Robert Wood Johnson Medical School, University of Medicine and Dentistry of New Jersey, New Brunswick
University of Medicine and Dentistry of New Jersey, Robert Wood Johnson Medical School, Division of Rheumatology and Connective Tissue Research, 1 Robert Wood Johnson Place, MEB-484, New Brunswick, NJ 80903-0019
Affective balance, relative levels of negative affect (NA) and positive affect (PA), better describes emotional functioning than NA or PA alone. Affect balance styles and their relationship to clinical outcomes were compared between patients with fibromyalgia (FM) and controls.
FM patients (n = 79) were compared with patients with other medical conditions (controls; n = 92). Patients underwent a physical examination, completed questionnaires, and were screened for clinical disorders such as depression, with diagnoses confirmed by structured interview. Affect balance style categories were calculated as follows: healthy (high PA/low NA), low (low PA/low NA), reactive (high PA/high NA), and depressive (low PA/high NA).
Compared with controls, FM patients had lower levels of PA (P = 0.0031; P values are adjusted for multiple testing), higher levels of NA (P = 0.0061), lower levels of functioning (P < 0.0001), and more clinical disorders (P = 0.0031). Groups differed regarding affect balance style (P = 0.0061), with FM patients being more likely than controls to be categorized as depressive (odds ratio 5.60) and reactive (odds ratio 3.81). FM patients and controls with reactive and depressive affect balance styles reported poorer functioning (P < 0.0001) compared with patients with healthy affect balance style. Finally, there was an association between affect balance style and psychiatric comorbidity (P < 0.0001), with patients with depressive and reactive affect balance styles having a 9.00 and 4.75 odds ratio, respectively, of having psychiatric comorbidity compared with patients with healthy affect balance style.
Depressive (low PA, high NA) and reactive (high PA, high NA) affect balance styles were predominant in FM patients and related to poor functioning and psychiatric comorbidity.
Fibromyalgia (FM) is a rheumatologic condition characterized by widespread musculoskeletal pain and multiple tender points (1). Persons with FM also frequently present with muscle stiffness, fatigue, sleep disturbance, paresthesia, cognitive impairment (1, 2), and depression (3–5). It is not clear whether depression is the result of chronic symptoms and subsequent impaired social role functioning, e.g., inability to work, or whether FM and depression potentially share a similar genetic basis (6–9), or some combination of the 2. In contrast, there is little debate regarding the relationship between depression and/or the observable expression of negative emotion, or negative affect, and poor outcomes in patients with FM including greater pain intensity (5, 10–13) and fatigue (14, 15), as well as poor functioning (16) and quality of life (17).
Recently, Zautra and colleagues examined the role of both negative affect (NA) and positive affect (PA) in patients with FM. Compared with patients with osteoarthritis, those with FM experienced significantly higher levels of fatigue (15), pain, and NA (13). It was also demonstrated that increases in NA (e.g., tension, nervousness, irritability) were often related to pain episodes (18), higher levels of pain (13), and fatigue (15) in patients with FM. Conversely, increased PA (e.g., feeling strong, enthusiastic, determined) was related to lower levels of pain (13) and fatigue (15). Interestingly, PA appeared to weaken the pain–NA relationship (18), yet PA was found to be lower in patients with FM than in other pain patients (13, 19). Zautra et al hypothesized that dysfunctional PA regulation, a lack of ability to sustain PA during times of increased pain or stress, is a key feature of FM (19).
Compelling literature links NA and medical outcomes, but the literature describing the relationships between PA and clinical outcomes in any illness is sparse. Notwithstanding, Pressman and Cohen (20) reviewed the existing studies examining PA and self-reported health outcomes and concluded that individuals with high NA report more symptoms than would be expected from the underlying disease, whereas those with high PA report fewer and less severe symptoms. Still, there remains confusion and criticism regarding the affect and medical outcomes literature because many studies fail to distinguish between positive and negative emotions, making it difficult to draw firm conclusions. Furthermore, little attention is paid to affective balance, relative levels of NA and PA, which may be a more informative way to understand the relationships between affect and physical and psychological functioning.
