The BUILD awards differ from other NIH-funded training grants in that they aim to impact simultaneously students, faculty, and institutions by implementing a variety of interventions targeted toward research capacity-building, mentorship, and institutional change. In 2014, ten competitive 5-year BUILD awards were issued to undergraduate institutions across the country. Eligibility for BUILD primary institutions included having less than $7.5 million in total NIH research project grant funding and a student population with at least 25 percent Pell Grant recipients. The funded BUILD institutions serve geographically, socioeconomic, and racially diverse populations, and include historically Black colleges and universities, Hispanic-serving institutions, Asian American/Native American/Alaska Native/Pacific Islander-serving institutions, and institutions that target outreach to special populations. Higher Education Research Institute (HERI) surveys of incoming BUILD students from 2015–2019 documented that many BUILD scholars were learners from groups historically underrepresented in biomedical sciences [9]. Funded BUILD institutions partner with nearly 100 other institutions (S1 Appendix), some of which had basic science education programs and that were research-intensive, to broaden the opportunities for the students participating in biomedical research training and maximize opportunities for faculty and staff development. The ten funded BUILD institutions with links to their programs are:

All institutions except UMBC developed a pilot project program designed to support faculty in writing research grants while simultaneously providing an educational platform to promote the pursuit of research by underrepresented students. The approach taken by BUILD primary institutions followed the steps of pilot Request For Application (RFA) development and announcement, using program support tools, implementing the research, and measuring key outcomes (Fig 1).

Each BUILD primary institution had some program tools already in place, which provided background support for the development of their pilot project programs; details of their approaches vary slightly. It was not possible to separate out faculty use of the BUILD funded program tools from research infrastructure (e.g., Clinical and Translational Science Institutes, etc.) that already existed in the funded institutions because the baseline HERI survey did not differentiate between the two. We, therefore, treated all such tools as potentially supporting BUILD funded pilot projects. These ten tools included 1) workshops on teaching; 2) developing research skills, and grant writing; 3) funds for sabbaticals; 4) institutional funds for travel and research; 5) training for administrative leadership; 6) incentives for developing new courses; 7) incentives for integrating technology in the classroom; and 8) incentives to integrate culturally competent and student-centered practices in the classroom. Although details differ in implementation, all sites shared the aim of supporting faculty research productivity and each element that is reported here was used by at least some HERI faculty survey participants at every site. In our analysis we included an evaluation of the role of program support tools. Seven quantitative productivity outcomes of interest were measured at each institution as reported by participants associated with the BUILD pilot programs: 1) numbers of BUILD pilots funded, 2) students included, 3) abstracts, 4) presentations, 5) publications, 6) submitted grant proposals, and 7) funded grant proposals.

The BUILD Coordination and Evaluation Center (CEC) annually collected data directly from faculty as a component of the Consortium-Wide Evaluation Plan (CWEP). In 2016–2017 faculty data were gathered using the HERI Faculty Survey, which is administered triennially [1]. Survey participants included faculty participating in various BUILD interventions including the pilot project program and a comparison group of faculty at the same institution. The target was 50 responses per BUILD program: 25 participating in the BUILD program and 25 for the comparison group. However, some interventions targeted most or all faculty in a department, college, or institution, leaving few faculty without exposure to the interventions for comparison, so the intent to have an internal comparison group was not feasible.

The HERI Faculty survey is a self-assessment comprising 55 questions encapsulating individual demographics, teaching and research experience, current scope and scale of roles in teaching, research or administration, and attitudes and uses of program support tools [12]. The survey included validated instruments from HERI so data could be compared with other studies that also used these validated instruments. Table 1 summarizes aspects of sample selection for the 2016–2017 survey. Most programs administered the survey during Fall 2016; the UDM administered the survey in Winter 2017. Some programs offered raffled incentives, some offered incentives to departments with high response rates, and others offered no incentives. The Survey data we analyze here are only from faculty who participated in BUILD (Table 1).

The nine BUILD primary institutions, excluding the UMBC, collected data from faculty participants of the pilot project program on productivity as part of their annual progress report to the NIH. We collected and analyzed the productivity data from progress reports submitted from 2015–2018 that summarized data per institution; counts of BUILD pilots funded, students included, abstracts, presentations, publications, grant proposals submitted, and grant proposals funded.

To describe the ten program support tools (not to be confused with the ten BUILD sites), we calculated the proportion of participants who used each tool at each institution. We characterized institutions by the distributions of gender, race/ethnicity, and faculty experience with education.

To assess which program support tools might be aligned with which productivity measures, we first designated each institution as a binary high or low use of the tool which, depending on how commonly the tool was used overall, was based on a cutoff of at least one half, one quarter, or one tenth of faculty. We then fit a Poisson regression model to the productivity count data and tool use data at the nine institutions in the study. The standard error of the regression coefficient was corrected for overdispersion by using the dispersion parameter, and the regression coefficient was exponentiated to present the association as a rate ratio of an increase in productivity for institutions that had high use of a program support tool compared to low. We used the Bonferroni adjustment to the 77 association tests to be conservative in concluding that any tool was significantly associated with any productivity measure. All tests were two-sided and considered statistically significant at the Type I error rate below 0.00065. Analyses were done using SAS software, Version 9.4, SAS Institute Inc. Data will be made available upon request.

