When molecular biologist and Princeton University President Shirley M. Tilghman first sounded the alarm about the need for major overhauls to the way the United States trains its biomedical workforce in the 1998 National Academies of Science report Trends in the Early Careers of Life Scientists, many of her proposals fell on deaf ears. Fourteen years later, Tilghman is arguing again for training reform, this time as chair of the National Institutes of Health (NIH) Biomedical Research Workforce Working Group.
Last week, Tilghman presented a draft of her group's latest report to NIH's Advisory Committee to the Director (ACD) at NIH headquarters in Bethesda, Maryland. In the report, the group calls on NIH to divert funding from research grants to training grants for graduate students, support more postdocs on training grants, increase pay and improve benefits for postdocs, and boost the prestige and remuneration of staff scientist positions in academic labs.
At the presentation, Tilghman and the other members of the working group argued that in its present state, the graduate training system at our nation's universities and the workforce that graduates enter into are dysfunctional and unsustainable. At the root of that dysfunction, Tilghman said, is a mismatch between the training most graduate students receive and the careers most Ph.D. graduates end up in.
The number of academic jobs has shrunk dramatically compared to the number of new graduates. NIH estimates that 26% of biomedical Ph.D. recipients end up in tenure-track academic positions, down from 34% in 1993; meanwhile, the proportion of nontenure-track academic positions has remained constant. The growth in jobs for Ph.D. biomedical scientists, the working group concluded, is outside academia, so new graduates must be prepared to work in other roles: in industry, in government, or in positions tangentially related to their degrees, such as science writing or policy, Tilghman said.
How can universities prepare graduate students better for the careers they're most likely to wind up in? One way, Tilghman said, would be for NIH to shift funding from R01 research grants, which currently support the majority of graduate students in biomedical sciences, to NIH training grants, which are peer-reviewed by NIH for their training-related virtues. The total number of graduate students supported by NIH, the report says, should remain constant.
While the number of graduate students supported by research grants has been higher than the number supported on training grants since the early 1980s, the gap steadily widened as NIH's research budget grew—then shot up in the early 2000s when NIH's budget doubled over 5 years (see graph below).
The report's authors argue that many graduate students are ill-served by this approach because it limits the ability of NIH to hold principal investigators (PIs) accountable in their roles as mentors. Without oversight, Tilghman argued, it's easy for PIs to see and treat their graduate students as laborers rather than scientists in training. If a larger proportion of the graduate student population were supported on training grants, she said, NIH could better monitor students' training and ensure broader exposure to careers outside of academia—and better training in the skills needed to perform well in those careers.
The members of the working group "are, I think, unanimously of the view that the most effective training dollars that the NIH has to expend are those in their training grants,” Tilghman said. “Training grants are immensely effective at inducing good behavior on the part of graduate programs. … It is the only mechanism we have to really peer review the quality of graduate training.”
Some members of the ACD weren't buying it. Biologist Robert Horvitz of the Massachusetts Institute of Technology in Cambridge argued that shifting funding away from R01s takes away too much autonomy from PIs. "One wants to be sure that the principal investigators, who are supposed to be doing the research, continue to have enough flexibility to be able to support the research they want to do," he said. Taking away that flexibility, he argued, could reduce research productivity.
Other ACD members, including Haile Debas, director of the University of California Global Health Institute in San Francisco, were more supportive of the recommendations. While such a shift would be bold, Debas said, "you can also do harm by doing nothing." He proposed that NIH launch experiments to determine whether graduate students who get industry experience during their traineeships, for example, go on to have successful careers in industry.
Judith Bond, incoming president of the Federation of American Societies for Experimental Biology (FASEB) and a biochemist at Pennsylvania State University, Hershey, also disagrees with this recommendation, saying in an interview with Science Careers that "oversight of student training should be left to the universities, not the federal government." Bond is not a member of the ACD.
The situation is equally grim, if not grimmer, for postdoctoral researchers, Tilghman and her colleagues argue in the report. The report recommends that more postdocs be supported by training grants and fewer by PIs' research grants, with the total number of NIH-supported postdocs remaining constant or perhaps decreasing.
One way of reducing the number of postdocs—and decreasing the intense competition for jobs—would be to increase postdoc salaries from $39,264 to $42,000 and provide benefits equal to those of employees at their institutions, the report says. It also recommends that NIH mandate a 4% raise before the third year of postdoctoral work and a 6% raise before the seventh. The idea, Tilghman said, would be to motivate PIs to help their postdocs move as quickly as possible into jobs rather than toil away as a postdoc.
“One of the things the committee really grappled with is: To what degree are these [people] trainees … and to what extent are they worker bees who are the producers of the research in our lab?” Tilghman said. The working group felt strongly, she said, that emphasizing training is the best way to produce well-trained future PIs.
Cato Laurencin, an ACD member and CEO of the Connecticut Institute for Clinical and Translational Science in Farmington, agreed with the working group's postdoc recommendations. “We've gotten into a mindset where postdocs last 5, 6, 7 years,” he said. “After 5 or 6 years of Ph.D. training, people are spending their careers in training. I am very concerned about that.”
Bond, too, agreed with the postdoc salary recommendation. “In general, FASEB is in favor of increasing postdoc salaries. … Postdocs are essential to work in the lab, and they should be paid a living wage,” she said.
But ACD member Horvitz was skeptical. The money to raise postdoc salaries "has to come from somewhere,” he said, and given NIH's current budget woes, it might be impractical to raise postdoc pay. If PIs were forced to make do with fewer (but better paid) postdocs, he argued, lab productivity would probably decline.
One way to provide more job opportunities for Ph.D. scientists would be to increase the number and stature of staff scientists in university labs. (See "A Hidden Academic Workforce." ) One way this could be accomplished would be to have universities shoulder a larger percentage of researcher salaries than most currently do, Tilghman said. That would make the positions more stable and less vulnerable to changes in NIH budgets and competitive grant renewals.
Those salaries should also be increased, the report argues, to be commensurate with the training levels of staff scientists and their value to the lab. If the number of postdocs drops as a result of raising postdoc salaries, staff scientists could fill the gap, which should help attract talented scientists to these positions.
Finally, Tilghman recommended that NIH award grants preferentially to PIs who employ staff scientists. “When I think about the tradeoff of a graduate student for a staff scientist who is already extremely well trained, who can work without constant supervision, who can really help train the younger people in the laboratory, … I actually think we'll be more productive,” she said.
Two of the key recommendations of the report—shifting funding away from R01s to create more training grants and increasing postdoc pay and benefits—met with resistance from members of the ACD. Yet Tilghman believes that these recommendations will gain more traction with NIH leadership than when she proposed similar reforms in 1998. Times are much tougher now, she said, which makes it easier to make larger changes. "The only time when it's going to be possible to make hard decisions that would … have a long-term, beneficial effect on all the players in the biomedical workforce is … during tough times,” Tilghman said. “Doing nothing, in my view, is not an option."
NIH Director Francis Collins said that the ACD appeared generally supportive of the report and that NIH would collect more data, build models, and run pilot programs so that they can better predict the impact of implementing the report's recommendations. “I do think the NIH will want to take some action here,” he said. “I like the idea of doing some experiments to get some early indications of whether the interventions are achieving the goals that we hope for. It would be a very good thing before we do something more systematically disruptive in ways that we didn't intend.”