‘Tis a bittersweet post that marks the end of using this blog for my class.   While I will miss the opportunity to blog about epidemiology I yearn for sun, fun, and a break from homework.  I have every intention of a self imposed week long hibernation following the semester, to catch up on my sleep and days that don’t involve me staring at my computer screen till my eyes begin to water and I drop my head on my laptop.

(can everyone take a minute look at that cell phone in this photo!)

But perhaps it is fate (or purely planned by my professor) that the last blog may be one of the most importance, at least to me.  I know there are lots of people who turn the other direction when the words “politics” or “policy” comes out of anyone’s mouth.  I mean there is a reason they say it’s never good class or etiquette to talk politics. But while some want to pull their hair out as people talk scream or yell politics, I am the person who frequents ALL political websites and has the cable news from all sources on in the background at all times.

This is what I think I am sometimes watching on the daily news cycle

So it would seem this would be the easiest blog to write, but for some reason this has been taking me a long time to write (and I don’t think it’s just because at the end of the semester my brain feels like goo).  I think this has been hard to write because when you peel away all the other stuff in policy and try to look at what role epidemiology plays, you realize that it can be the most important role and sometimes gets overlooked or ignored because of all the other hoopla.

So when I googled (why “googled” is not yet allowed in my spell check I don’t know, but it is a fatal flaw to Word and my BlackBerry) “epidemiology and policy” this very easy to read yet very cool article came up from Merrill & White titled “Why Health Educators Need Epidemiology” (citation below—the full article is on google). I read  one section of the article that talks about how epidemiology helps health educators entitled: “Assessing Individual and Community Needs for Health Education” (Merrill & White, pg 218) and I instantly had my topic.

A topic close to my heart is sexual education.  This may be because I am on the board of ChoiceUSA, which among other things is a strong proponent of young people getting involved in trying to advocate comprehensive sex education (I know shameless plug, but they do some great things, and if you are a college student wanting to get involved they are the place to start).

But how does epidemiology fit into sex ed?  Epidemiology is everywhere in this topic.

Throughout this blog we have talked about the different things epidemiology is used for and the wide range that the study of epidemiology encompasses, and in the push for policy and federal funds for sex education in high schools epidemiology has been cited by many.  Politicians and advocates for sex education point to epidemiological studies that show correlations between lower teen pregnancy rates or lower teen STI rates and sex education.  Or conversely show no statistical change or decrease in rate of teen pregnancy or STI rates in areas that only teach abstinence only education.  Much of this was used to encourage more federal funding for sex education programs in the Patient Protection and Affordable Care Act (Kaiser 2010—full citation below). The proponents of sex education argue that the evidence based results pointing towards sex education in schools is the most important reason to make policy changes nationwide and increase the funding.  Though it’s important to note sex education isn’t perfect (because the programs are not universal) and there is data that suggest there is important differences on the effect on young girls and boys (here is some audio about some new research released from the government about sex education).

But that doesn’t mean other stakeholders, or those for abstinence only education, don’t try to point out more than just the moral issues surround teaching sex ed to those in schools.  In social health behavior we talk about harm reduction programs and arguably (if you paint harm reduction with a broad stroke) sexual education could fit in this category (but very much to a lesser degree than other programs), which is why I understand the hesitation in teaching sex education, believing it could be encouraging teens to engage in sex (click here for a good site that looks at harm reduction theory and sexual health).

What is clear is the use of epidemiology by both sides.  Each side uses data about incidence rates (number of new cases) of teen pregnancy and STIs mainly from the CDC’s Youth Risk Behavioral Surveillance System, to advocate each side, to try measure effectiveness of programs.  And while there has been a decline in teen pregnancies over  time (though you would not know that if you watch shows like Teen Mom), there is still debate about what has caused these decline. There are some who use ecological studies to point to introduction to sexual education, some do not believe this to be true and there are studies that have tried to show there is no correlation.

