Psychosocial Risk Factors for Low Back Pain in US Workers: Data from the 2002–2018 Quality of Work Life Survey

Abstract

Background

Examining workplace psychosocial risk factors for back pain becomes increasingly important because of the changing nature of work and rising healthcare costs. Some psychosocial risk factors for back pain, such as work and family imbalance, the exposure to a hostile work environment and job insecurity, have been understudied for the working population in the United States.

Methods

Data used in this study came from the Quality of Work Life Survey (QWL), a supplementary topical module on the General Social Survey conducted in the United States. Data from the 2002, 2006, 2010, 2014 and 2018 QWL surveys were used in these analyses, giving a total sample size of 6,661. Five domains of workplace psychosocial risk factors for back pain were examined, including job strain, low social support, work-family imbalance, exposure to a hostile work environment (harassment and discrimination), and job insecurity. The adjusted odds ratio (aOR) of each psychosocial risk factor for back pain with a 95% confidence interval (CI) was estimated using a multivariable logistic regression model after controlling for job physical factors, occupation, and demographic and socioeconomic characteristics.

Results

Significant associations were found between back pain and several psychosocial factors including job strain (aOR 1.19, CI 1.00,1.41), work-family imbalance (aOR,1.42, CI 1.22,1.64), harassment (aOR 1.40, CI 1.15,1.71), and discrimination (aOR 1.20, CI 1.00,1.44).

Conclusion

This study contributes to the understanding of the relationship between a variety of workplace psychosocial factors and back pain. Our findings suggest directions in future longitudinal research to examine emerging workplace psychosocial factors for back pain.

1. BACKGROUND

Back pain, or more specifically low back pain, is a common and costly health condition in the working population of the United States and most workers are expected to experience back pain in their working life.^1,2 Back pain may be a precursor for underlying spinal or musculoskeletal disorders (MSDs), which have caused significant burden to individuals and the society as a whole. Katz (2006) estimated the total costs of low back pain in the United States to exceed $100 billion per year, with approximately one third accounted for by direct medical expenses and two thirds resulting from indirect costs, such as productivity loss and sickness absence.³ The annual direct cost of treating spinal pain was reported as the third highest treatment cost among all the major health conditions in the US, costing $87.6 billion in 2013. In 2016, spinal pain was ranked the highest among all major health conditions in amount of health care spending, costing an estimated $134.5 billion.⁴ The Global Burden of Disease 2017 study demonstrated that among 354 diseases and injuries, low back pain was ranked highest in terms of leading cause of disability and years lived with disability.⁵ Back pain also leads to suffering and interference with quality of life for individuals.^6–11 Investigating workplace related factors for back pain has been recognized as a critical area of research in the field of occupational health.

The major work-related risk factors for back pain have been seen as physical in nature, including heavy lifting, awkward postures, whole-body vibration, and prolonged sitting.^2,12 There has been substantial evidence in occupation specific studies linking back pain with these work-related physical risk factors.^13–15 The occupational groups with excess risk for back pain include manufacturing,¹⁵ construction,^16,17 transportation,¹⁸ healthcare,¹⁹ and office workers.¹⁴ However, many of these occupation specific studies also found back pain related to psychosocial stressors, such as high psychosocial demands, and low social support.¹⁹

In addition, population-based studies have demonstrated consistent evidence of associations between work-related physical risk factors and back pain. A longitudinal study of the general working population in Norway indicated that prolonged standing, awkward lifting and squatting/kneeling were significantly related to back pain.²⁰ A longitudinal study of young adult workers conducted between 1986 and 2007 reported that physically heavy work during young adulthood may have a long-lasting impact on back pain risk.²¹ The 2012–2018 Work Environment and Health study of Denmark reported that combined ergonomic exposures (standing/walking combined with lifting/carrying or twisted/bent back) increased risk for back pain and back pain intensity from baseline to a two-year follow-up.²²

