Sunday, September 21, 2014

You are at RISK!

Open a newspaper or read a medical journal or look at your Twitter feed and you'll learn a sad fact: something you're doing -- and probably something rather fun -- puts you at increased risk of some bad disease.  It sounds very ominous and you should stop doing those fun things.  Sometimes, I have to scratch my head and say 'Really? What does that actually mean?"  And while I've decided that my Diet Coke habit probably isn't particularly healthy and have mostly, sort of, cut way back, it caught my attention when I saw that one diet soda a day increases the risk of stroke by 16%.  Actually, it caught my attention because it was in my social media feeds about a zillion times tweeted by the Cleveland Clinic then retweeted by countless others.  Really?  How can anyone know that?  

So the first thing I wondered was how common is stroke?  I mean, a 16% increase is not huge, it's not like it makes you 5 times (a 500% increase) likely to have a stroke, and I didn't know how common stroke is.  A small increase for an unlikely event (stroke, for any given person for any given year) caused by a common habit (diet soda ingestion -- or actually any soda ingestion as I would soon learn) would seem like a difficult thing to quantify and measure.  And then you might wonder if 2 diet drinks/day increase the risk by 32% and if 6 diet drinks/day double it, and what about adding artificial sweeteners to coffee or tea? 

So I went to look for the prevalence of stroke and what I discovered was 2.6% for adults over age 18 have had a stroke.  But what surprised me was how the data was collected, per the CDC website:

BRFSS is a state-based, random-digit_dialed telephone survey of the noninstitutionalized, U.S. civilian population aged >18 years and is administered by state health departments in collaboration with CDC. In 2005, the median response rate among states, based on Council of American Survey and Research Organizations guidelines, was 51.1% (range: 34.6%--67.4%). This rate accounts for the efficiency of the telephone sampling method used and participation rates among eligible respondents who were contacted. A total of 356,112 respondents from all 50 states, DC, Puerto Rico, and USVI participated in the survey. State (including DC) and territory sample sizes ranged from 2,422 (USVI) to 23,302 (Washington). The racial/ethnic national sample sizes ranged from 5,535 (AI/ANs) to 279,419 (whites). All prevalence estimates in this report have a numerator >50 and a relative standard error <30 are="" ensure="" estimates="" stable.="" sup="" that="" to="">†
Survey respondents answered the question, "Has a doctor or other health professional ever told you that you had a stroke?" Differences in prevalence were assessed by age group, sex, race/ethnicity, education level, and state or territory of residence. Data were weighted to reflect the population aged >18 years in each state and territory and were age adjusted to the 2000 U.S. standard population to allow for more meaningful comparisons between states and between demographic groups. The weighted state prevalence values were used to estimate the number of persons with a history of stroke in various demographic groups and in each state or territory. Respondents provided racial/ethnic identification; those who identified themselves as multiracial were included in a separate category.

Ok, so first off,  half the people don't answer.  Are those who refuse to participate more or less likely to have had strokes?  The survey requires that someone answer the phone, so all the people who were too disabled to answer the phone are excluded from the survey results.  People with lower educational levels were more likely to have had strokes.  Could it be that people with lower educational levels were less likely to understand the question, or to confuse a doctor's warning "you could have a stroke" with "you have had a stroke?"   And could it be that people who have had strokes don't understand the question or have anosognosia so they don't know they've had a stroke?  I'm so confused. In the study below, the incidence of stroke was also self-reported but the report of a stroke was then verified by review of medical records.  I suppose they still might not capture those who have fatal strokes or those who don't know they've had a stroke.

And back to the Cleveland Clinic Study -- actually it's diet soda and sugar containing soda that increases risk. Switching to skim milk apparently decreases risk by 11%.  Go figure.  In this study, the participants were asked for a self report with forced choice responses (none, up to one/week, one per week up to one/day, more than one per day, etc).  Coffee consumption, by the way, lowered risk, but I couldn't figure out if drinking coffee and skim milk would counteract the risk of drinking soda.  Before I wash away....


The Alienist said...

It's always difficult to interpret correlations. You never know what cause (or causes) are being observed. Is the connection observed due to obesity, anxiety, inability to say "no" to a pollster, obsessive phone answering, intense need to have one's opinions registered, or unemployment leaving one free to talk on the telephone?

I also sometimes wonder when you totaled all the strokes attributed to all risk factors, if you would wind up with more statistical strokes than actual strokes?

a psychiatrist who learned from veterans said...

As the Alienist says re: correlations:
Someone asked me the other day about a report linking taking Xanax and dementia. I had seen it but couldn't find it. Anyone with the reference?

Sunny CA said...

The world of nutrition information seems to be plagued by studies like this.

I think it would be pretty safe to categorize Diet Coke as "junk food" in that there is no redeeming nutritional value. Most people would put high-fructose corn syrup Coke (it is not made with cane sugar these days) into the junk food category for everyone who is not so deprived of calories that they need fructose calories bare of fiber, minerals, vitamins. So there you have two junk food beverages.

Many people who drink junk beverages would eat junk food, so might well consume more McD bacon burgers with fries, and other fried, high transfat, high saturated-fat fast foods. The "sides" eaten along with the beverages might be causal for stroke, but it seems unlikely the beverages are causal.

However, both those beverages are mostly sold as caffeinated beverages. How does caffeine affect stroke risk? Mostly though, they are looking at weak data for correlations, not causation. This is science at its worst.

Knot Telling said...

@a psychiatrist who learned... - This is probably the study you heard about. Benzodiazepine use and risk of Alzheimer's disease: case-control study.(
The lead researcher published this one in 2012, which is also interesting and relevant. Benzodiazepine use and risk of dementia: prospective population based study. (

With regard to the diet soda - stroke risk correlation, this kind of report in the general press always seems suspect to me. Were confounding variables controlled for? Is the statistical analysis valid and reliable? And so on.

In other words, meh.

a psychiatrist who learned from veterans said...

Pretty sobering studies on benzos and dementia, for me, especially the first one. Thanks, Knot Telling.

form said...

If diet soda is unhealthy, changing its name from "diet" to something else would be helpful in convincing the public to drink less of it.