There are existing measures for the assessment of affect balance, but their accuracy and utility are limited. For example, the Bradburn Affect Balance Scale (21) has been criticized for its outdated language, vague items, and simplistic method for calculating affect balance (NA minus PA = affect balance) resulting in the same score for excessive affective activation (high NA/high PA) and minimal affective activation (low NA/low PA) (22, 23). Another instrument, the Balanced States of Mind Model (24), utilizes a slightly more complex formulation, ratio of PA to total affect (PA/[PA + NA]), but again this mathematical formula does not differentiate between patients with excessive affective activation and those with minimal affective activation. This distinction is important because activated affect and nonactivated affect are linked to level of physiologic arousal, which is one pathway through which affect may relate to physical health (20).
We suggest that affect balance style can be defined by 4 distinct patterns. Individuals with high PA and low NA are classified as having a healthy affect balance style because the literature supports the benefits of high PA and low NA (25–27). Those with both low PA and low NA are classified as having a low affect balance style because they likely do not report strong emotions in either direction or have strong physiologic arousal reactions. In contrast, individuals with more intense affective responses, high PA and high NA, are classified as having a reactive affect balance style and may experience high levels of physiologic arousal and stronger emotions, both positive and negative. Last, individuals with low PA and high NA are classified as having a depressive affect balance style, as they generally have few positive emotions and can be prone to the negative thoughts and feelings common among depressed individuals.
We hypothesized that patients with FM differed from control patients by having lower levels of PA, higher levels of NA, worse functioning, and greater psychiatric comorbidity. We also hypothesized that patients with FM were more likely than control patients to be categorized as having reactive and depressive affect balance styles and less likely to be categorized as having a healthy affect balance style. Finally, we hypothesized that patients with reactive and depressive affect balance styles would have worse functioning and more psychiatric comorbidity compared with those with a healthy affect balance style.
PATIENTS AND METHODS
Participants were drawn from a cohort of 240 patients evaluated for the Living with Lyme Disease (LLD) study conducted at the University of Medicine and Dentistry of New Jersey–Robert Wood Johnson Medical School (UMDNJ-RWJMS). For the LLD study, all English-speaking patients ages 18–70 years seen at the Lyme Disease Center between September 2002 and August 2006 were invited to participate. This study considered for analysis all patients meeting the American College of Rheumatology criteria for FM (1) upon physical examination by the physician (LHS) (n = 79), plus patients for the control group (n = 92). Control patients either had fully recovered from Lyme disease after antibiotic treatment or met criteria for another readily identifiable medical condition such as rheumatoid arthritis, multiple sclerosis, or osteoarthritis. This study was approved by the Institutional Review Board of UMDNJ-RWJMS. Participants received $10 for participating in the LLD study.
Positive and negative affect.
Affect was measured by the Positive and Negative Affect Scale (PANAS). The PANAS consists of 2 mood scales with 10 items each for the assessment of PA and NA. Scores for each scale range from 0 to 50. The scales have been shown to be internally consistent, uncorrelated, and stable over a 2-month period; good convergent and discriminant validity have also been demonstrated (28).
We used a slightly modified version of the Fibromyalgia Impact Questionnaire (FIQ) to assess functioning. The FIQ is a 19-item self-report instrument that has been found to be reliable and valid for the assessment of general health status in patients with FM (29). Raw scores are adjusted to a mean ± SD of 50 ± 10, with higher scores indicating greater disability. The modified version of the FIQ substitutes the term Lyme disease for fibromyalgia and has been found to be valid for the purposes of this study (30).
Psychiatric comorbidity screening
The Patient Health Questionnaire (PHQ) was used to screen patients for psychiatric disorders. The PHQ consists of 59 items that assess potential psychiatric disorders in 5 domains: depression, anxiety, somatoform, eating, and substance use/abuse (31). The validity of the PHQ is adequate, with good agreement between the PHQ and mental health professionals (for any diagnosis, κ = 0.71; overall accuracy rate 88%) (31).