A series of studies based on data from the 2002 to 2014 Quality of Work Life (QWL) modules of the General Social Survey (GSS) reported that lifting (repeated lifting, pulling, or pushing) and hand movement (repetitive or stressful hand movements or awkward postures) were work-related physical risk factors significantly associated with back pain in the US working population.^23–26 Although the authors reported some associations between workplace psychosocial factors and back pain, these psychosocial factors were reported in individual questionnaire items, instead of using developed scales with multiple questionnaire items.^23–26 Workplace psychosocial stress domains have been increasingly recognized to be important factors for both the initial development of back pain and subsequent disabilities in the past two decades.^27,28 Job strain, defined as a combination of high job demands and low job control,^29,30 has been frequently studied as a psychosocial risk factor for cardiovascular conditions.^31,32 Job strain has also been seen as a psychosocial risk factor for back pain in some recent studies.^33,34 Substantial research evidence from northern European countries has demonstrated significant associations between job strain and back pain. A study based on the Swedish Longitudinal Occupational Survey of Health (SLOSH) indicated that job strain was associated with increased back pain severity.³⁵ A study of the Swiss working population based on the Fourth European Working Conditions Survey (EWCS) reported that physically or psychologically demanding jobs were associated with higher prevalences of back pain.³⁶ A population-based study conducted in the largest province of China showed that high perceived job demands were associated with sickness absence due to low back symptoms.³⁷ Recent research on workplace back pain has identified new psychosocial risk factors associated with back pain, such as work-family imbalance,^38,39 exposure to hostile work environments,^40,41 and job insecurity.^42–44 These new psychosocial risk factors have gained attention in the field of occupational health research on back pain in the past few years.⁴⁵

Haemmig and his colleagues pointed out work-family conflict or imbalance as a “blind spot” in the field of occupational medicine research on back pain, based on survey data covering workers from four large companies in Switzerland.⁴⁶ With data from the 2011 Korean Work Condition Survey, Kim and her colleagues demonstrated significant associations between work-family imbalance and increased risk of occupational injury and musculoskeletal pain.⁴⁷ Yang and her colleagues reported significant associations between back pain and work-family imbalance (aOR 1.27, CI 1.15,1.4) among workers in the US, based on the 2010 National Health Interview Survey (NHIS).⁴⁵

A population-based study with a prospective design conducted in Norway reported a linkage between exposure to bullying and back pain in male workers.⁴⁸ Sabbath and her colleagues identified workplace abuse as a risk factor for musculoskeletal injuries among health care workers.⁴⁹ Vignoli and her colleagues conducted a study of retail sales workers in Italy and found associations between bullying and musculoskeletal disorders, including back pain.⁵⁰ In analyses of the 2010 NHIS, Yang and her colleagues also found significant associations between back pain and exposure to a hostile work environment (aOR 1.39, CI 1.25,1.55) in the working population of the United States.⁵¹

Job insecurity has also been seen as an important psychosocial risk factor for back pain in the workplace. With data from the 2006 Korean Work Conditions Survey, Lee and his colleagues showed that job insecurity was linked to a number of health problems including back pain among workers with permanent jobs, while the association was not significant among workers with temporary jobs.⁵² Two studies based on data from the 2010 NHIS demonstrated that American workers with job insecurity were likely to have back pain.⁵³

Psychosocial risk factors for back pain have not been well studied in the working population of the United States. The goal of this study is to explore the relationship between back pain and a wide range of workplace psychological factors, including job strain, low social support, work-family imbalance, harassment, discrimination and job insecurity, while controlling for the confounding effects of physical risk factors, occupation, and demographic characteristics.

2. METHODS

2.1. Data

Data used in this study came from the QWL survey that was part of the General Social Survey (GSS) which is designed to monitor social changes in the US population. The GSS is a nationally representative survey of non-institutionalized adults in the United States, conducted every 2 years through face-to-face interviews.⁵⁴ The QWL is a supplementary topical module on the GSS, sponsored by the National Institute of Occupational Safety and Health (NIOSH), and included on the GSS every 4 years since 2002. The QWL includes a set of variables on workplace risk factors and a limited number of health outcome variables.⁵⁵

The study used data from 5 QWL survey waves conducted in 2002, 2006, 2010, 2014 and 2018. The response rates ranged from 60.2% -71.2% from 2002 to 2014, and dropped to 59.5 % in 2018.^56,57 Two inclusion criteria for this study were participants in the GSS who (a) took part in the QWL and (b) worked 20 hours or more per week. The five survey waves resulted in sample sizes of 1,614, 1,549, 1,075, 1,144 and 1,279 respondents respectively. The combined sample size for the present study was 6,661. Table 1 summarizes the number of respondents from each survey using the inclusion criteria.

Table 1.

Sample Sizes of QWL Surveys that Met the Participant Inclusion Criteria

Number of participants	Year
	2002	2006		2010		2014			2018	Total
Total GSS participants	2765	4506		2041		2536			2346	14194
Did not take part in the QWL	21	1049		7		277			25	1379
Not full-time or part-time worker	1021	1744		890		1033			953	5641
Not worked 20 hours or more per week	109	164		69		82			89	513
Final sample size	1614	1549		1075		1144			1279	6661
Response rate (%)	70.1	71.2		70.3		60.2			59.5

There were minor variations in the prevalence of back pain during the study period with similar levels from 2002 (27.6%), 2006 (27.0%), 2010 (25.2%), 2014 (22%) to 2018 (26.0%). These minor prevalence variations in back pian were assumed to be steady and independent of the effect of time for statistical modeling. Myers and her colleagues who monitored the changes in nature of work with the QWL data over 12 years (2004–2014) found that job strain and work-family conflict significantly increased over the 12 year period, but no significant changes were found in other psychosocial factors focused on in this study.⁵⁸ Moreover, because the study population for each NHIS survey was different, the effect of time and the variations in the back pain and psychosocial variables from different study populations were difficult to distinguished. On the basis of the limited variations in back pain outcome and most psychosocial factors, as well as different study populations over the study period, the effect of time was ignored in our statistical model.

