Introduction

The present article reports a study that examined the possible relationship between psychological and social factors – anxiety, depression, current stress, quality of life, childhood trauma, as well as pain – and Body Perception Disturbance (BPD) in Complex Regional Pain Syndrome (CRPS). Complex Regional Pain Syndrome is a painful condition affecting the extremities. It is usually caused by physical trauma and characterized by severe pain. The intensity of the pain stands in no relation to the initiating incident. Pain, as well as the concomitant motor, sensory and autonomic disturbances, typically extends distally beyond the original area of the injury and does not follow the innervation areas of nerves (Schwartzman et al. 2009). The rate of CRPS is three to four times higher in women than in men (de Mos et al. 2007) and is associated with negative consequences in terms of psychological and social functioning (Lohnberg and Altmaier 2013; Sohn et al. 2016). The connection between chronic pain and adverse psychological (Dersh et al. 2002; Wilson et al. 2002) and social (Gatchel et al. 2007; Turk et al. 2008) effects is well established.

The hypothesis that psychological and social factors influence the development of CRPS has received mixed support. Stressful life events in the year before CRPS onset have been suspected to play an important role; however, results have not been conclusive (Geertzen et al. 1998; Monti et al. 1998). Beerthuizen et al. (2009) conducted a comprehensive review and found no relationship between chronic CRPS and psychological and social factors. The research group then conducted a large prospective multicenter study (Beerthuizen et al. 2011) and found that depression, anxiety and other psychological factors, as well as stressful life events, did not predict the development of CRPS in patients suffering a physical trauma. Although such a finding needs to be replicated, it suggests that anxiety, depression, and pre-existing stress are not causative factors for CRPS. Rather, these psychological conditions are likely to be sequelae of CRPS. Studies have shown that chronic pain leads to elevated depression and anxiety (Gore et al. 2012), spatial working memory deficits (Kim et al. 2012), and problems in emotional decision making (Apkarian et al. 2004; Gupta et al. 2009). These results corroborate the idea that instead of serving as predisposing psychological risk factors, psychological abnormalities in CRPS seem to be a consequence of chronic pain and disability and the resulting changes in the brain.

Indeed, studies employing structural and functional magnetic resonance imaging (Barad et al. 2014; Hotta et al. 2017; Maihöfner et al. 2004; Pleger et al. 2014; Shokouhi et al. 2017) and magnetoencephalographic recordings (Buntjen et al. 2017) have shown cortical reorganization of the primary somatosensory cortex, altered grey matter structure, abnormal brain system morphology, and changed neural activities in brain areas dealing with sensorimotor functions, pain perception, and emotional experiencing. The prefrontal cortex is thinner in patients suffering from CRPS than in healthy controls (Lee et al. 2015), resulting in impairment of executive functions (Libon et al. 2010) as well as disinhibited pain perception. These results indicate that psychological and cognitive changes in patients with CRPS are due to physiological alterations in the brain (Schwartzman et al. 2009).

This neuroplasticity is particularly crucial to understanding the change in body perception in CRPS. It is estimated that between 50% to 80% of patients with CRPS develop so-called “neglect-like symptoms” (Förderreuther et al. 2004; Galer and Jensen 1999; Lewis et al. 2007), similar to those seen in patients who have had a right-sided stroke (Förderreuther et al. 2004; Frettlöh et al. 2006; Kolb et al. 2012). However, studies have shown that neglect-like symptoms in CRPS are not quite the same as in patients who have had a stroke (Förderreuther et al. 2004; Frettlöh et al. 2006; Kolb et al. 2012). Therefore, the term body perception disturbance (BPD) was introduced (Lewis et al. 2007; Lewis and McCabe 2010). Body perception disturbance encompasses elements of body schema (for example, reduced proprioceptive acuity and mislocalization of tactile stimuli) and body image disturbance (for example, the affected limb is perceived to be larger and not belonging to the body (Moseley et al. 2012; Trojan et al. 2019). Body schema is defined as a “non-conscious system of processes that constantly regulate posture and movement” (Gallagher 1986), and is maintained through tactile, visual and proprioceptive stimuli. The primary (S1) and secondary (S2) somatosensory cortices (Maihöfner et al. 2004; Pleger et al. 2014) as well as the motor cortex (M1) and occipitotemporal cortex (OTC) (Orlov et al. 2010) play an important role in forming and maintaining body schemata.