While I think there is a lot of epidemiological data out there that would lead to policy that promotes sex education, I think we can’t ignore the lessons from abstinence education either.  Perhaps that’s why I am more inclined to be impressed and interested in the new studies coming out that look at the relationship between abstinence education and comprehensive sexual education (Jefferies, Dodge, Bandiera, & Reece—full citation below).   I think this could a call for more epidemiological research that better defines comprehensive sex education (exposure) to make sure it includes abstinence and the impact on sexual health outcomes (disease).

I definitely think the evidence points to comprehensive sex education, but I also think we live in a time where the increased polarity of ideas causes more strife than anything.  If deep down the two methods share a common goal, even if they don’t want to agree on that (to keep teens healthy), maybe this is a time to take lessons from both.  At first I was very frustrated that the new health care bill still had funding for abstinence only education, but I really think if we teach them both together at the same time there is room for us all to agree (if only it wasn’t “abstinence only” and instead “abstinence &” sex ed).  For some reason I think we have forgotten what compromise really means in politics, and maybe we can use epidemiology to show that we can provide comprehensive sex education to teens that also teaches them that abstinence is the only way we know to avoid teen pregnancy and STIs 100%.   I think what needs to be clear is that real comprehensive sex education is that, comprehensive, and should include abstinence as well. No matter what I am still going to go out there and get involved to allow comprehensive sex education in schools, but even with the data and the funding there are still those who don’t want to do it in exchange for abstinence only. So maybe this  blog is a plea for us all to remember how to compromise.

Hand shakes just in case we aren't all huggers

Here are some great articles about the complexities of sex education and abstinence only education, so you can make your own decisions:






Kaiser Family Foundation. (2010). Impact of health reform on women’s access to coverage and care. Focus on Health Reform, Retrieved from http://www.kff.org/womenshealth/upload/7987.pdf

Merrill, R.A., & White, G.L. (2002). Why health educators need epidemiology. Education for Health, 15(2), Retrieved from http://www.educationforhealth.net/EfHArticleArchive/1357-6283_v15n2s14_713665105.pdf

Jeffries, W. L., Dodge, B., Bandiera, F. C. & Reece, M. (2010). Beyond abstinence-only: relationships between abstinence education and comprehensive topic instruction. Sex Education: Sexuality, Society and Learning, 10(2), 171-185. doi:10.1080/1468181100366631


One thing about epidemiology I have noticed is sometimes its more about questions then it is about answers.  That is the reason we have studies; and in some studies, like an ecological study, we just have more questions or more ideas for ways to answer new questions. So lets play a question game with a topic that I think is important, marginalized groups, in particularly the LGBTQ community.

A significant amount of data on LGBTQ (lesbian, gay, bisexual, transgendered, and queer) population health statistics suggests an increase risk of risky sexual health behaviors and poor health outcomes.  But despite the increased risk LGBTQ individuals are less likely to receive consistent and/or routine care.  Some data haves shown that the lack of care is because LGBTQ individuals fear discrimination, retaliation, and views that physicians are unaware of the specific needs to LGBTQ patients.  So in focusing on LGBTQ health must we go beyond the expensive and wide stretching public health campaigns?  For regardless of awareness of the need to seek care, many don’t feel comfortable actually receiving care or do not disclose their identity as a member of the LGBTQ community, which has been associated with withholding useful health information.  The problem stems from the gap of knowledge of LGBTQ health; there is far too little epidemiological data on many health indicators besides sexual health and mental health, more importantly this data most often focuses on gay men alone.  But if we are aware of the fear of health visits and some of the causes, wouldn’t it be important to look at this through the ecological model (the social ecological model is a model that looks at the different influences on health behavior and suggest interventions at several levels can impact behavior change–click on the link the CDC has a great explanation) and not put the responsibility solely on the individual to make behavior change, which is an especially important idea for marginalized communities?

