Tuesday, May 27, 2014

Can the US Biomedical Research Enterprise Be Sustained? Implications for Informatics

Although I focus a good deal of my writing in this blog on the educational and career aspects of informatics, I believe that the research mission of academic informatics is equally important. Education and research are synergistic in successful academic departments, even in those where a majority of students pursue professional careers.

I have described some of the important issues in informatics research in various postings. For example, there is a need to solve the issue of data entry in order for informatics systems to provide usable data that can feed a host of functions. I have also noted that while data science is important to informatics, we cannot ignore the workflows that surround the best capture and use of data.

Unfortunately there are some challenges to the academic biomedical research enterprise in the US, not only in informatics, but in all fields. These were laid out nicely recently by Alberts et al. in a recent paper in Proceedings of the National Academy of Sciences [1]. The authors note there are "deep flaws" in US biomedical research, stemming mostly from the consequences of the system, operating under a perception of sustained growth in funding, whereas in realty federal research support has slightly declined in real dollars over the last few years and is unlikely to increase substantially in the years ahead.

The current situation has, in essence, reached a "Malthuisan" situation, due to the slowing growth outstripping resources. This leads to a number of consequences. The first of these is a hyper-competitive funding environment, where the success rate of grants being funded has fallen from 30-40% to the low teens. This results in scientists having to spend more time writing proposals, not to mention worthwhile and important science failing to get funded. It also leads scientists to be more conservative in the work they propose, sticking to tried and true, incremental science rather than bold but riskier innovation.

Another consequence of the current situation is increasing demands on scientist time. Because proposals are competitive, and increasing numbers of scientists are competing for funds, it means that more time must be spent in peer review of proposals. The hypercompetitive environment also requires that funders and others be more vigilant that funds are being spent properly, resulting in more regulations, paperwork, and so forth.

A final consequence of the current situation is an over-supply of trainees. In many biomedical research labs, graduate students do the lions share of the work. This has always been done under the assumption that this would train the next generation of scientists. The problem is that the hyper-competitive environment means it will be difficult for these students to launch successful careers as they start to compete with their mentors for a fixed or possibly shrinking amount of funding. In fact, as the article notes, only 20% of all PhD students in biomedical science in the US will get faculty positions or achieve independent research funding, as the system just cannot accommodate so many new researchers seeking funding.

The article notes that the enterprise needs to be brought into "sustainable equilibrium." They advocate new approaches for evaluating scientific proposals that are more efficient and reward the best and most innovative science, although it will always be a challenge to determine how these exalted few will be selected. The authors also call for less work of grants being done by graduate students and more by "staff scientist" types of positions by well-trained researchers who are not PIs of research. They also note that graduate programs need to prepare students for different career paths beyond being grant-funded academic faculty. One of those paths could be staff scientists while others might involve working in industry.

How does this problem play out in informatics and for the OHSU informatics program in particular? There is certainly no dearth of funding for research involving informatics, although much of it is applied and specific to projects. One downside to this is that there is little funding of core research in informatics that might lead to novel techniques. A related problem is that a good portion of  funding goes to very short-term ends, which is good for demonstrating progress to Congress and others who authorize funding but less so for advancing the field in the long run. As for informatics research locally at OHSU, we are doing relatively well but maintaining funding is always a struggle. One outgrowth that has achieved some success is looking for opportunity beyond government funding, as exemplified in our Informatics Discovery Lab that is developing partnerships with industry and others. We also have a track record of our advanced trainees (PhDs and postdocs) obtaining diverse employment for their skills.

One goal I have for the informatics field is raising awareness of this challenge, and how we might collectively work to advance the case for a more comprehensive approach to the important research that we undertake. Informatics is an interdisciplinary science but also requires attention to advancing its core scientific methods and results.


1. Alberts, B, Kirschner, MW, et al. (2014). Rescuing US biomedical research from its systemic flaws. Proceedings of the National Academy of Sciences. 111: 5773-5777.

Wednesday, May 14, 2014

Square Pegs into Round Holes - Challenges for the Clinical Fellowship Model for Clinical Informatics Subspecialty Training

Although the development of the clinical informatics subspecialty is an important accomplishment for the informatics field, I have a number of continued concerns about the optimal development of the subspecialty, especially now that the Accreditation Council for Graduate Medical Education (ACGME) program requirements for clinical informatics fellowships have been released. There are a number of aspects of training to be an informatician that just do not fit into the traditional model of clinical training, requiring those of us applying for program accreditation to fit proverbial square pegs into round holes.

Some of these concerns are also noted by Don Detmer and Ted Shortliffe in a new Viewpoint published in JAMA [1]. They raise concerns about:
  • The accreditation process that may fragment as a result of it being administered by nine different primary specialties
  • Program funding that will compete within healthcare institutions for other fellowship programs
  • National capacity for training subspecialists once the "grandfathering" period ends in 2018 and the only path to board certification is a full-time two-year fellowship
I have shared these and related concerns last year and the year before, and have also raised others. In the following sections, I will describe my major concerns, grouped under headings of the challenges with the rotations model, a more detailed description of the funding issues, and timing and flexibility issues.


Many of these are related to traditional model for clinical training versus the graduate educational model that the field has historically employed. For example, clinical training historically is based on the notion of "rotations," which are blocks of weeks that trainees spend in a given setting. This makes sense in clinical training, where one learns by being exposed to a steady stream of new and returning patients, whether on a hospital ward or in a clinic. An internal medicine trainee, for example, can learn quite a bit about intensive care medicine by taking care of the steady number of patients who are admitted, treated, and then discharged from the intensive care unit over weeks or months of time. This is the traditional "steeped tea" model of clinical training, where the assumption is that steeping the learner among patients in a given specialty will result in learning, a model that has been challenged in recent years [2]. (And for which informatics tools may help, in better tracking the cases seen.)

