Secondary Logo

Journal Logo

Prognosis After Whiplash Injury: Where to From Here? Discussion Paper 4

Sterling, Michele, PT, PhD*; Carroll, Linda J., PhD; Kasch, Helge, MD, PhD; Kamper, Steven J., PT, PhD§; Stemper, Brian, PhD

doi: 10.1097/BRS.0b013e3182388523
Focus Papers: Roundtable

Study Design. Nonsystematic review and discussion of prognosis after whiplash injury.

Objective. To summarize the research and identify a research agenda for improving prognostic models after whiplash injury.

Summary of Background Data. With up to 50% of individuals failing to fully recover after whiplash injury, the capacity to determine a precise estimate of prognosis will be important. Systematic reviews note inconsistencies and shortcomings of research in this area.

Methods. A nonsystematic review and discussion.

Results. Most prognostic whiplash studies are phase 1 (exploratory) studies with few confirmatory or validation studies yet available. It is recognized that whiplash is a heterogeneous condition and clinicians require prognostic indicators for clinical use. Although the evidence is not sufficiently strong to make firm recommendations, there are some prognostic factors that have shown consistency across studies and could be considered as preliminary flags or guides to gauge patients potentially at risk of poor recovery. These include pain and/or disability levels, neck range of movement, cold and mechanical hyperalgesia and psychological factors of recovery beliefs/expectations, post-traumatic stress symptoms, depression, and pain catastrophizing. It is not known whether these factors can be modified or whether modification will improve outcomes, thus they should not be considered directives for management. Research priorities identified to develop improved predictive models include confirmation and validation of factors identified in phase 1 studies; investigation of the interaction between variables; investigation of the predictive value of changes in variables over time; the inclusion of validated outcomes including measures of pain and disability as well as perceived recovery and psychological outcomes.

Conclusion. The current evidence is not sufficiently robust to be able to confidently predict outcome after whiplash injury. A preliminary set of consistent factors has been proposed to assist clinicians in identifying individuals at risk of poor recovery. Directions for the development of improved prognostic models are discussed.

Current evidence is not strong enough to be able to confidently predict outcome after whiplash injury. There are some factors that have shown consistency across studies that may be useful as a preliminary guide only and not as directives for specific interventions. These include pain and/or disability levels, neck range of movement, cold and mechanical hyperalgesia and psychological factors of recovery beliefs/expectations, post-traumatic stress symptoms, depression, and pain catastrophizing. Further research is required to confirm and validate predictive factors identified in phase 1 studies.

*Centre of National Research on Disability and Rehabilitation Medicine (CONROD), The University of Queensland, Brisbane, Queensland, Australia

Department of Public Health Sciences and Alberta Centre for Injury Control and Research, School of Public Health, University of Alberta, Edmonton, Alberta, Canada

The Danish Pain Research Center, Department of Neurology, Aarhus University Hospital, Aarhus, Denmark

§The George Institute for Global Health, University of Sydney, Sydney, New South Wales, Australia

Department of Neurosurgery, Medical College of Wisconsin and Veterans Affairs Medical Center, Milwaukee

Address correspondence and reprint requests to Michele Sterling, PhD, Centre of National Research on Disability and Rehabilitation Medicine, The University of Queensland, Edith Cavell Bldg, Royal Brisbane and Women's Hospital, Herston, Queensland 4029, Australia; E-mail:

Acknowledgment date: June 30, 2011. First revision date: September 4, 2011. Second revision date: September 13, 2011. Acceptance date: September 19, 2011.

The manuscript submitted does not contain information about medical device(s)/drug(s).

Professional Organizational funds and Foundation funds were received to support this work. No benefits in any form have been or will be received from a commercial party related directly or indirectly to the subject of this manuscript.

M.S. and S.K. are supported by the National Health and Medical Research Council of Australia. L.C.'s participation was made possible by a Senior Health Scholar Award provided by Alberta Innovates–Health Solutions, funded by the Alberta Heritage Foundation for Medical Research Endowment Fund.

