Development of a regression model using socio-demographic, behavioural and occupational risk factors for lumbar disc herniation (LDH) and lumbar disc degeneration (LDHD)-A case control study.

Background: Although many studies have been conducted on risk factors associated with lumbar disc herniation (LDH), only few studies reported on the association of these factors in comparison to LDH and lumbar disc herniation and degeneration (LDHD). There are no reported studies on a regression model incorporating these factors. As the risk factors are better described in regression models, present study aimed to develop a regression model associated with LDH and LDHD in relation to socio-demographic, behavioural and occupational factors. Methods: A case control study conducted using 104 cases with LDH and controls (n=104) without LDH. Pre-tested questionnaire was administered to all participants to gather information. Results: Among the cases with LDH, 35.6 % presented with LDHD while 64.4 % had only LDH. Among the socio-demographic characteristics, body mass index <25 kgm -2 was a signicant protective factor for both LDHD (OR=0.31; 95% CI=0.13-0.72) and LDH (OR=0.39; 95% CI=0.20-0.77). Involvement in daily activities with heavy (OR=5.1; 95 % CI=2.1-11.8) and moderate strain (OR=3.1; 95 % CI=1.5-6.6) to back, sitting more than eight hours per day (OR=5.1; 95 % CI=1.0-25.7), smoking (OR=5.0; 95 % CI=1.5-16.4) and sleeping in supine position (OR=2.09; 95% CI=1.09-4.06) were signicant risk factors for LDH. Only daily physical activities with heavy strain act as a signicant risk factor (OR=3.1; 95 % CI=1.1-8.5) for the development of LDHD. Types of mattresses used did not have signicant difference among cases and controls. Majority of cases (56.7 %) did not know the causative factor that led to LDH. According to the regression model, BMI, smoking and involvement in physical activities with moderate and heavy strain to back were considered as signicant risk factors for the development of LDH or LDHD. Conclusion: BMI, smoking and daily physical activities with moderate and heavy strain to back are signicant risk factors for development of LDH or LDHD in regression model.


Introduction
Lumbar disc herniation (LDH) is considered as a major socio-economic problem not only in Sri Lanka but also in many developed countries. Although there are numerous determinants for lower back pain (LBP), lumbar disc herniation (LDH) is the most common concern [1,2]. Evidences suggest that annual cost spent on LBP associated with LDH in United States exceeds 100 billion dollars each year whereas in United Kingdom it is estimated as 12 billion pounds per annum. In Netherlands it was reported as 1.7% of the gross national production, these data support the evidence that LBP is a major burden for the economy of the country [2,3]. Studies have reported that 60-80% of the population have suffered of LBP associated with LDH at least once during their lifetime [2,4]. In addition to the direct costs, indirect costs such as, signi cant percentage of sick leaves among affected employees, costs of lost wages, reduced productivity, psychological distress and costs for additional care given are major problems linked with LDH associated LBP [1]. It is suggested that LDH is one of the main determinant in limiting activities among adults below 45 years of age [5].
Despite several studies carried out worldwide to determine the cause for LDH, yet the exact cause/s are unknown [4,6]. It is hypothesized that some conventional factors such as age, gender, severe mechanical and physical loading, trauma, strenuous sporting activities, vibrations and smoking as main causative factors for LDH [7][8][9][10]. Evidences suggest that degeneration usually start at a very early stage of life where mild changes could be seen in rst decade of life and more signi cant changes from second decade onwards. However, studies emphasized that LBP associated LDH are most common during fourth to fth decade of life [3,[11][12][13]. Furthermore, studies carried out on LDH have indicated that males are more vulnerable for LDH compared to females [13][14][15]. Further, evidences con rm that there is a signi cant association of sporting activities with LDH [16][17][18].
Although there is an increased trend for hospitalization of patients with LDH, the information pertaining to behavioural and occupational risk factors that lead to LDH is not available in Sri Lankan context. Thus, the present study was carried out to identify the association of selected socio-demographic, behavioural, sports and occupational factors contributing to LDH. Although several biochemical factors have been considered in the risk of developing LBP associated with LDH, there were less information available on the effect of different rmness of mattresses and sleeping positions with this regard. Therefore, more importantly the present study also focused on sleeping positions and the sleeping systems used by the study participants as these factors were not considered much in reported literature pertaining to LDH, but conventionally considered as crucial factors for LBP associated with LDH. Further, there are no reported studies on development of a regression model incorporating these factors. As regression models give a better view on risk factors, present study aimed to investigate possible socio-demographic, behavioural and occupational factors and development of a regression model associated with LDH and lumbar disc herniation and degeneration (LDHD).

