Shortage of physician-geneticists in the United States

Update 1/2/07: I added a box toward the end which summarizes the genetics workforce in more detail. Caveat emptor: The maps featured in this post are preliminary works-in-progress and do not capture all physician-geneticists working in the United States, just those for which good data are available (board-certified physician-geneticists that see patients and are members of the ACMG).

There are very few health care professionals in the United States with extensive training in genetics. The most qualified professionals are board-certified physician-geneticists. According to the American College of Medical Genetics (ACMG), in the United States, there are approximately 509 physician-geneticists that see patients and are registered with the ACMG, which includes MDs, PhDsMD-PhDs, and DOs. This number excludes PhD geneticists, genetic counselors, and physician-geneticists who are either not registered with the ACMG or do not see patients (see summary box toward the end).

For perspective, there are nearly as many professional astronauts in the world as there are board certified geneticists that see patients in the United States. These 509 geneticists are not distributed evenly across the United States. Four states have no physician-geneticists at all. California has the most, with 84. The following image shows the distribution of physician-geneticists by state:

Number of Geneticists United States

If we divide the population of the U.S. by these 509 physician-geneticists, we get a back-of-the-napkin estimate of 600,000 U.S. citizens per board certified physician-geneticist that sees patients and registers with the ACMG (300M/500). This ratio varies significantly, depending on the state. For example, there are approximately 2.8M people per physician-geneticist in Arkansas, where there is only one ACMG-registered, board certified geneticist that sees patients in the entire state.

I’ve plotted this data on a map of the United States:

View Larger Map

The limited supply of health care professionals with expertise in genetics should be considered a risk associated with obtaining a genome sequence. Services such as 23andMe, Navigenics, deCodeMe, DNA Direct, and Knome will certainly put pressure on the capacity of the existing professional infrastructure. Of course, some of these companies (and others) may also come to fill the void with innovative models for patient triage.

***Big thanks to the ACMG for providing zip code data that enabled me to generate the maps above. Update 1/2/07: Although all errors are mine! See comments.


Professional Genetics Workforce Stats
~1100 MD geneticists in the U.S.
~2440 genetic counselors in the U.S.
~500 PhD geneticists in the U.S.
*The 509 physician-geneticists described above include MDs, MD-PhDs, and DOs that see patients and are members of the ACMG.

Some background papers with more detailed information:

This month I am doing an extended meditation about the risks of obtaining and sharing personal genome sequences. This exercise will be cathartic for me. Beyond that, I hope it may help to reorient the conversation about risks into something that one day might resemble practical guidance for individuals considering obtaining and sharing personal genetic information.

2007 Holiday Reading List

Here are the books I’ll be toting around this holiday season:

property_rights.jpg Jasper A. Bovenberg. Property Rights in Blood, Genes & Data: Naturally Yours? (Nijhoff Law Specials). Brill Academic Publishers, 2006.
science_salvation.jpg Rebecca Dresser. When Science Offers Salvation: Patient Advocacy and Research Ethics. Oxford UP, 2001.
politics_of_life.jpg Nikols Rose. The Politics of Life Itself: Biomedicine, Power, and Subjectivity in the Twenty-First Century. Princeton University Press, 2006.

2005 Holiday Reading List

Compelled Disclosure, Richard Nixon and your genetic information

In addition to the three general modes of sharing genetic information I described yesterday, there is another type of disclosure individuals considering personal genome sequencing should be aware of: compelled disclosure by subpoena.

For genetic sequence data to be useful, it will need to be shared with individuals, companies, or organizations. These individuals might take extraordinary care to maintain the confidentiality of this information and to be good data stewards. But, they might receive a subpoena that requests confidential records to be made available for a court proceeding. For individuals who are concerned about the confidentiality of their genetic information, compelled disclosure by subpoena should be considered a risk.

The good news is that subpoenas of medical research data about individuals who were promised confidentiality is uncommon. There are a few famous cases, one involved lawyers hired by tobacco companies to go looking for data about asbestos raising the chances of lung cancer in smokers. A subpoena allowed them to gain access to the confidential files of a Dr. Irving Selikoff, although the files were redacted to protect identities of research participants prior to (compelled) disclosure (a few details here).

Don’t get Dicked Richard’ed
So, what does Richard Nixon have to do with compelled disclosure? Interesting trivia mainly. The Supreme Court case United States vs. Nixon is typically cited by the courts as the precedent for supporting compelled disclosure by subpoena.

Endote: Boo Hoo
Incidentally, there is a very nice article in the New England Journal of Medicine about scientific researchers who get subpoenaed. The article is not open access, unfortunately, so you probably can’t read it without paying $10.

It has been at least a few months since my last mention of frustration about the lack of access to quality information for Joe Schmoes, like myself.

