David Murray, Ph.D.
Associate Director for Prevention, National Institutes of Health, and Director, Office of Disease Prevention
Dr. Murray outlined the intent of the Office of Disease Prevention (ODP) to increase the scope and impact of prevention research at the NIH and support prevention activities across all research phases, based on the best available evidence; to support both cross-Institute and public/private collaborations; and to be a trusted source of information in the field of prevention.
Dr. Murray also emphasized ways in which this workshop could help. The attendees were multidisciplinary, and therefore provided ideas for unique partnerships. They also were there to offer and spark new ideas about ways to move the science along in an area where the guidelines are fairly clear, but still are not being followed in the desired numbers.
Workshop Co-Chair CAPT Rick Troiano, Ph.D.
Senior Scientist, Risk Factor Monitoring and Methods Branch
National Cancer Institute
Dr. Troiano introduced the subject of physical activity (PA) by asking that attendees strive for semantic clarity. Since there were behavioral researchers, PA researchers, and interventionists around the table, he encouraged everyone to be clear and explicit about what was meant when using terms such as “physical activity,” “physically active,” and “sedentary,” and to define how these concepts were measured.
Workshop Co-Chair Russ Pate, Ph.D.
Professor, Department of Exercise Science
Public Health Research Center, University of South Carolina
Dr. Pate gave a brief history of the study of PA and public health that included a current status report on the National Physical Activity Plan and the 2008 Physical Activity Guidelines for Americans. He concluded by outlining research needs and the coordinating role NIH can provide by unifying and translating the research into practice, developing program evaluation methods, and performing efficacy studies.
One of the topics included the way PA is often equated with obesity. There is an educational charge to make sure people understand that PA is not only about obesity, but rather about multiple outcomes that cut across various Institutes and Centers, and that might tap into various funding sources. This would work if PA itself is considered a valid standalone outcome measured in an objective way.
A second topic was how to make PA and exercise an important societal issue, not just an individual issue, the way that second-hand smoke made tobacco a societal issue, and now people are embarrassed to smoke. When it is perceived as beneficial for everyone, it will shift the social norm so that living an inactive life just doesn’t seem right, and that will be the driver for societal change. Lifecourse issues also need to be examined in order to understand where PA occurs in different sex and age groups.
A third topic was the need for coordination of PA research within and across the NIH. Spending on the topic is at present not tracked among 200+ congressionally required topic areas. This is important to tackle, especially since the health effects are clear and the public response is underwhelming. An attendee recommended finding a “home” for PA coordination and funding at NIH.
The final topic discussed was initiated by a participant who advocated thinking of PA as a “positive gateway behavior” for youth to enhance family systems and self-efficacy. Inner-city programs such as Soccer for Success and Chicago Playworks have affected children’s drop-out rates, pregnancy rates, and gang activity. The sociological effects of PA need to be measured as well as physiological effects. In addition, the social and contextual ways to make PA fun and attractive need to be included in the study and promotion of PA.
Ross Hammond, Ph.D.
Director, Center on Social Dynamics and Policy
The Brookings Institution
Dr. Hammond discussed techniques from the field of systems science as they might apply to improving public health approaches to PA. He focused on a technique that is relatively new to public health and preventive medicine called agent-based modeling (ABM). ABM, a form of computational modeling, is used to study complex problems that lend themselves to systems science methodologies that draw on diverse research disciplines.
This technique can take into account an agent (in this case a person) within a society and all the interrelated influences both “below the skin” (genetic, psychological, neurobiological) and “above the skin” (environmental, social, geographical) with the ability to manipulate variables over time, context, and the life course. ABM tools have had a strong track record in social science, biological science, and infectious disease studies. The patterns generated by these models can be compared to existing data, and can draw in data at different levels.
Dr. Hammond showed an animated simulation movie to demonstrate the co-evolution of influences with individual behavior, and mentioned that there is a one-week summer intensive training session on these techniques sponsored jointly by the NIH and the CDC.
Deborah Cohen, M.D., M.P.H.
Senior Natural Scientist
Dr. Cohen’s central question was how to make PA ubiquitous and appealing in the way that coffee, junk food, and electronic media are now.
Research shows that people are more interested in avoiding losses than they are in getting gains. So if we want people to adhere to PA guidelines, we must phrase messages in the negative: You can avoid the losses (harm to your health) by getting rid of inactivity.
