Person-centred management of upper limb complex regional pain syndrome: an integrative review of non-pharmacological treatment

Abstract

Introduction

Complex Regional Pain Syndrome (CRPS) is most common in the upper limb and associated with high disability. The purpose of this review was to critically appraise and synthesise literature exploring non-pharmacological treatment for upper limb CRPS, to guide upper-limb-specific management.

Methods

Using an integrative review methodology, 13 databases were searched to identify all published studies on non-pharmacological management of upper limb CRPS. The Crowe Critical Appraisal Tool was used to provide quality ratings for included studies, and analysis employed a qualitative descriptive approach.

Results

From 236 abstracts reviewed, 113 full texts were read, and 38 articles selected for data extraction. Designs included single case (n = 14), randomised controlled trial (n = 8), prospective cohort (n = 8), case series (n = 4), retrospective (n = 3), and mixed methods (n = 1). Interventions were categorised as sensory retraining (n = 13), kinesiotherapy (n = 7), manual therapies (n = 7), physical modalities (n = 6), and interdisciplinary treatment programmes (n = 5). All studies measured pain intensity, and most (n = 24) measured physical parameters such as strength, movement, or perceptual abilities. Few measured patient-rated function (n = 13) or psychological factors (n = 4). Quality ratings ranged from 30% to 93%, with a median of 60%.

Conclusion

Methodological quality of non-pharmacological treatment approaches for upper limb CRPS is overall poor. Movement, desensitisation, and graded functional activity remain the mainstays of intervention. However, despite the impact of CRPS on wellbeing and function, psychological factors and functional outcomes are infrequently addressed. Further robust research is required to determine which aspects of treatment have the greatest influence on which symptoms, and when and how these should be introduced and progressed.

Introduction

Complex Regional Pain Syndrome (CRPS) is a pain condition that often becomes chronic. ¹ It is characterized by severe pain and sensory, autonomic, trophic, and motor changes, arising most commonly after injury to a limb. ² A complex interplay of aberrant inflammatory responses, vasomotor dysfunction, and central nervous system sensitisation is thought to occur. ² Combinations of signs and symptoms vary between individuals and change over time, challenging and delaying diagnosis and the start of appropriate treatment. ³ Such delays frequently lead to poorer outcomes,^1,4 and complications with insurance cover and other medico-legal factors, increasing stress for the individual.^3,5 In addition, clinicians differ in how they treat CRPS, based on individual experience or training, risking the provision of inconsistent and even inappropriate treatment. ⁵ Guidelines for CRPS management recommend an early interdisciplinary approach including non-pharmacological strategies (e.g. rehabilitation, psychological therapy, and education) alongside pharmacological management, however, determining the optimal approach to therapy remains challenging. ⁶ Addressing the underlying mechanisms has been proposed as one treatment algorithm to guide therapy. ⁷ Key mechanisms implicated in the maintenance of CRPS signs and symptoms include inflammation, sensitisation, maladaptive plasticity, and psychological factors.^7,8 Non-pharmacological treatments commonly used to address these include movement therapies, desensitisation, perceptual training (e.g. mirror therapy), education, and psychological interventions. ⁶ However, one challenge to prescribing intervention based on assumed mechanisms is that CRPS pathophysiology is still not completely understood, with new insights frequently emerging and challenging theories on which interventions were based.^9,10 Researchers have therefore called for increased phenotyping of cohorts in future studies, to enable a greater understanding of mechanisms involved in this complex condition.^11–13

Complex regional pain syndrome most frequently affects the upper limb, with high associated disability.^14,15 Upper limb disability has profound consequences for activities of daily living^16,17 and wellbeing; dysfunction can be associated with grief and lost self-identity, ¹⁸ while visible changes can elicit a sense of shame and embarrassment. ¹⁹ Given the differences in both form and function of the upper and lower limb, a one-size fits all treatment approach for CRPS may not be appropriate. However, no evidence synthesis has critically appraised non-pharmacological treatments for only upper limb CRPS. This integrative review aims to critically appraise and synthesise evidence for non-pharmacological treatment specific to upper limb CRPS.

Methods

We chose an integrative methodology for this review to enable the inclusion of experimental, non-experimental, and theoretical literature while retaining the transparent and systematic approach required for a high-quality review with minimally biased recommendations. ²⁰ The study design was guided by Whittemore and Knafls’ five-stage methodological approach ²¹ and the Preferred Reporting Items for Systematic Reviews and Meta-analysis Extension for Scoping Reviews (PRISMA-Scr) checklist. ²² The protocol was registered prospectively with Prospero in December 2021, with the following identifier: CRD42021295606.

Problem identification

The question guiding this integrative review was: “What is the evidence for non-pharmacological treatment of adults with upper limb CRPS, reported in the literature between 2005 to 2021?” CRPS was defined according to the Budapest diagnostic criteria (Figure 1), ²³ due to their validated sensitivity and specificity. ²⁴ Articles that referenced the Budapest research criteria were also considered eligible. ²⁵ The starting date for this review was set to reflect the emerging use of these criteria. Given the wide range of non-pharmacological treatments shown in a preliminary literature search, we decided to define and exclude surgical and pharmacological treatment rather than attempting to define and include non-pharmacological treatments. Surgical and pharmacological treatments were defined as sympathectomy, amputation, invasive neurostimulation, and any drug administered either topically, orally, intrathecally or via injection (e.g. nerve blocks, cortisone). Upper limb was defined as spanning from the fingertip to shoulder, including the axilla.

Figure 1.

Budapest diagnostic criteria. ¹²

Literature search

A systematic search was conducted in December 2021, using the following databases: CINAHL, Cochrane Library, OVID (AMED, Embase, Emcare, JBI, Medline, PsycInfo), PEDro, ProQuest, PubMed, Scopus, and Web of Science. The search was limited to peer-reviewed English publications available from 2005 onwards (see Figure 2 for the list of search terms). After deduplication using EndNote, one author (GG) independently reviewed titles, abstracts, and full texts for inclusion, and conducted hand searches of reference lists. A second author (BLT) performed 25% of tasks at each stage and compared with GG for consistency (see Figure 3 for flow diagram). The Rayyan web application (Qatar Computing Research Institute) was used to facilitate decision blinding. ²⁶

Figure 2.

