Chronic widespread pain (CWP) is the cardinal symptom of the fibromyalgia syndrome, one of the most common reasons for referral to a rheumatologist. Symptoms are persistent and individuals with such symptoms are long-term frequent consulters for healthcare1 resulting in high direct and indirect care costs.2 Epidemiological studies have demonstrated the main risk factors for symptom onset are psychological distress, psychosocial factors (including adverse life events), and markers for future symptom onset are the report of other somatic symptoms and a history of health-seeking behaviour.3

Descriptive studies of pain are hampered by the problem that there are few standard definitions for regional or widespread syndromes, and minor differences in the definition can make important changes to prevalence estimates.4 Notwithstanding this problem, the available evidence suggests that there may be relatively little variation in the prevalence of CWP internationally. Studies in the United States, United Kingdom and Sweden have all suggested, using self-completed questionnaires, that the prevalence is approximately 10–13% in adults overall, that prevalence increases with age until approximately 60 years and decreases slightly thereafter, and that prevalence rates are marginally higher at all ages in women compared with men.5^–7 One study in Canada reported a slightly lower prevalence of 7% but this study was conducted partly by telephone, rather than by self-completed questionnaire, which may have influenced the prevalence estimate.8

If psychological and psychosocial factors influence the reporting of symptoms, and subjects at higher risk can be identified by past health and illness attitudes and beliefs then it would be surprising if there were not differences in prevalence between countries, because such risk factors or markers are likely to differ between populations of countries with distinct cultures. Indeed, in a study in the United States, African-American young women were more likely to report CWP than young white women,9 whereas in a study of south Asians within the United Kingdom, cultural attitudes were demonstrated to influence the likelihood of pain reporting.10 In addition, there is emerging evidence that genetic factors may be important in terms of pain reporting and differences in genotypes associated with a greater likelihood of CWP, such as val-158-met catechol-O-methyltransferase, may also give rise to differences between populations of different ethnic backgrounds.11

We therefore took advantage of a major epidemiological study being conducted among men in eight countries of Europe, the European Male Ageing Study (EMAS), to determine whether among middle-aged and elderly men there was evidence of differences in prevalence and whether any such differences could be explained by psychological, psychosocial or general health factors.

METHODS

The EMAS study is a population-based prospective study of men aged 40–79 years in eight European countries. The methods of the study have been described in detail previously.12 Recruitment to the study took place in Belgium (Leuven), England (Manchester), Estonia (Tartu), Hungary (Szeged), Italy (Florence), Poland (Lodz), Spain (Santiago) and Sweden (Malmo). The sampling frame in each centre was a population register that fulfilled the following criteria: (1) all men aged 40–79 years resident in the study area must be eligible for inclusion on the register; (2) the register should be of sufficient size to yield the required number of participants for the study; (3) the information held in the register must be current and the date of the latest update known; (4) the register must hold, as a minimum, gender, name, date of birth and current address and (5) permission must be granted to access the register. Each of the eight centres aimed to recruit a target of 400 men: the sample was age-stratified to have similar numbers representing the four decades of age. The current analysis uses data from the baseline health questionnaire, which was collected in 2004–5. Ethics approval was obtained from local ethics committees of each of the participating centres.

Subjects were mailed an invitation letter, study information sheet and a questionnaire that had been translated and back-translated from English to ensure consistency across languages. For subjects from whom no reply had been received after 4 weeks, a further invitation letter and questionnaire was sent. In case of non-response, a check was undertaken on whether the individual was still registered at that address on the most recent version of the population register used to draw the study sample.

Pain

Information on pain was collected in the questionnaire by asking subjects, “In the past month have you had any pain which has lasted for one day or more?”. If subjects answered positively then they were asked to indicate the site(s) of pain on four-view body mannikins. To assess chronicity they were asked whether they had been aware of the pain for 3 months or more. In order to be classified as having CWP, we adopted the criteria for widespread pain included in the American College of Rheumatology criteria for fibromyalgia.13 These criteria require pain lasting at least 3 months, both above and below the waist, on the right and left sides of the body and in the axial skeleton. This instrument has frequently been used in population surveys and its construct validity demonstrated.14 The methods chosen to assess pain (body mannikins) were less language dependent than alternatives. Nevertheless the stem question to the mannikins did require translating into each centre’s language. This was done (for the whole questionnaire) by professional translation of the questionnaire from English, and back-translation by the centres’ staff with a consensus conference arranged thereafter to resolve any discrepancies between the original and the back-translated version.

Other health information

Information was collected on mental health using the Beck Depression Inventory (BDI). The BDI was originally designed to measure the depth or intensity of depression in psychiatric patients15 but has subsequently been revised and used as a community screening instrument.16 It evaluates 21 symptoms of depression (15 cover emotions, four behavioural changes and six somatic symptoms). Each symptom is rated on a four-point intensity scale giving a total score from 0 to 63: higher scores indicate more severe depression. For the purposes of analysis the scores have been categorised into six bands with labels “normal”, “mild”, “borderline”, “moderate”, “severe” and “extreme”.

