The PhenX Toolkit was developed as a source of high-priority, low-burden, and evidence-based measures for use by a wide variety of researchers.7,8
Although investigators are likely to be knowledgeable in selecting and implementing measures in their own content area, they may quickly become overwhelmed by the myriad of measures available in other fields. Further, once a measure is identified, the appropriate implementation of that measure must also be understood.
The PhenX Toolkit provides investigators with measures, background information, and use guidance that allow the inclusion of measures that could enhance their studies. The Toolkit provides standard measures related to complex diseases, traits, and environmental exposures. Use of PhenX measures not only may facilitate combining results from different studies but also could enable secondary analyses to expand studies beyond their primary research focus. The Toolkit can be accessed by a researcher who is planning a new study or looking to add measures to an ongoing study, with particular emphasis on those researchers seeking to add measures outside of their primary research focus.
The PhenX Steering Committee identified 21 research domains—Alcohol, Tobacco and other Substances, Anthropometrics, Cancer, Cardiovascular, Demographics, Diabetes, Environmental Exposures, Gastrointestinal, Infectious Disease and Immunity, Neurology, Nutrition and Dietary Supplements, Ocular, Oral Health, Physical Activity and Physical Fitness, Psychiatric, Psychosocial, Reproductive Health, Respiratory, Skin, Bone, Muscle, and Joint, Social Environments, Speech and Hearing—and provided guidance to a Working Group (WG) of content experts in each domain during the measure selection process.
In the Toolkit, researchers can browse by domains or measures or search using keywords. The selected PhenX measures are saved in a cart from which the researcher can generate a report with information about the measures and protocols of interest. In addition, data collection worksheets can be generated for each measure to facilitate data collection. Data Element Dictionaries are available and Toolkit users can also use the Collections search strategy. Collections have been developed to facilitate identification of measures related to a specific topic. For example, under the heading of Risk Factors, Behavior and Attitudes, a Collection will contain a grouping for Health Promotion measures, including the exercise measures selected by the PA/PF WG. The Health Status Collection also includes measures of PA/PF in combination with measures from several other domains.
Available in the Toolkit are a glossary of terms, frequently asked questions, a basic guide document, links to supplemental information about the other measures considered by the WGs, and additional resources. Toolkit users can provide direct feedback to the Toolkit project team through a link on the website.
The Physical Activity and Physical Fitness PhenX Domain
As is true in any research endeavor, instruments are designed to assess particular measures in a specific context. It is important to note that for complex domains such as PA/PF, it is especially important to match the assessment tool to the study objectives and population. A questionnaire designed to assess population levels of physical activity is not likely to do well at measuring change in activity level for an individual. The PhenX PA/PF WG provided measures and protocols for clearly delineated content areas and subpopulations.
The PhenX PA/PF Domain was designed to meet the needs of researchers interested in measures of PA/PF as outcomes, predictors, or covariates. The Federal Advisory Committee for the 2008 Physical Activity Guidelines noted that its attempts to synthesize the evidence relating physical activity to health outcomes were hampered by the variety of questionnaires used to assess physical activity and different approaches to data analysis and presentation.
Working Group Process
The PA/PF WG followed a predefined 6- to 8-month consensus process to come to agreement on a set of high-priority, low-burden, and evidence-based measures with a measure defined broadly as a standardized way of capturing data on certain characteristics of a study subject. Measures include exposures, clinical assessments, and quantitative or qualitative traits. PA/PF WG members were selected based on their experience in the development, evaluation, and use of PA/PF measures in studies investigating the health benefits and risks of physical activity, sedentary behavior; and physical fitness in youth (aged 5–17 years), adults (aged 18–65 years), and older adults (aged >65 years) across various races and ethnicities. The WG consisted of two co-chairs, four content experts, a steering committee liaison, and a WG manager. PhenX also has Liaisons appointed from many of the NIH Institutes and Centers who are invited to participate in any of the relevant WGs’ deliberations and meetings. In addition, the WG members consulted with numerous domain experts regarding the availability and selection of specific protocols.
