Advertisement

Association Between Sedentary Work and BMI in a U.S. National Longitudinal Survey

  • Tin-chi Lin
    Correspondence
    Address correspondence to: Tin-chi Lin, PhD, Center for Injury Epidemiology, Liberty Mutual Research Institute for Safety, 71 Frankland Rd, Hopkinton MA 01748
    Affiliations
    Center for Injury Epidemiology, Liberty Mutual Research Institute for Safety, Hopkinton, Massachusetts

    Environmental and Occupational Medicine and Epidemiology Program, Department of Environmental Health, Harvard T.H. Chan School of Public Health, Boston, Massachusetts
    Search for articles by this author
  • Theodore K. Courtney
    Affiliations
    Center for Injury Epidemiology, Liberty Mutual Research Institute for Safety, Hopkinton, Massachusetts

    Environmental and Occupational Medicine and Epidemiology Program, Department of Environmental Health, Harvard T.H. Chan School of Public Health, Boston, Massachusetts
    Search for articles by this author
  • David A. Lombardi
    Affiliations
    Center for Injury Epidemiology, Liberty Mutual Research Institute for Safety, Hopkinton, Massachusetts

    Environmental and Occupational Medicine and Epidemiology Program, Department of Environmental Health, Harvard T.H. Chan School of Public Health, Boston, Massachusetts
    Search for articles by this author
  • Santosh K. Verma
    Affiliations
    Center for Injury Epidemiology, Liberty Mutual Research Institute for Safety, Hopkinton, Massachusetts

    Department of Family Medicine and Community Health, University of Massachusetts Medical School, Worcester, Massachusetts
    Search for articles by this author
Open AccessPublished:October 01, 2015DOI:https://doi.org/10.1016/j.amepre.2015.07.024

      Introduction

      Technological advancements have made life and work more sedentary, and long hours of sitting are known to be associated with many health concerns. Several studies have reported an association between prolonged sitting time at work and weight gain, but the results are inconsistent. This study examined the relationship between sitting time at work and BMI using data from a large prospective cohort of U.S. men and women from 2002 to 2010. Initial analyses were performed in 2013, with additional analyses in 2014 and 2015.

      Methods

      The sample size at the base year (2002) was 5,285 and the age range 38–45 years. The outcome, BMI, was based on self-reported measures of height and weight. Estimates of workplace sitting time were linked from an external database (Occupational Information Network), and the occupation-wide rating for sitting time was linked to survey participants by occupation. Fixed-effects models controlling for time-invariant effects of all time-invariant characteristics were employed to examine the association, controlling for age, education, work hours, and hours of vigorous and light/moderate physical activities.

      Results

      Longer sitting time was significantly associated with higher BMI for the overall sample (β = 0.054; p<0.05) and men (β = 0.086; p<0.01). For women, the association was not statistically significant.

      Conclusions

      The findings provide further support for initiatives to reduce workplace sitting time as a means of reducing the risk of weight gain and related health conditions.