Dr. Cohen co-authored a study using SOPARC (System for Observing Play and Recreation in Communities) to gain nationally representative data on parks and their relationship with getting people active in communities. This study was funded by the National Heart, Lung, and Blood Institute (NHLBI). They found that 50 percent of vigorous activity occurs in parks. But parks are underutilized, and there is a gap in information or data on park usage. Some techniques to draw people in include creating “fitness zones,” implementing park programming, including classes and organized activities, and improving signage. Other proposals include implementation of large demonstration projects that include collaborators from many other sectors, such as transportation, housing, etc. She also mentioned a need to demonstrate the cost-effectiveness of PA interventions—dollars spent per MET (metabolic equivalent task) hour or minute of moderate-to-vigorous PA. A priority on PA has to be tied to the parks, and they must remain open during leisure hours.
James Sallis, Ph.D.
Distinguished Professor, Department of Family and Preventive Medicine
University of California, San Diego
Dr. Sallis began with an analogy of what a villain like “Dr. Evil” might do to keep people away from PA. The answer: put everyone in a car on a busy street. The presentation focused on policy surrounding environment changes to increase PA.
If a community is designed correctly, you should be able to bike or walk to wherever you want to go. And if you affect policy and environment, you affect everyone. The environments that encourage activity are walkable. These are denser, mixed-use areas with destinations to walk to and cul-de-sacs with sidewalks that cut through to the next cul-de-sac.
In addition, we need bicycle-friendly policies so people use their bikes for transportation. People in Minneapolis do four times more cycling than those in San Diego. This shows that PA-related transportation has less to do with weather and more to do with policy. But the people making the policy decisions are focused on money, cost, and return, so we have to have data that is focused on money. We need to improve the rigor of the research, develop simple environmental measures that community groups can use, develop economic analyses, make the policy makers a part of the research teams, and communicate well by supplementing the data with stories and case studies for lay audiences.
Some challenges that Dr. Sallis identified include: collecting data that researchers are interested in, connecting the data streams, and understanding how to operationalize changes while improving measurement. He also noted that policy, environment, and organization interventions have higher impact than individual-level ones, and yet NIH funds more individual and social research compared to other types. Dr. Sallis also mentioned using combined and “packaged” approaches.
Those at the workshop brought up the following responses and topics:
The NIH currently has program announcements to do natural experiments on green space, and there are studies underway that are policy purposed. However, with NIH funding cycles, it’s hard to get consistent policy information over the long term.
Questions about systems modeling indicated the ability to identify a few factors that work and model them together. It was suggested that the corollary in a real setting would be to try a lot of things, see what works, and replicate those that work.
The point was brought up that there is much talk regarding the availability of leisure time, but that large segments of the population do not have much control over their time. This was countered by citing the enormous amount of leisure hours spent in front of screens for all segments of the population.
An attendee asked if systems science could be related to systems biology, since systems biology seems to be having no problem getting funding. A speaker replied that it’s important to have people on the review panels who understand systems science. Another participant suggested the Office of Behavioral and Social Science Research (OBSSR) might have an interest in the systems science approach to PA.
There was a discussion about the relative merits of self-report and objective measures and, since each has pluses and minuses, it was suggested not to throw either out, but to use them in combination.
There was a question suggesting that people think not only of infrastructure, but also of cyber-infrastructure. Since there will be a good amount of data coming through the developing cyber-infrastructure, one key element for PA research will be to know what questions to ask of the data and how to analyze it according to those research questions.
Bess Marcus, Ph.D.
Professor and Chair , Department of Family and Preventive Medicine
University of California, San Diego
Dr. Marcus began by asking, “What do we know?” From her perspective, we know that land use mix and residential density are the most consistent environmental predictors of PA in children and adolescents. Also, self-efficacy has been shown repeatedly to be an important determinant of the overall effectiveness of interventions to promote PA in children and adults.
There are many interventions that work, but how do we make them happen? How do we implement them? What are the effective interventions that get people motivated? She noted that theory-based interventions should be combined with communication technologies to produce changes in PA behavior and increase effectiveness; with site-based approaches, just because we build it doesn’t mean they will come! The “digital divide” is narrowing, even among underserved populations.