In turn, body schema influences a person’s body image, defined as “a system of sometimes conscious perceptions, attitudes, and beliefs pertaining to one’s own body” (Gallagher 1986). Body image is easily manipulated, not only in persons with chronic pain conditions but also in people with no pain or other medical conditions (Giummarra et al. 2008; Lackner 1988; Ramachandran 1998). Because personal beliefs, memories, physical changes, as well as psychosocial factors can have modulating effects, a person’s body image is considered unstable. For example, it has been shown that depression and anxiety as well as physical and emotional trauma affect a person’s body image (Fang et al. 2015; Levine et al. 2005; Sabik 2017; Sack et al. 2010; Taylor and Pooley 2017). Bean and colleagues (Bean et al. 2015) found that patients with CRPS who had low levels of anxiety and disability at the time of diagnosis also reported the least pain – and conversely, those with the highest levels of anxiety had the highest amount of pain. This held true for the twelve months following the diagnosis. Poorer outcome after one year was associated with anxiety and pain-related fear. Depression, anxiety and pain intensity are also predictors of disability in CRPS (Bean et al. 2014). Symptoms of a changed body image are prevalent in other chronic pain conditions although not as severe as in CRPS (Frettlöh et al. 2006; Michal et al. 2017). Studies have shown that pain severity and the degree of felt body image distortion are linked (Senkowski and Heinz 2016; Wittkopf et al. 2018). Punt and colleagues (Punt et al. 2013) argue that due to pain and immobility in CRPS, fear-avoidance behavior develops, resulting in suppression of movement. This so-called “learned nonuse” leads to modifications in the cortical regions representing the diseased body part.

Traumatic childhood experiences are another possible factor influencing the development of CRPS and therefore, BPD. An increasing number of studies support the hypothesis that traumatic experiences in childhood, i.e. sexual, physical and emotional abuse, significantly increase the risk of developing psychological and physical diseases in later life, in particular diseases with inflammatory processes (Burke et al. 2017; Davis et al. 2019; Dube et al. 2009; Felitti 2009; Kraynak et al. 2019; Van Niel et al. 2014). Similar to CRPS, the affected population is predominantly female. Patients with a history of abuse often suffer from body image disturbance, which is also associated with other comorbidities, such as depression, anxiety, borderline personality disorder, and eating disorders (Dyer et al. 2013; Malecki et al. 2018). Therefore, it seems possible that individuals with traumatic experiences may be more likely to develop CRPS.

There is no consistent evidence to determine whether BPD symptoms arise as a direct result of CRPS-induced changes in specific brain areas; or if these symptoms occur in response to maladaptive learning (e.g. fear-avoidance behavior); or if BPD is influenced by psychological and social factors (Michal et al. 2017). Based on the literature, the present study sought to examine the possible role of psychological and social factors in BPD.

Based on the research described above, we tested the following three hypotheses:

  1. 1.

    In chronic CRPS Type I of the upper extremity, the severity of body schema disturbance correlates positively with psychological and social factors, including depression, anxiety, current stress, quality of life, traumatic experiences during childhood, stressful life events in the 12 months before onset of the disease;

  2. 2.

    In chronic CRPS Type I of the upper extremity, severity of body schema disturbance correlates positively with pain intensity;

  3. 3.

    In chronic CRPS Type I of the upper extremity, severity of body schema disturbance correlates positively with duration of the disease.

Methods

Recruitment and subjects

This cross-sectional study was approved by the research ethics committee of the Ludwig-Maximilian-University of Munich, Germany. Written informed consent was obtained from all participants. Sixty-eight adult patients with chronic (longer than six months) CRPS Type I of the hand were contacted, and sixty (fifty female and ten male subjects) agreed to participate. Recruitment took place in two large outpatient interdisciplinary pain clinics in Germany as well as through two internet-based CRPS self-help network sites (crps-netwerk.org; unfallopfer.de). In order to be included, the patient had to be diagnosed by following the IASP Budapest clinical criteria for the diagnosis of CRPS. The physician who made the CRPS diagnosis submitted written documentation of the diagnosis. Patients were excluded from the study if they had additional, severe medical or psychiatric comorbidities, such as multiple sclerosis, Parkinson’s disease, schizophrenia, or any form of personality disorder.

