Low back pain (LBP) and neck–shoulder pain (NSP) are very common disorders in industrialised countries today.1 2 These disorders are often associated with substantial consequences for both the individual and society.3^–5 Nonetheless, the aetiology of both LBP and NSP is to a large extent unknown, but is considered to be multi-factorial. Environmental factors (e.g., high physical workload) and individual factors (e.g., genetic influence) are both factors considered to impact the occurrence of LBP and NSP as well as the prognosis of these disorders.6

Many studies in the past decades have reported that high physical workload is of importance for the occurrence of LBP and of NSP.2 7^–9 However the results vary, often due to methodological diversity such as choice of study population, study design, and exposure assessment methods.10 11

It can be suggested that the associations between high physical workload and LBP or NSP might be confounded by genetic factors. The genetic influences on self-reported LBP and self-reported NSP have been investigated in earlier twin studies. All studies reported, to a various degree, a genetic influence on LBP, but for NSP the results were contradicting.12^–19 One earlier twin study that found an influence of genetic factors for the occurrence of LBP also concluded that there was an association between heavy workload and LBP. The latter by showing that the proportion of both monozygotic (MZ) and dizygotic (DZ) twins with back pain was significantly higher among those exposed to heavy workload.13

However, the authors have found only one earlier study that has investigated the association between high physical workload and LBP when taking confounding from genetic influence into consideration.20 Hartvigsen et al found that even when adjusting for genetic influences there was a graded association for increasing workload and LBP. No corresponding studies concerning NSP have been found.

Earlier studies have not taken into account the fact that among individuals with LBP there are many that have concurrent NSP and little is known about the influence of physical workload for the occurrence of pain in more than one area of the spine. A recent study found that the influence of genetic factors seems to be of greater importance for the occurrence of concurrent LBP and NSP, compared with for solely LBP and solely NSP (Nyman T, Mulder M, Iliadou A, et al. High heritability for concurrent low back and neck-shoulder pain – a study of twins. Manuscript submitted for publication, 2008). Sickness absence also seems to be more common among subjects with concurrent LBP and NSP.21 22 It might be that genetic and/or shared environmental factors have a greater influence for the association between high physical workload and subjects with pain in more than one area of the spine, than for the associations between high physical workload and subjects with solely LBP or solely NSP.

The purpose of the study is to investigate if high physical workload is associated with LBP and/or NSP when taking into account the influence of genetic and shared environmental factors. Further, the study aims to explore the potential influence of genetic and shared environmental factors in the associations between high physical workload and the three disorder subgroups: solely LBP, solely NSP, and concurrent LBP and NSP.

MATERIALS AND METHODS

Subjects

The Swedish Twin Registry (STR) is a national resource maintained at the Karolinska Institutet. It is the world’s largest registry of its kind and contains information on more than 170 000 twins born during 1886–2000.23

The present study was based on the Swedish Twin Study of Adults: Genes and Environment Study (STAGE-study). The study targeted 43 088 twins born during 1959–1985 who had not participated in any STR-administered data collection before. Data were collected through a web-based self-administered questionnaire and concerned among other things present illnesses, medical history, behaviour and occupational history. The data collection started in May 2005 and finished in June 2006. Those twins who did not answer the web-based questionnaire were offered a telephone interview or a postal questionnaire. The total response rate was 60%.

Pairs where at least one twin had answered that they had rheumatoid arthritis, systemic lupus erythematosus, or pelvic spondylitis were excluded from the study. This was because the aim in the present study was to investigate the association between high physical workload and non-specific LBP and NSP, and also because subjects with these rheumatic disorders are more vulnerable regarding exposures to high physical workload.

Subjects that were included in the present study were those complete 1802 MZ twin pairs, 1334 same-sex DZ twin pairs, 1282 opposite-sex DZ twin pairs, and 7271 singleton twins between the age of 21 and 47 years who had given complete answers on the questions concerning pain and pain-related disability in the low back and neck–shoulder regions and who could be categorised into one of four different groups regarding exposure to physical workload (for definition see below). Individuals with unknown zygosity (n = 341) were excluded from the analyses. However, they did not differ from the study group with regard to exposure to high physical workload and to the prevalence of LBP and/or NSP. In all, the study group consisted of 16 107 twins.