Dr. Marcus outlined the pros and cons of individually targeted interventions. She found certain behavioral and psychosocial approaches generally effective, low-cost, and sustainable. These included social support, such as buddy systems and walking groups. Contracts people sign to affirm their exercise resolve worked well, as well as programs located at the workplace. Messages that were tailored to readiness to change had an impact, as did goal-setting, monitoring, and self-reward. And again, self-efficacy-based messages showed the most promise. The challenge would be in widespread dissemination, through key communication channels and a system-based approach.
Site-based approaches, such as worksite interventions, community-based classes, and clinic-based interventions, have the potential to be sustainable and have a broad reach while capitalizing on existing social networks. However, they can be more costly. Community-wide mass media campaigns have a wide reach, but also a high cost. Point-of-decision prompts, such as flyers, can have a wide reach and low cost, but there is a question about how much these translate into actual health benefits for the people they reach.
School-based interventions, such as physical education classes, are consistently shown to be effective and can instill habits of lifelong PA and resulting reduction of cardiovascular disease (CVD) risk factors. An effective school-based physical education program must include frequency, quality, and proper equipment, as well as capacity-building and self-training. This intervention reaches large numbers of people from all segments of the population.
Dr. Marcus reiterated that “If you build it, they will come” is something of a fantasy, and that environmental modification, such as facility enhancement alone, is not always effective, but there are environmental modifications that do make a difference. The size of the park, the establishment of organized activities, the facilitation of walkable transport, and neighborhoods for both adults and children are highly effective in getting people moving.
Something that is highly motivating and effective is a combination of environmental changes and psychosocial interventions. Dr. Marcus emphasized that we have to ask how we can look at multiple approaches at the same time. We tend to pick one area and go in that direction, but successful behavior change depends upon a combined approach. One example is Ciclovía, a car-free, bike-only day that takes on a festive mood and gets people thinking about PA as well as moving on that day. Another intervention that works is community trail construction projects. Community projects are more challenging to evaluate, but the evidence has shown that combined strategies offer the most promise. A walkable neighborhood, combined with the social supports mentioned above, may be necessary to change communities.
The National Physical Activity Plan involves strategies for mass media dissemination of interventions and, with the changing media landscape, it makes sense to consider the kinds of media that would be most effective. We have to take what we know about emerging technologies and combine this with what we know about social science, in order to keep up with people’s needs. Among underserved communities, who rely disproportionally on mobile phone communication, and the general population, mobile devices may prove to be the most effective means of disseminating promising interventions and programs.
Abby King, Ph.D.
Professor, Health Research and Policy and Medicine
Dissemination Research: Dr. King emphasized that dissemination is key, and that reaching underserved populations has to be an important focus. Are we really targeting the different population groups in a way that makes sense? We must target populations not just in terms of culture, but also in terms of the subpopulations who have or are at risk for diabetes and other diseases. She said that the important question to ask is: Where do we need to go to get the biggest bang for our buck for population-wide PA?
She listed several potential growth areas for PA research, including dissemination and translation of research at the top of the list. She said we really don’t have an understanding of how to disseminate programs across populations, and we need a better understanding of the delivery channels and settings that work best. There might be interventions that need to be individually adapted according to audience and, for now, we could use the good examples as models.
Dr. King said that some effective programs include settings where we teach and excite and innovate, such as trainings in cooperative extension centers and community organizations, teaching group-based behavioral skills such as diabetes prevention, weight loss, and exercise instruction. School interventions can be translated for underserved populations. When looking at such programs, reach and cost-efficiency should be priorities. These might be achieved through training lay workers to promote PA outreach, or through information technology methods that do not need onsite training, such as automated delivery systems. Telehealth, virtual advisors, smartphone platforms, and social media all fall into that category.
Comparative Effectiveness Research: Comparative effectiveness research (CER) provides a means of clarifying the value-added and cost-effectiveness of alternative PA intervention strategies. We need to use participants and settings that are typical of our day-to-day lives. How often does the average person use the car rather than walking, for instance? If we put what we think are efficacious automated programs head-to-head with proven alternatives, then we can clarify the cost effectiveness of each, as well as the value-added benefits. In the current CER report, PA is barely mentioned. And it will be important to compare different PA formats and delivery channels side by side.
