Happy Thanksgiving everyone! After all that food and fun, Healthday and Babycenter.com are back to remind us that when it comes to our children, now, more than ever there are more reasons to be terrified all day, everyday. Hyper-vigilance seems like the only answer. ENOUGH IS ENOUGH. This post will illustrate the difference in literal meaning of rate of incidence vs. rate of observation/diagnosis (and it's a BIG one) and examine what I call the REALLY small denominator problem.
"Rate of eating disorders in kids keeps rising
Eating disorders have risen steadily in children and teens over the last few decades, with some of the sharpest increases occurring in boys and minority youths, according to a new report."
http://www.babycenter.com/204_rate-of-eating-disorders-in-kids-keeps-rising_10343138.bc
My friends at Babycenter are back with a doosey. Eating disorders are among the most serious of mental illnesses and not to be trifled with. If you or a loved one are exhibiting any of the signs of an eating disorder as described here http://www.nimh.nih.gov/health/publications/eating-disorders/complete-index.shtml , seek help immediately. Go directly to a qualified professional; Do not pass go; Do not collect $200.
http://www.babycenter.com/204_rate-of-eating-disorders-in-kids-keeps-rising_10343138.bc
My friends at Babycenter are back with a doosey. Eating disorders are among the most serious of mental illnesses and not to be trifled with. If you or a loved one are exhibiting any of the signs of an eating disorder as described here http://www.nimh.nih.gov/health/publications/eating-disorders/complete-index.shtml , seek help immediately. Go directly to a qualified professional; Do not pass go; Do not collect $200.
Now that that's out of the way, back to the skewering.
According to this headline, more and more children have eating disorders every time we turn around. The evidence cited is a recent study that shows that , "hospitalizations for eating disorders jumped by 119 percent between 1999 and 2006 for kids younger than 12."
RATE OF INCIDENCE AND RATE OF OBSERVATION
So the correct headline is "Frequency of hospitalizations for eating disorders in kids increased." Definitionally, more children may have eating disorders, but that's not what we see here. The other thing is that we don't really know if the rate of INCIDENCE of eating disorders is on the rise. All we DO know is that more cases are being diagnosed, documented and treated. It could even be that eating disorders are less prevalent than before, but because of better reporting and diagnostic tools, the rate of OBSERVATION is increasing. The rate of incidence implies that the absolute rate of eating disorders in the entire population is X. The rate of OBSERVATION refers the number that were counted and classified as an INCIDENCE of an eating disorders in the entire population. So if Susie had anorexia nervosa in 1998, but was untreated and undiagnosed, she still HAD anorexia nervosa. The INCIDENCE of anorexia nervosa was one. However, the number of observations for Susie would be zero.
So the correct headline is "Frequency of hospitalizations for eating disorders in kids increased." Definitionally, more children may have eating disorders, but that's not what we see here. The other thing is that we don't really know if the rate of INCIDENCE of eating disorders is on the rise. All we DO know is that more cases are being diagnosed, documented and treated. It could even be that eating disorders are less prevalent than before, but because of better reporting and diagnostic tools, the rate of OBSERVATION is increasing. The rate of incidence implies that the absolute rate of eating disorders in the entire population is X. The rate of OBSERVATION refers the number that were counted and classified as an INCIDENCE of an eating disorders in the entire population. So if Susie had anorexia nervosa in 1998, but was untreated and undiagnosed, she still HAD anorexia nervosa. The INCIDENCE of anorexia nervosa was one. However, the number of observations for Susie would be zero.
What we see here is that more children are being hospitalized for eating disorders. This may not even be a bad thing. What if, in the past, children with eating disorders were not hospitalized and instead were under-treated? Couldn't that mean that children were dying, carrying their disease longer into adolescence and adulthood, remaining in a sub-clinical state of misery for untold periods of time? Now that's awful. While hospitalizing anyone for anything is serious, not treating an illness that should be treated by hospitalization shows much more serious systemic problems (ignorance and neglect on the part of adults, mostly).
On the other hand, maybe it's even worse. What if, in the past children were treated by mental health professionals and their pediatricians at an early phase and because of increasingly stringent and bizarre rules pertaining to health insurance plans limiting mental health benefits, children are having to reach death's door (literally) before they receive the treatment that they need? That's a different awful problem.
On the third hand (foot, I guess), what if these children were being hospitalized with a different primary diagnosis? Perhaps they have other illnesses such as depression, severe abdominal distress, or anything else for that matter and the eating disorder was not considered the primary cause of the hospitalization?
And yet another foot, what if eating disorders weren't listed as the diagnosis code because some doctors knew that insurance would be more difficult to deal with? According to someone I know, in the 1950's many doctors admitted Mrs. Smith with acute appendicitis and performed an appendectomy at the same time as an abortion.
