This is the author’s version of a work that was submitted/accepted for pub-
lication in the following source:
Tudor-Locke, Catrine, Washington, Tracy L., Ainsworth, Barbara E., &
Troiano, Richard P. (2009) Linking the American Time Use Survey (ATUS)
and the Compendium of Physical Activities : methods and rationale. Jour-
nal of Physical Activity and Health, 6(3), pp. 347-353.
This file was downloaded from: http://eprints.qut.edu.au/41172/
c ? Copyright 2009 Human Kinetics
Notice: Changes introduced as a result of publishing processes such as
copy-editing and formatting may not be reflected in this document. For a
definitive version of this work, please refer to the published source:
Journal of Physical Activity and Health
© Human Kinetics, Inc.
Linking the American Time Use Survey
(ATUS) and the Compendium of Physical
Activities: Methods and Rationale
Catrine Tudor-Locke,Tracy L.Washington,?
Barbara E. Ainsworth, and Richard P. Troiano?
Background: The 2003 Bureau of Labor Statistics American Time Use Survey
(ATUS) contains 438 distinct primary activity variables that can be analyzed with
regard to how time is spent by Americans. The Compendium of Physical Activities is
used to code physical activities derived from various surveys, logs, diaries, etc to
facilitate comparison of coded intensity levels across studies. Methods: This paper
describes the methods, challenges, and rationale for linking Compendium estimates
of physical activity intensity (METs, metabolic equivalents) with all activities reported
in the 2003 ATUS. Results: The assigned ATUS intensity levels are not intended to
compute the energy costs of physical activity in individuals. Instead, they are intended
to be used to identify time spent in activities broadly classifjed by type and intensity.
This function will complement public health surveillance systems and aid in policy
and health-promotion activities. For example, at least one of the future projects of this
process is the descriptive epidemiology of time spent in common physical activity
intensity categories. Conclusions: The process of metabolic coding of the ATUS by
linking it with the Compendium of Physical Activities can make important
contributions to our understanding of Americans’ time spent in health-related physical
Keywords: metabolic coding, surveillance, secondary analysis
The prevalence of physical activity behaviors and their associated health cor
relates have been observed in US public health surveillance systems for nearly 25
years.1,2 The purpose of surveillance is to provide a timely, ongoing system to col
lect, evaluate, and disseminate data about the trends and magnitude of health
behaviors and problems in populations and to provide a basis for intervention,
education, and evaluation of related promotion activities.3 Primary data-collection
Tudor-Locke is with the Walking Behavior Laboratory, Pennington Biomedical Research Center,
Baton Rouge, LA 70808. Washington and Ainsworth are with the Dept of Exercise and Wellness,
Arizona State University, Mesa, AZ 85212. Troiano is with the Division of Cancer Control and
Population Sciences, National Cancer Institute, Bethesda, MD 20852.
????? ? ?? ??????
Tudor-Locke et al
methods used in public health surveillance are telephone surveys, physician
reports, and clinical evaluations for selected risk factors and health conditions.1–3
Physical activity behaviors have traditionally been measured using question
naires1,2; however recently, accelerometers4 and time use data5 have become avail
able to provide more detailed assessments of the type, intensity, and duration of
physical activities performed throughout the day.
Time use data and methods have been extensively validated6 and are based on
an international history of social research.7 Time use surveys are typically designed
to collect a detailed time-defjned trace of activities over the course of 24 hours.
The American Time Use Survey (ATUS) is a nationally representative survey con
ducted by the Bureau of Labor Statistics (BLS) as a means of assigning value to
nonmarket productivity based on the kinds of activities engaged in and the time
spent doing them.8 The 2003 ATUS contains 438 distinct primary activity vari
ables as part of a classifjcation system modeled after the more contemporary
activity classifjcations used in Australian time use surveys.8 Beyond its original
purpose, however, the ATUS presents a unique and valuable opportunity to explore
Americans’ activity engagement from a public health perspective.
