The Effectiveness of Conservative Management for Acute Whiplash Associated Disorder (WAD) II: A Systematic Review and Meta-Analysis of Randomised Controlled Trials

Objective To evaluate the effectiveness of conservative management (except drug therapy) for acute Whiplash Associated Disorder (WAD) II. Design Systematic review and meta-analysis of Randomised Controlled Trials (RCTs) using a pre-defined protocol. Two independent reviewers searched information sources, decided eligibility of studies, and assessed risk of bias (RoB) of included trials. Data were extracted by one reviewer and checked by the other. A third reviewer mediated any disagreements throughout. Qualitative trial and RoB data were summarised descriptively. Quantitative syntheses were conducted across trials for comparable interventions, outcome measures and assessment points. Meta-analyses compared effect sizes with random effects, using STATA version 12. Data Sources PEDro, Medline, Embase, AMED, CINAHL, PsycINFO, and Cochrane Library with manual searching in key journals, reference lists, British National Bibliography for Report Literature, Center for International Rehabilitation Research Information & Exchange, and National Technical Information Service were searched from inception to 15th April 2015. Active researchers in the field were contacted to determine relevant studies. Eligibility Criteria for Selecting Studies RCTs evaluating acute (<4 weeks) WADII, any conservative intervention, with outcome measures important to the International Classification of Function, Disability and Health. Results Fifteen RCTs all assessed as high RoB (n=1676 participants) across 9 countries were included. Meta-analyses enabled 4 intervention comparisons: conservative versus standard/control, active versus passive, behavioural versus standard/control, and early versus late. Conservative intervention was more effective for pain reduction at 6 months (95%CI: -20.14 to -3.38) and 1-3 years (-25.44 to -3.19), and improvement in cervical mobility in the horizontal plane at <3 months (0.43 to 5.60) compared with standard/control intervention. Active intervention was effective for pain alleviation at 6 months (-17.19 to -3.23) and 1-3 years (-26.39 to -10.08) compared with passive intervention. Behavioural intervention was more effective than standard/control intervention for pain reduction at 6 months (-15.37 to -1.55), and improvement in cervical movement in the coronal (0.93 to 4.38) and horizontal planes at 3-6 months (0.43 to 5.46). For early (<4 days) versus late (>10 days) interventions, there were no statistically significant differences in all outcome measures between interventions at any time. Conclusions Conservative and active interventions may be useful for pain reduction in patients with acute WADII. Additionally, cervical horizontal mobility could be improved by conservative intervention. The employment of a behavioural intervention (e.g. act-as-usual, education and self-care including regularly exercise) could have benefits for pain reduction and improvement in cervical movement in the coronal and horizontal planes. The evidence was evaluated as low/very low level according to the Grading of Recommendations Assessment, Development and Evaluation system.


Introduction
Whiplash Associated Disorder (WAD) is a consequence of whiplash injury caused by rapid acceleration-deceleration of the head and neck, leading to bony and soft tissue injuries. [1] Road traffic accidents are the most common cause of whiplash. [2] Over the past 20 years, the incidence of traffic related whiplash has risen in most western countries. [3] Prevalence has been reported as 3/1,000 people in North America and Western Europe, [3] with 300,000 individuals experiencing WAD annually in the UK. [4] 40%-60% of WAD patients develop chronicity [5][6][7][8][9][10][11] with approximately 30% of patients experiencing moderate to severe pain and disability. [12] Systematic reviews report limited effectiveness of chronic WAD management. [13][14][15][16] Consequently, effective intervention in the acute stage is required to prevent chronicity.
WAD contributes to a substantial economic burden throughout the industrialised world. Increased direct and indirect costs have been reported, including health care costs, reduced work productivity, lost earning capacity, higher socioeconomic costs and time contributed by caregivers. [17,18] The annual economic cost related to WAD is estimated as $3.9 billion in the US [19] and €10 billion in Europe. [20] Insurance costs are also high in the western world, [3,[21][22][23][24][25] with the UK described as the 'whiplash capital of Europe' by the Association of British Insurers, who estimate that one person in 140 claims for whiplash injury annually. [24] In the UK, the cost of claims has risen from £7 to £14 billion over the past decade. [24] Although there are five grades of whiplash classification, approximately 93% of patients post whiplash can be classified as WADII. [26] A neck complaint and musculoskeletal sign(s) are characteristic of WADII patients who are commonly managed by physiotherapists. [1] Conservative management (non-invasive treatment) is commonly utilized to manage acute WADII, and mainly focuses on physical treatment in terms of active exercise, manual techniques and physical therapy. [27,28] Currently, the effectiveness of conservative interventions is still reported as limited in managing acute WADII. [29][30][31][32][33][34][35][36][37][38] Patients with WAD exhibit both physical (e.g. pain and disability) and psychological (e.g. fear of movement, anxiety and depression) problems. [8,25,[39][40][41][42][43][44] Currently, the psychological components (e.g. cognitive behavioural therapy and other behavioural approaches) of WADII management are under-explored, and this may be a factor contributing to the limited success of some approaches to management. Some clinical guidelines have suggested psychological strategies in managing chronic WAD II. [27,28] However, these psychological components are not commonly recommended for management in the acute stage. Effectiveness of conservative management of acute WADII, employing both physical and psychological strategies is therefore important to prevent chronicity. No systematic review to date has specifically addressed this issue. The objective of this study was to evaluate the effectiveness of conservative management for acute WADII in order to summarise what we know about effective management in the acute stage.

