Methods

The 1958 British Birth Cohort Study (or “National Childhood Development Study”) recruited persons born in Great Britain during 1 week of March in 1958.15 16 They have been followed-up through childhood and adulthood with the most recent follow-up at approximately 45 years. In all, 17 638 individuals were eligible, while immigrants with the same birth dates were recruited up to age 16 years (n = 920), giving a total sample of 18 558. At 45 years a target sample of 12 069 participants who were still in contact with the study, and who at 42 years had not required a proxy interview, were invited to a nurse interview and examination in their home, from which a total of 9377 participants were seen September 2002 to March 2004. A flowchart of participants from birth to most recent follow-up has previously been presented elsewhere.17 Among participants in the latest follow-up, fewer had a low social class (IV or V) in childhood compared with the original sample, although the difference was small: 22.3% versus 26.9%. Further, previous work has shown that the 45-year sample remains broadly representative of the surviving cohort.18 Ethical approval for the current follow-up was obtained by South-East England MREC (ref 01/1/44).

The 45-year follow-up included a questionnaire which the participant self-completed at home. Participants were asked “During the past month, have you had any ache or pain which has lasted for 1 day or longer?”. Participants who responded positively were asked to indicate the site of the pain on four-view body manikins and to indicate whether they had been aware of the pain for more than 3 months. This allowed the identification of persons with pain in the shoulder, forearm, knee and low back. In addition, chronic widespread pain was defined using the American College of Rheumatology (ACR) criteria for fibromyalgia, namely pain in contra-lateral body quadrants and in the axial skeleton, which had lasted for more than 3 months.19 Socioeconomic position was classified using the Registrar General’s classification of social class. This comprises six classes: (I) professional, (II) managerial/technical, (IIInm) other non-manual, (IIIm) skilled manual, (IV) partly skilled and (V) non-skilled. In childhood this was measured on the basis of the father’s occupation in 1958 (ie, at birth of the study participant) or, where this information was missing, on the father’s occupation at the first follow-up in 1965 (n = 422). Adult socioeconomic position was based on the participant’s occupation, or most recent occupation, at age 42 years. If this information was missing, occupation at age 33 years was used (n = 1142). This approach classifies most individuals, although a small number of permanently disabled/institutionalised individuals are excluded. Participants lacking information on either child and/or adult socioeconomic position (n = 367), and thereafter those who did not return the self-complete questionnaire or did not complete the pain questions (n = 1084), were excluded.

Information was available in adulthood on factors that may be associated with low socioeconomic status and be related to pain reporting. At the 45-year follow-up, depression and anxiety in the previous week were measured by the revised Clinical Interview Schedule (rCIS) administered in participants’ homes by a trained nurse.20 Body Mass Index (BMI) was determined by measurement of participants’ height and weight. At 45 years, participants were also asked to self-complete a questionnaire that included items on adverse life events (such as serious illness, injury or assault; death of a close relative; or problems with the police) during the previous 6 months. Information was available, at 42 years, on participants’ psychological distress from the General Health Questionnaire (GHQ), which has been validated for use in population settings21 and frequency of regular exercise, and, at 45 years, current smoking status.

Statistical analysis

Poisson regression analysis was used to determine the relationship between socioeconomic status and the reporting of chronic widespread pain and four regional pain syndromes: low back, shoulder, forearm and knee pain, in five separate statistical models. In the first instance, for each pain outcome, the relationship between adult socioeconomic position and pain reporting was determined and, using an interaction term, whether it varied by sex. Given that there were no statistically significant interactions (p⩽0.05), all models are presented for both sexes together. It was then assessed whether there were additional risks associated with child social class by mutual adjustment for child and adult socioeconomic status in a single model. Finally we determined to what extent any observed relationships were mediated through adverse psychological status in adulthood by adding information on whether respondents reported more than one symptom on the depressive symptom or anxiety modules of the rCIS and psychological distress. Relationships are described using relative risk (RR) and the 95% confidence interval (CI).

Results

Social class mobility was substantial: in 1958 approximately 70% of participants had a manual social class at birth compared with approximately 43% men and 29% women at ages 45 years (table 1). Low back pain (27.9%) and shoulder pain (26.2%) were the most common regional pain syndromes, followed by knee pain (18.6%) and forearm pain (7.8%). In total 11.8% of people reported chronic widespread pain.

The prevalence of shoulder pain, low back pain, knee pain and chronic widespread pain increased with lower adult social class, while for forearm pain prevalence increased across social classes I–IV with a small decrease in social class V (table 2). Persons in the lowest social class (compared to the highest) experienced nearly a threefold increase in the risk of chronic widespread pain: unadjusted RR 2.9 (95% CI 1.8 to 4.6). The relationship was similar, although of slightly smaller magnitude, for shoulder pain RR 1.5 (95% CI 1.1 to 2.0), low back pain RR 1.6 (95% CI 1.2 to 2.2), forearm pain RR 2.0 (95% CI 1.1 to 3.6) and knee pain RR 2.0 (95% CI 1.4 to 2.8) (table 2). Childhood social class also demonstrated a relationship with most regional pains and chronic widespread pain. With the exception of forearm pain, the magnitude of the association between childhood social status and the reporting of pain in adulthood was less than that of adult social status. The effect in the lowest social class ranged from RR 1.2 (95% CI 0.9 to 1.5) for shoulder pain, RR 1.5 (95% CI 1.5 to 1.8) and RR 1.5 (95% CI 1.1 to 2.1) for low back pain and knee pain, respectively, to RR 1.8 (95% CI 1.2 to 2.6) for chronic widespread pain. However, participants in social class V at birth experienced almost treble the risk of forearm pain at 45 years: RR 2.7 (95% CI 1.5 to 4.9) (table 2).