Psychiatric diagnosis confirmation
Participants who screened positive for ≥1 psychiatric disorder were interviewed using the corresponding module(s) of the Structured Clinical Interview for the Diagnostic and Statistical Manual IV (SCID) (32). In the hands of a trained interviewer, the SCID is a reliable and valid assessment instrument for identifying current and past psychiatric disorders (33, 34).
Consent was obtained from participants before completing questionnaires, interviews, a medical examination, and serologic testing when indicated. In most cases, participants completed questionnaires before seeing the physician (LHS). Blind to results of the questionnaires and interviews, the physician examined, diagnosed, and treated patients no differently than if they were not study participants. Lyme disease was diagnosed using the Centers for Disease Control and Prevention surveillance criteria for Lyme disease (35). Participants with active Lyme disease received antibiotic treatment and were contacted 6 months later for followup. At 6 months, participants reporting complete recovery were included in the control group, whereas those with persistent symptoms were classified as having post-Lyme disease syndrome (PLDS) and were included in this study only if they also qualified for a diagnosis of FM. In the absence of symptoms explainable by a past or current infection with Borrelia burgdorferi, alternate explanations were sought. In many cases, patients met criteria for FM or another readily identifiable medical condition.
Statistical analyses were conducted using SPSS 14.0 software (36) and the R Statistical Environment (37). Fisher's exact test and t-tests were used to compare sample characteristics. One-way analysis of variance (ANOVA) procedures were used to compare FM and control patients' levels of PA, NA, and functioning using age, educational level, sex, and marital status as covariates. Using the same covariates, logistic regression and multinomial regression were used to determine the presence of psychiatric comorbidity (yes/no) and endorsement of affect balance style (healthy, low, reactive, and depressive) with likelihood ratio tests of nested models. Confidence intervals of odds ratios were determined using profile likelihoods. One-way ANOVAs and Tukey's honestly significant difference procedure were used to compare levels of functioning among affect balance style categories. Adjustment of P values for multiple testing was done separately for the sample characteristics and other results using Holm's method (38). Post hoc analysis results were not adjusted for multiple comparisons.
A total of 79 patients with FM comprised our experimental group; of these patients, 7 had PLDS and had Lyme disease prior to FM symptoms. A total of 92 patients were included in our control group; their medical diagnoses are presented in Table 1 and demographic characteristics of all participants are presented in Table 2. This heterogeneous group of medical patients allowed us to control for pain, fatigue, and the effects of having symptoms and seeking medical care. Participants (mean ± SD age 43 ± 13 years) were primarily white (87%) and female (74%). Most had some college education (mean ± SD 15 ± 2 years), were employed at least part time (58%), and were married (63%). Compared with controls, FM patients were slightly younger (P = 0.021, adjusted P = 0.094), were less likely to be married (P = 0.0187, adjusted P = 0.094), and were more likely to be female (P < 0.0001, adjusted P = 0.0006), less educated (P = 0.0057, adjusted P = 0.034), and have a lower income (P = 0.019, adjusted P = 0.094).
Values are the number (percentage) unless otherwise indicated. FM = fibromyalgia.
Age, mean ± SD years
43.1 ± 13.4
45.2 ± 13.7
40.6 ± 12.6
Education, mean ± SD years
15.2 ± 2.3
15.7 ± 2.4
14.7 ± 2.1
Affect balance style
Any axis I
Significantly more FM patients (53.2%) had at least 1 axis I clinical disorder compared with controls (21.7%; P = 0.0008, adjusted P = 0.0031). In FM patients, depression and anxiety were common, occurring in 25.3% and 29.1% of patients, respectively. Among controls, only 6.5% met the criteria for depression and 14.1% for anxiety. Somatoform disorders were identified in 15.2% of FM patients compared with only 2.2% of controls (Table 2).
Affect balance and outcomes.