Zygosity was determined based on two questions in the questionnaire relating to similarity and mistaken identity during childhood: (1) “During childhood, were you and your twin partner as like as two peas in a pod or not more alike than siblings in general?” and (2) “How often did strangers have difficulty in distinguishing between you and your twin partner when you were children?”. This is an often used and acknowledged method for determining zygosity in large twin studies, and comparison of the suggested method with laboratory studies has shown less than 5% misclassification.23

LBP and NSP

The questionnaire contained screening questions concerning pain in the low back or neck–shoulder region in the preceding 6 months. These questions were formulated according to the Standardized Nordic Questionnaire (SNQ) and were accompanied by a drawing marking the areas of the neck–shoulder and low back regions.24 25 If the subject answered “yes” for pain on any of the two screening questions concerning pain in the low back or neck–shoulder region, six questions concerning the body region of interest followed. Three questions concerned pain intensity and three questions concerned pain-related disability. These questions were phrased according to von Korff.26 The rating scale for each of these attendant questions ranged from 0 to 10 where 0 meant no pain/disability at all and 10 meant pain/disability as bad as it can be.

A pain intensity score for each subject and each of the two body regions was constructed by calculating the mean value of the three questions that concerned pain intensity.26

A pain-related disability score was constructed in the same manner as the pain intensity score. A subject was defined as having LBP, respectively having NSP if he/she had a pain intensity score ⩾3 and/or a disability score ⩾1 in the anatomical region in question.7 21 Using these cut-off points the 6-month prevalences of LBP and of NSP in the present study were 22% and 28%, respectively, which is in line with what has been found in earlier studies.1 27 28 Subjects were then classified into four mutually exclusive groups: no LBP or NSP, solely LBP, solely NSP, or concurrent LBP and NSP during the last 6 months (table 1).

Physical workload

The questionnaire contained two questions relating to occupation. In the first question the subject was asked to state present/most recent occupation and in the second question what occupation the subject had had for the major part of their working life. In the present study the question concerning occupation for the major part of their working life was used to classify the occupations into high and low physical workload. For those subjects that had missing data on this question (598 twins), the value for the exposure in present/most recent occupation was used instead. The occupations were coded according to Nordic Occupation Classification (NYK), which is comparable to the three-digit International Classifications (ISCO-88).29 30

The classification of occupations into high and low physical workload was carried out using a previously developed gender-specific exposure classification system.31 This system was developed using the Swedish Annual Level-of-Living Surveys (1977 and 1979–1981), which are based on a nationwide representative sample of the Swedish population and contain questions concerning, among other things, occupation, and working environment. These surveys were conducted by Statistics Sweden and quality issues and technical details have been described elsewhere.32 33 In the development of the classification system, the occupations were coded according to NYK in the Swedish Annual Level-of-Living Surveys, and the proportions of subjects reporting different physical strains at work were calculated.

By using factor analysis the following eight factors were then extracted: heavy lifts daily, repetitive and one-sided working movements, unsuitable working postures, heavy shaking or vibration, daily perspiring from physical exertion, contact with dirt, deafening noise, and risk of exposure to accidents. For each occupation, factor scores for the eight factors were calculated and summarised into an index. Based on these index scores, the occupations were then ranked from lowest to highest, and grouped by quartiles into four groups from low (level 1), low–medium (level 2), high–medium (level 3) to high physical workload (level 4).

In the present study, due to problems with statistical power, this variable was dichotomised in all the statistical analyses and subjects were classified into two groups: (1) low physical workload (levels 1–2) and (2) high physical workload (levels 3–4). Further, the STAGE questionnaire contained one question concerning the duration (in years) in that occupation the subject had had for the major part of their working life.

Statistical methods

Descriptive statistics

The prevalence of exposure to low and high physical workload as well as the mean exposure duration (years) with the corresponding 95% CI were calculated, stratified by sex and zygosity. Also, the 6-month prevalences of no LBP or NSP, solely LBP, solely NSP, and concurrent LBP and NSP as well as the mean age (years) with the corresponding 95% CI were calculated, stratified by sex and zygosity.