Research Study Control Arms: Another growth area Dr. King suggested for PA research was developing consensus in the field around the control arms of NIH-funded studies. At the moment, there are differences of opinion about how to handle this issue. There’s a need to convene a panel to recommend a framework for guiding the choice of the most appropriate and efficient controls. Consider “practical trials” to reduce control reactivity, increase external validity, and reduce respondent burden.
Environmental and Policy Research: Trans-generational approaches are needed. It’s not enough just to put out RFAs for natural experiments. We need to teach researchers how to do this kind of research, and incentivize these experiments. They can then result in less costly and more efficient interventions by devising experiments to evaluate environmental and policy activities such as park-based exercise.
Dr. King also suggested that researchers compare different clinical referral schemes to PA providers to enhance and encourage health behavior counseling in primary care. It is empowering, she said, for patients to be given these choices by their doctors. And since physical inactivity can be a “gateway” signal to other conditions or diseases, it is important to explore the synergies between PA and other health behaviors and treatments. For instance, when a person is referred to a dietician, why not marry the expertise on diet with a similar expertise on PA?
Dr. King argued that all types of research need to be funded, not just a narrow spectrum. We need to build interventions from the bottom up. For instance, the IT interventions that could be utilized to reach underserved populations need to be considered with the same kind of care as a randomized controlled trial (RCT). For all interventions, an action plan must be devised. The question is not only “Does it work?” but also “For whom does it work over time?” Writing a white paper is fine, but what we really need is to let those in the field know what they need to do. She recommended making a list of action steps of what researchers would need to do to move forward.
Enhancing Systematic Reviews: Another area for improvement, according to Dr. King, is enhancing the quality of systematic reviews. There are shortcomings as to the way many studies are selected and coded. She recommended that standards and guidelines for search terminology and coding be developed for literature reviews and meta-analyses.
In closing, Dr. King remarked that she favors, as others have mentioned, improving the NIH grant review process. Specifically, PA needs to be considered an outcome in and of itself. Additionally, we need to engage senior scientists working in the field to participate in the grant review process in the same way they have done for the American Recovery and Reinvestment Act (ARRA) grants. As our international colleagues have done, we should also consider funding international networks that are aimed at accelerating PA science.
Linda Collins, Ph.D.
Professor and Director, The Methodology Center
The Pennsylvania State University
Dr. Collins focused on an engineering-inspired approach to the development, optimization, and evaluation of behavioral interventions called the Multiphase Optimization Strategy (MOST). The advantage of MOST, as opposed to conventional methods, is that this approach makes it possible to find an intervention that meets specific criteria.
MOST is not a treatment package approach. Rather, MOST is used to engineer a preventive intervention to meet specific criteria that provides an optimal framework for interventions rather than an “off-the-shelf” package. The process entails looking at the components that go into interventions. This is in contrast to a randomized controlled trial in which all of the components are considered simultaneously. If the trial is successful, the investigators cannot determine which component(s) are beneficial. Additionally, if the RCT shows a non-significant effect, the investigator does not know if any of the components are worth retaining.
One example Dr. Collins described was a Northwestern University study funded by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK). The objective was to select intervention components that would provide the most effective weight reduction intervention that could be implemented for less than $500.
MOST permits researchers to screen and refine their interventions. Then, when you find the components that meet your criteria, that’s the intervention you take to an RCT. MOST is a framework for optimizing the interventions, not an off-the-shelf procedure. There is no one best design. But factorial experiments can be very economical.
Dr. Collins cited other studies that have used the MOST approach funded by the National Institute on Drug Abuse (NIDA) and the National Cancer Institute (NCI).
The co-chairs facilitated an open discussion of issues related to PA as a primary outcome when submitting grants.
NIH Grant Review Process: One participant expressed frustration that outside the field of behavioral health, the perception of PA as a bona fide outcome measure is not appreciated across the NIH. PA as a primary outcome is thought not to be a valid measure. Some Institutes and Centers (ICs), notably NCI, have research portfolios that include PA as the primary outcome, but this is the exception rather than the rule. There should be a mechanism to communicate PA as a valid outcome across Institutes to both NIH staff and reviewers. Then NCI can help other Institutes “get it” about PA being important.
David Murray said that suggestions for fixing the NIH review process were encouraged, especially when there are specific problems that are identified, and specific changes are requested. These types of changes are included in the review of the ODP strategic plan. He said that he had a conversation with Dr. Nakamura (Richard Nakamura, Ph.D., the new director of the NIH’s Center for Scientific Review [CSR]), and that ODP would be happy to serve as a conduit to CSR.
