Understanding recovery pathways for individuals following

Transcription

1 CARRIE RITCHIE, PhD 1 MICHELE STERLING, PT, PhD 1 Recovery Pathways and Prognosis After Whiplash Injury Understanding recovery pathways for individuals following whiplash injury continues to be a challenge. Over the past few decades, recovery rates have remained unchanged, and it is generally accepted that approximately 50% of individuals with whiplash-associated disorder (WAD) TTSYNOPSIS: Recovery from a whiplash injury is varied and complex. Some individuals recover quickly and fully, while others experience ongoing pain and disability. Three distinct patterns of predicted recovery (trajectories) have been identified using disability and psychological outcome measures. These trajectories are not linear, and show that recovery, if it is going to occur, tends to happen within the first 3 months of the injury, with little improvement after this period. Identification of factors associated with poor recovery is accumulating, and since 2000 there have been at least 10 published systematic reviews on prognostic factors for whiplash-associated disorder. Poor recovery has been consistently reported to be associated with high initial neck pain intensity and neck-related disability, posttraumatic stress symptoms, pain catastrophizing, and, to a lesser extent, low self-efficacy and cold hyperalgesia. Evidence regarding factors, including compensation will fully recover, while the remaining 50% will continue to experience symptoms to some degree. 9,23,41 If recovery occurs, it tends to happen within the first 3 months of the injury, with little improvement after this period. 9,23,41 This high rate of nonrecovery and the consequent personal and societal costs have continued to drive research into this disabling condition. Understanding the transition from acute injury to ongoing chronic WAD will contribute to a greater understanding of those at risk of poor outcomes and, of equal importance, help to identify those who are likely to fully recover. Groupbased trajectory analytical techniques have been used to identify meaningful clusters of individuals who have a similar response to an outcome over time. 9,41,42 Identification of factors associated with membership in more homogeneous subgroups identified with these trajectories could facilitate development of appropriate screening and profiling processes. Numerous well-designed research studies status, psychological factors, structural pathology, and preinjury health status, remains equivocal. Given the huge number of predictive factors and various interpretations of recovery, adapting these data for use in clinical practice is difficult. Tools such as clinical prediction rules (CPRs), by statistically quantifying relevant data, may help to predict the probability of diagnosis, prognosis, or response to treatment. Numerous CPRs have been derived for individuals with whiplash; however, to date, only 3 prognostic CPRs have undergone external validation, and none have yet undergone impact analysis, a necessary step in providing information about the rules ability to improve clinically relevant outcomes. J Orthop Sports Phys Ther 2016;46(10): Epub 3 Sep doi: /jospt TTKEY WORDS: cervical spine, clinical prediction rule, neck, recovery, WAD have evaluated the association of a plethora of prognostic factors with ongoing nonrecovery from a whiplash injury. 7,23 Consolidation of these study results is difficult to adapt for use in clinical practice. Tools such as prognostic clinical prediction rules (CPRs) may help consolidate research data by quantitatively analyzing the contributions of specific patient characteristics and breaking down varied recovery patterns into more homogeneous subsets. 28 There has been growing interest in these types of predictive tools for individuals with musculoskeletal pain, given that therapeutic benefits may be more likely to occur if subgroups of patients are identified and better matched to treatments. 10 The purpose of this clinical commentary is to present an overview of the research published to date describing recovery and recovery trajectories in WAD, the current state of prognostic factors for nonrecovery from acute WAD, and, finally, the emerging evidence about prognostic clinical prediction tools for WAD. Definition of Recovery The concept of recovery has an enormous array of meanings. Recovery can be related to performance, ability, or function, and recovery can be defined as both a process and an end point. Not surprisingly, numerous outcome measures have been used to define recovery post whiplash injury, including constructs such as pain, disability, perceived recovery, psychological symptoms, and return to 1 Recover Injury Research Centre, NHMRC Centre of Research Excellence in Recovery Following Road Traffic Injuries, Menzies Health Institute Queensland, Griffith University, Parklands Drive, Australia. The authors certify that they have no affiliations with or financial involvement in any organization or entity with a direct financial interest in the subject matter or materials discussed in the article. Address correspondence to Dr Carrie Ritchie, Recover Injury Research Centre, Menzies Health Institute Queensland, Griffith University, Parklands Drive, QLD 4222 Australia. carrie.ritchie@griffith.edu.au t Copyright 2016 Journal of Orthopaedic & Sports Physical Therapy journal of orthopaedic & sports physical therapy volume 46 number 10 october