2.2. Measurements

The QWL surveys included a broad range of workplace psychosocial risk factors, which allowed us to assess different domains of workplace psychosocial factors, in particular those used in the job strain model.³⁰ However, the measurements (yes/no questions) of exposure to physical risk factors and the health outcomes are relatively limited by comparison.⁵⁵ These variables are described in detail below:

2.2.1. BACK PAIN

In the QWL, back pain is defined as self-reported non-specific back pain every day for a week or more within the past year using a single questionnaire item, “In the past 12 months, have you had back pain every day for a week or more?” This item was developed based on the Nordic Questionnaire for Musculoskeletal Disorders.⁵⁹

2.2.2. Workplace characteristics

The workplace characteristics examined in this study were: psychosocial factors, physical factors and occupation. In this study, these factors were treated as the risk exposure variables for back pain. The psychosocial risk factors included job strain, low social support, work-family imbalance, harassment, discrimination, and job insecurity. The workplace physical factors included lifting (repeated lifting, pulling, or pushing) and hand movements (repetitive/forceful hand movements/awkward arm positions). Table 2 displays information on the constructions of these psychosocial and physical variables by domain, sub-domain, questionnaire item, response options, and coding. Reversed coding was done for items to match the direction of the negative effects (high demands, low control and poor social support) of the psychosocial variables. The higher the score, the more negative direction (high, low or poor) of the variables.

Table 2.

Workplace Related Risk Factor Measurements, Domains and Question Items, and Response Formats

Domain	Item	Question	Response format
Psychosocial risk factors
	Job strain
	Job demands 1
	(a) Work fast	My job requires that I work very fast	4-point Likert scale, “Strongly agree” to “Strongly disagree.” Reverse coded.
	(b) Over work	I have too much work to do everything well.	4-point Likert scale, “Strongly agree” to “Strongly disagree.” Reverse coded.
	(c) Work time	I have enough time to get the job done.	A 4-point Likert scale, “Very true” to “Not at all true.”
	(d) Conflicting demands	I am free from the conflicting demands that other people make of me.	4-point Likert scale, “Very true” to “Not at all true.”
	(e) Too few workers	How often are there not enough people or staff to get all the work done?	4-point Likert scale, “Very true” to “Not at all true.” Reverse coded.
	Job Control 2
	(a) Learn new things	My job requires that I keep learning new things.	4-point Likert scale, “Strongly disagree” to “Strongly agree.”
	(b) Work freedom	I am given a lot of freedom to decide how to do my own work	4-point Likert scale, “Very true” to “Not at all true.”
	(c) Work variety	I get to do a number of different things on my job.	4-point Likert scale, “Strongly agree” to “Strongly disagree.”
	(d) Develop own ability	I have an opportunity to develop my own abilities.	4 point Likert scale, “Very true” to “Not at all true.”
	(e) Allows own decisions	In your job, how often do you take part with others in making decisions that affect you?	4-point Likert scale, “Often” to “Never.”
	Social support 3
	(a) Co-worker help	The people I work with can be relied on when I need help.	4-point Likert scale, “Very true” to “Not at all true.”
	(b) Supervisor care	My supervisor is concerned about the welfare of those under him or her.	4-point Likert scale, “Very true” to “Not at all true.”
	(b) Supervisor help	My supervisor is helpful to me in getting the job done	4-point Likert scale, “Strongly agree” to “Strongly disagree.”
	(c) Co-worker interest	The people I work with take a personal interest in me	4-point Likert scale, “Very true” to “Not at all true.”
	Work-family imbalance 4
	(a) Level of difficulty for taking time off from work	How hard is it to take time off during your work to take care of personal or family matters?	4 point Likert scale, “Not at all hard” to “Very hard.”
	(b) Job interfered with family life	How often your job interfered with your family life?	4-point Likert scale, “Often” to “Never,” reverse coded.
	(c) Family interfere with job	How often do the demands of your family interfere with your work on the job?	4-point Likert scale, “Often” to “Never,” reverse coded.
	Harassment 5
	(a) General harassment at work	In the last 12 months, were you threatened or harassed in any other way by anyone while you were on the job?	Yes/No.
	(b) Sexual harassment	In the last 12 months, were you sexually harassed by anyone while you were on the job?	Yes/No.
	Discrimination 6
	(a) Discrimination based on race or ethnic origin	DDo you feel in any way discriminated against on your job because of your race or ethnic origin?	Yes/No
	(b) Discrimination based on gender	DDo you feel in any way discriminated against on your job because of your gender?	Yes/No
	(c) Discrimination based on age	DDo you feel in any way discriminated against on your job because of your age?	Yes/No
	Job security	The job security is good	4 point Likert scale, “Very true” to “Not at all true.”
Physical work environment
	(a) Lift heavy loads	Does your job require you to do repetitive lifting, pushing, pulling or bending?	Yes/No.
	(b) Awkward body positions and awkward arm positions	Does your job regularly require you to perform repetitive or forceful hand movements or involve awkward postures?	Yes/No.

