Role of road traffic accidents and other traumatic events in the onset of chronic widespread pain: Results from a population-based prospective study




To determine the relationship between physically traumatic events and the onset of chronic widespread pain (CWP).


This was a case–control study nested within a large prospective cohort. CWP was determined, by questionnaire, as per the American College of Rheumatology fibromyalgia classification criteria. Data were also collected on psychological health, health behavior, and sleep problems. Participants without CWP were then followed up at 4 years, and (new-onset) CWP was determined in the same manner. At followup, participants were also asked to report whether they had experienced any of a series of physically traumatic events between baseline and followup.


A total of 2,069 individuals (46.6%) participated at followup, and 241 of these individuals (11.6%) reported CWP. More than one-third of the study population reported at least 1 physically traumatic event; although these individuals were more likely to develop CWP, this relationship was completely attenuated after adjustment for confounding (odds ratio 1.01, 95% confidence interval 0.73–1.40). However, there was some evidence to suggest that involvement in a road traffic accident, specifically, may confer an increase in the risk of CWP onset.


This study provides support for the “at risk” phenotype hypothesis, where individuals characterized by poorer health and psychological variables may be predisposed to develop CWP following a traumatic trigger. However, although this has been seen with road traffic accidents, it is not the case with other events. Future research should examine what is peculiar about an accident, or about one's reaction to it, that confers this increase in the risk of CWP onset.


Chronic widespread pain (CWP) is defined by the American College of Rheumatology (ACR) as pain present for 3 months or longer, above and below the waist, on both the left and right sides of the body, and in the axial skeleton (1). It is a common and frequently disabling condition, and prevalence studies have provided consistent findings. In the UK, authors have reported prevalence rates of 11.2% (2) and 12.9% (3), with similar levels in Sweden (4), the US (5), and Germany (6). CWP increases with age, until ages 60–70 years, and is more common, at all ages, in women than in men (2–4, 5). Along with widespread tenderness, CWP is the principal feature of fibromyalgia, one of the most common reasons internationally for consultation with a rheumatologist (7).

Although the literature examining the etiology of CWP highlights the role of work-related postures/manual handling activities (8), psychological distress (9), features of somatization (10), and workplace psychosocial factors (8), individuals with CWP often attribute symptoms to a precipitating physically traumatic event, e.g., a surgical operation or road traffic accident (RTA) (11). However, there is little robust epidemiologic evidence to support this. Buskila et al (12) conducted an examination for fibromyalgia among 161 patients with traumatic injuries who were attending a clinic. These authors report that individuals presenting with neck injuries were 13 times more likely to meet the ACR criteria for fibromyalgia than those with leg injuries (12). However, this might have been expected since the major distinguishing feature between the 2 groups was not pain but tender point count, and 10 of the 18 fibromyalgia tender point sites are found in the area of the neck and shoulder (1). Al-Allaf et al (13) demonstrated that individuals with fibromyalgia were twice as likely to report a traumatic event in the 6 months prior to survey (odds ratio [OR] 2.1, 95% confidence interval [95% CI] 1.2–3.4). In particular, those with fibromyalgia reported an excess of injuries at work, fractures, and surgery.

Recently, several studies have demonstrated that among persons experiencing an RTA, the greatest predictors of widespread pain are individual-level factors and the psychological milieu, rather than aspects of the trauma itself. Wynne-Jones et al demonstrated prospectively that motor insurance claimants who had been involved in an RTA experienced approximately a doubling in risk of future widespread pain (albeit nonsignificant) compared with claimants not similarly involved in an RTA (risk ratio [RR] 1.9, 95% CI 0.7–4.8) (14). However, this was largely explained by levels of preaccident psychological distress and somatic symptom reporting (adjusted RR 1.4, 95% CI 0.5–3.2). The same authors, in a second study, identified 5 independent risk factors (older age, preaccident adverse health behavior and somatic symptoms, postaccident symptom count, and perceived injury severity) for the onset of widespread pain among 695 individuals involved in RTAs. In combination, these risk factors discriminated between different groups of individuals whose risk of widespread pain onset varied 18-fold, and of the 5 factors, none were accident-specific (15).

