Testing Interventions for Patients With Fibromyalgia and Depression

Chronic pain is a word-wide major health problem, which seriously affects the quality of life of those who experience it and causes important socio-economical costs (Tsang et al., 2008). According to the results of the last Chilean National Health Survey (2010), 34.2% of Chileans older than 15 years present pain ranging from moderate to severe. The pain experienced by a considerable proportion of these individuals (73.5%) is chronic. Among Chileans, chronic pain has been found to be related to more sickness leaves, a five-time increase in the use of primary care services, and the presence of depressive disorders (Miranda et al., 2013). In fact, around 35% patients with chronic pain suffer from depression (Gracely, Ceko, \& Bushnell, 2012). In the case of some specific pain disorders such as fibromyalgia, this comorbidity is even greater (life prevalence around 90% for depressive symptomatology and 62-86% for major depressive disorder; Gracely et al., 2012). As a result, some authors posit that fibromyalgia and depression are two manifestations of a single affective spectrum disorder (Gracely et al., 2012). In fact, according to some evidence, both disorders share pathophysiological aspects. For example, pharmacological treatment of both fibromyalgia and depression includes the same dual serotonergic and noradrenergic agonists, namely, amitriptyline and duloxetine (Maletic \& Raison, 2009). Depressive symptoms impair chronic pain treatment and are associated with a worse chronic pain prognosis (Li, 2015). The increasing and concomitant prevalence of depression and chronic pain represents a burden for the primary care systems. Providing effective and appropriate primary care interventions to address this comorbidity is indeed needed, considering that chronic pain and depression lead to great disability. Taking into account that reducing disability is part of the second strategic objective of the Health Goals of the Decade 2011-2020 for Chile, the design and dissemination of therapeutic strategies aimed at the treatment of these problems is a priority in this country. Behavioral Activation (BA) is an evidence-based psychological intervention for depression (Cuijpers, Van Straten, \& Warmerdam, 2007; Dimidjian, Barrera, Martell, Muñoz, \& Lewinsohn, 2011). It has been shown to be as effective as comprehensive cognitive behavioral therapy (Cuijpers et al., 2007) and antidepressant therapy (Dimidjian et al., 2006), with a lower dropout rate than the latter (Dimidjian et al., 2006). There is also evidence that BA works as well as medication and better than cognitive behavioral therapy in the treatment of severe depression (Dimidjian et al., 2006). Moreover, several meta-analyses support the efficacy of BA for the treatment of depression (Cuijpers et al., 2007; Ekers, Webster, Van Straten, Cuijpers, Richards, \& Gilbody, 2014; Mazzucchelli, Kane, \& Rees, 2009). According to the result of the last meta-analysis conducted (Ekers et al., 2014) - in which 26 randomized controlled trials (RCTs) were analyzed with a total of 1,524 participants - the treatment with BA is superior to the control condition (usual treatment, placebo, or waiting list), showing a high effect size (standardized mean difference, SMD: -0.74, 95% CI -0.91 a - 0.56). In addition, the examination of four RCTs with 283 participants showed that CT is also superior to medication. In this case the effect size was moderate (SMD -0.42, IC -0.83 to -0.00). Finally, according to the results of different studies conducted with several populations, BA can be used in multiple contexts and populations (Dimidjian et al., 2011) such as adolescents (Jacob, Keeley, Ritschel, and Craighead, 2013), seniors (Quijano et al., 2007), university students (Gawrysiak, Nicholas, \& Hopko, 2009), and primary care patients (Uebelacker, Weisberg, Haggarty, \& Miller, 2009). More research on the efficacy of BA is still needed, especially with regard to its effect after follow-up; however the results reported so far are promising. Furthermore, BA is much easier to apply and disseminate than other psychological treatments. For these reasons, there is now an increasing interest in examining the efficacy of this intervention in populations with different comorbidities such as obesity, cancer, anxiety, borderline personality, substance abuse, childhood abuse history, smoking, and post-traumatic stress disorder, among others (Dimidjian et al., 2011). BA could be particularly suitable for treating depression among patients with chronic pain, as some of its main purposes are fighting inactivity, increasing environmental reinforces, and decreasing aversive experiences. Indeed, inactivity, loss of reinforces, and excessive unpleasant experiences seem to play a key role in the etiology of both depressive disorders and chronic pain (Dimidjian et al., 2011, Leeuw et al. 2007). According to Lewinsohn's integrative model of depression (Lewinsohn, Hoberman, Teri, \& Hautzinger, 1985), when a person with certain vulnerabilities faces a stressful event, that event and the emotions associated with it interrupt their normal behavior patterns, reduce the availability of the reinforcers in the environment, and increase the rate of aversive experiences. All this finally leads to depression. Depressive symptoms have a series of behavioral, cognitive, and emotional correlates that make the subject more vulnerable to new stressful events, thus creating a pathological cycle that is self-perpetuating in time. Something similar occurs when a person suffers a painful injury. According to Vlayen and Linton's Fear and Avoidance Model (FAM, 2000), in the face of a painful injury, vulnerable individuals (with increased pain catastrophizing or increased pain-related fear) develop strategies aimed at avoiding pain and the activities that cause it. This leads to inactivity, loss of reinforcers, disability, and depression, and consequently, to the exacerbation of the pain intensity and pain-related anxiety and catastrophizing. The patient is trapped in a vicious circle of symptoms that feed-back each other and perpetuate over time. Taking into account the FAM assumptions, as well as the purpose of the BA, this treatment may likely not only be useful for reducing depression in patients suffering from pain, but also for reducing the intensity of their pain and the anxiety and catastrophism associated with it. BA could break the chain of perpetuation in both disorders. In fact, there are preliminary data that support this premise. In a case study, in which BA was used with a woman who had been suffering from chronic pain associated with a diagnosis of fibromyalgia for 11 years, a clinically significant decrease in depressive symptoms was observed after treatment. In addition, the degree in which the pain interfered with the patient's daily activities was reduced (in a 100%), as well as pain-related anxiety and the use of pain medication use. These results (with the exception of reduction in medication use) were maintained within three months of the intervention (Lundervold, Talley, \& Buermann, 2006). These same authors replicated these findings in another study, in which they applied the same treatment to another woman who had been suffering from fibromyalgia for 22 years. In this case, the decreases of pain intensity, pain-related anxiety, depression, and medication use were maintained three and six months after the intervention (Lundervold, Talley, \& Buermann, 2008). Behavioral activation is part of some cognitive behavioral pain management treatments (Ehde, Dillworth, \& Turner, 2014) that include - in addition to BA - cognitive re-structuring. There is evidence that such treatments are effective in reducing symptoms associated with chronic pain (Ehde et al, 2014; Williams, Eccleston, \& Morley, 2012). The size of the effect of such treatments varies from small to moderate, depending on the outcome variable analyzed. Thus, according to the results of the meta-analysis conducted by Williams and colleagues (2012), compared with the usual treatment, cognitive-behavioral interventions developed for pain management have small but significant effects on pain (SMD = -0.21, CI 95% CI = -0.37 to -0.05, Z = 2.35, p \<0.05) and disability (DME = - 0.26, CI 95% = -0.47 a -0.04, Z = 2.35, p \< 0.05), and moderate effects on mood (DME = -0.38, 95% CI = -0.57 to -0.18, Z = 3.84, p \<0.01) and catastrophism (DME = -0.53, 95% CI = -0.76 to -0.31, Z = 4.58, P \<0.01). However, in relation to other active treatments, their effects on disability (DME = -0.19, 95% CI = -0.33 to -0.05, Z = 2.66, p \<0.01) and catastrophism (SMD = -0.18, 95 % = -0.36 to 0.00, Z = 1.92, p = 0.05) appear to be small, and no significant effects have been found for pain and mood. In addition, further studies are needed to examine the active components of cognitive behavioral therapy for pain (Ehde et al., 2014). Although BA may be one of the effective components of cognitive behavioral therapy, the intensity and emphasis placed on behavioral activation within cognitive behavioral treatments is small, and is generally limited to use in one session, so it is not known yet whether BA in isolation could have beneficial effects on pain. Furthermore, to our knowledge, no studies have been conducted to examine the efficacy of BA to reduce depressive symptoms in patients with chronic pain (beyond the abovementioned case studies). Finally, it is important to highlight that BA has multiple advantages that make it very efficient in terms of costs and benefits. First, it is a brief therapy, very easy to apply and disseminate (Dimidjian et al., 2011). In addition, it can be implemented by non-psychologists after appropriate training (e. g., nurses or social workers; Ekers, Richards, McMillan, Bland, \& Gilbody, 2011). It can also be implemented in group format (Houghton, Curran, \& Saxon, 2008; Porter, Spates, \& Smitham, 2004) and in the context of primary care (Dimidjian et al., 2011). Finally, there are preliminary results that support its effectiveness in the Latino population (Bianchi \& Henao, 2015; Kanter, Santiago-Rivera, Rusch, Busch, \& West, 2010). Taken together, these advantages make BA of particular interest in the Latin American context, where economic limitations often do not allow for long and complex treatments, and access to trained therapists is usually limited. Therefore, the purpose of the present study is to examine the effectiveness of BA in a sample of Chilean women with chronic pain and comorbid mayor depression. For several reasons, the investigators will study the efficacy of this treatment in women with the same diagnosis of chronic pain, namely fibromyalgia. First, because this is the first RCT examining the efficacy of BA in patients with chronic pain, it is advisable to include patients with the same diagnosis to increase the homogeneity of the sample. Secondly, BA may be more effective in this type of patients, as there is a high comorbidity between fibromyalgia and depression and both disorders have been hypothesized to be manifestations of a single affective spectrum disorder (Gracely et al., 2012), Finally, preliminary results on the effects of BA in patients with pain have been reported in women with fibromyalgia. The investigators expect to find that, after treatment, the women in the experimental arm will experience greater decreases in depressive symptoms severity as well as in the pain-related variables assessed (pain intensity, fibromyalgia impact, pain-related anxiety, catastrophism, and physical health symptoms severity) than women in the comparison arm. In addition, the percentage of women in remission from the diagnosis of depression (as well as the percentage of women responding to treatment) will be greater in the experimental arm than in the comparison arm. Similarly, the percentage of women who show a clinically significant reduction in pain intensity (decreases greater than two units in the pain intensity scale) will be greater in the experimental arm than in the comparison arm. Finally, the investigators hypothesize that the decreases observed in the pain-related variables will be mediated by decreases observed in depressive symptoms severity. Regarding the outcomes assessed at a three months follow-up, the investigators expect to find that the differences found between the two groups in the depressive symptoms severity and the pain-related variables decreases will be maintained at follow-up. Furthermore, at follow-up the percentage of women in remission from the diagnosis of depression (as well as the percentage of women responding to treatment) will continue to be greater in the experimental arm than in the comparison arm. Similarly, after follow-up, the percentage of women who show a clinically significant reduction in pain intensity will continue to be greater in the experimental arm than in the comparison arm. The differences found after the follow-up in the pain-related variables will be mediated by the differences found in depressive symptoms severity. Method Participants: Women diagnosed with fibromyalgia and major depression attended at the Service of Family Medicine, Reumathology or Traumatology of the UC-Christus Health Network (N = 90) will participate. The study sample size was calculated for a Hierarchical Linear Model (HLM) in which the primary outcome (changes in depressive symptoms severity) was the dependent variable. Using Mplus, the investigators determined that a sample of 90 participants was needed to reach a power of .80. In this power analysis, the correlation between the intercept and the slope was specified to be -0.4. The average slope was assumed to be negative, as depression scores are expected to drop, and was specified as -0.5. To simplify the model, the variances of the intercept, slope, and scores of all the variables in the model (depressive symptoms severity, and covariates) were specified as 1, so as to have the standardized model solution. The treatment effect on the slope (standardized Beta) was specified as 0.45. This corresponds to a medium-to-large effect size (Cohen, 1992). In addition, the effect of two covariates was included. The effect of each of the covariates on the slope was specified as 0.1. The correlation between treatment and each covariate was specified as 0.1, and the correlation between covariates was specified as 0.3. Design The present study is a randomized clinical trial with a parallel design. After the diagnosis interview, participants will be randomly assigned to two intervention arms: an experimental arm - that will be treated with BA