The capacity to be able to determine a precise estimate of prognosis is important for a condition such as whiplash where up to 50% of those injured will not fully recover within a year1 and where there is substantial personal, economic, and social burden associated with the condition. The early identification of factors or characteristics that predict good or bad recovery may allow the institution of management approaches specific to the injured individual. It may also facilitate decision making and planning by injured people, clinicians, and other stakeholders.

There are now numerous studies available that have investigated prognosis after whiplash injury, yet all systematic reviews of whiplash prognosis highlight inconsistencies and shortcomings of research in this area.15 This leaves clinicians and stakeholders in a difficult situation in their attempts to apply the results of prognostic studies to the clinical scenario. A further reason for the confusion surrounding this issue is the nature of the predictive study conducted and what information it can actually provide to add to knowledge in this area.

Therefore, this review will first outline the types of predictive studies and phases of prognostic investigation and where the status of whiplash prediction sits with regard to this framework. It will then outline the current problematic issues with the currently available evidence for whiplash prognosis before exploring where to go from here for both research and clinical practice.

Back to Top | Article Outline


Various frameworks for the conceptualization of the stages of investigation of prognostic studies have been put forward.6,7 These frameworks recognize prognostic studies as being either exploratory in nature (phase 1 studies) with subsequent studies attempting validation of a previously identified prognostic factor or model (phase 2 studies).6,7 Exploratory or phase 1 studies aim to identify associations between potential prognostic factors and outcome via either univariate and/or multivariate analysis of a large set of predictors. These studies are hypothesis-generating studies and for this reason cannot be well extrapolated to individual patients. The majority of studies of prognosis after whiplash injury would be consistent with an exploratory or phase 1 study.813 Validation or phase 2 studies are confirmatory in that they test the independent effect of a predictor while controlling for confounders.6 In addition, they may validate a previously identified prognostic model in a new and independent cohort.14 There are some whiplash studies that are confirmatory in design. For example, in Sweden, Holm et al15 demonstrated the predictive capacity of recovery expectations while controlling for other factors such as symptom intensity and levels of distress. This factor has also been demonstrated to have predictive capacity in a cohort in Canada,16 providing some validation for this factor. We could find no studies that have attempted external validation of a previously identified model, that is, studies that have verified the strength of the associations (the coefficients) between predictors and the outcome in a different cohort. External validation is important to establish the generalizability of a predictive model with most models performing less well when tested in a new cohort.14 Thus, at this point in time, it is unknown whether identified prognostic models will show a similar performance in different sites (countries, jurisdictions) and thus clinicians should be cautious in their application of such models.

Hayden et al6 go one step further to include phase 3 studies in their general prognostic framework that aim to understand prognostic pathways via explanatory studies. These studies are based on an explicit theoretical framework and aim to provide information on mechanisms of action of predictive factors on outcome. These studies require complex designs, for example, involving multilevel modeling or structural equation modeling. There are no whiplash studies yet conducted under this category.

As can be seen, most studies of prognosis after whiplash injury are phase 1 or exploratory studies. The results of these studies cannot be automatically translated to individual patients and this likely contributes to the lack of certainty about the prognosis of this condition. Further confirmatory and validation studies as well as explanatory studies are required before clear prognostic indicators for recovery emerge and can be confidently used in the clinical environment.

Back to Top | Article Outline


Available systematic reviews of prognosis after whiplash injury have noted shortcomings in many of the primary cohort studies including inconsistencies between studies in time from injury until baseline data collection, use of various and sometimes invalidated outcome measures, and the lack of blinded outcome assessment among others.15 In addition, where prognostic models have been identified, most have failed to explain much of the variance in the outcome of interest, usually levels of pain and/or disability.17,18 Of course, it is highly unlikely that a prognostic model for whiplash will ever explain 100% of the variance in outcome, but models need to show improved performance to be confidently applied to clinical practice.