Questionnaires

Demographic data and pain

The German Pain Questionnaire, version 2015, is a 24-item self-report instrument developed by the German Association of Pain (Nagel et al. 2015). It includes demographic questions and assesses all aspects of pain (description, strength, exacerbating and palliative factors, mental model of pain) as well as the development of the disease in question, medications, forms of therapies used, medical and psychiatric comorbidities, the patient’s living situation, educational background, and work career.

Current depression, anxiety and stress

The German Pain Questionnaire also incorporates a screening instrument for depression, anxiety, and stress, the Depression Anxiety Stress Scales (DASS-21, Nilges and Essau 2015). This 21-item self-report questionnaire consists of three subscales, each comprised of 7 items rated on a Likert-type scale (0–3). The range of each subscale is therefore 0–21. The average Cronbach’s alpha coefficient for the DASS depression subscale is reported as 0.91, for anxiety 0.78–0.82, and for stress 0.81–0.89. The German version of the Hospital Anxiety and Depression Scale (HADS-D) is a 14-item self-report questionnaire used to assess depression and anxiety symptoms (Herrmann et al. 1994). Each subscale comprises seven items rated with a Likert-type scale (0–3) with total subscale scores, therefore, ranging from 0 to 21. The scale has good psychometrics for internal consistency (α = 0.80) and test-retest reliability (rho = 0.80).

Past psychological trauma

The short version of the Childhood Trauma Questionnaire (CTQ) is a 28-item self-report instrument, which was used to assess childhood maltreatment, including sexual, physical, and emotional abuse, as well as emotional and physical neglect (Bernstein et al. 1994; Spinhoven et al. 2014). Each subscale score is divided into four categories of abuse or neglect: none to low; low to moderate; moderate to severe; severe to extreme. Klinitzke and colleagues used the scale in a representative German sample and found that the subscales had a Cronbach’s alpha coefficient above 0.80. The physical neglect subscale showed high intercorrelations with the other subscales but weak internal consistency (Klinitzke et al. 2012). Klinitzke and colleagues (Kinitzke et al., 2012) hypothesize that this is probably due to some items not being exclusively dependent on the family context, but also on historical circumstances for a certain age cohort. For example, the items stating “when I grew up, I did not have enough to eat” and “when I grew up, I had to wear dirty clothes” describe the reality for large parts of the German population during and following World War II and are not due to physical neglect by caregivers. The authors strongly caution against the application of this subscale in the German population. Therefore, the physical neglect subscale was not used in the present study. The CTQ has high construct validity, as demonstrated by significant positive correlations with measures of anxiety and depression and negative correlations with measures of life satisfaction.

Past stress

The Social Readjustment Rating Scale (SRRS), commonly known as the Holmes and Rahe Stress Scale (Rahe 1978; Rahe and Arthur 1978), is a well-known standard measure of stress in relationship to illness. This 43-item dichotomous questionnaire asks subjects to indicate the number of stressful events experienced. Each event is assessed in “life change units” and given a different weight. For example, death of a spouse equals 100 points, taking out a bank loan 32 points, change in sleeping habits 16 points. The SRRS has been validated in thousands of people with different cultural backgrounds and was found to be highly reliable, with rank ordering remaining consistent both for healthy adults (r = 0.96–0.89 and patients r = 0.91–0.70) (Gerst et al. 1978). The scale was used to screen for stressful life events within the past year before the onset of CRPS.

Quality of life

The Veterans RAND 12-item Health Survey (Kazis et al. 2006) is a widely-used, well-validated measure of health-related quality of life to estimate disease burden. The twelve items are summarized into two scores, a physical health summary measure (PCS-physical component score) and a mental health summary measure (MSC-mental component score). Answers are assigend a score between 0 and 100. In the statistical analysis they receive another, weighted score (Selim et al. 2009). For both subscales, a higher score denotes better quality of life. The RAND survey is included in the German Pain Questionnaire. The German version is currently being tested, therefore psychometric data are not yet available. However, in earlier versions of the German Pain Questionnaire, the comparable 12-Item Short Form Health Survey (Ware et al. 1996) consisting of 12 questions, and assessing physical (PCS) and mental (MCS) quality of life states, was used successfully (n = 1086; PCS: M = 30.5, SD = 8.8; MCS: M = 44.7, SD = 9.4).

