Neck pain is a highly prevalent condition and a leading cause of disability worldwide,66 responsible for enormous economic burden attributable to both health care and indirect expenses.42 For many people, neck pain resolves quickly, whereas others report recurrent flare-ups,1 and at least 50% of people report pain and disability 1 year after the precipitating event.14–16Neck pain can arise after a traumatic injury (eg, a road traffic crash—whiplash-associated disorder [WAD]) or can be insidious in onset (nontraumatic neck pain [NTNP]). Current guideline-recommended treatments for neck pain, such as education, advice, and exercise, demonstrate only modest effects.22,30,78
Psychological factors such as low pain self-efficacy, stress, pain catastrophising, depression, and anxiety are associated with poorer health outcomes in patients with neck pain of both traumatic and nontraumatic onset.10,79 Clinical guidelines recommend the evaluation and treatment of psychological factors in the management of patients with neck pain.9,21,77 Consequently, psychological interventions, such as cognitive behavioural therapy (CBT), are commonly used for neck pain and musculoskeletal pain conditions more broadly.54
Although psychological interventions can benefit patients with musculoskeletal pain conditions,91 barriers exist for implementation of such treatments. There can be challenges associated with limited availability of pain psychology services in some settings,5,40 as well as potential financial5,26 and time costs24 to the patient. Nonetheless, the need to address both psychological and physical factors in the management of musculoskeletal pain is recognised and valued by patients41,72 and clinicians.43 To mitigate these barriers, nonpsychologist practitioners (eg, physiotherapists) have been used to deliver psychological interventions for patients with musculoskeletal pain conditions.11,19,71,80 Physiotherapists are well-placed to deliver integrated psychological and physical interventions as they commonly provide care to patients with neck pain in community primary care and hospital outpatient settings.65
Previous systematic reviews have evaluated the effectiveness of physiotherapist-delivered psychological interventions for musculoskeletal pain,71,93 post spine, hip, or knee surgery,20 and low back pain.35,38,84 Meta-analyses revealed mostly small-to-moderate effects35,71 on pain and disability in favour of psychological interventions compared with physiotherapy alone (typically exercise, advice, or manual therapy). One review included a WAD/neck pain subgroup71 and found no effects on pain and disability. However, their meta-analysis did not distinguish between acute and chronic neck pain nor consider the aetiology of neck pain. Patients with WAD report higher pain and disability,62 more psychological distress,62 greater hyperalgesia70 and hypoesthesia,17 and poorer outcomes at follow-up compared with those with nontraumatic neck pain,4 suggesting that they may respond differently to psychological interventions. Similarly, there may be a differential effect of psychological interventions depending on the stage of the condition, acute vs chronic. Consequently, nuanced evaluation of effectiveness of psychological interventions incorporating these critical clinical distinctions is necessary.
Physiotherapists do not commonly receive training in psychological care at the preprofessional level,29 making implementation of this approach problematic.6 Authors of recent reviews have argued that descriptions of interventions in the reports of trial results in low back pain are inadequate to achieve adequate training and replication in the clinical setting.35,73 Whether this is also the case for neck pain is not known.
The aims of this systematic review are to (1) determine the effectiveness of psychological interventions delivered by physiotherapists on neck pain, disability, and quality of life in people with acute or chronic WAD or NTNP; (2) determine if interventions are described sufficiently to enable replication by physiotherapists in the clinical setting.