The multivariable models for each of the four regional and chronic widespread pain included adult and childhood social class, and information on other adult psychological, physical and lifestyle factors (table 3). This demonstrated adult poor mental health (rCIS symptoms) was strongly associated with each of the regional pain syndromes and chronic widespread pain as was (with the exception of forearm pain) psychological distress (GHQ). Recent life events were associated with an excess risk of reporting forearm, knee and shoulder pain. Being either a current or ex-smoker was associated with all regional pains and chronic widespread pain, as was a high BMI (with the exception of shoulder pain). In contrast, frequency of exercise showed no relationship with any reported pain. When adjustment was made for all these factors there remained a statistically significant association between adult social class and forearm pain, knee pain, low back and chronic widespread pain, with excess risks ranging from 30% to 80%. The relationship with childhood social class was strong and statistically significant for forearm pain (lowest versus highest social class, RR 2.3 (95% CI 1.2 to 4.3)). Although there were some increased risks of childhood social class noted for knee pain, low back pain and chronic widespread pain there were of the order of 20% to 30% and were non-significant.

Discussion

This national birth cohort study has found that low social class in adulthood is associated with all major regional musculoskeletal pains and chronic widespread pain. The RR associated with lowest (compared to highest) adult social class ranged from 1.5 to 2.9 depending on pain syndrome. In addition, for the first time, we have demonstrated that there is also an additional effect of low social class in childhood. However these associations were partly explained by adult mental health, psychological distress, adverse life events and lifestyle factors.

There are some methodological issues to consider in interpreting the results. Firstly, loss to follow-up: there was only approximately half of the original sample who participated at 45 years. Although this is still a large number of participants, could non-participation bias explain the results obtained? In order for us to have artefactually observed an effect for low social class, persons of a higher social class who developed musculoskeletal pain must have been about 50% to 80% more likely to drop out than persons of high social class who remained pain free. This seems unlikely. The data are consistent with other studies of musculoskeletal pain and more generally of poor health in demonstrating a relationship with adult low social class. Further, loss to follow-up shows only small differences between childhood social classes.18

Secondly, there are a number of issues relating to the measurement of social class. The assessment method used in the current study, the Registrar General’s classification, is a broad and imperfect measure of socioeconomic status. Therefore, we would expect that the true effect of socioeconomic status may be attenuated. Other markers of socioeconomic status were available (for example: highest formal level of education) although these are likely to be highly correlated with social class and there is limited value in including both in the analysis. The Registrar General’s classification is based on occupation, and if occupation (a woman’s occupation in particular) is a poor measure of her socioeconomic status then it may overestimate the effect of childhood social class. Additionally, in addition to survey attrition, we were unable to assign social class to a small number of individuals (ie, those permanently disabled and/or institutionalised). We expected that there may be an issue with colinearity of measures of social status in adulthood and childhood thus making it difficult to determine the relative contribution of each. However, we observed substantial change in social class between childhood and adulthood.

Finally, we have collected very basic information on pain from participants by means of shaded areas on a body manikin of pain experienced for at least 1 day during the past month. The pain will include those whose pain is arising from the shaded area as well as pain arising elsewhere that is radiating to that area. We also have not collected information on disability arising from pain and we would expect that social class may have an even greater impact on functional limitation. More importantly, there may be substantial overlap between the pain categories (ie, many individuals have reported several regional pain sites and chronic widespread pain). Further, since the association between social class and pain is strongest for chronic widespread pain, it might be that the association between social class and regional pain is driven by chronic widespread pain. A sensitivity analysis (data not shown) revealed that the risk relationships were broadly similar between regional pain models with/without the inclusion of persons with chronic widespread pain. All models showed a modest reduction in effect, suggesting that the effects with regional pain were at least partly explained by chronic widespread pain. However, the exclusion of persons with chronic widespread pain did not fully explain the association between social class and regional pain.

In support of results from this study, preliminary results from a recent UK cross-sectional study of chronic widespread pain found an excess of reported symptoms associated with low socioeconomic status and this excess of pain could be explained principally by higher levels of psychological distress.22 Further, a cross-sectional study of 1287 adults with functional limitations due to musculoskeletal pain found that coping strategies, which have been shown to influence outcome of an episode of musculoskeletal pain,23 24 varied between persons of different social class.25 In a prospective study of approximately 1000 persons in northern Sweden, musculoskeletal disorders were strongly related to social class (as ascertained by white/blue collar worker status) at age 30 years. This inequality in prevalence with social status was mostly explained by adverse behavioural (eg, lack of physical activity, smoking, low educational achievement) and social (eg, being single) circumstances in adolescence and early adulthood. In contrast to other studies, we have been able to examine the role of social class across a spectrum of regional and chronic widespread pain; the results have shown generally consistent patterns although the magnitude of association has varied between sites: with adult social class being more important for forearm, knee and chronic widespread pain and childhood social class more so for forearm pain. It remains to be determined whether these differences are simply due to chance or whether social influences, possibly related to occupation, are particularly strong for some pain syndromes.

In summary, therefore, we have confirmed an influence of low adult social class on adult musculoskeletal pain and demonstrated an additional influence of childhood social class. We have been able to explain this partly (for most regional and chronic widespread pain) on the basis of adult behavioural and emotional factors and lifestyle factors in adulthood. There is consistency emerging across studies of different design, in different geographical settings and with different measures of social class as well as mediating factors, contributing to our understanding of how differences in occurrence of musculoskeletal pain by social class may occur. This emphasises the importance and potential impact of measures to reduce social adversity, which will have the effect of improving musculoskeletal health in adult life and also other major causes of morbidity.