Literature search terms.

Figure 3.

PRISMA flow diagram.

To be included, studies had to present original research describing non-pharmacological intervention for CRPS of the upper limb. If studies included both upper and lower limbs, they were included if results reported these data separately. Studies were excluded if a description of signs and symptoms did not match Budapest criteria; if treatment was solely pharmacological or surgical; if they discussed paediatric populations only; if they studied lower limb CRPS only; or if they did not explicitly state CRPS location. Studies including only persons with CRPS post-stroke were also excluded, due to differences in features (such as increased likelihood for motor dysfunction in the post-stroke population), indicating reduced validity of the Budapest criteria with this cohort. ²⁷

Data evaluation

Data were extracted from included studies using standardised Excel forms. We based these on Cochrane data extraction forms for intervention reviews and included risk of bias assessment. ²⁸ Data extracted included the title, author(s), country, aim, design, participant description, eligibility criteria, recruitment process, baseline characteristics, intervention, outcome measures used, and conclusions. Studies were then critically reviewed using the Crowe Critical Appraisal Tool (CCAT) (v1.4), ²⁹ chosen for its reliability and validity^30,31 and inclusion of a wide range of designs. The CCAT appraises studies on their own merits across eight quality categories (preliminaries, introduction, design, sampling, data collection, ethical matters, results, and discussion), not against a ‘gold standard’ methodological design. Each category can score up to five points, and these are then summed into a total score and a percentage provided. GG extracted data from and critically appraised all articles, with 25% of tasks independently conducted by BLT at each stage and discussed for consensus.

Data analysis

A qualitative descriptive analysis approach was used to determine categories without abstracting too far from the raw data. ³² This process consisted of repeated reading, reviewing, and refining of categories while keeping in mind the whole study. ³³ GG performed data analysis and verified truthfulness by discussing emerging patterns with BLT, and regularly returning to the original articles to confirm these.

Positionality statements

Researcher reflexivity is essential to pursuing rigour within the critical realist philosophical framework underpinning this study. ³⁴ The authors hold dual roles as clinician-researchers, ³⁵ with professional backgrounds in occupational therapy (GG and BLT), physiotherapy (JD) and psychology (DS), including specialisation in upper limb rehabilitation (GG and JD) and pain management (BLT). To increase objectivity and promote study rigour, we met regularly to reflect upon our interests, beliefs, and assumptions, and thus make explicit how these influence the research process, with notes recorded from these meetings and referred to frequently throughout the stages of data gathering, critical appraisal, and analysis. ³⁵

Results

A total of 236 studies were identified for abstract review, after which 113 full texts were assessed for eligibility. After exclusions, a total of 38 studies were included in the review (Figure 3).

Characteristics of included studies

Study design, characteristics of participants, interventions, outcome measures, and CCAT quality rating were collated (see Table 1). Six different designs were identified; 14 single case studies,^36–49 eight randomised control trials,^50–57 eight prospective cohort studies,^58–65 four case series,^66–68 three retrospective notes reviews,^69–71 and one mixed methods study. ⁷² 13 interventions were categorised as sensory retraining,^{37,38,40,42,54,55,57,58,63,64,67,69,71} seven as kinesiotherapy,^{36,39,45,48,65,66,68} seven as manual therapies,^{43,44,46,49,53,56,73} six as physical modalities,^{50–52,59,62,72} and five as interdisciplinary treatment.^{41,47,60,61,70}

Table 1.

Study interventions and results with quality scores.

