Psychological functioning of people living with chronic pain: A meta-analytic review
Abstract
Objectives
Chronic pain (CP; >3 months) is a common condition that is associated with significant psychological problems. Many people with CP do not fit into discrete diagnostic categories, limiting the applicability of research that is specific to a particular pain diagnosis. This meta-analysis synthesized the large extant literature from a general CP, rather than diagnosis-specific, perspective to systematically identify and compare the psychological problems most commonly associated with CP.
Methods
Four databases were searched from inception to December 2013 (PsychINFO, The Cochrane Library, Scopus, and PubMed) for studies comparing the psychological functioning of adults with CP to healthy controls. Data from 110 studies were meta-analysed and Cohen's d effect sizes calculated.
Results
The CP group reported experiencing significant problems in a range of psychological domains (depression, anxiety, somatization, anger/hostility, self-efficacy, self-esteem and general emotional functioning), with the largest effects observed for pain anxiety/concern and somatization; followed by anxiety and self-efficacy; and then depression, anger/hostility, self-esteem and general emotional functioning.
Conclusions
This study demonstrates, for the first time, that individuals with CP are more likely to experience physically focussed psychological problems than other psychological problems and that, unlike self-efficacy, fear of pain is intrinsically tied to the CP experience. This challenges the prevailing view that, for individuals with CP, problems with depression are either equal to, or greater than, problems with anxiety, thereby providing important information to guide therapeutic targets.
Practitioner points
Positive clinical implications
- This is the first time that the CP literature has been synthesized from a general perspective to examine psychological functioning in the presence of CP and provide practical recommendations for assessment and therapy.
- Individuals with CP were most likely to experience psychological problems in physically focussed areas – namely pain anxiety/concern and somatization.
- Although fear of pain was intrinsically tied to the CP experience, self-efficacy was not.
- CP was more strongly associated with anxiety than with depression.
Limitations
- The study focuses on the general CP literature, adults and research-utilizing self-report measures.
- Meta-analyses are limited by the empirical literature on which they are based.
Background
Approximately 20% of the world's population experiences persistent or chronic pain (CP; Goldberg & McGee, 2011) – Pain that occurs on most days for 3 or more months (IASP, 1994). With low back pain now the largest contributor to global disability (Hoy et al., 2014), CP has been shown to negatively impact on many areas of life, including work attendance and productivity (Stewart et al., 2010), physical function and quality of life (Douglas, Graham, Anderson, & Rogerson, 2004), engagement in social or recreational activities (Haythornthwaite & Benrud-Larson, 2000), and medication and health service usage (Blyth, March, Brnabic, & Cousins, 2004; Douglas et al., 2004).
The subjective experience of pain involves a complex interplay between physical, psychological, and environmental variables (Flor & Hermann, 2004; Nicholas, 2008). Consequently, many studies have examined the psychological aspects of CP, with research consistently demonstrating that psychological problems are common in people with this condition (Dersh, Polatin, & Gatchel, 2002). Although American data indicate that CP is associated with comparable rates of anxiety and depression (Von Korff et al., 2005), Australian data suggest that depression and adjustment disorders are more common (Access Economics, 2007).
Numerous aspects of psychological functioning may impact on, and be affected by, an individual's experience of CP including, but not limited to, mood (e.g., depression, anxiety, stress), feelings of self-mastery (e.g., self-efficacy, self-esteem), attributions about responsibility (e.g., guilt, shame), and grief/loss. Although these have been explored to varying degrees in the CP literature, most research has focussed on depression and anxiety. Similarly, research has tended to focus on specific diagnoses, such as neuropathic pain (Attal, Lanteri-Minet, Laurent, Fermanian, & Bouhassira, 2011) and fibromyalgia (Homann et al., 2012), or on specific subgroups, such as older people (Falsarella et al., 2012) and trauma survivors (Peterlin et al., 2009). This focus on specific diagnoses/subgroups is limiting because many people with CP do not have a clear diagnosis or precipitant for their pain. Indeed, one survey found that 65% of people had no clear medical diagnosis for their CP and 33% identified no clear precipitant (Blyth, March, & Cousins, 2003). Moreover, CP is associated with a large number of different medical diagnoses, not all of which have been well-researched, and many people with CP experience a range of medical and psychological comorbidities that are not specific to a particular CP diagnosis or subgroup.