Training Programs: A question was raised about how training programs and grants could fit into future funding opportunities. How do we train future researchers for promoting PA interventions, novel approaches, and measurement of the behavior and physical outcomes? One of the panelists answered that the solution has to be a trans-Institute solution. There is a good case for this if principal investigators and program officers are working together and integrating across centers.
Another participant commented that university graduates need know how to conduct research before they leave school. NIH-funded training opportunities are an important component of investigator training.
Timelines Related to Formative Research: Another discussion point about multi-component interventions brought up the fact that there is evidence that they are important, yet much of the interconnectedness of these approaches has to do with relationships, which do not occur overnight, and therefore do not fit within traditional R01 time constraints. R01 timelines do not allow enough time for program development and recruitment or a formative phase; they also do not allow enough time for researchers to examine the sustainability phase. The NIH needs to get more proactive with pilot funds. Researchers need funding up front to show initial efficacy, to build relationships, and then, when all that is established, to go for the R01. During the final intervention stage, as well, there is not enough time to analyze the data, nor is there sufficient time to determine if the outcomes would be sustained over time.
Funding Opportunities: As for funding, it was suggested that the group look to the Small Business Innovation Research (SBIR) program as a possible model for funding phases of research. That is, investigators would receive funds during Phase I for formative research and intervention development. Upon successful completion of Phase I, investigators would apply for Phase II funds for full implementation and evaluation.
One participant suggested that rather than limit the amount of money you can ask for in a year, maybe NIH could limit the total amount of money you can ask for over time. Factorial experiments would help alleviate some of these problems, because there are many combinations, but it doesn’t take as much time.
Follow-up and Evaluation Activities: A suggestion about evaluating and following up on interventions was to build pieces into these studies that include tracking. Are they going to a particular website? How long are they staying there? How much self-monitoring is taking place? If Internet-delivered interventions are used, it is easier to develop factorial experiments. Sometimes, on a grant schedule, this can be difficult, because the technology (hardware and software) becomes obsolete before things are reviewed and can be implemented. When you want more time to study things, it will take that much longer to get it into practice, since the average is now 17 years to get an intervention into clinical practice.
One of the panelists said that researchers should work smarter, not longer, and efficiency was necessary. Another participant said that there is a structural barrier that limits the dissemination of NIH-funded intervention findings. You put a lot of energy and resources into implementing the intervention, but then in order to replicate the results, someone has to invest the same amount of resources again. In the real world, when something is exciting, it takes off like a rocket. This is not true at NIH. The funding and review process limits things from taking off as they do in the real world.
Patty Freedson, Ph.D.
School of Public Health and Health Sciences, Department of Kinesiology
University of Massachusetts, Amherst
Dr. Freedson spoke about self-report and objective measures of PA, noting that the specific assessment method used in research should be determined based on the research question and the needs of the particular study. There are strengths and limitations of self-report methods and objective measures. For instance, self-report is low cost and easy to administer. It can also assess context and exacts a low burden on the participant.
On the other hand, self-report may have measurement errors or recall bias. It may also leave out some dimensions of behavior, or it may not measure certain activities. It also lacks validation for assessment of change in behavior following interventions, which needs to be more clear-cut with self-report tools. If certain behaviors are missing for certain populations, that could be rectified with this type of validation. At the moment, there is a lack of standards and harmonization among types of self-reports.
Still, self-report is a useful method and sometimes the optimal choice. What is still needed, according to Dr. Freedson, is a system to match the specific instrument for the needs of a particular project. If a study is conducted for PA and the tools don’t match the needs of the study, then we haven’t been as effective as we need to be in improving the quality of the information we’ve collected. But the information in a self-report tool can complement what’s in an objective tool.
Dr. Freedson then described the pros and cons of wearable sensors to assess PA and sedentary behavior. They do provide an objective measure over a long period of time, they can gain estimates of many measures, and in some cases they can provide real-time feedback. In addition, many of the devices can be adapted to mobile phones. But there are no standards of practice yet, and there are many different devices, although with raw signals this problem shouldn’t matter. Then one must depend on compliance by users, which may not be consistent, as well as variability in data management and processing. Lastly, these devices still need to be validated in natural settings.