Notes:

¹The scores for the 5 items were added and scores above the mean were defined as “high demand.”

²The scores for the 5 items were added and scores below the mean were defined as “low control.”

³The scores for the 4 items were added and scores below the mean were defined as “low social support.”

⁴The scores for the 3 items were added and scores above the mean were defined as “work-family imbalance.”

⁵“Yes” for “threatened or harassed in any other way by anyone” or “sexually harassed by anyone on the job” was defined as harassment.

⁶Yes” for discriminated against on job “because of race or ethnic origin,” “because of your gender,” or “because of your age” was defined as discriminations.

Job strain was assessed with the theoretical integrity of the job strain model and with items that were the same for all 5 data collection waves included in the study. Job strain included 2 domains: job control (5 items) and psychological demands (5 items). The Job Control scale had a Cronbach alpha of 0.663 and the Job Demand scale had a Cronbach alpha of 0.6435. Low social support was measured with 4 items with a Cronbach alpha of 0.7425. The assessment of both job strain and low social support was consistent with that used by Myers and her colleagues in describing changes in work characteristics in the US with the 2002–2014 QWL data series.⁵⁸ Job strain was constructed with the quadrant method (high demand and low control) with the mean as the cut-point. Work-family imbalance was assessed with 3 items (Cronbach alpha of 0.5724). Harassment was assessed with 2 items (Cronbach alpha of 0.3433). Discrimination was assessed with 3 items (Cronbach alpha of 0.4750). Job insecurity was assessed with one item.

Occupation was coded into 8 groups: (a) Professional (Computer and mathematical; Architecture and engineering; Life, physical, and social science; Community and social services; Legal occupations; Education, training and library; Arts, design, entertainment, sports and media; Healthcare practitioners and technical), (b) Management (business and financial operations), (c) Sales and related (d) Production related (Farming, fishing, and forestry, manufacturing), (e) Office and administrative support, (f) Services and related (Healthcare support; Protective service; Food preparation and serving related; Building and grounds cleaning and maintenance; and Personal care and service), (g) Construction related (Construction and extraction; Installation, maintenance and repair) and (h) Transportation related. This classification was derived from the 2000 Occupation Subgroups and Major Occupation Groups, respectively, as determined by the U.S. Census Bureau and the Bureau of Labor Statistics. The occupation group, Professional, was used as the reference group in the logistic regression analyses described in the statistical analysis section.

2.2.3. Demographic characteristics

Demographic variables used in the analysis included gender, age, and race/ethnicity. Age was coded into five age groups: (a) 18 – 25, (b) 26 – 40, (c) 41 – 55, (d) 56 – 65 years, and (e) 66 and over. The 18–25 age group was used as the reference group in the logistic regression analyses.

Race/ethnicity was coded into four groups: (a) Non-Hispanic White, (b) Non-Hispanic Black, (c) Hispanic, and (d) Non-Hispanic Others. The Non-Hispanic White group was used as the reference group in the logistic regression analyses.

Socioeconomic status (SES) variables were education and personal earnings. Education was coded into five categories: a) less than high school, b) high school, c) some college, d) college and e) master’s degree and above. The master’s degree and above was used as the reference group in the univariate logistic regression analyses, so an inverse association between education level and back pain could be seen. The high school group was used as the reference group in the multivariate logistic regression analyses.

Personal earnings were defined as personal yearly earnings coded into 6 categories, a) under $15,000, b) $15,000 - $24,999, c) $25,000 – 34,999, d) $35,000 - $49,999, e) $50,000 - $74,999, and f) $75,000 and above. The $75,000 and above group was used as the reference group in the univariate logistic regression analyses to demonstrate an inverse association between earning level and back pain. The lowest earning group was used as the reference group in the multivariate logistic regression.

