Heritability of neck pain: a population-based study of 33,794 Danish twins
Neck pain is very common in Western countries, where more than one in two adults have had neck pain during their lifetime. Neck pain is a great socio-economic burden in terms of sick leave, disability and workers’ compensation benefits, and it is therefore important to study what causes neck pain in order to identify possible preventive measures. Studies on the aetiology of neck pain have largely focused on occupational and psychosocial risk factors. Also, comorbidities and a previous history of neck injury have been associated with neck pain, but much confusion exists regarding the true causes of this very common symptom.
Recently, a few studies have focused on whether genes play a role in relation to neck pain. Sambrook et al. found an important genetic influence in degenerative changes of the cervical intervertebral discs based on MRI. However, in women such changes are probably not associated with neck pain. In a recent study, MacGregor et al. showed a genetic influence on neck pain in women. Unfortunately, the analysis was based on a non-population-based sample of volunteer twins and is therefore probably subject to selection bias. Hartvigsen et al. studied neck pain in Danish twins aged 70 and older, and found that genes play only a minor role in the liability to neck pain in seniors. However, these results cannot be extrapolated to a younger twin cohort and so the heritability of neck pain in a population of young and middle-aged twins remains to be investigated.
The aim of this study was to determine the genetic and environmental contributions to neck pain in men and women aged between 20 and 71 yr. Population-based twin registries offer a valuable tool for investigating the effect of genes and environment in complex diseases such as neck pain. This paper will therefore add to the understanding of the aetiology of neck pain.
With this study we have demonstrated that the overall heritability of lifetime neck pain is 44%. Heritability differs between men and women, being highest for women (33 vs 51%). In other words, nearly half of the variation with regard to neck pain in the young and middle-aged population is a result of genetic variation; in women, slightly more than half of the liability to neck pain is caused by genetic variation, whereas in men the proportion is only one-third. Hence, environmental influence plays a larger role with regard to the development of neck pain in men than in women. The gender difference in the perception and experience of pain is well documented and is believed to be a combination of psychological and biological factors. Recently, willingness to report pain has been raised as an explanation for the gender differences in pain reporting. Interestingly, when willingness to report pain was measured in a questionnaire—and subsequently controlled for—ambiguous results were seen: sex was not a predictor of the pain threshold, but remained a significant predictor in pain tolerance. Therefore, psychosocial factors as well as biological factors are viable explanations for gender differences in (experimental) pain. However, our heritability estimates were performed on each gender separately and thus the willingness to report neck pain will not have influenced the gender-specific heritability estimates. However, the significant difference in the heritability of neck pain between genders may be influenced by the willingness to report neck pain. Further studies are needed to elucidate whether this difference can be explained by this phenomenon.
The environmental influences increase with age and the genetic component becomes negligible in the oldest age groups, especially among females, in whom the liability to neck pain is almost entirely due to environmental factors. Thus, environmental influences are more likely to be the major cause of the liability to neck pain in older age groups. This is not surprising, since neck pain is associated with several physical and psychosocial risk factors that may manifest later in life. Unfortunately, it was not possible to measure and subsequently adjust for these well-known psychosocial risk factors in our study.
In order for biometric modelling to be valid, two major assumptions need to be fulfilled. First, the heritability estimates calculated from twins are based on the assumption that the heritability can be extrapolated to the general population. This generalizability has been supported by several studies, as it has been shown that the Danish twin cohort is representative of the Danish population in terms of many common diseases and mortality rate. However, the lifetime neck pain prevalence in the Danish population remains to be confirmed in other studies and it is therefore impossible to determine whether the prevalence of neck pain in our twin cohort is equal to the prevalence of neck pain in other cohorts. Second, the assumption of equal environment assumes that MZ and DZ twins are equally exposed to environmental factors aetiologically related to the trait under study (in this case neck pain). Monozygotic twins are treated more similarly than DZ twins by their parents. However, this similarity is not due to their greater phenotypic similarity, but is rather a consequence of their genetic identity and the greater similarity in responses that this elicits from the environment. Thus, twins are treated according to their actual zygosity, not their perceived zygosity. In our study, the prevalence estimates among MZ and DZ twins were identical, and therefore no violation of the equal environment assumption was detected.