Affect balance style categories were calculated using cut points based on the population means, so that PA >35 was classified as high and NA >18.1 was classified as high. Calculations resulted in the following distribution of patients (both FM and control) within affect balance style categories: 40 classified as healthy (high PA, low NA), 31 classified as low (low PA, low NA), 28 classified as reactive (high PA, high NA), and 72 classified as depressive (low PA, high NA). Table 3 displays the mean PA and NA scores for each affect balance style, as well as each group (FM versus control), and the distribution of patients by group to affect balance style categories. As hypothesized, after controlling for age, sex, marital status, and years of education, there was a highly significant difference between groups for affect balance style categories (P = 0.0032, adjusted P = 0.0061).
Table 3. Mean ± SD PANAS and FIQ scores by affect balance style (ABS) and group*
PANAS = Positive and Negative Affect Scale; FIQ = Fibromyalgia Impact Questionnaire.
General population norm
35.0 ± 6.4
18.1 ± 5.9
Healthy (n = 40)
39.3 ± 3.2
13.7 ± 2.3
32.3 ± 18.1
Low (n = 31)
28.7 ± 4.9
14.5 ± 2.7
41.8 ± 16.2
Reactive (n = 28)
39.6 ± 4.4
25.8 ± 7.2
52.0 ± 19.5
Depressive (n = 72)
25.5 ± 5.5
26.1 ± 5.2
56.2 ± 16.8
Fibromyalgia (n = 79)
29.1 ± 8.4
23.2 ± 7.4
56.5 ± 16.5
Control (n = 92)
33.8 ± 6.9
19.2 ± 7.3
39.4 ± 19.2
The odds ratio (OR) was 5.60 (95% confidence interval [95% CI] 2.12–14.77) for depressive affect balance style as opposed to healthy style, 3.81 (95% CI 1.20–12.04) for reactive affect balance style as opposed to healthy, and 3.91 (95% CI 1.23–12.46) for low affect balance style as opposed to healthy for FM patients relative to controls. More specifically, in the FM group the distribution of affect balance style was healthy (n = 8), low (n = 14), reactive (n = 14), and depressive (n = 43), whereas for the control group the distribution was healthy (n = 32), low (n = 17), reactive (n = 14), and depressive (n = 29). Raw and adjusted affect balance style proportions by FM group versus control group are shown in Figure 1.
One-way ANOVAs, controlling for age, marital status, sex, and education, revealed that when compared with controls, FM patients reported significantly lower levels of PA (P = 0.0010, adjusted P = 0.0031), higher levels of NA (P = 0.0031, adjusted P = 0.0061), and worse functioning (P < 0.0001, adjusted P < 0.0001) (see Table 3 for means). Furthermore, one-way ANOVAs and Tukey's tests based on data for all participants revealed a significant main effect for affect balance style category (P < 0.0001, adjusted P < 0.0001) where patients with depressive affect balance style reported significantly lower levels of functioning than those with healthy and low affect balance styles (see Table 3 for mean FIQ scores). Furthermore, patients with healthy affect balance style reported significantly higher levels of functioning than those with a reactive affect balance style. Last, there was an association between endorsement of affect balance style category and psychiatric comorbidity (P < 0.0001, adjusted P < 0.0001). Compared with patients with a healthy affect balance style, those with a depressive affect balance style had an OR of 9.00 (95% CI 3.27–29.58) for psychiatric comorbidity, and those with a reactive affect balance style had an OR of 4.75 (95% CI 1.43–17.79) (Figure 2).
