White rice and risk of type 2 diabetes

BMJ 2012; 344 doi: http://dx.doi.org/10.1136/bmj.e2021 (Published 15 March 2012) Cite this as: BMJ 2012;344:e2021
  1. Bruce Neal, senior director
  1. 1George Institute for Global Health, University of Sydney, PO Box M201, Sydney NSW 2050, Australia
  1. bneal{at}george.org.au

New study highlights old challenges of nutritional epidemiology

A marked rise in the incidence of type 2 diabetes is a characteristic feature of populations that have undergone nutritional transition.1 First seen in the developed Western world, the same pattern is now being replicated in low income and middle income countries, where most new cases of diabetes now occur.2 The role of poor diet in the diabetes epidemic is undisputed, but the details have long been debated.3 Although overconsumption of energy and accumulation of excess body fat is a common cause of type 2 diabetes,4 diet almost certainly has other unknown effects, and specific foods with particular adverse effects may have a direct role in the development of type 2 diabetes.

The linked paper by Hu and colleagues (doi:10.1136/bmj.e1454) sheds new light on the relation between diet and diabetes, and the authors bring a new level of rigour to efforts to answer the question of the possible effect of higher consumption of white rice.5 They systematically searched for prospective studies of the effect of consumption of white rice on the risk of diabetes and quantitatively summarise the findings of individual studies. The study includes a large number of incident cases of diabetes recorded over a long follow-up period and reports a dose-response association between white rice consumption and risk of diabetes. However, as is common in nutritional epidemiology,6 interpretation of the observed association and, in particular, determination of the likelihood of causality are problematic.7

It is often difficult to quantify consumption of the dietary component of interest and collect adequate information about factors that might confound the association. Large scale studies must rely on inexact tools to measure dietary factors, and errors in measuring individual levels of consumption can be large.7 Broadly, imprecise measurement of the exposure of interest (in this case rice consumption) will usually produce an underestimate of the true strength of the association with the outcome (diabetes). The same cannot be said for imprecise or incomplete measurement of potential confounding factors. Inadequate adjustment of the model for potential confounding factors may deliver results that are wrong in terms of the strength of the association and also in the direction of the effect identified. The confounders adjusted for varied between studies,5 and the impact of each on the final estimate of effect cannot be determined from the data presented. A meta-analysis based on the individual participant records from the included studies, rather than the published summary estimates, would enable a much deeper exploration of the associations and greater insight into the adequacy of the adjustments made.

Hu and colleagues calculated risk ratios for each included study by comparing the rates of diabetes in the group with the highest consumption of white rice with those in the group with the lowest consumption. However, the highest and lowest levels of rice consumption varied greatly between studies. For example, in the primary analysis, a difference in consumption of 33 g/day (56 g/day v 23 g/day) is plotted on the same scale as a difference in consumption of 250 g/day (750 g/day v 500 g/day). Such massive differences in consumption are unlikely to produce the same effects on the risk of diabetes. Although these discrepancies may explain some of the heterogeneity in the size of the effect estimates obtained, what is really needed is a more sophisticated analysis based on primary rather than summary data.

Although the findings of the current study are interesting they have few immediate implications for doctors, patients, or public health services and cannot support large scale action. Further research is needed to develop and substantiate the research hypothesis. Ideally, incident diabetes should be investigated in participants of an adequately powered randomised controlled trial in which white rice consumption is substantially modified in an intervention group.8 Such a study would need to be large and would be a challenge to complete, but it would not be impossible to conduct in the right setting with moderate resources.

The real problem for the field of nutritional research is one of attracting the kind of resources that are available for the development of a promising drug treatment. Diet related ill health is now widely believed to be the leading cause of chronic disease around the world,9 but definitive research that precisely and reliably defines the effects of plausible, scalable, and affordable interventions is almost completely absent. Public health nutrition awaits the discovery of the model that will secure the investment needed to answer questions about the role of nutrition in health using large randomised studies. Until then, the effect of the consumption of white rice on the development of type 2 diabetes will remain unclear.


Cite this as: BMJ 2012;344:e2021


  • Research, doi:10.1136/bmj.e1454
  • Competing interests: The author has completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declares: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work.

  • Provenance and peer review: Commissioned; not externally peer reviewed.