Author	Country	Methods	Intervention	Outcome measures	Number (F/M)	Age range/mean	Mechanism	CRPS duration	Result	CCAT score, %
Bakilan et al. (2018)	Turkey	Single case study	Kinesiotherapy: 30 sessions consisting of contrast bath, TENS, stretching and strengthening exercises. No further information provided	Clinical signs and symptoms	1 M	78	Herpes zoster	>10 days	After 6 weeks “...the complaints of the patients were markedly resolved and, dryness, blue colour of skin were disappeared [sic]” (p. 64)	35
Benedetti et al. (2020)	Italy	Randomised control trial	Physical modality: TAU (physiotherapist-delivered rehabilitation 2 h per day over 10 consecutive days) compared to TAU plus bio-electro-magnetic-energy-regulation (BEMER) therapy, provided for 20 min per day, daily for 10 days. Protocol described in text	VAS, DASH and grip strength	n = 12		Fracture		Participants who received BEMER therapy in addition to TAU achieved statistically significant improvements in pain and grip strength, but not in DASH scores	78
Bultitude and Rafal (2009)	UK	Single case study	Sensory retraining: 9 sessions over 15 weeks of prism adaptation, and mirror therapy. Treatment occurred for 3 weeks, followed by a 13-days washout period, a week of treatment with the unaffected hand, and a further 9 weeks of treatment with the affected hand	NPRS, straight ahead pointing, clinical signs and symptoms	1 F	53	Fracture	>3 months	Treatment resulted in reduced pain and improved symptoms, however these results were not maintained long term	40
Cameron (2009)	Scotland	Single case study	Sensory retraining: Three-stage motor imagery programme (hand laterality, guided imagery, and mirror therapy), 2 weeks per stage, including practice of the relevant task (stage-dependant) once every waking hour	NPRS (defined as VAS), Pain Self-Efficacy Questionnaire, clinical signs and symptoms	1 M	42	Post-chemotherapy	>3 years	Reduction in pain and limb signs, and an improvement in self efficacy	50
Christophe et al. (2016)	France	Prospective cohort	Sensory retraining: 8 sessions of prism adaptation (2 sessions per day, separated by at least 5 h)	Visual straight ahead, manual straight ahead (L/R), line bisection, VAS, Sickness Impact Profile	n = 7 (6/1)	24–58 (M 44.4)	Trauma, surgery	4–36 months (M 13.2)	Statistically significant improvements in pain and quality of life following intervention, and non-significant differences between pre/post adaptation measures of spatial reference frames	60
David et al. (2015)	Germany	Prospective cohort	Physical modality: Single 45-min session of repetitive sensory stimulation (rSS) provided to fingertips over 5 consecutive days. Protocol described in text	Two-point-discrimination-threshold, mechanical-detection-threshold, NPRS	n = 20 (10/10)	28–71 (M 55.2 ± 2.1)	Trauma, surgery, fracture	2–27 months (M 8.7 ± 5.8)	Low Frequency-rSS stopped after 4 patients due to no improvement in pain and impairment in two-point discrimination. After 5 days of high Frequency-rSS, two-point discrimination was improved significantly, and mechanical detection threshold was significantly decreased	65
Devrimsel et al. (2015)	Turkey	Randomised control trial	Physical modality: TAU (ROM and stretching exercises and underwater ultrasound therapy) with either whirlpool bath or neuromuscular electrical stimulation (NMES). One session per day, 5 days a week, over 3 weeks. Protocol described in text	Edema, ROM, VAS, grip and pinch strength	n = 60 (35/25)	26–58 (M39.53 ± 9.37)	Fracture, tendon injury	M 5.46 ± 0.73 and 5.11 ± 0.73 weeks	Statistically significant improvements in all outcomes in both groups. The group treated with whirlpool bath had significantly better improvements than the group treated with NMES.	38
Dimitrijevic et al. (2014)	Serbia	Randomised control trial	Physical modality: TAU (active and active assisted exercises up to pain threshold for 30 min twice a day) with either low level laser therapy (LLLT) or interferential current therapy (IFC), for 30 min, twice daily, daily for 5 out of 7 days over 2 weeks, then alternative days for 10 sessions. Protocol described in text	VAS, circumference, ROM	n = 21	M 53.90 ± 13.36 and 57.80 ± 10.75			Both groups showed statistically significant reduction in pain and edema and improved ROM. LLLT was significantly better than IFC at reducing pain intensity and edema	63
Duman et al. (2008)	Turkey	Randomised control trial	Manual therapy: TAU (therapeutic ultrasound, stellate ganglions and therapeutic exercises) compared to manual lymphatic drainage therapy (MLDT) provided as a 45-min session 5 days a week for 3 weeks. Both groups then underwent a 2-months maintenance period including exercises (plus self-massage for the study group)	VAS, edema, ROM (defined as function)	n = 34	M 20.6 ± 0.8	Fracture, soft tissue trauma, laceration, spontaneous	M5.1 ± 1.3 months	Both groups achieved statistically significant changes in pain reduction, but non-significant improvements in function (measured as ROM). The group receiving additional MLDT achieved statistically significant improvements in edema compared to the TAU group	60
Elomaa et al. (2019)	Finland	Prospective cohort	Interdisciplinary treatment: Physiotherapy: 9 weeks of weekly alternating individual and group sessions (GMI and exercises for grip, dexterity, two-point discrimination, balance, postural control, and body perception. Psychology: 7 weeks of weekly 90-min group sessions (CBT: Relaxation, behavioral activation, exposure to previously avoided situations. ACT: Body scan, breathing, attention management, mindfulness in activities of everyday living)	Health-Related Quality of Life (15-D), Beck Depression Inventory (BDI-II), Chronic Pain Acceptance Questionnaire (CPAQ), DASH, Pain Anxiety Symptoms Scale (PASS-20), NPRS, Nine Hole Peg Test, Grip and Pinch Strength, ROM, and CRPS symptom Count	n = 10 (10/0)	43–57 (M49)	-	8–144 months (M61.4)	Pain rating and motor function significantly improved following intervention. BDI-II, CPAQ, and PASS-20 ratings improved post-intervention but did not reach statistical significance. 15-D measures were unchanged	73
Enax-Krumova et al. (2017)	Germany	Prospective cohort	Interdisciplinary treatment: 6-months treatment including pharmacotherapy and “positional therapy, sensorimotor training (occupational treatment, physiotherapy) as well as psychological treatment and, if indicated, invasive treatment by sympathetic blocks” (p. 97). No further information provided	NPRS, ROM, clinical signs and symptoms, Quantitative Sensory Testing battery	n = 24 (15/9)	25–70 (M52.4 ± 10.1)	Fracture, trauma, surgery	0.2–15.8 months (M3.5)	Significant increase in pressure pain threshold for all participants, however improvements in pressure hyperalgesia only occurred in treatment responders - there was no improvement in non-responders	65
Ergan et al. (2012)	Turkey	Single case study	Kinesiotherapy: Contrast bath, TENS, and ultrasound in combination with active-assisted ROM and stretching exercises daily for 20 days. No further information provided	Clinical signs and symptoms	1 M	66	Snake bite	>4 months	“The patient’s complaints decreased at the end of his hospital stay, but no significant improvement was achieved in his joint ROMs and the finding of demineralization in the X-rays persisted” (p. 70)	55
Foncelle et al. (2021)	France	Single case study	Sensory retraining: 7 prism adaptations in first week (2 left hand sessions per day for 3.5 days). 8 adaptations in second week (2 right hand sessions per day for 3.5 days plus 1 left hand session for 0.5 days)	VAS, visual straight ahead, manual straight ahead, open-loop pointing and line bisection	1 F	50	Surgery	>3 years	Some reduction in pain occurred over the course of the treatment, but this was not statistically significant. Spatial reference frames changed significantly only when activities were performed with the affected hand	53
Gaertner et al. (2018)	USA	Prospective cohort	Physical modality: Transcranial magnetic stimulation (TMS) compared between 1 or 5 consecutive sessions. Protocol described in text	Formal VAS, verbal NPRS	n = 9 (8/1)	27–71 (M47.8)	Fracture, surgery, soft tissue trauma	12–186 months (M67.3)	TMS treatment significantly reduced pain immediately following treatment and 1 week later. There was no difference in response at 1-week follow up between those who had received 1 compared to 5 treatments. Multiple adverse effects were additionally noted, primarily, headache and nausea	53
Glassey et al. (2013)	UK	Single case study	Interdisciplinary treatment: Physiotherapy 2x weekly for 11 months (GMI, desensitisation, functional use, passive ROM with sedation). Pharmacological interventions provided throughout the 11 months. Psychology input for 1 month to address coping with chronic pain	Clinical signs and symptoms	1 F	-	Soft tissue trauma	>6 days	The patient’s condition deteriorated and treatment was considered ineffectual. She requested an amputation and was referred for inpatient high-dose ketamine infusion. 11 months after her initial presentation, and three weeks following ketamine infusion, the patient presented with full resolution of signs and symptoms	58
Gutiérrez-Espinoza et al. (2019)	Chile	Prospective case series	Kinesiotherapy: Two 1 h sessions weekly for 6 weeks. Included 15 min of active exercises in whirlpool bath, joint mobilisations, and motor skill training exercises (grip, reverse dart-throwing, and scapular retraction)	VAS, grip, DASH, and Patient Rated Wrist Evaluation	n = 54 (46/8)	M 65.3 ± 3.9	Distal radius fracture	M 5.5 ± 1.2 weeks	Statistically significant improvements in wrist/hand and upper extremity function, statistically and clinically significant pain reduction, and non-significant increase in grip strength	58
Halicka et al. (2021)	UK	Randomised control trial	Sensory retraining: Prism adaptation compared to sham treatment. Treatment dose was one training session followed by 2 weeks of 2x daily home-based sessions (29 total)	CRPS Severity Score, NPRS, Brief Pain Inventory, Pain Detect Questionnaire, Bath CRPS Body Perception Disturbance Scale, Tampa Scale for Kinesiophobia, Profile of Mood States, Revised Life Orientation Test, Patient Centred Outcome Questionnaire, Patient Global Impression of Change Scale	n = 41 (36/5)	43.20–51.95 (M47.35) and 39.85–50.85 (M45.31)	Fracture, surgery, soft tissue trauma, laceration	48.57–75.91 (M61.71) and 34.72–68.56 (M51.25)	Overall, CRPS severity significantly decreased over time for both groups, but no difference in outcomes between prism adaptation and sham treatment	93
Hirakawa et al. (2020)	Japan	Single case study	Sensory retraining: 20 sessions of 3-staged “body shadows” over 48 days: 1) observe movements of the shadow of unaffected right hand and therapist’s left hand, 2) observe shadow of own left hand moving gently and progressing to touching shadow thrown by own face, 3) move both limbs while observing own body shadows	NPRS, Short Form McGill Pain Questionnaire neglect-like-symptoms	1 F	30s	Soft tissue trauma	>6 years	Pain reduction and improvements in neglect-like symptoms, and skin colour and swelling asymmetries	48
Hommer (2012)	USA	Single case study	Manual therapy: Chinese scalp acupuncture once a day for 3 consecutive days, followed by one session a week later. Method described in text	NPRS, clinical signs and symptoms	1 M	18	Fracture	>19 months	Treatment resulted in complete resolution of pain, colour, and temperature changes, except for remaining allodynia over the patient’s surgical scar	43
