Potential Risk Factors for the Onset of Complex Regional Pain Syndrome Type 1: A Systematic Literature Review

Anaesthetists in the acute and chronic pain teams are often involved in treating Complex Regional Pain Syndromes. Current literature about the risk factors for the onset of Complex Regional Pain Syndrome Type 1 (CRPS 1) remains sparse. This syndrome has a low prevalence, a highly variable presentation, and no gold standard for diagnosis. In the research setting, the pathogenesis of the syndrome continues to be elusive. There is a growing body of literature that addresses efficacy of a wide range of interventions as well as the likely mechanisms that contribute to the onset of CRPS 1. The objective for this systematic search of the literature focuses on determining the potential risk factors for the onset of CRPS 1. Eligible articles were analysed, dated 1996 to April 2014, and potential risk factors for the onset of CRPS 1 were identified from 10 prospective and 6 retrospective studies. Potential risk factors for the onset of CRPS 1 were found to include being female, particularly postmenopausal female, ankle dislocation or intra-articular fracture, immobilisation, and a report of higher than usual levels of pain in the early phases of trauma. It is not possible to draw definite conclusions as this evidence is heterogeneous and of mixed quality, relevance, and weighting strength against bias and has not been confirmed across multiple trials or in homogenous studies.


Introduction
There is a growing body of literature addressing a variety of disorders known as Complex Regional Pain Syndrome (CRPS). It is a condition that presents with a pain experience that is severe and disproportionate to the inciting event and is accompanied by highly variable signs and symptoms of inflammatory, sensory, autonomic, trophic, or motor features. Anaesthetists in the acute and chronic pain teams are often involved in treating Complex Regional Pain Syndromes. The onset of CRPS can follow injuries ranging from minor injuries to fracture(s), from lesions of the central nervous system, or from surgery [1][2][3]. Its prevalence is low, ranging from 5.46 to 26.2 per 100 000 [4,5]. This low prevalence has led to difficulty in research where robust statistical analysis necessitates larger sample sizes [6,7]. Furthermore, CRPS nomenclature continues to be debated and remains controversial [8][9][10][11]. Research had shown that this condition is not wholly a problem of the sympathetic nervous system. The old terms "reflex sympathetic dystrophy" and "causalgia" needed to be changed [12]. In 1994, the committee for taxonomy of the International Association for the Study of Pain (IASP) identified specific diagnostic criteria for this syndrome that were termed the IASP criteria. This IASP committee changed the name to Complex Regional Pain Syndrome or CRPS. The term CRPS Type 1 (CRPS 1) applies, if there is no nerve damage, or CRPS Type 11, if the nerve is physically and permanently damaged, and this nomenclature replaced the terms "reflex sympathetic dystrophy" and "causalgia, " respectively [13]. Other diagnostic criteria developed were the Veldman [14] and Harden/Bruehl [15] criteria that 2 Anesthesiology Research and Practice continue to be used in clinical practice and research. The Harden/Bruehl criteria became known as "The Budapest Criteria" with minor modifications. Though published in an IASP-sanctioned book, the Harden/Bruehl criteria have not been officially endorsed by the IASP. The "Budapest Criteria" are used in clinical diagnosis. Here a report of at least one symptom in 3 or 4 categories (sensory, vasomotor, sudomotor/oedema, motor/trophic) with at least one sign at time of evaluation in 2 or more of the categories (sensory, vasomotor, sudomotor/oedema, motor/trophic) confirms a clinical CRPS diagnosis. There must be no other diagnosis that better explains the signs and symptoms. Budapest Clinical Criteria have retained sensitivity almost identical to the IASP criteria but with much improved specificity.
In the "Budapest Research Criteria, " diagnostic decision rule is at least one symptom in all four symptom categories and at least one sign (observed at evaluation) in two or more sign categories. The intent of the Budapest Research Criteria was to maximize specificity (minimize false positives) at the expense of sensitivity. They have a high specificity but a low sensitivity [16,17]. This systematic review explores the literature since 1999. It therefore includes a level of variation for diagnostic criteria. CRPS 1 is considered by most to be overdiagnosed [6,18,19]. There are a few, however, who still consider it underdiagnosed [20]. The precise pathophysiological mechanisms and predictive factors underlying CRPS 1 or subsets of CRPS 1 remain unknown [6,[21][22][23]. A standard diagnostic test is unavailable and the absence of a gold standard makes the validation of diagnostic criteria difficult [6,24,25]. Effective treatment strategies (in both the research and clinical fields) have moderate evidence [26][27][28][29]. A variety of medical and physiotherapy interventions and a multidisciplinary approach to the management of CRPS 1 continue to be widely used [23,30,31]. These factors contribute to the difficulty in determining potential risk factors for CRPS 1 in a reliable and statistically valid way.
There has been no systematic review of risk factors which may contribute to the onset of CRPS 1. This is the first systematic review to address factors posing as possible risk factors for the onset of CRPS 1. This paper selects from the current literature to systematically describe factors which expose a potential risk factor for a possible relationship to the onset of CRPS 1. This paper defines a risk factor as a factor contributing to a likely association of the onset of CRPS 1. This association is not necessarily causal.