Materials and Methods
A systematic review of randomised controlled trials (RCTs) was conducted according to a predefined protocol using the method guidelines of the Back Review Group of the Cochrane Collaboration, [45] the Cochrane handbook, [46] and is reported in line with PRISMA. [47] Eligibility criteria Table 1 details study eligibility criteria using Population Intervention Comparison Outcome Study Design (PICOS). [47] Information sources and searches Two independent reviewers (TW/MM) searched: • PEDro, Medline, Embase, AMED, CINAHL, PsycINFO, and Cochrane Library databases from inception to 15 th April 2015

Study selection
After searching, the two independent researchers (TW/MM) evaluated identified studies for eligibility by screening 1] title and abstract, then 2] full texts, grading each study as eligible/ not eligible/ might be eligible at each stage. [45] Included studies were agreed by the two reviewers (TW/MM). The third reviewer (AR) mediated in situations of disagreement. Only full articles in English were included.

Data collection process
Data were extracted by one researcher (TW) and checked by a second (MM). Trial authors were contacted for additional data when data were missing or ambiguous.

Data items
Trial authors, countries, study design, stage of whiplash patients, WAD classification, sample size, interventions, study setting, power calculations, outcome measures, follow-up period, loss of follow-up, intention to treat and main results were extracted for each trial. Data relating to key outcome measures including pain, disability, function, patient satisfaction, social impact, and physical impairment based on the International Classification of Functioning, Disability and Health were extracted. [48]

Risk of bias (RoB) in individual trials
Training and a pilot of RoB assessment was carried out by the two reviewers (TW/MM). The two reviewers evaluated RoB for each included trial independently. The Cochrane RoB assessment tool, which is informed by empirical evidence, was utilised to assess the internal validity / risk of bias. [47,49,50] The third reviewer (AR) mediated situations of disagreement following discussion. Each component of RoB was reported in terms of unclear, low or high risk of bias in tabular form. [49,50] The Kappa Measure of Agreement was utilised to assess the agreement between the two reviewers using SPSS version 21.

Summary measures
Risk of bias was assessed using the Cochrane RoB assessment tool. [49,50] The recommendation for overall RoB was in line with the Cochrane handbook. [46,49,50] Quantitative data analysis was conducted in situations of comparability of interventions, outcome measures and assessment points across trails. Meta-analyses compared effect sizes with random effects as the primary analyses and were conducted using STATA software version 12. [51]

Synthesis of results
Standardised mean difference (SMD) and standard error of SMD were calculated for metaanalyses. The results of meta-analyses were graphically demonstrated in forest plots. Summary statistics including 95% Confident Interval (CI), p-value, and heterogeneity (I 2 ) were tabulated.

Risk of bias across studies
RoB assessment across studies was tabulated. The criteria of judgement for overall RoB followed recommendations in the Cochrane Handbook. [50] A consensus in overall potential risk of bias was explored to evaluate the level of evidence. As the number of studies in each meta-analysis was less than 10, evaluation of publication bias using Funnel plots was not helpful. [52,53] Results

Study selection
15 RCTs (n = 1676 participants, 9 countries) were included. The process of study selection is detailed in Fig 1.

Study characteristics
Trial characteristics are summarised in Table 2. A range of conservative treatments were employed across included trials (see Table 2 for detail of interventions). Interventions could be grouped to inform analyses into conservative (broad group), active, passive and behavioural interventions.

Risk of bias within and across trials
Agreement of RoB assessment between two reviewers was very good (Kappa 0.87). [60] The RoB of individual trials is detailed in Table 3. All trials were assessed as high RoB for a range of reasons including: inadequate allocation concealment, selective outcome reporting, no intention to treat analysis, no specification of primary outcome, no specification primary endpoint, no reporting statistical analysis, no reporting reasons of drop-out, difference in loss to follow up between groups, and no reporting of some outcome measures or/and information.

Results of individual trials
Based on the comparability of interventions, outcome measures and assessment points across trials, meta-analyses were conducted on 4 intervention comparisons (please see Table 2 [37,55] A summary of the meta-analyses is detailed in Table 4.