The rotations approach makes even less sense in clinical informatics. Trainees, for example, will derive value from spending time in the clinical informatics department, the IT organization, the compliance (privacy and security) departments, and so forth. As we hope to create a multi-instutional fellowship program, fellows in our program will also gain exposure to informatics in other settings (e.g., managed care organizations, the Veteran's Administration Hospital, safety net clinics, and even industry settings). It is not clear, at least to me, that prescribing fixed numbers of weeks in specific settings makes any sense.

A related issue with rotations is that most informatics/IT projects evolve over months, if not years. Just as primary care physician trainees need continuity experiences, so do informatics trainees. Because of the long-term nature of informatics projects, these trainees should be, in by opinion, spending significant amounts of time in them. But with all the rotations, clinical work, courses, and so forth, there will be only intermittent hours in the week for any sort of longitudinal informatics project experience.


I applaud Detmer and Shortliffe for raising concerns about funding, although the brevity of their paper does not allow detailed exploration of the complex challenges. Many institutions fund clinical fellowship programs with the understanding that the cost will at least partially be offset by the clinical work of the fellows. Clinical informatics fellows will certainly be able to make tangible contributions to the organizations that sponsor fellowships. However, they will have competing demands for their time, such as clinical work, rotations, education, teaching, and so forth, that will reduce the value they can provide to their sponsoring institutions.

An additional issue is that some, perhaps many, positions in these training programs will be funded by specific medical specialties within healthcare organizations. Those funding the programs are likely to want to see some tangible contributions to their informatics/IT efforts, just as clinical trainees contribute to the clinical mission of organizations (i.e., internists see patients in the Internal Medicine Clinic). But again, with the myriad of clinical work, rotations, education, teaching, and other demands for their time, the amount that fellows will be able to contribute to informatics work in their specialties may be limited.

Also a funding issue is that most institutions are interpreting Centers for Medicare and Medicaid Services (CMS) rules to prohibit these board-certified or board-eligible physicians, practicing in their primary specialties, to bill for patient care. Thus one source of financial value that trainees could provide to their organizations cannot be financially realized. Some have suggested that fellows do their clinical work in other settings, i.e., moonlighting, but to me, not being part of the informatics team at the same place they are providing patient care is a real lost opportunity.

A final funding issue concerns ACGME requirements for faculty, staff, and program leadership time. In these times of ever-tightening budgets for academic medical centers, the requirement for 2.0 FTE of faculty time for programs that may only have a handful of fellows is not realistic. I know this issue has historically been "fudged" by clinical programs where faculty "supervise" trainees in clinical settings while simultaneously providing patient care, but this will be challenging for clinical informatics fellowship programs.

Timing and Flexibility

I have other concerns beyond funding that I have raised in the past about timing and flexibility. The two-year, full-time commitment will block the post-2018 pathway for many physicians who have established jobs, practices, and families. Our distance learning program has more than a decade of experience of allowing physicians (and others) to transition into informatics training and careers at their own pace. One of the other subspecialty fellowship program directors at OHSU doubted that many physicians in training would want to complete a clinical subspecialty (such as oncology) and then spend another two years pursuing informatics. He did believe that there would be great interest in a joint fellowship where a trainee could get "credit" for more than one subspecialty by, for example, embedding clinical informatics training in the 18-24 months that subspecialty physicians often get for more flexible portions ("research") of their fellowship.

Clearly there are some people who should train solely in informatics, at full-time and for extended periods. These should be the researchers and educators who will work in academia, industry, and leadership roles. But clearly many "practicing" informaticians will need to maintain strong ties to their clinical specialties as well as be efficient in their training to become a subspecialist. The ability to overlay or combine informatics training with training in other specialties is appealing to my colleague program director mentioned above (not to mention myself).


In the long run, these issues will need to be resolved, especially if the new subspecialty is to thrive. Our approach at OHSU is to move forward, get our fellowship established and accredited, and then hope that ACGME and others will have the flexibility to allow clinical informatics fellowships and the larger field thrive.


1. Detmer, DE and Shortliffe, EH (2014). Clinical informatics: prospects for a new medical subspecialty. Journal of the American Medical Association. Epub ahead of print.
2. Hodges, BD (2010). A tea-steeping or i-Doc model for medical education? Academic Medicine. 85: S34-S44.

Friday, May 9, 2014

Accolades for the Informatics Professor - Spring 2014 Edition

I periodically toot my own horn in this blog, and enough interesting things have accumulated for me to do so now.

One accolade comes not from myself, but the data! My information retrieval (IR) colleague Jimmy Lin has developed a new tool, scholar-scraper, which builds a list of citation metrics of researchers in a given discipline (or from any list). One of his original lists is from the IR field, and some other colleagues, Allison McCoy and Dean Sittig, have created a list for biomedical informatics. The accolade for me is that I rank well both in biomedical informatics (15th as of today) as well as IR (18th as of today). Of course these rankings may change as the data changes as well as new scholars are added to each list.

I also have been quoted in articles on various topics, including an article on the new clinical informatics subspecialty that seems to have appeared in a number of clinical news publications:
(And probably others! To access the complete articles, free registration is required to get beyond the first page, although there is one version that seems to reliably appear without registration.)

I was also quoted in an article about the resignation of former Health and Human Services Secretary Kathleen Sebelius.