However, all of the previous work in this area has identified study participants after the traumatic event. Therefore, even in those that have attempted to adjust for preaccident factors, all have had to rely on all measurements being taken postaccident. It is possible, therefore, that various factors related to the trauma, such as the psychological response to the accident, may have influenced the reporting of preaccident health, pain, and distress. As such, it is not possible to disentangle the potential confounding effect of preaccident versus postaccident distress.

We hypothesize that there are persons defined by prior physical and psychological health who, in the event of a traumatic trigger, are particularly likely to develop CWP. Under this hypothesis, the precise nature of the traumatic event may even be immaterial. The aim of the current study was to examine, prospectively, the relationship between a number of different physically traumatic events and the onset of CWP among persons already well characterized for aspects of physical and psychological health.


The Epidemiology of Functional Disorders Study is a large prospective cohort study in Northwest England, the design and primary results of which have been published previously (16, 17). Briefly, at baseline, 6,290 individuals were recruited from 3 general practices in Northwest England and were sent a self-completion questionnaire. Of these, 6,244 participants answered the question: “Thinking back over the past month, have you had any pain that has lasted for one day or longer?” Those answering positively were asked to shade in the location of their pain(s) on a 4-view body manikin. These manikins, widely used in epidemiologic studies of pain, were then scored in accordance with the definition of CWP in the 1990 ACR classification criteria for fibromyalgia (1).

Potential confounding variables that were measured in the baseline questionnaire included psychological distress (General Health Questionnaire) (18); depression and anxiety (Hospital Anxiety and Depression Scale) (19); illness behavior and illness attitudes (Illness Attitudes Scale) (20); sleep problems (Sleep Problem Scale) (21); and somatic symptoms (Somatic Symptoms Scale) (22).

Persons without CWP were followed up 4 years later by self-completion questionnaire, when the occurrence of (new-onset) CWP was assessed in the same manner as at baseline. Data were also collected concerning the recent experience of 6 physically traumatic events: RTAs, workplace injury, surgery, fracture, hospitalization (for any reason other than the above) and, in women, childbirth. We were interested in traumatic events occurring after the baseline survey; participants were asked about traumatic events in the previous 6 years and were asked to provide the date of these events (or the date of the most recent event in instances where participants had 2 events of the same type [e.g., 2 RTAs]). Events occurring prior to the baseline questionnaire completion were then excluded post hoc.

Statistical analysis.

The association between traumatic events and CWP was examined using logistic regression; therefore, results are expressed as ORs with 95% CIs. All analyses were conducted using Stata, version 10 software. Initially, the relationship was examined (unadjusted) between exposure (traumatic event) and outcome (new-onset CWP). Thereafter, results were adjusted where appropriate for age, sex, general practice, baseline pain status, and other potential confounding variables. The latter were variables assessed at the time of the baseline survey that were associated with both the exposure and outcome with a P value less than 0.25 (by chi-square test). This high cutoff value was chosen to allow adjustment for variables with even a modest potential confounding effect on the relationship of interest.

Ethical approval.

The study received ethical approval from the South Manchester and East Cheshire Local Research Ethics Committees. Approval was also granted from the University of Manchester Ethics Committee.


Of the 6,244 participants at baseline, 5,181 were without CWP and therefore were eligible for followup. Of these, 737 were known to have died or moved away and, of the remainder, 2,069 were successfully followed up at 4 years and provided pain data (adjusted response rate 46.6%). Basic descriptive characteristics of the study population are shown in Table 1. Of the responders, 241 (11.6%) reported new-onset CWP. There was no significant difference between CWP onset in men and women (χ2 = 0.25, P = 0.62) or with age (χmath image = 1.16, P = 0.28). However, there was a statistically significant difference in the prevalence of new-onset CWP between the 3 general practices (range 8.9–17.5%; χ2 = 25.3, P < 0.001) and also between individuals of different baseline pain status: 4.5%, 16.9%, and 17.9%, for those with no pain, regional pain, and widespread (but not chronic widespread) pain, respectively (χ2 = 75.4, P < 0.001).