To better understand the impact of interpersonal relationships between LGBTQ patients and their primary care physicians, why not create a study  that will measure the impact of making health providers offices more LGBTQ friendly? Would one also want address the gap in knowledge, and look at whether making sure physicians are aware of the distinct needs to LGBTQ populations and the importance of a patients disclosure of LGBTQ identity to their physicians impacts retention of LGBTQ patients?  Medical history reports still use heterosexist language that can potentially impact an individual’s view of the health visit. There has been data that suggest physicians are not aware of unique health issues of LGBTQ individuals.   What impact does this gap in knowledge have overall on health outcomes? And how does a patient’s perception that their health provider knows little about LGBTQ health impact the relationship between the two (patient and physician)?  I know as an African American woman I tend to want to seek care from a provider who is African American or has expressed knowledge of the different health needs for my population.  This impacts how much we trust our physicians with health information and health history.   This is important because there are arguably more than 4 million of the United States population identifying as LGBTQ, and many more either not “out” or identifying within these labels, health care for sexual minorities, transgendered, and non cis-gendered individuals.  And with knowledge of the increased risk of poor health indicators as described in Healthy People 2010 we have responsibility to help those who are affected by health inequities.

Sometimes the best way to learn is to create something, and as we well know from mytreemakingskillsand my ability to manipulateAndersonCoopersface (not that he needs any touch ups on that pretty face) I can definitely create things.  Creating things allows you to not only figure out how to build whatever it is you are creating, but you really have to understand the concepts to get whatever you create to make sense.  This is also true for epidemiology, and since I have blogged about what epidemiology is, epidemiological study designs, and discussed media interpretation of epidemiological studies, I will now take the time to learn and perhaps teach through creating something.  So I have been told asked to make my own study design.

Naturally in making a study design there is a process of thought.  First you have to find something to study; and public health’s fight against obesity is not lost on me, particularly childhood obesity. I think at this point to say there is an obesity problem is an understatement (that’s why they call it an epidemic).  In the US as a whole more than one third of children ages 10-17 are overweight or obeseAndasCNNnoteswecantkeepcallingbabyfatjustthat, babyfat, becauseresearchershavefoundthatactuallyonefifthofAmerican 4-year oldsareobese.(I definitely believe this because I have seen some BIG babies eating cheeseburgers; thank you Maury Show for the baby eats a cheeseburger episode!)  But when you are looking at children, obesity becomes about more than just about their lifestyle choices. There is definitely a need for research on the relationship between childhood obesity and the adults around them.

So, I wanted to know more about the role mother’s play when dealing with childhood obesity in their children at a young age–clearly I just like blaming parents [hence the last few studies putting the blame on the parents and not the children (It’s all your fault mom!)].  Anyway, my thinking about a mother’s role in childhood obesity led me to the question: what is the association between mothers’ physical activity, eating habits, and stress level on the prevalence of obesity in their children?

In designing a study, you must go back to the exposure and disease, as with anything else in epidemiology.

Exposure Mother’s physical lifestyle (physical activity, stress level, eating habits)
Disease Overweight and obese child (Obesity in a child is characterized when a body fat in a child is more than 20% its body weight; BMI≥ 30 constitutes obese and BMI=25-29.9 constitutes overweight)

Once you know your exposure and disease you have to decide what study design to use. Rememberanearlierblogcoveredallthedifferenttypesofepidemiologicalstudydesigns. Given the topic, the best study is a prospective cohort study. Remember cohort studies follow individuals who do not have the disease, separates them into categories based on those exposed and those not exposed, and studies the incidence of disease over time.  Gordis notes on page 172 of Epidemiology (the 4th edition) that most studies involving children use cohort studies, probably because you can follow the incidence of disease (assuming a child isn’t born with the disease) and you can study multiple variables.  Also, when dealing with pregnant women and with children there is always a question of ethics, and certainly an experimental design could be unethical especially if you are trying to advocate poor health habits for a particular group. A prospective cohort study, also allows the research to monitor people in a natural lifestyle (I should say natural habitat in my best Australian accent like I am on Animal Planet).

Study Design Type Prospective Cohort Study
Exposure Mother’s physical lifestyle (physical activity, stress level, eating habits)
disease Overweight and obese child (Obesity in a child is characterized when a body fat in a child is more than 20% its body weight; BMI≥ 30 constitutes obese and BMI=25-29.9 constitutes overweight)

So now we have the research question, exposure, disease, and the study design type.

Next we need to explain the method in this study.