Thus, while it is undeniably important to understand CP from these specific perspectives, especially where CP is known to be associated with syndrome-specific sequelae (e.g., cancer: Urbaniec, Collins, Denson, & Whitford, 2011; and spinal cord injury: North, 1999), a more global understanding of the psychological problems that are associated with CP is also needed, one that includes the majority of people with CP and better meets the needs of clinicians working with this group. Therefore, this study sought to meta-analyse all quantitative research that used standardized assessment tools to examine the psychological functioning of adults with CP (including specific and non-specific CP diagnoses), relative to their healthy peers. We sought to synthesize this very large literature in order to provide a better clinical understanding of the CP experience and the underlying evidence base. Self-report measures of emotional functioning were targeted because they are the most frequently used method of assessment in the literature and are also commonly used in clinical practice.
Method
Literature search
Information sources
Four databases (PsychInfo, Scopus, PubMed, and The Cochrane Library) were searched for studies that examined the psychological functioning of individuals with CP, published prior to 2014. The searches included singular and plural forms of each term and regional variations in spelling (e.g., behaviour/behaviour; see Supporting information, Table S1a, for search terms). The initial search was very broad to ensure that all relevant papers were identified because CP is an umbrella term that encompasses a variety of labels, is attributed to a range of conditions and has been examined by a number of disciplines.
Eligibility criteria
Studies were only included in the current meta-analysis if they met the following criteria: they recruited a CP sample that (1) was aged ≥16 years (commonly used age for adult pain services) and (2) experienced CP – specifically defined as pain on most days for a period of ≥3 months; and the study (3) included a healthy control group that was matched to the CP group, (4) investigated the psychological functioning of both groups utilizing standardized self-report assessment tools, (5) was published in English and documented original quantitative research (excludes reviews), (6) was not a case study (n > 1), and (7) provided data in a format that permitted the calculation of Cohen's d effect sizes (e.g., mean and SD, t-statistic, or exact p-value).
Medical and psychiatric conditions that have syndrome-specific sequelae, and therefore require separate consideration, were excluded because the aim was to examine the general CP experience. Specifically, excluded conditions were as follows: (1) spinal cord injuries, (2) particular medical disorders/conditions (e.g., cancer, cardiac, renal), (3) neurological disorders (e.g., stroke, traumatic brain injury), (4) terminal/palliative conditions, (5) psychiatric conditions (e.g., factitious disorder, psychosis), and (6) personality disorders. Studies examining acute pain onset and/or tolerance were similarly excluded, because the current study was designed to examine people living with CP.
Study appraisal and selection
Critical appraisal and eligibility assessment were performed by the primary author (ALJB). If there was any ambiguity, papers were independently appraised by the full panel of authors and eligibility determined by consensus, following group discussion. If it was not possible to definitively determine eligibility based on the available information, clarification was sought from the corresponding authors. If no response was obtained, that study/variable was excluded from analysis. Where data for the same sample were reported in multiple papers, the paper with the largest sample was included. Matching of the CP and healthy control groups was a key factor in the critical appraisal of studies as it served to minimize extraneous between-group differences, thereby increasing confidence in the validity of the results. CP and healthy control groups were deemed to be matched if there was either a deliberate attempt to match groups on at least one demographic variable (e.g., age, gender) or post-hoc analyses indicated that groups were demographically comparable. Studies that did not meet either of these criteria for matching were excluded from the meta-analysis. Similarly, all studies were required to use standardized assessment tools to ensure only good-quality studies were included. The scope and size of the current study precluded more detailed qualitative evaluation and rating of individual studies. However, both the application of strictly defined eligibility criteria and the weighting of data by the inverse variance (Lipsey & Wilson, 2001) served to minimize the potential impact of poor-quality empirical studies.