Specifjcally, a valuable aspect of the ATUS for surveillance research on phys
ical activity is inclusion of activities in all locations, circumstances, and contexts:
occupation, transportation, household, and leisure time. Some aspects of these
data have already been reported in a news release from the BLS9:
•? Employed adult women (18 years and older) spent about an hour more per
day than employed men doing household activities and caring for household
•? Adults in households without children spent about 1.4 hours more per day
engaged in leisure and sports activities than those with children.
•? Men were more likely than women to participate in sports on any given day,
19% versus 16%. Men also spent more time in sports activities on the days
they participated (2.0 versus 1.3 hours).
Although these time estimates for various categories of activity are valuable,
research to link physical activity to disease risk or prevention requires going
beyond estimates of time duration to estimates of the intensity of physical activi
ties and ultimately their associated metabolic or energy cost.
The ATUS collects data in suffjcient detail to allow for coding of primary
activities for activity intensity using available estimates of physical activity inten
sity. The National Cancer Institute within the National Institutes of Health saw
this opportunity to link the ATUS primary activities codes with metabolic equiva
lents (METs) cataloged in the Compendium of Physical Activities10,11 and com
missioned this effort to capitalize on available and future ATUS data. Linking
primary activities reported in the ATUS to estimates of physical activity intensity
will allow improved assessment of the extent of physical activity in the popula
tion, identifjcation of major sources of energy expenditure, and examination of
factors associated with differing degrees of energy expenditure, to name but three
opportunities available from this process. This article describes the methods, chal
lenges, and rationale for linking Compendium MET estimates of physical activity
intensity with all primary activities reported in the 2003 ATUS.
Time Use Metabolic Coding
Microdata from the 2003 ATUS were released in January 2005. Details about the
ATUS methods are available at http://www.bls.gov/tus/. Briefmy, the ATUS
response sample represents a subsample drawn from households that have previ
ously completed the Current Population Survey (CPS; http://www.bls.gov/cps),
a federal survey that provides the source of the nation’s unemployment rate,
among other vital statistics. Specifjcally for the ATUS, a single individual from
each selected household is interviewed by telephone once (on a single, preas
signed reporting day) about their personal time use over the previous 24-hour day
(anchored by 4:00 AM). Both weekdays and weekend days were considered, but
users are advised to use ATUS-constructed weights to ensure appropriate interpre
tation of time spent between these types of days. The actual interview is con
ducted (after obtaining verbal consent) using a Computer Assisted Telephone
Interviewing (CATI) system to standardize progress and prompting through a
combination of structured general background questions and conversational inter
viewing representing the designated recalled day. Responses about activities (and
their durations) are captured verbatim. The 2003 ATUS sample consisted of about
21,000 interviews. The ATUS is authorized by Title 13, United States Code sec
tion 8 (population statistics) and 9 (confjdentiality).
Interviewers are trained to use software to assign a 6-digit code to each ATUS
primary activity based on an organizational system that classifjes activities from
broad categories to more specifjc ones using hierarchical 2-digit tiers. ATUS vari
ables are named according to specifjed rule, and the fjrst letter T stands for time
and the second letter U indicates that the variable is unedited. Values of unedited
variables are generally produced by the CATI or Computer Assisted Personal
Interview (CAPI) instruments and are either collected or assigned during the
interview. The fjrst tier, indicated by the fjrst 2 digits of the 6-digit code (the actual
ATUS variable name is TUTIER1CODE, but for simplicity herein is called the
Major Category), represents 17 Major Categories of activities (eg, personal care;
household activities; sports, exercise, and recreation; and traveling). The second
tier is indicated by the second 2 digits (TUTIER2CODE herein called General
Category). For example, General Category activities under household activities
include housework; lawn, garden, and houseplants; and household management.
