Thursday, October 7, 2010

Diet and Mood Disorders

Found two more studies measuring diet pattern and mood disorders!

The first one is Dietary pattern and depressive symptoms in middle age from the British Journal of Psychiatry in 2009 (free full text, go take a look!). This paper is part of the large Whitehall II epidemiological study, where some 10,000 people were screened at baseline (phase 1) for all sorts of demographic characteristics, stress levels, health, lifestyle factors, blood pressure, and some labwork. Every 2&1/2 years or so, the study subjects received a postal questionnaire to fill out (phases 2,4,6,8), and every 5 years a questionnaire and a clinical examination were done (phases 3,5,7). In this paper, the data was taken from 3486 White European participants with data on dietary patterns and all covariates at phase 5 and depression at phase 7. The 175 Black and 331 Asian participants were excluded due to "differences in eating patterns." (!!!)

Dietary "pattern" was determined by a Food Frequency Questionnaire based on the one used in the Nurses Health Study (but changed up to include British sorts of foods - like bangers and mash, or crisps, or yorkshire pudding). Here's a good write-up addressing the (actually rather rigorous, but still hopeless) validity of the FFQ used in the Nurses Health Study. 127 food items were rated according to how often they were consumed - "never or less than once per month" up to "six or more times per day" - the 127 items were divided into 37 groups, and dietary patterns were identified using "principal component analysis" of the 37 groups, and statisticians gleefully addressed the data with graphs and scree plots and tweaks and nudges until a score pops out about each person's diet, falling into "whole foods" or the "processed foods" patterns.

"Whole foods" were diets consisting of vegetables, fruits and fish.

"Processed foods" were diets consisting of "high consumption of sweetened desserts, chocolates, fried food, processed meat, pies, refined grains, high fat dairy products, and condiments."

Depression measures were determined by a 20 question "Center for Epidemiologic Studies Depression Scale." People with a score higher than 15 were considered depressed.

Then the statisticians got a whack at the data again, adjusting for covariates such as age, gender, marital status, employment grade, education, smoking, physical activity, health status (based on clinical findings such as a high hemoglobin A1C (a measure of average blood glucose), blood pressure, being on antidepressants, etc.), a second "GHQ" depression subscale, and a cognitive score based on 65 question test.

Ready for the results yet? What do you think will turn out?

Patients with the highest intake of whole foods were less likely to be depressed (and this association wasn't affected much, actually, by adjusting for all the covariates, which is interesting - suggesting that the association with diet could be as important or more important than all those other covariates.). Patients with a high intake of processed food were much more likely to be depressed.

There is another interesting thing about this study. See, from just the above data, you don't know whether depressed people are too blue to do anything but sit around gorging on meat pies, sugar, and refined grains, or if the diet actually leads to the depression. The depression measures were repeated at phase 5 and phase 7, and when the 427 participants who were depressed at stage 5 were excluded, the data showed that the same people who had crappy diets but were not depressed (yet) at phase 5 were more likely to be depressed at phase 7. In addition, the researchers went back to the phase 3 data, and found no evidence that the dietary patterns in phase 5 were worse for the participants who were depressed at phase 3. These are clues that crappy diet precedes depressive symptoms, not the other way around.

All in all, a rather cool study, considering the limitations of epidemiology.

The second study is hot off the presses - Diet quality in bipolar disorder in a population-based sample of women from the Journal of Affective Disorders (I think this is the November issue - these are the corrected proofs available online before print). The subjects used were women from the Australian Geelong Osteoporosis Study (I blogged about them back in June). 1046 women randomly recruited from compulsory voting roles volunteered for this particular piece of the study. Each of the women were given the gold standard interview for diagnosis of psychiatric disorders - the SCID (non-patient edition). Patients with depression or anxiety were excluded, allowing for patients with bipolar disorder to be compared to those with no current (or lifetime) psychopathology.

Once again, a food frequency questionnaire was used, and the diets were broken up into "traditional" (vegetables, fruit, beef, lamb, fish, whole grains), "western" (meat pies, processed meats, pizza, chips, hamburgers, white bread, sugar, flavored milk drinks, and beer), and "modern" (fruits, salads, fish, yogurt, nuts, beans, tofu, and red wine). Blah blah blah logistic regression analyses and exposure variables were studiously applied and accounted for (can you tell I'm not an epidemiologist?).

Here's an interesting tidbit that I don't remember from the previous paper I blogged about in June - the researchers note that it was the absolute amount of western style foods that seemed to be related to the risk of depression rather than the amount as a proportion of overall energy consumption. Therefore, the researchers examined the data in this grouping as both absolute amounts, percentage of energy, and covariate confounding. Whew.

Bipolar disorder is less common than depressive or anxiety disorders, so only 23 women in the whole study qualified for the diagnosis (there were 332 with depression or anxiety and therefore excluded from this analysis, and 691 without depressive or anxiety disorders). Participants with bipolar disorder were younger and had higher average daily energy intake, though there was no difference in education, BMI, alcohol intake, education, socioeconomic status, smoking, and physical activity compared to those with no psychopathology. (Interesting - bipolar disorder and substance abuse are highly correlated so it is something to keep in mind that these 23 bipolar ladies didn't seem to smoke or drink more than the 691 "normals.")

Now the diet results! Individuals with a "modern" pattern were more likely to have a diagnosis of bipolar disorder, as were individuals with a "western" pattern, before and after adjustments for energy intake. A "traditional" dietary pattern was protective. These data held even when accounting for age.

