Epidemiology is a branch of research which analyses data to look for associations between things. This brand of research will very often throw up ‘facts’ like ‘red wine lowers risk of heart disease’, because of evidence that red wine drinkers are at reduced risk of heart disease. However, epidemiological evidence can never tell us if red wine actually reduces the risk of heart disease. It might be, for instance, that lower risk of heart disease seen in red wine drinkers is actually because they smoke less or eat more healthily than those who don’t drink red wine. In other words, the relationship between red wine drinking and heart disease may have nothing to do with the red wine per se, but due to other factors (referred to as ‘confounding factors’) that are common in red wine drinkers.
I don’t have much faith in epidemiology but, like it or not, I recognise a lot of common beliefs about health come from this type of research.
Many concepts that are taken as ‘fact’ and feel intuitively right are based on epidemiological research. A classic example relates to the benefits of physical exercise on weight control. It’s been noted that people who are more physically active have a lower risk of being overweight than those are more sedentary. For many this observation will tend to reinforce the preconceived idea that exercise is good for weight control.
And this might be right, but it’s also perhaps true that thinner people may be more inclined to take exercise compared to larger people. Actually, there is some evidence in children that this may indeed be the case: one study which followed children over time found a link between sedentary behaviour and increased weight, but that the weight gain preceded the sedentary behaviour . In other words, weight gain appeared to cause the sedentary behaviour rather than the other way round.
The particular type of research used here is known as a ‘prospective cohort’ study where a group is followed over time and developing associations are looked for. The other major type of epidemiological studies are known as ‘case-control’ studies. An example of this would be researchers taking a group of overweight children, and then looking back in time to find out how active they have been compared to their slimmer counterparts. This particular study would not be able to discern which came first (the sedentary behaviour or the increased weight).
It is generally accepted that prospective cohort studies are quite superior to case-control studies, though neither will tell us about cause and effect (they only tell us that two things are associated with each other, not that one is causing the other).
If you’re wondering where I’m bothering with this, it’s because I want to discuss a recent piece of research which used different types of epidemiological evidence to assess the relationship between constipation and risk of ‘colorectal cancer’ (cancer of the large bowel and/or rectum). The idea that constipation causes bowel cancer is quite firmly-entrenched in many people’s minds (including many doctors). That was certainly the case for me. I remember at medical school being taught about the research of Dr Denis Burkitt, who worked in Africa in the middle of the last century and observed lower rates of colon cancer there compared to people in the West. He put at least some of the difference down to dietary fibre. I remember at medical school being shown slides of typical stools (faecal waste) from someone living in the West (cigar-shaped) and Africa (more like a cow-pat).
My naïve and not very scientifically-savvy brain lapped this stuff up. Only later did I realise that the idea that constipation causes colon cancer was based on really very weak case-control-type epidemiological evidence which, worse still, compared people in different countries (this enhances the risk of confounding factors coming into play).
In the recent research I want to write about today, several similar studies looking at the relationship between constipation and colorectal cancer were amassed . The researchers found that, overall, case control studies (remember, a weak type of epidemiological evidence) did find constipation to be associated with a significantly increased risk of colorectal cancer.
However, they found this was not the case with prospective cohort studies (the stronger type of epidemiological evidence).
They analysed another sort of epidemiological evidence too, known as ‘prospective cross-sectional’ surveys, which also have the benefit of following people over time. These studies actually found that constipation was associated with a reduced risk of colorectal cancer.
So, this review looked at three types of epidemiology, and each one threw up a different result. In a way, I think this review highlights the unreliability of epidemiology. This study also found that the best epidemiological evidence does not support the idea that constipation can cause colon cancer. The authors write: “Prospective cross-sectional surveys and cohort studies demonstrate no increase in prevalence of [colorectal cancer] in patients or individuals with constipation.”
So, what might explain the association seen in the weaker, case-control studies? The authors offer up a potential explanation in the form of ‘recall bias’. This is essentially people being more likely to recall events because of pre-conceived ideas. If you’ve got it in your head that constipation is a cause of colon cancer, then once you have this diagnosis you are perhaps more likely to say you’ve suffered from constipation (compared to someone who does not have colorectal cancer).
Epidemiology is not completely devoid of value, but it is seriously hampered by its very nature. The real test of a theory is to expose people to things and compare the effect to a non-exposed group. For example, we could give people fibre supplements over the long term and see if this reduces the risk of growths in the colon compared to those who don’t take the supplement. Actually, two such studies have been done, and neither found any benefit from the fibre [3,4].
1. Metcalf BS, et al. Fatness leads to inactivity, but inactivity does not lead to fatness: a longitudinal study in children (EarlyBird 45). Arch Dis Chil 2011;96(10):942-7
2. Power AM, et al. Association Between Constipation and Colorectal Cancer: Systematic Review and Meta-Analysis of Observational Studies. Am J Gastroenterol advance online publication 12 March 2013; doi: 10.1038/ajg.2013.52
3. Jacobs ET, et al. Intake of supplemental and total fiber and risk of colorectal adenoma recurrence in the wheat bran fiber trial. Cancer Epidemiol Biomarkers Prev. 2002 11(9):906-14
4. Alberts DS, et al. Lack of effect of a high-fiber cereal supplement on the recurrence of colorectal adenomas. Phoenix Colon Cancer Prevention Physicians’ Network N Engl J Med. 2000;342(16):1156-62