey there found only boys with the disease — four of them.
Far end of smile flattened
It was the only group in the survey of 51 regions for which there were no male/female numbers provided. This created an incidence rate of 20 per 100,000. The Finnish group responded to this by an averaging of males and females — my guess based on the rates in the rest of the country. This reduced the new incidence rate to eight per 100,000 — the same as in Sao Paulo.
If you use the eight figure, the far end of the smile graphic which now goes through Tierra del Fuego gets flattened. The visual punch of the findings, the cause-and-effect line, dissolves.
Given all these contradictions, complications and confusions, why does Garland insist on a vitamin D connection? Why use an Occam's razor approach to reduce diabetes Type 1 to a simple explanation when the data doesn't suggest simplicity?
You must understand that Garland and his brother Frank have been arguing since the early 1980s that lack of vitamin D causes cancers of various sorts. They are now part of a group of scientists and nutritionists leading the charge to increase vitamin D dosages for people for a host of health reasons. That is to say, they look at the world through a vitamin D deficiency lens, and I think what they see in the data is what they want to see.
While there is a statistically valid trend toward increasing Type 1 incidence as you leave the poles, there are literally dozens of places that go against that trend. Accordingly, it seems to me to insist on a vitamin D/sunlight connection is to turn Occam's razor into Occam's sledgehammer.
Media ignored apparent disconnect
And how did the media report this tangle, these contradictions, these mysteries? How did they temper the paper's "take vitamin D now" message for parents worried about their children's future health?
I could not find a single media source that reflected on the apparent disconnect between sunlight and disease incidence. The words China or Cuba or Sardinia or Novosibirsk never appeared anywhere (media coverage can be seen at Google News using the search terms "Garland + vitamin D + diabetes"). No one talked about the contradictions highlighted by the Finns in their paper — a paper, by the way, freely available on the internet. No one seemed to ask why only boys were included in the graph.
Why? My only conclusion is that it is easier to report and get published a simple truth than a complicated one, and that when given a chance journalists would rather look at headline-pleasing scientific conclusions than headline-displeasing scientific data.
Think of that as the basis of what I call Strauss' Media Blindfold Principle. And think of me as an old data grump.