NIH Extramural Nexus

The National Postdoctoral Association defines  a postdoctoral scholar (or a postdoc) as “an individual holding a doctoral degree who is engaged in a temporary period of mentored research and/or scholarly training for the purpose of acquiring the professional skills needed to pursue a career path of his or her choosing.” 

As the data show, postdocs are more prevalent in most of the top fields receiving NIH funding (genetics, biochemistry, developmental biology, and neuroscience) than in those fields that receive less NIH funding (nursing, public health, and pharmaceutical science). In light of this, the experiences and future paths of postdocs obviously are an essential part of any study of the biomedical research workforce. 

As we started delving more deeply into the data, however, it became clear that we lack reliable information about the postdoc population in the US. There are many reasons for this. First and foremost, we do not collect much information about foreign-trained PhDs who come to the US to do a postdoc, and we have no idea how long they stay or how many leave after their training. These foreign-trained postdocs comprise about 2/3 of the total postdoc population. In addition, postdocs have many titles, and some institutions require they change their titles after a certain number of years. That is why the PhD snapshot I presented last week includes a range of numbers, and they are colored red, meaning that we have little confidence in their accuracy. 

Again, we had heard anecdotal information suggesting that the postdoc training period has lengthened over time. However, data from the NSF Survey of Doctorate Recipients suggest that most US-trained biomedical PhDs spend fewer than 5 years in postdoctoral positions. Some do remain in postdoc training a lot longer, though. There is some indication those who do the longest postdocs are the ones who go on to tenure-track academic research careers. For example, in the figure below, the age at first non-postdoctoral job (many of which are in industry) has been consistently a year or two lower than the age of obtaining the first tenure-track job. Note that the latest data in this graph (2002-2003) may be underreported due to delays in reporting that result in a lag time bias.

With all that said, what can we glean from the data we have? 

First, data in the figure below from the NSF Survey of Graduate Students and Postdoctorates (which includes all sources of support, not just NIH and surveys US degree-granting institutions about their US- and foreign-trained PhDs) show that the vast majority of basic biomedical postdocs are supported on federal research grants, and this number has grown considerably over the past decade. This is perhaps not surprising, as it parallels the growth of basic biomedical graduate students supported on research grants that I showed in the previous post. Similar to the data for graduate students, the numbers of postdocs supported on federal fellowships and traineeships have remained remarkably stable over the same time period. Once again, these data are supported by the NIH-specific data posted on the RePORT website.

As shown below, the other source of postdoc support that has been growing over the last five years is nonfederal support, defined as support from state and local government, institutions, foreign sources, foundations, industry and other private sources. 

Combining the average ~6.5 years of PhD training and 4-5 years of postdoctoral research means that it takes approximately 10 years before a person with a biomedical PhD is ready to begin his or her first or post-training job, and even longer if he or she chooses the academic tenure-track research path. 

So what does this postdoc have to look forward to? 

We looked at earnings potential as one (but by no means the only) attribute of the career path of biomedical PhDs and compared it to other scientific fields and professions. As can be seen in the table below, starting salaries of biomedical PhDs (pooled SDR data in 2008 dollars) are lower than in other fields. However, later in the career stage, 30 years after the PhD, this is no longer the case.   

Table 1. Salary Across Broad Fields by Years of Experience

Source: NSF Survey of Doctoral Recipients

A more comprehensive timeline of earnings is shown in the report and on the website, and Paula Stephan, who was on the modeling subcommittee of our working group, has a very interesting discussion of this in her recent book How Economics Shapes Science. 

The data I’ve presented in this and my earlier posts, and the information included in the working group report should be of interest to anyone considering a career in the biomedical sciences and those of us responsible for ensuring the availability of a well-trained biomedical research workforce in the future. They are important for making informed decisions about graduate training, sources of federal support, and institutional policies that will attract and retain the best and brightest in biomedical science careers.

Now that I’ve discussed the consolidated snapshot of the PhD biomedical research workforce, I’d like to focus on some trend data about graduate students gathered by the biomedical workforce working group. 

Before I get to the data, however, I want to provide some details about these analyses. As you will see in the report and on the accompanying website, many of our analyses separated out basic biomedical science, clinical sciences, and behavioral and social sciences (see what these fields include). There are a large number of PhDs in behavioral and social sciences research but only a portion conducts NIH-funded research. Therefore, the framework summary in the previous post included only basic biomedical and clinical sciences. That said, because the behavioral and social sciences research we fund is an essential part of NIH’s mission, we included these fields in as many of the trend analyses as we could. In addition, to put the data in context, wherever possible we compared biomedical research to chemistry, which seems to be more constant in numbers of students and faculty. 

