Dev.SN

Rise in Brain Birth Defect in Washington State

posted by girlwhowaspluggedout on Monday March 03 2014, @09:00AM   
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.

• Re:Good summary (Score: 3, Informative) by frojack on Monday March 03 2014, @03:26PM

  by frojack (1554) on Monday March 03 2014, @03:26PM (#10169)

  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:

  The anencephaly rate was 8.4 per 10,000 live births (95% confidence interval [CI] = 4.5–12.0), compared with a national estimate of 2.1 per 10,000 live births (CI = 1.9–2.2)

  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.

  --
  Discussion should abhor vacuity, as space does a vacuum.

  Parent

  Starting Score:	1		point
  Moderation		+1
  Informative=1, Total=1
  Extra 'Informative' Modifier		0
  Karma-Bonus Modifier		+1
  Total Score:		3