The third tier is indicated by the last 2 digits (TUTIER3CODE herein called Spe
cifjc Category). For example, Specifjc Category activities under housework
include interior cleaning; laundry; and storing interior household items, includ
ing food. The ATUS lexicon further provides extensive lists of activity examples
that would fall further under these Specifjc Categories, but there are no codes
assigned at this (fourth) level and, therefore, they do not constitute an actual ATUS
variable. Uncoded example activities included under the Specifjc Category of
interior cleaning include vacuuming, scrubbing, dusting, and emptying the ash
tray. Trained ATUS coders refer to these examples when assigning the 6-digit
primary activity codes that represent the full 3-tier system (but, to emphasize, this
fourth level is not coded).
Tudor-Locke et al
Compendium of Physical Activities
The Compendium of Physical Activities10,11 is a comprehensive list of 605 physi
cal activities used to code the type, purpose, and intensity of physical activities
performed in daily life. The Compendium was developed to facilitate comparison
of intensity levels across studies. The activities listed in the Compendium were
identifjed from various surveys, logs, diaries, and occupational task lists, and their
associated MET levels were obtained from existing charts and tables, published
research studies, and, for some activities, MET levels from similar activities pre
viously listed in the Compendium. The Compendium uses a 5-digit coding scheme
to categorize activities. The fjrst 2 digits represent the major purpose of the activ
ity or Major Heading, for example, self-care and home activities. There are 21
Major Headings for physical activity behaviors ranging from inactivity to sports
and exercise. The last 3 digits indicate a specifjc activity within each Major Head
ing. For example, under the Major Heading of home activities there are separate
3-digit codes for vacuuming, scrubbing fmoors, and light cleaning (including dust
ing, straightening up). Each 5-digit activity is associated with a 2- or 3-digit MET
intensity level. A MET is defjned as the activity metabolic rate divided by the rest
ing metabolic rate, with lying or sitting quietly classifjed as 1 MET. A 3-MET
activity requires 3 times the energy expenditure at rest. For example, scrubbing
fmoors is a 3.8-MET activity, vacuuming is a 3.5-MET activity, and light cleaning
is a 2.5-MET activity. Intensity categories are broadly interpreted as light (<3
METs), moderate (3–6 METs), and vigorous (>6 METs).12 It is possible to also
separate the light-intensity category into sleeping activities (<1 MET) and seden
tary/lying/sitting activities (≥1 and <3 METs) based on Compendium coding.
Linking the ATUS and the Compendium
The Compendium was used to assign MET values to the (fourth-level) example
activities as presented in the ATUS lexicon. We began by producing 17 Excel
spreadsheets representing each of the 17 Major Categories, their associated Gen
eral and Specifjc Categories, and ultimately example activities provided in the
ATUS lexicon of activities (ATUS now provides this lexicon in Excel format).
Compendium activity codes and MET values were then assigned to each example
activity independently by one of the researchers (TLW) who regularly verifjed and
clarifjed problematic activities as they emerged with the lead author (CT-L). An
example of linking a single ATUS primary activity variable with the associated
Compendium activity codes and MET value is shown in Table 1. The fjrst author
of the Compendium publications10,11 (and another researcher in this project, BEA)
then independently reviewed all activity codes and MET values to ensure confor
mance with the Compendium. Noted inconsistencies were rarely discrepant by 1
or more METs, and most discrepancies related only to the suggested linked 5-digit
Compendium activity code. Specifjcally, when coding varied, it was a result of
how the coders interpreted the setting and purpose of the ATUS activities and how
the coders assigned Compendium codes and MET intensities to ATUS activities
missing from the Compendium. Coding differences were resolved to provide an
appropriate MET intensity for the intended purpose of the activity. The greatest
challenge was in making sure the MET intensity for specifjc activities refmected
General Category Specific Category
Table 1 Example of Linking ATUS and Compendium Activity Variables
(and associated Lexicon example
Personal care (02)
Doing own hair
and grooming oneself
Abbreviations: ATUS, American Time Use Survey; METs, metabolic equivalents.