Post hoc analyses were conducted to determine the types of psychiatric comorbidity that were more common among FM patients. The relationship between reactive and depressive affect balance styles and the various types of psychiatric comorbidity was also examined. Results indicated that FM patients were more likely than control patients to meet criteria for current depression (OR 4.19, 95% CI 1.55–12.83, P = 0.0040) and somatoform disorder (OR 4.59, 95% CI 1.12–31.30, P = 0.033). Affect balance style appeared to affect the risk of current depression (P = 0.0003), somatoform disorder (P = 0.018), pain disorder (P = 0.0034), anxiety disorders (P = 0.0017), and generalized anxiety disorder (P = 0.0021). Compared with the healthy affect balance style, the depressive affect balance style resulted in an OR of 15.13 (95% CI 2.81–283.35) for current depression, 10.26 (95% CI 1.82–193.21) for somatoform disorder (comparison group of healthy or low affect balance style), 11.44 (95% CI 1.94–221.21) for pain disorder (comparison group of healthy or low affect balance style), 4.93 for anxiety disorders (95% CI 1.50–22.43), and 12.53 (95% CI 2.35–233.42) for generalized anxiety disorder. Similarly, the reactive affect balance style resulted in an OR of 22.33 (95% CI 3.01–472.39) for pain disorder (comparison group of healthy or low affect balance style), 5.75 (95% CI 1.45–29.23) for anxiety disorders, and 13.87 (95% CI 2.16–273.63) for generalized anxiety disorder. Furthermore, among those with a depressive affect balance style, FM patients were more likely than controls to screen positive for psychiatric comorbidity (P = 0.026), with 71% of FM patients with depressive affect balance style screening positive for some axis I disorder compared with only 38% of control patients. For example, 38% of FM patients with depressive affect balance style screened positive for current depression compared with only 10% of control patients with depressive affect balance style (P = 0.026). However, not all patients with current depression had a depressive affect balance style: 35% of FM patients and 66% of controls with depression had another affect balance style. The sample size in this study was sufficient for 80% power to detect an OR of 4.2 for depressive versus healthy affect balance style with respect to FM or control status, or 75% of the effect size actually observed.
The objective of this study was to examine a potential role for affect balance in the physical functioning and mental health of patients with FM. As hypothesized, patients with FM demonstrated lower levels of functioning, higher rates of psychiatric comorbidity, higher levels of NA, lower levels of PA, and more disturbances in affective balance compared with our medical control group. Depressive (low PA, high NA) and reactive (high PA, high NA) affect balance styles were predominant in patients with FM and related to functional impairment and the presence of psychiatric comorbidity.
The tendency of FM patients to report lower levels of functioning and higher rates of psychiatric comorbidity than other medical populations is consistent with most studies addressing these questions (3, 5, 39, 40), as is the tendency for FM patients to report higher levels of NA (13, 15) and lower levels of PA (13, 15, 19). In contrast, our control group, consisting of many patients reporting pain, demonstrated levels of PA and NA more consistent with those found in a healthy population. Changes in NA in relation to pain episodes are not always consistent between and within individuals and groups (18), e.g., not all patients with pain experience high levels of affective problems (27). For some patients, even though there may be an increase in pain, greater PA may blunt any related increase in NA. In addition, by looking at the relative balance between NA and PA, we found that a more complete understanding of the influences of affect on both physical and psychological outcomes can be achieved.
As predicted, FM patients were more likely than patients with other medical conditions to have a depressive or reactive affect balance style. Fifty-four percent of our FM sample had a depressive affect balance style compared with only 32% of controls. What is more striking is that 40% of FM patients with a depressive affect balance style qualified for a diagnosis of major depressive disorder (MDD) compared with only 10% of controls with a depressive affect balance style. This could be explained in part by a potential genetic predisposition to MDD in FM patients rendering them more vulnerable (6–9). For both groups, having a depressive affect balance style was associated with higher rates of somatoform disorder compared with the other 3 affect balance styles. Also, patients with a depressive affect balance style in our sample reported significantly lower levels of functioning when compared with those with low and healthy affect balance styles. This is consistent with other research reporting the relationship between depression and higher levels of disability in patients with FM (16, 41). Interestingly, we also found that having a low affect balance style appears to decrease the risk of having a pain or somatization disorder. These individuals could be thought of as “mellow,” experiencing few emotional lows or highs.
One of the pathways linking affect to health may be physiologic reactivity (20). Crawford and Henry (42) noted that the affective states measured by the PANAS are more aptly named positive activation (e.g., excited, enthusiastic) and negative activation (e.g., irritable, jittery). We hypothesized that patients with depressive and reactive affect balance styles, whether these styles preceded or were the consequences of living with chronic pain, would experience lower levels of functioning in part because they likely experience greater and more persistent physiologic activation of the stress response systems, resulting in dysfunction of these systems and other physiologic consequences (43). Depression has been associated with abnormalities of the hypothalamic–pituitary–adrenal axis (HPA axis) (44) and autonomic nervous system (ANS) (45). Similarly, investigations into the pathophysiology of FM addressing findings of abnormalities of the HPA axis, ANS, and pain processing pathways have been well reviewed (46).