      Introduction

      Society has benefited from many technological advancements, but people’s body weight may have increased as a side effect of technological progress.
      • Lakdawalla D.
      • Philipson T.
      The growth of obesity and technological change.
      • Cutler D.
      • Glaeser E.
      • Shapiro J.
      Why have Americans become more obese?.
      Work has become less strenuous because of innovations in technology and production process,
      • Dempsey P.G.
      • Mathiassen S.E.
      On the evolution of task-based analysis of manual materials handling, and its applicability in contemporary ergonomics.
      • Straker L.
      • Mathiassen S.E.
      Increased physical work loads in modern work — a necessity for better health and performance?.
      and studies have shown that at the population-level energy expenditure related to work has declined steadily over recent decades.
      • Church T.S.
      • Thomas D.M.
      • Tudor-Locke C.
      • et al.
      Trends over 5 decades in U.S. occupation-related physical activity and their associations with obesity.
      • Ng S.W.
      • Popkin B.
      Time use and physical activity: a shift away from movement across the globe.
      As life in developed economies has become more sedentary and people’s weight has increased concomitantly,
      • Church T.S.
      • Thomas D.M.
      • Tudor-Locke C.
      • et al.
      Trends over 5 decades in U.S. occupation-related physical activity and their associations with obesity.
      it raises questions as to whether increasing sedentariness is a contributing cause to obesity trends.
      The relationship between obesity and certain sedentary behaviors, such as watching TV, has been analyzed in the literature.
      • Ching P.
      • Willett W.C.
      • Rimm E.B.
      • et al.
      Activity level and risk of overweight in male health professionals.
      • Hu F.B.
      • Li T.Y.
      • Colditz G.A.
      • et al.
      Television watching and other sedentary behaviors in relation to risk of obesity and type 2 diabetes mellitus in women.
      • Proper K.I.
      • Singh A.S.
      • Van Mechelen W.
      • et al.
      Sedentary behaviors and health outcomes among adults: a systematic review of prospective studies.
      • Thorp A.A.
      • Owen N.
      • Neuhaus M.
      • et al.
      Sedentary behaviors and subsequent health outcomes in adults: a systematic review of longitudinal studies, 1996-2011.
      However, most of the previous research has been limited to leisure-time behaviors. Work is a major domain of life for many adults, and lack of information on sedentary behaviors in the workplace limits the understanding of potential causes of the obesity problem.
      Whether sedentariness at work leads to weight gain has received some attention.
      • Church T.S.
      • Thomas D.M.
      • Tudor-Locke C.
      • et al.
      Trends over 5 decades in U.S. occupation-related physical activity and their associations with obesity.
      • Brown W.
      • Miller Y.
      • Miller R.
      Sitting time and work patterns as indicators of overweight and obesity in Australian adults.
      • Chau J.Y.
      • van der Ploeg H.P.
      • Merom D.
      • et al.
      Cross-sectional associations between occupational and leisure-time sitting, physical activity and obesity in working adults.
      • Choi B.
      • Schnall P.L.
      • Yang H.
      • et al.
      Sedentary work, low physical job demand, and obesity in U.S. workers.
      • Ishizaki M.
      • Morikawa Y.
      • Nakagawa H.
      • et al.
      The influence of work characteristics on body mass index and waist to hip ratio in Japanese employees.
      • King G.A.
      • Fitzhugh E.
      • Bassett Jr, D.
      • et al.
      Relationship of leisure-time physical activity and occupational activity to the prevalence of obesity.
      • Mummery W.K.
      • Schofield G.M.
      • Steele R.
      • et al.
      Occupational sitting time and overweight and obesity in Australian workers.
      However, as review articles
      • Proper K.I.
      • Singh A.S.
      • Van Mechelen W.
      • et al.
      Sedentary behaviors and health outcomes among adults: a systematic review of prospective studies.
      • Thorp A.A.
      • Owen N.
      • Neuhaus M.
      • et al.
      Sedentary behaviors and subsequent health outcomes in adults: a systematic review of longitudinal studies, 1996-2011.
      • Van Uffelen J.G.
      • Wong J.
      • Chau J.Y.
      • et al.
      Occupational sitting and health risks: a systematic review.
      • Hu F.
      Obesity Epidemiology.
      noted, most previous research has used a cross-sectional design that cannot ensure temporal precedence, and a recent review paper
      • Van Uffelen J.G.
      • Wong J.
      • Chau J.Y.
      • et al.
      Occupational sitting and health risks: a systematic review.
      found that nearly half of the cross-sectional studies did not find a relationship. The associations reported by cross-sectional studies may include both causal effect of prolonged sitting and selection bias; obese people tend to increase sedentary behaviors and potentially may pre-select themselves into more sedentary jobs.
      A few prospective studies have investigated this topic,
      • Hu F.B.
      • Li T.Y.
      • Colditz G.A.
      • et al.
      Television watching and other sedentary behaviors in relation to risk of obesity and type 2 diabetes mellitus in women.
      • Wilsgaard T.
      • Jacobsen B.K.
      • Arnesen E.
      Determining lifestyle correlates of body mass index using multilevel analyses: the Tromsø Study, 1979-2001.
      • Andersen U.O.
      • Jensen G.
      The CCHS Group
      Decreasing population blood pressure is not mediated by changes in habitual physical activity. Results from 15 years of follow-up.
      • Bak H.
      • Petersen L.
      Sørensen TIA. Physical activity in relation to development and maintenance of obesity in men with and without juvenile onset obesity.
      • Graff-Iversen S.
      • Selmer R.
      • Sørensen M.
      • Skurtveit S.
      Occupational physical activity, overweight, and mortality: a follow-up study of 47,405 Norwegian women and men.
      • Pulsford R.M.
      • Stamatakis E.
      • Britton A.R.
      • et al.
      Sitting behavior and obesity: evidence from the Whitehall II study.
      but their results were inconsistent. Four of these studies
      • Wilsgaard T.
      • Jacobsen B.K.
      • Arnesen E.
      Determining lifestyle correlates of body mass index using multilevel analyses: the Tromsø Study, 1979-2001.
      • Andersen U.O.
      • Jensen G.
      The CCHS Group
      Decreasing population blood pressure is not mediated by changes in habitual physical activity. Results from 15 years of follow-up.
      • Bak H.
      • Petersen L.
      Sørensen TIA. Physical activity in relation to development and maintenance of obesity in men with and without juvenile onset obesity.
      • Pulsford R.M.
      • Stamatakis E.
      • Britton A.R.
      • et al.
      Sitting behavior and obesity: evidence from the Whitehall II study.
      did not find a prospective association, one study
      • Hu F.B.
      • Li T.Y.
      • Colditz G.A.
      • et al.
      Television watching and other sedentary behaviors in relation to risk of obesity and type 2 diabetes mellitus in women.
      found a positive relationship among female nurses but another
      • Graff-Iversen S.
      • Selmer R.
      • Sørensen M.
      • Skurtveit S.
      Occupational physical activity, overweight, and mortality: a follow-up study of 47,405 Norwegian women and men.
      showed that occupational physical activities were correlated with overweight (only among its female participants), indicating a negative association. These studies, though innovative, were not without limitations. One
      • Hu F.B.
      • Li T.Y.
      • Colditz G.A.
      • et al.
      Television watching and other sedentary behaviors in relation to risk of obesity and type 2 diabetes mellitus in women.
      used samples from a single occupational group (nurses) and the other
      • Pulsford R.M.
      • Stamatakis E.
      • Britton A.R.
      • et al.
      Sitting behavior and obesity: evidence from the Whitehall II study.
      contained government workers only; it is unclear whether their results are generalizable to workers of other occupations or the broader workforce. Another
      • Wilsgaard T.
      • Jacobsen B.K.
      • Arnesen E.
      Determining lifestyle correlates of body mass index using multilevel analyses: the Tromsø Study, 1979-2001.
      was based on a cohort living in a municipality well above the Arctic circle, so the lifestyle of the residents might not be comparable to those living in the continental U.S. or other temperate regions. Several other studies from Nordic countries
      • Andersen U.O.
      • Jensen G.
      The CCHS Group
      Decreasing population blood pressure is not mediated by changes in habitual physical activity. Results from 15 years of follow-up.
      • Bak H.
      • Petersen L.
      Sørensen TIA. Physical activity in relation to development and maintenance of obesity in men with and without juvenile onset obesity.
      • Graff-Iversen S.
      • Selmer R.
      • Sørensen M.
      • Skurtveit S.
      Occupational physical activity, overweight, and mortality: a follow-up study of 47,405 Norwegian women and men.
      used samples consisting of workers of varying occupations. Nonetheless, the data may be dated; none of these studies had a follow-up later than 1995, and the results may not be reflective of the circumstances faced by the contemporary U.S. workforce.
      The relationship between sedentary work and weight gain warrants further analysis with a more representative sample, a study design that ensures temporal precedence of exposure, robust statistical methods, and data that are more contemporary. Sedentary behavior refers to activities that do not increase energy expenditure substantially above the resting level.
      • Van Uffelen J.G.
      • Wong J.
      • Chau J.Y.
      • et al.
      Occupational sitting and health risks: a systematic review.
      • Owen N.
      • Healy G.N.
      • Matthews C.E.
      • et al.
      Too much sitting: the population-health science of sedentary behavior.
      As sitting appears to be the most common workplace sedentary behavior, it was hypothesized that longer sitting time at work would contribute to weight gain.