Study design and setting
This was a case-control study. Cases were recruited from a hospital in the capital of Sri Lanka which drains patients from all over the country and thereby representing almost all districts of Sri Lanka.
Controls represented several districts of Sri Lanka. The analysis was carried out at University of Sri Jayewardenepura Sri Lanka. Ethical approval was obtained by the Ethics Review Committee of Faculty of Medical Sciences, University of Sri Jayewardenepura, Colombo, Sri Lanka (29/14). After detailing out the study protocol, informed written consent was obtained from all participants.

Study population
The 104 cases were patients who had low back pain with lumbar disc herniation con rmed by Magnetic Resonance Image (MRI) by a consultant neurosurgeon and consultant radiologist. Inclusion criteria for controls (n=104) were adult volunteers without low back pain at least for the past one month period of the study and did not have LDH. Both case and control subjects were between 18-74 years of age. The concomitant presence of other bone disorders such as osteoarthritis, osteoporosis and pregnancy and, malignancies were exclusion criteria for both cases and controls, while cases with trauma and accidents related LDH were also excluded.

Data collection
A pre-test was carried out to validate the questionnaire (data not included in the study). A standardized, interviewer administered questionnaires lasting 10-15 minutes was administered to each patient by the principal investigator enquiring their demographic data, daily activities, physical and behavioural activities, occupational status, sleeping pattern, type of mattress used, general health status and current health condition. Major causative factor stated by the patients for disc herniation was noted from the clinical history. Participants were given the opportunities to clarify their doubts in a familiar and comfortable language of the individual (English, Tamil or Sinhala).
Age had been calculated to the nearest completed year and educational levels were categorized according to the national criteria. BMI was calculated according the standard formula and height of each participant was measured without shoes in straight standing style and when the heels of the foot, back of the hops and head was stuck to the wall using a standard stadiometer with a moveable ruler to the closest 0.1 cm while weight measured without shoes, using an electronic weighing scale to the closest 0.1 kg. Smokers were de ned as individuals who smoked any tobacco in the past twelve months of the study and included those who stopped smoking within past year and consumption of alcohol was de ned as individuals who consumed alcohols usually or occasionally in the past twelve months of the study and also included who stopped consuming alcohol within past year [19]. Severity of daily physical activities were categorized according to the Yusuf et al (2004) and Seidler et al (2003) [20].

Statistical data analysis
Data were coded and captured on Excel and SPSS version 20.0. Frequencies and percentages were calculated for all data. Odds ratio was calculated to assess the risk factors for LDH in the study population and p value ≤ 0.05 was considered statistically signi cant. Risk estimates for developing either LDHD or LDH was calculated using logistic regression analysis. Crude odds ratio (OR) and 95 % con dence interval (CI) were calculated for age, BMI, gender, marital state, education and employment at the enrollment for the study using chi-square test. An unadjusted ORs were calculated for severity of daily physical activities, occupational exposure, engaged in sports, duration of sitting, smoking and consumption of alcohol. The exposure categories (0, 1, 2 etc.) were included as internal scaled variables in logistic regression model. Measures of exposure were based on self-declaration of the subjects. For the purpose of analysis, cases were further categorized into two sub groups; patients who presented with both degeneration and herniation (LDHD group) and patients who presented only with lumbar herniation (LDH group). Missing values were analysed as a separate category (results not shown here).