2 work, among others. In 31 independent cohort studies investigating outcome prediction for WAD, Walton 47 found 30 different methods used to define recovery. These variations create difficulties when attempting to interpret and consolidate the literature, describe natural recovery, and identify risk in clinical practice. 47 While it is outside the scope of this paper to provide a full discussion on outcome factors used to define recovery in WAD, this section will provide a brief overview of outcome measure selection. Self-report measures of relevant symptoms (eg, pain, dizziness), neck pain related disability, and/or individual perceptions of recovery are among the most common outcome measures used to define recovery from WAD. 50 These outcome measures relate to health and/ or functional goals for recovery, are relevant to the whiplash condition, and have been shown to be important indicators of recovery in individuals with WAD. 5,50 Qualitative data have revealed that reduced neck-pain symptoms and a desire to return to preinjury health are important markers of recovery for individuals with neck pain. 5,50 Psychological measures such as pain catastrophizing, 9 posttraumatic stress symptoms, 42 and anxiety and depression 4 have also been used to measure recovery in WAD. 26 The estimated prevalence of psychological symptoms, such as posttraumatic stress disorder, ranges from 6% to 45% 22,41 in individuals with chronic WAD compared with their prevalence in the general population, which is approximately 10% for Australian adults. 27 In addition, individuals with WAD and psychological symptoms such as posttraumatic stress report higher levels of pain and disability than individuals with WAD and no psychological symptoms. 43 Furthermore, there is some evidence to show that psychological symptoms such as pain catastrophizing 45 and posttraumatic stress 16 are amenable to change with appropriate intervention. Inclusion of an outcome measure that is sensitive to change is beneficial when assessing the effectiveness of a targeted therapy. 45 It should be noted that there is some evidence to suggest that preinjury psychological health may influence postinjury psychological health outcomes, emphasizing cautious interpretation of conclusions. 30 Returning to work after an injury is important for injured individuals, employers, insurers, and regulators. Numerous iterations of return to work have been used as outcome measures of recovery, including proxy measures such as sick leave and insurance claim closure. 24,47 While all of these may be important goals, return to work is a surrogate measure of recovery. Return to work is not the same as health recovery, and return to work, as an outcome measure, prevents inclusion of individuals who are unemployed or retired. Furthermore, the true duration of disability may be underestimated, as some individuals return to work prior to feeling like they are recovered. 34 As a result, research utilizing return-to-work outcome measures is oftentimes limited to narrow populations and jurisdictions. Consideration of the justifications used to define outcome measures and the criteria for recovery will help with critical appraisal of the literature. In general, standardized outcome measures that have been shown to be reliable, valid, and able to detect minimal clinically important difference are preferred over unique measures that have not undergone these fundamental statistical tests. It is also important to consider the course of recovery and outcome as a process. Understanding the developmental course of symptoms enables interventions to be designed and implemented at critical time points. Recovery Trajectories Some work has been done to examine the longitudinal course of recovery following whiplash injury. 9,41,42 Distinct postinjury recovery pathways for individuals with WAD have been identified using groupbased trajectory analytical techniques. 9,41 These statistical processes assess repeated measures of an outcome over time. 31 Individuals who respond similarly to an outcome measure over time are clustered, thereby proposing distinct recovery trajectories for more homogeneous subgroups. 31 These techniques also enable exploration of potential nonlinear changes in outcomes over time. 31 For example, initial change in an outcome may be rapid, whereas change may be slower at later time points. Another unique advantage of these statistical analyses is that a priori cutoffs of outcomes are not required; that is, distinct groups are not categorized by arbitrary outcome thresholds. 31 Sterling and colleagues 41 analyzed recovery trajectories for individuals with WAD (n = 155) recruited within 1 month of their injury and followed for 12 months postinjury. Three distinct patterns of predicted recovery (trajectories) were identified using both disability and psychological outcome measures. These recovery patterns were categorized generally as following a mild trajectory, moderate trajectory, or severe trajectory (FIGURE 1). Casey and colleagues 9 confirmed 3 distinct recovery trajectories using disability and psychological outcomes in a cohort of individuals with compensable WAD (n = 246) (FIGURE 2). Individuals with WAD who had lodged a compensation claim were recruited within 3 months of their injury and followed for 24 months postinjury. A similar predictive proportion to each of the mild, moderate, and severe recovery trajectories for both disability and psychological outcomes was found with both studies. 9,41 These data confirm that approximately 50% of individuals who sustain a whiplash injury are likely to fully recover, and the remaining 50% are likely to report some degree of ongoing neck-related disability or psychological dysfunction long term. Furthermore, these recovery trajectories were not linear and appear to plateau, with very little change in recovery status after 3 months (FIGURES 1 and 2). Both research teams also used dualtrajectory modeling analyses to assess possible commonalities between the physical (eg, disability) and psychological trajectories. 9,42 The results showed strong 852 october 2016 volume 46 number 10 journal of orthopaedic & sports physical therapy

3 FIGURE 1. Predicted NDI trajectories with 95% confidence limits and predicted probability of membership (percent). Suggested cutoffs for the NDI are 0% to 8% (no pain and disability), 10% to 28% (mild pain and disability), 30% to 48% (moderate pain and disability), 50% to 68% (severe pain and disability), and greater than 70% (complete disability). Abbreviation: NDI, Neck Disability Index. Reprinted with permission from Sterling et al. 41 Copyright 2010 Wolters Kluwer Health, Inc. FIGURE 2. Predicted disability trajectories using the FRI. Reprinted with permission from Casey et al. 9 Copyright 2015 Elsevier. Abbreviation: FRI, Functional Rating Index. linkages for the mild disability and mild psychological symptom trajectories, and for the severe disability trajectory and severe psychological symptom trajectory. The conditional probability of dual membership in the moderate disability and psychological symptom trajectories was more variable. To explain further, an individual experiencing mild to no symptoms of posttraumatic stress had a high likelihood of having low levels of neck-related disability and was likely to fully recover within a few months of the injury. 