Tudor-Locke et al
the energy cost for specifjc activities listed in the ATUS. Inconsistencies were
discussed among authors, resolved by consensus, and corrected in the fjnal data
set. The process was repeated as necessary to identify and resolve problems.
As an example of this process, a notable inconsistency was identifjed under
caring for household members (ATUS Major Category = 03), caring for house
hold adults (General Category = 04), providing medical care to household adult
(Specifjc Category = 03). ATUS example activities under this collective 6-digit
code included giving household adult medicine and bandaging household adult.
Initially, the closest Compendium code considered was 05187 home activities/
elder care, disabled adult, only active periods, with a MET value of 4, which
appeared high considering the implied ATUS activity. Following discussion, we
arrived at a consensus that the ATUS example activities implied sitting or standing
and, therefore, were more aligned with the Compendium code 05185 home activi
ties, child care: sitting/kneeling, dressing, bathing, grooming, feeding, occasional
lifting of child—light effort, general (MET value of 2.5). This coding challenge
illuminated a limitation of the Compendium; additional detailed MET values are
required to more accurately code diverse activities related to elder care.
Once linked codes and MET values were fjnalized, aggregate intensity values
were computed for each of the ATUS third-tier Specifjc Category activities based
on a process of averaging the MET variables assigned to the underlying and asso
ciated example activities. Essentially, MET values were averaged over those
example activities categorized under shared 6-digit codes representing the fjnal
ATUS Specifjc Category activity MET value. Exceptions included treatment of
General or Specifjc Categories that ATUS assigned “99” as a 2-digit code. ATUS
uses these as not elsewhere classifjed (n.e.c.) indicators, that is, the example activ
ity was deemed to be representative of the relevant activity (ie, General or Specifjc
Category, depending where the 99 was indicated in the 6-digit series) but not else
where classifjed. In these cases, MET values were averaged over similar 2-digit
General Category variables (or, in some cases, the Major Category if the General
Category was also coded as 99). As a result of averaging the MET levels assigned
to individual activities (provided as examples within each Specifjc Category activ
ity), summary MET values for ATUS activities might differ exactly from similar
activities reported in the Compendium. An illustration of this process is presented
in Table 2.
Through this iterative process, challenges in assigning MET levels occurred.
It became evident early on that sports, exercise, and recreation (Major Category
= 13) was an ATUS amalgamation of both participating in (represented by Major
Category = 01; participating in sports, exercise, or recreation) and spectating at
(represented by General Category = 02; attending sporting/recreational events)
such activities. An aggregate MET value would ultimately underestimate the
intensity value of participation and overestimate spectating. We, therefore, deter
mined it was necessary to produce aggregated MET values for ATUS primary
activities under sports, exercise, and recreation separately by the associated Gen
The ATUS category sports, exercise, and recreation proved challenging to
code for additional reasons. Specifjcally, with regard to participating in sports,
exercise, or recreation (Major Category = 01), the ATUS lexicon combined exam
ple activities of active engagement in sports, exercise, or recreation and talking to
Table 2 Illustrative Example of Process of Imputed Summary MET Values
Summary MET value
Major Category 01 Personal Care
General Category 01 Sleeping
01 Specifjc Category Sleeping
0.92 (average of all
Specifjc Category =
01 MET values)
getting some shut-eye
02 Specifjc Category Sleeplessness
1.00 (average of all
Specifjc Category =
tossing and turning
02 MET values)
99 Specifjc Category Sleeping, n.e.c.
(no assigned value)
0.96 (average of all
General Category =
01 MET values)
Abbreviations: MET, metabolic equivalent; n.e.c., not elsewhere classifjed.
Tudor-Locke et al
involved individuals in 15 cases. For example, a Specifjc Category under this
structure is called doing aerobics. Lexicon example activities include step aero
bics (METs = 8.5), high-impact aerobics (METs = 7.0), low-impact aerobics
(METs = 5.0), and talking to an aerobics instructor (MET = 1.8). Strictly applied,
talking to an aerobics instructor would pull the average intensity down for this set
of example activities, and the summary MET value would underestimate the likely
intensity of doing aerobics. To illustrate, the summary MET value for the Specifjc
Category doing aerobics was 5.58 when talking to an aerobics instructor was
considered in the average and 6.83 when it was removed from the equation.
