Study design and participants
This study is part of a large longitudinal cohort study of six medium-sized manufacturing companies in Connecticut, designed to assess changes over time in an aging workforce, focusing in particular on musculoskeletal, psychosocial, and work-related variables. The full study protocol was approved by University of Connecticut Health Center’s Institutional Review Board. Eligibility criteria for study sites were: medium company size; broad age distribution centered on late 5th and 6th decades, and a workforce engaged in skilled light-manufacturing with high degrees of repetition. Four of the organizations had labor unions. Details of site identification and study procedures at each company are available in a prior publication .
The current study used data on BMI and BFP collected from physical performance testing performed at two time points, time 1 and time 2, approximately 36 months apart (average time between collections 33 months), and demographic, health-related, and work-related factors collected from paper-and-pencil surveys conducted at time 1. During the workday, following informed consent, surveys were distributed and collected by members of the research team. Participants were given a small financial incentive for completing the survey or physical testing measurements. All employees at selected sites were considered eligible and invited to participate in the study; no exclusion criteria were specified. Employees of all job classifications participated (e.g., production, sales, administrative, managerial staff).
BMI and BFP
BMI was calculated based on objective measurements of each participant’s height and weight. A vertical anthropometer was used to measure height in centimeters. Participants were barefoot for measurement. Weight was determined with the use of a standard balance scale with the balance was calibrated to zero. Values for height were recorded to the nearest tenth of a centimeter, and weight was recorded to the nearest quarter kilogram .
BFP was estimated through bioelectrical impedance [25, 26]. A Bioelectrical Body Composition Analyzer (Quantum X, RJL Systems, Clinton Township, MI) captured reactance and resistance for conversion to proportional body fat content. All testing was performed in accordance with the manufacturer’s instruction: shoes, socks, and jewelry or clothes with metal appurtenances were removed, and subjects were supine for 5 min prior to testing.
Demographic, health-related, and work-related factors
Demographic variables included age, gender, race (White/European Descent, Black/African American/African, American Indian/Alaska Native, Asian /Asian American. Other), marital status (married or live with partner, widowed, divorced or separated, single or never married), education level (less than high school, high school graduate or GED, some college, 2 or 4 year college degree, graduate degree),family income ($10,000–24,999, $25,000–49,999, $50,000–74,999, $75,000–99,999, More than $100,000), childcare responsibility, and elder care responsibility. Childcare responsibility was measured with one question: “How much responsibility do you personally have for any children under 18 in your household?” Respondents checking that they had primary or shared responsibility were defined as having a high level of childcare responsibility, while those who indicated that they had no children under 18 at home or that another adult had primary responsibility were defined as having a low level of childcare responsibility. Elder care responsibility was measured with one question: “How many adults age 65 and older depend on you in any way to help them due to disability or chronic illness? “Respondents checking 1 or greater were defined as providing elder care, while those responding “zero” were defined as not providing elder care.
We examined eight health-related factors including hours of sleep, depressive symptoms, leisure time physical activity, musculoskeletal pain, weight perception, and work-life balance. Hours of sleep was assessed with a single-item measure from the Pittsburgh Sleep Quality Index that asked: “During the work week, about how many hours of sleep do you typically get per 24-h period?”. There were eight response options (<4 h, 4–5 h, 5–6 h, 6–7 h, 7–8 h, 8–9 h, 9–10 h, > 10 h). Depressive symptoms were assessed with an 8-item version of the CES-D scale, which has shows excellent reliability in studies of adults (; α = .80). The measure listed several symptoms of depression (e.g., sad, lonely) and asked respondents how often they experienced each symptom on a 4-point rating scale from 0 (less than 1 day per week) to 3 (5–7 days per week); scores are calculated by summing across the item ratings. Leisure time physical activity was assessed with one item: “Outside of work, in an average week during the past year, how many hours did you spend on… physical exercise such as fitness, aerobics, swimming, jogging, cycling, tennis, etc.?” adapted from the EPIC Physical Activity Questionnaire . Response options included: 0 h per week, 1–3 h per week, 4–6 h per week, 7–9 h per week, 10–12 h per week, greater than 12 h per week.
Musculoskeletal pain was assessed with the question: “During the past 3 months, how much pain, aching or stiffness/limited motion have you had in the areas shown on the diagram below?”[30, 31]. The measure listed seven areas of the musculoskeleture (e.g., low back, knee) and asked respondents to rate how severely each area was affected on a 5-point rating scale from 0 (mild) to 4 (extreme). Participants were considered to have musculoskeletal pain if they indicated a score of 2 (moderate) or more in any body area. Weight perception was assessed with one item: “Tell us whether you are interested in making changes or improvements in your health in the following area… lose weight or maintain healthy weight”. Response options were: 0 (not interested in changing), 2 (interested in changing), and 3 (currently doing this to my satisfaction). Work-life balance was based on one question, “How successful do you feel at balancing your paid work and your family life? Do you feel…?” Response options ranged on a 5-point scale from 1 (not at all successful) to 4 (completely successful) .