42 Similarly, an individual classified as a noncatastrophizer was very likely to have low neck-related disability and to fully recover quickly. 9 These individuals are likely to follow an uncomplicated recovery path and unlikely to require significant intervention. On the other hand, individuals with high levels of neck-related disability were likely to also experience high levels of psychological dysfunction and to experience ongoing neck-related disability long term. While it is recognized that the severe trajectory cluster is not the largest, this group may require more concerted management and is likely the group that incurs the majority of costs. The synchrony was less clear with individuals identified as following a moderate trajectory for psychological symptoms and individuals identified as following a moderate recovery trajectory for neck-related disability. Individuals grouped within the moderate trajectory pathways either fully recover or experience ongoing disability and/or psychological symptoms. It would appear important to continue to assess these individuals for changes in recovery and target interventions accordingly. These linkages support the influence of biopsychosocial factors on recovery from a whiplash injury, and add to understanding that there may be common mechanisms that underlie the development of chronic moderate/severe disability and psychological symptoms after a whiplash injury. This information is important for the design of specific interventions to target multiple factors associated with recovery and/or their mechanism of action. Given that interventions to date have had minimal success and that there is some evidence to show that early intensive intervention may be detrimental for some, 14,35 it is of clinical interest to be able to identify factors associated with membership in these distinct trajectories and more homogeneous subgroups. Sterling and colleagues 42 demonstrated that cold pain threshold, initial pain level (measured with a visual analog scale), and age predicted membership in both the severe disability and severe posttraumatic stress syndrome trajectory groups. Although dual-trajectory modeling indicated linkages between the severe disability and clinically significant catastrophizing (severe) trajectories, Casey and colleagues 9 found different factors to be predictive of membership in each group. Further research is needed to identify factors associated with trajectories of poor recovery and, of equal importance, to identify predictor vari- journal of orthopaedic & sports physical therapy volume 46 number 10 october

4 TABLE 1 Summary of Findings From Systematic Reviews About Prognostic Factors Associated With Poor Recovery From a Whiplash Injury Study Summary of Included Studies Predictors of Outcome No Effect on Outcome Inconsistent Evidence Côté et al cohort studies Older age and sex, baseline neck pain, baseline headache intensity, radicular signs/symptoms Scholten-Peeters et al articles from 29 cohorts High initial pain intensity Older age, female sex, high acute psychological response, angular deformity of the neck, rear-end collision, compensation Williams et al 52 Carroll et al 7 38 articles from 26 cohorts (physical factors only) Best-evidence synthesis; 47 articles Abbreviation: WAD, whiplash-associated disorder. High initial neck pain intensity, high initial neck pain related disability, cold hyperalgesia Greater initial symptom severity (greater initial pain, greater number of symptoms, more parts of the body in pain, painrelated limitations), postinjury psychological distress and passive coping styles, WAD grade II Kamper et al articles from 38 cohorts High neck pain intensity, high disability, WAD grade II or III, greater number of complaints, psychological distress (depression, anxiety), low coping, female sex, lower education, reduced cervical range of motion Williamson et al articles from 17 cohorts (psychological factors only) Limited evidence for association with lower self-efficacy, limited evidence for greater posttraumatic stress Walton et al cohorts Baseline neck pain greater than 55/100, headache at baseline, no postsecondary education, WAD grade II or III, presence of neck pain at baseline, catastrophizing, no seatbelt use at time of collision, female sex, history of previous neck pain Goldsmith et al 20 Walton et al 49 Daenen et al 15 6 studies from 4 cohorts (cold hyperalgesia only) Additional 4 cohorts to previous review of 11 cohorts 10 articles from 5 cohorts (cervical motor dysfunction only) Moderate evidence that cold hyperalgesia is an independent prognostic factor High pain intensity (>5.5/10) (most robust predictor), high Neck Disability Index score (>14.5/50) (may prove equally as useful), report of headache at inception, less than postsecondary education, no seatbelt in use during the accident, report of low back pain at inception, preinjury neck pain, high catastrophizing, female sex, WAD grade II or III... Limited evidence for compensation and legal factors; other demographics include work, education, etc; preinjury headache and neck pain; collisionrelated factors Limited evidence for reduced pressure pain thresholds Most collision-related factors, age, sex Neurological symptoms, radiographs taken, collision-related factors, prior neck pain, quality of life, posttraumatic stress syndrome, catastrophizing, stress, fear avoidance, older age, body mass index Personality traits, general psychological distress, well-being, social support, life control, psychological work factors Older age, disturbed sleep, other collision-related factors, history of headache, depressive symptoms Preaccident history of headache, rear-end collision, older age, collision severity, restricted cervical range of motion, report of disturbed sleep, presence of radicular symptoms at inception, depressive symptoms Restricted range of motion, high number of complaints, previous psychological problems, nervousness, accident on highway, need to resume physical therapy Limited evidence for preinjury chronic widespread pain Preliminary evidence that compensation or legal factors are associated with recovery; greater health care utilization in first month after injury is associated with slower recovery... Inconclusive: injury-related stress, previous psychological problems, blame and anger, perceived threat, cognitive function, anxiety, depression, irritability, familiarity with whiplash symptoms, fear avoidance, catastrophizing, coping, somatization Reduced cervical mobility, disturbed kinesthesia, altered muscle activity october 2016 volume 46 number 10 journal of orthopaedic & sports physical therapy