In order to evaluate the general health of participants, they were asked whether they were currently being treated for any of a list of 14 morbidities: heart conditions, high blood pressure, pituitary disease, testicular disease, chronic bronchitis, asthma, peptic ulcer, epilepsy, diabetes, liver conditions, kidney conditions, prostate disease, adrenal disease, thyroid disease.

Life events

The list of threatening experiences, a 12-event inventory initially modified by Brugha et al17 from a 67 life events inventory,18 was used to gather information on recent adverse life events. The 12 categories are associated with a significant marked or moderate long-term contextual threat rating. The categories ask about personal relationships, employment, illness and financial and legal problems in the past 6 months. Each item response scores 0–1 giving a maximum life events score of 12.

Finally, respondents were asked if they were currently being treated for any of a series of 14 conditions and a count of these provided a measure of co-morbidity.

Statistical analysis

All statistical analyses have been undertaken using STATA (http://www.stata.com). In addition to the presentation of prevalence rates across centres, the relationship of the effect of the centre and individual factors on CWP has been evaluated by logistic regression. The cluster subcommand has been used to take account of the clustering of individuals in centres that may be similar with respect to health, lifestyle life experiences and cultural aspects of reporting. The relationship has been expressed as odds ratios (OR) and 95% CI. Finally, the effect of centre adjusted for other potential confounding factors has been evaluated in a multivariable model.

RESULTS

Of the 8416 men surveyed across eight international centres, 3963 returned a study questionnaire (table 1): this represents a 47.1% participation rate. Excluding individuals from the denominator who were thought not to have received a questionnaire (died or moved house), the adjusted participation rate was 48.7%. Adjusted participation rates ranged from 24.1% in Szeged (Hungary) to 71.7% in Tartu (Estonia).

CWP prevalence across centres

Of the 3963 participants, 3757 provided sufficient information to allow us to classify them as having CWP or not. Overall, 313 men reported CWP (prevalence 8.3%, 95% CI 7.5% to 9.3%). Prevalence was broadly similar across the four decades of study. It increased slightly from 40–49 years (7.4%) to 50–59 years (9.6%) and then decreased at 60–69 years (8.5%) and 70 years and over (7.8%). There were significant differences in prevalence between centres. Prevalence was between 5% and 7% in Florence (Italy), Manchester (England), Leuven (Belgium) and Malmo (Sweden), 9% to 15% in Santiago (Spain), Lodz (Poland) and Szeged (Hungary) and 15% in Tartu (Estonia). Manchester (England) was chosen as the reference centre because it was in the group of centres with the lowest prevalence and had the largest sample. In comparison, the odds of reporting CWP was significantly increased in all centres except Florence (Italy): in Lodz (Poland) (OR 2.0, 95% CI 1.8 to 2.2), Szeged (Hungary) (OR 2.3, 95% CI 2.0 to 2.6) and Tartu (Estonia) (OR 3.7, 95% CI 3.3 to 4.1) the odds were at least twofold (table 2).

Distribution of potential risk markers and risk factors across centres

The distribution of BDI classifications, the number of adverse life events and physical morbidities are shown in table 3. The BDI band score was significantly different between centres. The percentage of subjects categorised in bands other than “normal” was between 12% and 18% in all centres except Santiago (21%), Szeged (22%), Tartu (35%) and Lodz (39%). Likewise, the number of life events reported in the past 6 months differed between countries. Across all centres, 28% of men reported at least two recent life events, with Szeged (30%), Tartu (34%) and Lodz (38%) reporting higher than average proportions. Szeged (Hungary) and Lodz (Poland) also had the highest level of multiple (more than one) specific morbidities (from a list of 14) reported (29% and 38%, respectively) together with Santiago (Spain) (28%), whereas Tartu (21%) had multiple morbidity levels similar to the remaining centres (range 16–22%).

The relationship between potential risk markers and factors with CWP

All of the above factors showed a relationship with CWP. The band of the BDI was strongly associated with the report of CWP. In comparison with individuals scoring in band 1 (normal) increasing score was associated with a significantly increased likelihood of reporting CWP (OR (bands 2–5 + 6): 1.6, 3.9, 4.4, 6.6; table 3). The number of recent life events was strongly associated with the likelihood of reporting CWP. In comparison with individuals who reported none, individuals reporting two events had a doubling of odds, three to five events had a three to fourfold increased odds, whereas those reporting more than six recent life events had a ninefold increased odds. Similarly there was a monotonic increase in the likelihood of reporting CWP with the number of morbidities, such that, for example, those reporting three morbidities had a threefold odds and those reporting five or more morbidities had a sixfold increased odds of CWP compared with those reporting no morbidities (table 2).

Given the strong relationship between BDI, recent life events, the number of morbidities and CWP, we thereafter used a multivariable model to determine whether the differences in prevalence of CWP between centres could be explained by differences in the distribution of these psychological, psychosocial and general health factors. When entered into such a multivariable model, the excess risks for reporting CWP in Leuven (Belgium), Malmo (Sweden), Santiago (Spain), Szeged (Hungary) and Tartu (Estonia) centres remained. The magnitude of increased risk was reduced in most cases: Leuven OR 1.4 to 1.2; Santiago OR 1.7 to 1.5; Lodz 2.0 to 1.1; Szeged 2.3 to 1.6 and Tartu 3.7 to 2.6, but all remained statistically significant. Only the magnitude of excess in Malmo did not decrease, changing from 1.5 to 1.7 (table 2).