Data extraction and organization of studies
Data relating to the study (e.g., sample size, country), participants (e.g., age, gender, race), pain (e.g., diagnosis/location, duration), and the measures of psychological functioning (test, scoring method and direction, means and SDs, or exact t- or p-statistics) were extracted from each study via a detailed data extraction form. As different aspects (e.g., cognitive, emotional, somatic) of psychological functioning have been assessed by a large number of alternative measures, the most meaningful way to explore the data was to group measures on the basis of psychological domains (e.g., depression). Different measures varied in their focus, assessing either positive or negative mood states; consequently, scores were rescaled where necessary (sign inverted but scores unchanged) so that they could be consistently interpreted. For positive domains (self-esteem and self-efficacy), higher scores indicated better outcomes. For all other domains, higher scores indicated greater psychological distress (poorer outcomes). In all cases, positive effects indicated greater levels of the domain in the control group and negative effects indicated greater levels in the CP group. Thus, for depression, a positive effect indicates greater depression in the control group, whereas a negative effect indicates the CP group is more depressed.
Data analysis
Data were analysed using the ‘Comprehensive Meta Analysis’ programme (Borenstein, Hedges, Higgins, & Rothstein, 2005). Where a study recruited multiple CP and/or control groups and reported data separately, scores were pooled to provide an overall score for each group (CP, controls). If studies provided multiple scores for a single domain (e.g., more than 1 depression score), a mean effect was calculated, thereby ensuring that each study only contributed a single score to the overall mean when the effect sizes from different studies were averaged (Lipsey & Wilson, 2001). Information relating to pain-related litigation and socio-economic status was collected to evaluate whether these variables acted as moderator variables; however, there were insufficient data to analyse these relationships. Similarly, although active treatment seeking may have reflected an important difference between the CP groups (treatment seekers may have had more severe conditions and/or comorbidities, and more psychological problems), this could not be examined because it was frequently unclear whether samples recruited from non-treatment settings were receiving care. Thus, any analysis of recruitment source was unlikely to be reliable or informative.
The effect size used in the current analysis was Cohen's d (Cohen, 1988); providing a measure of the standardized difference between the means for the CP and control groups. Effect sizes were interpreted using Cohen's (Cohen, 1988) guidelines, with d = 0.2, 0.5 and 0.8 indicating small, medium, and large effects, respectively. Consistent with Hopkins and colleagues (Hopkins, Marshall, Batterham, & Hanin, 2009), d = 2.0 and 4.0 were labelled very large and extremely large effects.
Heterogeneity analyses, using the I2 statistic, were performed to assess whether differences in the effect sizes from individual studies reflected chance-based sampling differences (homogeneous effects) or additional sources of variance, possibly reflecting methodological differences between studies (heterogeneous effects). As might be expected, given CP was examined from a general, rather than diagnosis-specific, perspective, many of the heterogeneity analyses (I2) were significant (I2 > 50), indicating substantial variability in the findings. Therefore, a more conservative random-effects model was used instead of the traditional fixed-effects model (Higgins, Thompson, Deeks, & Altman, 2003). The latter assumes a ‘true’ effect that is common across all studies, while a random-effects model assumes that there are differences between studies (e.g., samples and/or diagnoses). In addition, when calculating mean effects, individual effects were weighted by the inverse of their variance in order take into account differences in the precision of the effect size estimates obtained from different studies (Lipsey & Wilson, 2001).
Ninety-five per cent confidence intervals (95% CIs) were calculated to test whether the population mean group difference differed significantly from zero, indicated by a CI that does not span 0. Finally, fail-safe N's (Nfs) were calculated to examine the impact of publication bias on the results obtained: Nfs indicating the number of unpublished studies with non-significant findings that would be required to render the current results insignificant (Zakzanis, 2001). Calculations were based on the formula outlined by Lipsey and Wilson (2001), using a 0.2 (small effect; Orwin, 1983). A Nfs statistic was considered adequate if it was larger than the number of studies meta-analysed for a given measure.