Because, as stated earlier, vigorous intensity is broadly interpreted as >6 METs,
censoring talking to an aerobics instructor was necessary to truly represent doing
aerobics as a vigorous activity. Following careful consideration of alternative
strategies (eg, weighting was determined not to be possible), the authors arrived at
a consensus that it was prudent, for the purposes of assigning an appropriate MET
value, to ignore those example activities that included talking to individuals under
the General Category participating in sports, exercise, or recreation.
We also quickly identifjed that traveling (Major Category = 17) did not pro
vide suffjcient detail about the mode of traveling to assign MET values without
linking it further to another ATUS variable, TEWHERE. The second letter E indi
cates that the variable has gone through an editing process, or consistency checks.
Values of edited variables are almost always equal to values of the corresponding
unedited variables. Data differ when a value is allocated or imputed by the pro
cessing system based on allocation rules specifjed in CPS or ATUS processing.
This variable corresponds to the interview question “Where were you while you
were [activity]?” For simplicity, it will be referred to as a Place/Transit variable
herein. When asked in the context of traveling, this Place/Transit variable can
provide important information necessary to assign an intensity variable. Specifj
cally, Place/Transit represents either place (eg, home/yard, workplace, grocery
store, etc) or in-transit (eg, walking, bicycling, bus, car, etc) indicators. In-transit
indicators within Place/Transit could be directly assigned a Compendium MET
value. After carefully considering the place indicators, it became evident that in
each case, the traveling would likely be walking; it was considered unlikely that
any individual would be traveling by any other mode in the home/yard, work
place, grocery store, etc.
A fjnal challenge in the process of linking the ATUS and the Compendium
was with regard to assigning MET values corresponding to occupation time. The
ATUS Major Category = 05, working and work related activities, does not seg
ment out individual employment tasks necessary to assign specifjc MET values;
almost all associated 6-digit codes are exactly the same and simply indicate that
the respondent was working. To emphasize, respondents are not asked to break
down the activities they did while at their main job. The explanation for the omis
sion of detailed occupational activities from the ATUS is due to the fact that the
BLS developed the detailed time use survey to identify nonoccupational activities
and to provide a monetary estimate of time spent in such activities.8 However,
because linked Occupational Category variables (based on 2002 Census Occupa
tion Codes located at http://www.bls.gov/tus/census02iocodes.pdf) from the
related CPS fjles were available for each respondent, coauthor BEA considered
the types of movements characterizing over 500 example occupations listed under
Time Use Metabolic Coding
these broader categories to assign underlying corresponding MET values using
the Tecumseh Occupational Physical Activity Questionnaire classifjcation sys
tem.13 This system assigns MET levels based on the considered body position (sit,
stand, walk, heavy labor) and intensity (light, moderate, vigorous). This approach
is a better refmection of the energy cost of occupational activities as done during a
usual workday. A single example is a fjrefjghter who might be active fjghting a fjre
for a couple of hours at a 10 to 12 MET level but does a light and moderate mix
of activities the rest of the shift. The eventual output of this separate process was
a single summary MET value (using a similar process as described earlier) linked
to each of the 22 Occupation Categories (Table 4, which can be viewed at http://
riskfactor.cancer.gov/tools/atus-met/). For example, the Occupational Category
building and grounds cleaning and maintenance has a mean MET summary value
of 3.58, based on a range of values (2.5 to 4.5) for all the associated occupations
listed. Other examples of the summary MET values assigned include management
(1.73); production (2.6); and farming, fjshing, and forestry (3.67). As with the
ATUS-coded activities, the METs assigned to the occupation codes might differ
from the Compendium of Physical Activities because of differences in the pro
cesses used to assign MET levels and averaging of METs across several activities
and specifjc movements that refmect an occupational activity category.