We examined ten work-related factors including job tenure, job type, work shift, overtime, time standing at work, job satisfaction, civility norms, decision latitude, procedural justice, psychological demands, social support, and stress in general. Job tenure was assessed with the open-ended question “How many years have you worked at your organization?” to which respondents entered a numeral. Job type was measured with an item to assess whether employees were either production workers on the shop floor or administrative employees in office jobs (i.e., managers, sales and administrative staff); each job type places distinct biomechanical and psychosocial demands on workers. Work shift was measured using one question “What shift do you typically work?” with three possible response options (firstshift, second shift, third shift). Work overtime was assessed with one question “Thinking of the past year, which best describes the amount of overtime or extra hours you work in an average month?” that had six response options (0–4 h, 5–12 h, 13–24 h, 25–36 h, 37–50 h,51 h and above). Work time standing was measured with one question: “Please check the box that best describes how much standing/walking you do on your job, from always sitting (0 %) to always standing or walking (100 %)” followed by 11 response options (0 % always sitting, 10, 20, 30, 40, 50 % Half & Half, 60, 70, 80 90, 100 % always standing or walking).
Job satisfaction was assessed using a 3-item measure ; a sample item was “I am satisfied with the overall quality of work done in my workgroup” to which participated responded using a 5-point scale that ranged from 1 (strongly disagree) to 5 (strongly agree) and a score was calculated by averaging ratings across the items. Civility norms was assessed using a 4-item measure ; a sample item was “Respectful treatment is the norm in my department” to which participated responded using a 5-point scale that ranged from 1 (strongly disagree) to 5 (strongly agree) and a score was calculated by averaging ratings across the items. Decision latitude was measured with a subscale from the job content questionnaire  consisting of seven items that assess skill discretion and decision authority. Sample items include: “My job requires me to be creative,” and “My job allows me to make a lot of decisions on my own.” Response options ranged on a 4-point scale from 1 (strongly disagree) to 4 (strongly agree) and a score was calculated by averaging ratings across the items. Procedural justice was measured with four items  that assess work experiences. A sample item is: “Job decisions are made in an unbiased manner.” Response options ranged on a 5-point scale from 1 (strongly disagree) to 5 (strongly agree) and a score was calculated by averaging ratings across the items. Psychological job demands were assessed with a subscale from the job content questionnaire . A sample item was: “My job requires working very hard.” Response options ranged on a 4-point scale from 1 (strongly disagree) to 4 (strongly agree) and a score was calculated by averaging ratings across the items. Stress was assessed with a six-item version of the Stress in General scale (SIG; ; α = .91), which instructs respondents to indicate whether several words or phrases describe their work (e.g., irritating, hectic, hassled). Each item was rated with a 0 (no), 1.5 (cannot decide), or 3 (yes), and a score was calculated by averaging ratings across the items. Social support was measured with a subscale from the Job Content Questionnaire (JCQ: ) consisting of four items that assess instrumental and socioemotional social support from supervisors and coworkers including “(My supervisor is)/(People I work with are) helpful in getting the job done” and “(My supervisor/People I work with) take a personal interest in me”. Response options ranged on a 4-point scale from 1 (strongly disagree) to 4 (strongly agree) and a score was calculated by averaging ratings across the items.
BMI and BFP were treated as continuous variables for all analyses. Age was grouped into three categories (under 45 years old, 45–54 years old, 55 or more years old) with about a third of the sample in each group. All other demographic, health-related, and work-related factors were dichotomized in order to reduce the number of degrees of freedom to be included in the models. When dichotomizing variables, we aimed to choose standard cutoffs or to divide data into two categories as equally distributed as possible in order to optimize power.
Demographic variables dichotomized included race (white, other), marital status (married or living with partner, other), education level (at least some college, no college), family income (less than $75,000, $75,000 and over), childcare responsibility (some or complete responsibility, none or another adult responsible), and eldercare responsibility (responsible for at least one adult, no responsibility).
Health-related variables that were dichotomized included sleep hours (less than 6 h, 6 or more hours), depressive symptoms (1 day per week or less, more than 1 day per week), leisure time physical activity (at least some, none), musculoskeletal pain (none to mild, moderate to severe), stress (low, high), weight perception (interested in changed, not interested), work-life balance (not or somewhat successful, very or completely successful), and social support (disagree, agree).
Work-related variables that were dichotomized included job tenure (five years or more, less than 5 years), work shift (first shift, other), overtime (less than 24 h per month, 24 h per month or more), time standing at work (standing 30 % of the time or less, standing more than 30 % of the time), job satisfaction (agree, neutral/disagree), civility norms (agree, neutral/disagree), decision latitude (agree,disagree), procedural justice (agree, neutral/disagree), and psychological demands (agree,disagree).
We used chi-squared tests to evaluate differences in the distribution of factors, BMI, and BFP by age. To identify factors associated with BMI and BFP, we performed multivariate linear regression analyses, stratified by age, using all demographic, health-related, and work-related factors to assess associations with baseline and change in BMI and BFP. Before performing the multivariate analyses, we used kappa tests to assess correlation among demographic, health-related, and work-related factors, but because no factors were highly correlated (kappa coefficient > 0.7), we did not restrict the factors included in the multivariate regression models. All statistical analyses were performed in SAS version 9.4 (Cary, NC). Significance was defined as two-tailed p < 0.05.