Results
After removal of duplicate records, the initial search identified 11,211 records, for which the title and abstracts were assessed against the inclusion criteria by the first author (ALJB). This initial review excluded 10,525 articles because they failed to meet the specified criteria (see Figure 1). Full-text versions of the remaining 686 articles were then sourced and the contents systematically evaluated against the inclusion/exclusion criteria. Following thorough review, 110 papers were retained and their data extracted and meta-analysed.
Study characteristics
All studies were published between 1986 and 2013, with most published in the past 13 years (n = 86, 78%). The final sample comprised 67,554 participants (CP + controls), aged 17–102 years, most of whom were women (n = 30,981, 57%). As seen in Table 1, few studies provided information relating to relationship and employment status, education, ethnicity, BMI and surprisingly, pain duration (see Table S1b, for demographic/background details for individual studies). Some studies provided data in a format that did not permit between-study comparisons (e.g., categorical data for age). Where reported, the majority of participants were Caucasian, married/partnered, had some form of employment and had completed secondary schooling. The CP group had experienced pain for one to 15 years (M = 8.0 years, SD = 3.7). Most studies originated from Europe or America, explored mixed pain conditions, examined deliberately matched samples and recruited their CP group from treatment settings and controls from the general community (see Table 1).
| Chronic pain | Healthy controls | |||
|---|---|---|---|---|
| N (%) | N studies | N (%) | N studies | |
| Participant information | ||||
| N participants | 25,084 (37) | 110 | 42,470 (63) | 110 |
| Age * | 45.6 (11.0) | 94 | 44.4 (11.1) | 92 |
| Gender | ||||
| Male | 6,901 (36) | 96 | 16,375 (47) | 96 |
| Female | 12,318 (64) | 18,663 (53) | ||
| Relationship status | ||||
| Married/partnered | 6,514 (63) | 28 | 5,809 (62) | 28 |
| Not married/partnered | 3,775 (37) | 3,587 (38) | ||
| Employment status | ||||
| Employed | 1,372 (64) | 14 | 1,845 (71) | 14 |
| Not employed | 780 (36) | 753 (29) | ||
| Years of education* | 12.7 (3.3) | 14 | 13.7 (3.4) | 14 |
| Ethnicity | ||||
| White | 7,220 (76) | 20 | 6,928 (69.7) | 20 |
| Not White | 2,339 (24) | 3,010 (30.3) | ||
| Body Mass Index (BMI)* | 27.0 (1.3) | 14 | 25.6 (1.3) | 14 |
| Pain duration (years)* | 8.0 (3.7) | 37 | ||
| Study Information | ||||
| Location of origin | ||||
| Europe | 48 (44) | 110 | ||
| America | 46 (42) | |||
| Australia | 9 (8) | |||
| Asia | 6 (5) | |||
| Other | 1 (1) | |||
| Pain conditions explored | ||||
| Mixed | 44 (40) | 110 | ||
| Back | 16 (15) | |||
| Fibromyalgia | 12 (11) | |||
| Pelvic/abdominal | 12 (11) | |||
| Head/neck | 8 (7) | |||
| Arthritis | 7 (6) | |||
| Facial | 7 (6) | |||
| Whiplash | 3 (3) | |||
| Neuropathic | 1 (1) | |||
| Sample matching by study design | ||||
| Deliberately matched | 75 (68) | 110 | ||
| Not deliberately matched | 31 (28) | |||
| Not reported | 4 (4) | |||
| Recruitment source | ||||
| Treatment seeking | 58 (53) | 110 | 3 (3) | 110 |
| Community based | 17 (15) | 55 (50) | ||
| Primary care | 4 (4) | 8 (7) | ||
| Mixed | 20 (18) | 13 (12) | ||
| Students | 3 (3) | 3 (3) | ||
| Not reported | 6 (5) | 21 (19) | ||
| Other | 2 (2) | 7 (6) | ||
Note
- Figures presented are N (%) except where indicated by * to be X (SD).