Probandwise concordance rate and tetrachoric correlations

To explore if there is a genetic and/or shared environmental influence in the choice of occupation, probandwise concordance rates and tetrachoric correlations were calculated. The probandwise concordance rate estimates the risk that a twin in a pair is affected, given that the co-twin in the pair is affected.34 A genetic component is indicated when the probandwise concordance rate for MZ twins is higher than for DZ twins. Also, 95% CIs for the probandwise concordance rates were calculated as exact confidence limits for a binomial proportion using the statistical package SAS version 9.1.35

A tetrachoric correlation is correspondent to an intraclass correlation. Hence, a higher correlation for MZ twin pairs compared with DZ twin pairs suggests a greater similarity within MZ twin pairs. A genetic component is indicated when the tetrachoric correlation for MZ twins is more than twice that of DZ twins.36 The tetrachoric correlations and the corresponding 95% CIs were calculated using the PROC FREQ procedure in the statistical package SAS version 9.1.35 All analyses were stratified by zygosity and sex.

Cohort analyses

To analyse the association between high physical workload and the group with any one symptom (LBP and/or NSP), as well as the associations between high physical workload and the three disorder subgroups: solely LBP, solely NSP, and concurrent LBP and NSP, odds ratios (ORs) and 95% CIs were calculated using generalised estimation equations (GEEs) in the PROC GENMOD procedure in the statistical package SAS version 9.1.35 In this analysis all MZ twin pairs, same-sex DZ twin pairs, opposite-sex DZ twin pairs, as well as singleton twins were used. The analyses were adjusted for sex and age. Also, physical activity (during leisure time ) and duration of exposure (years) in the occupation were considered as potential confounders.

Co-twin control analyses

In the co-twin control analyses, complete MZ twin pairs and same-sex DZ twin pairs discordant for LBP and/or NSP were selected from the cohort of twins. The design resembles a matched case-control design, where one twin (case) has the disorder of interest and the co-twin (control) has not. The advantage of this design is that the association between high physical workload and LBP and/or NSP as well as the three disorder subgroups: solely LBP, solely NSP, and concurrent LBP and NSP can be controlled for confounding factors shared by the twins in a pair, such as common genes and environment shared by the twins early in life. The analyses were made separately for same-sex DZ and MZ twins. ORs and corresponding 95% CIs were calculated using the PROC PHREG procedure in the statistical package SAS version 9.1.35

Interpretation of the cohort and co-twin control analyses

By studying the changes in ORs from the cohort and co-twin control analyses, we can conclude about the importance of common genes and environment shared by the twins early in life for the association studied.

For example, a decrease in the association between the cohort and co-twin control analyses, but where the ORs for DZ and MZ twins remain similar, indicates the importance of shared environmental effects. Also, since DZ twins share approximately 50% and MZ twins share 100% of their segregating genes, a graduate decline between the associations for DZ and for MZ twins compared with the cohort analyses indicates a genetic influence for the association between exposure and outcome (i.e., OR_cohort>OR_DZ>OR_MZ).

RESULTS

Exposure to physical workload

The distribution of exposure to physical workload did not differ across zygosity (table 2). However, a higher proportion of men (48%, 95% CI 47 to 49) than women (43%, 95% CI 42 to 44) was exposed to high physical workload. The mean exposure duration (in years) did not differ across zygosity. For men, the mean exposure duration was longer for those with exposure to high physical workload (12 years, 95% CI 12 to 12), compared with those with exposure to low physical workload (10 years, 95% CI 10 to 10). For women, no corresponding differences in mean exposure duration could be seen (table 3).

The probandwise concordance rates and the tetrachoric correlations regarding exposure to high physical workload were consistently higher for MZ twins compared with DZ twins. The same pattern was seen for both men and women (table 4).

Prevalence of LBP and NSP

The prevalence of no LBP or NSP, solely LBP, solely NSP, or concurrent LBP and NSP, did not differ across zygosity (table 2). The prevalence of solely LBP was similar between women (10%, 95% CI 9 to 11) and men (12%, 95% CI 11 to 12); however women had a higher prevalence of solely NSP, 20% (95% CI 19 to 21) compared with 12% (95% CI 12 to 13) for men. Also for concurrent LBP and NSP, women had a higher prevalence, 14% (95% CI 13 to 14) compared with 8% (95% CI 7 to 8) for men.

