The Relation Between BMI and Musculoskeletal Symptoms
The Relation Between BMI and Musculoskeletal Symptoms
Based on a large working population cohort, we examined BMI in association with prevalence of musculoskeletal symptoms in employees, with adjustment for potential confounders. Additionally, within a subcohort, transitions in musculoskeletal symptoms were longitudinally investigated in relation to BMI.
Data were obtained from The Netherlands Working Conditions Survey (NWCS). This dataset constitutes of a representative sample of the Dutch workforce in the 15–64 years age group, but excluded self-employed individuals. Each year, 80,000 individuals were sampled from the Dutch working population database by Statistics Netherlands. This database contains information on all jobs that fall under the worker national insurance schemes and are liable to income tax. Sampling was random, except for a 50% over-sampling of employees with lower response rates, namely employees under the age of 25 years and employees with a non-western background. Individuals in the sample received the questionnaire mailed to their home address. After three to four weeks, reminders were sent to those who had not yet responded. Data collection was stopped after two months. To be representative for employees in the Netherlands, the response was weighted for gender, age, sector, ethnic origin, level of urbanization, geographical region and level of education.
The sample was extensively informed about the study in a letter that accompanied the questionnaire. The burden for respondents was low given the topics covered in the questionnaire. Consequently, and in accordance with ethics regulations in the Netherlands, ethical approval was not required for this study.
A total of 44,793 employees completed the NWCS questionnaire in 2008 or 2009 (2008: n = 22,025, 2009: n = 22,768; overall response rate: 28%) and these employees were eligible for the cross-sectional analysis. In addition to the regular annual survey, respondents of the NWCS questionnaire in 2007, who gave consent for being contacted in the future, were invited to respond to follow-up questionnaires in 2008 and 2009 (Netherlands Working Conditions Cohort Study (NWCCS)).
In this cohort, a total of 7,909 completed the NWCCS questionnaire in 2009 (response rate: 35%). Respondents who participated at follow-up were more often higher educated and slightly older than expected based on the NWCS sample. No selective differences were found for the dependent variables BMI and musculoskeletal symptoms. Data retrieved from the NWCCS of these 7,909 respondents were used for the second research aim (i.e., to investigate the transition in musculoskeletal symptoms).
Self-reported body weight in kilogrammes (kg) and body height in centimetres (cm) were used to determine BMI. BMI was computed as weight (kg)/height (m). Subsequently, BMI was classified into three categories (normal weight (BMI 18.5–24.9 kg/m), overweight (BMI 25.0–29.9 kg/m), and obese (BMI ≥ 30 kg/m)), which is in accordance with the international classification system of the WHO.
The questions on musculoskeletal symptoms were based on the Dutch Musculoskeletal Questionnaire. Employees were asked to rate the occurrence of pain or discomfort in the neck, shoulders, back, arms/elbows, hands/wrists, and lower extremity, in the previous 12 months using 6 questions with five answering categories ('never', 'only once, of short duration', 'only once, prolonged', 'frequently, of short duration', 'frequently and prolonged'). Employees who answered 'never' or 'only once, of short duration' on all questions were classified as having no musculoskeletal symptoms. Those who answered 'prolonged' or 'frequently' for one or more locations were classified as having musculoskeletal symptoms overall. Hence, this overall prevalence is reported for any location, in addition to location-specific prevalences for which the responses on neck and shoulders were combined (neck/shoulder), as were those on arms/elbows and hands/wrists (upper extremity).
Employees were asked questions on current use of force, work in awkward positions, use of vibrating tools (tools, machines or vehicles), and repetitive motions on a 3-point scale ('never', 'yes, occasionally', yes, regularly'). Employees who answered 'yes, regularly' on use of force or work in awkward positions were classified as having high physical workload. Those who answered 'no, never' or 'yes, occasionally' on both questions were classified as having low physical workload.
Additional potential confounders were gender, age, education (categorized into low, intermediate, and high educational level), contractual working hours (part time/full time), current smoking (yes/no), and physical activity (days a week physically active for at least 30 minutes and of at least moderate intensity). Physical activity was dichotomized as physically active (yes/no) according to the Dutch public health recommendation for moderate intensity physical activity.
For the first research aim, using the weighted cross-sectional data, logistic regression analyses were carried out to investigate the association between BMI and musculoskeletal symptoms. The measure of association was expressed by the Odds Ratio (OR) and its 95% confidence interval (CI). In the categorical analyses involving BMI, the interval 18.5–24.9 was considered as the reference group. In adjusted analysis potential confounders were added to the regression model (full model).
Effect modification was defined as a significant interaction term (p < 0.05) between potential effect modifiers (age, gender, physical workload) and BMI. Analyses were presented stratified for age, gender, or physical workload if the associations between BMI and musculoskeletal symptoms differed based on significant interaction terms.
For the second research aim, using the cohort data (no weighting), the analyses were stratified for respondents without symptoms and those with symptoms in the baseline survey. To determine the difference in the risk of developing symptoms (occurrence) between employees who are overweight and those who are not, outcome was the 12-month incidence of musculoskeletal symptoms. Cases of musculoskeletal symptoms were identified as those who reported frequent or prolonged symptoms at follow-up. To study the influence of BMI on recovery from symptoms, a separate analysis for employees who reported frequent or prolonged symptoms in the last 12 months was performed. Hence, the OR expressed the association between the risk factor at baseline (high BMI) and transition from symptoms to no symptoms, or the reverse, at follow-up.
