Research Plan

1.I often defer to what already has been written by others about crafting the best possible research plan since the advice is timeless. In fact, over 25 years ago, the surgeon, John Niederhuber, wrote (long before he became Director of the National Cancer Institute) that “Anything that can be done to make the reviewer’s job easier can only help the applicant. Brevity, clarity and organization are essential to the application.”7

2.Similarly, over 15 years ago, Charles Lowenstein, M.D. (currently Professor of Medicine at University of Rochester Medical Center), emphasized to his graduate students at Johns Hopkins University the following:

Focus, focus, focus your Specific Aims and then organize all sections of the research plan around the aims. The most important aspect of the research grant is that it be focused on a single issue and not cover multiple issues. Connect the background, the preliminary data, and the experimental design back to the Specific Aims focus.

3.Lowenstein continues about the research plan: note that the experimental design is the most important section and must tell a narrative, namely how the work will move forward from the preliminary data with anticipated results and alternative approaches if problems occur. A generalized approach would be “I discovered X, now I will do Y. If the result is A, then I will do B; but if the result is C, then instead I will do D. My controls will be E, F, and of course, G. If problems occur, I will do Z.” Although it is a story, be mechanistic and focused, not descriptive.

4.I often combine the Niederhuber and Lowenstein approaches by telling my surgical residents and postdocs to keep their narrative simplistic and avoid becoming technically abstruse. A useful strategy, before tackling especially difficult sections, is to explain your research plan verbally to a 10 year old – “I am trying to attack the cancer cells, but if this doesn’t work, I have a back-up plan.”

194 M.V. Brock

5.Although the final text should be independently written, the experimental design and the main narrative of the research plan should be conceived soups to nuts with the mentor. This means allocating the necessary time for him/ her to read this section and then the time to comment extensively on the working drafts before submission. Beware that research plans not reviewed by mentors are quite obvious to reviewers, reflecting poorly on both the applicant as well as the mentor/co-mentor team. A knowledgeable mentor, for example, is often very skilled at including pertinent, critical references. Experts on some study section will immediately recognize when important references have been overlooked (often their own), and judge accordingly.

6.As Lowenstein implies, the Specific Aims abstract and the research plan, although physically separated in the grant, should be thought of as being one continuous entity. Lowenstein often stresses that a reviewer should be able to read the abstract or Specific Aims page and have an important, real appreciation of the nuts and bolts of the research plan. In the abstract, organize each Specific Aim, and summarize the background as well as the preliminary data succinctly so that all reviewers can read the grant’s Specific Aims as a single page. The following structure is an example: Hypothesis-1 sentence, brief background summary of the Specific Aim, truncated preliminary data summary, likely experimental anticipated results, interpretation of results, and future directions.

7.Remember that this is a Career Development Award and not a R01 award. Don’t be too ambitious when listing Specific Aims, at most, have three or four. Don’t fall for the mistake of listing three Specific Aims and then subdividing them into four or five subaims. Be careful not to list three time-consuming Specific Aims either, such as three randomized multicenter clinical trials, all to be completed over the course of the grant. Moreover, the modest resources of a K Award alone may not be enough to fund the salaries of the numerous support staff and research

nurses needed to carry out even one large clinical trial.

Remember what I have iterated before, that a K Award is meant not only to protect your time for engaging in research but also is equally, or even more importantly, designed to give you the opportunity to spend time pursuing scholarship that would enable you to do independent research.

8.Although you must be focused, as Lowenstein stresses, avoid having “a one hypothesis grant” with all of the Specific Aims hinging on the success of the very first Specific Aim. Many reviewers see it as a fatal flaw if each Specific Aim is unable to stand alone on its own merit. For example, if Specific Aim 2 is totally reliant on the successful completion of Specific Aim 1, and then Specific Aim 1 fails, Specific Aim 2 as well as the full execution of the grant is jeopardized.

9.Choosing an appropriate research design with your mentor’s assistance is critical, and be sure to justify why a particular design was employed. For example, why was a modeling approach chosen instead of a randomized clinical trial-based design. What are the biases inherent in one vs. the other, and how will this impact your anticipated results?

10.When designing the research plan, take advantage of the infrastructure of your institution. Reviewers often discuss why a candidate did not utilize the full resources of his institution especially if the university has been designated a “Center of Excellence” in a particular area.

11.Also early on in designing the research plan, ask a statistician to be actively involved. A statistician can facilitate sample size calculations, adjust over optimistic effect sizes of odds ratios, and most importantly, tighten the writing in the statistical sections. On every study section, at least one statistician is normally present. Even if a statistician is not assigned to review your grant, the concise nature of the statistical section allows them to comment on the adequacy of the statistical plan of analysis. I found that statistical reviewers sense after reading

196 M.V. Brock

one or two sentences if the candidate has consulted with a statistician or not.

12.Statisticians can help advise if your patient accrual goals are too ambitious. Be prepared in your consultations with them to ask questions such as, “What would be the power of my study if only 75% of accrual goals are attained?”

13. Does your research plan or question push the field forward or fill an unknown in our knowledge? How important is the research grant in terms of scholarship? What are the applications of this research? An exciting, innovative research idea is attractive to a reviewer even for a K Award.

14. When writing the research plan, remember that the grant should be self contained and thus self explanatory. The reviewers should not need to refer to supplementary material outside of the grant to understand the background or significance of the research.

15.Be very forthright about the genesis of the preliminary data you employ in your grant application especially if it is generic to your mentor’s laboratory. If the data are not the result of your own hand and you do not acknowledge this explicitly with the mistaken belief that you are impressing the reviewers, almost invariably the result will be the exact opposite. Reviewers are surprisingly adept at knowing and finding out what your mentors are doing (especially in the closed confines of a group NIH study section with free internet availability).

16.If databases or websites are used in the preliminary data, how reliable are these sources to base a research idea for a K Award? Ask oneself, how have these sources been validated? For example, have these survey data been replicated? If not, should this be one of your Specific Aims?

17.Toward the end of the research plan, acknowledge its limitations. For example, with help of your statistical colleague, you can write about biases such as recruiting English-only speaking patients in a survey from a population with a growing Hispanic community or the deficiencies in using a mouse model that does not quite adequately reflect human disease.