For both women and men, the overall prevalence of LBP and/or NSP was lower in the group with exposure to low physical workload, and higher within the group with exposure to high physical workload (table 3).

Cohort analyses

The initial cohort analysis was performed using the group with any one symptom (LBP and/or NSP). The results showed an association between high physical workload and LBP and/or NSP, OR 1.47 (95% CI 1.37 to 1.57) (fig 1). The associations between high physical workload and the two subgroups: solely LBP (OR 1.41; 95% CI 1.27 to 1.57) and solely NSP, respectively (OR 1.31; 95% CI 1.20 to 1.43), were similar. The association between high physical workload and the subgroup with concurrent LBP and NSP was stronger (OR 1.80; 95% CI 1.62 to 1.99) than for those with solely LBP or solely NSP. All analyses were adjusted for age and sex. Adjusting for physical activity did not change the ORs. Duration of exposure (years) in the occupation was strongly correlated with age (r = 0.75) and was therefore not included in the analysis. Further, including duration of exposure in the analyses did not change the ORs (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Odds ratios (ORs) with 95% CI for the association between high physical workload and the group with any one symptom (low back pain (LBP) and/or neck–shoulder pain (NSP)) as well as OR with 95% CI for the three disorder subgroups: solely LBP, solely NSP, and concurrent LBP. Results from the cohort (all) and co-twin control analyses, respectively (dizygotic (DZ); monozygotic (MZ)). Total number of subjects and (cases) included in each analysis are also presented.

Co-twin control analyses

In the co-twin control analyses, the association between high physical workload and the group with any one symptom (LBP and/or NSP), was OR 1.34 (95% CI 1.02 to 1.75) for DZ twins and OR 1.44 (95% CI 1.06 to 1.95) for MZ twins. The OR did not decrease in these analyses indicating an association between high physical workload and LBP and/or NSP not confounded by genetic and shared environment factors (fig 1).

Concurrent LBP and NSP was the only disorder group that showed a stepwise decrease of the point estimates from the initial cohort analysis (OR 1.80; 95% CI 1.62 to 1.99), and then further between DZ and MZ twins, where the corresponding associations were OR 1.71 (95% CI 1.00 to 2.94) for DZ twins, and OR 1.29 (95% CI 0.64 to 2.59) for MZ twins. This might suggest that there is genetic confounding in the association between high physical workload and the occurrence of concurrent LBP and NSP. However, the 95% CIs were wide and overlapping between the zygosity groups (fig 1). Due to the sex differences with regard to both exposure to physical workload and prevalence of LBP and/or NSP, only DZ twins and same-sex MZ twins were used in the co-twin control analyses.

DISCUSSION

The results in the present study indicated a significant association between high physical workload and the occurrence of pain in the low back and/or neck–shoulder region even after adjusting for genetic and shared environmental factors. The findings are in line with Hartvigsen et al, who after adjusting for genetic factors found a graded association for increasing workload and LBP.20 Further, two earlier studies focusing on the genetic influence for intervertebral disc degeneration, also found some evidence for an independent association between physical workload and intervertebral disc degeneration.37 38

Concurrent LBP and NSP was the only disorder subgroup that exhibited a stepwise decrease in OR from the initial cohort analysis and the co-twin control analyses between DZ and MZ twins, indicating that the association between high physical workload and concurrent LBP and NSP was confounded by both shared environmental and genetic factors.

However, in the co-twin control analyses of concurrent LBP and NSP, the 95% CIs were wide and overlapping and no statistically significant differences between the analysis of DZ twins and the analysis of MZ twins were found. This can be explained by problems with statistical power, since the numbers of discordant twin pairs regarding LBP and/or NSP in the study were few.

The findings regarding the group concurrent LBP and NSP could suggest a presence of genetic factors that are specific for the group with pain in more than one area of the spine. This might reflect a musculoskeletal pain syndrome with somewhat different underlying causes than for solely LBP and solely NSP (Nyman T, Mulder M, Iliadou A, et al. High heritability for concurrent low back and neck-shoulder pain – a study of twins. Manuscript submitted for publication, 2008).