The predominant theory for the pathogenesis of FM involves pain pathway dysregulation resulting in central sensitization (47, 48). The origin of the pain processing aberration remains unknown, although reports of sympathetic hyperactivity in patients with FM (46, 49, 50) suggest that FM could be a sympathetically maintained pain syndrome. Another potential explanation is based on the observation that mediators of the stress response system (e.g., cortisol, catecholamines, interleukin-6) are often altered in FM. These stress mediators have been found to modulate N-methyl-D-aspartate receptor physiology thereby increasing sensitization to pain (51). These are intriguing hypotheses because they relate to linking affect balance style to FM, but physiologic processes were not studied herein. Future studies should explore the relationship between a particular affect balance style and physiologic correlates including levels of cortisol and catecholamines or heart rate variability, to name a few potential targets.
Although we found that FM patients from our sample had overall higher levels of NA and lower levels of PA than controls, close to 28% of FM patients had either a healthy or low affective balance style, suggesting the presence of an emotionally resilient subgroup. The healthy style was related to less functional disability and psychiatric comorbidity than the depressive and reactive styles, but was not significantly less than the low affect balance style. Also, the low affect balance style was associated with a decreased risk of pain or somatization disorder diagnoses. More resilient subgroups of patients with FM have been previously described (52, 53). For example, Giesecke et al (52) identified a cluster of patients with FM with extreme tenderness on evoked pain testing who did not react negatively as exhibited by normal mood, low levels of catastrophizing, and high perceived control over pain. It is important to consider and better understand affectively healthy patients with FM with regard to conceptualization of FM and the provision of appropriate treatment.
There are several limitations to this study. First, the FM patients evaluated herein were drawn from patients presenting to a Lyme disease specialty center; thus the belief that one might have Lyme disease could distinguish these patients from others who never had this belief. Some of these patients readily accepted the diagnosis of FM, while others eschewed the diagnosis, certain that chronic Lyme disease was the only explanation for their symptoms. Because patients who cling to the chronic Lyme disease diagnosis may differ from those who do not, future studies should assess patients with FM in more typical settings. Second, affect balance style assessed in the manner we described is a new approach that will benefit from future studies focusing on validation. Also, this study relied on self-report measures, and although well-validated instruments were chosen, objective reports could provide supportive information. In a separate study, we demonstrated that PA and NA can be reliably observed by others (54). Third, patients were categorized as having 1 of 4 affect balance styles without consideration for scores that fell within the standard deviation. Therefore, future studies using larger samples might consider analyzing a neutral affect balance style. Finally, due to the cross-sectional design of this study, it is inappropriate to infer causality. Although we hypothesize that affect balance styles contribute to pain and other FM symptoms, it is equally possible that affective dysregulation is the result of living with a chronic painful condition. Future studies using a longitudinal design would best identify causal factors that can become the targets of intervention.
In conclusion, the clinical implications of considering NA and PA in terms of affective balance styles include evaluating those at greater risk for adverse outcomes and formulating targeted treatment strategies. For example, we found that depressive (low PA, high NA) and reactive (high PA, high NA) affect balance styles were predominant in patients with FM and related to poor functioning and psychiatric comorbidity. We have also reported elsewhere that depressive and reactive affect balance styles were associated with higher levels of catastrophizing (55). Thus a multidisciplinary approach that addresses mood and anxiety disorders and promotes improved coping and functioning could be particularly effective. FM is one of a number of medical conditions where outcomes are closely tied to psychological factors. Therefore, interventions targeting increasing PA and decreasing NA might prove to be particularly helpful for patients with FM.
Dr. Hassett 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 design. Hassett, Sigal.
Acquisition of data. Hassett, Radvanski.
Analysis and interpretation of data. Hassett, Simonelli, Radvanski, Buyske, Savage, Sigal.