Body perception disturbance

Two scales were used for the assessment of motor and cognitive aspects of BPD. The Galer and Jensen Neurobehavioral Questionnaire (GJ) in the German translation (Frettlöh et al. 2006), is a seven-item Likert-type self-report scale (0 = never to 6 = always). It comprises two subscales, motor (Items 1, 3 and 4) and cognitive neglect (Items 2 and 5). The scale has excellent internal consistency (Cronbach’s alpha α = 0.86) and an effect size of d = 0.57 in distinguishing CRPS patients from normal controls. In the present study, Cronbach’s alpha was found to be 0.88.

The BATH CRPS BPD Scale (Lewis et al. 2007; Lewis and McCabe 2010) was modified and translated into German. This 7-item scale comprises two subscales: cognitive (Items 1, 3, 4, 6, and 7) and motor neglect (Items 2 and 5). Additionally, the BATH scale required the patient to draw their body (Item 7). This drawing was rated by the investigator on a scale from 0 to 2, as follows:

  • 0 = no distortion; the body is drawn anatomically correct, no fingers are missing.

  • 1 = distortion; size or/and form of the hand is anatomically incorrect, but no fingers are missing.

  • 2 = severe distortion; size or/and form of the hand is anatomically incorrect, fingers are missing.

In the original study, the questionnaire had an acceptable internal consistency (α = 0.66) and excellent inter-rater reliability (Cohen’s Kappa κ = 0.87). In the present study, alpha was found to be 0.79.

The GJ and BATH scales were strongly correlated (rho = 0.71) in this study, indicating that the two scales measure the same variables. Nevertheless, the statistical analyses utilized the respective individual scores of each scale to measure the severity of BPD.

Statistical analysis

Data analysis was conducted using IBM SPSS Statistics, Version 24 for Microsoft Windows. Descriptive statistics were used to gain a general overview of the demographic, clinical and psychometric data in terms of means and frequencies. The Shapiro-Wilk test was employed to test for normality of the data. Demographic, clinical and psychometric variables were not normally distributed, with a slight positive skewness; therefore, only non-parametric statistical tests were applied. Spearman’s rank-order correlation was used to calculate the strength of association between BPD and all other variables. For variables showing highest correlation with BPD, a multiple linear regression analysis was performed to gain a more complete understanding of these associations. Multicollinearity was assessed using Variance Inflation Factor (VIF) tolerances.

For the correlation and regression analyses, the depression and anxiety subscale scores of both the HADS-D and DASS-21 questionnaires were combined to form a single depression score and a single anxiety score. Combination of the subscales from different measures has been done elsewhere (Covic et al. 2012; Covic et al. 2009).

Results

Demographic and pain characteristics

We recruited 60 patients. The mean age was 57.8 years (± 11.82). Twenty-seven patients were affected on the left side, and 33 patients were affected on the right. The average disease duration was 4.5 years (± 3.68), and most (65%) reported a surgical procedure to be the triggering event for their CRPS. The patients reported a mean pain intensity of 4.4 (based on a Numerical Rating Scale of 0–10). There were no differences between female and male participants concerning any of the assessed variables.

Past trauma and stressful life events

The sample had significantly higher rates of childhood trauma (sexual and physical abuse, emotional neglect) compared to the overall German population (Hauser et al. 2011). Data presented in Table 1 only show severe to extreme abuse/neglect.

Participants had on average low levels of stress twelve months prior to the onset of the disease, as measured by the Social Readjustment Rating Scale (M = 97.1, SD = 95.9). A score below 150 indicates low levels of stress and a less than 30% probability of developing a stress-related illness in the next two years.