Kotsougiani-Fischer et al. (2020)	Germany	Retrospective longitudinal study	Interdisciplinary treatment: Input from physiotherapists and occupational therapists (ROM, strength, proprioception, desensitisation, mirror therapy, massage), psychologists (CBT and relaxation techniques), pain medicine specialists, and specialised hand surgeons	ROM, grip, pinch, VAS	n = 89 (34/55)	17–70 (M45.4±12.2)	Fracture, soft tissue trauma	Initial trauma to CRPS days M160.0 ± 183.0. CRPS and time to admission days M70.5±101.5	All outcomes showed statistical improvements, correlating with the duration of therapy; mean treatment duration was 53 days	73
Kulbida et al. (2019)	Germany	Single case study	Manual therapy: Five sessions of leech treatment every 6–7 weeks with the exception of a 14-weeks interval after the third session. Farm cultured Hirudo medicinalis were used, and usually two leeches were applied per session	NPRS, three quantitative sensory tests as per German Research Association for Neuropathic Pain, clinical signs and symptoms	1 M	54	Fracture	-	Remarkable improvement in pain, oedema, use of the hand, and skin and temperature asymmetries	60
Lagueux et al. (2012)	Canada	Prospective case series	Sensory retraining: Modified GMI, consisting of 4 phases at 1–3 weeks duration each, and 3 at-home 10-min sessions per day, 6 days a week. The 4 phases were hand laterality, imagined hand movements, mirror therapy with mobilization of the nonaffected hand, and mirror therapy with mobilization of both hands	Short Form McGill Pain Questionnaire (including VAS), DASH, grip, Patient Global Impression of Change	n = 7 (6/1)	34–57 (M45 ± 9.36)	Fracture, surgery, soft tissue trauma, laceration	-	Significant improvements in pain, grip, and impression of change. Improvements in upper limb function as measured by the DASH, but these were neither clinically or statistically significant	75
Lewis et al. (2021)	UK	Randomised control trial	Sensory retraining: Mediated Virtual Reality (VR) sessions once weekly for 4 weeks, with a final follow-up session 2 weeks later. As participants viewed their affected hand within the VR device, the operator digitally altered its appearance (for treatment group only) based on descriptions provided regarding preferred shape, size, and/or colour	Body perception disturbance, NPRS, 7-point Likert scale perceptual statement rankings (Schaefer et al. 2007)	n = 45 (29/16)	32–78 (M52 ± 13)	Fracture, soft tissue trauma, surgery, infection, spontaneous	4–264 months (M56 ± 54)	Body Perception Disturbance scores and pain intensity improved significantly compared to controls after a single exposure, however at follow up only pain reduction was maintained	85
Luković et al. (2012)	Serbia	Prospective longitudinal case series	Kinesiotherapy: Duration averaged 91.5 ± 42.16 days. Included education, interference current, magnetic therapy, graded activity, isometric exercises, gentle active ROM or active supported exercises, strength and mobility exercises	VAS, swelling, AROM, clinical signs and symptoms (skin colour and temperature)	n = 36 (26/10)	23–73 (M42.6 ± 14.6)	Fracture, soft tissue trauma, stroke	3–48 days (M19.6)	Measurements were summed to report that over the course of treatment statistically significant improvement in overall functional status occurred. Statistically significant reduction in pain, oedema, and skin asymmetries were also reported	58
Manal et al. (2021)	Egypt	Randomised control trial	Manual therapy: TAU (TENS, mirror therapy, and exercises) compared to TAU plus thoracic manipulation (TM) (Maitland screw technique grade V at the T3–T4 level), provided 3 days per week for 3 months	DASH and VAS	n = 30 (21/9)	40–60 (M53 and 50)	Surgery post fracture	10–18 weeks (M16 and 15)	Both groups had statistically significant reduction in both pain and function (DASH) post treatment. There was a significant difference between groups after treatment for DASH scores in favour of the TM group	63
Packham et al. (2017)	Switzerland	Retrospective cohort study	Sensory retraining: Somatosensory Rehabilitation Method, comprised of three elements - distant vibrotactile counter stimulation (provided 8x daily for <1min at home), application of therapeutic vibration (provided 1x weekly for 10 min in the clinic), and avoidance of pain-evoking touch stimuli	McGill Pain Questionnaire (French version), allodynic territory	n = 48 (34/14)	18–74 (M45.4 ± 13.4)	-	1–335 months (M31.2 ± 57.5)	Statistically significant reduction in pain and allodynia	88
Park et al. (2018)	South Korea	Single case study	Kinesiotherapy: 9 days of cryotherapy, TENS, contrast bath, gentle flexibility, and isometric strengthening exercises. No further information provided	VAS, clinical signs and symptoms	1 M	61	Injection	>7 days	Clinical signs and symptoms subsided rapidly after treatment, with disappearance of swelling and heat, significant pain reduction, and a return to previous function, with complete pain resolution 3 months after discharge	50
Ryan et al. (2017)	UK	Mixed methods feasibility study	Physical modality: TAU (physiotherapist-provided advice, education, exercise, CBT, motor imagery, hand laterality recognition training, desensitizing and hydrotherapy) compared to TAU plus TENS, self-administered for 90 min per day, daily for 3 weeks. Protocol described in text	DASH, Body Perceptual Disturbance, VAS, hand laterality perception	n=4	-	-	-	Recruitment for statistical analysis was below target, however qualitative findings suggested that some participants found the TENS unit easy to use and beneficial	75
Safaz et al. (2009)	Turkey	Single case study	Manual therapy: Manual lymphatic drainage therapy daily, 5 days a week, for 3 weeks. Technique described in text	Oedema, grip, clinical signs and symptoms	1 M	28	Fracture	>5 years	Dramatic reduction of oedema, and improvements in grip, AROM, and function	50
Sato et al. (2010)	Japan	Pilot case series	Sensory retraining: Virtual reality mirror visual feedback system sessions once weekly, no set duration. Participants reached, grasped, transferred, and placed objects seen on screen, while focusing on the movements of the virtual forearm (matching the affected side but controlled by the non-affected side)	VAS	n = 5 (4/1)	46–74 (M55.8)	Fracture, surgery	1–3 years (M2.2)	4 out of 5 participants achieved a greater than 50% pain reduction following treatment. Treatment duration differed for all participants	53
Schmid et al. (2016)	Switzerland	Prospective cohort	Sensory retraining: Sensory-motor training at home over 2 weeks. Involved pattern recognition with fingertips on specially-designed discs at different levels of difficulty, and three training modes (bi-manual, speed and memory training)	Pain Disability Index, Centre of Epidemiologic Studies Depression Scale Short Form, VAS	n = 10 (7/3)	40–73 (M58.4 ± 11.23)	-	2–264 (M42.7 ± 79.7)	Analyses found a statistically significant reduction in pain, associated with both the total time (minutes) and duration (days) spent training	68
Seyrek et al. (2014)	Turkey	Single case study	Interdisciplinary treatment: Physiotherapy input (TENS, contrast bath therapy for pain relief and desensitisation, stretching exercises) and occupational therapy input (functional and fine motor exercises). No further information provided	Clinical signs and symptoms (pain, swelling, limited ROM, skin colour changes, allodynia and hyperalgesia)	1 F	26	Vaccination	>5 days	“The patient’s symptoms started to resolve at the end of second week [sic] and significant improvement was observed at the end of 4 weeks of therapy” (p. 214)	43
Solcá et al. (2018)	Switzerland	Prospective cohort crossover	Sensory retraining: Single 1.5-h experiment measuring effect of heartbeat-enhanced immersive virtual reality. Participants were immersed in a virtual environment and shown a virtual depiction of their affected limb flashing in synchrony (or in asynchrony in the control condition) with their own heartbeat	VAS, grip, heart rate variability (HRV), virtual hand ownership (10 item questionnaire with likert scale), proprioceptive drift (ruler)	n = 24 (14/10)	23–73 (M49.7)	Soft tissue trauma, laceration, fracture, infection	1–22 months (M5.08)	The treatment resulted in statistical improvements in pain ratings and grip strength, and modulated HRV. There were non-significant changes to virtual hand ownership and proprioceptive drift	65
Strauss et al. (2021)	Germany	Randomised control trial crossover	Sensory retraining: Waiting compared to spending 2 weeks on each of 3 GMI phases (left/right judgements, imagined movement, and mirror therapy), with home exercises >10 min every waking hour over these 6 weeks. Further information in text	CSS, quick DASH, VAS, somatosensory testing with monofilaments, manual dexterity Roeder Manipulative Aptitude Test, spatial tactile resolution (two-point resolution (TPR) and the Grating Orientation Task)	n = 21 (17/4)	19–80 (M54.71 ± 14.13)	Fracture, surgery, soft tissue trauma	4–172 months (M58.24 ± 43.88)	Pain with movement improved significantly following GMI, but not after waiting. In addition, GMI, but not waiting, resulted in a decrease of clinical scores (CSS), increased usage of the affected hand side (DASH), and an improvement of motor function (Roeder) and spatiotactile performance (TPR)	78
Szynkowicz and Petrucci (2020)	USA	Single case study	Kinesiotherapy: Chiropractic treatment 2× weekly for 3 weeks, reducing in frequency for 36 appointments over 5 months. Included soft tissue mobilization, massage, and HEP (passive ROM, self-massage, active fist, functional grasp, and functional use of the hand as tolerated)	NPRS, clinical signs and symptoms	1 F	56	Fracture	>3 months	Increase in function and complete resolution of motor, trophic, sensory, and vasomotor signs after 5 months, with only mild oedema remaining in the proximal and distal interphalangeals	68
Van de meent et al. (2011)	The Netherlands	Prospective cohort	Kinesiotherapy: <6 one-hour sessions across <3 months. Included cognitive-behavioural approaches to motivate participants towards activity/participation-related goals, and progressive-loading HEP.	ROM, grip, DASH, VAS	n = 11 (9/3)	27–77 (M42.9)	Fracture, soft tissue trauma, surgery, infection, spontaneous	3–13 months (M6.54)	Statistically significant improvements for all measures	75
Vas and pai (2016)	India	Single case study	Manual therapy: Ultrasound guided dry needling targeting muscles of the right upper extremity. Provided 2× weekly under sedation for 3 sessions, progressing to 2× weekly without sedation for a month, then tapering off	painDETECT, DASH, Patient Health Questionnaire	1 F	24	Overuse	>3 years	Improvement in pain and function scores	30
Vas et al. (2016)	India	Retrospective cohort study	Manual therapy: 3x weekly ultrasound-guided-dry-needling of the neck and upper extremity, for <45–60 days. Programme included electrical modalities, myofascial release, muscle stretching, and, after resolution of some pain, strengthening exercises	NPRS, ROM, grip, DASH	n = 44 (24/20)	24–80 (M51 ± 18.8)	Fracture, surgery, soft tissue trauma, spontaneous	3–9 months	Sonographic changes and improvement, if not complete resolution, of signs and symptoms measured by outcome measures for all patients	43

