BP: Back Pain; CLBP: Chronic Low Back Pain; EDA: Exploratory Data Analysis; HRA: Health Risk Appraisal; LBI: Low Back Injury; LBMD: Low Back Musculoskeletal Disorder; OSHA: Occupational Safety and Health Administration; SOBP: Self-reported on-going back pain; SPSS: Statistical Package for Social Science; WC: Workers’ Compensation; WPS: Workplace Preventative Strategies

Back Pain (BP) is by far the most prevalent and costly musculoskeletal disorder among United States (U.S.) industries today. BP, also referred to as Low Back Musculoskeletal Disorder (LBMD), is defined as “pain in the lower back area that can relate to problems with the lumbar spine, the discs between the vertebrae, the ligaments around the spine and discs, the spinal cord and nerves, muscles of the low back, internal organs of the pelvis and abdomen, or the skin covering the lumbar area” [1]. The duration of LBMD varies from a few days, more than a few days to a few weeks (acute or short-term BP) or persists for more than three months (chronic BP) [2]. Eighty percent of people will experience BP over their life time [3]. BP is the fifth most common health problem for physician visits in the U.S. [4,5], with 26% of American adults reporting pain on at least one day every three months [6]. LBMD creates a substantial personal, community, and financial burden [7-9] where the direct and indirect costs incurred cost Americans approximately $50 billion each year [10]. In an occupational setting, LBMD accounts for a significant amount of pain and suffering, and workers’ compensation (WC) claims which often lead to an increase in worker absenteeism rates [11]. LBMD claims are the most common category of WC losses, accounting for 15-25% of all claims and up to 40% of costs [12,13]. The recurrence rate of LBMD is significantly high as reported in various studies, with a lifetime recurrence rate even higher ranging from 70-80% where 60-70% need up to six weeks to recover from back pain, and 80-90% need up to 12 weeks [14].

Research has shown that when ergonomic prevention strategies are applied appropriately, often times they can result in substantial cost savings for companies [15-18] and strong empirical evidence suggests that early prevention and intervention are more effective at preventing chronic pain and disability than attempts to treat pain and disability once it has been established [19]. Available treatments for LBMD focus on detecting relevant subgroups of patients with BP with a different prognosis and susceptibility to specific treatments [9]. However, the cause of LBMD problems remains obscure in most patients, and the generalized primary prevention does not appear to be feasible [20,21]. It is the author’s intention to make a contrast between single-cause models of health accident outcomes and the need for approaches that are more complex to raise awareness for prevention or mitigation methods, as other studies have done [22-27].

A range of individual conditions, such as psychosocial and occupational factors, have been identified as risk factors either for the occurrence of LBMD or for the development of chronicity [9,28,29]. However, it is not necessarily clear whether an individual difference is a cause or an effect. Hence, LBMD is best explained in relation to three categories, which are the topics of the author’s three-part series: low back injury (LBI) documented in WC claims [30], Chronic Low Back Pain (CLBP) diagnosed by a medical professional [29], and selfreported on-going back pain (SOBP). The goals of this manuscript are to: 1) Outline the occupational and lifestyle risk factors that may contribute to SOBP; and 2) Discuss a more in-depth understanding of LBMD based on the risk factors previous determined for CLBP and LBI based on a similar population and the risk factors for SOBP from this study.

The findings from this study are focused on making a contribution to Workplace Prevention Strategies (WPS) to help catch LBMD early, when treatment is most effective, resulting in healthier employees, higher productivity [31], fewer sick days [32], and a greater sense of well-being [33]. This study does not attempt to present a hypothesisdriven model (testing specific factors) to understand the risk factors associated with SOBP but rather a visual picture based on a data mining process. This process provides an opportunity to “learn from data” where information (i.e., important patterns and trends) are extracted from a data set and transformed into an understandable structure for future use.

Study approach

The approach used an analytic process outlined by Bidassie [34] to explore the large amounts of retrospective data available in the Health Risk Appraisals (HRA) [35], WC and Occupational Safety and Health Administration (OSHA) Logs dataset; in search of consistent patterns, and/or systematic relationships between variables based on the conceptual principles of statistics including the traditional Exploratory Data Analysis (EDA).

This SOBP study used the “potential effects” fatigue, stress, physical health (e.g., bodily pain), and emotional health (depression and anxiety) as outlined by Bidassie [34] to understand the association of occupational and lifestyle risk factors associated with SOBP. Depending on an individual’s physical [36] or emotional health [29], potential effects may contribute to impairment [37,38] or inhibit concentration [39].