2.3. Statistical analysis

We developed our research model based on two bodies of literature on workplace related psychosocial risk factors: the traditional risk factors of job strain and the emerging psychosocial risk factors of work-family imbalance, exposure to a hostile working environment, and job insecurity.⁴⁵ The measurements of job strain in this study were based on questions common to both the Job Content Questionnaire (JCQ) and the QWL that were theoretically consistent with the job strain scale.⁶⁰ The job strain model has been a frequently used framework for understanding how job structures impact many health problems of workers, including back pain.^33,61,62 The emerging psychosocial risk factors included in this study were based on the recent research conducted in this area by this research team.⁴⁵

We did not use all the QWL data collected regarding other working conditions that may be related to back pain. For example, work dissatisfaction has been cited as a work related psychosocial variable linked to back pain. ^24,63 This study did not include this variable in the analysis model for back pain after examining its multicollinearity with another psychosocial risk factor, job strain.

First, we conducted descriptive analyses of the demographic and socioeconomic characteristics of study participants. Descriptive analyses included distributions of demographic characteristics and socioeconomic status; workplace psychosocial and physical factors; and occupation by back pain. Second, we calculated each independent variable’s crude odds ratio (OR) for back pain with 95% confidence intervals (CIs) using univariate logistic regressions.

Third, we developed a full multivariate logistic regression model to evaluate the associations between back pain and a set of psychosocial factors (job strain, low social support, work-family imbalance, harassment, discrimination, and job insecurity) while controlling for workplace physical factors (lifting and hand movements), occupation, demographic (age, gender, and race, and ethnicity), and socioeconomic variables (education and earning).

Adjusted ORs (aORs) were estimated with 95% CIs. STATA 12 was used to compute descriptive statistics and measures of associations with the logistic regression models. The P value for statistical significance was set as < 0.05.

This study was reviewed by CDC and conducted consistent with applicable federal law and CDC ethics policy.

3. RESULTS

The total number of working adults included in this study was 6,649, with 68.1% Non-Hispanic Whites, 14.3 % Non-Hispanic Blacks, 13.7% Hispanics, 50.5 % females, 94.6 % in the age range of 18–64 years, and, 35.7% with less than a high school or a high school education, and 23.7% with under $15,000 in personal earnings per year.

The prevalence of back pain indicated by this pooled data analysis was 26.4%. Table 3 shows the demographic distribution of respondents by back pain. The female workers reported a slightly higher rate for back pain (26%) than the male workers (25%). There was a decreasing trend in reporting back pain from the youngest age group to the oldest age group. The youngest age group (18 – 25) had the highest percentage of reported back pain (28%) and the oldest age group (65 and above) had the lowest percentage (22%). Compared with those with high school education, those with college (OR 0.64, p < 0.001) or Master’s and above education (OR 0.59, p < 0.001) were less likely to report back pain. Compared with those with earnings of $75,000 and above, the unadjusted ORs for reporting back pain were significantly higher for those with lower earnings.

Table 3.

Distributions of the Study Population by Demographic Characteristics and BACK PAIN, Univariate Odds Ratios

	Demographic Characteristics	% in study pop	% with BACK PAIN	Odds Ratio	95% Conf.	P
BACK PAIN			25.9
Age
	18 – 25	10.6	28.2
	26 – 40	36.1	26.7	0.93	(0.77,1.12)
	41 – 55	35.2	25.7	0.88	(0.73,1.06)
	56 – 64	12.7	24.0	0.81	(0.64,1.01)	**
	65+	5.4	21.6	0.70	(0.53,0.95)	*
Sex
	Male	49.5	25.3	1.00
	Female	50.5	26.4	1.06	(0.95,1.18)
Ethnicity and race
	Non-Hispanic White	68.1	26.7	1.00
	Non-Hispanic Black	14.3	20.9	0.72	(0.61,0.86)
	Hispanic	13.7	27.7	1.05	(0.89,1.25)
	Non-Hispanic Others	4.1	24.6	0.90	(0.68,1.19)
Education
	Less than high school	10.3	36.0	2.43	(1.95,3.04)	**
	High school	25.5	28.2	1.70	(1.40,2.05)	**
	Some college	29.9	27.8	1.66	(1.38,2.00)	**
	College	18.9	20.1	1.09	(0.88,1.34)
	Master and above	15.5	18.8
Earning
	<  $15,000	23.7	31.3	2.11	(1.62,2.75)	**
	$15,000 - $24,999	19.3	26.6	1.68	(1.28,2.21)	**
	$25,000 - $34,999	15.7	25.7	1.61	(1.21,2.13)	*
	$35,000 - $49,999	19.2	26.0	1.63	(1.24,2.15)	**
	$50,000 - $74,999	13.3	23.6	1.43	(1.07,1.92)	*
	> =$75,000	8.9	17.7	1.00