TENS: Transcutaneous Electrical Nerve Stimulation; TAU: Treatment as Usual; VAS: Visual Analogue Scale; DASH: Disabilities of the Arm, Shoulder, and Hand; NPRS: Numerical Pain Rating Scale; ROM: Range of Movement; CBT: Cognitive Behavioural Therapy; ACT: Acceptance and Commitment Therapy; GMI: Graded Motor Imagery; HEP: Home Exercise Program.

All studies (n = 38) measured pain intensity, and most (n = 24) measured objective physical parameters such as pinch or grip strength, ROM, oedema, or perceptual abilities (e.g. laterality/pointing accuracy, and tactile detection, discrimination and dexterity) (see Table 1). Several case studies did not describe the use of any measurement tools but mentioned clinical signs and symptoms present before versus after treatment.^36,39,41,47 13 studies measured patient-rated function. The Disabilities of the Arm, Shoulder and Hand (DASH) questionnaire was used in ten.^{49,50,56,57,60,65–67,72,73} Alternative measures of function included the Sickness Impact Profile, ⁵⁸ Patient-Rated Wrist Evaluation, ⁶⁶ Patient Centred Outcome Questionnaire and Patient Global Impression of Change Scale, ⁵⁴ and the Pain Disability Index. ⁶³ Four studies measured psychological factors.^38,54,60,63 These all used standardised self-report measures and measured kinesiophobia, ⁵⁴ mood,^54,60,63 quality of life, ⁶⁰ and optimism, ⁵⁴ as well as pain-related self-efficacy, ³⁸ acceptance and anxiety. ⁶⁰

