Saturday, February 27, 2010

UK DonorLink "mistake" exposes serious concerns about registry effectiveness

Last week the Daily Mail reported on the story (Despair of the DNA 'sisters' - 22 Feb 2010) of the two donor-conceived women who had been matched as siblings by the UKDonorLink, a government-funded national volunteer donor offspring registry, but now the agency claims they are in fact NOT related.

To sum up what was published:

Keeley Hall of Perth, Western Australia and Elizabeth Howard of Cambridge, England were matched by the UKDL registry in 2007. After meeting, they developed a close friendship despite the distance between them. However, several months ago Elizabeth received a shocking piece of information from UKDL. It turns out that another offspring matched with her, but this girl was apparently un-related to Keeley. While Liz and Keeley only matched with a statistic of 56.6 times greater of being sisters than not related, Liz and the new donor offspring matched with an index over a million times greater! Yet, Keeley did not match the new offspring. Thus, according to the UKDL, it made Liz and Keeley's match less likely of being a positive result.


Now for some science background information:

The UKDL, as I have ranted about in the past, has recently changed their criteria of what they call a sibling match to a whooping 99% probability. To bring this into perspective, here in the USA, if a siblingship test is needed in court to prove relatedness, 90% is considered conclusive and admissible. The 56.6 index that Liz and Keeley were assigned in 2007 is still at a highly significant 98% probability, well over what is considered conclusive by most testing companies and the US court system. However, they were not able to now reach the new excessively high threshold established by the UKDL of 99% probability.

So instead of admitting that they have changed their criteria and that Liz and Keeley, while they were obviously considered sisters on the old criteria, have been pushed below the new criteria, they argue that the testing measures and analysis have changed (in 3 years) and now they doubt that Liz and Keeley are actually siblings!! It is as if they do not want to admit that their unbelievably high threshold rating is the cause of this "mistake" as it calls into question their own practices and ethics. If Liz and Keeley have been now considered as not being siblings, how many others are in the same situation who have also not been contacted and informed in this "ch???

But the biggest concern for me, however, is that a significant portion of registered offspring DO have siblings on the UKDL registry, but they are NOT being contacted about the possibility!! While setting the threshold at 90%, as most labs and courts believe, provides enough proof that the two individuals are in fact related, it gives some room for leeway. Remember, half-siblings only share approximately 25% of their DNA, so out of the 15-20 markers tested in most labs, that means that true half-siblings should on average only match one allele at 7-10 of the markers!! If none of those matched alleles are very rare in the general population a siblingship index in the 50s, or even lower, as Liz and Keeley's results, could easily ensue.


So what needs to be done?

The UKDL needs to fess up to their outrageous criteria, apologize to those who have been wronged in the process, and set their criteria at a discriminatory yet not impossible level (i.e. 90% probability), so that the offspring registered with the UKDL have a fighting chance to find their relatives, rather than be played by a government-funded system designed to be as little work as necessary for their employees at the expense of human beings right to search, find, and know their biological kin.

It is a situation like what has been made public in the circumstances of Liz and Keeley that has prompted Damian Adams and I to look towards creating our own DNA database for donor offspring, that contacts individuals with any positive (over 50% probability) siblingship index, where they can either consult further testing arrangements, or compare physical genetic traits such as eye color, blood type, location of conception, etc, to determine in their own minds if they are siblings or not.

Siblingship tests are not perfect. They are not 100% conclusive, but that does NOT mean that they cannot provide this valuable information to two individuals. If two potential siblings receive an 'inconclusive' result of say 80%, it should be their choice, not some bureaucratic exec in an office, that decides if they want to believe this as a true match, as unrelated, or as something that needs to be further investigated using techniques like X and Y chromosome testing, or an extra panel of STR markers.

1 comment:

Anonymous said...

I can kinda ee why they want the answer to be "all but certain" (99%) before they give it their official and legally actionable imprimatur. It's chickenshit, but that is bureaucracy for you. Like you said, though, there is the assumption that they are making that the call is, in fact, even theirs to make.

It seems to me that there MUST be some anonymous format to let people see the data in context, so the get a visual feel for what it means. Off the top of my head, something like a a set of concentric circles where the dot at the center is you (perfect match index) and the outermost circle is everyone in the dataset. Then you can do a plot where there is a fog of dots at the outside that thins out as it gets towards you. For any one person, you can see where their dot would be and decide if that is "close enough" in the context of all that other data that you want to do something about it. OK, there are probably a dozen better and more information-rich ways to do it, but you see the point? -Hernan