Let’s take a look at some data on graduate and medical students. As you can see in the figure below, the number of PhDs granted in biomedical science in the US has risen substantially over the past decade. In contrast, the number of PhDs awarded in behavioral and social sciences and in chemistry has been stable over the same period, as has the number of MDs. The steep increase in the number of biomedical PhDs awarded began in 2004, reflecting an increase in graduate enrollments that began with the doubling of the NIH budget from 1999 to 2004 (see the second figure). Given an average training period of 6 years or so, this underlines the close relationship between the size of the NIH budget and the number of biomedical PhD slots. 

Source: NSF Survey of Earned Doctorates and AAMC Data Book

How are these graduate students supported? 

In the following figure, data from the NSF Survey of Graduate Students and Postdoctorates (which includes all sources of support, not just NIH) show that the vast majority of basic biomedical PhDs are supported as research assistants, and this number has grown considerably over the past decade, again in parallel with the doubling of the NIH budget. The number of students supported on fellowships and training grants has remained stable over that time, as also is evident from the NIH-specific data posted on the RePORT website.

Source: NSF Graduate Student Survey

How long does it take to get a biomedical PhD? 

Among the anecdotes we heard when we started this study was that training for a career in biomedical research is getting longer every year. When we looked at the data shown below, however, we saw that the average time to a basic biomedical PhD has remained relatively stable since 1985—between 6 and 7 years. In recent years the training period has been shorter. Both clinical and behavioral and social sciences degrees take somewhat longer, but these too have decreased over the past decade. This is reflected in the average age at degree, which is about 30 or so years of age.

These data do beg the question of how people get their PhD at 30 or 31 when the degree only takes 6 or so years on average. There seem to be couple of years missing. We did not find data to address this question specifically, but anecdotes from working group members and others suggest that more and more people are working or doing postbaccalaureate research before going to graduate school.

Even though the time to degree of biomedical researchers has not increased significantly, it is important to remember that most biomedical research fields also require a period of postdoctoral training, so the total period before people embark on their career is considerably longer. We’ll discuss postdoctoral researchers in my next post.

Update 6/27/12: The full report is now posted on the ACD website.

As I blogged last week, and most of you have heard by now, a working group of the Advisory Committee to the NIH Director (ACD) that I co-chaired with Shirley Tilghman from Princeton just completed a study of the biomedical research workforce. We reported our findings to the ACD last Thursday (you can find a link to the videocast here). 

We gathered a lot of data during this study, which are included in the report (see the ACD site for the executive summary and instructions for obtaining a copy of the full report). The data also are posted on an accompanying website. I plan to highlight some of the specific data in future posts, but first, I’d like to discuss the outcome—the conceptual framework that presents a snapshot of the biomedical research workforce, incorporating the latest available data. The framework of the PhD workforce is presented below, and a companion framework for MDs and MD/PhDs in the biomedical research workforce can be seen in the report and on the website.

First, 9,000 biomedical PhDs graduated in the US in 2009 (including basic biomedical and clinical sciences), and 70% of these went on to do postdoctoral research. As we conducted our analysis, it became clear that there are few reliable data on the number of biomedical postdoctoral researchers in the US. We lack solid information on foreign-trained postdoctoral researchers, and many postdoctoral researchers change their title as they proceed through their training, complicating the data collection. That’s why the estimate of postdoctoral researchers ranges from 37,000 to 68,000. 

Looking at the career paths taken by these US-trained biomedical PhDs, we can see that fewer than half end up in academia, either in research or in teaching, and only 23% of the total are in tenured or tenure-track positions. Many other people are conducting research, however, with 18% in industry and 6% in government. 

The science related non-research box includes individuals working in industry, government, or other settings who do not conduct research but are part of the scientific enterprise. Many of the career paths represented by this box contribute to the scientific research enterprise and require graduate training in biomedical science. For example, program and review officers at NIH and managers in many biotechnology companies would be included in this group. This is my box too. It’s interesting to note the 18% included in this group is made up of PhDs employed in industry (13% of the total workforce), in government (2.5%), and in other settings (2.5%). This means that all individuals working in industry (research plus non-research occupations) represent about 30% of the workforce, and all those working in government represent about 9% (more than 10,000 individuals).   

That leaves 13% in non-science related occupations and 2% unemployed (this does not include retirees or those who choose not to work). These are 2008 data, the latest available from the NSF Survey of Doctoral Recipients.

If you’re a graduate student or postdoc looking at these numbers, particularly the proportion of people in industry and government settings, it makes sense to learn as much about these career paths as possible. I’m very proud that we were able to develop this framework, as it seems that for the first time we have an idea of where domestically trained biomedical researchers are going. I was quite surprised by the idea that the majority of our trainees do not end up in academia. Did this surprise you?