While this model gave insight into valid predictions, it did not identify the specific nature of the interrelations between the risk factors. The focus produced a solution approach (the role of the input variables in explaining the outcome in a search for a parsimonious model involving a subset of the variables) that can generate useful predictions in future studies, rather than determining the nature of the underlying functions or the types of interactive, multivariate dependencies between risk factors. The model can be used for classifying highrisk groups for guiding early-detection screening for SOBP among university employees’.

Conceptual framework

Figure 1 is a modification based on the original conceptual framework outlined by Bidassie [29,33,34] their time and engagement with this study.

Figure 1: Systematic problem solving process.

Similar to employees with LBI and/or CLBP, employees with BP may experience difficulty in the execution of tasks in the workplace, and inability to participate in social activities and routine work both in and outside of the workplace [29,30,34,37-39].

Study design

The Human Resource Services at a Midwest university unveiled a university-wide health improvement initiative for benefit-eligible faculty, staff, and their spouses [29,34]. Participants were given the opportunity to complete a voluntary standardized HRA questionnaire to evaluate their health risk factors [34]. Once the employee completed an HRA and a wellness screening, he/she received a financial incentive, a personal health report, one-on-one telephonic coaching for health and lifestyle risk factors, and additional print resources.

This retrospective study was based on three years of data from January 2006 through December 2008. Available WC claim and OSHA 300 Logs data for employees who completed an HRA were also included. In the United States, an approved WC claim allows an employee who is injured at work or acquires an occupational disease to receive benefits; including wage replacements, medical treatment, vocational rehabilitation and others [40]. When an injury occurred in the workplace (e.g., LBI), an authorized safety professional documented the first incident and cause of the injury in the OSHA 300 Logs. If the incident received an approved WC claim for LBI, the case was added to the study. Details of the incident location, a reported cause of the injury, and a description of the nature of the injury were stored in the OSHA logs data. This research was approved by the university’s Institutional Review Board (IRB).

Study sample

From January 2006 through December 2008, 9,149 employees participated in the university’s work-life program, of which 15.5% (n=1,414) reported that they had an on-going problem with back pain that was serious enough to interfere with their daily activity. These people were then compared to the remaining 84.5% (n=7,735) of the employees who also participated in the university’s work-life program and reported that they did not have an on-going problem with back pain that was serious enough to interfere with their daily activity.

Data collection

This study modeled the data collection process outline by Bidassie [34]. The encrypted de-identified final dataset used in the analysis consisted of OSHA logs, WC claims, and HRA data for the employees who participated in the university’s work-life program from January 2006 through December 2008.

OSHA 300 logs: OSHA 300 logs provided information on employees’ job status, department, date of injury or onset of illness, location where incident occurred, description of injury or illness, cause of accident, type of injury, eligibility of injury for WC, job transfers, missed or restricted workdays, and employee death [11].

Workers’ compensation (WC) data: WC data provided the following data: age, gender, employment status, job status, marital status, number of dependents, years of work experience, cause of workplace injury, and part of body affected, date of workplace injury, lost days and WC paid.

HRA data: The HRA data provided biometric data and the following lifestyle risk factors: 1) Occupational and Lifestyle Risk Factors: sociodemographics, occupational, physiological, psychological, psychosocial factors, family health history, alcohol use, smoking or tobacco use, sleeping habits, self-care, suggested examination/immunization, medication usage, physical activity, and attitude toward daily safety precautions; 2) Biometric data: height, weight, cholesterol, glucose, and blood pressure measurements; 3) Potential Effects: Fatigue, level of stress from minor annoyances to fairly major pressures, problems or difficulties; physical health (perceived health and physical condition), and emotional health (feeling depressed, down or hopeless); and 4) Impairment: Emotional health and/or physical capability.

Framework for statistical analyses

Dependent (response) variable: The dependent variable had two-response option: “Do you have an on-going problem with back pain that is serious enough to interfere with your daily activities?” The responses are Y=1 (reported have on-going problem with back pain was serious enough to interfere with daily activities) and Y=0 (they did not have an on-going problem with back pain that was serious enough to interfere with their daily activity).

Independent (predictor) risk factors: Independent risk factors included: demographic information, occupational factors, and nonoccupational factors as listed in section 2.5.

