“Eating a diet high in processed foods increases the risk of depression,” BBC News reported.
This report is based on data from a long-running study of middle-aged civil servants. An analysis found that eating processed foods was associated with depression five years later, even after other social and health factors were taken into account.
However, the study design has some limitations, and although this type of study (called a cohort study) can make a strong case for causation, it cannot prove that one thing causes another. In addition, it is possible that depression affects a person’s diet rather than the other way around.
A link between diet and depression seems plausible, but further research that gives more conclusive evidence is needed.
Where did the story come from?
The study was carried out by Dr Tasnime Akbaraly and colleagues from University College London. The study was based on data from the Whitehall II study, which was funded by grants from the Medical Research Council, British Heart Foundation, UK Health and Safety Executive, Department of Health and several national funding organisations in the US. The study was published in the peer-reviewed British Journal of Psychiatry .
BBC News gives a balanced report of the study and points out that this sort of study cannot prove cause and effect, but can only show associations.
What kind of research was this?
This was a cohort study, which used data from a larger, long-running cohort study called the Whitehall II study. Whitehall II is a well-established and well-regarded study that was set up to investigate how social class, lifestyle and psychosocial factors contribute to the risk of disease. Many subsequent studies have used its data to produce or dismiss several theories regarding risk factors for disease.
This particular research investigated whether there is an association between diet and depression.
As a cohort study, it can make a strong case for causation, but it cannot prove cause and effect, in this case that poor diet causes depression. In addition, it cannot rule out reverse causation, in other words that depression may have affected the participants’ diets.
Other factors, measured or unmeasured, may also confound the association between an exposure and outcome. The researchers attempted to account for some of these factors by collecting certain sociodemographic factors and health behaviours and adjusting for them in their analysis. This was a strength of the study.
What did the research involve?
Between 1985 and 1988, the Whitehall II study enrolled 10,308 London-based civil servants aged between 35 and 55. When they signed up, the participants were given a physical examination and a broad questionnaire about their diet and lifestyle. At five-year intervals after this, they were invited for clinical examinations and between these visits were sent postal questionnaires.
This particular study involved 3,486 white European participants who had data collected on dietary patterns and related factors from 1997 to 1999, and on depression from 2002 to 2004.
Food intake was measured using a food frequency questionnaire adapted from another study that asked how much of 127 items the participants ate during the past year. It is not clear whether this food frequency questionnaire had been validated in the UK population, although the researchers report that the questionnaire was ‘anglicised’ (presumably meaning it was made relevant to UK foods). Each participant was given a score according to their responses. This score was used to measure how well they fit two dietary patterns: ‘whole foods’ (a high intake of vegetables, fruits and fish) or ‘processed foods’ (including fried food, chocolate, pies, processed meat and refined grains). Within each group, scores for each pattern were divided into thirds to indicate how well the person fit the pattern.
A statistical method called logistic regression was used to examine the association between dietary pattern and depression. This is an appropriate analytical method for these types of data. Factors that could have affected this link, including sociodemographic factors (such as age, gender and education) and health behaviours (such as smoking and exercise) were taken into account in the analyses. The researchers also carried out analyses that excluded people who had depression at the time of the dietary assessment (defined as having a score above a cut-off point on a depression scale, or receiving antidepressants).
What were the basic results?
People with the highest intake of whole foods were less likely to have depression. This was the case even after all the factors that may have influenced this link were taken into account (odds ratio [OR] 0.74, 95% confidence interval [CI] 0.56 to 0.99). People who ate the most processed foods were more likely to have depression (OR 1.58, 95% CI 1.11 to 2.23).
This link between processed foods and depression remained statistically significant after those who already had depression when they completed the dietary questionnaire were excluded from the analysis. This was not the case for the whole foods group, where the association with less depression was no longer statistically significant.
How did the researchers interpret the results?
The researchers conclude that, in middle-aged people, processed foods are a risk factor for depression five years later, while whole foods can protect against it.
This study suggests that a healthier diet protects against depression, but it cannot prove this due to several limitations:
- It is plausible that depression affected the participants’ diet rather than the other way around. The researchers argue that this is unlikely to have happened as no association was found between the participants' early reports of depression (between 1991 and 1993) and their diet six years later. The results were also not affected by excluding people who already had signs of depression when their diets were measured between 1997 and 1999. Although this may be the case, different methods were used to assess depression at these times and this reduces the reliability of these results.
- Depression was assessed with a short questionnaire, and participants who scored above a certain cut-off were classified as having depression. Although the researchers used a commonly used questionnaire to measure depressive symptoms, the best way to diagnose depression would be a full clinical interview with a doctor.
- By excluding black and Asian participants and people with missing data, biases may have been introduced into the study. The researchers note that people in the included group (white Europeans) were less likely to have depression or to be in a low social class, and were more likely to be men than all study participants who were alive in 2002-04.
- Food intake was measured using a food frequency questionnaire, which asked how much of 127 food items the participants ate during the previous year. This method of assessing diet has limitations as not everyone will remember exactly what and how much they ate in the past 12 months. There may also be a systematic difference in the way that people with depression and those without it remember their food intake.
- It may not be possible to apply these findings to populations other than white European civil servants in the UK.
- The researchers did take some factors other than the consumption of whole and processed foods into account. However, it is possible these adjustments may not have completely removed their effect or the effects of other unmeasured or unknown factors.
The researchers conclude that processed foods are a ‘risk factor’ for depression rather than specifically labelling them a ‘cause’. This is a balanced conclusion, considering that unmeasured factors may contribute to this association. A healthy diet has a range of proven benefits and the suggestion from this study that there is a link with improved mental health seems plausible. Randomised controlled trials would provide more conclusive evidence for this.
Analysis by Bazian
Edited by NHS Website
Links to the headlines
BBC News, 2 November 2009
Links to the science
The British Journal of Psychiatry 2009, 195: 408-413