Dr. Freedson’s talk also mentioned an analytical approach for device-based data called Artificial Neural Network, or ANN. ANN is a set of nonlinear statistical modeling tools that change structure based on internal and external information, which are often used in prediction models. She closed by briefly reviewing other measures of interest to PA that involve geographical imaging like GPS, GIS, and Sensecam, useful tools in measuring where PA occurs. Location is important because it can inform intervention development and monitoring.
Steven Intille, Ph.D.
Associate Professor, College of Computer and Information Science and Bouvé College of Health Sciences
Dr. Intille said that there were many “gadgets” or devices that might be used to measure and study PA versus sedentary behavior. He urged researchers to think long term and therefore to build effective and cost-effective devices. And it’s quite important that these devices measure the context in which the PA takes place. We don’t want to know just whether an individual is exercising, walking, etc., but also why and where. These are the questions that will inform future interventions.
He said that passive sensing assumption doesn’t go very far. Interventions require interactions. The end users (the public) can provide information that will guide data collection — where and when to collect data. It’s also important to gather data by asking only what you need to know.
Researchers need to care even about sedentary activities like making the bed, cooking, and so on, so there can be interventions developed to address these activities as well. This is why we need long-term, detailed data on individuals to build the models. He urged workshop participants to imagine how having data on PA for many, many people for the last year would change what they were doing.
Objective methods that are used to measure PA include activity monitors such as accelerometers and pedometers worn on the belt, the shoe, or the wrist; heart rate monitors; and smart phone apps, among others. Dr. Intille favored passive methods in which subjects wear a device and need no training to begin registering data. As an example from his own research, he described the prototype “Wockets” system. This has components that are worn on various parts of the body that measure the type of energy expended by capturing both upper and lower body motion. Data from the sensors can be uploaded from the wearer’s cell phone. When coupled with data from a phone, it can give context for the user’s activity. For example, the data reflects where the sensor is placed and the type of activity (walking, running, biking, etc.). He predicted that mobile phones will become the hub device for PA data collection because they facilitate long-term compliance monitoring and permit researchers to gather information about context from multiple types of sensors, including self-report. The technology in this field is rapidly developing such that data collected by wearable cameras, sensor devices mounted in eyeglasses, and in-home “object use” sensors are likely to add to the repertoire of PA measurement tools.
Population Assessment: One question centered on the collection of this kind of PA data from disparate populations. What information or results do we already have? We need to think about which tool is most appropriate for which population. The panelists responded that if there is an important study, it will be easier to convince members of a certain population to participate. It’s a different question if you are going to send an intervention out into the world, but for research studies, people are willing to participate regardless of the tool.
There was a hope expressed that this data could be collected in real time. There are a lot of calibration studies in the literature, but much less information about validity when the instruments are taken into the field.
Validation Research: A question was posed about the validation framework. How is it decided which new opportunities for intervention devices or methodology are the best and can be used in the field? A panelist responded that while the measurement might begin in the lab, we work to make the measurement work in real time and the real world. Then the participant can come back into the lab or communicate with the researcher electronically, and the researcher can test his or her assumptions: “I think you were doing X. Were you doing X?” This helps to validate the tool.
Real Time PA Assessment Methods: There is dramatic progress being made in the ability to measure PA in real time. Device assessment helps define behavior. For instance, in a baseball game, some players are batting, some are sitting on the bench, some are standing in the field, and some are running. In terms of observation or self-report, they are all playing baseball. But when combined with the device assessment, the different activities can begin to be differentiated from one another.
And while there is value in determining the type of behavior, GPS can also begin to determine purpose of behavior, the “why.” Is someone taking a leisurely walk, or transporting him- or herself to a destination? Are people walking, or are they biking? This is why we need both self-report and objective measures. We need creative use of all the tools in the toolbox. We can create patterns, and the computer can detect those things that we cannot, based upon input going into the system. Then we can add in energy expenditure, blood pressure, and cholesterol, create models, and see what makes sense for certain profiles. What makes sense given the patterns people follow if they are more active?
Researchers have been careless because we have lumped lots of things into PA. [PA researchers] are now starting to be clearer in the field about what we mean. So what you measure should be closely related to what you care about. The devices don’t measure the behavior, they measure movement. And we’re trying to map to specific activities.