Statistical analysis

The data mining process using the Statistical Package for Social Science (SPSS) 16.0.1 consist of three stages: (1) The initial exploration; (2) Model building or pattern identification with validation/verification; and (3) recommendations in deployment (i.e., the application of the model to new data in order to generate predictions).

Stage 1: Stage 1 (Exploration) focuses on data preparation which may involve cleaning data, data transformations, selecting subsets of records since our data sets consisted of large numbers of variables. To bring the number of variables to a manageable range, this first stage may involve anywhere between a simple choice of straightforward predictors for a regression model to elaborate exploratory data analysis using a wide variety of graphical and statistical methods (e.g., descriptive and t-test) to identify the most relevant variables and determine the complexity of the models that can be taken into account in the next stage.

Descriptive statistics are used to describe the main features of a collection of occupational and lifestyle risk factors (predictor variables) with the aim to summarize this sample, rather than use the data to learn about the population that the sample of data is thought to represent. A two-sample t-test is used to compare means to determine if two sets of data are statistically significantly different from each other. Pearson Chi-square (χ2) tests are used to determine the relationship between SOBP (y) and predictor risk factors (x’s) with two or more categories.

Stage 2: Stage 2 (Model building and validation) focuses on an elaborate process of applying different models to the same data set and comparing their performance to choose the best model based on their predictive performance (i.e., explaining the variability in question and producing stable results across samples). Backward stepwise logistic regression method factors [41,42] (a model for classification rather than regression) will be used to predict the probability [43,44] whether an employee has SOBP and measure the relationship based on observed characteristics of the individual; for example: age, sex, body mass index, blood cholesterol level, systolic blood pressure, relative weight, etc. [41,45,46] using probability scores as the predicted values of the dependent variable being positive [47-53]. If the odds of SOBP increase when the predictor risk factor (independent variable) increases, this is signified by an odds ratio greater than one. Conversely, if the odds of SOBP decrease when the predictor risk factor increases, this is indicated by an odds ratio less than 1 [54,55]. A probability level of p<0.05 was considered statistically significant. Lastly, risk factors in the final model with β>1 will be considered manageable risk factors that are recommended to be considered and incorporated into preventative strategies in the workplace.

Stage 3: Stage 3 focuses on the understanding of LBMD by comparing the risk factors from the LBI model [30], the CLBP model [29] and the SOBP model to gain insight into LBMD for consideration for future BP studies.

The following are the results of the statistically significant risk factors that were considered the model based on 56 risk factors investigated in stage 1 to determine each association with SOBP. Detailed statistics of all the variables are included in the Appendices.

Risk factors for SOBP

Employees with SOBP tended to work as service and operations staff, worked the evening shift and their regular job required regular lifting at work (Appendix 1). They tended to consume more than one alcoholic drink per day and tended to smoke one or more packs of cigarettes a day. Their physical condition tended to limit their ability to get physical exercise (moderate, vigorous, strength-building); however, they tended to do stretching exercises to improve flexibility. The number of hours that participants slept varied (Appendix 2). Biometric indices such as high blood pressure and body mass indexes in the obese range were also associated SOBP (Appendix 3). The majority rated themselves to be in poor to fair health; they tended to have been diagnosed with more than 3 chronic diseases, (i.e., arthritis, low back pain, insomnia), suffered from bodily pain and reported other on-going problems with wrist pain, tingling and numbness, and neck pain (Appendix 4). Participants’ daily lives were also affected by their health conditions (Appendix 5). Participants suffering from SOBP tended to have the knowledge about how to treat CLBP and reported regular medication usage, such as use of prescriptions, non-prescriptions and/or herbal remedies (Appendix 6).

Participants with SOBP tended to be more stressed than participants with no SOBP. They reported an average of three major sources of stress (SD=2) with the most commonly cited stressors being financial difficulties, work responsibilities & relationships, death and/ or family illness, care of love ones, and coping with stress (Appendix 7). They indicated feeling depressed, feeling down or having a lack of interest or pleasure in doing thing (Appendix 8).

Backward stepwise logistic regression analysis

The final model for SOBP (Table 1) consists of 22 risk factors with coefficients (β)≥0.1; containing 18 socio-demographic, occupational, lifestyle and physical and emotional health risk factors, and four potential risk factors (χ2=2593.99, df=21, p<000). Five (5) risk factors appeared to have the most impact in this study of SOBP: diagnosis of CLBP, persistent neck pain, persistent wrist pain including sensations of tingling or numbness, previous LBI, and on-going bodily pain.