DISCUSSION

This is the first study to compare the prevalence of CWP across several countries. It has found that overall there are important differences in prevalence between European countries—the main differences being a higher prevalence in centres in eastern Europe. Although varying between country, between half and all of the excess was explained by higher levels of depression, the number of recent life events and levels of physical morbidity in these populations.

There are a number of methodological points to consider. First, the study was population based and therefore unselected apart from the personal decision of whether to participate or not. Participation rates varied considerably between centres, but importantly there was no relationship between the participation rate and the prevalence of CWP. For example, Szeged, Leuven and Manchester all had low participation rates but respectively reported high, average and low prevalence rates, whereas Tartu with a high participation rate recorded, by a considerable margin, the highest prevalence of CWP in the study. A supplementary analysis of the prevalence of CWP among individuals who responded after the initial mailing compared with those responding only after a reminder showed no significant difference in prevalence (8.3% and 8.4%, respectively; p = 0.96) further supporting the view that the decision to participate was not importantly influenced by CWP. Second, this study was cross-sectional rather than prospective and therefore we cannot determine the temporal relationship of the onset of CWP with the level of depression and adverse life events. The temporality of the association between pain and depression has been the subject of much study and discussion; in particular, whether pain leads to depression and/or depression leads to pain, or both arise as a result of a common pathway. In a review of the pain–depression relationship, Fishbain et al19 reported multiple studies whose results were consistent with each of these mechanisms. In particular, studies of CWP that have tested the hypothesis that depression/mood disorders in patients free of CWP increase the risk of its onset have been supportive.14 20 Third, this study has measured only some of the risk factors for CWP. For example, although we have measured anxiety and depression, and recent life events, these are only two of the psychological and psychosocial factors that have been shown to be important predictors of the onset of CWP. In addition, illness attitudes and beliefs, early life events, the propensity to somatise and the propensity to seek healthcare for symptoms have all been found to be important.21 22 There are likely also to be differences in these domains internationally, which will therefore account for some of the remaining differences between centres. There are also other, non-psychosocial risk factors for the development of CWP. A previous study, conducted principally in England, has suggested that aspects of occupational mechanical load, posture or repetitive movements are predictors of the onset of widespread body pain.23 Although the EMAS study did not have a detailed evaluation of workplace physical exposures, subjects were asked to identify the best descriptor of the physical activity involved in their current job or voluntary work. The risk of CWP was increased only among those reporting the highest level of physical exposure: mainly sitting (OR 1.0, reference group), sitting or standing with some walking (OR 0.9, 95% CI 0.6 to 1.3), walking with light manual work (weights of less than 25 kg) (OR 0.99, 95% CI 0.7 to 1.5), walking with heavy manual work (weights of more than 25 kg) (OR 1.7, 95% CI 1.1 to 2.6). Adjustment for workplace physical factors did not, however, explain further the differences between centres (data not shown).

The final methodological issue that pertains to all epidemiological studies of pain, but is important to emphasise in interpreting this study, is that we are studying the reporting of CWP rather than the experience of CWP itself. One cannot thus distinguish between differences in the pain experience and influences on this from differences in the willingness to report negative events that could include pain, mood disturbance, health and adverse life events.

We are not aware of any other studies that have compared the prevalence of CWP across countries, but there have been studies that have compared chronic pain (generally) or specific types of pain across two or more countries. Raspe et al24 noted higher rates of back pain in Germany compared with Britain, whereas a large study of 46 394 individuals throughout Europe and Israel demonstrated rates of chronic paini ranging between 12% and 13% in Spain, Israel and the United Kingdom to 26% to 30% in Italy, Poland and Norway.25

The study has confirmed the important relationship between anxiety/depression and CWP. Although previous work has demonstrated that adverse life events are associated with CWP (and other pain syndromes) the populations studied had experienced relatively few such events.26 In this international study with a much greater number of individuals with multiple recent events, we have shown their important influence: the risk of reporting CWP increases monotonically with the number of such events experienced.

Studies conducted in single regions within a country have reported rates of CWP between 10% and 15% among general population adult samples including both men and women. Whether the differences reported are real or a result of small differences in definition is not clear—several of these studies have been conducted in one area of north-west England but there are also studies from north America, Scandinavia and Russia.5^–7 27 28 The data reported in this study of a prevalence of 8% among men aged 40–79 years is consistent with the data reported for men in this age range in previous studies, but the use of consistent methods across centres has revealed that there does appear to be differences in prevalence.

In summary, this study is the first study, of which we are aware, which has been able to compare the occurrence of CWP internationally and been able to investigate the potential aetiology behind differences in occurrence in Europe. We have demonstrated that there is an excess prevalence particularly in centres of eastern Europe and that some of this excess is related to higher levels of adverse psychosocial factors as well as the influence of poorer psychological health and physical morbidities.