It could be argued that in a clinical context, predictive models can be useful in one of two ways. First, to alert clinicians to those at high risk of poor recovery without necessarily offering information to guide management and second, models that can direct treatment decision making. The latter is more useful but requires the inclusion of factors that can be modified with specific intervention approaches. In the case of whiplash, there are several predictive factors that could potentially be modified, but it is yet unknown whether such modification can be achieved or if the modification of certain factors will actually improve outcomes after injury. For example, the most consistent predictive factor is initially higher levels of pain.5 Yet, there are no studies that have specifically investigated how acute whiplash pain may be effectively decreased and whether this would improve more long-term outcomes. The same scenario applies to other potentially modifiable prognostic indicators such as sensory disturbances, movement loss, and psychological factors. Subsequent to the current lack of knowledge in this regard, there are no clinical prediction or decision-making tools or algorithms yet available for the management of whiplash-associated disorders (WAD).

Most, if not all, prognostic studies measure a predictive variable at one point in time, when in fact many of these variables will change over time. Examples of time-changing variables would include pain levels, disability, and psychological measures among others. It is likely that it is the change (or lack thereof) in variables over time that may be a more relevant prognostic indicator. This is more in keeping with clinical practice where such variables are measured over the course of a patient's management not just at one time point. This would be a promising area for future research.

It should also be noted that prognostic models developed and tested at one site (e.g., country and jurisdiction) and in one cohort might not have the same predictive capacity when used or tested at another site. Inherent differences in population demographics, access to medical interventions, litigation environment, and consensus ideas regarding WAD may influence patient perceptions and outcomes. As outlined previously, a prognostic model is not ready for clinical application until robust validation is achieved.14

Back to Top | Article Outline


It is now widely acknowledged that WAD is a heterogeneous condition with diverse physical and psychological characteristics variably present in injured people1,4,19 and variable clinical pathways for recovery.20 The heterogeneous nature of the condition has led to calls from clinicians for prognostic indicators that would be useful for clinical practice to assist in gauging an individual patent's prognosis as well as assisting with clinical decision making. However, as illustrated earlier, the current state of the evidence is not strong enough to make firm recommendations as to the clinical use of prognostic indicators. Nevertheless, there are some prognostic indicators that have shown some consistency across studies that could be considered as “flags” or guides for clinicians to gauge which patients may be at risk of poor recovery and which patients show good potential for recovery. It is important to emphasize that these factors are a clinical guide only and it is not known whether treatment decisions made according to their presence or absence will alter an individual patient's outcome. It is also important to understand that further studies are required to confirm and validate most of these factors.

This preliminary “core set” of predictors could include a measure of reported pain levels, a measure of reported disability, neck range of movement, the presence of cold and mechanical hyperalgesia, and measures of psychological factors (probably post-traumatic stress symptoms but also other factors if indicated such as depression, pain catastrophizing, recovery beliefs, and expectations among others). Higher initial pain intensity is the most consistent predictor of poor recovery and has been identified by all systematic reviews15; thus, at this stage, it would be the most important factor to include in the clinical evaluation of acute WAD. Initial levels of pain-related disability have been identified by some primary cohort studies1,4 and are recommended for clinical assessment by current clinical guidelines for whiplash management.21 Poor patient expectations of recovery have been shown to be predictive of poor recovery in two phase 2 studies in separate cohorts, indicating that this factor should also be included in early clinical assessment.

The other variables have been less well investigated to date. Conflicting results have been found for the predictive capacity of neck range of movement,3,5 but again this measure is commonly used in the clinical environment and may provide some prognostic information and for this reason could continue to be used in the clinical assessment of acute WAD. The presence of hyperalgesic sensory responses has been investigated in several cohort studies. In general, it would seem that increased sensitivity to cold stimulation tested via pain threshold responses or the cold pressor test has been shown to be a significant predictor in several studies.9,22,23 Mechanical hyperalgesia usually tested by responses to pressure stimuli has been shown to be predictive in some studies but not others.9,10 The mechanisms underlying the hyperalgesic responses are not clearly understood but are thought to reflect augmented nociceptive processing in the central nervous system.24