in a group setting in addition to usual care (UC) for fibromyalgia and comorbid depression- and a comparison arm that will receive only UC for fibromyalgia with comorbid depression. The research design is longitudinal and mixed (within and between subjects repeated measures). Outcomes will be assessed in both groups before, during and after the intervention, as well as after a three-month follow-up, by a research assistant who will be blind to group assignment. Procedure Participants will be recruited through several procedures. First, posters and brochures with information about the study will be available at the medical attention boxes, so that patients interested in participation could contact the research team by email or phone. In addition, one of the nurses of the medical center will contact by phone all the fibromyalgia patients registered in the medical center to ask them for permission to put them in contact with the study research team. Participants who agree to be contacted will received a phone call by a research assistant, who will provide them with further information about the study and - in order to corroborate that they indeed meet the inclusion criteria for fibromyalgia and mayor depression -ask them to respond to the Fibromyalgia Survey Questionnaire (FSQ; Wolfe et al., 2011; Carrillo de la Peña et al., 2015) and the Patients Health Questionnaire-9 (PHQ-9; Baader et al., 2008). Participants who meet inclusion criteria will be scheduled for a diagnoses interview according to Diagnostic and Statistical Manual of Mental Disorders IV (DSM-IV) criteria. The Mini-International Neuropsychiatric Interview (MINI; Ferrando, Bobes, \& Gilbert, 1999; Sheehan \& Lecrubier, 1999) will be applied in order to make the diagnoses. In addition, the Mini-Mental State Examination (Folstein, Folstein, \& Mchugh, 1975, González-Hernández et al., 2009) will be administered. All participant will be asked for informed consent before participation. Participants who meet all the inclusion criteria after this interview will be randomized to the two arms. Randomization will be conducted by a research assistant (other than those who will perform the assessments, treatment and data analysis) using a computer program. As a compensation, patients will receive gift card (with a value of 10.000 Chilean pesos) after each outcome assessment interview. Data analyses In order to test whether there are differences between the groups in the sociodemographic and clinical characteristics assessed at baseline, Student's t-tests will be performed for continuous variables and Chi-square test for categorical variables. If there are small or empty cells in the categorical tests, the Chi-square test will be replaced by an Exact Fisher's Test. Those variables in which statistically significant differences are found will be introduced in the analyses as covariates. In order to examine the differences between the groups in the reduction of the primary and secondary outcome, HLM will be performed using MPlus. This analysis involves two levels of analysis: intra-subject (Level 1, which analyzes the change of subjects over time) and inter-subject (Level 2, through which the effect of treatment is examined). Repeated measures of symptoms are nested in each subject, and the analysis allows estimating a change parameter along repeated measurements. The analysis tests whether the change parameter varies according to the treatment group, and other covariates. HLM allows to consider in the analyses all participants (including those who abandoned the study). Therefore, these analyses will be conducted according to the principle of intention to treat. In order to examine whether there are differences between the groups in the remission of the diagnosis of depression, the investigators will perform two Cochran-Mantel-Haenszel tests (one for the results obtained after the intervention and another for the results obtained after the follow-up). Similarly, in order to examine whether there is a response to treatment (reduction of at least 50% in the severity of depressive symptoms; Dimidjian et al., 2006), two other Cochran-Mantel-Haenszel testse will be carried out. Finally, in order to examine whether there are any differences between the groups in the percentage of women responding to treatment with regard to the intensity of pain, two more Cochran-Mantel-Haenszel tests will be performed. Treatment response will be considered if there are reductions of at least two units on the pain scale from 0 to 10, as this is the criteria usually employed for clinically significant pain improvement (Michener, Snyder, \& Leggin, 2011). Finally, in order to test whether the decreases found after the follow-up in the pain-related variables are mediated by the decrease in depressive symptoms severity, regression analysis with bootstrapping will be conducted (Preacher \& Hayes, 2008).