5 TABLE 2 Characteristics of Prognostic Clinical Prediction Rule Studies for Acute Whiplash Study Stage Sample* Included Variables Outcome Accuracy Posttest Probability Atherton et al 1 Derivation n = 535, acute (median time since injury, 8 d), emergency department Carroll et al 6 Derivation n = 1858, acute ( 6 wk postinjury), motor vehicle insurance claimants and health providers Gabel et al 19 Derivation n = 30, duration of injury, general practitioner and primary care Hartling et al 21 Derivation n = 334, acute ( 2 wk postinjury), emergency department Kasch et al 24 Derivation n = 141, acute ( 1 wk postinjury), emergency department Kasch et al 25 Validation n = 625, acute (median duration, 5 d), emergency department or general practitioner Nederhand et al 32 Derivation n = 90, acute (mean duration, 8.1 d), emergency department Olsson et al 33 Derivation n = 114, acute ( 4 wk postinjury), emergency department ables for the mild and moderate recovery trajectories. 5 variables: precollision widespread pain, vehicle type other than car, 5 WAD symptoms, NDI 19, GHQ 6 2 variables: PMI passive coping subscale of and CES-D 16 2 variables: modified ÖMPQ 109 and cervical rotation at impact 4-variable decision tree: MVC occurred other than at an intersection in the city, upper back pain since MVC, still experience neck pain, still experience shoulder pain 3 variables: total CROM <266.7, 7/15 nonpainful complaints, baseline pain VAS 54/100 mm Score chart (0-10) with 3 variables to create strata (1-7): total CROM, number of nonpainful complaints (0-11), baseline pain VAS (0-10) 2 variables: NDI >15/50 and TSK 40 1 variable: MPI interference variable >1/6 (pain) and 1/6 (no pain) 12-mo self-report neck pain (yes/no) lasting 1 d in last week Time to recovery (specific measure ) 6-mo NDI >28% or self-report symptoms or impairments 6-mo WAD pain classification III 12-mo self-report work handicap: reduced work hours and capacity from injury, dismissed, change in job or receiving disability pension 12-mo self-report work handicap: >3 mo of sick leave during last 6 mo, or work inability in last month, or not working anymore because of accident 24-wk NDI 15/50 12-mo self-report residual pain related to the accident (yes/no) Current State of Knowledge of Predictors Gauging prognosis is an important part of clinical decision making for musculoskeletal pain conditions. Given the complexity of WAD, the high rate of nonrecovery, and the growing evidence that recovery appears to plateau after approximately 3 months postinjury, 23,41 the ability to identify early an individual s likelihood of recovery may help direct treatment. More than 2 decades ago, the Quebec Task Force 40 recognized the potential importance of early identification of poor recovery and recommended that research be increased into the identification of prognostic factors to enable early identification of individuals less likely to recover and, therefore, to help direct therapeutic resources to those most in need. Researchers have heeded that recommendation, and, since publication of the Quebec Task Force report, 40 many 75% slower recovery if PMI was and CES-D was 16 (adjusted HRR = 0.25; 0.17, 0.38) Sensitivity, 100%; specificity, 87%; positive LR = 7.7 (CIs ) Sensitivity, 91.5% (86.5%, 96.6%); specificity, 51.4% (44.7%, 58.1%) Sensitivity, 30.0%; specificity, 99.2% (CIs ) Positive/negative LR for stratum 6: 3.49/0.57; for stratum 7: 7.84/0.73 Positive LR = 4.3; AUC, 0.77 (0.63, 0.91) PPV, 50.7% (44%, 57.4%); NPV, 91.7% (86.8%, 96.6%) PPV, 75.0%; NPV, 94.7% (CIs ) Stratum 6, 80.9%; stratum 7, 87.4% (CIs ) 83.3% (70.3%, 91.3%) Pain, 81%; no pain, 94% (CIs ) Table continues on page 855. potential factors have been examined for their predictive association with recovery from WAD. For clinicians and researchers alike, interpreting this vast amount of research with the aim of identifying important prognostic factors is difficult. It should be noted that identified prognostic factors are not necessarily causal of a poor outcome, but merely show an association with the outcome of interest (eg, pain-related disability). The most reliable evidence of prognostic factors comes from systematic journal of orthopaedic & sports physical therapy volume 46 number 10 october

6 TABLE 2 Characteristics of Prognostic Clinical Prediction Rule Studies for Acute Whiplash (continued) Study Stage Sample* Included Variables Outcome Accuracy Posttest Probability Radanov and Derivation n = 117, acute (mean, 7.2 Sturzenegger 36 d postinjury), general practitioner Validation n = 16, acute (mean, 23.1 d postinjury), insurance company Ritchie et al 38 Derivation n = 262, acute (<4 wk in duration), emergency department, primary care practices, and community Ritchie et al 37 Validation n = 101, acute (<4 wk in duration), emergency department, primary care practices, and community Williamson et al 54 Derivation n = 430, acute (duration, 6 wk), emergency department or physical therapy Equation including 9 variables: impaired neck movement, pretraumatic headache, history of head trauma, age, initial neck pain intensity (0-10), initial headache intensity (0-10), FPI nervousness, FPI neuroticism, focused attention (measure ) As above Full recovery, 2 variables: initial NDI 32%, age 35 y Ongoing disability, 3 variables: NDI 40%, age 35 y, PDS hyperarousal subscale 6 12-mo recovered or symptomatic (specific measure ) Recovered or symptomatic (specific measure ) Full recovery: 12-mo NDI 10% Ongoing disability: 12-mo NDI 30% As per Ritchie et al 38 Full recovery: 6-mo NDI 10% Ongoing disability: 6-mo NDI 30% 5 variables: baseline NDI 50%, self-predicted long (>6 mo) recovery or nonrecovery, GHQ 4/12, PCQ passive coping 5/12, CSOQ 6/15 12-mo NDI 30% Abbreviations: AUC, area under the curve; CES-D, Center for Epidemiologic Studies