The ultimate product of this extensive and iterative process was a single
summary MET value linked to most of the ATUS 6-digit primary activity codes.
The exceptions were only the primary activities related to traveling, and these
could be imputed by further linkage to the ATUS Place/Transit variables (Table 5,
which can be viewed at http://riskfactor.cancer.gov/tools/atus-met/). Finally, it
is necessary to link ATUS occupational activities with CPS occupational codes
(Table 4). These summary MET values are available in Tables 3, 4, and 5 at http://
The process of metabolically coding the ATUS by linking it with the Compen
dium of Physical Activities can make important contributions to our understand
ing of Americans’ time spent in health-related physical activity. The most funda
mental public health application of these metabolically coded time use data is to
provide nationally representative estimates of Americans’ time spent in intensities
of physical activity deemed to be related to health. Time spent in at least moderate
intensity has readily apparent health ramifjcations since public health recommen
dations endorse such behavior for at least 30 minutes daily.14,15 Time spent in
lower-intensity physical activity might still be important in terms of energy bal
ance.16 A recent focus has emerged on sedentary time and its potentially detrimen
tal (and possibly independent) effects on health.17 Descriptive statistics captured
as part of the ATUS include sex, age, education level, employment status, marital
status, race/ethnicity, and whether the respondent has a child living in the house
hold, among others. Therefore, it is possible to examine time spent in health-
related physical activity across different groups. For example, it will be possible
to compare physical activity patterns of the following: men and women at all ages,
single women or men living in households with children and single women or
Tudor-Locke et al
men living in households without children, students and nonstudents, etc. At least
one of the future projects of the process described herein is the descriptive epide
miology of time spent in common physical activity intensity categories (ie, light,
moderate, and vigorous) based on some of these important demographic vari
ables. The availability of time use data from other countries would permit impor
tant international comparisons. In addition, because the ATUS is an ongoing
survey, it will be possible to track changes in time spent in different intensities of
In terms of surveillance strategies, the ATUS will provide important informa
tion to compare and verify existing surveillance data that have been based primar
ily on longer-term recalled physical activities (ie, over the past week, over the past
month). One challenge that will need to be addressed is the comparability of these
various scales of monitoring frame. Specifjcally, the ATUS is based on a single
day of recalled activity. It is likely that an individual who runs 3 days per week
will not be captured within an ATUS prevalence estimate of time in vigorous-
intensity activity if they are solicited on a nonrunning day; longer-term recall
would more likely capture this nondaily (ie, episodic) activity, resulting in dis
crepant prevalence estimates by instrument.
Another possible use of this linked product is the identifjcation of major
sources of energy expenditure because the ATUS captures primary activities rep
resenting the general domains of occupation, transportation, household, and lei
sure time. However, a simple review of the MET levels associated with the activi
ties shows that most are within the sedentary and light (1–2.9 METs) and moderate
(3–6) MET intensities. This is a result of averaging MET intensities from indi
vidual activities reported by ATUS participants into a composite database pre
sented for use. An example is in the Major Category participating in sports, exer
cise, or recreation where activities ranging from 3.0 to 11.0 METs were averaged
to yield a summary MET value of 5.87 for playing sports, n.e.c. (defjned earlier).
Thus, for respondents who engage in vigorous sports and exercise activities, the
energy cost of their daily activities might be underestimated. However, given the
lower prevalence of respondents engaging in sports and exercise activities in the
2003 ATUS database (19% of men and 16% of women9) and given that the overall
intended use for the ATUS data in aggregate is to describe how population groups
spend their time, the level of participant misclassifjcation should be low. Further
more, as stated earlier, there is growing interest in the contribution of lower-
intensity activities to energy balance.16
The ATUS has a number of limitations as a physical activity surveillance
instrument. First, only one activity at a time is captured. Primary activities are
collected and coded; respondents are not systematically asked about concurrent or
secondary activities. In fact, if the respondent volunteers two concurrent activi
ties, interviewers are trained to probe for the primary activity and only include it
in the coded data (so as to not collect time in activities that would sum over 24
hours).8 An implication of the ATUS design strategy is that any physical activities
that are performed concurrently but secondary to another activity would not be
counted and, therefore, would be underestimated. For example, it is possible that
a respondent’s report of riding a bicycle after school with a child (ATUS 6-digit
code 030105, MET summary value 5.0) could have been otherwise coded as hear
ing about their child’s day (ATUS 6-digit code 030106, MET summary value 1.5).