Study Selection.
Key words for CRPS (such as diagnosis, epidemiology, aetiology, genetics, history, pathophysiology, rehabilitation, risks, fractures, osteoporosis, or predictors) were combined in searches of Web of Science and OVID Medline for articles dated 1996 to April 2014. All abstracts were screened. Inclusion criteria for data extraction were articles written in English with reference to risks or predictors associated with the onset of CRPS 1. Exclusion criteria included articles written in other languages or no mention of CRPS 1 risks for either onset or outcomes or prognosis. A   CRPS 1. Exclusion criteria incorporated CRPS Type 11, methodology used in animal studies, case studies, and case control studies or cross-sectional studies. Sixteen articles met these inclusion criteria (10 prospective studies, 6 retrospective studies). Twenty-five articles were excluded due to the direction of effect not being able to be determined. These inclusion and exclusion criteria were used to determine evidence for a direction of the effect specific for the likelihood or not for the potential risk for the onset of CRPS 1. A prospective or retrospective study can provide evidence for the likelihood (or not) of a risk towards the onset of a disease by determining a direction of effect. Animal studies, case studies, case control studies, and cross-sectional studies provide evidence of a relationship. They are not able to determine the direction of effect of this relationship over a period of time. Longitudinal, prospective, or retrospective studies by nature of their design are more able to determine this necessary direction of effect to reveal the potential risks for the onset of disease [7]. Figure 1 summarises the data extraction.

Data Synthesis.
No randomised controlled trials (RCTs) were found describing either possible risk or predictive factors for the onset of CRPS 1. Ten prospective studies and six retrospective studies (total of 16) were included for the data synthesis. These data were synthesised from the following Six studies used a variation between the Veldman and Harden/Bruehl criteria. One study did not define their criteria stating their use of "standard criteria. " Criteria were not stated at all in one study. In the prospective studies, follow-up time varied from 3 months to 2 years. In the retrospective studies, follow-up time varied from 3 to 10 years. Table 3 describes the data from the literature synthesised from prospective studies. Table 4 describes the data from the literature synthesised from retrospective studies.