Earlier research has found sex differences in the associations between work-related risk factors and musculoskeletal complaints.39 40 In the present study, the size of the study sample did not allow for analysing women and men separately, and hence nothing can be concluded about possible sex differences in the association between high physical workload and solely LBP, solely NSP, and concurrent LBP and NSP. However, the descriptive analyses displayed sex differences in the prevalence of solely LBP, solely NSP, and concurrent LBP and NSP. Further, even though the exposure classification was carried out separately for women and men, there were sex differences in the distribution of exposure to physical workload, with a smaller proportion of women in the group with high physical workload. These findings are in line with previous studies.41 Further studies are needed to explore possible sex differences in the association between high physical workload and NSP and LBP.

A strength in the present study is that the exposure to high physical workload was assessed based on a classification system which had been developed using another data material where consideration had been taken to eight different physical strain dimensions, in addition to heavy lifting, repetitive work, and awkward work postures.31 This classification system was developed based on self-reported data on work-related exposures which could lead to differential misclassification, as subjects with LBP and NSP tend to overestimate the exposure to physical strains, which in turn may result in overestimations of the association between exposure and outcome.42 The influence of any potential differential misclassification should be minor since the exposure classification system was developed using a different data material, and subjects in the present study only reported their occupation (and not the exposure to physical workload).

Still, using an exposure classification system based on occupation codes also has limitations. First, it has to be assumed that the level of exposures to physical workload in an occupation is stable over time. Second, exposure differences between individuals within an occupation are not taken into account. However, there is no reason to assume that this exposure misclassification would differ between the exposure groups. Further, in the present study, the duration of exposure was not included in the analyses, and the mean exposure duration was slightly longer for men with exposure to high physical workload (12 years), compared with those with exposure to low physical workload (10 years) which could lead to an overestimation of the association between high physical workload and LBP and/or NSP. Further studies with more precise exposure assessment methods are needed.

The lack of precision in the exposure assessment and/or in the outcome measurements can lead to non-differential misclassification, diluting the association.42 43 This may be one reason for the moderate size of the ORs found in the present study. Nevertheless, the point estimates of the ORs in the present study consistently indicated an association between exposure to high physical workload and the three disorder groups: solely LBP, solely NSP, and concurrent LBP and NSP. Still, if the sample size would have allowed, using four levels of exposure to physical workload instead of the dichotomised variable would have been preferred, since the prevalence of LBP and/or NSP showed a graded increase with increasing exposure to physical workload (data not shown).

In the present study, the probandwise concordance rates and the tetrachoric correlations were consistently higher for MZ twins than for DZ twins, which might point to some genetic influence for the choice of occupation. However, the results in the present study only found a genetic confounding in the association between high physical workload and concurrent LBP and NSP. Further, the probandwise concordance rates as well as the tetrachoric correlations for exposure to high physical workload were less than twice as high for MZ twins compared with DZ twins, which suggest that shared environmental factors are of great importance for the choice of occupation. This is something which needs to be further investigated.

To be able to extrapolate the findings in twin studies to the general population, it has to be assumed that the prevalence of LBP and NSP among twins does not differ from non-twins. For the Swedish twin cohort this generalisation has been found to be valid for a variety of common disorders in several previous studies where the cohort has been shown to be representative for a Swedish population.23 Further, in the present study, the prevalence of LBP and NSP (without taking concurrent complaints into consideration) were similar to what has been found in earlier studies on the general population.1 27 28

Clinical significance

In a clinical view, the results in the present study emphasise the continuous need for preventive workplace interventions in occupations with high physical workload. Further, the results also support earlier findings that concurrent LBP and NSP seems to differ in some respects from subjects with solely LBP or solely NSP. It may well be that subjects with pain in more than one area of the spine constitute a group where it is especially important to take early signs of pain/discomfort seriously. An early onset of counselling about choice of occupation or ergonomic interventions might be of great importance for these individuals.

Conclusions

High physical workload was associated with LBP and/or NSP even after adjusting for genetic or shared environmental factors. Only for concurrent LBP and NSP, genetic and shared environmental factors seemed to have an influence on the association with high physical workload.