Depression Scale; CI, confidence interval; CROM, cervical range of motion; CSOQ, Cervical Spine Outcomes Questionnaire; FPI, Freiburg Personality Inventory; GHQ, general health questionnaire; HRR, hazard rate ratio; LR, likelihood ratio; MPI, West Haven-Yale Multidimensional Pain Inventory; MVC, motor vehicle collision; NDI, Neck Disability Index; NPV, negative predictive value;, not reported; ÖMPQ, Örebro Musculoskeletal Pain Questionnaire; PCQ, Pain Coping Questionnaire; PDS, Posttraumatic Stress Diagnostic Scale; PMI, Pain Management Inventory; PPV, positive predictive value; TSK, Tampa Scale of Kinesiophobia; VAS, visual analog scale; WAD, whiplash-associated disorder. *Number of participants, duration since injury, and place of recruitment. Values in parentheses are 95% CI. Full recovery: sensitivity, 45.3 (35, 54); specificity, 84.5 (77, 90); positive LR = 2.9 (1.9, 4.5); negative LR = 0.6 (0.5, 0.8) Ongoing disability: sensitivity, 43.5 (31, 55); specificity, 93.8 (89, 96); positive LR = 7.0 (3.8, 12.9); negative LR = 0.6 (0.5, 0.7) Full recovery: sensitivity, 54.9 (40.5, 68.6); specificity, 86.0 (72.6, 93.7); positive LR = 3.9 (1.9, 8.1); negative LR = 0.5 (0.4, 0.7) Ongoing disability: sensitivity, 43.5 (22.9, 65.1); specificity, 98.7 (92.9, 99.9); positive LR = 33.9 (4.6, 251.2); negative LR = 0.6 (0.4, 0.8) Relative risk for 1 factor, 3.46 (1.04, 11.45); for 2 factors, 7.44 (2.37, 23.40); for 3 factors, 8.11 (2.56, 25.68); for 4 factors, (5.36, 49.27) 96% (CIs ) 88% (CIs ) Full recovery: PPV, 70.7% (59%, 80%); NPV, 64.5% Ongoing disability: PPV, 71.4% (55%, 84%); NPV, 82.3% (76%, 87%) Full recovery: PPV, 80.0% (62.5%, 91.7%) Ongoing disability: PPV, 90.9% (58.7%, 98.5%) reviews. Systematic reviews address important methodological issues and help synthesize research in a specific domain. Attributable to the large number of cohort studies examining prognostic factors for recovery from WAD, there have been at least 10 systematic reviews published since 2000 (TABLE 1). 7,13,15,20,23,39,49,51-53 In addition, there has been 1 overview 856 october 2016 volume 46 number 10 journal of orthopaedic & sports physical therapy

7 of systematic reviews for acute neck pain. 48 Although this overview was not restricted to individuals with WAD, it did include a specific discussion about prognostic factors associated with recovery from a whiplash injury. 48 While drawing an overall conclusion from these reviews is challenging, for the most part, these systematic reviews confirm the presence of biopsychosocial factors in WAD recovery. As shown in TABLE 1, initial moderate to high levels of neck pain have been the most consistently reported prognostic factor for poor recovery from WAD. 7,13,23,39,49,51,52 Moderate to high neck-related disability has also consistently been shown to be predictive of poor recovery. 23,48,49,52 There is a moderate level of evidence that psychological factors, including posttraumatic stress symptoms, 48,53 pain catastrophizing, 49,51 and possibly low self-efficacy, 53 are associated with delayed recovery. In addition, there is a moderate level of evidence that cold hyperalgesia predicts poor recovery after a whiplash injury. 20 Conversely, collision-related factors have been the most consistently reported factors to have no association with recovery from WAD. 7,39,49,51 The association of factors such as age and sex with recovery remains equivocal. 13 To date, the conclusions from these systematic reviews provide the best consolidation of evidence toward understanding prognostic factors associated with poor recovery from a whiplash injury. However, there are, and will continue to be, emerging factors that have not been examined systematically and require further consideration. For example, there are factors, such as compensation-related processes 7,41 and perceived injustice, 46 that are difficult to evaluate consistently in the acute injury stage. Issues associated with the compensation claims process, including poor communication, problematic treatment approvals, and the burden of delayed claims settlement, have been suggested to impede recovery. 17 It has also been suggested that individuals with higher levels of perceived injustice may focus their attention on the injustice of their injury or collision rather than on the rehabilitation process, 44 resulting in a psychological barrier to recovery. Some of these factors are not possible to measure in the early days after a motor vehicle collision (MVC); for example, an injured person may not yet have submitted a claim for compensation, but consideration at the later subacute or chronic stage may be important for understanding ongoing nonrecovery. 17 Factors that have varied contextual associations, such as expectations of recovery, 34 also show some predictive capacity. One of the difficulties in assessing the literature about expectations of recovery is that the outcome may measure expectations in terms of reduced pain, or expectations of return to work, or expectations of full recovery. Consequently, it is difficult to review these data systematically. Additional factors that have not yet been reviewed systematically include those requiring specialized equipment or processes that are not readily available to an adequate number of research teams. 3,8,18,30 The availability of large epidemiological data sets 30 and access to jurisdictional administrative databases 3,8 have allowed specific prognostic factors collected before the whiplash injury to be assessed for their influence on recovery following an MVC. For example, higher use of sick leave prior to an MVC has been shown to be predictive of prolonged recovery following whiplash injury, 8 and preinjury use of multiple health services and elevated use of medications have been shown to predict ongoing pain postwhiplash. 30 While clearly difficult to prospectively assess in many cohorts, these results highlight the potential impact of preinjury health on recovery from a whiplash injury and may be important to understanding WAD recovery patterns. Although, to date, no verifiable structural change (eg, via radiological findings) has been shown to be associated with the transition to chronicity, advancements in magnetic resonance imaging technologies may begin to provide diagnostic or prognostic information. 