Notes on the figure

The main sources of the original data, from which the graphs in the report were made and these numbers were derived, come from three NSF surveys: the Survey of Graduate Students and Postdoctorates, the Survey of Earned Doctorates, and the Survey of Doctorate Recipients. You can see the specific sources of each number by clicking on the relevant box on the website.

The color of the numbers reflects our confidence in the accuracy of the data: high (green), medium (yellow), or low (red). For more details see colors. In this case, the red numbers in the post-training workforce box are accurate, but the color reflects the fact that we know almost nothing about the distribution of foreign-trained PhDs in the workforce, so the overall picture is an under-estimate.

The post-training workforce boxes are color coded, with light blue denoting those in research positions and academic teaching positions. The science related non-research box is colored dark blue to indicate that many of the careers represented in this box are closely related to the conduct of biomedical research.

It has been an exciting time at the NIH over the past two days. The Advisory Committee to the NIH Director (ACD) met for its regular biannual meeting on campus, and the agenda was jam-packed. Three separate working groups tasked with assessing important issues—the future of the biomedical research workforce, diversity in the workforce, and data and informatics—concluded their work and presented their recommendations to the ACD.   

First, on Thursday afternoon, Shirley Tilghman presented the conclusions of the work done by the working group she and I chaired that evaluated the future of the biomedical research workforce (read previous posts here and here). The presentation included a snapshot of the current biomedical research workforce, as well as recommendations to the ACD about funding and training of graduate students and postdoctoral researchers, and about how to improve and maintain robust data collection on the biomedical research workforce to provide accurate information to those in the field and those thinking about joining it.

Immediately after that, Reed Tuckson presented the results of the working group he chaired with Larry Tabak and John Ruffin that was tasked with making recommendations to improve the recruitment and retention of underrepresented racial and ethnic groups, people with disabilities, and people from disadvantaged backgrounds in biomedical research careers. The recommendations are focused on data collection and evaluation; mentoring, and career preparation and retention; institutional support; bias-related research and intervention testing within the context of peer review; and a NIH diversity strategy and infrastructure.  

Last, but definitely not least, on Friday morning Dave DeMets presented the findings of the working group he and Larry Tabak chaired that was asked to provide expert advice on the management, integration, and analysis of large biomedical research datasets. The recommendations of this group covered both the extramural and the intramural communities and touched on data sharing, informatics methods and applications, training in the quantitative disciplines, and an NIH-wide IT strategic plan.  

You can read all the recommendations and obtain copies of the reports on the ACD website. We also posted additional data on the biomedical research workforce on the RePORT website.

At this point in time, all of these recommendations are under consideration by Dr. Collins. He has a lot to consider, and I will update you on developments as they happen. In the meantime, I will be dissecting and discussing some of the data behind these reports in future blogs.

Earlier this year, I announced the reauthorization of our small business program and the immediate changes that were made for fiscal year 2012. We now have more information from the Small Business Administration, as they recently released a proposed rule that will make some substantial changes in the program. The major change centers on eligibility and the participation of small businesses majority-owned by venture capital operating companies, hedge funds, or private equity firms, which would now be eligible to apply under certain situations. If this change affects you and your research, now is the time to let your voice be heard. The SBA is accepting comments on the proposed rule through July 16, 2012. In addition, don’t forget to look out for draft SBIR and STTR policy directives, which the SBA will release for public comment soon.

I wanted to let you know about a request for information we just issued to collect opinions on a proposal to modify the NIH biographical sketch that is used as part of your grant application. We are concerned that the biosketch, as currently configured, may not allow applicants to adequately describe the nature, significance, or impact of their scientific accomplishments and capabilities in the most effective manner to illuminate a person’s contribution to their field. 

The biosketch portion of the NIH grant application is limited to 4 pages, and applicants are encouraged to list no more than 15 publications. The proposed modification would allow investigators to include a short explanation of their most important scientific contributions. We haven’t decided on a format, but scientific accomplishments and contributions could be described in a short narrative section, annotated with references. References could be peer-reviewed publications or other types of scientific output, for example, data sets, videos, crystal coordinates, patents, licenses, or changes to standard medical practice. Alternatively, the biosketch could permit a short descriptive paragraph associated with an individual’s most significant scientific publications documenting the advances and subsequent scientific impact. The Howard Hughes Medical Institute  uses both of these approaches successfully.  

I hope that you will take a few minutes and let us know how you feel about this potential change. Please submit your responses here and not to my blog so they are included in the analysis.