      Methods

      Study Sample

      This study’s primary data source was the National Longitudinal Survey of Youth 1979 (NLSY79), an ongoing prospective study that began in 1979 with a nationally representative sample of 12,686 U.S. men and women then aged 14–22 years. Although a primary focus of the survey is labor force behavior, the survey’s content is considerably broader, including obesity
      • Bhattacharya J.
      • Bundorf M.K.
      The incidence of the healthcare costs of obesity.
      and occupational injury.
      • Berdahl T.A.
      Racial/ethnic and gender differences in individual workplace injury risk trajectories: 1988-1998.
      • Lin T.
      • Verma S.K.
      • Courtney T.K.
      Does obesity contribute to non-fatal occupational injury? Evidence from the National Longitudinal Survey of Youth.
      The interview was conducted biannually after 1991. Detailed information on sampling and data collection is published elsewhere.
      • Lin T.
      • Verma S.K.
      • Courtney T.K.
      Does obesity contribute to non-fatal occupational injury? Evidence from the National Longitudinal Survey of Youth.

      Miller S. NLSY79 user’s guide: a guide to the 1979-2000 National Longitudinal sSrvey of Youth data. www.bls.gov/nls/79guide/2001/nls79g0.pdf. Washington, DC: US Department of Labor; 2001

      The NLSY data were analyzed from 2002 to 2010, the latest publicly available data at the initiation of the study, and 2002 was selected as the starting year because of the substantial differences between the occupational codes used by NLSY79 prior to 2002 and the exposure measure from another source (see Measures section). From 1991 onward, the number of NLSY79 respondents eligible for interview was actually 9,964.

      National Longitudinal Survey of Youth. Retention and Reasons for Noninterview. www.nlsinfo.org/content/cohorts/nlsy79/intro-to-the-sample/retention-reasons-noninterview. Accessed August 7, 2015.

      Figure 1 details the selection process for the participants at the 2002 base year. The data were an unbalanced panel. Those who had missing values in the base year were included in later years if they had valid data in those years. The number of observations in 2002, 2004, 2006, 2008, and 2010 was 5,285, 5,003, 5,062, 5,029, and 4,717, respectively.
      Despite >30 years of follow-up, the NLSY79 has a relatively low degree of sample attrition and has maintained a good response rate. Retention rate, defined as the percentage of base year respondents remaining eligible who were interviewed in a given survey year, was 78.1% in 2002 and 75.3% in 2010; most sample attrition was due to elimination of subsamples by design in the program.

      National Longitudinal Survey of Youth. Retention and Reasons for Noninterview. www.nlsinfo.org/content/cohorts/nlsy79/intro-to-the-sample/retention-reasons-noninterview. Accessed August 7, 2015.

      Response rate, defined as the percentage of respondents remaining eligible and alive who were interviewed in a given survey year, was 80.3% in 2002 and 80.6% in 2010.

      National Longitudinal Survey of Youth. Retention and Reasons for Noninterview. www.nlsinfo.org/content/cohorts/nlsy79/intro-to-the-sample/retention-reasons-noninterview. Accessed August 7, 2015.

      Measures

      The outcome was BMI, based on self-reported height and weight from NLSY79 participants. Between 2002 and 2010, a total of 268 observations reporting a BMI <14 or >50 kg/m2 were considered implausible and were set to missing. Sensitivity analyses indicated that setting different cut-off points for implausible values did not influence the study results.
      The primary explanatory variable was “time spent sitting at work,” extracted from the Occupational Information Network (O*NET) and then linked to the main NLSY data by occupation. Using a data source external to (and then linking it back to) the NLSY to evaluate exposure was necessary, as no U.S. national longitudinal survey documents both participants’ body weight and time spent sitting at work.
      The O*NET is a database developed for the U.S. Department of Labor that comprehensively describes occupations and is provided to the public and private sectors at no cost.

      O*NET Resource Center. About O*NET. www.onetcenter.org/overview.html. Accessed August 7, 2015.

      It collects a wide range of jobs characteristics such as abilities, skills, and personal attributes required for a job, as well as activities performed in a job and working conditions. Data are collected for >800 different jobs through national surveys or by occupational experts.

      U.S. Department of Labor. O*NET data collection program. Office of Management and Budget clearance package. Supporting statement Part A: Justification. 2012. www.onetcenter.org/dl_files/omb2011/Supporting_StatementA.pdf. Accessed August 7, 2015.