Quality ratings of included studies ranged from 30% to 93%, with a median of 60%. Eight studies scored 75% or more. 19 studies scored between 51 and 74%, and 11 studies scored 50% or less. Common methodological limitations included failure to report information required to determine quality, lacking sample size power estimates, questionable ethical practices, and drawing inferences inconsistent with the strength of the data. Overall, the populations tested were heterogeneous, with CRPS duration ranging significantly, sometimes from a few weeks to several decades within a single study. Many studies lacked rationale for the intervention used or did not describe the setting, personnel, or process involved in administering the intervention. Outcome measures frequently lacked justification, and many were non-standardised. The full characteristics of the included studies and their CCAT scores are described in Table 1.

Discussion

This review synthesised and critically appraised the existing evidence for non-pharmacological management of upper limb CRPS. Five main intervention categories were identified: sensory retraining, kinesiotherapy, manual therapies, physical modalities, and interdisciplinary treatment. The quality of included evidence was generally poor, indicating a need for careful consideration of treatment options against an individual’s goals and presentation.

Sensory retraining

The largest category of interventions addressed common sensory symptoms of CRPS, including allodynia, body perception disturbances, and changes to sensorimotor integration. These included Prism Adaptation,^37,40,54,58 Graded Motor Imagery,^38,57,67 Virtual Reality,^55,64,69 graded somatosensory stimulation,^63,71 and body shadows. ⁴²

The four studies of Prism Adaptation (PA) ranged in quality from 40-93% with a median of 56.5%.^37,40,54,58 PA is a sensorimotor training technique thought to influence pain and “neglect-like” symptoms of CRPS by increasing attention to the CRPS-affected side, thus restoring normal sensorimotor integration. ³⁷ Some researchers have proposed that people with CRPS show stroke-like sensorimotor and spatial neglect toward the affected limb.^74,75 However, recently researchers have challenged this assumption, suggesting that people with CRPS may instead demonstrate hyper-attention toward the affected limb, and a greater propensity for sensorimotor realignment.^10,58,76 The highest quality study in this review concluded that there was no difference in outcomes between PA and sham treatment and that immediate, short-term reduction in pain post-intervention in both groups could be explained by placebo effects or general benefits of moving the affected arm. ⁵⁴ While these effects may occur in many prescribed exercises and patient-therapist interactions, ⁷⁷ most studies in this review did not discuss these as potential confounding factors or employ them for their therapeutic benefits.