Data
Analysis. The data show a high level of heterogeneity. There are no particular variables consistent across these studies with evidence strong enough to comprise a risk factor. Rating criteria for quality and relevance and weighted strength against bias were based on published recommendations [7,62,63]. Quality and relevance criteria used included the following: the sample had to be representative of the CRPS 1 population; an adequate control group was needed; study attrition rate was required; adequate description of study and measurements used were necessary (to identify a potential risk factor); the statistical analysis needed to be appropriate. The data were analysed for a weighted strength against the risk of possible bias. The criteria used included bias risk in sample selection, study design, funding provision, detection, and measurement. Two authors (Tracey Pons, Roger T. Mulder) independently assessed each paper for quality, relevance, and weighted strength against potential bias. Any disagreement was discussed and resolved by consensus. Where a disagreement could not be resolved by consensus, the two other authors (Edward A. Shipton, Jonathan Williman) arbitrated disagreement and facilitated consensus amongst all four authors. Observer expectancy was considered to be reduced since 2 authors' (Roger T. Mulder, Jonathan Williman) expertise is outside the pain management field; hence, no external observer was included as they were considered objective enough with no historical or current involvement with CRPS 1 diagnosis or management.
Quality and relevance were measured against six factors. Ratings were graded as poor, adequate, or good with the following algorithm: good = five or six factors rated as yes; adequate = three or four factors rated as yes; and poor = less than two factors rated as yes. The weighting against bias was measured against five factors. Ratings were graded as weak, acceptable, or strong with the following algorithm: strong = all five factors rated as no risk; acceptable = three or four factors rated as no risk; and weak = two or less factors rated as no risk.      These data provide a broad and heterogeneous research platform towards probing for possible risk factors for the onset of CRPS 1. In this systematic review, 2 prospective studies and 3 retrospective studies were rated as good (total of 5). Four prospective studies and 1 retrospective study were rated as adequate (total of 5). Four prospective studies and 2 retrospective studies were rated as poor (total of 6). For the weighted strength against bias, 7 prospective studies were weak, 2 were acceptable, and 1 was strong. In the retrospective studies, 3 were weak, 2 were acceptable, and 1 was strong. In summary for the weighted strength against bias, 10 were weak, 4 were acceptable, and 2 were strong. However, these findings should be treated with caution as their statistical reliability and consistency have not been established across multiple or homogeneous studies.
The quality and relevance data are outlined in Table 5 for prospective studies and in Table 6 for retrospective studies. The weighting strength against bias data is illustrated in Table 7 for prospective studies and in Table 8 for retrospective studies.
The following are shown not to be risk factors for the onset of CRPS 1: namely, preoperative psychological distress; preoperative pain levels (with poor quality, poor relevance, and weak weighting against bias); psychological behaviour and depression (with adequate quality and relevance but with weak weighting against bias); and a diagnostic bone scan (with adequate quality and relevance but with weak weighting against bias). The factors not considered to be risk factors for the onset of CRPS 1 are summarised in Table 9.
The potential risk factors identified with a strong weighting against bias as well as good quality and relevance are being female (particularly postmenopausal), a fracture of the distal radius, and dislocation or an intra-articular fracture of the ankle. The factors presenting as possible risks for the onset of CRPS 1 are summarised in Table 10.

Discussion
Potential risk factors identified (strong weighting against bias, good quality, and relevance) across the 16 papers    Sandroni et al. 2003 [4] Electronic medical records of one region, population gender mix not given    [41] Acceptable Adequate are as follows: being female (particularly postmenopausal); obtaining a fracture of the distal radius; suffering an ankle dislocation or intra-articular fracture; and reports of higher than usual levels of pain in the early phases after trauma. The findings of this systematic review should be treated with caution as their statistical reliability and consistency have not yet been established across multiple or homogeneous studies and diagnostic criteria were mixed with Budapest Criteria not being used. Age was accounted for in most studies. Age as a consistent potential risk factor for the onset of CRPS 1 could not be identified. This is shown in the population studies by Sandroni et al. [4], Moseley et al. [41], and de Mos et al. [5], as well as in the study by van Rijn et al. [22]. Although most of these data show that the risk increases in postmenopausal women [5,[34][35][36][37][38]40], the retrospective studies by Allen et al. [42] and by Anderson and Fallat [44] show a lower age. This might be due to average age of the group sample with the inclusion of both genders. The average age of the female sample groups is, unfortunately, not provided in either of these studies. Females at any age pose a higher risk for the onset of CRPS 1. However, the study of males in the armed forces by Duman et al. [43] shows that males are vulnerable as well. This systematic review shows that the onset of CRPS Type 1 is higher in females than in males in the mixed gender studies.
The data show the cause of the inciting event to be mixed. It can be related to surgery, fractures, or soft tissue injuries. The presence of other comorbidities is neither predictive nor a risk factor for the onset of CRPS 1. Reports of higher than usual levels of pain in early phase of trauma were cited as strong evidence of a risk factor by Beerthuizen et al. [35] but as weak evidence by Jellad et al. [38].
Psychosocial factors are weakly weighted as a risk factor for the onset of CRPS 1. Moseley et al. find catastrophising not to be predictive for the onset of CRPS [41]. This is confirmed by other reviews and studies investigating psychological influences on the onset and progression of CRPS [58,59,[64][65][66]. Psychological behaviour, depression, and preoperative psychological distress or pain levels are not predictive of the onset of CRPS. The earlier literature described the "Sudeck A personality, " a personality of high anxiety, as a likely risk factor towards the onset of CRPS 1 [61]. A high anxiety personality trait was identified by this systematic review as only a weak potential risk [37]. In other persistent pain conditions, these complex interactions between the onset of the pathogenesis of CRPS 1 and psychological factors are predictive of level of function [45,60]. Their interaction in CRPS 1 continues to be investigated by clinicians and researchers [5,64].
These data show that a positive diagnostic bone scan is not a risk factor for the onset of CRPS 1 [46] and has been confirmed by other studies [42,47,48]. Interobserver consistency with interpretation of bone scans appears to be variable [49]. However, a diagnostic bone scan has been found to be helpful towards a diagnosis of CRPS 1 in some observations [50,51], but since recovery of bone mass following ankle fractures remains variable, it is not necessarily indicative of CRPS 1 [52].
Two recent systematic reviews have collated the prognostic findings about CRPS 1 [53,67]. Both agreed that the quality of evidence is poor. Our systematic review has confirmed this regarding the risks of onset of CRPS 1. One review retrieved 1648 relevant papers of which twelve were robust enough for qualitative analysis [53]. Prognostic factors for poor outcomes were grouped within 7 clinical clusters as follows: (1) gender, where two studies show the male gender and one study shows the female gender; (2) age, where there is a high variation in age of onset affecting prognosis; (3) inciting event, such as polytrauma, inciting event other than fracture, severe initial injury, and distal articular location; (4) localisation site, either upper or lower extremity; (5) clinical features, such as exercised induced pain, sensory disturbances, initial cold skin temperature, complications of infection, skin ulcers, chronic oedema, dystonia or myoclonus, algodystrophy score > 7 out to 10, low SF-36 general health score, disease duration > 1 year, and coexistence of misdiagnosed nerve injury and compression; (6) associated comorbidities included alcoholism and psychological background in nontraumatic CRPS 1; and (7) diagnosis where a delay of >2 months after inciting event was shown to be associated with poorer outcomes. The other review found that many CRPS 1 patients recover in 6-13 months but that a significant number continue to experience persistent pain and disability [67].