Barry Portnoy, Ph.D.
Senior Adviser for Disease Prevention, Office of Disease Prevention, National Institutes of Health
Jessica Wu, Ph.D.
American Association for the Advancement of Science (AAAS) Science & Technology Policy Fellow, Office of Disease Prevention, National Institutes of Health
In this session, Drs. Portnoy and Wu described the distribution of projects that the NIH funds on the topic of PA. The two main challenges have to do with the difficulty in tracking multiyear projects and in accurately reflecting the budget on long-term PA experiments.
After Dr. Portnoy briefly described the role of portfolio analysis in strategic planning and analyzing budgets, Dr. Wu presented the process used to perform the portfolio analysis and described the NIH PA portfolio retrieved from this process. The limitations of the portfolio analysis include: the restrictive nature of the Research, Condition, and Disease Categorization (RCDC) computerized coding process (results were limited to the “prevention” RCDC category), the inability to track multiyear projects, inaccuracy of the budget numbers reported, and the lack of details in further categorizing the portfolio. At this time, because of the preliminary nature of the study, this presentation was only a very rough overview of the PA portfolio. The results of the analysis were more qualitative than quantitative.
The presenters estimate that 6 percent of the NIH prevention budget relates to PA in some way. If PA were its own RCDC category, its budget would be on the same order of magnitude in funding as the RCDC categories Chronic Pain, Tobacco, and Injury.
According to the number of projects in the portfolio, the top five Institutes that host PA projects, in order, are NHLBI, NIDDK, NCI, Eunice Kennedy Shriver National Institute for Child Health and Human Development (NICHD), and the National Institute on Aging (NIA). Most of the PA projects were research grants (R grants), with most of the grants awarded as R01s. When topic or text analysis was performed on the portfolio, the expected topics and terms we found to frequently appear in grant titles and abstracts: obesity/children, diabetes, weight, and cancer.
Comments and questions from the audience during the presentation included:
The portfolio analysis rubric should be narrowed to projects that include PA as a primary or secondary outcome. These can be identified by looking at the titles and abstracts of the grants to see if the grantees state that PA is a targeted outcome of the study.
When doing a portfolio analysis, different strategies can be used to identify current NIH-funded PA projects. Although search terms can be used to identify projects, the text search tools that are currently used lack the desired specificity, thus adding significant burden to the portfolio analysis process. A proposed solution from the workshop attendees to improve and standardize the coding process would be to ask the author or grant submitter to pick key words from four or five pre-assigned categories that NIH would compile. This would assist in searching for projects according to the categories the grantees/PIs think they should fit into.
Drs. Portnoy and Wu then asked for feedback about where the NIH might focus in the next 5 years. They asked, “Are there themes or areas of research that should receive increased or decreased emphasis?” The responses included the following:
Data Gathering: NIH should put out a grant to ask for computer scientists to suggest ways of gathering data related to Big Data projects.
Funded Grants: What about using RePORTER to see what grants are in what areas?
Communication Issues: A participant asked if there would be a mechanism for getting this information out, specifically the outcomes of this workshop? Dr. Portnoy replied that on the ODP website the meeting summary and PowerPoint slide presentations will be posted, along with relevant program announcements. If there is support for having another type of statement related to this portfolio analysis, we will do that. This is a collaborative effort and, if this helps the field to understand what NIH is doing, we’d be happy to do that and keep talking about it.
What about posting information on PA as a blog, to give others who look at data in the community a place to have an ongoing discussion? A participant replied that if there were a blog limited to NIH staff, it would be very interesting, given the different definitions of PA that people at NIH have now.
One of the participants said that $340 million for PA research is only 1 percent of the total NIH budget, so he was surprised to hear that PA funding is comparable to that of tobacco research. Another attendee said that 20 percent of the NIH budget is spent on prevention, though this is not the public perception, which means maybe we need to get the word out more about PA and what we are doing on it.
PA Research Database Development: It was suggested to use the Human Nutrition Research Database (HNRIM) as a model and as a tool to garner support for a stronger voice for PA research.
Workshop Co-Chair CAPT Rick Troiano, Ph.D.
CAPT Troiano laid out the research priorities and next steps or future directions, based upon the discussions from the workshop. He outlined six main areas and several additional concerns.