Psychological function
Although there are many forms of psychological functioning that may be relevant to CP, the areas found to be most commonly examined by researchers were depression and anxiety (general and pain-related) and, to a lesser degree, somatization, anger/hostility, self-efficacy, self-esteem and general emotional functioning (overall mental/emotional health and perceived stress) – thus, subsequent analyses focussed on these areas. The mean effects for each domain are summarized in Figure 2. As can be seen, there were moderate to large, significant and robust differences in the psychological functioning of persons with CP, relative to their healthy peers, in all areas. Moreover, the limited overlap in the CIs indicates that there were significant differences in the extent to which many of these domains were affected.
Depression
Depression was the most commonly assessed area of psychological functioning (Nstudies = 82, see Table S1c and Figure S1a), with the majority of studies (n = 77) using one of 14 measures. Only five studies used multiple measures. The most frequently used measure was the Beck Depression Inventory (BDI; Nstudies = 33, 40%) (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961), followed by the Centre for Epidemiological Studies Depression scale (CES-D; Radloff, 1977), and the Hospital Anxiety and Depression Scale (HADS-D; Zigmond & Snaith, 1983). Although the overall mean effect for depression was moderate (d = −0.46; see Figure 2), there was considerable variation between the effects for different measures. However, with the exception of the Kessler Psychological Distress Scale (K-10; Kessler et al., 2003), which was only used by one study and had an unsatisfactory Nfs statistic, all effects were moderate to very large. Moreover, they were significant and negative, indicating consistently higher levels of depression in the CP group.
Anxiety
Anxiety was examined by a total of 40 studies (see Table S1d and Figure S1b): 36 of which used a single measure of anxiety, while others used two scales. The HADS-A was the most commonly used measure, reported in 12 studies, followed by the State-Trait Anxiety Inventory (STAI-S, Nstudies = 8; Spielberger, Gorsuch, Lushene, Vagg, & Jacobs, 1983). Across all measures, the CP group showed consistently higher levels of anxiety than the controls (negative d), resulting in a large and significant overall mean effect for this construct (d = −0.82; see Figure 2). Significant moderate to very large effects were found for the majority of measures, with the two exceptions having unsatisfactory Nfs statistics, raising concerns about the reliability of those findings.
Pain anxiety/concern
‘Pain anxiety/concern’ comprised measures assessing catastrophizing, fear of pain/movement and pain anxiety. In total, 18 studies investigated various aspects using 15 different measures (see Table S1e and Figure S1c); most commonly the Pain Catastrophising Scale (PCS; Sullivan, Bishop, & Pivik, 1995), with total and/or subscale scores being reported by 10 studies (56%). Twelve studies employed a single measure of pain anxiety/concern, while the remainder used two, three, or six scales (Nstudies = 3, 2, 1, respectively). A large overall mean effect (d = −1.15) was found for this construct (see Figure 2), with most measures returning large to very large and significant effects. Of note, the only measure to yield a small and non-significant effect for pain anxiety/concern – the Fear of Pain Questionnaire (FPQ-III; McNeil & Rainwater, 1998) – was only used in a single small-scale study. Moreover, unlike other measures that examine fears relating to the specific CP experience (e.g., it will make my back pain worse), the FPQ-III explores fear of pain in relation to a range of activities (e.g., gulping a hot drink before it has cooled), none of which are specifically tied to CP.
Somatization
Somatization was assessed in 16 studies using eight measures (see Table S1f and Figure S1d), the most common being the somatization subscale of the Revised Symptom Checklist-90 (SCL-90-R-S; Nstudies = 6; Derogatis, 1994). The overall effect for somatization was large, negative and significant (d = −1.2; see Figure 2), indicating that the CP group consistently reported higher levels of somatization than controls. With the exception of the Wahler Physical Symptom Inventory (WPSI; Wahler, 1968), which had a moderate and significant effect that was susceptible to publication bias, all other measures revealed large and significant effects.