Table 1 Comparison of Past Trauma; Numbers Listed are the Percentages of the Study Participants and General Population Samples, Respectively
Full size table

Current depression, anxiety, and stress

Based on a clinical cut-off score of 8, established by Bjelland and colleagues (Bjelland et al. 2002), scores on depression and anxiety as measured by the HADS indicate that 55% of the subjects met the screening criteria for depression and 52% met the screening criteria for anxiety. Based on a clinical cut-off score of 10 for depression, 6 for anxiety and 10 for stress, established by the German Pain Questionnaire Handbook (Nagel et al. 2015), scores on depression and anxiety as measured by the DASS indicated that 30% of the subjects met the screening criteria for depression and 45% met the screening criteria for anxiety. Moreover, 50% of the patient showed abnormal levels of stress (Table 2).

Table 2 Means and Standard Deviations for all Scales
Full size table

Quality of life

Participants showed a physical component score (PCS) of 33.8 (SD = 8.2) and a mental component score (MCS) of 43.7 (SD = 14.2), indicating no severe deterioration in quality of life (Table 2). However, due to CRPS, 63% had lost their job and 23% were in early retirement.

Body perception disturbance

Generally, participants exhibited low levels of motor and cognitive levels of neglect (Table 2). The Likert-type scale was divided into two: scores between 0 and 3 meaning low levels, and scores between 4 and 6 meaning higher levels of neglect. For the GJ scale, 62% of participants had low motor and 78% had low cognitive neglect. For the BATH scale, 67% scored low in motor and 88% low in cognitive neglect. However, the majority of participants drew either a distorted (47%) or severely distorted (37%) body image (Fig. 1a and b, respectively).

Fig. 1
figure 1

Example for a Distorted (a) and a Severely Distorted (b) Body Image Drawing

Full size image

Correlational and regression analyses

It was hypothesized that in chronic CRPS of the upper extremity, the severity of BPD would correlate with psychological and social factors (hypothesis 1); pain intensity (hypothesis 2), and duration of disease (hypothesis 3). Hypothesis 1 was partially confirmed: BPD showed a strong correlation with depression and anxiety. Stress suffered in the past week was also significantly, but to a lesser extent, correlated with BPD. Both subscales assessing quality of life were significantly correlated with BPD. Hypothesis 2 was confirmed: Average pain intensity showed a strong correlation with BPD (Table 3).

Table 3 Correlations between Body Perception Disturbance, Pain, Disease Duration, Psychological and Social Variables
Full size table

There was no significant correlation between BPD, disease duration (hypothesis 3), and stress suffered before the developing of CRPS, or with childhood trauma (Table 3). However, one of the four subscales of the CTQ measures emotional neglect and fully 68% of patients reported emotional neglect at a low (20%), medium (10%) or high (38%) level. In addition, emotional neglect in childhood correlated with current depression (rho (60) = 0.29, p < .05) and anxiety (rho (60) = 0.25, p < .05).

As will be discussed in the next section, the confirmatory findings are striking because the majority of participants showed neither severe motor nor cognitive neglect as measured with GJ and BATH scales. This suggests that the assessments of both BPD and psychological variables were highly sensitive, because even within a limited range of scores, correlations and regressions were obtained between BPD and several psychological and social variables.

Multiple linear regression analysis was performed to test if anxiety, depression, current stress and pain predicted participants’ BPD ratings. In the case of the GJ scale, pain, anxiety and current levels of stress predicted 47% of the variance. In the case of the BATH scale, two predictors, namely pain and depression explained 44% of the variance (Table 4). The values of tolerance for the independent variables predicting both body perception disturbance scores were higher than 0.3 and, simultaneously, the values of Variance Inflation Factor (VIF) were lower than 3.2. Therefore, it is prudent to assume that no multicollinearity was present in the data.

Table 4 Summary of Multiple Linear Regression Analysis for Variables Predicting Body Perception Disturbance Ratings
Full size table

Another multiple linear regression analysis was utilized to test which variables (BPD, pain, depression, anxiety, stress) influenced quality of life. The physical aspect of quality of life (PCS) was only predicted by pain, which explained 25% of the variance. The mental aspect of quality of life (MCS) was predicted by depression and stress, which together explained 72% of the variance (Table 5). The values of tolerance for the independent variables predicting MCS were higher than 0.2 and, simultaneously, the values of Variance Inflation Factor (VIF) were lower than 3.3. Therefore, it is prudent to assume that no multicollinearity was present in the data.