      The program is designed for career planning and workforce developments, but is also used by government agencies for administrative purposes.
      • Hilton M.L.
      • Tippins N.T.
      A Database for a Changing Economy: Review of the Occupational Information Network (O* NET).
      Version 15.1 of the O*NET database,

      O*NET Resource Center. O*NET® database releases archive. www.onetcenter.org/db_releases.html. Accessed August 7, 2015.

      based on data collected and updated gradually between 2002 and 2010,

      National Center for O*NET Development. Data dictionary: O*NET 15.1 database. 2011. www.onetcenter.org/dl_files/DataDictionary15_1.pdf. Accessed August 7, 2015.

      was used for this study.
      Below is the O*NET question (extracted from Question 34 of the Work Context questionaire

      O*NET Resource Center. Appendix A: Questionnaires. 2012. www.onetcenter.org/dl_files/omb2011/AppendixA.pdf. Accessed August 7, 2015.

      ) used for the analysis:
      How much time in your current job do you spend sitting?
      There were five possible responses: 1 (never), 2 (less than half of the time), 3 (about half of the time), 4 (more than half of the time), and 5 (continuously or almost continuously). The data value from the database is an average of individual ratings (not actual hours spent sitting) sampled from an occupation; Table 1 lists the ratings for the most and least sedentary jobs.
      Table 1The Most and Least Sedentary Jobs, Occupational Information Network (O*NET 15.1)
      RankingOccupationHow much time do you spend sitting (min=1, max=5)?
      1Telephone operators4.98
      2Insurance underwriters4.92
      3Tax preparers4.91
      4Telemarketers4.90
      5Statisticians4.87
      6Computer programmers4.86
      7Atmospheric and space scientists4.85
      8Technical writers4.81
      9Editors4.81
      10Budget analysts4.79
      –1Manufactured building and mobile home installers1.09
      –2Bakers1.12
      –3Tire builders1.15
      –4Drywall installers, ceiling tile installers, and tapers1.18
      –5Brickmasons, blockmasons, and stonemasons1.18
      –6Pressers, textile, garment, and related materials1.20
      –7Cabinetmakers and bench carpenters1.21
      –8Maids and housekeeping cleaners1.26
      –9Textile knitting and weaving machine setters, operators, and tenders1.26
      –10Bartenders1.27
      The O*NET data value (a constant for a given occupation) was assigned to NLSY participants in each survey year by occupation. O*NET and the NLSY used different occupational taxonomies, and a crosswalk

      Bureau of Labor Statistics. 2012 national employment matrix/SOC to CPS crosswalk (XLS). www.bls.gov/emp/classifications-crosswalks/NEM_OccCode_CPS_Crosswalk.xls.

      was used to convert the occupational codes of one to the other. To ensure the temporal precedence of the exposure, the O*NET measurement was matched to the job(s) that the respondent held 6 months prior to the interview. The job information was extracted from NLSY79’s employment data. Then, the repeated measurement for the exposure (6 months prior to the interview) was used to predict BMI in each wave from 2002 to 2010. The choice of 6 months was made to allow enough time for observing weight change.
      Each analysis controlled for participants’ age (centered at 40 years), education (in years), weekly work hours of all jobs, and hours of vigorous and light/moderate physical activity (PA) per week based on self-reported frequency and duration of the activities, respectively. Further analysis suggested that a non-negligible proportion of the PA measures had exceedingly high data values. To reduce the impact of extreme values, an upper limit for the data values (90th percentile of the original distributions) was placed for both PA measures. For all observations whose original data values were greater than the threshold, their data values were replaced with this upper limit. After the primary analyses, sensitivity analyses were performed by conducting the analysis for the upper limit set at the 80th, 85th, and 95th percentiles, respectively.

      Statistical Analysis

      Fixed-effects longitudinal models
      • Allison P.D.
      Fixed Effects Regression Models.
      were used to examine the association between sitting time and BMI, because they can control for the time-invariant effects of all time-invariant factors (e.g., ethnicity).
      • Allison P.D.
      Fixed Effects Regression Models.
      • Houle J.N.
      • Light M.T.
      The home foreclosure crisis and rising suicide rates, 2005 to 2010.
      Additional analysis was performed using random-effects models to investigate the potential interaction effect between gender and sitting time on BMI. Because men and women differ in labor market activities
      • Stigler G.J.
      Information in the labor market.
      and body metabolism,
      • Blaak E.
      Gender differences in fat metabolism.
      the fixed-effect model was run first for the overall sample, and then for women and men respectively. All data analyses were performed using Stata, version 13; the command “xtreg, fe” was used to run fixed-effects models, and a p-value of <0.05 was considered statistically significant. Initial analyses were performed in 2013, with additional analyses in 2014 and 2015. The New England IRB determined that this study was exempt from review.

      Results

      Table 2 presents characteristics of the participants in the base year, 2002. The demographic profile of these 5,285 participants was similar to the U.S. population of the same age range in 2002,

      CDC WONDER. Bridged-Race Population Estimates 1990-2012 Request. wonder.cdc.gov/Bridged-Race-v2012.HTML. Accessed August 7, 2015.