Numerous studies have now investigated psychological factors and their association with outcome after whiplash injury. The more recent systematic reviews indicate that there is some evidence for predictive capacity of symptoms of post-traumatic stress,25 lower self-efficacy,25 pain catastrophizing,5 depressed mood,1 and fear of movement.1 Recent studies indicate that lower expectations of recovery are also predictive of poor recovery.15,16

This “core set” of prognostic indicators is by no means exhaustive and will likely change as new evidence comes to light. Furthermore, the optimal scores or cutoffs on these questionnaires that would assist clinicians in decision making are not clear, also emphasizing the preliminary nature of these factors. However, a recent systematic review suggested a cutoff score of 55 of /100 for a visual analog pain scale.5

Back to Top | Article Outline


It is clear that the current situation and knowledge base of prognostic indicators after whiplash injury need to be improved. Clear validated prognostic indicators for both good and poor recovery would benefit both clinicians and patients. The literature at present has numerous primary cohort studies and systematic reviews of these studies, yet we can still not clearly identify patients at risk of poor recovery. There are very few validation or explanatory studies available. Systematic reviews all make comment on quality issues of primary cohort studies and problems include low sample sizes, lack of validated outcome measures, inconsistent ways to measure prognostic variables, and poor reporting standards.13,5,27 In fact, only one review attempted meta-analysis5 with others citing heterogeneity as a reason for being unable to pool data.3,4

Progress in this research area would be facilitated by a more consistent approach to the design, conduct, analysis, and reporting of predictive studies. On the basis of current available data, a priority set of variables has been proposed for inclusion in future predictive studies (Table 1). The use of these variables in future research would enable clarification and validation of predictive factors in various sites, countries, or jurisdictions and consolidate the results of the numerous primary studies available. Multicenter studies could be useful to achieve the sample sizes required for larger studies. Interactions between variables should also be explored, as should the time period at which the prognostic factor has the maximal predictive value.



The priority measures considered for inclusion are broad ranging to encompass the biopsychosocial construct of chronic pain development and include variables of self-reported pain and disability, psychological factors, physical factors, social and societal factors, and biomechanical crash-related factors. For reasons outlined earlier, they would include self-reported pain and pain-related disability levels as well as physical factors of neck range of movement, cold and mechanical pain thresholds. Psychological factors suggested for use in the clinical assessment of WAD would also be included and these are post-traumatic stress symptoms, self-efficacy, pain catastrophizing, depressed mood, and beliefs/expectations of recovery. In addition, variables including depression precursors such as health status before the injury28 and feelings of perceived injustice29 have shown recent prognostic promise and should also be further investigated. Crash-related biomechanical factors have also been included as although these have received some investigation, the measures taken have often been imprecise. For example, research participants’ self-perceptions of vehicle speed or whether or not the person's head was turned at the time of impact.1 Factors that may be important include vehicle change in velocity (i.e., Delta-V), head restraint positioning, acceleration at the time of impact, crash anticipation, and opportunity to pretense muscles. Existing accelerometer technology providing the ability to record and analyze vehicle accelerations during the crash event can remove some of the ambiguity regarding magnitude and direction of acceleration and Delta-V. Likewise, emergency response personnel and police training to record vehicle-related factors such as head restraint positioning, vehicle damage, and seat belt use may provide a better estimate of the vehicle crash environment. The relationship between crash-related factors and recovery will not be realized until improved methods of measurement are made.

Last, social and societal factors, for example, compensation-related factors should also be considered. The association of these factors and outcome is arguably one of the most debated in the whiplash literature with some reviews finding support for the predictive capacity of such factors1 and others reporting no prognostic capacity.3 Further investigation of these factors is required and as such they should be included in future studies of this type.