Time Use Metabolic Coding
Unless a more systematic approach to collecting secondary activities is imple
mented, it appears that this loss is the unfortunate trade-off for a nonreactive
survey that does not specifjcally probe physical activity behaviors.
Second, because the ATUS did not obtain detailed information about occupa
tional behaviors, the database has limited application in identifying total physical
activity behaviors, which include those engaged in during work. Although the
entire process of linking the ATUS and Compendium was based on imputation,
we acknowledge that the resulting summary MET values associated with the 22
Occupation Categories is most crude and should be used and interpreted only with
Third, the ATUS obtains detailed data about daily activities of randomly
selected adults in the United States for a single 24-hour period. Although the data
are collected on an ongoing basis throughout the year, accounting for seasonal
differences in physical activity, the data cannot be used to characterize habitual
physical activity behaviors of individuals or selected population groups. Instead,
the ATUS is most effective for characterizing patterns and trends in population
groups for the types and durations of activities and the behavior settings for the
activities (eg, at home or in the car; alone or with families or others).
Fourth, the MET levels assigned to ATUS activities might ultimately differ
for similar activities in the Compendium of Physical Activities. This a result of the
process of averaging MET levels for several example activities listed in the lexi
con, a document originally intended as a guide to identify and code Specifjc Cat
egory activities. The fjnal product of this process is a summary MET variable, to
emphasize, one that might not be exactly the same as comparable Compendium
Finally, despite the process described herein undertaken to metabolically
code ATUS primary physical activities, we are not able to compute their actual
energy cost, at least in terms of kilocalories expended. The ATUS did not collect
respondents’ self-reported height or weight data (necessary for such imputation)
before 2006. More importantly, however, the aggregated MET summary values
are too crude for such specifjc estimates at the level of the individual. Although
additional ATUS coding at the level of the lexicon examples would allow more
detailed estimates of time spent in the various intensities of physical activity, it
still can only be interpreted on a population basis. It is possible, however, that a
descriptive epidemiology paper based on the currently metabolically coded data
will be able to compute MET-min, a compilation of activity duration and intensity.
Starting with 2006, height and weight data were collected by ATUS. Regardless
of whether the end output is expressed as MET-min or kilocalories, however, both
will be crude, refmecting the limitations of the ATUS activity coding.
ATUS microdata are now available for 2003 to 2007 at http://www.bls.gov/
tus/. There have been some minor changes to the lexicons relevant to the process
of linking ATUS primary activities and the Compendium MET values; however,
overall summary MET values have been little affected (eg, <±0.74 METs when
comparing similar categories between 2003 and 2005). Beyond changes to lexi
con examples, 2004 additions to the Place/Transit variable described earlier
included bank, gym/health club, and post offjce. In terms of assigning MET values,
these additions are considered only when linked to the Major Category traveling.
As before, we determined that traveling undertaken within any of these place
Tudor-Locke et al
indicators would likely be walking. Finally, in 2005, a change in digits assigned to
the fjrst-tier code for traveling had no effect on assignment of MET values.