18 At this stage, the data are not robust enough to suggest that early imaging is necessary, and, in fact, clinical guidelines recommend that early imaging not be provided to the vast majority of individuals postwhiplash. 29 However, there is emerging evidence to show that levels of muscle fatty infiltrates in cervical spine muscles measured 2 weeks postinjury are associated with neck disability at 3 months postinjury. 18 The mechanisms that cause the muscle fatty infiltrates are not known; however, in the future, this type of research may prove important to our understanding of processes underlying nonrecovery from a whiplash injury. 18 Finally, another challenge of translating the findings of prognostic-type research into clinical practice is that quantifiable cut points are rarely provided for prognostic variables. It is difficult for clinicians to use these potential prognostic factors at a clinical level. Tools such as CPRs may help to consolidate prognostic research by providing quick and usable estimates of probability. Clinical Prediction Tools Tools such as CPRs use quantitative methods to analyze the contributions of specific patient characteristics to help predict diagnosis, prognosis, or response to treatment. All types of CPRs should undergo 3 stages of development: derivation, validation, and impact analysis. 10,28 Derivation involves rigorous statistical analyses to determine a set of clinical variables with the greatest predictive power. Derived CPRs are not recommended for use in clinical practice because of the possibility that the results may reflect chance statistical associations or be specific to the study population. Following validation with a new patient cohort, the validated CPR may be applied with similar populations with some level of confidence in its predictive accuracy. The final stage of development, impact analysis, involves testing to see whether application of the CPR results in changed clinician behavior and whether that change improves patient outcomes. journal of orthopaedic & sports physical therapy volume 46 number 10 october

8 Only after impact analysis can a clinician be fully confident that using a CPR may improve outcomes. 10,28 While the recommendations for the development of diagnostic, prognostic, and prescriptive CPRs are the same, each type of CPR has specific advantages and limitations. This clinical commentary will present information on prognostic CPRs. Prognostic CPRs break down varied recovery patterns into more homogeneous subsets and assist clinicians in making predictions about recovery and nonrecovery using specific patient outcomes. 2,28 The predicted pathways are different from recovery trajectories in that these tools usually provide specific numeric cutoff values for each predictive factor, facilitating use in clinical practice. Several methodological issues should be considered when reviewing prognostic CPRs. First, data should be collected prospectively from longitudinal studies. 2 Second, CPRs are derived for a specific outcome measure, and predictor variables may not generalize to other measures of recovery. 12 Third, CPRs should be derived with adequate statistical power (eg, 10 to 15 study participants for each prospective predictor variable) and, although the statistical strength of the association of predictor variables (eg, R 2 or R 2 equivalent) is rarely reported, 11 examination of accuracy statistics with 95% confidence intervals (CIs) will provide some information about the extent of accuracy in classifying individuals (TABLE 2). 2,28 To our knowledge, there have been at least 10 statistically derived prognostic prediction tools for WAD (TABLE 2). 1,6,19,21,24,32,33,36,38,54 To ensure these statistically derived CPRs are not unique to the study population or do not reflect chance associations, validation is recommended with a unique population cohort. 2,28 Only 3 of these CPRs have begun the validation process. 25,36,37 No prognostic CPRs have undergone impact analysis. The whiplash CPR (FIGURE 3) has undergone derivation and initial validation processes. 37,38 This CPR analyzed 8 FIGURE 3. Whiplash clinical prediction rule (reference) to predict both continued moderate/severe disability and full recovery following an acute whiplash injury. Abbreviations: NDI, Neck Disability Index; PDS, Posttraumatic Stress Diagnostic Scale. Reprinted with permission from Ritchie et al. 38 Copyright 2013 Wolters Kluwer Health, Inc. previously identified predictor variables for poor recovery from 262 individuals who participated in 2 prospective, longitudinal studies. A dual-pathway CPR was derived using the Neck Disability Index (NDI) as the outcome measure that predicted ongoing moderate/severe disability and also predicted full recovery. An increased probability of developing chronic moderate/severe disability was predicted in the presence of older age (35 years or older), higher baseline NDI score (40% or greater), and hyperarousal symptoms (6 or greater on the hyperarousal subscale of the Posttraumatic Stress Diagnostic Scale). An increased probability of full recovery was predicted for individuals younger than 35 years of age and with a lower baseline NDI score (32% or less). External validation confirmed the reproducibility and accuracy of the whiplash CPR for individuals seeking health care for neck pain following an MVC, with a positive predictive value for the ongoing moderate/severe disability pathway of 90.9% (95% CI: 58.7%, 98.5%) and a positive predictive value for full recovery of 80% (95% CI: 62.5%, 91.7%). 37 Interestingly, although not the aim of a recent prospective cohort study, Elliott and colleagues 18 found that all participants classified at 3 months as having ongoing moderate/severe disability in their study (22% of the study population) fit the criteria for this CPR s prediction of developing moderate/severe disability, thereby providing further support for this CPR s prognostic value. The CPR provides a probability of prognosis and also indicates pathways that may assist in developing various treatment options based on likelihood of recovery. For example, the recommendation for individuals who meet the full-recovery criteria may be minimal treatment and reassurance, whereas a targeted treatment approach involving physical therapy, psychological treatment, and possibly medication may be recommended for individuals who meet the criteria for development of ongoing moderate to severe pain-related disability. Although prospective validation and impact of inclusion on practice are still needed to fully understand the usefulness of the whiplash CPR, the predictive probabilities for each pathway are significantly higher than estimated probabili- 858 october 2016 volume 46 number 10 journal of orthopaedic & sports physical therapy

