from the some-correlation-is-still-better-than-none dept.
GungnirSniper writes:
"In the US State of Washington, the rare birth defect anencephaly has become slightly more common, worrying would-be parents and baffling epidemiologists. TechTimes.com reports that the health records of a single three-county area in Washington State 'revealed 23 cases of anencephaly in 36 months, between January 2010 and 2013. This translates to a rate of 8.4 births out of every 10,000. That is four times the normal occurrence for the rare disorder.'
A group of epidemiologists working for the state's Department of Health reported finding no clear cause for the exceptional prevalence of this fatal birth defect. But they are now accused of not looking hard enough for the cause. Dr. Beate Ritz, who has done several studies on birth defects, told CNN that the data quality on medical records, which were the primary source of data used in the study, 'is so low that it's not really research'.
Washington's Department of Health has admitted that 'Medical record reviews might not have captured all information, preventing a cause from being identified,' and says its officials will continue monitoring births, and look for possible causes.
(Score: 5, Insightful) by Sir Garlon on Monday March 03 2014, @01:09PM
Kudos to the submitter for putting proper numerical perspective in the summary - "rare defect" becoming "slightly more common," along with actual number. In TFS! This is the sort of thing I, as a nerd, thrive on.
[Sir Garlon] is the marvellest knight that is now living, for he destroyeth many good knights, for he goeth invisible.
(Score: 4, Interesting) by girlwhowaspluggedout on Monday March 03 2014, @01:32PM
I'd like to second those kudos. Although I -- as a stickler for accuracy and details -- sometimes add such data to submissions, this time both the numbers and the non-sensationalistic tone were already present in GungnirSniper's FS.
Keep 'em coming!
Soylent is the best disinfectant.
(Score: 3, Informative) by Grishnakh on Monday March 03 2014, @04:14PM
This is definitely something you wouldn't see on Slashdot. The summaries there have been atrocious for years.
(Score: 0) by Anonymous Coward on Monday March 03 2014, @06:35PM
about 20 cases, and the state health department has already done an early investigation? I was hoping that something new could have been learned, such as a given chemical causes birth defects, I guess this is merely an overactive local TV station, and state health officials.
(Score: 3, Informative) by frojack on Monday March 03 2014, @07:26PM
Four times the norm is still pretty big, but without any indication of how widely the norm varies it is still not enough to male a judgement.
Only by digging another level deeper into the CDC analysis [cdc.gov] do you find:
Now it starts to look a little more clear that while the results were significant, the confidence interval is quite wide (probably due to the tiny size of the study). The calculate rate may not be reliable at all, because the time frame is too small, the area sampled is too small, and the study was done without detailed interviews or site assessments.
Randomly distributed results may still occur in clusters, and if you base your statistical analysis on such an accidental cluster, you tend to come up with results that look way more significant than they really are.
Still when the lowest boundary of your confidence interval is twice the upper level if the national CI, further study is warranted.
No, you are mistaken. I've always had this sig.
(Score: 0) by Anonymous Coward on Wednesday April 09 2014, @11:30AM
sYDyr7 http://www.qs3pe5zgdxc9iovktapt2dbyppkmkqfz.com/ [qs3pe5zgdx...kmkqfz.com]
(Score: 5, Interesting) by MrGuy on Monday March 03 2014, @01:13PM
...is the comments that the "investigators" never even interviewed the mothers (or even the families) to identify common patterns.
That's like trying to trace the outbreak of food poisoning without asking the patients where and what they ate.
Sure, there's at least some merit in thinking asking mothers about their children with birth defects might be upsetting. But if you're too squeamish about that to actually, y'know, RESEARCH, then you're an absolute fraud to claim you "investigated" anything.
(Score: 5, Insightful) by hubie on Monday March 03 2014, @04:42PM
Going back through medical records to find out the number of cases is the first step in the research. The more telling part is:
The article does not say whether the State funding is limited, and it doesn't put the numbers in context enough for us to know much at all. Certainly if you see 8.4 events when you expect to see 2.4 events, this is statistically significant (and that assumes the incidents are statistically independent, which goes against what is implied in the tone of this story), but that 2.4 is a national average. Is the number of incidences more or less the same nationally? Were those 8.4 cases uniform over the 3 years, or did they all clump? And if they did clump, it could be just plain old coincidence. This article [nbcnews.com] gives a better description of the situation than the one in the article summary. Given the resources of the State and what other priorities they have, 8.4 cases per 10,000 births might not be above the threshold for doing a full-out investigation.