With initial guidance from the PhenX steering committee regarding potential areas of inclusion to be considered, WG members completed their review and recommendations of measures and protocols for the Toolkit between September 2009 and February 2010. The initial list of possible measures underwent substantial discussion and revision over a period of approximately 4 months. These discussions took place during several conference calls and one in-person meeting involving WG members, steering committee members, NIH liaisons, National Human Genome Research Institute (NHGRI) staff, and PhenX staff. Further discussions took place via e-mail and on a web portal, providing a secure space for WG member interaction.
After defining the general scope of the PA/PF domains, a listing of possible measures was developed and discussed. From this broad list of measures, a target of no more than 25 measures with associated protocols was set for review by the PhenX steering committee and eventually submitted to the scientific community for their review and comment. The PA/PF WG was limited to selecting no more than 15 measures.
Desired Measure and Protocol Characteristics
It was a requirement of the PhenX steering committee that a majority of protocols needed to have low subject and investigator burden and implementation costs consistent with data collection in large population studies. No more than two measures were to be considered high burden. The goal was to select high-quality and well-established protocols recommended by domain experts. Also, the protocols needed to have utility for investigators who are not PA/PF domain experts and be relevant for at least the next few years.
During the initial WG deliberations, extended discussions were held regarding which PA/PF measures should be included in the Toolkit. Priority was given to measures demonstrated to be related to major health outcomes and for which well-established protocols could be identified. Protocols needed to be in the public domain or available at a low cost from the source and include implementation instructions published in sufficient detail that replication was possible. Several measures were considered as desirable for the Toolkit, but existing protocols did not meet selection criteria, especially the need for broad validation, demonstrated utility, or reproducibility in the target population. As a result, the WG carefully examined the strengths and limitations of the protocols proposed for each measure, with particular emphasis on those such as lifetime physical activity for adults and older adults, and any PA/PF measure for young children (aged <6 years).
It was decided that both PA objective measurement protocols and self-report protocols be included. The WG agreed to refer to measures using devices such as accelerometers or heart rate monitors as “objective” because of the common use of this term in the PA/PF literature. The WG also agreed that “objective” does not automatically mean better than “subjective” self-report measures: the information obtained is simply different. Also, for some PA/PF measures, separate protocols would be included for youth, adults, and older adults.
Physical fitness measures selected by the WG included cardiorespiratory fitness (three protocols—laboratory, field, and nontest estimate); integrated fitness; muscle strength; and physical functioning (objective and subjective protocols). For physical activity measures, the WG selected total physical activity (three protocols—screener, comprehensive, objective); walking/ambulation (objective); and sitting/sedentary (self-report). Protocols also were included for measures of physical activity self-efficacy; neighborhood environments (as a determinant of physical activity behavior); and physical activity readiness (safety screening questionnaire). Measures and protocols were selected by the WG based on a consensus process.
Although the recommendations for PA/PF measures and protocols were made by the WG, from early in the process, outreach to and consensus gathering from numerous experts in the field took place. Individual WG members made contact primarily through e-mail with scientists who had substantial experience developing or implementing various protocols for evaluating a specific measure. In some cases, these scientists directed WG members to useful documents, especially regarding the nature of protocol reliability and specifics of protocol implementation.
Near the end of the process, PhenX staff posted 14 measures on the Internet via the PhenX Toolkit for review and comment by the scientific community. Each PhenX domain used this outreach process to obtain feedback from domain experts. In addition to general notices being sent to the scientific community, e-mails were sent to individuals and groups identified by the WG members who would provide the most useful feedback on the measures.
Researchers had 2 weeks to respond with their thoughts on the value of the measures and if they should be included in a core set of genome-wide association measures that could be used by researchers in genetic and epidemiologic research fields. WG members also contacted their colleagues from other institutions and organizations for suggestions and received helpful feedback from a total of 34 researchers. The WG members reviewed and carefully considered this feedback as they chose the final set of measures and protocols.