Fibromyalgia (FM) is a disorder characterized by chronic, widespread pain (CWP) with increased sensitivity for mechanical pressure in at least 11 out of 18 defined tender points (TP)[1]. FM is associated with female gender, low income and education, fatigue, irritable bowel and bladder symptoms, headache, negative mood and sleeplessness[2, 3], previous pain[4] and stress induced HPA axis reactivity[5]. The prognosis is uncertain, and there are few and inconclusive prospective studies. FM pain appear stable from two to ten years in some natural course studies[6–13], while others demonstrate aggravation[14], improvement[15] or mixed aggravation and improvement[16]. Long-term follow-ups after psychological interventions have shown small effects on pain, but improvement in quality of life (QoL)[17]. Guided imagery has improved FM pain in the short term[18], but not in the long-term[19]. Cognitive behavioral therapy (CBT) has improved worry, pain behavior, pain coping, depression and reduced healthcare-seeking behavior in prospective studies[20, 21].

According to the neuromatrix pain theory, the perception of pain is dynamically influenced by emotions, motivations, memory and other sensory and homeostatic inputs[22–24]. Thus, pain is always experienced in a context and influenced by it[25]. Context affects a variety of health measures including blood-pressure, which appears lower at home than to a doctor[26, 27]. Also memory and learning[28] as well as anxiety and fear-conditioning[29, 30] are affected by context. The impact of contextual factors for the experience of pain has been demonstrated in brain imaging studies and laboratory experiments[31, 32]. For example, animal studies have shown that one drug has different effects in dissimilar contexts[33, 34].

In a Norwegian study, healthy males reported less laboratory pain in a context with female investigators, compared to a context with male investigators, while females reported the same pain regardless of the investigators’ gender[35]. Perhaps the most obvious example of context influencing pain is the discrepancy observed in clinical and experimental contexts[36]. However, there are few studies investigating variations in the reporting of pain by contextual factors among patients with chronic pain. In one study, variations in weather were related to daily variations in pain among FM patients[37]. Other studies have shown that the social contexts are important for the acceptance of chronic pain[38] and the appraisals of pain threats[39].

In this study we monitored a cohort of female FM patients where pain was measured at home and in a doctor office the same day (baseline) and reexamined 14 years later. Our main research question was to examine if pain was context-dependent or time-dependent, i.e. if pain would vary by two different social contexts in one day and by a 14 years’ time lag from baseline to follow-up in one context. As a secondary aim we wanted to investigate if vulnerability factors were associated with pain in the different contexts and time points.

None of the baseline variables showed any differences between the completers and the 14 years FU non-completers on one-way ANOVA for the continuous variables and Pearson’s Chi-square and Fisher exact test for the categorical variables. Pain improved with psychological treatment (guided imagery) after 30 days, as shown in the original study[18].

Mean pain was worse to a doctor than at home the same day, but appeared unchanged after 14 years in the same context. The pain recordings at the doctor’s office were not significantly correlated, indicating that pain was not stable within individuals, and our findings suggest that some factors may explain the individual differences for pain in context and time.

Pain measured to the doctor was positively associated with the trait-anxiety (STAI-T) and insomnia measures at baseline, findings which are in accordance with other studies about pain and anxiety[49]. For blood-pressure measurements, which tend to be higher to a doctor than at home, (the so-called white-coat syndrome), anxiety of previous visits to the physician’s office probably has an effect on subsequent recordings[26]. Previous research supports our findings that insomnia is associated with FM pain in the short-term perspective[3]. Until recently, it has been unknown whether poor sleep contributes to FM in the long-term. In a new study by Mork & Nilsen, sleeplessness predicted fibromyalgia during 10 years in a previous healthy population[50].

When pain was measured at home the same day, previous pain mattered for pain worsening, a finding in accordance with other studies[13, 51], while amitriptyline appeared as a resilience factor. Some analogue studies have shown that drug-effects may depend on their context, the so called place preference or place aversion (classical) conditioning[33, 52].

Baseline depression, long pain-duration, being single and OCD-traits were associated with worse pain 14 years ahead. These findings support previous literature that depression is a risk factor for long-term FM symptoms, e.g.[14]. However, depression may act both as a risk factor and a consequence of pain[53] and vary with third factors[54, 55]. Self-critical perfectionism, an active generation of stress, stress sensitivity and levels of depression are features which may explain the association between OCD and forthcoming pain[56]. In contrast, education seemed to be a resilience factor for future pain in our study, a finding in accordance with previous findings that FM patients have lower educational level than non-patients, e.g.[57]. Guided imagery also appeared as a resilience factor in this 14 years follow-up of the Fors et al. 2002 study[18], which supports findings that relaxation, imagery and cognitive training may predict reductions in pain compared to untreated patients[58]. Syrjala et al.[59] and van Kuiken et al.[60] have revealed results which show that guided imagery have positive effects the first five to seven weeks after treatment, but that the effects seem to yield after 18 weeks[61] in reducing persistent pain[19, 62], but it has not been investigated in a long-term prospective studies previously.

This study brings new information about how pain vary with context among patients with FM. A strength of the study was that pain was measured in two different contexts the same day and two similar contexts with long term follow up.

Several limitations of this study needs to be addressed: The statistical model used for associations and prediction of pain (in a doctor context, in a home context, and in a doctor context after 14 years) was an ANCOVA linear regression model, which was fitted to the data using a backwards stepwise selection using F-tests with 0.05 as critical value, so that we ended up with a model where each covariate was significant in presence of the others. The low number of participants (n = 29), leading to low power may have prevented discovery of potential predictors. Also, the strict significance level of 0.05 may have possibly omitted important predictors from the models. On the other hand, the multiple testing procedure to select covariates in stepwise regression in general may lead to too small p-values and possibly false findings. Hence the findings may be regarded as hypothesis-generating findings more than true predictions. There was great individual variability among the pain sufferers, which may explain why individual baseline current pain did not predict future pain. Another weakness in the study was that the frequency of the guided imagery use was not measured, so the effect of this intervention may be spurious. Another possible limitation of our study was the measurement of just current pain. In chronic pain studies many investigators measure pain scores as the mean sum of current+ last week + maximal + minimal pain, as seen in e.g. the Brief Pain Inventory[63]. However, we would expect a current pain measure to be more sensitive to variations in context than an aggregated measure of pain. Thus, contextual pain cues should be accounted for when pain is monitored and treated, e.g. by focusing more on home-measured pain and doctor office context.

Our study suggests that contexts should be accounted for when reporting current pain. It seems to matter where and how the pain is recorded. Our study also suggests that the factors explaining the variance in pain may vary by contexts. The implication of our findings would advocate a new approach to pain by focusing more on home-measured pain and modifying the context or environment of the doctors’ office if possible, e.g. to emphasize comfort and empathy more[64] in order to reduce pain.