Synthesis of results
Conservative intervention was more effective for pain reduction than standard/control intervention at 6 months (95% CI: -20.14 to -3.38, p = 0.005, I 2 = 63.8%) and 1-3 years (-25.44 to -3.19, p = 0.012, I 2 = 0.0%) (Fig 2). Moreover, conservative intervention was superior to the standard/control intervention for improvement of cervical mobility in the horizontal plane at <3months (0.43 to 5.60, p = 0.022, I 2 = 0.0%) (Fig 3). However, there were no significant differences between interventions for pain reduction at <3months, other follow-up periods in the horizontal plane of Cervical Range of Motion (CROM), nor any follow-up periods in terms of other planes of CROM, including the number of days of sick leave. Active intervention was more effective than passive intervention for pain reduction at 6 months (-17.19 to -3.23, p = 0.004, I 2 = 56.9%) and 1-3 years (-26.39 to -10.08, p<0.001, I 2 = 0.0%) (Fig 4). However, there was no significant difference in pain reduction at <3 months. Also, improvement of cervical mobility and days of sick leave were not significantly different between interventions.
There were no significant differences between early and late interventions for pain reduction, CROM, and days of sick leave at any follow-up period. and allocation concealment significantly improved between 2000 and 2006, [65] the quality of reporting blinding, and descriptions of approach, exclusion, treatment and missing data is still frequently inadequate, [65][66][67] contributing in 2010 to the revised CONSORT statement. [68,69] In this systematic review, only three trials were published after 2010. [36,54,56] Due to the high RoB across all trials, confidence in findings is reduced. The meta-analyses findings are more powerful and reliable than individual trials because of minimised biases from the individual trials. [70] The results of the meta-analyses were influenced by individual trials demonstrating conflicting conclusions. For example, the meta-analysis demonstrated that conservative intervention was more effective than standard/control intervention for pain reduction long term, despite some trials reporting negative finding. [35,37] Another example is that some trials [32,38,58] found the active intervention was more effective than the passive short term, but there was no effect demonstrated in the meta-analysis.
The level of heterogeneity was evaluated to determine the credibility of the evidence. [71] For example, in the meta-analyses demonstrating an effect for pain reduction at 6 months, the heterogeneity ranged from moderate (I 2 = 44.2%, behavioural intervention) to substantial (I 2 = 63.8%, conservative intervention; I 2 = 56.9%, active intervention), and this was acceptable overall.
Although this systematic review identified some interventions demonstrating an effect, the size of the effect was not clinically significant. The minimal clinical significant differences in improvement of pain intensity (VAS) and CROM are at least 20 millimetres [72] and 10°, [73] respectively. Currently, therefore, there is no evidence of an effective intervention for acute WADII management. However, conservative intervention (non-invasive treatment inclusive of both physical and psychological components such as active mobilisation exercise, manual techniques, physical agents, multimodal therapy, behavioural approaches, and education, except for drug therapy) seems to be a useful intervention for acute WADII management in terms of pain reduction in the medium (95% CI: -20.14 to -3.38, p = 0.005, I 2 = 63.8%) to long term (95%CI: -25.44 to -3.19, p = 0.012, I 2 = 0.0%), and improvement of cervical mobility in the horizontal plane in the short term (95%CI: 0.43 to 5.60, p = 0.022, I 2 = 0.0%) compared with standard/control intervention.  From these findings, there are two interesting interventions for acute WADII management worthy of further consideration. Firstly, active intervention (involving range of movement, mobilising exercises, and strengthening of the neck and scapular muscles) is strong recommended from whiplash guidelines [27,28] and may be useful for pain reduction medium (95% CI: -17.19 to -3.23, p = 0.004) to long term (95%CI: -26.39 to -10.08, p = <0.001). Secondly, behavioural intervention (e.g. act-as-usual, education and self-care including regularly exercise) may be effective for pain reduction medium term (95%CI: -15.37 to -1.55, p = 0.016) and improvement of cervical mobility in the coronal (95%CI: 0.93 to 4.38, p = 0.003) and horizontal planes (95%CI: 0.43 to 5.46, p = 0.27) short-medium term. The combination of the two into an active behavioural intervention may be a good strategy to manage acute WADII.

Strengths
This study's strengths are its design and specific focus to acute WADII using a pre-defined protocol and attention to potential sources of bias such as: a minimisation of errors from searching, using two independent reviewers, decreased publication bias through searching both published and unpublished trials, assessment of RoB using two independent reviewers, and data extraction using two reviewers.

Limitations
This study's limitations include the small number of trials identified and their high RoB. Furthermore, effectiveness for the outcome of NDI which is a key outcome measure [11,74] with high validity and reliability [75]could not be calculated in a meta-analysis due to an insufficient number of trials evaluating this outcome.
According to GRADE (the Grading of Recommendations Assessment, Development and Evaluation system), [76] the evidence reviewed in this study is low/very low (low in conservative and active interventions, very low in behavioural intervention). Consequently, an adequately powered, low risk of bias, and well-reported trial to evaluate effectiveness of acute WADII management is warranted to enable confidence in evidence for clinical practitioners, health policy-makers and researchers.

Conclusions
This rigorous systematic review found that conservative and active interventions may be useful for pain reduction in acute WADII management in the medium-long term. Additionally, improvement of cervical movement in the horizontal plane short term could be promoted by the employment of a conservative intervention. The employment of a behavioural intervention (e.g. act-as-usual, education and self-care including regularly exercise) may be an effective treatment in reducing pain and improving cervical mobility in patients with acute WADII in the short-medium term. Finally, there was no significantly difference between treating in early (<1 week) and late (2 weeks) interventions. The level of evidence from this systematic review is evaluated as low/very low level according to GRADE.