^*<=0.05

^**<=0.001

Table 4 shows the characteristics of the workplace psychosocial and physical factors and the occupational groupings of the study population. For workplace psychosocial factors, the proportion of respondents who reported job strain was 34.1%, low supervisory support 29.0%, work-family imbalance 28.3%, harassment 37.4%, discrimination 32.7% and job insecurity 33.0%. The univariate logistic regressions showed that all 5 workplace psychosocial factors were significantly associated with back pain: job strain (OR 1.63 CI. 1.42,1.87), low social support (OR 1.36 CI. 1.21,1.52), work-family imbalance (OR 1.42 CI. 1.27,1.60), harassment (OR 1.85 CI. 1.58,2.17), discrimination, (OR 1.48 CI. 1.28,1.72), and job insecurity (OR 1.49 CI. 1.27,1.73).

Table 4.

Proportion of Workers with BACK PAIN by Work-related Psychosocial and Physical Risk Factors, Univariate Odds Ratios

	Variable	% in study pop	% with BACK PAIN	Odds Ratio	95% Conf.	P
BACK PAIN		25.9
Psychosocial risk factors
	Job strain	19.0	34.1	1.63	(1.42,1.87)	**
	Low social support	48.7	29.0	1.36	(1.21,1.52)	**
	Work-family imbalance	63.5	28.3	1.42	(1.27,1.60)	**
	Harassment	11.4	37.4	1.85	(1.58,2.17)	**
	Discrimination	15.0	32.7	1.48	(1.28,1.72)	**
	Job insecurity	13.1	33.0	1.49	(1.27,1.73)	**
Physical risk factors
	Lift heavy loads	47.8	31.4	1.74	(1.56,1.95)	**
	Repetitive/forceful hand movements/awkward arm positions	48.7	32.9	2.05	(1.83,2.30)	**
Occupation
	Professional	22.5	19.9	1
	Management, business and financial related	16.7	20.4	1.03	(0.85,1.26)	*
	Sales and related	9.1	25.2	1.36	(1.08,1.70)	**
	Production related	6.6	28.7	1.62	(1.26,2.07)	**
	Office and administrative support	12.9	27.8	1.54	(1.27,1.88)	**
	Services and related	17.7	30.1	1.73	(1.45,2.07)	**
	Construction related	8.8	34.3	2.10	(1.69,2.60)	**
	Transportation related	5.7	31.1	1.81	(1.41,2.34)	**

^*<=0.05

^**<=0.001

For workplace physical factors, 48% of the respondents answered yes to lifting heavy loads in their work and almost half (49%) of the workers reported yes to repetitive or forceful hand movements or awkward arm positions at work. The univariate logistic regressions demonstrated that these workplace physical factors were also significantly associated with back pain: lifting (OR 1.74 CI. 1.56,1.95) and hand movements (OR 2.05 CI. 1.83,2.30).

For occupational patterns of back pain, the top three occupation groups with significantly higher percentages of back pain were: construction (34.3%), transportation (31.1%) and service related (30.1%). Compared with the professionals, all other occupation groups had a significantly increased risk for back pain in the univariate analyses.

In the multivariate analyses shown in Table 5, workers that reported high job strain were 19% more likely to report back pain, compared with those that reported low job strain (aOR 1.19, CI 1.00,1.41). Workers that reported work-family imbalance (aOR 1.42, CI 1.22–1.64), harassment (aOR 1.40, CI 1.15,1.71), or discrimination (aOR 1.20, CI 1.00,1.44) were significantly more likely to report back pain.

Table 5.

Logistic Analysis of Psychosocial and Physical Risk Factors for Back Pain^***

	Variable	Odds Ratio	95% confident interval	P
Psychosocial risk factors
	Job strain	1.19	(1.00,1.41)	*
	Low social support	1.05	(0.91,1.22)
	Work-family imbalance	1.42	(1.22,1.64)	**
	Harassment	1.40	(1.15,1.71)	**
	Discrimination	1.20	(1.00,1.44)	*
	Job insecurity	1.16	(0.95,1.41)
Physical risk factors
	Lift heavy loads	1.22	1.03,1.43)	*
	Repetitive/forceful hand movements/awkward arm positions	1.62	(1.39,1.89)	**
Occupation
	Professional	1.00
	Management, business & financial related	1.06	(0.84,1.34)
	Sales and related	1.17	(0.89,1.55)
	Production related	1.22	(0.90,1.65)
	Office and administrative support	1.40	(1.10,1.77)	*
	Services and related	1.21	(0.96,1.53)
	Construction related	1.56	(1.18,2.07)	*
	Transportation related	1.42	(1.03,1.96)	*

^*<=0.05

^**<=0.001

^***controlling for demographic (age, gender, and race, and ethnicity), and socioeconomic variables

Two physical job factors, lifting heavy loads (aOR 1.22, CI 1.03,1.43) and repetitive/forceful hand movements, or awkward arm postures (aOR 1.62, CI 1.39,1.89), were also significantly associated with back pain. Three occupation groups that had a significantly higher risk for back pain were construction (aOR 1.56, CI 1.18,2.07), transportation (aOR 1.42, CI 1.03,1.96), and office and administrative support (aOR 1.40, CI 1.10,1.77).