PA Research Coordination:
First, he said, there is a need for a central NIH coordinating unit on PA, preferably within the NIH Office of the Director, and that this coordinating unit could ensure adequate and appropriate representation of PA experts for application reviews. This would mean defining exactly what constitutes a PA expert, and perhaps training the NIH Center for Scientific Review as to who is an expert.
Perhaps PA should form an interest group that meets informally every few months, as the tobacco area has [there is an Exercise Interest Group at NIH that meets regularly]. Or perhaps it should be at a higher level, similar to the Division of Nutrition Research Coordination group. That would require a committed leadership contact point from multiple ICs who could discuss the multiple small portfolios spread across the NIH. In addition, we need more workshops involving the extramural community to develop program announcements and implementation steps.
PA Research Priorities:
PA Research Partnerships and Collaborations:
There should be public-private partnerships for research support through the Foundation for NIH. Possible non-health sectors that could play a role include the U.S. Departments of Education, Defense, Transportation, Housing and Urban Development, and Interior/National Park Service; also the National Science Foundation.
PA Research Methods:
Among the overarching points was that research on measurement of PA should generate various instruments that are applicable in public health practice and/or a wide range of research applications. The new methods should support multimodal or self-report methods targeted to purpose, especially for large community or cohort studies. In addition, the new methods should support statistical approaches to PA data, especially with regard to distributions and variability. The same considerations should be applied to environmental measures.
In addition to these main take-away ideas, other issues that came up in the workshop include the importance of the following:
ODP Director David Murray, Ph.D.
In his closing remarks, Dr. Murray thanked the co-chairs, participants, and the planning team for their thoughtful input, presentations, and candid and constructive feedback. Dr. Murray reiterated his office’s commitment to advancing the NIH PA prevention research agenda and improving coordination and communication of NIH research activities and opportunities. The concept of a central coordinating group for PA research is something that comes under the ODP purview, and he expressed his intention to have a conversation with senior leadership about a more focused approach to PA and the related research issues that were discussed during the workshop.
Dr. Murray is interested and committed to supporting all stages of prevention research. Based on his personal experiences and observations, he added that too often, scientists embark on large, expensive trials prematurely—that is, before having the science to support new experiments. To that end, Dr. Murray plans to identify alternative mechanisms to explore early and late-stage activities. At the same time, he would like to promote trans-IC activities that engage multiple ICs during the initial phase funding announcement planning. Projects that are related to PA must be encouraged, and thus, Dr. Murray expressed his support for innovative PA research design, methodologies, and training opportunities for both extramural and intramural staff, as well as scientific review officers (SROs) who review PA research applications and funding requests. He envisions developing a full complement of PA research resources, including future workshops to address the key themes that emerged from this meeting.
The interest for promoting PA as a stand-alone outcome already has precedent at NCI, for instance. If it is properly presented, it might not meet with the resistance that one might anticipate.
As for dealing with Big Data, NIH is actively engaged in new initiatives to build and integrate extremely large biomedical research datasets and to develop new informatics tools for managing and analyzing these resources. A key goal of the Big Data effort at the NIH is to have all of the ICs work together to identify and achieve Big Data management and data analysis requirements.
He acknowledged that changes are needed to improve the NIH review system. ODP welcomes suggestions to improve the process. For instance, if the right people are not on the study sections, he implored the participants to be part of the solution and to serve on study sections. He said, “We can’t go in and tell them what to pay attention to,” but we could put together lists of candidate reviewers. That’s how to get better people reviewing.
When presenting portfolio analyses, NIH as a whole needs to devise a better system for examining its research portfolio. The Office of AIDS Research has a strategic plan with objectives and measures, and all grants are coded against their strategic plan objectives. ODP is in the process of developing its first strategic plan. Specific research portfolio strategic objectives and outcomes could be included in the plan as a means of improving accountability, communication, and reporting.
Dr. Murray is supportive of efforts to strengthen partnerships and collaborations. He supports working with professional organizations and other groups to share their experiences and research with the NIH. Professional groups that give a common message and a set of recommendations are very powerful.
In closing, Dr. Murray thanked all those involved in the workshop, those who planned and led it, the speakers and those around the table. Barry Portnoy announced that the slides and the meeting summary would be on the website, and he hoped all those involved would consider the workshop as a first step in an ongoing collaboration.