      except that the working data set had a higher proportion of men than the national average (53.3% vs 50.4%), presumably because employment history was required for a participant to be included in the analyses. On average, at the baseline the participants were aged 41.5 years, had completed 13.7 years of education, and performed vigorous and light/moderate PA 3.0 and 4.7 hours a week, respectively. They spent 41.1 hours working for all jobs, and the average rating of workplace sitting time was 3.0 (about half of the time), both evaluated at 6 months prior to the 2002 interview. The body weight of the cohort increased with time: The average BMI increased from 27.68 kg/m2 in 2002 to 28.45 kg/m2 in 2010, equivalent of a weight gain of 4.9 pounds (2.28 kg) for a 5 foot, 7 inch–tall person (172.7 cm).
      Table 2Characteristics of NLSY79 Respondents in Working Dataset in Base Year, 2002 (N=5,285)
      VariableMean (weighted)SDMinMax
      Female46.7%01
      Age41.52.283845
      Black13.4%01
      Hispanic6.5%01
      Education (in years)13.72.49020
      BMI27.75.4215.549.9
      Ratings of workplace sitting time, 6 months prior to interview3.01.1305.0
      Work hours (all jobs), 6 months prior to interview41.114.110105
      Hours of vigorous physical activities per week3.04.34014
      Hours of light/moderate physical activities per week4.76.81021
      NLSY79, National Longitudinal Survey of Youth 1979.
      Table 3 presents coefficient estimates for the relationship between sitting time at work and BMI from fixed-effects models; longer sitting time was significantly associated with higher BMI (β=0.054, p<0.05) for the overall sample. Thus, if one’s sitting time at work were to change from never (1 rating) to continuously or almost continuously (5 rating), BMI would increase by 0.216. But the results differed substantially by gender. For men, the association was statistically significant (β=0.086, p<0.01); for women, the coefficient of sitting time was not statistically different from zero. Gender differences in the association were consistent when the upper limit for the PA measures was set at different levels (Appendix Table 1, available online); the estimated association was significant among men, but no significant associations were found among women.
      Table 3Fixed-Effects Regression Assessing the Association Between Workplace Sitting Time and BMI, National Longitudinal Survey of Youth 1979 (NLSY79) in 2002–2010
      All (N=25,096)Male (n=12,625)Female (n=12,471)
      Coef.SECoef.SECoef.SE
      Age (centered at 40 years)0.099***0.0050.090***0.0060.109***0.008
      Education (in years)0.0020.049–0.0240.0750.0030.067
      Workplace sitting time0.054*0.0250.086**0.0320.0250.038
      Work hours, all jobs0.0020.0010.0030.0020.0010.002
      Hours of vigorous physical activities–0.0090.005–0.0020.005–0.024*0.009
      Hours of light/moderate physical activities–0.008*0.003–0.009*0.004–0.0070.006
      Constant27.755***0.66528.257***1.00527.599***0.917
      Note: Boldface indicates statistical significance (*p<0.05; **p<0.01; ***p<0.001).
      Coef, coefficient.
      Consistent with the fixed-effects model, the random-effects model (Model 1 of Appendix Table 2, available online) showed that the association between workplace sitting time and BMI was significant for the overall sample (β=0.056, p<0.05). Model 2 indicated that gender and sitting time interacted in their relationship with BMI (β=–0.112, p<0.05), and the association was significant for men (β=0.120, p<0.001) but not for women (β=0.007, p=0.82).

      Discussion

      Data from NLSY79 and O*NET were used to examine the association between sedentary work and BMI. The results showed that longer sitting time at work was significantly associated with higher BMI for the overall sample and for men. For women, the association was not statistically significant. Gender differences in coefficient estimates have been reported by at least two previous studies.
      • Mummery W.K.
      • Schofield G.M.
      • Steele R.
      • et al.
      Occupational sitting time and overweight and obesity in Australian workers.
      • Graff-Iversen S.
      • Selmer R.
      • Sørensen M.
      • Skurtveit S.
      Occupational physical activity, overweight, and mortality: a follow-up study of 47,405 Norwegian women and men.
      The findings corroborated a 2003 Australian cross-sectional study,
      • Mummery W.K.
      • Schofield G.M.
      • Steele R.
      • et al.
      Occupational sitting time and overweight and obesity in Australian workers.
      but contrasted with a Norwegian cohort study
      • Graff-Iversen S.
      • Selmer R.
      • Sørensen M.
      • Skurtveit S.
      Occupational physical activity, overweight, and mortality: a follow-up study of 47,405 Norwegian women and men.
      that did not find a prospective association among men. Neither of these studies provided detailed explanations for the observed gender differences.
      For the present study, the difference in coefficient estimates is not believed to suggest distinct biological mechanisms between men and women, but rather potential residual confounding and selection bias. The NLSY79 did not comprehensively document participants’ PA and did not record participants’ diet. These unmeasured factors may differ between men and women, which may have contributed to the observed gender difference. Selection bias might confound the coefficient estimates for women. People tend to spend less time sitting at work in general if their jobs have higher physical demands, and three previous studies
      • Wilsgaard T.
      • Jacobsen B.K.
      • Arnesen E.
      Determining lifestyle correlates of body mass index using multilevel analyses: the Tromsø Study, 1979-2001.
      • Graff-Iversen S.
      • Selmer R.
      • Sørensen M.
      • Skurtveit S.
      Occupational physical activity, overweight, and mortality: a follow-up study of 47,405 Norwegian women and men.
      • Lakdawalla D.N.
      • Philipson T.
      Labor supply and weight.
      showed that women with higher BMIs were more frequently employed in jobs requiring a higher level of physical demand. This indicates a (counterintuitive) relationship between higher BMI and physically strenuous jobs among female workers. A strong selection bias at work would attenuate the coefficient estimates, as the coefficient would be influenced by the causal effect of long sitting time on BMI (a positive association) and selection bias (a negative association).