2. Materials and methods 2.1. Protocol registration and study designThe protocol for this systematic review was registered on International Prospective Register of Systematic Reviews (PROSPERO) database (CRD42021242180) and is available at https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42021242180. We report the following deviations from the original protocol: We used amended definitions of short-, medium-, and long-term follow-up to more effectively represent the follow-up time points reported in the included studies. We also added quality of life outcomes and assessed the description of trial interventions as relevant for replication in a clinical setting. This systematic review is reported in accordance with the guidelines detailed in the PRISMA statement.57
2.2. Eligibility criteriaRandomized controlled trials (RCTs) published in English were considered for inclusion, comprising participants (aged ≥18 years) with a diagnosis of acute (<3-month duration) or chronic (≥3-month duration)83 WAD (grades 1, 2 or 3)76 or nontraumatic neck pain. Participants could not have a specific cervical spine pathology, such as bony injury (fracture or dislocation [WAD 4]) or rheumatoid arthritis. Included RCTs must have investigated effectiveness of a psychological intervention delivered by physiotherapists only, compared with physiotherapy alone or no treatment. Physiotherapists delivering the psychological interventions must have undergone training by a psychologist or other relevant health professional with expertise in psychological interventions. When such training was not reported, details were sought by correspondence with the authors. Psychological interventions were defined as any approach that incorporated the following: cognitive behavioural strategies, acceptance-based interventions, relaxation, mindfulness, hypnosis, coping skills, problem-solving, stress management, and depression interventions or similar,71 in combination with physiotherapy. Studies of treatment delivery in individual or group settings were eligible. Studies were ineligible if the psychological treatment was delivered by other health professionals.
2.3. Search strategy and information sourcesA search strategy was developed for the following databases: CINAHL, EMBASE, PubMed, Cochrane. Search terms were related to physiotherapy, neck pain, and psychological interventions, based on previous relevant systematic reviews23,28,71 (Supplementary File 1, available at https://links.lww.com/PR9/A191). Databases were searched up to November 4, 2021. Identified studies were exported to EndNote 20 (Clarivate Analytics, New York City, NY) and uploaded to Covidence. Duplicates were removed using the “remove duplicates” function in Endnote and then again in Covidence. Forward citation searching was carried out on the included studies using Web of Science and reference lists of included studies were hand searched for potentially eligible studies.
2.4. Selection of studiesUsing Covidence, titles and abstracts were initially screened, followed by the full texts, by 2 independent reviewers (D.E., J.F., H. Martine, and H. Mohamed—randomly allocated between review team). Any disagreements were settled by consensus discussion or inclusion of a third reviewer (S.F.F.) if needed.
2.5. Data items soughtThe primary outcomes were pain (eg, intensity rated on a visual analogue scale [VAS]) and disability (eg, Neck Disability Index [NDI]). Secondary outcomes were quality of life measures (eg, Short-Form 36 [SF-36]). Outcomes were collected at the following time points: short term (immediately posttreatment), medium term (3–6 months postbaseline), and long term (≥12 months postbaseline).
2.6. Data extractionData were extracted from each included study by one author (S.F.F.) and checked by another author (J.L.), using a standard form (Excel; Microsoft, Redmond, WA) prepared by the research team. Descriptive data were extracted regarding study design and setting, sample size, demographics of participants, details of the psychological and control interventions, follow-up time points, clinical outcome measures and main results, and limitations and conclusions as presented in the published articles. Supplementary materials and published protocols or other studies were accessed for additional details as required. If data were missing or not clear, authors were emailed to request data, or if applicable, data were used from an earlier related systematic review with meta-analysis undertaken by our research team.71 If authors did not respond to our request for data, their article was not excluded from the review. Rather, findings were reported descriptively for outcomes with missing or unclear summary data. The Template for Intervention Description and Replication (TIDieR)39 checklist was used to assess descriptions of psychological interventions (such as name, rationale, materials, delivery, providers, training, dosage, and fidelity). Where a study contained more than one group receiving an eligible psychological intervention, data from these groups were pooled to create one psychological intervention group using Review Manager 5.4 (The Nordic Cochrane Centre, Copenhagen, Denmark). If group summary statistics were presented as median (IQR), these data were converted to mean (SD) using the calculator tool provided by Wan et al.,87 as described in the Cochrane Handbook.36 If group SD for an outcome at a follow-up time point could not be calculated or acquired for a study, SD was imputed using the baseline SD for that group.