Anger/Hostility
Eight studies explored anger/hostility using four different measures (see Table S1f and Figure S1d). Two studies used multiple measures. While the overall effect for this domain was moderate (d = −0.38; see Figure 2), there was marked variability in the range and significance of findings for individual measures. Interestingly, while the largest effect was found using the Brief Symptom Inventory (BSI-H; Derogatis & Melisaratos, 1983), its longer counterpart (SCL-R-90-H) yielded non-significant results (p = .125). Non-significant results were also found for the State-Trait Anger Expression Inventory (STAXI-AE; Spielberger, 1988; p = .463). Of note, the findings for these latter two measures were susceptible to publication bias (Nfs < Nstudies). Overall, the findings lacked consistency, suggesting that the relationship between anger/hostility and CP is unclear.
Self-efficacy
Self-efficacy was examined by four studies, with one using multiple measures to examine both general and pain-related self-efficacy (see Table S1f and Figure S1d). Not surprisingly, there was a large and significant difference (d = 0.96) between the self-efficacy levels of CP patients and their healthy peers, with the CP group showing substantially less confidence in their ability to bring about change and demonstrate mastery in their lives (see Figure 2). When the specific measures were considered, although not significant, the between-group difference for pain-related self-efficacy, as measured by the Pain Self-Efficacy Questionnaire (PSEQ; Nicholas, 2007), was considerably larger (d = 1.12) than that found for the more general measures of self-efficacy (General Self-Efficacy Scale; GSES, d = 0.64; Schwarzer & Jerusalem, 1995) or life control (Multidimensional Pain Inventory Life Control subscale; MPI-LC, d = 0.80; Kerns, Turk, & Rudy, 1985). Thus, as might be expected, while the pain group consistently reported lower levels of self-efficacy than did controls, they indicated feeling somewhat better able to exert control over their life in general, than they did to exert control over the pain itself and their ability to function in its presence.
Self-esteem
Self-esteem was examined by three studies using one of two measures (see Table S1f and Figure S1d). Overall, a moderate and significant positive effect (d = 0.44) was found (see Figure 2). Although the total sample used to explore this construct was limited (CP: n = 146, controls: n = 162) compared to other domains, these results suggest that healthy controls consistently reported having more positive feelings about themselves and their overall self-worth, than did the CP group.
General emotional functioning
Finally, general emotional functioning (sometimes conceptualized as distress) – a more global construct – was examined by 36 studies (see Table S1g and Figure S1e). Most of the 14 measures of this domain were used by between one and four studies, the exception being the Mental Health subscale of the 36-item Short-Form Health Survey (SF-36-MH; Stewart, Hays, & Ware, 1988), which was used by over 50% of studies. After inversing the effect sizes for specific scales (SF-12, SF-20, SF-36, Profile of Mood States: POMS – McNair, Lorr, & Droppleman, 1992; World Health Organisation Quality of Life Assessment – Brief: WHOQOL-BREF – Hawthorne, Herrman, & Murphy, 2006) so that they all measured impairment in psychological functioning (rather than positive mood), the weighted overall effect for this domain was moderate and significant (d = −0.54), indicating that the CP group consistently reported experiencing greater levels of emotional distress than healthy controls (Figure 2). The effects for individual measures varied greatly (d = −0.04 to d = −3.85), with very low Nfs statistics indicating that the results for four measures were vulnerable to publication bias.
Discussion
Life with CP is a complex experience that cannot be adequately understood in purely physical terms or reduced to neat diagnostic groupings. Therefore, it is important that clinicians, planners and policy-makers understand the psychological aspects of the CP experience from a general perspective because comorbidities, symptom variation, and/or unclear diagnoses confound interpretations of diagnosis-specific CP research. Consequently, this meta-analysis was designed to synthesize the large CP literature from a broad perspective in order to systematically document the psychological functioning of people living with CP.
