The results of the study are consistent with previous research showing that scores on a measure of pain intensity are significantly correlated with scores on a measure of fatigue . The findings of the present study extend previous research in showing that, through the course of a behavioral activation intervention, early treatment changes in pain predicted later changes in fatigue; but early treatment changes in fatigue did not predict later changes in pain intensity. The findings do not support a bi-directional relation between pain and fatigue. To our knowledge, this is the first study to show that, during a rehabilitation intervention, early changes in pain severity predict later changes in fatigue.
Temporal Relation between Pain and Fatigue.
The results of systematic reviews suggest that fatigue is a common symptom reported by individuals with persistent pain conditions [1, 23]. Several prospective studies have reported that changes in pain precede changes in fatigue [8, 24, 25]. For example, Feuerstein et al  reported that changes in pain intensity were prospectively associated with changes in the severity of fatigue in patients with low back pain. Similar findings were reported by Nicassio et al  in a sample of patients with fibromyalgia. The findings of Feuerstein et al  and Nicassio et al  suggest that the functional linkage between pain and fatigue is proximal in time; in both studies, variations in pain significantly predicted next-day pain. Christie et al  examined the time-lagged relation between pain and fatigue in patients with rheumatoid arthritis over a one-week period. Their findings suggested that pain prospectively predicted later fatigue only in a subset of patients. For the majority of patients, the relation between pain and fatigue was synchronous.
Several different mechanisms have been proposed to account for a temporal relation between pain and fatigue. Resource depletion accounts have been proposed where it has been suggested that the physical or psychological energy required to manage or cope with persistent pain ultimately depletes energy resources contributing to the experience of fatigue . Olson and colleagues have conceptualized fatigue as a component of the stress response . Van Damme et al  suggest that pain-related fatigue is the subjective experience of disengagement from goal pursuits that occurs when the costs of pursuing activity goals exceed the benefits. The disruptive effects of pain on sleep have also been discussed as contributing to pain-related fatigue .
Research has also pointed to a number of physiological processes that might give rise to fatigue associated with pain. It has been suggested that systemic inflammation might underlie the association between pain and fatigue [30, 31]. For example, the inflammatory phase accompanying musculoskeletal injuries of the spinal column induces proinflammatory cytokines implicated in the multifidus muscle remodeling. These changes can lead to muscle fatigue as replacement of muscle fibers by adipose tissues weakens the muscle and decreases muscle endurance . Other potential mediators of the relation between pain and fatigue include reductions in heat shock protein (HSP), brain-derived neurotrophic factor (BDNF)  and lower dopamine levels .
Temporal Relation between Fatigue and Pain.
Fewer studies have provided support a prospective relation between fatigue and pain. Studies examining the trajectory of symptoms reported by primary care patients have revealed that symptoms of fatigue prospectively predict the later emergence of pain symptoms [11, 12]. In a community health survey of high school students, Siivola et al  reported that participants who reported symptoms of fatigue were more likely to have developed a musculoskeletal pain condition when followed up 7 years later. Halder et al  reported that high scores on a measure of fatigue predicted onset of abdominal pain in a community sample of adults when re-assessed 1-year later. Aili et al  reported that symptoms of fatigue pre-dated the onset of pain in a large sample of individuals who later (5 years) developed chronic widespread pain.
Our failure to replicate previous findings showing a temporal relation between fatigue and pain might be due of several factors. First, the time frame of our study might not have been adequate to capture the time-dependent nature of the relation between fatigue (as a causal factor) and pain. Studies that have yielded findings supportive of a temporal relation between fatigue and pain have had follow-up periods ranging from 1 to 7 years. The timeframe of the present study extended only over a 10-week period. As well, studies reporting evidence of a temporal relation between fatigue and pain sought to determine the role of fatigue symptoms as a prognostic indicator of the onset of a pain condition at a later point in time. The present study addressed whether early treatment-related reductions in fatigue predicted later reductions in pain. It is possible that the mechanisms that underlie the prognostic value of fatigue for the onset of a future pain condition differ from those that are associated with treatment response.
It is important to consider that the analytic procedures used in all studies supporting a temporal relation between fatigue and pain have used group-based procedures, and have not controlled for within-person associations. In all studies, the analytic approach involved the use of logistic regressions using categorized groups of fatigue with a binary outcome (presence of pain or not) to calculate odds ratios. As noted earlier, group-based procedures are associated with a higher rate of Type I error. In support of this explanation, when the data from the present study were analyzed using traditional CLPM, which does not control for within-person variance, results supported a bi-directional relation between pain and fatigue. However, when analyzed using RI-CLPM procedure, controlling for within-person variance, the paths from fatigue to pain were no longer significant.
Pain is an important determinant of magnitude of disability associated with musculoskeletal conditions . Emerging research suggests that fatigue is a symptom that frequently co-occurs with musculoskeletal pain, and further adds to the burden of disability [36, 37]. Indeed, research shows that treatment-related reductions in fatigue prospective predict resumption of occupational activities in individuals with musculoskeletal conditions . Surprisingly, there is a paucity of research that has addressed how best to manage debilitating symptoms of fatigue to promote rehabilitation progress in individuals with musculoskeletal conditions.
In the present study, participation in a behavioral activation intervention was associated with a 20% reduction in fatigue, corresponding to a moderate effect size (d = .66). The effect size was higher when analyses were conducted only on the subsample of participants who initially presented with severe (> 6/10) symptoms of fatigue (d = .71). In previous research, reductions in fatigue of 20% (medium effect size) have been shown to be sufficient to translate into meaningful clinical outcomes such as return to work .
The pattern of findings of the present study suggests that early-treatment reductions in pain might contribute to reductions in fatigue in the rehabilitation of musculoskeletal injury. In turn, reductions in fatigue might promote fuller engagement in treatment, and contribute to more positive rehabilitation outcomes . The data also suggest that only minor reductions in pain severity were required to achieve meaningful reductions in fatigue. It is not clear however, that all techniques, modalities or medications that reduce pain severity will also be associated with reductions in symptoms of fatigue in individuals with musculoskeletal conditions. It will be of interest to determine whether other treatments typically used in the management of chronic pain, such as medication, heat/ice, laser, dry needling or electrical stimulation are also associated with reductions in fatigue.
Some degree of caution must be exercised in the interpretation of the study findings. First, data records were drawn from the clinical files of individuals referred to an occupational rehabilitation service. Only a minority of individuals with debilitating pain conditions are referred for occupational rehabilitation services. In addition, all participants were receiving long-term disability benefits. These sample characteristics necessarily have implications for the generalizability of findings. The modest sample size also limited the nature of analytic procedures that could be applied to the data. As well, it was not possible to examine the possible mediating or moderating role of other variables that could potentially influence the relation between pain and fatigue (i.e., type of medication, sleep, co-morbid health or mental health conditions). It is important to note that many participants in the present study were receiving other concurrent treatments such as medication or physical therapy. As such it is not possible to unambiguously attribute changes in pain and fatigue to the behavioural activation intervention.
In spite of these limitations, the results of the present study suggest that there is a temporal relation between symptoms of pain and fatigue in individuals with musculoskeletal conditions. The findings suggest that early treatment reductions in pain severity might contribute reductions in fatigue, and in turn, foster fuller participation in rehabilitation interventions. More research is needed to better understand the pathways underlying the temporal relation between pain and fatigue. More research attention also needs to be directed toward the development of more effective means of reducing fatigue in individuals with chronic musculoskeletal pain.