On the third hand (foot, I guess), what if these children were being hospitalized with a different primary diagnosis? Perhaps they have other illnesses such as depression, severe abdominal distress, or anything else for that matter and the eating disorder was not considered the primary cause of the hospitalization?
And yet another foot, what if eating disorders weren't listed as the diagnosis code because some doctors knew that insurance would be more difficult to deal with? According to someone I know, in the 1950's many doctors admitted Mrs. Smith with acute appendicitis and performed an appendectomy at the same time as an abortion.
The truth is that we don't know if more children have eating disorders than in the past. We do know that more are being hospitalized at a young age with the primary reason for hospitalization being cited as an eating disorder.
THE TINIEST OF PROBABILITIES AND THE REALLY SMALL DENOMINATOR PROBLEM
The other disturbing part of this article is the idea that more and more boys and minorities are really seeing more eating disorders. Here we go again. Is it that these kids didn't have eating disorders, or that they went untreated and unreported? My guess is the latter, but I can only speculate because there is no base control to work form. But here's the more important lesson. the rate of anything occurring is a probability. That is a fraction reduced to 1 in X. A three fold increase sounds REALLY big but what if that is the increase from 1 in 1 Billion to 3 in 1 Billion? In the grand scheme of things, it's not a very significant change. It certainly wouldn't warrant a change in policy for the group that is 4x the population of the United States. (That's not to say those 3 cases aren't important, you aren't likely to come across the on or the three in your everyday life).
In this article we hear that the sharpest increase is for boys and minorities. Let's do the math.
THE TINIEST OF PROBABILITIES AND THE REALLY SMALL DENOMINATOR PROBLEM
The other disturbing part of this article is the idea that more and more boys and minorities are really seeing more eating disorders. Here we go again. Is it that these kids didn't have eating disorders, or that they went untreated and unreported? My guess is the latter, but I can only speculate because there is no base control to work form. But here's the more important lesson. the rate of anything occurring is a probability. That is a fraction reduced to 1 in X. A three fold increase sounds REALLY big but what if that is the increase from 1 in 1 Billion to 3 in 1 Billion? In the grand scheme of things, it's not a very significant change. It certainly wouldn't warrant a change in policy for the group that is 4x the population of the United States. (That's not to say those 3 cases aren't important, you aren't likely to come across the on or the three in your everyday life).
In this article we hear that the sharpest increase is for boys and minorities. Let's do the math.
"...experts estimate that between 0.8 to 14 percent of Americans generally have at least some of the physical and psychological symptoms of an eating disorder, according to the report.
Boys now represent about 5 to 10 percent of those with eating disorders, although some research suggests that number may be even higher, said Lisa Lilenfeld, incoming president of the Eating Disorders Coalition for Research, Policy and Action in Washington, D.C."
Ooh - catch that the first paragraph and the second are using 2 different data sets. Yikes (that's a problem for another post).
If math makes your eyes glaze over skip to the BIG bold part for a summary...
US population clock :
310,814,706
For simplicity let's split the difference between 0.8 and 14 and say 7%...
310,814,706 * 7% = 21,757,029
21,757,029 * 10% (high-end estimate for boys) = 2,175,703
Let's just say that in the next year we saw 2,300,000 and the population stayed flat. the percentage increase would be (2,300,000-2,175,703)/2,175,703 = 5.7% increase.
Since non boys represent the rest of the eating disorders population is 21,757,029 * 90% = 19,581,326
If the same increase in CASES was seen there (about 124,297 CASES), we would see that there was an increase of (124,297/19,581,326) = 0.63%.
5.7% vs. 0.63% HUGE difference but the same number of cases. So when you hear that the increase is sharper for boys this makes sense, they are starting off a much smaller base.
If you sold one car today and two cars tomorrow, you had a 100% increase in the number of cars sold. The next day you sold three cars but you only have an increase of 50%, and so on... As the denominator increases the % change will continue to decrease.
Really, what we see is a statistically insignificant difference wherein, 1 more person in 2,300 (2 people in 2,300 - which is the difference of 0.04% and 0.09%) may attempt suicide. Naturally this is something that one may take into consideration when deciding if drug therapy for acne is right for them. Perhaps, candidates for treatment may need to be screened for depression prior to and during treatment. But to assume that an otherwise happy, well-adjusted person with acne will double their risk of attempting suicide because they took an acne drug would be illogical and irrational.
One thing that I liked about the researchers was that rather than trying to defend the idea that the drugs WERE somehow linked to suicidal behavior, even though the data said otherwise, was that they actually considered other reasons why this sample may have had even a slightly higher incidence of depression. Turns out, having a painful and visible illness might be enough after all, without blaming "big pharma."
Who knows? The point is the headline is MISLEADING and in order to get to the headline, the writer needed another doctor to cough up a quote. Watch for my next post about necessary and sufficient conditions and why we need both together in order to draw rational conclusions.