Methods
Sample/Study Population
Based on a large working population cohort, we examined BMI in association with prevalence of musculoskeletal symptoms in employees, with adjustment for potential confounders. Additionally, within a subcohort, transitions in musculoskeletal symptoms were longitudinally investigated in relation to BMI.
Data were obtained from The Netherlands Working Conditions Survey (NWCS). This dataset constitutes of a representative sample of the Dutch workforce in the 15–64 years age group, but excluded self-employed individuals. Each year, 80,000 individuals were sampled from the Dutch working population database by Statistics Netherlands. This database contains information on all jobs that fall under the worker national insurance schemes and are liable to income tax. Sampling was random, except for a 50% over-sampling of employees with lower response rates, namely employees under the age of 25 years and employees with a non-western background. Individuals in the sample received the questionnaire mailed to their home address. After three to four weeks, reminders were sent to those who had not yet responded. Data collection was stopped after two months. To be representative for employees in the Netherlands, the response was weighted for gender, age, sector, ethnic origin, level of urbanization, geographical region and level of education.
The sample was extensively informed about the study in a letter that accompanied the questionnaire. The burden for respondents was low given the topics covered in the questionnaire. Consequently, and in accordance with ethics regulations in the Netherlands, ethical approval was not required for this study.
A total of 44,793 employees completed the NWCS questionnaire in 2008 or 2009 (2008: n = 22,025, 2009: n = 22,768; overall response rate: 28%) and these employees were eligible for the cross-sectional analysis. In addition to the regular annual survey, respondents of the NWCS questionnaire in 2007, who gave consent for being contacted in the future, were invited to respond to follow-up questionnaires in 2008 and 2009 (Netherlands Working Conditions Cohort Study (NWCCS)).
In this cohort, a total of 7,909 completed the NWCCS questionnaire in 2009 (response rate: 35%). Respondents who participated at follow-up were more often higher educated and slightly older than expected based on the NWCS sample. No selective differences were found for the dependent variables BMI and musculoskeletal symptoms. Data retrieved from the NWCCS of these 7,909 respondents were used for the second research aim (i.e., to investigate the transition in musculoskeletal symptoms).
Measurement of BMI
Self-reported body weight in kilogrammes (kg) and body height in centimetres (cm) were used to determine BMI. BMI was computed as weight (kg)/height (m). Subsequently, BMI was classified into three categories (normal weight (BMI 18.5–24.9 kg/m), overweight (BMI 25.0–29.9 kg/m), and obese (BMI ≥ 30 kg/m)), which is in accordance with the international classification system of the WHO.
Measurement of Musculoskeletal Symptoms
The questions on musculoskeletal symptoms were based on the Dutch Musculoskeletal Questionnaire. Employees were asked to rate the occurrence of pain or discomfort in the neck, shoulders, back, arms/elbows, hands/wrists, and lower extremity, in the previous 12 months using 6 questions with five answering categories ('never', 'only once, of short duration', 'only once, prolonged', 'frequently, of short duration', 'frequently and prolonged'). Employees who answered 'never' or 'only once, of short duration' on all questions were classified as having no musculoskeletal symptoms. Those who answered 'prolonged' or 'frequently' for one or more locations were classified as having musculoskeletal symptoms overall. Hence, this overall prevalence is reported for any location, in addition to location-specific prevalences for which the responses on neck and shoulders were combined (neck/shoulder), as were those on arms/elbows and hands/wrists (upper extremity).
Potential Confounders and Effect Modifiers
Employees were asked questions on current use of force, work in awkward positions, use of vibrating tools (tools, machines or vehicles), and repetitive motions on a 3-point scale ('never', 'yes, occasionally', yes, regularly'). Employees who answered 'yes, regularly' on use of force or work in awkward positions were classified as having high physical workload. Those who answered 'no, never' or 'yes, occasionally' on both questions were classified as having low physical workload.
Additional potential confounders were gender, age, education (categorized into low, intermediate, and high educational level), contractual working hours (part time/full time), current smoking (yes/no), and physical activity (days a week physically active for at least 30 minutes and of at least moderate intensity). Physical activity was dichotomized as physically active (yes/no) according to the Dutch public health recommendation for moderate intensity physical activity.
Analysis
For the first research aim, using the weighted cross-sectional data, logistic regression analyses were carried out to investigate the association between BMI and musculoskeletal symptoms. The measure of association was expressed by the Odds Ratio (OR) and its 95% confidence interval (CI). In the categorical analyses involving BMI, the interval 18.5–24.9 was considered as the reference group. In adjusted analysis potential confounders were added to the regression model (full model).
Effect modification was defined as a significant interaction term (p < 0.05) between potential effect modifiers (age, gender, physical workload) and BMI. Analyses were presented stratified for age, gender, or physical workload if the associations between BMI and musculoskeletal symptoms differed based on significant interaction terms.
For the second research aim, using the cohort data (no weighting), the analyses were stratified for respondents without symptoms and those with symptoms in the baseline survey. To determine the difference in the risk of developing symptoms (occurrence) between employees who are overweight and those who are not, outcome was the 12-month incidence of musculoskeletal symptoms. Cases of musculoskeletal symptoms were identified as those who reported frequent or prolonged symptoms at follow-up. To study the influence of BMI on recovery from symptoms, a separate analysis for employees who reported frequent or prolonged symptoms in the last 12 months was performed. Hence, the OR expressed the association between the risk factor at baseline (high BMI) and transition from symptoms to no symptoms, or the reverse, at follow-up.