Maybe there is something here, maybe not. Given only what is presented in the news article, I think calling them "absolute frauds" is rude and entirely unfair.
(Score: 3, Interesting) by MrGuy on Monday March 03 2014, @04:59PM
Disagree.
It's a reasonable position to take that "we don't have the resources to investigate every possible statistical anomaly in detail." I'd have no problem with such a statement.
A statement that "we've investigated and found no common factor" when you've done very little in terms of actual investigation is misleading at best and in my opinion is (as I said) fraudulent coming from someone who's a medical scientist who ought to know better. It's giving the public reassurance that "there's nothing to see here" when you don't know that at all.
(Score: 3, Informative) by frojack on Monday March 03 2014, @08:14PM
The study they did, amounted to a medical records pull, (and probably bent a few confidentiality laws in the process).
At best it was a cursory glance, and the numbers may not (yet) warrant much further.
HOWEVER: This gets much wider play because the the nurse who noticed this works in the Richland Kennewick Pasco area (commonly called the Tri-Cities area in Wa).
This area is just downstream from the Hanford nuclear site [google.com].
Hanford [wikipedia.org] has huge underground tanks, some of which are leaking radioactive waste quite badly and apparently have been for some time.
The three counties (not mentioned in national press, but common knowledge in Washington state) all border the Columbia River and may tap the same aquifers (or river water) being polluted by Hanford leaks.
State residents are more than a bit pissed off about the Federal Governments handling of the Hanford cleanup.
So you start to see the tie in here (rightly or wrongly) with resident rage over Handord.
This kind of cluster in Minot North Dakota might not not even be noticed.
No, you are mistaken. I've always had this sig.
(Score: 2) by hemocyanin on Monday March 03 2014, @09:13PM
Fuck the Feds. Free Cascadia.
(Score: 1) by demonlapin on Monday March 03 2014, @11:32PM
As long as the IRB approves it, not really. And dead babies don't have privacy interests.
(Score: 2) by frojack on Monday March 03 2014, @11:53PM
One does not research dead babies.
One researches living mothers.
No, you are mistaken. I've always had this sig.
(Score: 1) by demonlapin on Tuesday March 04 2014, @03:25AM
(Score: 1) by geb on Monday March 03 2014, @01:14PM
A brief look through the articles seems to show a few relevant numbers, including average rate, the increased rate observed, and confidence interval (i.e. percentage chance that the figures reflect actual reality)
There was a mention of statistical significance, but only to say the rise was not observed to be linked to any factor studied (smoking, alcohol, health of the mother, etc).
What I did not see was any mention of the chance that this rise in rate would happen without any unusual influence, or statistical significance of the spike. Did I miss it?
(Score: 0) by Anonymous Coward on Monday March 03 2014, @01:41PM
Some more numbers:
(Score: 4, Informative) by BradleyAndersen on Monday March 03 2014, @02:02PM
(Score: 2, Informative) by MichaelDavidCrawford on Monday March 03 2014, @01:21PM
My ex-wife's older sister, a deeply faithful member of the Salvation Army church, got the bad news late during her pregnancy that her child was anencephalic.
At least in her case, this could have been prevented by taking Folic Acid. You don't need a prescription, it's over-the-counter, all drugstores will have it, most grocery stores will too.
Ask around among your relatives whether anyone at all - in the entire history of your family as well as your spouse's - has been born with birth defects. If so, take extra, it's safe. A pharmacist will tell you what your dose should be if you need to take extra.
My sister-in-law was faced with the excrutiating decision to have a late-term abortion. Because of her age, that was the last chance for her to have a baby.
Oh yeah to change the subject: I read in the Vancouver Washington paper the other day, that some new liquid radioactive waste tanks at the Hanford Nuclear Reservation are leaking.
I thought that was a solved problem decades ago. My father used to be a Nuclear Engineer for Mare Island Naval Shipyard in Vallejo, California. He told me that when they change the submarine reactor water, they make concrete blocks out of, then the concrete and the cement mixer are all buried at hanford.