9 TABLE 3 ties suggest in previous research, hence clinicians may have a certain level of confidence in using the whiplash CPR in individuals similar to those in the derivation and validation studies. Another CPR-like tool that has undergone derivation and validation analyses is the Danish Whiplash Study Group risk assessment tool (TABLE 3). 24,25 Eight factors were analyzed, using 12-month selfreport work handicap as the outcome measure for prospectively collected data in 141 study participants. 24 The derived tool was then converted into 7 risk strata, using an index based on measures of baseline intensity of neck pain and headache, cervical range of motion, and number of nonpain symptoms. Categories for each of these factors are linked to points on a 0-to-10 scale, and the total number of points is then categorized into 7 strata. Higher scores indicate greater risk of 1-year work disability. Although validation studies have altered the original screening tool, the results showed that only 4% (CIs not reported) of individuals within stratum 1 were work disabled after 12 months, compared with 68% (CIs not reported) of those in stratum 7. In addition, the risk score and number of sicklisted days were significantly related. 25 The Danish Whiplash Study Group Risk Assessment Score* CROM, deg > <200 Neck-head VAS (0-10) Number of nonpain symptoms Risk strata Points Stratum 1 = 0 points; stratum 2 = 1-3 points; stratum 3 = 4-6 points; stratum 4 = 7-9 points; stratum 5 = points; stratum 6 = points; stratum 7 = points. Abbreviations: CROM, cervical range of motion; VAS, visual analog scale. *Reprinted with permission from Kasch H, Qerama E, Kongsted A, Bach FW, Bendix T, Jensen TS. The risk assessment score in acute whiplash injury predicts outcome and reflects biopsychosocial factors. Spine (Phila Pa 1976). 2011;36:S263-S Copyright 2011 Wolters Kluwer Health, Inc. A total of risk points is given based on assessment of each of 3 risk factors: neck mobility (active CROM), present score of pain (maximum value of either neck pain or headache rated by VAS or by 11-point box scale), and number of nonpainful complaints (paresthesia, dizziness, vision disturbances, tinnitus, hyperacusis, globulus, fatigue, irritation, concentration disturbances, memory difficulties, sleep disturbances). Hence, this risk stratum system may be useful to consider as a tool to assess likelihood of return to work following a whiplash injury. A third example of a CPR-like tool to undergo derivation and initial validation analyses is a tool that considers 9 variables, including initial severity of injury (assessed by impaired neck movement and intensity of neck pain and headache), preinjury variables (history of head trauma and preinjury headache), and initial injury-related changes in psychological and cognitive function (nervousness and impaired focus attention), to predict recovery 12 months after a whiplash injury. 36 While this tool demonstrated good posttest probability on validation, the CPR was derived from the analysis of more than 40 predictor variables using only 117 participants, and CPR accuracy, CIs, and a definition of outcome were not provided. Furthermore, only 16 participants were included in the validation study. The omission of these details and use of a very small sample size make interpretation and generalization of these results difficult. These examples of research-generated tools to predict recovery following a whiplash injury provide a mechanism to translate some of the research regarding prognosis into a procedure that health care providers may find easy to implement and interpret. In addition, identification of subgroups may help to facilitate the design of treatment strategies to target specific subsets of individuals. To our knowledge, none of the screening tools derived to identify recovery pathways for WAD have undergone an impact analysis. Clearly, additional research is needed to assess the impact of inclusion of these tools in practice and to examine the efficacy of linking treatment strategies with predicted prognosis. SUMMARY The studies reviewed as part of this clinical commentary highlight the ongoing challenges associated with understanding recovery pathways for individuals following a whiplash injury. While 50% of individuals appear to follow a mild recovery trajectory and recover fully, the remaining 50% continue to experience some level of ongoing disability. Given this high rate of nonrecovery and the growing evidence that recovery appears to plateau after approximately 3 months postinjury, the ability to identify early an individual s likelihood of recovery may help inform the development of health care interventions aimed at preventing the transition from acute to chronic WAD. Poor recovery has been consistently reported to be associated with moderate to high initial neck pain intensity and neck-related disability, posttraumatic stress symptoms, pain catastrophizing, and, to a lesser extent, low self-efficacy and cold hyperalgesia. Evidence regarding factors including compensation status, some psychological factors, structural pathology, and preinjury health status remains equivocal, and further research is needed to understand the influence of these factors on recovery from WAD. Adapting these data for use in clinical practice is difficult. Tools such as CPRs may help by providing a mechanism to translate some of the research journal of orthopaedic & sports physical therapy volume 46 number 10 october