Table 1: Risk factors associated with SOBP.

Figure 2 represents the final model of occupational and lifestyle risk factors for SOBP employees’ serious enough to interfere with their daily activities.

Figure 2: Risk Factors associated with SOBP (Socio-demographic, occupational and non-occupational risk factors). †p<0.06; *p<0.05; **p<0.01; ***p<0.0001.

Insight into LBMD

It is important to note that LBMD is a combination of key risk factors within a complex system that consists of SOBP, CLBP and LBI. When we view the multifaceted LBMD (SOBP, CLBP and LBI) we can see that the risk factors are: gender, lifting/twisting/bending, stress, person with high blood pressure, physical health (bodily pain), emotional health (level of depression) and fatigue. While it is important to be specific when labeling the different facets of LBMD, because they all have unique risk factors, these are the ones in common. Table 2 shows the complete comparison of the risk factors of SOBP, CLBP and LBI

Table 2: Comparison of risk factors for LBMD (BP, CLBP, LBI).

The findings in this study offer a new perspective on the association of lifestyle risk factors for SOBP serious enough to interfere with daily activities and the risk factors associated with LBMD. These may help explain why traditional engineering solutions [56] may not have the desired impact. It also supports the findings in the literature where a comprehensive understanding of LBMD cannot be based solely on simple one-to-one relationships, but rather the combination of key risk factors within a complex system, and its relationship to the wellness and safety of the entire human system [28]. However, there seems to be an indication that people with SOBP serious enough to interfere with their daily activities tend to have greater odds of have had previous or may have future LBI or CLBP. Preventative strategies focus on the risk factors for SOBP, may reduce the incidence for future LBI and CLBP.

The results in this study support the following finding in the literature. The impact from psychological factors in the development of LBMD include depression, anxiety, passive coping strategies, and workrelated factors such as high physical job demand, low expectation of return to work, low job satisfaction, low social support, and perception of stress at work [57]. Especially noticeable are the reported physical conditions relevant to muscle, joints and skeleton problems, such as CLBP, neck pain, bodily pain, wrist pain, tingling and/or numbness, and fatigue [29,58]. BP could be affected by psychosocial factors like lack of social activities and contact/support from friends and relatives, but not significantly. Fatigue may come from three sources: time on task, lack of sleep or sleep interruption, and a justification for escaping from a task that could be stressful or onerous in other ways, e.g. a stress reaction [59-66].

In an attempt to increase employees’ awareness of ergonomics and physical work environment to prevent SOBP and LBI, employers may consider integrating lifestyle preventative strategies into their traditional engineering preventative strategies.

Due to the limitations of this retrospective dataset, each of the LBMD categories (SOBP, CLBP, LBI) was studied individually. Also, our analysis does not permit causative conclusions. Future studies should be designed so that more sophisticated statistical models can be applied. Other limitations are that the HRA was pre-designed, and there was a bias towards participants with an approved WC claim who also participated in a wellness program.

Findings from this study can be used to revise traditional approaches to workplace ergonomics and wellness programs and promote a new focus on the health and lifestyle risk factors associated with LBMD. In order to get a comprehensive understanding of the manageable risk factors associated with LBMD, it is suggested that future studies on occupational preventative strategies should include both lifestyle risk factors and occupational risk factors along with the impact of a previous LBI and/or CLBP (diagnosed by a medical doctor). In sum, in addition to implementing fitness programs and facilities, smoking cessation programs, and obesity programs in the workplace, employers should also offer programs designed to address the risk of emotional stress, improve attention and alertness, increase employee performance to minimize the risk of LBMD.

The author would like to thank Amanda Kovach for her work in editing and proofreading the manuscript. Additionally, the author would like to thank the following people for their support and contributions to this research project. Dr. James D McGlothlin, MPH, PhD, CPE, for the opportunity to work on this purposeful research topic. Mindy Paulet and Dr. Joseph Thomas for providing access to the data from the University’s wellness program. Teresa Wesner from Human Resources authorized research funding for two years to work on this project. Kristina Evans and Steve Gauger from the Radiological and Environmental Management provided access to the OSH 300 logs data-set. Tammy Synesael and Deborah Pope from Human Resources provided access to the workers’ compensation data.