Comments on: Association of Frequency of Organic Food Consumption with Cancer Risk


Comments on: Association of Frequency of Organic Food Consumption with Cancer Risk – Findings from the NutriNet-Santé Prospective Cohort Study

Paper by Baudry et al, JAMA Internal Medicine, October 22, 2018

For those who may be interested, I’ve reviewed this paper in some detail and have also read related reviews by others (listed below). Here’s a quick overview:

The authors conclude, “higher organic food consumption is associated with a reduction in the risk of overall cancer,” and “promoting organic food consumption in the general population could be a promising preventive strategy against cancer.” The authors also make it clear that they believe the differences they found are likely the result of higher pesticide residues in non-organic foods.

The study involved recording of the incidence of several types of cancer over an average of 4.5 years for 68 946 French volunteer adults who, on their own initiative, participated in a large national study on food consumption and health. Participants completed a questionnaire near the beginning about their consumption of 16 categories of foods and the extent to which consumption of each was ‘most of the time’ (assigned a numerical rating of ‘2’), ‘occasionally’ (rating of ‘1’) or ‘never’ (rating of ‘0’) of organic origin. Estimates of individual daily quantitative intake were also collected. Of the 68,946 volunteers, 1350 – or 2% – were afflicted with first-time cancer during the study, and the authors quantify the incidence of several forms of cancer.

The authors make no attempt to hide conflicts of interest, acknowledging an association with the entity, Fond de Dotation Institut de l’Alimentation Bio (an institute for organic food). The literature review includes dubious/erroneous statements such as this: “natural pesticides allowed in organic farming in the European Union exhibit much lower toxic effects than the synthetic pesticides used in conventional farming.” (For information to the contrary, check ).

The authors have been condemned by others for their conflict of interest and apparent bias, but I’ll not do that. Virtually every researcher has inherent bias and this study, like all others, must be judged on its scientific merit, not the personal views of the individuals involved.

The large number of participants is both an asset and a problem. No one can criticize it for inadequate sample size, but at the same time, the huge number of data points complicates the statistical analysis. Trivial differences may be shown as ‘statistically significant’ mainly because of the very large number of ‘degrees of freedom’ (a statistical term for those not familiar).  For example, Table 1 in the study shows people who consume more organic food are shorter at a P<0.001 (less than one chance in 1000 of being due to random chance).

There are a lot of confounding factors in the study. For example, the data show that those who consume more organic food are more likely to be female, older, better educated, wealthier, former smokers, have a history of family cancer, less over-weight, and eat less red meat and processed meat (Table 1). The authors claim to have removed the effect of such confounding by statistical adjustments though they don’t state how this was done; a deep suspicion exists, for both me and others, that serious confounding remains even after adjustment. For example, the most common way to remove the influence of confounding is by linear adjustment. But what happens if the confounding influence is not linear? (We know, for example, that the relationship between age and cancer incidence is nonlinear – an accelerating likelihood as you get older – and those who consumed the most organic food in this study were older on average.)

With all of the comparisons made in this paper between categories of organic consumption and population traits it is certain that some would be found to be statistically significant at P<0.05. Some results in this study are significant at this level – for example, a claimed relationship between organic food consumption and incidence of Non-Hodgkin lymphoma (NHL), at P=0.049 – but is this real or just what would be expected by random chance? (The data show no dosage response between reduced NHL affliction and increased organic consumption except for those consuming the most organic).

And there are some really oddball findings. For example, the incidence of cancer was no higher for those eating a low-quality, non-organic diet than with a high-quality-plus organic regime. However, if the ‘quality’ of the diet improved (less processed meat eaten, as an example), then risk of cancer increased unless there was a corresponding increase in organic consumption. (The implication is that if you don’t eat organic, then you should also eat a lower-quality diet to minimize the risk of cancer.)

While the number of people surveyed in the study is huge, it cannot be termed a representative sample of French society. For one, participants were 78% female. For another, individuals who participated were likely predisposed to respond to a voluntary survey on health. The results cannot be dismissed as meaningless because of this – but neither can they be assumed representative of a population other than those who self-selected to participate.

A major weakness in my view is the assignment of individuals to various categories of organic consumption based on a questionnaire self-completed about 2 months after enrolment and estimates of portion-size consumption made during three 24-hour periods. The assumption is that dietary patterns did not change during the nearly 4 ½ years which followed. There was no apparent effort made to confirm the accuracy of the initial self-assessment process, nor to ensure consumption remained the same a year or two – or four- after the initial assessment.

The results do show statistically significant negative relationships between certain types of cancer incidence and increased organic food consumption – for example, for older females with higher body mass indices who are more highly educated, former smokers and with a family history of cancer. By contrast, the authors note, “When considering different subgroups, the results herein were no longer statistically significant in younger adults, men, participants with only a high school diploma and with no family history of cancer, never smokers and current smokers, and participants with a high overall dietary quality.”

From my perspective, it is hard to understand how organic food would change susceptibility to cancer since most surveys show its nutritional composition to be virtually identical to its non-organic counterpart. See here for example. (There are some exceptions: for example, this study found higher levels of certain fatty acids in organic milk – which is most certainly related to a higher relative perennial forage intake for organically raised cows. The same would likely be achieved with a high-forage, non-organic diet.) While, detectable residues of synthetic pesticides are more common in non-organic crops/foods, the levels are almost always far below concentrations considered to have toxic effects according to some well-established science. And the above excludes recognition of organic pesticides which were not considered in the Baudry paper and which are also toxic at higher concentrations.

Carl Sagan is credited with the statement, “extraordinary claims require extraordinary proof” and that would certainly relate to the claim about organic foods reducing cancer. Unfortunately the paper by Baudry et al comes far from providing extraordinary proof – if, indeed, it provides any proof at all.

Other reviews of this paper include:

Organic Foods for Cancer Prevention—Worth the Investment? Hemler et al., JAMA Internal Medicine, October 22, 2018

Viewpoint: Chemophobia epidemic—Fanning fears about trace chemicals obscures real risks and ‘damages public health’ Jon Entine, Genetic Literacy Project, October 24, 2018

Organic Food Consumption and Cancer Jayson Lusk, Purdue University, October 24, 2018

No, Organic Food Doesn’t Reduce Cancer Risk. That’s Biologically Impossible Alex Berezow, American Council on Science and Health, October 22, 2018

Twitter thread by Bill Price (@pdiff1), University of Idaho, October 24, 2018

Twitter thread by @TamarHaspel, Washington Post columnist, October 23, 2018

Twitter thread by @AlanLevinovitz, James Madison University, October 23, 2018