2.7. Risk of bias of individual studiesMethodologic quality of included studies was assessed using the Cochrane Risk of Bias Tool 2.0.81 Each included study was independently assessed for bias by 2 reviewers (S.F.F., J.L.). Briefly, the Tool assesses risk of bias of an individual study with respect to 5 domains: (1) the randomization process; (2) deviations from the intended interventions (effect of assignment to intervention); (3) missing outcome data; (4) measurement of the outcome; and (5) selection of the reported result. Bias assessments were considered specific to each outcome of interest reported in the study81—that is, pain, disability, and quality of life. Ratings of overall methodological quality were considered as “low risk,” “some concerns,” or “high risk” of bias. Disagreements were resolved through consensus discussion.
2.8. Effect measures and synthesis of resultsRandom-effects meta-analysis was performed for studies reporting data appropriate to be pooled, using Review Manager 5.4. Effect measures were calculated as standardised mean difference (SMD) for continuous data (eg, pain intensity VAS, NDI, SF-36), with SMD = 0.2, 0.5, and 0.8 considered small, medium, and large effects, respectively.18 The minimum number of studies reporting comparable data for meta-analysis was 2.25 Meta-analysis findings were presented using forest plots. Statistical significance was set at P < 0.05.
2.9. Risk of bias across studies and certainty assessmentFor each meta-analysis comparison at each time point for pain, disability, and quality of life outcomes, overall certainty of evidence was assessed using the Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) approach. Briefly, certainty of evidence began as “high” for each meta-analysis comparison comprising RCTs (ie, all comparisons). Each comparison was then considered with respect to 5 domains7: risk of bias, inconsistency of results, indirectness, imprecision, and other applicable factors (eg, publication/reporting bias). The certainty of evidence rating was then reduced by one classification level for each domain not met (ie, if one or more of the following criteria are applicable)67,69: (1) >25% of participants were from studies with high risk of bias; (2) there was substantial heterogeneity (I2 > 50%7); (3) >50% of participants fell outside target group (ie, a general population with neck pain); (4) total sample comprised <400 participants for continuous variables and <300 events for binary variable; (5) other (eg, publication/reporting bias, assessed using funnel plots when ≥10 studies formed a comparison56). Conversely, certainty of evidence was raised one classification level in the case of a large effect size (ie, ≥0.818).7 Subsequently, a rating of overall certainty of evidence was determined as high, moderate, low, or very low, interpreted as per the study by Balshem et al.7
3. Results 3.1. Study selectionThe electronic search identified 932 records. After removal of duplicates (n = 298), 634 records were screened at the title and abstract stage, 47 of which were then screened as full texts (Fig. 1). Of the full-text articles assessed for eligibility, 29 were excluded. The most common reasons for exclusion were that the psychological treatment was not delivered by a physiotherapist (n = 6) and articles describing secondary analyses of included trials (n = 6). The remaining 18 articles8,12,31,32,34,45,47–49,51–53,59,75,80,82,86,89 (detailing 14 RCTs8,12,31,45,49,51–53,59,75,80,82,86,89) fulfilled the eligibility criteria and were included in the review. Neither forward citation searching nor hand searching reference lists of included studies yielded additional articles. Agreement between reviewers was 85% to 97% for title and abstract screening and 60% to 100% for full-text screening. One RCT was reported in 2 separate published articles,8,47 containing pain47 and disability8 results for slightly different samples (10 of 47 patients differed). We therefore use the article by Beltran-Alacreu et al.8 as the primary citation for this RCT and cite López-de-Uralde-Villanueva et al.47 when referring to results specific to that published report.
Figure 1.:Flowchart of studies through review. RCT, randomized controlled trial; Rx, intervention.