In summary, the major strength of the ATUS is in obtaining a cross-sectional
view of how American adults spend 24 hours of their lives. This provides research
ers and practitioners with detailed information about patterns of transportation,
home, family-care, leisure-time, and social-interaction activities that can be used
for hypothesis generation, program development, and comparison of current data
sets. The process of metabolic coding of the ATUS by linking it with the Compen
dium of Physical Activities can make important contributions to our understand
ing of Americans’ time spent in health-related physical activity. Specifjcally, these
data will provide us with a clearer picture of the proportion of Americans who
engage in any moderate or vigorous physical activity on any given day. Further
more, we will be able to generate more detailed estimates of time spent in these
specifjc (and other, eg, sedentary) activities on a populations basis, but also among
the subsample who report any engagement whatsoever in said activities. At least
one of the future projects of this process is the descriptive epidemiology of time
spent in common physical activity intensity classifjcations including lying, sitting,
and light, moderate, and vigorous intensities.
The authors declare that they have no competing interests.
Funding for this project was provided through a contract with the National Cancer Institute,
National Institutes of Health.
1. Centers for Disease Control and Prevention. Prevalence of sedentary lifestyle—
Behavioral Risk Factor Surveillance System, United States, 1991. MMWR Morb
Mortal Wkly Rep. 1993;42(29):576–579.
2. DiPietro L, Williamson DF, Caspersen CJ, Eaker E. The descriptive epidemiology of
selected physical activities and body weight among adults trying to lose weight: the
Behavioral Risk Factor Surveillance System survey, 1989. Int J Obes. 1993;17:69–
3. Halperin WE. The role of surveillance in the hierarchy of prevention. Am J Ind Med.
4. Troiano RP. Accelerometer-measured physical activity prevalence in NHANES 2003–
2004. Med Sci Sports Exerc. 2006;38(5):40.
5. Tudor-Locke C, Bittman M, Merom D, Bauman A. Patterns of walking for trans
port and exercise: a novel application of time use data. Int J Behav Nutr Phys Act.
6. Pentland WE, Harvey AS, Lawton MP, McColl MA. Time Use Research in the Social
Sciences. New York, NY: Kluwer Academic/Plenum Publishers; 1999.
7. Szalai A, ed. The Use of Time: Daily Activities of Urban and Suburban Populations in
Twelve Countries. Paris: Mounton; 1972.
8. Shelley KJ. Developing the American Time Use Survey activity classifjcation system.
Mon Labor Rev. 2005;128(6):3–15.
Time Use Metabolic Coding Download full-text
9. Bureau of Labor Statistics, US Department of Labor. Time-Use Survey, fjrst results
announced by BLS. http://www.bls.gov/news.release/archives/atus_09142004.pdf.
Published September 14, 2004. Accessed February 15, 2005.
10. Ainsworth BE, Haskell WL, Leon AS, et al. Compendium of physical activities:
classifjcation of energy costs of human physical activities. Med Sci Sports Exerc.
11. Ainsworth BE, Haskell WL, Whitt MC, et al. Compendium of physical activities: an
update of activity codes and MET intensities. Med Sci Sports Exerc. 2000;32(suppl
12. Pate RR. Physical activity and health: dose-response issues. ?Res Q Exerc Sport.
13. Ainsworth B, Richardson M, Jacobs DR Jr, Leon A, Sternfeld B. Evaluation of occu
pational activity surveys. J Clin Epidemiol. 1999;52:219–227.
14. Centers for Disease Control and Prevention. Participation in high school physical edu-
cation—United States, 1991–2003. MMWR Morb Mortal Wkly Rep. 2004;53(36):844–
15. Le Masurier GC, Sidman CL, Corbin CB. Accumulating 10,000 steps: does this meet
current physical activity guidelines? Res Q Exerc Sport. 2003;74(4):389–394.
16. Levine JA, Vander Weg MW, Hill JO, Klesges RC. Non-exercise activity thermogen
esis: the crouching tiger hidden dragon of societal weight gain. Arterioscler Thromb
Vasc Biol. 2006;26(4):729–736.
17. Hamilton MT, Hamilton DG, Zderic TW. Role of low energy expenditure and sitting
in obesity, metabolic syndrome, type 2 diabetes, and cardiovascular disease. Diabetes.