We have to know how we are going to measure our exposure and disease. The disease will be measured with BMIorbodymassindex, whichiscommonlyusedtomeasureobesity.  We will also measure percent body fat, because many have noted that BMI can be misleading depending on an individual’s muscle mass. For the exposure we have to measure the three components with three different measures.  Since it’s best to use measures that are proven validandreliable, these measures come from others who have proven those two things.  For diet, mothers will take a food-frequency questionnaire (FFQ) that has questions that ask about the frequency of consumption of several foods and drinks. This food frequency questionnaire was used by Northestone et. al in their study on diet and children’s IQ.  We will standardize the data for each variable because some foods and drinks are measured differently.  Since we want to measure categories for the food consumption patterns we will use principal-component analysis (PCA), as the Northestone et. al study does, which is often used to find dietary patterns (If PCA doesn’t make sense try looking at StatSoftsonlinetextbook, it’s very helpful).  The PCA will help us categorize food diets as either processed (high fat and high sugar processed food), traditional (meat, potatoes, vegetables), or health conscious (fish, vegetables, fruit, etc).  Physical activity will be measured with a 7 day physical recall sheet from the SanDiegoPreventionResearchCenter.  Participants will be asked to fill out a 7 day physical activity sheet, asking them to mark average activity a week.  To measure stress a 10 item self report questionnaire that was created by Cohen and Williamson will be used.  This questionnaire measures a persons’ evaluation of the stressfulness of their past months.

To start the study, you have to have a defined sample. For this study, I want to recruit a sample from University of Missouri Hospital and University of Mississippi Medical Center, because both of these states have had issues with obesity. Missouri is currently ranked as the ninth worst state in the United States for childhood obesity (Famuliner, 2010). The cost of health care in Missouri for obese children is $1.6 billion per year. Mississippi was named the MOST obese state in the USA in 2010 with the number one adult obesity rate of 33.8% and number one child obesity rate of 21.9% (Trust for America’s Health).  For the study to be relevant there has to be a significant sample size, so there should be 1500 mothers from the Missouri hospital and 1500 from the Mississippi Medical Center. That would be a total sample size of 3000 (n=3000).  I want to begin measuring the mothers’ lifestyle while she is pregnant and in her second trimester (between 4 to 6 months); the second trimester is used to avoid a number of miscarriages.  The exclusion criteria for the sample is that the mothers have to be between the ages 18 to 35 (35 and above is when mothers have special precautions for pregnancies), this must be their first child, they should not be on any required special diets, cannot have gestational diabetes, must not be having multiples, cannot be put on bed rest before the 8th month, and cannot engage in substance use, alcohol consumptions, or smoking.

Since we have the sample, exposure, disease, study design, we have to next have to explain how we will carry out the study. As previously stated data on eating habits, physical activity, and stress will start to be collected during the second trimester.  Data will also be collected every six months thereafter until the child reaches the age of 4 (we are looking till age 4 because age, which was decided by looking at the growth chart), and will have one follow up every year until they are the age of 8.  Six months may seem like a lot, but six months is more reasonable to measure how things change in people’s lives. The women will answer surveys for their food habits, physical activity, and stress; and children will be brought to the hospital every 6 months to be weighed and BMI and percent body fat measured.

We also need to look at potential confoundingfactors when collecting data.  We need to collect and account for age, marital status (married, cohabitating, single, same-sex), income level, insurance, education level, geographic (urban or rural or suburban), month of delivery, sex of the baby, work or stay at home, and mothers BMI.

There will be challenges with this study. There could be potential for recall bias, as we are asking the mothers to recall information from the past six months.  Because this is cohort study, we can also have problems with attrition or the ability to keep people participating in the program. Hopefully the incentive of receiving a free check up for their child every 6 months and free transportation will help to curb this.  There is also the potential that the sample won’t be as diverse as desired.   We also are picking a very distinct sample from two states that could impact whether the findings can be considered generalizable.