Results
Socio-demographic characteristics of the study subjects As investigators have checked the eligibility criteria for both controls and cases prior to the study by clinical examination (conducted by a consultant radiologist and a consultant neurosurgeon) and interviewer administered questionnaire, all eligible study participants (cases=104 ; controls=104) were included for the analysis of all variables mentioned except for variable BMI (Table 1) and sleeping positions ( Table 2).
Socio-demographic characteristics of the study subjects are presented in Table 1. Disc herniation and degeneration was present in 37 (35.6%) of the cases (LDHD) and rest of the subjects had only LDH (64.4 %). In cases, the mean age (SD) for subjects with LDH only was 41.5 (±14.8) years, while subjects with LDHD was 47.4 (±17) years. Control subjects showed a mean age of 43.2 (±15.2) years. Majority of the cases with LDHD (51.5%), LDH (64.2%) and controls (60.5%) were less than 50 years of age. However, there was no signi cant difference in the age between LDHD group (OR=0.69; 95% CI=0.32-1.46) and LDH group (OR=1.16; 95% CI=0.62-2.20) when compared to that of control group (Table 1).
Response rate for BMI in controls, cases with LDHD and cases with LDH was 95.1%, 83.8% and 86.6% respectively. Cases were reluctant to participate in height and weight measurement due to the severity of pain where control subjects declined to remove foot ware for the weight measurement were the limitations for calculating BMI. It was recorded that majority of cases with LDHD (67.7%) and LDH (62.1%) were in over weight and obese category (≥ 25 kgm -2 ). The mean (SD) BMI in both LDHD and LDH groups was above the normal limits, 39.1 (27.5) kgm -2 and 36.7 (24.3) kgm -2 respectively. The BMI of the control subjects (27.1±16.6 kgm -2 ) was also above the normal limits but signi cantly lower than the case subjects (LDHD and LDH group). BMI less than 25 kgm -2 acts as a protective factor for LDHD (OR=0.31; 95% CI=0.13-0.72) and LDH group (OR=0.39; 95% CI=0.20-0.77). Signi cant difference was not observed between groups for gender difference. In this cohort of study, majority of the subjects in both cases (LDHD=81.1%; LDH=77.6%) and controls (73.1%) were married. Similarly, there was no signi cant difference in marital status among the study groups. Majority of the subjects in cases (LDHD=67.6%; LDH=73.1%) and controls (62.5%) had secondary or higher level of education with no signi cant difference between study groups. In all study groups higher percentages of subjects were employed with LDHD (64.9%), LDH (65.7%) and controls (80.8%) respectively. Employment at the time of enrollment did not indicate a signi cant difference among the three study groups (Table 1).

Behavioural and occupational physical work load
It was reported that LDHD and LDH were associated with behavioural and occupational physical work load that caused strain to the lower back. When occurrence of LDH was assessed in relation to behavioural and occupational risk factors among cases and controls, majority of the cases (LDHD=56.7%; LDH=68.6%) were engaged in activities which causes heavy or moderate strain to back when compared to controls (36.5%) ( Table 2). The odds ratio (OR) for daily activities with heavy strain to back yielded a statistically signi cant risk value of 3.1 (95% CI=1.1-8.5) in cases with LDHD and odds ratio of 5.1 (95% CI=2. 1-11.8) in cases with LDH only. Although daily activities with moderate strain to back did not show a statistically signi cant association for LDHD group, it appeared as a risk factor for the cases with LDH only , with OR of 3.1 (95% CI=1.5-6.6) ( Table 2).
Results also indicated that engaging in sports and duration of sitting does not have a signi cant association in patients with LDHD. However, sitting more than 8 hours a day revealed a statistically signi cant OR of 5.1 (95% CI=1.0-25.7) in subjects with only LDH in Table 2.
Smoking showed a signi cantly elevated OR of 5.0 (95% CI=1.5-6.4) in patients with LDH only which acts as a strong contributory risk factor for LDH. Although there was a risk OR of 3.1 (95% CI=0.78-2.8) in patients with LDHD, it was not signi cant. Consumption of alcohol was not a signi cant risk factor for both LDH and combined degeneration ( Table 2). According to the regression model, BMI has been considered as a signi cant risk factor for the development of lumbar spine diseases for both LDHD (OR=1.02 (95% CI=1.0-1.04)) and LDH (OR=1.02 (95%CI=1.02-1.04)). Smoking and heavy to moderate strain to back revealed statistically signi cant ORs of 6.44 (95% CI=1.69-24.51) and 3.36 (95% CI=1.57-7.09) in the regression model for patients with only LDH (Table 3).