Table 1. Description of study population
CharacteristicNo. (%)
Men881 (42.6)
Women1,188 (57.4)
Age range, years 
 25–34.9250 (12.1)
 35–44.9514 (24.9)
 45–54.9639 (30.9)
 55–65663 (32.1)
General practice 
 1383 (18.5)
 2463 (22.4)
 31,223 (59.1)
Baseline pain status 
 None881 (42.6)
 Regional1,149 (55.5)
 Widespread (nonchronic)39 (1.9)
New-onset chronic widespread pain (primary outcome) 
 No1,828 (88.4)
 Yes241 (11.6)

A total of 684 participants (37.2%) reported at least 1 traumatic event in the period between baseline and followup, and there was no difference in the proportion of men (36.3%) or women (37.8%) reporting an event (Table 2). However, men were more likely than women to report an injury at work (4.1% versus 2.4%) and less likely to report surgery (14.6% versus 18.5%). Men were also slightly more likely to report involvement in an RTA (7.2% versus 5.2%), although this was only of borderline significance. In women, 5.0% (n = 59) reported childbirth by either natural or assisted delivery.

Table 2. Self-reported traumatic events during 4-year followup
Traumatic eventMen, no. (%)*Women, no. (%)*P
  • *

    Denominator differs depending on nonresponse to specific questions.

  • By chi-square test.

  • Only includes individuals answering “yes” to ≥1 exposure or “no” to all exposures.

Any event283 (36.3)401 (37.8)0.51
Road traffic accident62 (7.2)61 (5.2)0.06
Injury at work35 (4.1)28 (2.4)0.03
Fracture33 (3.9)50 (4.3)0.61
Surgery117 (14.6)198 (18.5)0.03
Hospitalization174 (21.3)266 (24.3)0.12
Childbirth59 (5.0)

Participants with new-onset CWP were approximately one-third more likely to report a traumatic event in the 4-year period since baseline than other individuals (OR 1.34, 95% CI 0.996–1.80). This association was largely driven by RTAs, fractures, and injuries at work. In contrast, no association was found with hospitalization, surgery, or (in women) childbirth (Table 3). However, after adjustment for baseline psychological variables and sleep problems, no association was seen between the reporting of a traumatic event and the onset of CWP (OR 1.01, 95% CI 0.73–1.40).

Table 3. Association between CWP onset and physically traumatic events*
Traumatic eventWith CWP, no.Without CWP, no.OR (95% CI)
CrudeAdjustedFurther adjusted
  • *

    CWP = chronic widespread pain; OR = odds ratio; 95% CI = 95% confidence interval.

  • For age, sex, general practice, and baseline pain status.

  • For psychological distress, anxiety, depression, adverse health behavior, and sleep problems.

  • §

    For anxiety and sleep problems.

  • For psychological distress and adverse health behavior.

  • #

    For psychological distress, anxiety, and adverse health behavior.

  • **

    For psychological distress, depression, adverse health behavior, and sleep problems.

  • ††

    For depression, adverse health behavior, health anxiety, and sleep problems.

  • ‡‡

    For psychological distress and health anxiety.

Any event     
 Yes885961.34 (0.996–1.80)1.22 (0.90–1.65)1.01 (0.73–1.40)
Road traffic accident     
 Yes201031.55 (0.94–2.56)1.84 (1.10–3.11)1.50 (0.89–2.52)§
 Yes13701.50 (0.82–2.76)1.46 (0.78–2.75)1.22 (0.64–2.33)
Injury at work     
 Yes9541.30 (0.63–2.66)1.31 (0.62–2.77)1.03 (0.47–2.24)#
 Yes483920.98 (0.70–1.39)0.85 (0.60–1.21)0.77 (0.54–1.11)**
 Yes332820.94 (0.64–1.40)0.89 (0.59–1.33)0.77 (0.50–1.18)††
Childbirth, women     
 Yes6530.84 (0.36–2.01)1.16 (0.42–3.15)0.83 (0.35–2.00)‡‡