The three studies of Graded Motor Imagery (GMI) ranged in quality from 50-78%, with a median of 75%.^38,57,67 GMI consists of three sequential phases, with practice of phase-specific tasks occurring three times every waking hour. ⁷⁸ Two studies in this review modified this protocol, including reducing the recommended task practice frequency,^38,67 with one stating that this was for reasons of practicality. ³⁸ Indeed, given the discrepancy between recommended and actual exercise practice ⁷⁹ the feasibility of translating GMI to the real-world may require further consideration. ⁸⁰ All studies provided a different rationale for the intervention. Re-establishing the pain-free relationship between sensory feedback and motor execution was mentioned in two,^38,67 of which one ⁶⁷ additionally referred to the concept that cortical changes may underpin pain, and thus pain may decrease by reorganizing the cortex through sequential activation. ⁸¹ However, insufficient evidence exists for the theory that smudging or shrinking in the cortical representation of the limb with CRPS influences the incidence or severity of sensorimotor symptoms and pain, and this has recently been challenged as an implicit assumption underlying CRPS treatment.^13,82 The mechanisms through which GMI may help to dissociate movement from pain may instead be better attributed to reducing conditioned fear and subsequent avoidance responses, and associated physiological arousal, as described by the highest quality study of GMI in this review. ⁵⁷

Three studies used Virtual Reality (VR) as a tool for their primary intervention.^55,64,69 The quality of these studies ranged from 53% to 85%, with a median of 65%. Studies differed significantly in their methods and purpose (see Table 1). Outcomes were suggestive of short-term pain reduction, however, long-term effects are unknown. In studies of other conditions, immersive, interactive VR is a useful tool to prompt sensorimotor changes without active use of the affected limb, ⁸³ and to influence nociceptive regulation via distraction. ⁸⁴ More studies exploring its use and clinical feasibility with CRPS are required.

Two studies used graded somatosensory stimulation techniques, with quality scores of 68% and 88% respectively.^63,71 The theoretical rationale for and methods of intervention differed substantially (see Table 1). One aimed to reduce pain and disability by gradually increasing the challenge of a home-based tactile recognition training, performed with the CRPS-affected hand. ⁶³ The second aimed to reduce allodynia by gradually increasing the intensity of sensory stimulation to a related area of normal sensation. ⁷¹ Both studies deliberately avoided causing pain while gradually increasing sensory input. Similarly, hand therapists treating CRPS tend towards a pain-contingent approach toward tactile desensitisation. ⁸⁵ Whether to avoid or provoke pain during intervention for CRPS remains unclear, as avoiding pain provocation may lead to greater protective behaviours and learned disuse, while inducing pain may provoke flare ups and thus increase disability.^86–88 The term “desensitisation” itself lacks clarity, as this may refer to pain-contingent approaches such as those described, to consciously attending to stimulation of the painful area to re-educate to normal sensation, or to flooding with intense stimulation to produce accommodation to the stimulus.^71,89 While in this review a pain-contingent approach appears to have had a short-term effect on pain, more studies are required to investigate the long-term results and impact on function, and thus provide guidance on when and how desensitisation techniques should be applied and progressed.

Finally, one study conducted a stepwise intervention using the shadows thrown by the participant’s own body to address their self-disgust, pain, and neglect-like symptoms, with a quality score of 48%. ⁴² Similar to studies of PA, the authors referenced the now challenged theory that people with CRPS exhibit stroke-like neglect, which in turn increases pain. Their stepwise intervention may rather have primarily contributed to a lessening of conditioned fear and avoidance responses as described by the highest quality study of GMI. ⁵⁷ As CRPS prognosis is associated with psychological factors related to threat perception, activity avoidance, and protectiveness toward the affected limb, ⁹⁰ further consideration of such factors may be indicated in treatment and research around body schema changes and CRPS.

Kinesiotherapy

There was significant heterogeneity across studies of kinesiotherapy, making it difficult to disentangle the most effective element of treatment. Quality scores ranged from 35% to 75%, with a median of 58%. Kinesiotherapy was provided by physiotherapists, chiropractors, and occupational therapists. All programmes included ROM exercises of the hand. Programmes also frequently incorporated physical modalities, including contrast baths, whirlpool baths, and ultrasound. It was often unclear whether these modalities were used as an adjunct to ROM exercises or whether these were intended to provide therapeutic benefits in isolation. Studies that purely used modalities and ROM scored 35% and 55%.^36,39 Studies that used modalities, ROM, and strengthening scored 50% and 58%.^45,66 There was most evidence for studies that used modalities, ROM, strengthening, and graded functional activity, with scores of 58%, 68%, and 75%.^48,65,68 Incorporating occupation as exercise may be a beneficial means of enhancing movement and addressing motor impairment after upper limb injury,^91,92 and a recent review of hand therapists found that ROM exercises and functional activity practice are the most frequently prescribed physical interventions with CRPS. ⁸⁵ This indicates a need for further robust research exploring the value of movement and occupation-based interventions in kinesiotherapy for CRPS.

Manual therapies

In this review manual therapies refer to therapies in which individuals were recipients of hands-on intervention. These included manual lymphatic drainage therapy,^46,53 acupuncture, ⁴³ dry needling,^49,73 and leech therapy. ⁴⁴ Quality scores of the seven studies ranged from 30% to 63%, with a median of 50%.

Manual therapies were used to address pain and limb signs, including oedema, colour/temperature asymmetry, and reduced ROM. Studies often used non-standardised measures to monitor outcomes,^43,44,46 and very limited rationale was provided for intervention, particularly for acupuncture ⁴³ and dry needling.^49,73 While manual therapies provided in isolation may help with immediate pain relief, research is lacking regarding their effect on CRPS. Manual therapies tend to be discouraged as a primary focus in chronic pain management because they may reinforce fear avoidance and feelings of helplessness. ⁹³ Instead, minimal use of passive treatments and transitioning toward an active model of care is recommended. ⁹³ Given that signs and symptoms of autonomic dysfunction are severe to the point of differentiating CRPS from other chronic pain conditions, accurately measuring these in future studies of manual therapies and CRPS is important, to more reliably indicate whether an intervention has been effective.