Now there are other kinds of liquids, but most liquids will react with SOMETHING to make a solid.
If there is no way to chemically solidify a liquid, you could always freeze it.
Yes I Have No Bananas. [gofundme.com]
(Score: 3, Interesting) by fadrian on Monday March 03 2014, @04:24PM
First of all, we're only talking three cases here - individually devastating, but statistically, given the small sample size, talking about X per 10K or Y per M doesn't really make a lot of sense, since over the next few years, the samples will probably revert back to the mean.
Second, if these happened (say) in Seattle, Yakima, and Vancouver, there might not be a "common cause", as these cities are separated by hundreds of miles. Even if they did all happen in one area, you'd have a pretty good chance that this is random and not correlated by anything other than the time of occurrence.
What you have here is almost certainly a statistical anomaly, not an epidemic. And this is what happens when you have doctors and reporters (who know so, so little about statistics) trying to make sense of random chance.
That is all.
(Score: 5, Informative) by hemocyanin on Monday March 03 2014, @05:30PM
Three cases?
http://www.cdc.gov/mmwr/preview/mmwrhtml/mm6235a5. htm [cdc.gov]
In that same CDC summary none of the counties are enumerated. Take a look, they aren't there.
After some googling, it turns out that we're talking Yakima, Benton, and Franklin counties. http://www.tri-cityherald.com/2014/02/22/2843413/b irth-defect-in-benton-franklin.html [tri-cityherald.com] This particular area of Washington state gets inundated with pesticides and the rates amongst hispanic peoples is higher (see link above). And of course, the Hanford site is smack dab in the middle.
Map showing counties:n -county-map-large.png [coilgun.info]
http://www.coilgun.info/rover_wa/images/washingto
Hanford reservation is a huge block of that (shaded gray), just east of dead center:6 .5124226,-119.4993278,10z/data=!3m1!4b1!4m2!3m1!1s 0x54985a3722b17805:0x350eb8746e4035a7 [google.com]
https://www.google.com/maps/place/Hanford+Site/@4
If it does have anything to do with Hanford, you could be certain the CDC would be under pressure to not figure that out, and rolling downhill, state health officials as well. With reports on the study being extra crappy ( http://www.cnn.com/2014/03/01/health/cohen-birth-d efects/ [cnn.com] ), well, it is reasonable to wonder why.
(Score: 3, Informative) by hemocyanin on Monday March 03 2014, @06:04PM
Some other interesting citations:
"Changing incidence of anencephaly in the eastern Black Sea region of Turkey and Chernobyl." http://www.ncbi.nlm.nih.gov/pubmed/2374745 [nih.gov]
"The prevalence at birth of congenital malformations in communities near the Hanford site." http://www.ncbi.nlm.nih.gov/pubmed/3337080 [nih.gov]
(Score: 5, Insightful) by mmcmonster on Monday March 03 2014, @05:23PM
I am a physician.
Most medical records from an outpatient office are useless. More so if they are paper charts/not electronic medical records (EMR). But even most outpatient records in an EMR system are useless. They make sure all the data required for adequate billing are correct, but things aren't updated regularly.
For instance: The EMR on the last patient I saw said that he had heart surgery 3 years ago. I asked him what he had. He said that he had a cardiac stent in 2008. No one had updated the '3 years ago' line since then until I saw him for the first time today!. A big part of my documentation is to sanitize everyone else's documentation so it makes sense in the future. I use absolute dates and get rid of the '3 years ago' and things like that.
And God help the researchers if the medical record was a paper chart. They'll just have the vital signs, and quick exam, and a few scribbles about what the physician thought was important that day.
Proper research on something like this has to involve the researcher (or maybe even the CDC) going personally and interviewing every parent and seeing if they can get a causal relationship based on things that would never make it into a medical record.
(Score: 1) by gottabeme on Tuesday March 04 2014, @05:19AM
Maybe you're a doctor or are extremely non-squeamish, but when I went to the Wikipedia link to confirm what I thought the disorder was, I wasn't expecting all the photos to be right there. Seeing the photos of the real, deceased, human-but-not-human fetuses was a bit disturbing. Recommend avoiding it.