Implications for Research.
This systematic review highlights potential risk factors that will contribute to future exploration about the onset of CRPS 1. Identifying risk factors associated with a poor prognosis is important as well. Risk factors for the onset of CRPS 1 identified in this systematic review may or may not be associated with a poor prognosis. The low prevalence of CRPS 1, its heterogeneous presentation, and its lack of highly specific or sensitive diagnostic criteria as well as the lack of clarity around consensus for these criteria create challenges in carrying out research [4,6,21,42,54,55]. In CRPS 1, more trials across different settings are needed. The cross-sectional and case control studies excluded from this systematic review may still offer insight into the development of future longitudinal studies to determine direction and strength of the effects.

Implications for Clinical Practice.
No specific or sensitive clinical sign or clinical symptom was shown in this review to pose a risk factor for the onset of CRPS 1. Clarity around the sensitivity and specificity of laboratory and imaging testing is needed [42,47,48,51,52]. This review confirms the importance of maintaining clinician awareness and of being aware of potential risk factors to enable the early diagnosis of CRPS 1 [56]. Evaluation by experienced clinicians hastens the diagnosis [36,57]. Early diagnosis and referral to pain management specialists and physiotherapists are related to better outcomes [53,68,69]. Moseley et al. [41] suggested that a pain score of ≥5 in the first week of fracture could be considered a "red flag" risk for the likely onset of CRPS 1.

Conclusion
This systematic review shows that the accurate potential risk factors for the onset of CRPS 1 remain elusive. Studies remain heterogeneous, of mixed quality and relevance, and with varied weighting against the risks of bias. The low prevalence of CRPS 1 accompanied by a lack of a gold standard for diagnosis contributes to the difficulties around determining potential risk factors for the onset of CRPS 1.
Potential risk factors identified with strong weighting against bias and good quality and relevance are summarised as follows: being female (particularly postmenopausal); obtaining a fracture of the distal radius; suffering an ankle dislocation or intra-articular fracture; and reports of higher than usual levels of pain in the early phases after trauma. Potential risk factors with much weaker weighting against bias and poorer quality and relevance include immobilisation, psychosocial barriers, and a positive diagnostic bone scan. Definite conclusions cannot be drawn as evidence remains inconsistent across multiple trials or in homogenous studies.