Although the consolidation and validation of the above set of promising measures is a priority, we are not arguing to stifle research in this area by limiting investigation to these variables. There will no doubt emerge new and innovative factors that will need to be explored for their predictive capacity in phase 1 study designs before further validation. Recent studies have identified interesting features of participants with acute and chronic WAD that warrant further investigation of possible prognostic capacity. Some of these include morphological fatty muscle changes identified using magnetic resonance imaging in participants with acute WAD and subsequent poor functional recovery30 and the presence of genetic variations in the catechol-O-methyltransferase gene in individuals with acute injury and more severe symptoms.31 Although factors such as these may not be practical for clinical use to gauge prediction, they will be useful to inform and develop etiological models. This may, in turn, inform future management directions.

There are several factors that need urgent consideration for the design and planning of future prognostic cohort studies after whiplash injury. Previous calls have been made for the inclusion of validated outcome measures3 and this is again reiterated. At a minimum, a validated measure of pain-related disability should be included with measures of other outcomes, for example, psychological and work-related outcomes also considered. It should be noted that recovery can be difficult to define and there is much debate about this in the literature.26 In view of this, a measure of perceived recovery should also be used. Investigators are also urged to consider the dynamic nature of many potential predictor variables and build changes in the variables into predictive models. The use of more sophisticated analysis techniques (e.g., multilevel modeling and factor analysis) will also facilitate an enhanced understanding of explanatory relationships between variables as well as allow site and system differences to be accounted for in multicenter studies.

Back to Top | Article Outline


Numerous studies have investigated factors for their capacity to predict outcome after whiplash injury. Because of shortcomings in many studies and that they are mainly exploratory in nature, predictive factors are still quite vague and further work is required before variables can be confidently adapted for clinical use. Nevertheless, some factors have shown consistency across studies and for this reason a preliminary predictive set has been proposed. These variables should be considered as a guide only and not as directives for specific management. Future research should aim to confirm and validate predictive models. Research centers need to work together, in terms of both sharing existing data sets and also planning new prospective studies together. Independent patient data meta-analysis is virtually unheard of in the whiplash field and may be helpful in this area.

Back to Top | Article Outline

Key Points

  • Because of the significant chronicity rates, it is important to be able to accurately identify injured people at high risk of poor recovery.
  • There are shortcomings in the current evidence of prognostic factors after whiplash injury, including that most studies are phase 1 or exploratory in nature.
  • Nevertheless, some factors have shown consistency across several prognostic studies and for this reason a preliminary set of factors is proposed as “flags” or guides in the clinical evaluation of acute WAD.
  • Further research is required to confirm and validate prognostic models as well as to develop improved models.
Back to Top | Article Outline


