June 28 - 29, 2007 Advisory Committee Meeting
Overview of Systematic Literature Review Approach
Dr. Kohl opened with an overview of four areas that will be covered in his
presentation: the rational utilized in CDC's overall approach in gathering data,
methodology in abstracting data, review how data was inputted and demonstrate
the database live.
Due to the time constraints of producing Physical Activity Guidelines within
a 2-year window, compared to the 5-year cycle for Dietary Guidelines, there was
a need to make certain assumptions and decisions, before the formation of the
Advisory Committee, in order to prepare data that can be reviewed immediately.
Three goals guided the CDC: Develop and implement a feasible approach for
systematic review of physical activity and health literature for maximal benefit
to the Federal Advisory Committee, provide an initial literature database in
preparation for first Committee meeting and support the Committee throughout its
The actual literature review is organized in four phases. The first phase is
a developmental phase that began in October 2006 which included resource
gathering and planning. Phase A is the initial literature review conducted by
CDC and subsequently Phase B will be the reactive phase starting at the end of
this Committee meeting through the next meeting in December. During Phase B,
Committee members will review the data, comment and request additional data.
Phase C, December – February, is the final phase wrapping up work in preparation
for the Committee's report to HHS.
The framework for the literature review consists of six exposure types as
they relate to eight health outcome areas. The exposures include intensity,
frequency, duration, pattern, type and an estimate of caloric expenditure. The
outcomes include cardiovascular-respiratory health outcomes, metabolic health
outcomes, musculoskeletal health, cancers, functional health, mental health,
all-cause mortality, and adverse events and risks of physical activity.
Additionally, the groups were broken down by age, children and adolescents (ages
6 – 18), adults (ages 19 – 64) and older adults (ages 65 and over) and were
reviewed functionally by three teams consisting of Quality Control, an External
Science Advisory Group — providing input on constructing the searches, and
the Scientific Review Team aided by a contractor responsible for database
systems development and outside data abstractors.
While it may have been preferable to pull all literature starting from 1953
it was apparent that would not be possible so the search started 1 year prior to
the 1995 Surgeon General's Report and goes through 2006. Functionally, a fourth
age group emerged — Mixed Ages — to account for studies that have very large
data sets or account for studies that occur over very long periods of time.
Again, due to the proactive approach of this strategy the CDC team could not
wait for research questions to come from the Committee so four questions were
framed to allow the process to move forward:
- Is physical activity associated with an outcome?
- What is the evidence of the dose of physical activities associated with a
- What is the evidence that physical activity actually might increase the risk
of a certain outcome?
- What is the evidence in exposure of physical activity, other than 30 minutes
per day, associated with an outcome?
Additional research questions from the Committee are possible.
In order to manage a very large amount of data an 8-step process was devised to
manage the abstracting process:
Develop Search Strategy
A professional library scientist and an external scientific advisory group
assisted in the development of key strategies. Through this process the group
unearthed idiosyncrasies working with Medline and search terms such as key
words, exploded terms and mesh headings.
Execute Search Strategy
In early January the strategy was implemented to come up with reference lists.
The material that made up the universe of literature from 1995 – January 2007
Data inputted into literature reference manager systems and retrieved for
Abstracts were reviewed to determine if they were relevant and if they met
pre-determined exclusion criteria. Excluded abstracts were filed away while
other abstracts were flagged for retrieval and coding. If an abstract was not
relevant towards diagnosable disease endpoints, with some exceptions, were
excluded and or deferred to Phase B. All studies including body composition
were included along with risk and adverse events dealing only with variability
in physical activity exposure and functional health studies in older adults.
Studies dealing with acute effects of single exercise bout were excluded.
Abstracts that met the following criteria were deferred to Phase B.
- Risk factors or risk markers as outcomes
- Physiologic adaptations without health outcome
- Fetal outcomes
- Balance – except where appropriate for functional health in older adults
- Addictive behaviors
- Cognitive function in youth
The primary reasons for exclusion included 14% excluded due to
inappropriate format, 27% excluded because of no health outcome and 59%
excluded because there was something wrong with their physical activity
exposure data. Of 12,428 papers reviewed, 1,105 papers were abstracted.
1,105 papers were recorded on DVD and sent to the Committee.
Assignment to Abstracters
In order to properly enter and code data into the database 49 individuals
that were masters-trained, exercise-students or professionals with public
health expertise were recruited and trained.
Coding of Papers
As papers were coded a quality assurance program was implemented to insure
accuracy. To get a sense of the quality of work performed 14% of the papers
were selected to undergo "double coding."
The strengths of the initial database include its specific focus on
physical activity and its ability to focus on multiple outcomes. It is
standardized, systematic and very searchable. The database is dynamic meaning
additional choices can be made and additional papers can be added. It is also
very inclusive and not limited to one type of study. The database also links
to the original PDF and strong quality assurance methods were employed in the
vetting of abstracts.
At the conclusion of the presentation, members of the Committee were led
through a hands-on demonstration of the database led by Susan Carlson from CDC
and other CDC team members.