Discussion
Evidences suggests that degeneration starts at a very early stage of life where mild changes are seen in the rst decade of life and more signi cant changes from second decade onwards [3,11,21]. It is reported that LBP, lumbar disc herniation and degeneration are common in the fourth to fth decade of life [5,12,22]. One study has indicated that mean age for LDH as 37 years [12], while other studies have reported mean ages as 45 ± 13 years and 42 ± 10 years [15] and 41 ± 10 years [21]. Mean ages in all study groups of the present study were in fth decade of life which was similar to the previous reported ndings mentioned above. However, contrast to our ndings, one study has recorded 61 -70 years as the peak age for LDHD in both genders [23].
We observed a signi cant difference in BMI between cases and control indicating majority of the patients in LDHD group (67.7%) and LDH group (62.1%) were in overweight or obese categories according to BMI. However, control group had 51% subjects with normal BMI. Although there were many heterogenous data available regarding the association of BMI and LDH, majority of the data emphasized that increased BMI or obesity is a risk factor for LDH. Present study also con rms the above fact as BMI less than 25 kgm -2 as a protective factor with odds ratio of 0.31 (95% CI=0.13-0.72) in LDHD group and 0.39 (95 % CI=0.20-0.77) in the LDH group. Studies conducted on histological assessment of intervertebral disc tissue further con rmed that high degree of degeneration is also associated with elevated BMI [23]. As overweight and obesity encounters an increased pressure and weight on the intervertebral tissue thus, initiate herniation and degeneration of the intervertebral discs.
Study ndings regarding gender and LDH in the present study are in accordance with reported similar studies. A study conducted with 205 surgical patients reported that men to women ratio in patients who are undergoing lumbar surgery was 1.5:1 in surgical setting [13]. This was in accordance to the study carried out by Kelsey and co-workers (1984). However, in non-surgical setting it was reported that men to women ratio was1:1. In addition, another study showed that prevalence ratio for male:female was 1:0.61 with a signi cance of p=0.0001 [23]. Similar observations were noted in a study which recruited 48 patients with LBP. Above study a rmed similar male prominent gender distribution with 67% males and 33% females [15]. Further, a reported study has also indicated that LDH is found in 4.8% men over 35 years and 2.5% women over 35 years suggesting that men are more prone to LDH [12]. The present study nding in Sri Lankan subjects with LDH also adds to the study ndings that males are more prone to develop LDH compared to females.
Majority of cases (both LDHD and LDH groups) and controls had secondary or higher educational level.
Experts suggest that subjects who are employed with higher education level having more sedentary life style and lack of exercise on back muscles weaken the power of the muscles. This could trigger the herniation of the intervertebral disc, when sudden load to the vertebral column increases. However, according to the present study control group also had a good educational level, hence this phenomenon cannot be applied to the present scenario.
In the present study there was a signi cant difference in smoking among cases and controls (p=0.012) with high frequency of smoking reported in cases (16.3%) compared to controls (4.8%). Our ndings are in accordance with previous studies which a rm the association between smoking and LDH. Studies have reported that smoking in past years is associated with increased risk of LDH [14]. Further studies have highlighted that nicotine in cigarettes may cause narrowing of blood vessels hence impair the blood ow to the disc tissue causing disc degeneration [22,24]. A twin study reported by Battie et al (1995) remarked that there was 18% greater mean disc degeneration scores in lumbar spine of smokers when compared to non-smokers. Interestingly, a study has stated that smoking cannot be regarded as a risk factor for disc degeneration although there was considerable percentage (41 %) of smokers in the study [25]. Therefore, this present study nding on smoking further adds evidence to previously reported studies on the positive association between LDH/LDHD and cigarette smoking. Further, studies have identi ed that intervertebral disc being the largest avascular tissue in the human body, narrowing of blood vessels by nicotine can interrupt the diffusion process via cartilage end plate, thus leading the disc to degenerate.
To further strengthen the study, the present study also attempted to distinguish the relationship between the sleeping postures and type of mattress used in LDH subjects. These factors are considered as critical conventional factors contributing to LBP associated with LDH. However, present study did not nd any signi cant association with types of sleeping systems used and LDH/LDHD. However, there were limited literature on these parameters. A study conducted in 313 adults with LBP has proven that medium form mattresses had better outcome for pain while in bed (OR=2.35; 95% CI=1.13-4.93) compared to the pain on rising on the same mattresses type (OR=1.92; 95% CI=0.97-3.86) than in patients using rm mattresses. Finally, authors have concluded that medium rm mattresses could improve the pain and disability in patients with chronic lower back pain [26]. Further, it was also believed that mattresses with soft surfaces increase LBP due to incorrect support to the vertebral column and decrease the quality of sleep [27].
It is believed that loading of the intervertebral disc as an important factor which determines the LDHD and LDH. Therefore, different impact on the disc by different sleeping postures could not be disregarded in the etiology of LDHD and LDH. However, studies done on direct measurement of spinal loading is limited and studies on sleeping postures are scarce. Interestingly, present study has observed that sleeping in supine posture as a signi cant risk factor with odds ratio of 2.09 (95% CI=1.09-4.06) in patients with LDH. However, this phenomenon could not be observed in patients with LDHD. It was stated that proper sleeping system could align the spine on to its neutral posture as do in upright position, whereas non-neutral postures can apply unbalanced loading on intervertebral discs and facet joints. Further, intervertebral discs tend to restore and grow through hydration during sleeping. As the gravity changes during sleeping, intervertebral disc tissues are unloaded and can rehydrate to restore its elasticity [28].
Therefore ndings related to sleeping postures and type of mattress used adds valuable insight to the studies on risk factors associated with LDH and LDHD.