A total of 450 individuals participated in the 2016–17 HERI Faculty Survey; 51 participants were excluded from the analysis since no productivity data was available from the participants’ institution (i.e., UMBC). As shown (Table 2), 220 (55%) participants identified as women, and 82 (20%) belong to racial/ethnic minority groups. Of the 313 respondents who reported race/ethnicity, 33 (11%) were Black, 40 (13%) Asian, 19 (6%) Latinx, 18 (6%) multiracial, 12 (4%) other races, and 181 (58%) White. Median age was 48. The largest academic group was Assistant Professors (152; 38%) with a close to equal distribution of Associate Professors (109; 28%) and Professors (104; 26%); out of those faculty 204 (52%) were tenured. Participants had a median of 11 years of teaching experience, teaching two courses per semester, mentoring four research students per years, and spending seven hours per week on research and scholarly writing.

The most used tools were workshops on teaching (59%), workshops on research skills (37%), workshops on grant writing (29%), institutional funds for travel (55%), grant funds from the institution (51%), incentives for technology (36%), and incentives for integrating culturally competent practices in the classroom (40%) (Table 3).

It is likely that some of these tools that faculty reported using were not newly developed for the BUILD program but are included because they characterize the institutional environment. Approximately three-quarters of participants reported using the teaching workshops at CSUN, Xavier, and UDM.

Approximately two-thirds of participants used workshops on research skills at MSU. At least one-half used institutional travel funds at CSULB, CSUN, MSU, UDM, PSU, and UTEP and, likewise, at least half used research grants provided by the institution at CSULB, CSUN, MSU, and UDM. At least one-half of participants used incentives for adding technology in the classroom at MSU and UTEP, and incentives for adding culturally competent practices in the classroom were used by at least one-half of participants used at MSU and UTEP and at CSUN, MSU, and UDM, respectively.

The pilot funding amount was approximately $10,000–50,000 per pilot and most were for a duration one or one and a half years, but others ranged up to five years (UAF 2–2.5 years; MSU 1–3 years (Table 3). The BUILD project grants amounts were on average at least $50,000 at four institutions: UAF, CSUN, MSU, and PSU. The average award amounts varied between institutions from $14,396 at CSULB to $73,552 at Portland State University.

The number of pilot projects supported by each BUILD program varied from five to 38 (Table 4). Most pilot programs were offered between 2015 and 2018; except for PSU where pilots were only supported in 2016–17. The average number of students engaged in the research supported by these projects overall was 42 with as many as 160 students involved at CSULB.

An average of 12.1 abstracts, 35 presentations, and 9.3 publications per institution were produced in BUILD projects. CSULB reported the most peer-reviewed publications (47) generated from their pilot projects. Among all BUILD sites, an average of 17 grants were proposed and 6.4 successfully funded. Almost all of proposals (73%) were submitted to the NIH (N = 41) or other federal funders (e.g., National Science Foundation) (N = 49) and the number of grants submitted or funded from other sources was thus too small for meaningful comparison. The largest number of grants funded because of these BUILD pilot projects was 15 at PSU.

Often institutions where more faculty used a program tool there was lower productivity (Table 5). However, there were also various program tools that aligned with more productivity. At institutions where more faculty used a workshop on teaching there were more students involved, more abstracts published, more presentations made, and more publications produced. Also, at institutions where more faculty used a workshop on grant writing, more students were involved and there were more publications. Sabbatical use was associated with more student involvement and more publications. In addition, at institutions where more faculty had research grants, more students were involved and there were more publications. When more faculty used training for administration there were more funded grants and where more faculty had incentives to develop a new course there were more abstracts published.

On the other hand, the associations tallied from this study suggested that fewer students were involved in projects at the institutions where more faculty used a workshop on research skills, incentives to develop a new course, incentives to integrate technology in the classroom, or incentives to integrate culturally competent practices in the classroom, and where BUILD grants were more than $50,000. Fewer abstracts were produced at the institutions where more faculty used a workshop on grant writing, sabbatical, travel funds, research grants, training on administrative leadership, or incentive to integrate culturally competent practices in the classroom, and where BUILD grants were more than $50,000. Fewer presentations were made at the institutions where more faculty used a workshop on research skills, incentive to develop a new course, incentive to integrate technology in the classroom, or incentive to integrate culturally competent practices in the classroom, and where BUILD grants were more than $50,000. There were fewer publications at the institutions where more faculty used a workshop on research skills, incentive to develop a new course, incentive to integrate technology in the classroom, or incentive to integrate culturally competent practice in the classroom, and where BUILD grants were more than $50,000. There were also fewer submitted grants at the institutions where more faculty used a workshop on research skills, incentives to develop a new course, incentives to integrate technology in the classroom, or incentives to integrate culturally competent practices in the classroom. Finally, there were fewer funded grants at the institutions where more faculty used a workshop on research skills, incentive to develop a new course, incentive to integrate technology in the classroom, or incentive to integrate culturally competent practice in the classroom.