We hypothesize that high stress, poor eating (yeah, my mom told me about those 3am trips to White Castle when she was preggers…), and little physical activity (self-prescribed bed-rest anyone?) by the mother from when she is pregnant and as the child grows will be associated with childhood obesity.  And let’s say that our hypothesis is correct, so what? Well there can be lots of things impacted by a study that suggests this.  If the study is big enough then the media could pick it up and print some of those grandiose headlines that make everyone totally believe that health food, low stress, and physical activity will help your baby/child not be overweight or obese.  If there truly is a strong association, then more physicians could begin telling pregnant women and mothers of young children they need to change their eating habits, stress levels, and physical activity because it will impact their child’s obesity rates (though I imagine telling a pregnant woman she can’t eat those pickles with ice cream three times a day isn’t going to necessarily stop her…plus I would be totally afraid, you know, hormones).   Public health services may work harder to get promotional health information about the need for a change in the mother’s diet, stress level, or physical activity to curb their child’s chance of obesity.

But overall what are the strengths and weaknesses?

Weaknesses Strengths
Potential for high costs Ability to have multiple variables at once for disease
Complicated logistics and complicated to organized Large amount of data
Attrition levels Able to start with a population who doesn’t have disease and track them overtime
Confounding factors Going from exposure to disease to find association Exposure   —————————->  disease
Potential for recall bias

The potential for confounding factors to impact our data can affect a study’s credibility.  Because we don’t live in a vacuum there is always this risk, and if you don’t necessarily measure all of them then when people critique your study they will note that when measuring the studies credibility.  Also with a cohort study you have to worry about attrition if you have a low attrition level then people will question the credibility of the study. Certainly its an issue when you are asking people to remember things and recall bias is going to be an issue , but you hope you develop measures and ask not to fair in the future about past behaviors.

So now let me take some time to reflect on this.

It is difficult to create your own study because, if you are like me, you tend to be more critical of your own thoughts and recognize more of the possible limitations.  One hard part about deciding upon a sample is the difficulty in picking exclusion criteria.  Certainly if you had all the money in the world (I wish I did!) and all the time in the world (Man I really wish I did…) then you could try to include everyone and measure for every confounding variable and develop the most comprehensive and amazing study ever, but logistically this just is not possible. The problem is as soon as you begin to exclude certain people from your study you wonder what you are missing.  For example, while women 35 and older may have a greater chance for difficulties in pregnancies women are having children later and later, and we don’t know if those over the age of 35 have different habits or what will have a different effect on their children.  We could be looking at the completely wrong age group. The first idea was to use diabetic pregnant women, to see the effect diabetes had on obesity rates of their children.  It seemed like a great idea, since we know there is a relationship between diabetes and weight, but once you try to figure out the logistics of such a study it just seems too difficult.  How would you get a large enough sample size?  How would you measure for different types of diabetes (type 1; type 2; gestational; gestational that turns into type 2…it’s confusing!)?  What about the fact that diabetic women who are pregnant usually have to have special diets?  Sometimes the plausibility of carrying out a study impacts the choices you make. It would have been interesting to see the effect culture and country of origin have on pregnant women’s lifestyles, and that impact on childhood obesity.  Different cultures and different countries have different traditions.  Perhaps a woman who is pregnant eats a completely different diet?  Or it is customary for an extended family to live in the same house?  What would be the impact on stress of the mom and on childhood obesity? (I don’t have in-laws, but from what I hear that would just make you more stressed out!) What if once a woman has a child she always stays home with the child, and what is that impact on childhood obesity?  While physical activity and eating were our exposures in our study, and we still recorded whether or not a mother works or stays at home as a confounder, culture is a different beast.  But culture is hard to measure.  How do you decide what is a different culture? (Try looking for a measure for that on google scholar!)

Just a reminder: “This blog was created for a graduate-level epidemiology project and does not officially represent the views of the University, professor, or other related entity.”


Cohen, Sheldon;Kamarck, Tom;Mermelstein, Robin.  A global measure of perceived stress.Journal of Health and Social Behavior. Vol 24(4) Dec 1983, 385-396.

Famuliner, R (2010). Missouri 9th worst state for child obesity:Education,Health & Medicine. MissouriNet. Retrieved March 2011, from http://www.missourinet.com/2010/09/13/42086/

Gordis L. Epidemiology. 4th ed. Elsevier. 2008.