Our search of the CP literature revealed a substantial amount of research using diverse constructs and measures to examine a range of different psychological domains, particularly in recent years. Compared with controls, the CP group consistently reported experiencing significant and substantial problems in all aspects of psychological functioning. Not surprisingly, the greatest impact was on those domains that are directly tied to the physical experience of pain, namely pain anxiety/concern and somatization. One measure of general emotional functioning (BSI) was also particularly compelling because a finding of this magnitude indicates almost no overlap (<3%) between the scores of CP and healthy individuals on this measure (Zakzanis, 2001).
Although somatization was associated with the largest group difference (effect size), this result is somewhat difficult to interpret because the term is used inconsistently throughout the literature. For instance, in their recent meta-analysis of somatic symptoms, Zijlema et al. (2013) interpret somatization in two quite different ways: the tendency to (1) report/emphasize physical symptoms in the absence of, or to a greater extent than would be expected by the, identified organic pathology; and (2) ‘experience and communicate somatic distress in response to psychosocial stress’ (p. 459). Using the first interpretation, it is not surprising that people with CP showed significantly elevated rates of somatization: CP is not purely a physical experience and commonly lacks clear organic causes. Further, the range of comorbidities often experienced with CP may impact scores on these scales. However, the second interpretation suggests something different: that individuals living with CP are more likely to experience emotional distress in physical ways, possibly due to a heightened tendency to notice (and respond to) physical sensations, especially ones that are directly related to their pain. The current analysis does not distinguish between these alternative interpretations.
Overall, our findings confirm that CP is associated with a range of impairments in psychological functioning. However, they do not support the assertion that depression is the most commonly experienced problem. Instead, we found that, although individuals with CP were consistently more depressed than their healthy peers, they were comparatively more anxious (see Figure 2) – both in general and in response to pain. Similarly, self-efficacy was also found to be broadly affected across general and pain-related areas. However, of note, the elevated levels of pain anxiety/concern found in the CP group were specifically tied to the CP experience – Although the CP group was understandably anxious about exacerbating their condition, they were not more anxious about general pain experiences (e.g., hitting your head).
If an adjustment disorder is defined as a larger-than-expected emotional response (of mixed symptomatology) that impairs a person's ability to cope with a stressful experience or significant life change (World Health Organisation, 1990), it could be argued that many of the domains considered here fall into this broad category. However, it is not possible to comment more definitely here about the frequency of adjustment disorders in CP because researchers have rarely assessed ‘adjustment disorders’, per se. Rather, our current findings suggest that, of the psychological domains that were assessed, physically orientated problems (somatization and pain anxiety/concern) are greater than depression and general impairments in emotional functioning.
It is well documented that the physical symptoms of CP overlap with the symptoms of depression, so much so that an accurate diagnosis of this type of mood disorder can be challenging in a CP setting (Cheatle, 2011; Wong et al., 2011). Indeed, it is possible that this overlap in symptoms made it difficult for individuals to determine the origin of their symptoms when completing the self-report scales. However, this is unlikely to explain why the CP group was comparatively less depressed than they were anxious, for two reasons. First, the measures do not ask respondents to identify the cause of their symptoms – they merely ask them to indicate whether they experience those symptoms – which should have resulted in higher depression scores (symptoms would be reported, irrespective of cause) and narrowed the difference between depression and anxiety. Second, measures that had fewer somatic items (e.g., HADS, Depression Anxiety and Stress Scale [DASS]; Lovibond & Lovibond, 1996) did not yield noticeably smaller effects than scales with higher somatic content (e.g., BDI). This suggests that, contrary to current clinical thinking, the level of somatic content in the measures did not have a major impact on the results.
With that in mind, there are a number of limitations to this study that warrant consideration. First, research has been inconsistent in its terminology and operationalization of various psychological domains (e.g., somatization, anger/hostility), making it difficult to interpret some of the current findings and, indeed, to select appropriate search terms (e.g., disease versus illness). Second, the study size precluded detailed qualitative evaluation of individual studies to exclude sources of potential bias other than publication bias, sample inconsistency and low-quality assessment. Third, because this study focussed on the adult CP population from a general perspective, results may be less applicable to specific groups, especially those with syndrome-specific sequelae. Similarly, this study focussed exclusively on self-report measures. Further research examining specific cohorts (e.g., older people, children, indigenous populations) and other methods of mood assessment (e.g., ICD-10 diagnosis) is now needed. Moreover, we are unable to comment on whether the identified difference predated or resulted from the CP due to the research designs of the original studies. Such information could help improve our understanding of the factors that may pre-dispose and/or protect individuals from transitioning from acute to CP.