10 regarding prognosis into a procedure that health care providers may find easy to implement and interpret. In addition, identification of more homogeneous subgroups may help to stimulate additional research to improve our understanding of recovery from WAD and to facilitate the design of treatment strategies to target specific subsets of individuals. To our knowledge, only 3 of 10 statistically derived prognostic CPRs for WAD have begun the validation process, and none have undergone impact analysis. Further research is needed to determine whether these types of tools help improve outcomes for individuals with WAD. t REFERENCES 1. Atherton K, Wiles NJ, Lecky FE, et al. Predictors of persistent neck pain after whiplash injury. Emerg Med J. 2006;23: org/ /emj Beattie P, Nelson R. Clinical prediction rules: what are they and what do they tell us? Aust J Physiother. 2006;52: Berecki-Gisolf J, Hassani-Mahmooei B, Collie A, McClure R. Prescription opioid and benzodiazepine use after road traffic injury. Pain Med. 2016;17: pme Berglund A, Bodin L, Jensen I, Wiklund A, Alfredsson L. The influence of prognostic factors on neck pain intensity, disability, anxiety and depression over a 2-year period in subjects with acute whiplash injury. Pain. 2006;125: Bostick GP, Brown CA, Carroll LJ, Gross DP. If they can put a man on the moon, they should be able to fix a neck injury: a mixed-method study characterizing and explaining pain beliefs about WAD. Disabil Rehabil. 2012;34: dx.doi.org/ / Carroll LJ, Cassidy JD, Côté P. The role of pain coping strategies in prognosis after whiplash injury: passive coping predicts slowed recovery. Pain. 2006;124: org/ /j.pain Carroll LJ, Hogg-Johnson S, van der Velde G, et al. Course and prognostic factors for neck pain in the general population: results of the Bone and Joint Decade Task Force on Neck Pain and Its Associated Disorders. Spine (Phila Pa 1976). 2008;33:S75-S82. org/ /brs.0b013e be 8. Carstensen TB, Fink P, Oernboel E, Kasch H, Jensen TS, Frostholm L. Sick leave within 5 years of whiplash trauma predicts recovery: a prospective cohort and register-based study. PLoS One. 2015;10:e org/ /journal.pone Casey PP, Feyer AM, Cameron ID. Course of recovery for whiplash associated disorders in a compensation setting. Injury. 2015;46: injury Childs JD, Cleland JA. Development and application of clinical prediction rules to improve decision making in physical therapist practice. Phys Ther. 2006;86: Cook C. Potential pitfalls of clinical prediction rules. J Man Manip Ther. 2008;16: dx.doi.org/ / Cook CE, Learman KE, O Halloran BJ, et al. Which prognostic factors for low back pain are generic predictors of outcome across a range of recovery domains? Phys Ther. 2013;93: Côté P, Cassidy JD, Carroll L, Frank JW, Bombardier C. A systematic review of the prognosis of acute whiplash and a new conceptual framework to synthesize the literature. Spine (Phila Pa 1976). 2001;26:E445-E Côté P, Hogg-Johnson S, Cassidy JD, Carroll L, Frank JW, Bombardier C. Early aggressive care and delayed recovery from whiplash: isolated finding or reproducible result? Arthritis Rheum. 2007;57: art Daenen L, Nijs J, Raadsen B, Roussel N, Cras P, Dankaerts W. Cervical motor dysfunction and its predictive value for long-term recovery in patients with acute whiplash-associated disorders: a systematic review. J Rehabil Med. 2013;45: Dunne RL, Kenardy J, Sterling M. A randomized controlled trial of cognitive-behavioral therapy for the treatment of PTSD in the context of chronic whiplash. Clin J Pain. 2012;28: AJP.0b013e318243e16b 17. Elbers NA, Akkermans AJ, Lockwood K, Craig A, Cameron ID. Factors that challenge health for people involved in the compensation process following a motor vehicle crash: a longitudinal study. BMC Public Health. 2015;15: dx.doi.org/ /s Elliott JM, Courtney DM, Rademaker A, Pinto D, Sterling MM, Parrish TB. The rapid and progressive degeneration of the cervical multifidus in whiplash: an MRI study of fatty infiltration. Spine (Phila Pa 1976). 2015;40:E694-E dx.doi.org/ /brs Gabel CP, Burkett B, Neller A, Yelland M. Can long-term impairment in general practitioner whiplash patients be predicted using screening and patient-reported outcomes? Int J Rehabil Res. 2008;31: MRR.0b013e3282f44e Goldsmith R, Wright C, Bell SF, Rushton A. Cold hyperalgesia as a prognostic factor in whiplash associated disorders: a systematic review. Man Ther. 2012;17: org/ /j.math Hartling L, Pickett W, Brison RJ. Derivation of a clinical decision rule for whiplash associated disorders among individuals involved in rear-end collisions. Accid Anal Prev. 2002;34: Heron-Delaney M, Kenardy J, Charlton E, Matsuoka Y. A systematic review of predictors of posttraumatic stress disorder (PTSD) for adult road traffic crash survivors. Injury. 2013;44: injury Kamper SJ, Rebbeck TJ, Maher CG, McAuley JH, Sterling M. Course and prognostic factors of whiplash: a systematic review and metaanalysis. Pain. 2008;138: org/ /j.pain Kasch H, Bach FW, Jensen TS. Handicap after acute whiplash injury: a 1-year prospective study of risk factors. Neurology. 2001;56: Kasch H, Kongsted A, Qerama E, Bach FW, Bendix T, Jensen TS. A new stratified risk assessment tool for whiplash injuries developed from a prospective observational study. BMJ Open. 2013;3:e bmjopen Kenardy J, Heron-Delaney M, Warren J, Brown EA. Effect of mental health on long-term disability after a road traffic crash: results from the UQ SuPPORT study. Arch Phys Med Rehabil. 2015;96: apmr McEvoy PM, Grove R, Slade T. Epidemiology of anxiety disorders in the Australian general population: findings of the 2007 Australian National Survey of Mental Health and Wellbeing. Aust N Z J Psychiatry. 2011;45: / McGinn TG, Guyatt GH, Wyer PC, Naylor CD, Stiell IG, Richardson WS. Users guides to the medical literature: XXII: how to use articles about clinical decision rules. Evidence-Based Medicine Working Group. JAMA. 2000;284: Motor Accidents Authority. Guidelines for the Management of Acute Whiplash-Associated Disorders for Health Professionals. Sydney, Australia: Motor Accidents Authority; Myrtveit SM, Skogen JC, Petrie KJ, Wilhelmsen I, Wenzel HG, Sivertsen B. Factors related to non-recovery from whiplash. The Nord-Trøndelag Health Study (HUNT). Int J Behav Med. 2014;21: s Nagin DS, Odgers CL. Group-based trajectory modeling in clinical research. Annu Rev Clin Psychol. 2010;6: annurev.clinpsy Nederhand MJ, Ijzerman MJ, Hermens HJ, Turk DC, Zilvold G. Predictive value of fear avoidance in developing chronic neck pain disability: consequences for clinical decision making. Arch Phys Med Rehabil. 2004;85: october 2016 volume 46 number 10 journal of orthopaedic & sports physical therapy