The zygosity determination must be accurate in order for the results to be valid. The zygosity of twins has previously been determined via self-reported questionnaires, showing a misclassification rate of 3%, which is considered acceptable. Since more MZ than DZ twins were misclassified, any effect on the accuracy of the heritability estimates is in the conservative direction.
Also, non-differentiated misclassification due to unspecific disease definition will tend to lower the heritability estimates. Even though the definition of neck pain was clearly stated and accompanied by a drawing marking the area of interest, a positive answer indicating lifetime occurrence of neck pain in an individual remains a source of subjective interpretation. Furthermore, this definition includes both specific and non-specific causes of neck pain. Consequently, the heritability estimates presented in this paper should be considered only as an overall estimate of the genetic contribution to neck pain and the real genetic effect in specific subgroups may be larger.
Another potential limitation of this study is non-response and selection bias. Although the unadjusted response rate was 73%, we did not see significant differences between non-responders in terms of zygosity. In addition, a potential non-response bias specifically regarding our parameter (lifetime neck pain prevalence) is unlikely as this question was nested in a large 20-page general health survey. Thus, if bias was present it is related to other factors (for example, the size of the questionnaire) and not specifically to one single parameter such as neck pain.
The use of lifetime neck pain raises the issue of recall bias, which affects the prevalence estimates. This may explain the decline in neck pain prevalence in the oldest age group. However, this age-related drop in prevalence estimates could also reflect a birth cohort effect due to different levels of exposure to risk factors. Only a follow-up study could further investigate possible explanations of recall bias. Nevertheless, recall bias is assumed to be equal between MZ and DZ twins and, as the heritability estimates are comparisons between MZ and DZ twin pairs, possible recall bias does not affect our heritability estimates.
This is the first study on heritability of neck pain in young and middle-aged men and women, and it has several major strengths. First, it is based on a twin cohort with nearly complete ascertainment, resulting in less biased heritability estimates. Second, the large sample size and the high response rate made it possible to estimate differences in heritability between men and women with high accuracy (i.e. small confidence intervals). Third, the wide age span made it possible to study how the genetic and environmental contributions change in different age groups. Finally, the Danish twin cohort has been shown to be representative of the general population and so our findings can be extrapolated to the background population.
Three previous studies have focused on genes in relation to neck pain. Hartvigsen et al. estimated the genetic and environmental contribution in neck pain in people aged 70 and older and concluded that the genetic effect plays only a minor role in old age. Their study supports our findings that the environmental effect gradually increases with age and that the genetic effect diminishes as people get old. Thus, environmental risk factors are more important in developing neck pain over time than any genetic component(s).
In the UK, MacGregor et al. investigated the heritability of neck pain in female twins. Although their heritability estimates (48%) were very close to ours (51%), this may be a coincidence as their study was based on volunteers found through media campaigns and thus subject to severe selection bias. Also, they did not stratify by age, making it impossible to evaluate how age might have affected their results.
Sambrook et al. showed a high genetic influence in cervical disc degeneration, with a genetic contribution varying between 63% and 79%. However, these results were based on two different cohorts (Australia and UK) with different selection criteria and with overrepresentation of females, and thus again selection bias may partly explain the high heritability estimates. Further, the outcome (cervical disc degeneration) is probably not directly associated with neck pain.
Although neck pain is heritable, it is not known how genes specifically influence the presence of neck pain. However, as genes play a lesser role with increasing age, efforts should be made to minimize any documented environmental risk factors that may cause or aggravate neck pain (e.g. physical and psychosocial risk factors). Neck pain may be considered a heterogeneous cluster of more specific conditions with pain localized in the neck area as a common feature. In addition, there may be genetic differences in, for example, the duration and severity of neck pain (i.e. chronic neck pain vs acute neck pain, or one episode of neck pain vs many episodes of neck pain). Unfortunately, it was not possible to investigate these factors because of practical limitations (i.e. it was impossible to include additional questions) in the survey. Future studies should therefore be directed at investigating whether the genetic contributions in the various subdefinitions of people with neck pain are significantly different from each other. Attempts should be made to adjust for well-known psychosocial risk factors if possible. Furthermore, studies should be conducted to investigate whether genetic differences between men and women are based on different genes or whether it is the same set of genes or shared environmental experiences that contributes to neck pain.