In addition, many CP studies that explored treatment outcomes used CP controls, rather than healthy controls, necessitating their exclusion from this meta-analysis. Somewhat surprisingly, an even larger number of studies were excluded because they did not report the basic data required to calculate Cohen's d effect sizes. Standards for data reporting have been under increasing scrutiny over recent years, with a strong push for authors to report more detailed data (including effect sizes) when publishing research (American Psychological Association, 2010; Moore et al., 2010). Adoption of these reporting principles in future research would facilitate more comprehensive meta-analyses.
It was intended that this meta-analysis assist in determining the clinical utility of specific measures, but this did not prove to be the case. With the exception of the measures used to assess anger/hostility – where inconsistent findings suggest that the definition and/or measurement of this domain require more careful consideration and examination – the most commonly used measures consistently discriminated between CP and their healthy peers, suggesting that they were suitable for use in clinical contexts.
Finally, there were insufficient data to examine the impact of other variables (e.g., employment, relationship status, etc.) on psychological functioning. Again, more detailed reporting would enable an examination of these variables.
Conclusions
In summary, CP is a common condition that is associated with a range of psychological problems. This study revealed that those aspects of psychological functioning that are most closely tied to the physical experience, namely pain anxiety/concern and somatization, are most affected in people who are living with CP. This challenges previous assertions that depression levels in this population are equal to, or greater than, anxiety levels (Access Economics, 2007; Von Korff et al., 2005). Not surprisingly, in all areas, the pain-related aspect of the impact was paramount. Although self-efficacy was globally eroded, pain anxiety/concern remained closely tied to the CP experience and did not extend to more general pain events.
Of note, this study supports the earlier findings of McWilliams, Goodwin and Cox (2004) who asserted that anxiety in CP populations requires greater attention because anxiety was often more strongly associated with CP than depression in their sample. Interestingly, the relative strengths of the depression–CP and anxiety–CP relationships have gone largely unchallenged over the last decade.
Overall, these findings suggest that people with CP are in a debilitating bind. The chronic physical pain that they experience is associated with considerable psychological distress, which is most commonly focussed on physical aspects of the overall experience. Although this physical focus is not surprising in the circumstance, it is likely to heighten their level of attention to, and lower their threshold for, physical symptoms. This may, in turn, further increase the chance that they will notice physical symptoms and interpret them as threatening, thus heightening their distress and discomfort, and perpetuating the cycle. This cycle is discussed in detail in the CP literature, but the present study is the first to quantitatively consolidate the existing research findings and, in doing so, enable a direct statistical comparison between different areas of psychological functioning. Although pain anxiety/concern, somatization and self-efficacy (particularly pain-related self-efficacy) are common considerations when working with individuals who experience CP, the degree to which they are prioritized in therapy varies greatly. The current meta-analysis suggests that, to help individuals break the pain cycle, evidence-based practice in CP should prioritize these aspects of psychological function.
Acknowledgements
This work has been partially supported by the Royal Adelaide Hospital Research Committee Allied Health Project Grant awarded to the first author.
Conflict of interest declaration
There are no relationships that may lead to conflict of interest. However, while not representing a conflict of interest, the primary author wishes to disclose the following commercial engagements. The primary author has received honoraria for GP/nursing education sessions and steering committee participation from Mundipharma and Pfizer/Elixir Health and serves as the South Australian State Director of the Australian Pain Society, which receives sponsorship for its Annual Scientific Meeting and Research Scholarship Program from a range of pharmaceutical and medical device companies including Mundipharma, Janssen-Cilag, Pfizer Australia, and Medtronic Australasia.