In the current analysis of the study, we specially focused on the association between physical workload and LDH. Accordingly, results of this case-control study on occupational risk factors associated with LDH are well correlated with the reported studies on similar theme. Heavy physical work such as lifting and carrying heavy objects are proposed risk factors for LDH associated LBP [16]. Another study on identical twins also found similar ndings [29]. It is also reported that heavy lifetime occupational and physical loading have an association with disc degeneration in upper lumbar levels (p=0.055 -0.01) whereas sedentary work was associated with less signi cant degeneration (p=0.006) [29]. Contrast to our ndings, observations by a different study conducted in monozygotic twins stated that there was no signi cant difference observed in the level of leisure time physical activities when the monozygotic twins were compared to entire twin cohort in Finland [30]. Similarly, a review has shown that workers with many sedentary activities had higher prevalence rates for LBP symptoms and sick leaves due to LBP [OR=1.46; (95% CI=1.18-1.29) for sedentary leisure activities)]. They have also indicated that physical activities in leisure time (either sports or daily physical activities) do not associate with prevalence rates for low back morbidity [31]. Review study concluded contradictory ndings stating that sedentary lifestyle and leisure time is not associated with LBP [32]. Our study further con rmed that severity of daily physical activities causing strain to back have a considerable effect on LDHD.
Occupation was recorded as a risk factor by Manek and MacGregor (2005). The authors stated that occupations with night shifts, lifting, bending, twisting, pulling and pushing favours LDHD [16]. According to the present study authors found heavy lifting, bending and twisting as severe or moderate risk occupations that had a strong signi cant association with LDH [OR=5.96 (95% CI=1. 22-29.18)]. Another study also emphasized that main causes for LBP associated with LDH in work place are heavy lifting, repeated loads from manual handling, work postures incurring postural stress and whole body vibrations [33]. Contradictory to our ndings a twin population study stated that there is no signi cant association with occupational loading and LDHD [34]. Therefore, our ndings with perceived work strains on LDH cannot be disregarded.
There are several reported literatures that suggest the relationship between sports and LDHD. Hence, present study also hypothesized sports as a contributory factor for LDH. However, authors could not nd signi cant association with LDH and sports. According to published literature, evidences have stated that there was high incidence of radiographic abnormalities of spondylolysis in college level football players (80.5%) [16]. In addition, above study also stated that spondylolysis as a signi cant risk factor for LBP in football players. Observations from another study was in agreement with previous studies stating that football players were at increased risk of developing LBP and disc degeneration [17]. Another study conducted in Japan among rugby players (n=327) also supported the above relationship of LBP and strenuous sporting activities. That study also emphasized radiographic abnormalities seen in spondylolysis as a signi cant radiological risk factor for LBP in high school rugby players [18]. A similar study conducted in elite athletes also revealed that disc degeneration is signi cantly higher in elite athletes (75%) when compared to non-athletes (31%) [35]. However, a similar case control study carried out in former elite athletes showed that odds ratios for back pain was signi cantly lower among athletes than among control subjects suggesting contradictory ndings of the above report. Authors have stated that LBP is less common in athletes when compared to control subjects [OR=0.62; (95 % CI=0.37 -0.98) for endurance sports: OR=0.60; (95% CI=0.44-0.82) for sprinting and games: OR= 0.67; (95% CI=0.47-0.96) for contact sports such as wrestling and boxing]. The study further commented that maximal weight lifting is associated with disc degeneration of the entire lumbar spine, whereas soccer associated degeneration con ned to lower lumbar spine region only. Authors further emphasized that there was no accelerated disc degeneration in runners and shooters [36]. Number of factors could have interfered with the results of present study with LDH and sports. Majority of participants of the present study were unable to mention the duration of involvement in sports, reason for stop playing and unable to recall the speci c sporting activities they were engaged during school time. Therefore, these factors could have greatly reduced the speci city of sports de nitions and might also lead to numerous misclassi cations of the type of sports (strenuous sports or mild strenuous or etc.). Though there is no signi cant association between sports and LDH, present study could highlight some valuable information regarding sports and LDH. According to the history of involvement in sports among the recruited subjects in our study emphasizes that improper training or lack of back muscle strengthening exercise may attribute for sports associated LBP and LDHD in Sri Lankan context. Further, according to expertise experience it is hypothesized that people who have engaged in sports have developed a good muscle tone during the period of active involvement in sports, but when they quit or stop regular sporting activities the developed muscle tone will decrease and as a result when they participate in strenuous work or sports, the load that comes to the body will directly pass through the vertebral column without involvement of back muscles. Hence, the intervertebral disc tends to herniate which is enhanced by the excessive load that triggers degeneration.
Traditionally it was believed that traumatic occupations and heavy physical/mechanical loading were the major contributing factors that leads LBP and LDHD [7,9]. However, according to the present study more than half of the study subjects (56.7%) did not have any of the above predisposing factors associated with LDH. Therefore, it is suggested that there could be other factors associated with regular or occupational behavior that is related to LBP in this cohort of patients.
Recurrence of LDH in the present study was 13.5% and was in agreement to previous ndings of recurrence of lumbar disc disease (5-15%) [37,38]. However, above published studies further commented that there was no signi cant association of age, sex and level of herniation and the recurrence of LDH.
The limitations of the present study include a convenience sample with case-control study design. Secondly, assessment of BMI had a limitation as some of the cases were reluctant to measurement of height and weight due to severity of pain while some controls were also refused remove foot ware to measure the weight. Further, social behavior also had a limitation as it was based on direct questioning of the participants only. Also, self-reported data on sleeping posture and data on daily physical activities were regarded as limitations of the study. There are several notable strengths in our study such as assessing of sleeping postures, types of sleeping systems and developing of a regression model associated with LDH considered as some strengths. Though the sample size was adequate to detect the hypothesized effects of socio-demographic, behavioural and occupational factors associated with LDH among Sri Lankan subjects, large studies would add more comprehensive ndings in the etiology of LDH.

