Up at 5AM: The 5AM Solutions Blog

Earlier this summer, I drove down to the Southeast Linux Fest in Spartanburg, South Carolina. One of the talks that stood out to me was given by Heather Holl, bioinformaticist and Slackware Linux team member. She talked about the open source tools she uses most in her work in equine genomics. I was especially impressed at how she used standard, open source Linux command-line tools to get her job done.

A national patient ID is one of the more taboo subjects in United States law. As one writer for the College of American Pathologists reports, “In some federal government offices, all one has to do to stop a conversation cold is mention a national patient identifier. That’s how adamantly Congress, in a 1998 bill, outlawed any plans for, consideration of, and even research on a national system of assigning patient ID numbers.”

In fact, no government money can be spent even investigating the possibility of a national patient identifier, effectively making the law difficult to reverse without a large push from private industry.

If it were created, a national patient identifier would provide a unique number to identify an individual in health systems across the United States. It could be used to pull or push health data for a patient across the country on-demand without complication or fuss.

Without a national patient identifier, industries face the challenge of mapping patients in their internal systems to patients in systems outside of their operating networks. Typically, this involves sending patient names, addresses, phone numbers, social security numbers, and other identifying data across secure channels to ensure that the systems really are talking about the same person. Mistakes and match-refusals are common. Spelling mistakes and formatting inconsistencies while entering data can prevent systems from recognizing somebody as the same person (“Bob” vs “Robert,” or “Ft. Worth” vs “Fort Worth”). Policy differences also can prevent successful matching. Some health vendors will outright refuse, by policy, to make a patient correlation without having been sent the social security number of a patient. Other vendors will refuse to ever send a social security number over the Internet, even if the channel is secure. Systems like this will never be able to communicate with such limiting policies in place.

A national patient identifier cuts through all of this red tape. It replaces complex methods of patient matching that may or may not be successful, and reduces the likelihood of confusing people with similar information for one another. Overall, both the quantity and quality of information that can be gathered under a unique identifier is greater (and cheaper!) than a haphazard industry approach trying many different solutions.

Why is it important to enable systems to “match” patients? Imagine you’re lying unconscious in an emergency room, and the doctor can’t find a record to indicate your penicillin allergy. While the idea of a ER physician being able to conduct an instant search across the globe to find your records is itself still largely theoretical, realizing that theory is much, much more difficult without a simple way to identify patients.

The current government policy feels contradictory. On one hand, government initiatives such as the Nationwide Health Information Network (and implementation in CONNECT) suggest that moving health information into the electronic space, to be readily accessible by consumers of health data, is a national priority. On the other hand, policies preventing investigation into a national patient identifier stifle the ease of rolling out such systems nationwide and limit their accuracy.  

Privacy is a major concern with a national patient identifier; however, both the concept of uniquely identifying numbers associated with people as well as the notions of ‘security breach’ are no different with a national patient identifier than with the systems that are currently in place. The information that could be gained from learning the national patient identifier would be little different from the information gained under the current approach of using names, social security numbers, and addresses. Once a person has this information, and somehow gains access to the health network, it will not matter if they have the NPI number or the separate bits of information. Also, the concept of a unique identifier exists nationally from birth in the US in the form of a social security number, and from the point of ID issuance at the state level. It is a little known fact that many states create drivers license numbers based off of the full name and date of birth of an individual. Maryland is one of those states, and based upon that number alone, anybody can tell who you are! You can try it yourself here. The real security is in maintaining the network and reducing fraud.

While in no way would the implementation of a national patient identifier be easy, it should at least be one of the possibilities that can be considered at the national level. It is in the interest of private industry to speak up and express their concern, since current government policy forbids any possible action internally.

Recently here on this blog, we discussed the issue of the growing racial divide in personalized medicine – and how it has the potential to leave non-whites behind as we move into a promising new age in medicine.

Since then, on a positive note, 23andMe has launched an initiative, called Roots Into the Future, giving away their service for free to 10,000 self-identified African-Americans. It’s a step in the right direction. Though with over 100,000 people already in their database, that’s still only 10% of their total. Of course there are already African-Americans in their customer base, but the fact that they feel the need to offer this initiative is indicative of the industry-wide divide and the general need for more diverse DNA in the study pool.

This week, the Washington Post published an article that adds yet another wrinkle. Federal patent examiners at the US Patent and Trademark office have been rejecting patents for molecular diagnostics (a mainstay of personalized medicine) that have not been validated across a variety of racial and ethnic groups. This is controversial partly because the ideas of “race” and “ethnicity” are not hard genetic concepts – a better breakdown would be along the lines of haplogroups, but even here we have the sticky problem of ancestry admixture and cultural identification. For instance, Latinos, a very large and diverse group, can come from a range of admixed ancestries, spanning Europe, Africa and native American haplogroups. The label “Latino” is not in and of itself particularly meaningful or specific enough when it comes to genetic markers. While the USPTO’s intentions are good (to make sure we don’t think that tests derived from genetic information heavily skewed to one population are necessarily valid for everyone) there is clearly more work to be done before we can claim to judge these tests fairly. And on the flipside, we’re going to need more initiatives like 23andMe’s to make sure the database we’re working from isn’t skewed in the first place.

It is important to note too that getting non-whites to participate in DNA testing is not always as simple as it sounds, especially outside of the developed world. The US Government’s use of DNA spying in the Osama Bin Laden case and some famously bungled efforts to work with the DNA of native populations have sown mistrust and bad blood that will be difficult to clear. But if we’re to bring the promise of personalized therapy to everyone, we will need to find ways to move past these differences and establish ethical guidelines that strike a balance between being open for constructive science while being respectful of heritage and tradition. We'll also need to get the regulators up to speed on what does and doesn't constitute race and ethnicity in a genomic world.