Chapter 5 Results

We’ll start our journey through the exciting world of digestive system cancer by looking at the trends in cancer incidence and mortality in recent years. Then, we’ll break down (digest) trends in diet at the level of food group and top food sources. That’ll give us a sense of each data set separately before we dive more deeply into possible relationships between diet and digestive cancer.

5.3 Diet and Digestive Cancer

The destination of our journey. Let’s see how foods, food groups, and caloric intake relate to digestive cancer.

5.3.1 Cancer Incidence

Even using a smoother to draw a line through the data, the correlations are unclear using the raw caloric intake values. We can see that both “sugar and sweeteners” and “flour and cereal products” correlate with digestive cancer incidence but the trends of the other groups are unclear. We’ll use percent difference from 2000 per capita caloric intake from now on to show the trends more clearly.

We found that the decrease in total calorie intake correlated with the decrease in digestive cancer incidence. We also found that food groups such as “sweeteners and sugars” and “cereal/flour products” positively correlated with digestive cancer incidence. Surprisingly, we found that fruits caloric intake and vegetable caloric intake correlated positively with digestive cancer incidence. We think that since total caloric intake is associated with cancer incidence, the other food groups appear to be correlated as well.

Looking closely at this scatter plot matrix, our hunch about other major food groups being correlated with total caloric intake was correct. All major food groups trend upwards when compared to both total caloric intake and total caloric intake excluding fats. That doesn’t mean we shouldn’t look at specific foods more closely. To a food-poor nation, high-calorie food sources are gold. To a food-rich nation, the culture or status associated with these foods (or, the markets, supply chains, and industries that make them so available) might drive the overall trends we’re seeing.

Decomposing the food groups into the top 25 foods by caloric intake yields more interesting results. As expected, high fructose corn syrup and beef are positively associated with digestive cancer incidence. Ice cream, potatoes, and orange juice trend positively with incidence, too. On the other hand, we found that butter, peanuts, cane/beet sugar, and mozzarella trended negatively with cancer incidence, meaning the higher the calorie intake of these foods, the lower the cancer incidence.

Each of these trends are worth investigating in their own right. Given the complexity of the problem, it isn’t likely that these data are saying that butter, or peanuts, or mozzarella cheese are protective against digestive cancer. But they might represent clusters of diets or cultures with sets of behavior and/or genetics that are protective against digestive cancer. For the most part, we’ll leave those extrapolations to the reader. For a couple of the most surprising results (namely, fruits and vegetables … super foods!) we’ll look more deeply into the data.

The clearest take-away here is that when the summary data say “vegetables” they really mean “potatoes.” Everything else, by volume of caloric intake, has a much smaller effect, but is still interesting to see. And of course, potatoes are strongly associated with digestive cancer. Head lettuce and sweet corn are also strongly associated with digestive cancer incidence while sweet potatoes, broccoli and chili peppers are not.

The same story can be told about fruits, but maybe not as well. Orange juice is strongly associated with digestive cancer while also making up a large portion of calories by fruit. Apple and pineapple juice, as well as a number of canned fruits, show similar results. In a sense, these food sources are smuggling added sugar into the fruits category in the summary data above. That doesn’t let the rest of the fruits off the hook. Some are “good” (strawberries, avocados, cranberry juice) while others are not (peaches, pears).

5.3.2 Cancer Mortality

These graphs are a bit of a sanity check. Our cancer incidence and cancer mortality data are very similar, and we would expect to see the same results if we had compared our calories data to mortality instead of incidence.

Albeit to different degrees, that is indeed the case.