I get frustrated most when articles deal with issues that are of direct relevance to the general public and they are not accessible, like the article about subpoenas above. Clearly, this article should be of interest to researchers. But what about research participants? It is, after all, their data that is potentially at risk. Research participants should not be treated like cows and keeping them outside the tent of good scholarly work on matters such as this does just that. I say this with the utmost sincerity, I’m not a dogmatic jerk in real life.

My favorite example in the last few years was an article published in Science with a title that included language about the “need for public dialogue“. If you click on the link, you’ll see that the article is trapped behind a firewall. For public dialogue to occur, the public needs access to the article. No?

One of the things I look forward to the most over the next 10 years or so, and one of the reasons I’m so happy to be a part of the Personal Genome Project with its open-access and open source ethos, is increased public access to science and the scientists who produce it. Support open access.

This month I am doing an extended meditation about the risks of obtaining and sharing personal genome sequences. This exercise will be cathartic for me. Beyond that, I hope it may help to reorient the conversation about risks into something that one day might resemble practical guidance for individuals considering obtaining and sharing personal genetic information.

3 modes of sharing genetic information

In order to understand the risks of obtaining and sharing personal genetic information, a few words are in order about the different types of sharing that are possible.

In general, there are three modes for sharing genetic information.

1) No disclosure

This might be referred to as the cupboard drawer strategy. A person gets sequenced, their data is shipped to them on a DVD, and they stick this DVD in the cupboard drawer. Data is kept private and not shared at all. Not very useful.

Conceivably, one day there might be software available for home computers, for those who are paranoid about sharing their genetic data but still seek knowledge that may be gained from genomic interpretation.

2) Controlled disclosure

Medicine, recreation, and the advancement of science are all reasons for sharing personal genetic information. Giving access permissions to specific individuals or groups of individuals, including organizations, is called controlled disclosure. Individuals with access permission will be responsible for being good data stewards.

Examples of data stewards might be:

3) Uncontrolled disclosure

This category of information describes the situation that personal genetic data is made publicly available, by publishing on a website for example. To be clear, publishing a personal genome sequence on a public website may be a conscious act, and is controlled in that way, but once the data is in a public forum where anyone has access, disclosure itself becomes more or less uncontrolled.

Uncontrolled disclosure also describes any situation where access is gained by individuals without explicit permission, through hacking or the carelessness of a data steward for example.


This month I am doing an extended meditation about the risks of obtaining and sharing personal genome sequences. This exercise will be cathartic for me. Beyond that, I hope it may help to reorient the conversation about risks into something that one day might resemble practical guidance for individuals considering obtaining and sharing personal genetic information.

Expectation setting about the risks of obtaining and sharing genetic information

Imagine being given the job of writing a manual that describes the risks associated with driving. All new drivers will be required to read this manual before they get behind the wheel.

The publisher wants a manual that is both comprehensive and concise. So, great care will need to be taken about the number of words dedicated to each risk. Spend too much time writing about flat tires, potentially more significant risks may get glossed over.

By the way, the year is 1908 and the Model-T is just beginning to roll off the assembly lines. Few people in the world actually own their own vehicles, but expectations are that many people will want to learn to drive even if its only recreational. They may even own vehicles themselves one day (heck, there might even be a system of interconnecting roads one day).

Model T


Foreseeing the potential mass appeal of vehicular transport, the publisher has requested that the manual take into account not only risks that exist today, but treatment should also be given to potential future risks that may arise.

The value of this manual will depend on how well the author is able to match the content with the driving situations readers are likely to face. Are they driving in the mountains of Coroico or in conditions similar to those I’m facing here in Boston, which look something like this:

dangerous roads
(Photo by Shuck)


The value of the manual will also depend on the drivers themselves: their experiences, their level of knowledge, their particular type of vehicle, and their expected behavior on the road. Are they risk-takers?

risky drivers
(photo by zerega)


Risks may even depend on an individual’s belief system:

amish drivers
(photo by cindy47452)


Setting reasonable expectations about risk
What is obvious to us now is that any manual produced in 1908 about the risks of driving very likely would be completely irrelevant today, except for those of us who find historical minutiae to be of great entertainment value.

Do you think this manual would have included “road rage” or the dangers of text messaging while driving? (BTW, in 1908, the Western Electric number 317 magneto wall phone did not even have an interface for dialing numbers, let alone text…calls required a voice-based interaction with an operator)

So, any manual developed today about the risks of obtaining and sharing a genome sequence will have a limited shelf-life. A few risks may come to be recognized more or less as “eternal truths” that go hand-in-hand with genome sequencing, while others are destined to become historical artifacts.

To get started with this meditation about risks, I’m going to state the obvious. Talking about the risks of obtaining and sharing genetic information is difficult to do in absolute terms for at least three reasons:

1) Risks are contextual
2) Risk-taking preferences vary widely
3) Many risks are poorly understood or are non-obvious

This month I am doing an extended meditation about the risks of obtaining and sharing personal genome sequences. This exercise will be cathartic for me. Beyond that, I hope it may help to reorient the conversation about risks into something that one day might resemble practical guidance for individuals considering obtaining and sharing personal genetic information.