Northstone, K., Joinson, C., Emmett, P., Ness, A., & Paus, T. (2011). Are dietary patterns in childhood associated with iq at 8 years of age? a population-based cohort study. J Epidemiology Community Health, Retrieved from http://jech.bmj.com/content/early/2011/01/21/jech.2010.111955.abstract doi: doi:10.1136/jech.2010.111955

Seven day physical activity recall (par) . (1997). San Diego Prevention Research Center, Retrieved from http://www.sdprc.net/lhn-tools/PAR-interview-instructions-English.doc

Trust for America’s Health (2010).  F as in Fat: How Obesity Threatens America’s Future 2010. Retrieved April 2011, from http://healthyamericans.org/reports/obesity2010/

I, like most people, have been glued to the TV watching all that’s developing in Japan.

In my failed attempts to pry my eyes away from TV and the internet news, I have found some interesting health related articles that I thought I would post, so I can keep procrastinating on doing the work I need to do for tomorrow and you can click some links.  I don’t know who is looking at this blog 543 times probably my mom 543 times, but even though I have nothing due for class I’m going drop some public health related news stories your way.

I was definitely unaware of the fact that Mississippi Polices are Fueling the HIV Epidemic: “Thousands of Mississippians are at risk for HIV, and many who are infected are denied lifesaving measures and treatment because of counterproductive state laws and policies, Human Rights Watch said in a report released today.”

A new study from researchers in STL came out that makes me want to get a refund on my women’s health magazine subscriptions cause I swear they all say that apple shaped people are at higher risk for heart attacks and strokes, as compared to pear shaped folks. But a study of 220,000 published last week says that the distribution of fat has no impact on risk and instead it’s all BMI.

I know I like to think I am allergic to the cold weather, especially in response to having to leave my home when there is a foot and half of snow on the ground with freezing rain in my face.  But there are actually people allergic to the cold, suffering from the allergy cold urticaria. A 9 year old was on the Today show talking about her allergy.  She is a rockstar because she says she doesn’t mind it because “it keeps her from having to go outside in the winter to walk the dog or shovel the snow.” I know I could NEVER survive without ice cream or popsicles or slushies or….. (clearly I have the taste buds of a 9 year old)

Enjoy Sunday Funday!

He takes a while to get to the punch line, but it is kinda funny…

Since everyone read my post giving the 411 on epidemiological designs I have found the perfect study to write about.  The reason I find this to be perfect is because it may present potential for a great excuse to blame your parents for any bad grades you get.  Just say, “It wasn’t me, it was all that processed food you gave me when I was a toddler”.  This is according to an article from CNN, and to be honest I usually take anything CNN says as fact as long as Anderson Cooper is saying it (That silver fox! Picture provided so everyone can swoon…).  But in this blog I will talk about whether we can take the study as is, and if CNN does a good job reporting (we might not be able to believe everything even if Anderson Cooper’s face is saying it, and I know am I not the only one with feelings about this man cause there are a ton of weird tribute videos).

Before I start Anderson wants to define two important things for you.


All loving of Anderson aside, the CNN article is actually from Sanjay Gupta’s blog and though short it does a pretty good job explaining the study and some limitations.

I read the published study (in all my free time…) that was conducted in England by researchers from University of Bristol, University of Toronto, and McGill University, from the Journal of Epidemiology & Community Health.  Was this a case of chance that the kids with high amounts of processed food had lower IQs? (Or crafty skills of those British folks and their accents)

The Study (This sounds like a super official title I know boring) The Avon Longitudinal Study of Parents and Children (ALSPC) recruited pregnant women with due dates between April 1991 and December 1992 in Southwest England and had a cohort (Cohort studies are when a group of people and followed overtime and then measured to see who develops the disease. This is used to see if a certain exposure or risk factor has an effect on the development of a disease—in this study we are following children to see the eating habit (exposure) effect IQ’s (disease) (Gordis, pg. 167)) of the 13988 children (screaming, no doubt) and their 14541 pregnant mothers (emotions, cravings…oy).  The mothers (the women whose children survived from the 14541 pregnant mothers) answered self reported surveys.  When their children (those participating from the 13988) were 7, they began to take physical and psychological tests.   IQ was measured with a shortened version of the Wechsler Intelligence Scale for Children (yeah I don’t know, I had to look this up… hope that doesn’t say something about my IQ).  7044 children from the 13988 in the study participated in these IQ tests at an average age of 8.5.   The frequency and range of the children’s diet was taken at ages 3, 4, 7, and 8.5 years with a food-frequency questionnaire that measured energy intake.