      Limitations

      One limitation of this study was the fact that O*NET data value for a given occupation was a constant, preventing the authors from assessing within-occupational changes in sitting time even though people may have increasingly spent more time sitting at work without changing their occupation. Misclassification of sitting time is possible, especially for occupations with a highly diversified job content
      • Cifuentes M.
      • Boyer J.
      • Lombardi D.A.
      • et al.
      Use of O* NET as a job exposure matrix: a literature review.
      ; for example, the rating for “physicians” may be too high for those working in emergency rooms but low for those who spend most of their time in research. Still, O*NET is one of the few data sources that have comprehensively evaluated job characteristics using national samples. Despite its limitations, many studies
      • Dembe A.E.
      • Yao X.
      • Wickizer T.M.
      • et al.
      Using O* NET to estimate the association between work exposures and chronic diseases.
      • Fisher G.G.
      • Stachowski A.
      • Infurna F.J.
      • et al.
      Mental work demands, retirement, and longitudinal trajectories of cognitive functioning.
      • Meyer J.D.
      • Mutambudzi M.
      Association of occupational trajectories with alcohol use disorders in a longitudinal national Survey.
      • Monfort S.S.
      • Howe G.W.
      • Nettles C.D.
      • et al.
      A longitudinal examination of re-employment quality on internalizing symptoms and job-search intentions.
      • Mutambudzi M.
      • Meyer J.D.
      Construction of early and midlife work trajectories in women and their association with birth weight.
      have used it to assess the association between job content and a variety of health conditions.
      In addition, the study has other limitations. The NLSY did not comprehensively record participants’ energy expenditures of varying intensity (e.g., time spent driving or sedentary behaviors off work). The observed association might be attributable to those unmeasured factors and the possibility of residual confounding cannot be ruled out. A non-negligible proportion of the PA measures had data values too high to be realistic. However, imposing a threshold for these extreme values or excluding them from the analyses yielded similar results. BMI is not a perfect measure for body adiposity, and factors other than increasing body fat could contribute to heightened BMI, such as developing muscle mass.
      • Heymsfield S.B.
      • Scherzer R.
      • Pietrobelli A.
      • et al.
      Body mass index as a phenotypic expression of adiposity: quantitative contribution of muscularity in a population-based sample.
      Missing data in the sample survey could also introduce bias. However, missing data owing to death (3.7% and 6% of eligible participants in 2002 and 2010, respectively), participant non-response (18.2% of all eligible observations), and item non-response (16.8% of all observations in the working subpopulation) was relatively small, and therefore it does not appear that the resulting loss of information would change the results dramatically.
      The study has several strengths. First, the analyses are based on a large and ongoing national cohort survey; the sample resembles the national demographic profile and consists of diverse occupational groups, instead of a single occupation.
      • Hu F.B.
      • Li T.Y.
      • Colditz G.A.
      • et al.
      Television watching and other sedentary behaviors in relation to risk of obesity and type 2 diabetes mellitus in women.
      • Pulsford R.M.
      • Stamatakis E.
      • Britton A.R.
      • et al.
      Sitting behavior and obesity: evidence from the Whitehall II study.
      The NLSY79 cohort has a relatively high participation rate and low sample attrition. The longitudinal nature and employment history information of NLSY79 allowed the authors to construct an exposure that ensured temporal precedence. Lastly, fixed-effects models have particular strengths in reducing omitted-variable bias, and robustness checks yielded similar results.

      Conclusions

      American adults spend more than half of their waking time in sedentary behaviors.
      • Matthews C.E.
      • Chen K.Y.
      • Freedson P.S.
      • et al.
      Amount of time spent in sedentary behaviors in the United States, 2003-2004.
      Prolonged sitting time is linked with mortality
      • Patel A.V.
      • Bernstein L.
      • Deka A.
      • et al.
      Leisure time spent sitting in relation to total mortality in a prospective cohort of U.S. adults.
      and many chronic diseases,
      • Hu F.B.
      • Li T.Y.
      • Colditz G.A.
      • et al.
      Television watching and other sedentary behaviors in relation to risk of obesity and type 2 diabetes mellitus in women.
      • Gerhardsson M.
      • Norell S.E.
      • Kiviranta H.
      • et al.
      Sedentary jobs and colon cancer.
      independent of PA level. The results of the study showed that longer sitting time at work was significantly associated with higher BMI for the overall sample and for men. The results provide support for measures to reduce the duration of sitting in the workplace. Future studies should measure sitting time with objective measures
      • Chau J.Y.
      • Van der Ploeg H.P.
      • Dunn S.
      • et al.
      Validity of the occupational sitting and physical activity questionnaire.
      and investigate the underlying biological mechanisms.
      • Shoham N.
      • Girshovitz P.
      • Katzengold R.
      • et al.
      Adipocyte stiffness increases with accumulation of lipid droplets.
      Given the deleterious consequences of sedentary behavior and its potential contribution to weight gain, the public health community should promote interventions to reduce time spent sitting both during and outside of work.

      Acknowledgments

      The authors appreciate the thoughtful comments of Dr. Elyssa Besen and Mr. Raymond McGorry.
      No financial disclosures were reported by the authors of this paper.