Physical modalities

Six studies reported the results of physical modalities alone, with critical appraisal scores ranging from 38% to 78%, and a median of 64%. Modalities included Bio-Electro-Magnetic-Energy-Regulation therapy, ⁵⁰ neuromuscular electrical stimulation, ⁵¹ extremity whirlpool bath, ⁵¹ low level laser therapy, ⁵² bipolar interferential current therapy, ⁵² transcranial magnetic stimulation, ⁶² and Transcutaneous Electrical Nerve Stimulation (TENS).^59,72 The treatment protocols of studies were diverse, offering limited guidance regarding optimal intensity, duration, frequency, and timing. Outcomes were similarly heterogeneous. In one study, participants found a TENS machine very easy to use, and mostly beneficial, even if only as a distraction. ⁷² However, the study using transcranial magnetic stimulation reported significant adverse events, primarily headache and nausea, ⁶² while another study noted impaired sensation after applying low-frequency stimulation (1 Hz). ⁵⁹ Only one study measured function as an outcome. ⁵⁰ Consistent with our findings, a recent scoping review of physical modalities for CRPS treatment found that ethical considerations are often lacking or not described, functional outcomes are rarely measured, and methodological quality is generally poor. ⁹⁴ While future research into physical modality treatment protocols for CRPS is required, further ethical considerations, and measurement of functional outcomes in addition to pain intensity, is recommended.

Interdisciplinary treatment

Interdisciplinary treatment approaches are globally recognised as best practice for CRPS management. ⁸⁹ However, only five studies in this review assessed the results of such approaches,^{41,47,60,61,70} with critical appraisal scores ranging from 43% to 73%, and a median of 65%. All programmes included pharmacological and physiotherapy treatment. Occupational therapy treatment was provided in three,^47,61,70 and psychology input in four.^41,60,61,70 While no relationship has been identified between psychological factors and CRPS onset, ⁹⁵ as CRPS pain intensity may be impacted by psychological distress, a vicious cycle can occur where pain, distress, and disability maintain each other. ⁹⁶ Assessing and treating pain-related beliefs, behaviours, and mood-disorders is therefore a critical aspect of disability-reducing treatment for CRPS; ⁹⁷ however, only one study of interdisciplinary treatment assessed psychological factors. ⁶⁰ Similarly, across the review, there was limited consideration of personal factors, activity, participation, or environmental factors, with a greater focus on physical impairments (such as pain and ROM) at the level of body function and structures. Increased use of the common language offered by The International Classification of Function, Disability and Health (ICF) ⁹⁸ to guide assessment and rehabilitation has been recommended for other conditions of the upper limb, ⁹⁹ and, indeed, one study in our review reported using the ICF to guide interdisciplinary treatment, however, exactly how this was used was unclear. ⁷⁰ The Core Outcome Measurement set for complex regional PAin syndrome Clinical sTudies (COMPACT) ¹⁰⁰ considers similar domains (such as participation and physical, emotional and psychological function), however, only five studies across this review used outcome measures recommended by COMPACT.^{38,42,54,57,67} As one barrier to CRPS management is the complexity of the condition, with multiple and changing diagnostic criteria and proposed treatment, ¹⁰¹ further use of existing tools such as ICF and COMPACT is suggested to guide consistent biopsychosocial interdisciplinary practice that captures the multidimensional nature of the CRPS experience.

Limitations

There are limitations inherent in the integrative review methodology used. Including studies of poor quality may lead to inaccurate conclusions; conversely, excluding these may bias the synthesis and reduce diversity. Therefore, we have attempted to differentiate between studies of high and low quality and to weight conclusions accordingly.

Further, using a single tool for critical appraisal across multiple designs may reduce specificity toward assessing bias implicit in individual designs. ¹⁰² However, in the spirit of the integrative review methodology, we agreed that the appraisal of studies of diverse designs should be based on the quality of each study within its methodological domain, rather than against a pre-determined hierarchy. ¹⁰³

While the integrative review methodology was deliberately chosen to enable inclusion and comparison between quantitative and qualitative studies, no qualitative studies met the criteria for inclusion in this review. We excluded qualitative studies primarily because they did not specify the diagnostic criteria used, except for one study, in which CRPS location was not specified. ¹⁰⁴

Several quantitative studies were also excluded because they did not state the diagnostic criteria used, or used older, less specific criteria. The strict adherence to Budapest criteria could be seen as excluding relevant studies, as other reviews of CRPS have not taken this approach.^8,105 However, the Budapest criteria have been the international gold standard for CRPS diagnosis for almost two decades. ¹² Given the diversity in studies of this population, adherence to recommended diagnostic criteria is a minimum expectation for future research, with the use of the CRPS Severity Score to confirm a diagnosis and further standardise cohorts. ¹² This review lends support to the efforts of organisations such as the International Research Consortium (IRC) for CRPS, which aim to encourage collaboration across multiple sites in order to increase sample size and improve rigour and quality of CRPS-related studies. ¹⁰⁶

Finally, inclusion of English-only studies may be seen as a limitation to this review. This was due to the lack of time and resources available for translation. However, after abstract review, only three non-English articles were found that would have otherwise gone to full text review. This suggests that no large volume of non-English data has been missed and so is unlikely to have biased this review.

Conclusion

CRPS most frequently affects the upper limb and is highly debilitating. Early provision of appropriate non-pharmacological therapies is a critical aspect of CRPS management, influencing long-term outcomes. Movement, desensitisation, and graded functional activity remain the mainstays of non-pharmacological interventions and are supported by the results of this review. However, the methodological quality of studies of treatment approaches for upper limb CRPS is poor overall. Despite the impact of CRPS on function and wellbeing, few studies measured function or psychological factors or considered patient perspectives on intervention feasibility or efficacy. Studies drew on a wide range of theories regarding pathomechanisms of CRPS to justify their interventions if a rationale was indeed provided. While further research into the pathomechanisms of CRPS is required, in the interim, given the multi-dimensional nature of the CRPS experience, we suggest prioritising people over mechanisms by assessing and treating the whole person in their context, with a clear rationale for chosen interventions and a link to individual goals. Further robust research is required to determine which aspects of treatment have the greatest influence on which symptoms, when and how these should be introduced and progressed, and whether these successfully address people’s treatment priorities.

ORCID iD

Grace S Griffiths https://orcid.org/0000-0003-3498-2383