What did they find? When adjustments were made for confounding factors (defined by reference.MD Factors that can cause or prevent the outcome of interest, are not intermediate variables, and are not associated with the factor(s) under investigation)

When this was done many of their associations between the three different eating categories: processed-high fat high sugar food, traditional-meat and potatoes, and health conscious-including lots of vegetables; and IQ levels became insignificant. However, there still showed an association between processed diets and lower IQ scores.   This was most significant when processed foods were consumed to at least age 3.   There was an association between higher IQ levels and health conscious diets, but the data wasn’t as significant. Though if you look at all of the data the associations are pretty weak across the board and the authors note this.

The question then is can we really blame our parents for how they fed us when we were younger (you know those happy meals with the cute toys—unless you are in San Francisco the crunchiest city in the world, I swear they must only eat granola) for our lower IQ scores? There are a few things the CNN article doesn’t point out that are mentioned in the journal article.  Firstly, most of the respondents were girls with educated parents who owned their homes.  This is an example of selection bias.  Also, anytime you are asking people to do a questionnaire there is a chance they won’t remember the information correctly or recall bias (Gordis, pg. 250). Respondents also might lie to write what they think the researchers want to hear, which is called reporting bias.  In this study you would be concerned about whether mothers would remember exact eating patterns of their children, or try to say their kids ate certain diets because they thought that’s what the researchers wanted to hears. The study says they adjusted for many factors, which I listed, among other things in a table below.

What to Make of the News Reporting on Studies The CNN article isn’t telling people to jump to any conclusions (which means you probably can’t sue your parents for bad grades), and they explain that the associations are weak.  The article even quotes that the American Academy of Pediatrics calls for more research:    “Dr. Sandra Hassink, chair of the American Academy of Pediatrics Obesity Leadership Workgroup, agrees with the study authors that “these are weak and novel associations, “which means it doesn’t actually prove that a diet of processed food causes a lower IQ. Hassink says there are “so many variables in a child’s life,” which makes it very difficult to tease out what exactly is leading to a drop in IQ assessments.”

Also, who knows if you can generalize this for an American population? America has such a diverse population compared to many other countries there are many other factors at play.  We are definitely the melting pot—though I prefer chunky stew—when it comes to race, culture, and economics even in a small geographical area (For example think St. Louis and all that great gentrification at its best *loaded with sarcasm).  All these differences mean different cultural backgrounds, genetics, and habits, making some of these studies that come from such homogenous areas not as easy to employ here or get the same results.

The CNN article does justice to being critical of the article.  What is interesting is to read some blurbs from other news sources who just took the study and ran with it (reference them in your attempt to sue your parents over this stuff).

FoxNews just straight up tells you Feeding Toddlers Junk Food Leads to Lower IQ (I will leave my politics aside, but I can’t act like I am surprised they didn’t do their research).  The Toronto Sun just says Processed foods are bad for kids’ brains: Study and assumes its fact (those Canadians *I refuse to make an Eh!  reference, but I do love the Blame Canada Song).                                                                   The CNN article doesn’t make such proclamations and tries to caution rushing to suggest the study’s claims of association are fact.

They note:   “The study authors note that in this paper ‘we report weak but novel associations between dietary patterns in early childhood…with general intelligence assessed at 8.5 years of age.’” And clarifies the impact on ages:  “Their research also suggests that what a child eats in the first three years of life is associated with a modest decrease in intelligence, but what a child ate at age 4 and 7 did not.”

It’s actually fun to Google the topic and see how many articles just ignore most detail of the study, and assume it shows a complete association between processed food and lower IQ.  I recommend this in your free time.  Trust me there are lots.

So out there in that empty space that is the internet: Do you think CNN did a good job?  Do you trust how the media reports studies?