      Appendix. Supplementary Materials

      References

        • Lakdawalla D.
        • Philipson T.
        The growth of obesity and technological change.
        Econ Hum Biol. 2009; 7: 283-293https://doi.org/10.1016/j.ehb.2009.08.001
        • Cutler D.
        • Glaeser E.
        • Shapiro J.
        Why have Americans become more obese?.
        J Econ Perspect. 2003; 17: 93-118https://doi.org/10.1257/089533003769204371
        • Dempsey P.G.
        • Mathiassen S.E.
        On the evolution of task-based analysis of manual materials handling, and its applicability in contemporary ergonomics.
        Appl Ergonomics. 2006; 37: 33-43https://doi.org/10.1016/j.apergo.2004.11.004
        • Straker L.
        • Mathiassen S.E.
        Increased physical work loads in modern work — a necessity for better health and performance?.
        Ergonomics. 2009; 52: 1215-1225https://doi.org/10.1080/00140130903039101
        • Church T.S.
        • Thomas D.M.
        • Tudor-Locke C.
        • et al.
        Trends over 5 decades in U.S. occupation-related physical activity and their associations with obesity.
        PloS One. 2011; 6: e19657https://doi.org/10.1371/journal.pone.0019657
        • Ng S.W.
        • Popkin B.
        Time use and physical activity: a shift away from movement across the globe.
        Obes Rev. 2012; 13: 659-680https://doi.org/10.1111/j.1467-789X.2011.00982.x
        • Ching P.
        • Willett W.C.
        • Rimm E.B.
        • et al.
        Activity level and risk of overweight in male health professionals.
        Am J Public Health. 1996; 86: 25-30https://doi.org/10.2105/AJPH.86.1.25
        • Hu F.B.
        • Li T.Y.
        • Colditz G.A.
        • et al.
        Television watching and other sedentary behaviors in relation to risk of obesity and type 2 diabetes mellitus in women.
        JAMA. 2003; 289: 1785-1791https://doi.org/10.1001/jama.289.14.1785
        • Proper K.I.
        • Singh A.S.
        • Van Mechelen W.
        • et al.
        Sedentary behaviors and health outcomes among adults: a systematic review of prospective studies.
        Am J Prev Med. 2011; 40: 174-182https://doi.org/10.1016/j.amepre.2010.10.015
        • Thorp A.A.
        • Owen N.
        • Neuhaus M.
        • et al.
        Sedentary behaviors and subsequent health outcomes in adults: a systematic review of longitudinal studies, 1996-2011.
        Am J Prev Med. 2011; 41: 207-215https://doi.org/10.1016/j.amepre.2011.05.004
        • Brown W.
        • Miller Y.
        • Miller R.
        Sitting time and work patterns as indicators of overweight and obesity in Australian adults.
        Int J Obesity. 2003; 27: 1340-1346https://doi.org/10.1038/sj.ijo.0802426
        • Chau J.Y.
        • van der Ploeg H.P.
        • Merom D.
        • et al.
        Cross-sectional associations between occupational and leisure-time sitting, physical activity and obesity in working adults.
        Prev Med. 2012; 54: 195-200https://doi.org/10.1016/j.ypmed.2011.12.020
        • Choi B.
        • Schnall P.L.
        • Yang H.
        • et al.
        Sedentary work, low physical job demand, and obesity in U.S. workers.
        Am J Ind Med. 2010; 53: 1088-1101https://doi.org/10.1002/ajim.20886
        • Ishizaki M.
        • Morikawa Y.
        • Nakagawa H.
        • et al.
        The influence of work characteristics on body mass index and waist to hip ratio in Japanese employees.
        Ind Health. 2004; 42: 41-49https://doi.org/10.2486/indhealth.42.41
        • King G.A.
        • Fitzhugh E.
        • Bassett Jr, D.
        • et al.
        Relationship of leisure-time physical activity and occupational activity to the prevalence of obesity.
        Int J Obesity. 2001; 25: 606-612https://doi.org/10.1038/sj.ijo.0801583
        • Mummery W.K.
        • Schofield G.M.
        • Steele R.
        • et al.
        Occupational sitting time and overweight and obesity in Australian workers.
        Am J Prev Med. 2005; 29: 91-97https://doi.org/10.1016/j.amepre.2005.04.003
        • Van Uffelen J.G.
        • Wong J.
        • Chau J.Y.
        • et al.
        Occupational sitting and health risks: a systematic review.
        Am J Prev Med. 2010; 39: 379-388https://doi.org/10.1016/j.amepre.2010.05.024
        • Hu F.
        Obesity Epidemiology.
        Oxford University Press, New York2008https://doi.org/10.1093/acprof:oso/9780195312911.001.0001
        • Wilsgaard T.
        • Jacobsen B.K.
        • Arnesen E.
        Determining lifestyle correlates of body mass index using multilevel analyses: the Tromsø Study, 1979-2001.
        Am J Epidemiol. 2005; 162: 1179-1188https://doi.org/10.1093/aje/kwi328
        • Andersen U.O.
        • Jensen G.
        • The CCHS Group
        Decreasing population blood pressure is not mediated by changes in habitual physical activity. Results from 15 years of follow-up.
        Blood Press. 2007; 16: 28-35https://doi.org/10.1080/08037050701190022
        • Bak H.
        • Petersen L.
        Sørensen TIA. Physical activity in relation to development and maintenance of obesity in men with and without juvenile onset obesity.
        Int J Obesity. 2003; 28: 99-104https://doi.org/10.1038/sj.ijo.0802525
        • Graff-Iversen S.
        • Selmer R.
        • Sørensen M.
        • Skurtveit S.
        Occupational physical activity, overweight, and mortality: a follow-up study of 47,405 Norwegian women and men.
        Res Q Exercise Sport. 2007; 78: 151-161https://doi.org/10.1080/02701367.2007.10599412
        • Pulsford R.M.
        • Stamatakis E.
        • Britton A.R.
        • et al.
        Sitting behavior and obesity: evidence from the Whitehall II study.
        Am J Prev Med. 2013; 44: 132-138https://doi.org/10.1016/j.amepre.2012.10.009
        • Owen N.
        • Healy G.N.
        • Matthews C.E.
        • et al.
        Too much sitting: the population-health science of sedentary behavior.
        Exerc Sport Sci Rev. 2010; 38: 105https://doi.org/10.1097/JES.0b013e3181e373a2
        • Bhattacharya J.
        • Bundorf M.K.
        The incidence of the healthcare costs of obesity.
        J Health Econ. 2009; 28: 649-658https://doi.org/10.1016/j.jhealeco.2009.02.009
        • Berdahl T.A.
        Racial/ethnic and gender differences in individual workplace injury risk trajectories: 1988-1998.
        Am J Public Health. 2008; 98: 2258-2263https://doi.org/10.2105/AJPH.2006.103135
        • Lin T.
        • Verma S.K.
        • Courtney T.K.
        Does obesity contribute to non-fatal occupational injury? Evidence from the National Longitudinal Survey of Youth.
        Scand J Work Environ Health. 2013; 39: 268-275https://doi.org/10.5271/sjweh.3347
      1. Miller S. NLSY79 user’s guide: a guide to the 1979-2000 National Longitudinal sSrvey of Youth data. www.bls.gov/nls/79guide/2001/nls79g0.pdf. Washington, DC: US Department of Labor; 2001

      2. National Longitudinal Survey of Youth. Retention and Reasons for Noninterview. www.nlsinfo.org/content/cohorts/nlsy79/intro-to-the-sample/retention-reasons-noninterview. Accessed August 7, 2015.