4. DISCUSSION

The QWL survey is designed to monitor the changing nature of work, to assess the relationship between job/organizational characteristics and worker health, and to identify strategies for workplace preventive interventions.⁶⁴ While the goal of this study is not to monitor the changes in nature of work, the primary goal of this study is to investigate the association between back pain and a set of workplace psychological factors (job strain, low social support, work-family imbalance, harassment, discrimination and job insecurity) with the QWL data.

The main finding of this study was the associations between back pain and psychosocial factors outside of the job strain model. These psychosocial factors were work-family imbalance, harassment, and discrimination. This study provided evidence to support the finding of the associations between back pain and a similar set of psychosocial factors (work-family imbalance and exposure to hostile work environment) in a previous study using the 2010 NHIS data.⁴⁵ The proportions of workers reporting work-family imbalance in the QWL were somewhat higher than in the 2010 NHIS.^44,45 These variations may be due to the differences in the measurements of these psychosocial factors and possible sampling differences between the QWL and the NHIS.

Although the back pain definitions were different in the QWL (back pain lasting 7 days or more in one year) and the NHIS surveys (any low back pain episode in past three months), the proportions of workers reporting back pain in the two study populations were similar.⁴⁵ Interestingly, the proportions of workers reporting back pain for each of the psychosocial risk factor categories were also similar. We do not have a specific explanation for this comparable reporting of different pain definitions and exposures to psychosocial stressors. The similar results may be due to the fact that both the QWL and NHIS surveys were drawn from the same US working population. The two national survey instruments do not assess the chronicity of back pain. The Research Task Force on Chronic back pain defined chronic back pain as “a low back pain problem that has persisted at least 3 months and has resulted in pain on at least half the days in the past 6 months” and also called for assessments of low back pain with chronicity, intensity, and interference.⁶⁵ These assessments of pain chronicity will be important for future studies on the impact on quality of working life.

The univariate OR (1.7) of lifting heavy loads for back pain in this study was comparable to those (OR 2.01, 1.69, 1.55 and 1.86 for 2002, 2006, 2010 and 2014, respectively) found in previous studies using the QWL survey data from 2002 to 2014.^23–26 The aOR of lifting heavy loads for back pain using the pooled dataset was attenuated slightly to 1.2 but the statistical inference remained significant. Similarly, repetitive hand movements as another physical risk factor for back pain remained significant after controlling for a set of psychosocial risk factors and other confounding variables. Their significance in the multivariate regression model indicates both hand movements and lifting heavy loads remain highly relevant to back pain regardless of exposures to various psychosocial stressors. It is worth noting that the magnitudes of the aORs of several psychosocial stressors were in a similar range of the aORs of the physical risk factors (heavy lifting and hand movements in Table 5), although a direct comparison of the aORs cannot be made.

Compared with the professional group, the univariate ORs for all other occupation groups were significantly higher. After controlling for workplace psychosocial and physical risk factors, and demographic characteristics in the multivariate analysis, construction workers, transportation workers, and office/administrative support workers had a significantly increased risk (a 56%, 42% and 40% increase, respectively) for back pain. The finding of the increased risk for back pain in these three occupation groups is consistent with physical risk exposures in the literature. A longitudinal study of construction workers of the United States showed that jobs that involved a great deal of physical effort significantly increased the risk of back disorders, and the longest-held jobs in construction also increased the risk for back pain.¹⁶ A Swedish study of a randomly selected sample of construction workers found a dose-response relationship between severe back pain and both stooping or kneeling.⁶⁶ A study of office workers conducted in Turkey showed that 53% of them complained about back pain, which could be attributed to prolonged sitting without rest breaks.⁶⁷ However, a review study of office workers and back pain found little evidence of an association between sitting with a poor posture and back pain among office workers, compared with studies of diverse occupations focusing on associations between exposure to various working postures and back pain. ⁶⁸ Nevertheless, indicators of socioeconomic status may attenuate the effect estimated for specific occupational exposures. These indicators may include financial or geographic access to healthier food or exercise facilities, health literacy and access to information, social prestige, and perceived relative deprivation.^69,70