3.2. Study characteristicsCharacteristics of the included RCTs can be found in Table 1. Four trials included patients with acute WAD,12,45,80,89 2 patients with chronic WAD,49,75 and 5 included patients with chronic nontraumatic neck pain.8,31,53,82,86 Three trials included patients with subacute59 or mixed51,52 duration neck pain, so we report these results separately to the trials in acute and chronic neck pain. Sample sizes ranged from 2889 to 599,45 with a total of 2028 patients comprising the review. Psychological treatments are detailed in Table 1. All trials used cognitive behavioural techniques in some form, such as assessing and challenging unhelpful thoughts and beliefs,8,12,31,45,49,51–53,59,75,80,82,86,89 problem-solving,51,80,82 goal setting,12,45,49,51,59,75,82 relaxation,8,31,45,49,75,80,89 or graded activity.8,12,53,59,86 Control treatments included advice,12,45,51–53,59,80,82,89 exercise,31,49,51–53,59,75,80,82,86,89 manual therapy,8,31,52,53,59,89 and electrophysical agents.31,52,75,89
Table 1 - Characteristics of included studies. StudyIQR, interquartile range; N/A, not applicable; NDI, Neck Disability Index; NDPS, Neck Pain and Disability Scale; NPQ, Northwick Park Neck Pain Questionnaire; NRS, numerical rating scale; NTNP, nontraumatic neck pain; PDI, Pain Disability Index; SF-12, Short-Form 12; SF-36, Short-Form 36; VAS, visual analogue scale; WAD, whiplash-associated disorder.
Pain was assessed by a VAS or numerical rating scale (NRS) in 11 trials.12,31,47,49,53,59,75,80,82,86,89 Disability was assessed by the NDI,8,31,45,49,59,80,86,89 Pain Disability Index (PDI),12,75 Northwick Park Neck Pain Questionnaire (NPQ),51,52,82 or Neck Pain and Disability Scale (NPDS)53 in 14 trials. Quality of life was assessed in 8 trials using the SF-36,52,53,59,80 SF-12,45,51 or EQ-5D.86,89 Ten trials8,12,31,49,51,53,59,75,80,86 included short-term follow-up (immediately posttreatment), 12 trials8,12,31,45,49,51,52,75,80,82,86,89 featured medium-term follow-up (3–6 months), and 10 trials12,31,45,49,51–53,59,80,86 reported long-term follow up (≥12 months).
3.3. Risk of bias of individual studiesDetails of risk of bias assessment of individual studies are presented in Figure 2. One study89 was assessed overall as high risk of bias across all outcomes, whereas 3 trials were assessed as high risk of bias for certain outcome measures—disability49 and quality of life59,86—because of incomplete reporting of multiple planned measures. The remaining trials were assessed as having “some concerns” overall. The primary reason for assessing studies as “some concerns” was because of inability to blind participants or treating physiotherapists to participant group allocations, meaning that the outcome assessors were not blinded (ie, pain and disability outcomes were self-reported and participants were typically not [able to be] blinded to treatment allocation). Agreement between reviewers in risk of bias assessment items was 91%. Disagreements were mostly related to domain 2 (risk of bias due to deviations from the intended interventions [effect of assignment to intervention]) and domain 3 (missing outcome data). All disagreements were resolved through consensus.
Figure 2.: Risk of bias of individual studies, assessed using the Cochrane Risk of Bias 2 tool. Traffic-light plot prepared using the robvis tool.503.4. Results of synthesesTwo trials8,12 featured 2 experimental groups with eligible psychological interventions, so data for the 2 experimental groups were pooled. For 3 studies,8,12,89 data were converted from median (IQR) to mean (SD) using the tool provided by Wan et al.87 The author of one study provided follow-up SD for pain and disability data,82 and in 2 trials,49,59 study data were sourced from our centre's previously published systematic review with meta-analysis.71 Summary data were requested from the corresponding author of one study for SF-36 but no response was received—this study did, however, have summary data available for pain and disability. In 2 older studies,
留言 (0)