      3. O*NET Resource Center. About O*NET. www.onetcenter.org/overview.html. Accessed August 7, 2015.

      4. U.S. Department of Labor. O*NET data collection program. Office of Management and Budget clearance package. Supporting statement Part A: Justification. 2012. www.onetcenter.org/dl_files/omb2011/Supporting_StatementA.pdf. Accessed August 7, 2015.

        • Hilton M.L.
        • Tippins N.T.
        A Database for a Changing Economy: Review of the Occupational Information Network (O* NET).
        National Academies Press, Washington, DC2010
      5. O*NET Resource Center. O*NET® database releases archive. www.onetcenter.org/db_releases.html. Accessed August 7, 2015.

      6. National Center for O*NET Development. Data dictionary: O*NET 15.1 database. 2011. www.onetcenter.org/dl_files/DataDictionary15_1.pdf. Accessed August 7, 2015.

      7. O*NET Resource Center. Appendix A: Questionnaires. 2012. www.onetcenter.org/dl_files/omb2011/AppendixA.pdf. Accessed August 7, 2015.

      8. Bureau of Labor Statistics. 2012 national employment matrix/SOC to CPS crosswalk (XLS). www.bls.gov/emp/classifications-crosswalks/NEM_OccCode_CPS_Crosswalk.xls.

        • Allison P.D.
        Fixed Effects Regression Models.
        Sage Publications, Thousand Oaks, CA2009
        • Houle J.N.
        • Light M.T.
        The home foreclosure crisis and rising suicide rates, 2005 to 2010.
        Am J Public Health. 2014; 104: 1073-1079https://doi.org/10.2105/AJPH.2013.301774
        • Stigler G.J.
        Information in the labor market.
        J Polit Econ. 1962; 70: 94-105https://doi.org/10.1086/258727
        • Blaak E.
        Gender differences in fat metabolism.
        Curr Opin Clin Nutr. 2001; 4: 499-502https://doi.org/10.1097/00075197-200111000-00006
      9. CDC WONDER. Bridged-Race Population Estimates 1990-2012 Request. wonder.cdc.gov/Bridged-Race-v2012.HTML. Accessed August 7, 2015.

        • Lakdawalla D.N.
        • Philipson T.
        Labor supply and weight.
        J Human Resour. 2007; 42: 85-116
        • Cifuentes M.
        • Boyer J.
        • Lombardi D.A.
        • et al.
        Use of O* NET as a job exposure matrix: a literature review.
        Am J Ind Med. 2010; 53: 898-914https://doi.org/10.1002/ajim.20846
        • Dembe A.E.
        • Yao X.
        • Wickizer T.M.
        • et al.
        Using O* NET to estimate the association between work exposures and chronic diseases.
        Am J Ind Med. 2014; 57: 1022-1031https://doi.org/10.1002/ajim.22342
        • Fisher G.G.
        • Stachowski A.
        • Infurna F.J.
        • et al.
        Mental work demands, retirement, and longitudinal trajectories of cognitive functioning.
        J Occup Health Psychol. 2014; 19: 231https://doi.org/10.1037/a0035724
        • Meyer J.D.
        • Mutambudzi M.
        Association of occupational trajectories with alcohol use disorders in a longitudinal national Survey.
        J Occup Environ Med. 2014; 56: 700-707https://doi.org/10.1097/JOM.0000000000000184
        • Monfort S.S.
        • Howe G.W.
        • Nettles C.D.
        • et al.
        A longitudinal examination of re-employment quality on internalizing symptoms and job-search intentions.
        J Occup Health Psychol. 2015; 20: 50-61https://doi.org/10.1037/a0037753
        • Mutambudzi M.
        • Meyer J.D.
        Construction of early and midlife work trajectories in women and their association with birth weight.
        Am J Public Health. 2004; 104: S58-S64https://doi.org/10.2105/AJPH.2013.301401
        • Heymsfield S.B.
        • Scherzer R.
        • Pietrobelli A.
        • et al.
        Body mass index as a phenotypic expression of adiposity: quantitative contribution of muscularity in a population-based sample.
        Int J Obesity. 2009; 33: 1363-1373https://doi.org/10.1038/ijo.2009.184
        • Matthews C.E.
        • Chen K.Y.
        • Freedson P.S.
        • et al.
        Amount of time spent in sedentary behaviors in the United States, 2003-2004.
        Am J Epidemiol. 2008; 167: 875-881https://doi.org/10.1093/aje/kwm390
        • Patel A.V.
        • Bernstein L.
        • Deka A.
        • et al.
        Leisure time spent sitting in relation to total mortality in a prospective cohort of U.S. adults.
        Am J Epidemiol. 2010; 172: 419-429https://doi.org/10.1093/aje/kwq155
        • Gerhardsson M.
        • Norell S.E.
        • Kiviranta H.
        • et al.
        Sedentary jobs and colon cancer.
        Am J Epidemiol. 1986; 123: 775-780
        • Chau J.Y.
        • Van der Ploeg H.P.
        • Dunn S.
        • et al.
        Validity of the occupational sitting and physical activity questionnaire.
        Med Sci Sports Exerc. 2012; 44: 118-125https://doi.org/10.1249/MSS.0b013e3182251060
        • Shoham N.
        • Girshovitz P.
        • Katzengold R.
        • et al.
        Adipocyte stiffness increases with accumulation of lipid droplets.
        Biophys J. 2014; 106: 1421-1431https://doi.org/10.1016/j.bpj.2014.01.045