The significant workplace psychosocial factors found in this study may increase the understanding of factors contributing to back pain in workers.^11,71 There have been several hypothesized mechanisms or pathways between psychosocial risk factors and back pain. Based on recent interdisciplinary physiology and neuroscience research, a conceptual framework was proposed by McEwen and his colleagues to better understand complex stress-related health problems and possible interventions.^72,73 Two major key terms in this framework are “allostasis” and “allostatic load.” Allostasis is defined as a fundamental process of adaptation to both predictable and unpredictable events to restore stability. Allostatic load is defined as the cumulative result of an allostatic state, and it could be understood as "the wear and tear” on the multiple bodily systems through collective exposures to psychosocial stressors, along with the impact of life experiences.^72,74 The brain is the central organ of stress and adaptation (allostatic load) and physiological dysregulation caused by wearing down of multiple body systems and psychosocial factors may also trigger pain.^75,76

In addition to McEwen’s conceptual framework of allostatic load that provides an explanation of the pathways between psychosocial factors and health problems, there has been research exploring the possible pathways between back pain and psychosocial factors, such as exposure to work-family imbalance, harassment, discrimination, and job insecurity. These speculations are not contradictory, but complementary. For example, work-family imbalance may induce muscle tension due to physical and emotional exhaustion and lack of rest and recovery may be an important risk factor for the development of muscular pain, including back pain.^46,77 The mechanism for back pain related to exposure to a hostile work environment is likely to involve increased psychosocial strain resulting from both biomechanical and physiological processes (e.g., by increasing muscle tension).⁷⁸ The “physiological vulnerability” caused by mental strain associated with job insecurity may contribute to the development of back pain.⁷⁹

As the characteristics of work have been changing in the past decades, continuing investigation of workplace psychosocial risk factors in relation to various adverse health outcomes is particularly important. Some driving forces of these changes are: intensifying global economic competition, growing information and communication technologies, an expanding service sector, blurring the boundary between day and night, increasing female labor force participation, increasing political and cultural openness and fluctuating economic growth.^58,80–82 The future of work may involve less manual labor but increasing cognitive demands resulting from collaborating with robots or smart production systems. These cognitive demands may translate into psychosocial stressors that mediate the development of back pain.

There are several strengths of this study. First, we constructed measurements of psychosocial factors by utilizing the QWL’s extensive measures of working conditions. The QWL survey instrument covers a wide range of psychosocial working conditions, including but not limited to workload, worker autonomy, skill utilization, role clarity and conflict, job security, work and family imbalance, discrimination, harassment, physical effort, supervisory support, job satisfaction, and physical health.⁶⁴

Second, the measurements of job strain in this study were based on questions common to both the Job Content Questionnaire (JCQ) and the QWL that were theoretically consistent with the job strain scale.⁶⁰ The measurements consisted of low social support with 4 items, work-family imbalance with 3 items, harassment with 2 items, discrimination with 3 items, and job insecurity with 1 item. Assessing these psychosocial factors with multiple questionnaire items improves the reliability and validly of the measures.^83,84

Third, we combined five waves of data from the QWL survey to assure a relatively large sample size for an analysis of associations between many psychosocial factors and back pain. Results of this pooled data analysis on the association between back pain and job strain filled a research gap in the linkage between job strain and back pain in the US working people, as most of the research evidence in this area was from Northern European countries.^69,85–87

Several limitations in this study are worth mentioning. First, this study was in cross-section design, which did not allow us to infer the causality between back pain and related risk factors. Thus, future research using a longitudinal study design is recommended for confirming the associations between psychosocial risk factors and back pain. Second, the health outcome in this study may have unclear onsets or recurrences of back pain that were not defined clearly in the QWL survey. Third, we pooled data from five recent QWL datasets to gain stability and generalizability of the associations between workplace risk factors and back pain. However, the pooled dataset cannot be used to assess the potentially changing nature of work over the years. The effect of time on the variations in both pain outcome and psychosocial variables in the statistical model was ignored. Workplace related health outcomes and risk factors in cross-sectional study waves may vary, even though they are being measured consistently. These fluctuations may be due to various reasons that are difficult to assess. For example, changes in sample size in the study population for each year may be related to variations in workplace related health outcomes. Finally, the respondents’ reporting of back pain in the past year may be subject to recall bias.

5. CONCLUSION

This is the first study to explore the associations between back pain and a wide range of psychosocial stressors in the US working population by analyzing a pooled dataset from five QWL surveys. Significant associations were found between new psychosocial factors (work-family imbalance, harassment and discrimination) and back pain, after controlling for job-related physical risk factors, traditional psychosocial risk factors (i.e., job strain and social support), occupation and demographic characteristics. An implication of the findings from this study is a need for future research that uses a longitudinal study design for identifying the potential causal pathways of the new psychosocial risk factors for back pain.