As in the previous study (which I addressed in the comments in my previous blog post), the diets were also given a total "DQS" or dietary quality score, associated with how close the diets came to the national recommendations for good diets (similar to the USDA recommendations). DQS score of diets had no correlation with bipolar disorder, protective or not protective.

Overall, the authors noted (as I do now!) that the association with depressive, anxiety, and bipolar disorders are similar - a "traditional" diet being protective, and modern and western diets being similarly dismal. My theory is that an individual's response to diet would be based on genetic risk, so the inflammation level related to similar dietary choices would come out as different psychopathology depending on your genes.

The authors of the second study make note of the results of the first study I mentioned above - they think diet is causative, not a choice as a result of specific psychiatric illness. They note systemic inflammation, oxidative stress, and neurotrophin levels as potential explanations of the demonstrated associations between dietary quality and bipolar disorder. I agree. Wholeheartedly.

Eat real food, folks. It won't hurt. And maybe it will help.


  1. So much for the much vaunted Mediterranean diet then!

  2. Well - beans, tofu and red wine are certainly a bust when it comes to epidemiological protection from major mood disorders.

  3. I wonder about some of these food classifications. In the first study, "high-fat dairy" e.g. butter is a processed food? In the second, "fruits" are modern but "whole grains" are traditional?

    A problem with these kinds of studies is that they can decide on their classification scheme after seeing the data. With 127 food items and 37 food groups, there are many possible principal component breakdowns. So it's easy to work back from a result you want to find, or to make the results appear much stronger than they really are.

  4. Yes, indeed. Another word for what you are suggesting is data mining. Agree absolutely.

  5. (I personally object to the inclusion of chocolate in the processed foods:)

  6. Wow, quite an unusual classification the latter one used for traditional -"vegetables, fruit, beef, lamb, fish, whole grains." Beef and lamb usually get lumped into the stereotypically "bad" diet as OMGREDMEAT. Awesome study.

  7. Melissa - while the groupings in these dietary studies always irk me a bit, I do find the Australian study grouping to be interesting. As a Texan, the "traditional" grouping is right in line with what I would consider traditional ranching/farming food - "meat and potatoes" sort of food it is not necessarily what people here where I live now in the northeast US would consider traditional, but beef has special status in Texas and the west, maybe....despite what CAFOs do to the beef. Here in the northeast US "traditional" tends to mean more ethnic varieties of food - Italian or Irish or that sort of thing. The "modern" group is less "Mediterranean" than what you might think is healthy from reading a ton of women's magazines and watching the news. Overall, I think the authors may have been influenced by Michael Pollan and his "eat what grandma eats" philosophy.

  8. "My theory is that an individual's response to diet would be based on genetic risk, so the inflammation level related to similar dietary choices would come out as different psychopathology depending on your genes."

    THIS rings so true with me and my family. my dad has big stress/worry/anger/depression/anxiety problems. has his whole life, on the typical western diet and he is addicted to carbs- always thought if he dropped wheat he would improve 10 fold. well he went on the atkins diet(a good version of it) with my mom one time and within 3 days i noticed an IMMEDIATE mood change in him. he isnt nor has ever been overweight but the chane in his food choices...oh my gosh i cannot even begi to explain. then, he decided 'i cant give up wheat, ice cream, chips' etc...and the problems came right back.

  9. Emily
    I also object to the inclusion of full fat dairy as "processed" foods.
    Lets see...milk the cow...drink the milk= full fat dairy. No processing involved.
    Milk the cow...ultra pasteurize...homogenize...add vitamin D...take out the fat...add in powdered dry milk for well as fillers and extenders,= low/no fat dairy. Now which one is processed?
    Sorry , just had to rant!

  10. Hi Emily,

    I've only done a quick survey but I really like the blog! Especially your "voice"-- it is accessible and professional. Get cracking on the book!

  11. Hmmm. This is very interesting. Obviously there are a lot of individual exceptions. My mother (lifelong manic depression, rapid cycling, treatment resistant, dozens of hospitalizations despite every drug/treatment known to medicine) ate a traditional diet (which revolted us children: rare meat, no fruit, eggs,a few vegetables, zero carbs, lots of cheese, Hovis brown bread, tons of butter, red wine etc. She was massively overweight and would become manic for months at a time. Probably just quantities excessive...In her case, carbs were a good thing as they helped calm her down, but she didn't like them as much as she liked greasy high protein foods. She HATED fish, especially things like salmon or kipper, which might have done her some good.

    It's also interesting that however dreadful a high carb, heavily processed diet may be for anyone over time (and especially for depressed people), depressed people tend to crave sweets, carbs, and especially chocolate in the short term. Because these foods affect serotonin levels, relatively quickly, for a short time. I am told. Perhaps habitually eating this way messes with these processes?

    I also wonder about the related issues of class and culture and the degree to which these affect diet. In the US it is primarily the poor who are extremely overweight, and certainly being poor sucks, and exacerbates any preexisting tendency towards mood problems.

    As far as the alcohol and substance use in bipolar patients, most people lie about it. A psychiatrist friend's general rule is whatever his patients tell him they drink, he multiplies by 3. A good test is probably the patient's weight. If they are thin and bipolar, they likely abuse alcohol and/or drugs. If they are plump to heavy, they may or may not. But there are a lot of sugars in alcohol, and to some extent people with mood problems are manipulating their mood with these as much as with the direct intoxicating effects of the alcohol. I think. If not, why would it be that people in recovery gain so much weight and immediately crave sugary, carb laden food? Just my unscientific observations...

    Now, back to making TOTALLY UNHEALTHY and delicious pecan pie...My family all crave carbs.