After adjusting for age, sex, general practice, and baseline pain status, those who reported an RTA experienced an 84% increase in the likelihood of new-onset CWP (OR 1.84, 95% CI 1.10–3.11). This relationship remained, although was no longer significant, after further adjustment for other potential confounding variables (OR 1.50, 95% CI 0.89–2.52). Whereas, although modest associations were initially observed between CWP and fracture, and with workplace injury (Table 3), these were greatly attenuated after multiple adjustment (OR 1.22, 95% CI 0.64–2.33 and OR 1.03, 95% CI 0.47–2.24, respectively).


We have demonstrated that the onset of CWP over a 4-year period is ∼12%. Further, we have shown that persons exposed to an RTA, but not those exposed to other traumatic events, experienced an increase in the likelihood of CWP onset.

There are a number of methodologic issues to consider in the interpretation of these results, with the first being sample attrition. Of the 4,444 baseline participants eligible for followup and not known to have died or moved away, only 46.6% were successfully followed up at 4 years. Loss to followup results in a smaller sample size, decreasing the likelihood of observing statistically significant findings and widening CIs around the OR. Also, in the current study, participants lost to followup were more likely to be young and male, and there was some evidence to suggest that they were more anxious and more depressed at baseline but reported fewer sleep problems. The key issue, however, is whether selective attrition could have introduced a bias to our results. If sample attrition has compromised the internal validity of the study, one would have to argue that the relationship between trauma(s) and CWP onset is different among persons who did, versus did not, participate at followup. We consider this to be unlikely. However, it is possible that those with a prior trauma and potentially with CWP are less likely to be followed up due to their pain levels. Alternatively, one could reasonably propose that the group least likely to participate at followup is the healthy nonexposed (i.e., the nontrauma, non-CWP group). In either instance, our results will be biased toward the null; we will have underestimated the occurrence of new-onset CWP in the trauma group(s) and the true effect sizes will be greater than those we present.

Second, all the traumatic events in the current study were assessed by self-report. It is possible, therefore, that a recall bias may exist, whereby individuals with/without CWP recall (or report) traumatic events to a different extent. One can easily conceive of a scenario where such recall bias could occur. What is not so obvious is why it should only occur with RTAs and not other traumatic events, although RTAs may be subject to psychological after-effects and insurance implications that are less prominent with other traumas. In particular, we inquired about hospitalization (although this is not necessarily a physically traumatic event). Persons with CWP are a population known for higher than average levels of consultation for health care (23). Although it seems probable that if a differential recall exists it will be particularly prominent for the reporting of this exposure, persons with CWP in the current study were no more likely to report a recent hospitalization than other individuals (OR 0.98, 95% CI 0.70–1.39).

Previous studies have been unable to adjust satisfactorily for potential baseline confounding variables, since all preRTA data have been collected postRTA. One of the methodologic strengths of the current study is that, unlike previous studies in this area (12–15), potential confounding variables were recorded prior to the RTA, therefore eliminating the possibility that psychological response to the RTA may have influenced the reporting of preaccident health, pain, and distress. In addition, the participants in the current study were without CWP prior to their RTA, i.e., it is known that they were without CWP at baseline. What is not known, however, is the temporal course of events between baseline, the RTA, and the onset of CWP. It is not possible to determine whether participants experienced an RTA and then developed CWP, or whether they developed CWP prior to their RTA. Clearly, in terms of etiology, this is an important issue. However, in the case of the latter, for this to have influenced our results, one would have to argue that CWP onset increases the likelihood of reporting involvement in an RTA, but not the reporting of other traumatic events, including hospitalization. Again, we would hypothesize that this is unlikely. What cannot be ruled out with certainty, however, is that any potential confounding variables are not associated with the risk of exposure, i.e., with the risk of a traumatic event. If this is the case, and these variables are, in fact, path variables and not confounders, then statistical adjustment is inappropriate.