Risks of obtaining and sharing your genome sequence

The era of personal genomics will transform our notions of risk in many ways. In the realm of health and medicine, we will soon have much more specific information about our baseline risk for a health outcome, such as a disease or a pharmaceutical drug response, and how it is influenced by our DNA and modulated by other factors like physiology, lifestyle, and environment. At some point, we will have health risk profiles that prioritize and sort risks according to severity, immediacy, and whether effective strategies are available for managing them with pharmaceuticals, lifestyle modifications, or other medical interventions. There might even be a richter scale for genomics one day. This vision of the future is often referred to as “personalized medicine”.

Genome sequences are an important source of fuel for powering the engine of translational medicine and driving the transformation of how we understand and manage health risks. But before anyone can step on the accelerator toward personalized medicine, people need to be willing to actually obtain and share personal genome sequences. Without this willingness, personalized medicine faces an energy crisis – a shortage of genomic fuel – and may putter or stall. Where will this willingness come from?

Roadmap to Willingness
Unlike personalized medicine, which has a coterie of visionaries and an abundance of roadmaps, each with their own devotees, no “roadmap to willingness” exists as far as I can tell. So, how do we build one? What are the ingredients of a roadmap?

The same three ingredients are needed to get nearly any enterprise off the ground, as I heard recently in a wonderful talk by Doug Solomon of IDEO (here). They are feasibility, desirability, and viability. And you really need all three, having just one or two doesn’t cut it.

To get to willingness, personal genomics needs to be cheap (feasible), useful (desirable), and safe (viable). As far as I’m concerned, two out of three ingredients are in the bag. There are tremendously productive forces working to make genomes both cheap and useful. While personal genomics may not be fantastically cheap or fantastically useful at this moment, we’re on a clear path to get there (and quickly).

Google Chart API Venn diagram of the three ingredients for making personal genomics a smashing success

The ingredient that could use some serious inspiration is safety. After all, what should be sought is not a “roadmap to willingness” that is just blinky lights and siren songs, but a “roadmap for informed willingness” that includes well-marked signs, off-ramps, and rest areas (for slow drivers with small bladders).

Where the rubber meets the road
Some influential policy-wonks are still spending their energy questioning whether or not we *should* as a society be pursuing personal genomics at all. Then there are those who have gotten past that and take for granted, as they should, that large numbers of people will desire and seek personal genome sequencing. Yet, I still have the overwhelming sense that we are very much stuck in a stage of facile hand-wringing about the implications of obtaining and sharing personal genome sequences and that these activities tend to be more oriented toward sending people into some sort of existential apoplexy than they are a potential source of pragmatic guidance.

We are at the point in history where the rubber is about to meet the road for personal genome sequencing. So, I would like to advocate that we spend more time rounding out this roadmap. We should begin with a meditation about risk and consider the calculus an individual would need in order to weigh the tradeoffs between taking the personal genomics plunge right now or standing on the sidelines for another day.

The month of December I plan to focus my blogging efforts, nearing one blog post per day (assuming I’m not paralyzed by some existential crisis, fingers crossed), on my thoughts about the risks of obtaining and sharing personal genetic information. I hope you’ll join me.

P.S. The Venn diagram above was created on the fly via the recently released Google Chart API — an incredibly easy-to-use way to add graphic bling to any blog. Here is the coded URL, in case you want to play with the Venn Diagram on your own.

Geek Doctor

Pssst. Another member of the PGP-10, John Halamka, has started blogging over at Geek Doctor: Life as a Healthcare CIO. So far most of the discussion is around issues related to being responsible for the IT needs of thousands of doctors and millions of patients (and gazillions of medical records). There has not been any mention yet of personal genomes. But, his recent release of a music recording might be construed (by me, with no apologies) as a quiet nod to emerging genre of “genome pop“. Very zen, as is John. This past summer I sent him an email and received an out-of-office reply that was a Basho poem. For a split second, there was serenity in the world. :)

The X-team

The scientists and engineers in the Church lab at Harvard Medical School, who are busy developing the sequencing technology that will be used for the Personal Genome Project, will be competing in the Archon X-Prize for Genomics. The announcement was made this morning in the Boston Globe.

To win the $10M prize all we need to do is to line-up about 200 of our machines and turn them on. Well maybe its not that simple. There are also a few items on the science and engineering “To Do” list, like figure out how to drop the price of whole genome sequencing by 1 or 2 orders of magnitude.

The founding members of the Personal Genome X-team are:

Richard Terry, Greg Porreca, Jay Shendure, and Kevin McCarthy. Go team!

P.S. I wonder what George plans to do with all of those machines after competing in the X-Prize?