All images, quotes, photos, and the likeness of Anderson Cooper are used for the fun of this blog and don’t reflect actual comments he has made (Duh!  I mean come on; he is way more suave than I am when it comes to speaking)

Article Citation: Northstone, K., Joinson, C., Emmett, P., Ness, A., & Paus, T. (2011). Are dietary patterns in childhood associated with iq at 8 years of age? a population-based cohort study. J Epidemiol Community Health, Retrieved from http://jech.bmj.com/content/early/2011/01/21/jech.2010.111955.abstract doi: doi:10.1136/jech.2010.111955

Okay, so I listened to the feedback from all the many readers of my super awesome blog (a.k.a. my friends who I guilt tripped into reading it by claiming it was the funniest thing I had ever written seen online).  But my skills of trickery didn’t elude anyone everyone, because people quickly realized it was written by me and suggested I break up the posts for the assignments.  Obviously you can’t handle the funny actual educational information in one long post.   So, en lieu of my promise to not be as boring as Wikipedia or worse, The Encyclopedia Britannica, I decided that I would first give you some useful information about epidemiological studies, in case, you know, you get bored and decided to read up on epidemiological studies.  What…? Just me…?

Since I have established that epidemiology is really a jack of all trades field, you shouldn’t be surprised that there are lots of ways to conduct epidemiological studies. The overarching two categories are descriptive and analytical, and then these two are broken down further.  I am a big fan of trees (all trees really… give me a tree and I will lay under it, dance around it, tie a bow on it, hug it, or cut it down and make paper… wait…I mean…).  So, in honor of all the trees that will soon (please soon) be budding with leaves, I will give you your very own tree of epidemiological designs.


My skills in Microsoft Paint are clearly amazing (this took all of 3 hours… ugh perfection).  And if that is not enough, I will break/saw the bottom roots of my tree (I know I am killing this tree thing).  If you click on the links there are some short videos for better explanation.

Case Series: In a case series you are collecting data on patients who have a disease.  You aren’t comparing different groups of people; the patients all have the same disease.   With a case series you can look at the characteristics of the groups.  Because this is just a collection of cases, there isn’t much to it alone, but it may make people want to investigate more.

Ecological: Ecological studies show potential associations between characteristics in a population. Data may show a positive or negative correlation between X characteristic and Y characteristic in a population.  So for example you can look at the incidence of ulcers with the consumption of alcohol products in college towns.  Perhaps you find a positive correlation (the more alcohol consumed in the town the more incidence ulcers in the town).  But if you have not figured out it yet, this kind of study can be iffy, because the data is not individually based so we don’t have the data for each individual’s characteristics (as with everything they have names of this, it’s called ecological fallacy-this is your word(s) of the day, who needs that dictionary.com app!).  So in our college example, we don’t know if the cases of ulcers have to do with everyone hitting the bottle, or could be something else (Gordis, pgs. 228-230).

Cross-sectional: These are basically a snap shot of time (I’m tempted to make a Polaroid reference, but they decided to stop making those things). This study looks at a population at a specific time so you can look at the incidence of some disease in a given population and characteristics of that population (Gordis, pg. 197).  But because we don’t know about the temporal characteristics of a disease, there are some factors you have to worry about when saying there is an association between the different factors.

Case control: This study involves matching individuals who have a disease (case) with people who don’t have the disease (control), and collect data on who was exposed and who wasn’t. It allows you to make a comparison between the two groups and see where the groups differ back in time (Gordis, pgs. 177-178).

Cohorts: Cohort studies are when a group of people and followed overtime and then measured to see who develops the disease. This is used to see if a certain exposure or risk factor has an effect on the development of a disease (Gordis, pg. 167).

Now if you have gotten to the bottom of this post you are probably thinking two things 1. WOW! You are so smart now and totally get epidemiological designs and 2. I am a huge liar and this is still a page and not a short post at all.  I will only believe you are thinking the first thing and ignore any vibes (or comments) I get towards the second because I have the shield of my computer screen and a snuggie (I am definitely lying about that; I refuse to wear a snuggie!).

If I wore a snuggie I would only wear this one, and look just as fly as this kid.