Conclusion
Physical strain, related to occupation and behavior, increased BMI, smoking and sleeping postures play vital role in pathogenic entities of LDH and LDHD. Remarkably, present study highlights that there was no signi cant association between type of mattress used by the study participants although these were considered as triggering factors for LDH and LDHD. According to the regression model BMI, smoking and daily physical activities with moderate to heavy strain to back are signi cant risk factors for development of LDH or LDHD.

Declarations Con icts of Interest
The authors report no con icts of interest either nancially or non-nancially. The authors alone are responsible for the content and writing of the paper.

Ethical approval
All procedures performed in studies involving human participants were in accordance with the ethical standards of the Ethics Review Committee, Faculty of Medical Sciences, University of Sri Jayewardenepura, Colombo, Sri Lanka (29/14) with the 1964 Helsinki declaration and its later amendments.

Informed Consent to participate
Informed consent was obtained from all individual participants included in the study.

Availability of data and materials
All data generated during this study are included in the published article.
Author's contributions LVA and SP conceived the study concept while all the authors contributed in the study design. NDW wrote the rst draft of the rst manuscript, conducted experimental studies and data analysis. SP supported in results interpretation. LVA, SP and HP contributed in manuscript editing. NDW and LVA are guarantors of the work and NDW had the full access to data and takes responsibility for the study.