We found the incidence of CWP to be 11.6% over a 4-year period. This is possibly higher than one might expect, given what is known about the prevalence of the condition. Other authors have reported a prevalence of 11–13% in a similar population (2, 3). The reason for the high prevalence of new-onset CWP is unknown, although it may be a function of surveillance, i.e., CWP may be better ascertained in a repeatedly monitored population than in the general population. However, in the current study, even the baseline prevalence of CWP was high (17%), although previous work has shown that this is explained, at least in part, by (lower) socioeconomic status of the current study population (24). Although it would have been preferable to have an individual-level marker of socioeconomic status, such as income or social class, in the current analysis we were only able to adjust for socioeconomic status using a cruder, group-level proxy variable, i.e., general practice.

Although there is an increase in the likelihood of CWP onset associated with involvement in an RTA, we have shown that this is not true for other physically traumatic events. The reason for the specificity of this relationship is unclear. It may be that the extent of the initial injury will determine who develops CWP, although this was not examined in the current study. Again, however, this would not explain the specificity of the relationship to RTAs. To explain the current findings, one would have to suppose that the RTAs resulted in greater initial injury than the other traumatic events, including events such as bone fracture. Also, it is possible that some of the events overlap, e.g., some of the fractures/hospitalizations/surgery resulted from RTAs, although it is not possible to distinguish between these in the current study.

The psychological response to the unplanned nature of RTAs may be important. This might also explain the small increase in the likelihood of CWP onset associated with fractures and injury at work (also unplanned) although, interestingly, these latter relationships were attenuated (completely so in the case of workplace injury) by adjustment for baseline psychological variables.

So might there be something peculiar to an RTA that leads people to expect a poor outcome and maybe influence natural history? There is some evidence in support of this question. A study conducted in Lithuania revealed that in a country where compensation for whiplash injury was infrequent, and expectation of long-term disability was minimal, there was no evidence that car accidents led to chronic neck problems (25). Or, it may be that there are a number of influences on the reporting of chronic symptoms, including perceived “trigger” events and even, in the case of RTAs, issues surrounding fault and compensation. A recent survey demonstrated that persons with fibromyalgia, of which CWP is a major symptom, commonly attribute their symptoms to a physically traumatic event: 16.1% of respondents reported that symptoms were triggered by an RTA (11). It may be that this perception of a causal pathway, and potential over-reporting of trigger events (compared to perceived nontrigger events) may partly explain the current findings. However, this previous study also found that the same proportion of respondents reported surgery as a trigger event, and slightly more (17.1%) reported a nonRTA physical injury.

In summary, previous work has shown that patients with CWP often attribute symptom onset to a precipitating physically traumatic event and, although some studies have reported associations between such events and the onset of CWP, they have been unable to adequately adjust for pretrauma psychological health, which may confound any observed relationship. The current study provides some support for the hypothesis of an “at risk” phenotype, where individuals characterized by poorer health and psychological variables may be predisposed to develop CWP following a traumatic trigger event. We have demonstrated that persons with new-onset CWP are 34% more likely to report a prior, recent, and traumatic event than individuals who remain without CWP. We have also shown that involvement in an RTA specifically does appear to confer a modest increase in the likelihood of symptom onset over the short to medium term, although in general this effect is removed after adjusting for potential confounding factors. Future research should examine what it is about involvement in an RTA, or about one's reaction to an RTA, that confers this increase in the risk of CWP onset, which does not seem to occur with other traumatic events.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Jones had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Jones, McBeth, Morriss, Dickens, Macfarlane.

Acquisition of data. Nicholl, McBeth.

Analysis and interpretation of data. Jones, Nicholl, McBeth, Davies, Macfarlane.


We are grateful for the participation and help of the doctors, staff, and patients of the 3 general practices in Greater Manchester, and Alison Littlewood, Ruth Fullam, Joanne Bradley, and Karen Schafheutle for survey administration.