Table 5 Summary of Multiple Linear Regression Analysis for Variables Predicting Quality of Life
Full size table

Discussion

This study found that Body Perception Disturbance (BPD) in Complex  Regional Pain Syndrome (CRPS) was significantly correlated with measures of depression, anxiety, current stress, quality of life, and pain intensity, but not with stress experienced in the twelve months preceding onset of illness, childhood trauma, or duration of illness. As was noted earlier, in their prospective multicenter study, Beerthuizen and colleagues (Beerthuizen et al. 2011) found that depression, anxiety and other psychological factors, as well as stressful life events, did not predict the development of CRPS in patients suffering a physical trauma. These psychological conditions are therefore likely to be sequelae of CRPS. Instead of serving as predisposing psychological risk factors, psychological abnormalities in CRPS seem to be a consequence of chronic pain and disability and the resulting changes in the brain. However, it should not be assumed that the directionality of physical brain changes to psychological symptoms is one way, with brain alterations resulting in depression, anxiety and stress. Psychological symptoms may also cause changes in the brain (Mah et al. 2016; Yaribeygi et al. 2017). For example, in a recent review (Sheng et al. 2017) concerning the link between chronic pain and depression, the authors conclude that research clearly shows the close connections between pain- and depression-induced neuroplasticity as well as changes in neurobiological mechanisms. Psychological symptoms interact with the physical illness, and with alterations in the brain, in a reciprocal, bidirectional manner, such that depression, anxiety, and stress may increase the severity of BPD and the intensity of pain and disability associated with CRPS, as well as alterations in brain activity and cortical mass, which then may amplify depression, anxiety, and stress in a vicious spiral, thereby diminishing experienced quality of life. The finding that the majority of patients had experienced emotional neglect may provide a partial explanation for BPD. Emotional neglect in childhood is often associated with, and believed to cause, insecure attachment (Benamer and White 2008). Insecure attachment is highly associated with emotion dysregulation, because children in an insecure relationship with their parents typically do not receive adequate soothing, leading to incapacity to self-soothe both as a child, and as an adult. The inability to self-soothe may then result in increased distress, such as catastrophizing thoughts, about CRPS, leading to worsened BPD.

These findings strongly suggest that psychological and social variables must be incorporated into a comprehensive understanding of the nature of BPD in CRPS. Of course, correlation does not establish causation, and the directionality of the relationship between these variables and the illness cannot be established without further prospective, longitudinal research.

From a methodological perspective, the present study demonstrated the usefulness of including two measures of BPD – the GJ and the BATH scales, the latter being a new scale with a particularly interesting behavioral assessment element -- having patients draw their afflicted limb. As they were highly correlated, using two scales also added important convergent validity.

Future research needs to identify which patients are likely to have these psychological conditions triggered by CRPS, so that early, preventive intervention is possible. Through such intervention, BPD might be avoided or ameliorated, resulting in improved quality of life.

Limitations of the present study

The sample was not randomly selected, opening it up to the possibility of unknown selection biases.

The majority of participants showed neither severe motor nor clinical neglect as measured with GJ and BATH scales. Although, as noted earlier, this suggests that the assessments of both BPD and psychological variables were highly sensitive, it would be preferable in a next study to recruit a sample with a wider range of motor and cognitive neglect. In retrospect, it might have been a limitation to eliminate patients with psychiatric disorders, since they are most likely to report higher levels of psychological distress surrounding BPD and CRPS. This would likely increase the already strong correlations obtained, and also result in a normal distribution, avoiding the problems encountered in the present study due to a non-parametric data set.

This was a cross-sectional study; as noted above, in order to better assess the potentially causative role of psychosocial variables on BPD, a prospective, longitudinal design would be necessary. However, it is always recommended that, as was done in the present study, cross-sectional studies be performed prior to expensive, complicated prospective studies.

Clinical implications

The present study suggests that comprehensive treatment of CRPS and BPD associated with CRPS must address patients’ risk for the depression, anxiety, and stress associated with it. Recovery from CRPS likely requires treatment of these associated psychological conditions. The possible relationship between a history of childhood emotional neglect and BPD also strongly suggests the need for trauma-based, attachment-based psychotherapy.