1. Carroll L, Holm L, Hogg-Johnson S, et al. Course and prognostic factors for neck pain in whiplash-associated disorders (WAD): results of the Bone and Joint Decade 2000–2010 Task Force on Neck Pain and Its Associated Disorders. Spine 2008;33:583–92.
2. Côté P, Cassidy D, Carroll L, et al. A systematic review of the prognosis of acute whiplash and a new conceptual framework to synthesize the literature. Spine 2001;26:E445–58.
3. Scholten-Peeters G, Verhagen A, Bekkering G, et al. Prognostic factors of whiplash associated disorders: a systematic review of prospective cohort studies. Pain 2003;104:303–22.
4. Kamper S, Rebbeck T, Maher C, et al. Course and prognostic factors of whiplash: a systematic review and meta-analysis. Pain 2008;138:617–29.
5. Walton D, Pretty J, MacDermid J, et al. Risk factors for persistent problems following whiplash injury: results of a systematic review and meta-analysis. J Orthop Sports Phys Ther 2009;39:334–50.
6. Hayden J, Côté P, Steenstra I, et al. Identifying phases of investigation helps planning, appraising, and applying the results of explanatory prognosis studies. J Clin Epidemiol 2008;61:552–60.
7. Moons K, Royston P, Vergouwe Y, et al. Prognosis and prognostic research: what, why, and how? BMJ 2009;338:b375.
8. Atherton K, Wiles N, Lecky F, et al. Predictors of persistent neck pain after whiplash injury. Emerg Med J 2006;23:195–201.
9. Sterling M, Jull G, Vicenzino B, et al. Physical and psychological factors predict outcome following whiplash injury. Pain 2005;114:141–8.
10. Kasch H, Bach F, Jensen T. Handicap after acute whiplash injury: a 1-year prospective study of risk factors. Neurology 2001;56:1637–43.
11. Pobereskin L. Whiplash following rear end collisions: a prospective cohort study. J Neurol Neurosurg Psychiatry 2005;76:1146–51.
12. Richter M, Ferrari R, Otte D, et al. Correlation of findings, collision parameters, and psychological factors in the outcome of whiplash associated disorders. J Neurol Neurosurg Psychiatry 2004;75:758–64.
13. Gun R, Osti O, O’Riordan A, et al. Risk factors for prolonged disability after whiplash injury: a prospective study. Spine. 2005;30:386–91.
14. Altman DG, Vergouwe Y, Royston P, et al. Prognosis and prognostic research: validating a prognostic model. BMJ 2009;338:b605.
15. Holm L, Carroll L, Cassidy D, et al. Expectations for recovery important in the prognosis of whiplash injuries. PLoS Med 2008;5:e105.
16. Carroll L, Holm L, Ferrari R, et al. Recovery in whiplash-associated disorders: do you get what you expect. J Rheumatol 2009;36:1063–70.
17. Rebbeck T, Sindhusake D, Cameron I, et al. A prospective cohort study of health outcomes following whiplash-associated disorders in an Australian population. Inj Prev 2006;12:93–8.
18. Söderlund A, Lindberg P. Whiplash-associated disorders—predicting disability from a process-oriented perspective of coping. Clin Rehabil 2003;17:101–7.
19. Sterling M, Jull G, Vicenzino B, et al. Characterization of acute whiplash-associated disorders. Spine 2004;29:182–8.
20. Sterling M, Hendrikz J, Kenardy J. Compensation claim lodgement and health outcome developmental trajectories following whiplash injury: a prospective study. Pain 2010;150:22–8.
21. South Australian Centre for Trauma and Injury Recovery. A Clinical Pathway for Best Practice Management of Acute and Chronic Whiplash-Associated Disorders. Adelaide, Australia: South Australian Centre for Trauma and Injury Recovery; 2008.
22. Kasch H, Qerama E, Bach F, et al. Reduced cold pressor pain tolerance in non-recovered whiplash patients: a 1-year prospective study. Eur J Pain 2005;9:561–9.
23. Sterling M, Hendrikz J, Kenardy J. Similar factors predict disability and posttraumatic stress disorder trajectories after whiplash injury. Pain. 2011;152:1272–8.
24. Curatolo M, Arendt-Nielsen L, Petersen-Felix S. Evidence, mechanisms, and clinical implications of central hypersensitivity in chronic pain after whiplash injury. Clin J Pain 2004;20:469–76.
25. Williamson E, Williams M, Gates S, et al. A systematic literature review of psychological factors and the development of late whiplash syndrome. Pain 2008;135:20–30.
26. Walton D. A review of the definitions of “recovery” used in prognostic studies on whiplash using an ICF framework. Disabil Rehabil 2009;31:943–57.
27. Kamper S, Ostelo R, Knol D, et al. Global Perceived Effect scales provided reliable assessments of health transition in people with musculoskeletal disorders, but ratings are strongly influenced by current status. J Clin Epidemiol 2010;63:760–6.
28. Phillips L, Carroll L, Cassidy D, et al. Whiplash-associated disorders: who gets depressed? Who stays depressed? Eur Spine J 2010;19:945–56.
29. Sullivan M, Thibault P, Simmonds M, et al. Pain, perceived injustice and the persistence of post-traumatic stress symptoms during the course of rehabilitation for whiplash injuries. Pain 2009;143:325–31.
30. Elliott J, Pedler A, Kenardy J, et al. The temporal development of fatty infiltrates in the neck muscles following whiplash injury: an association with pain and posttraumatic stress. Plos One 2011;6:e21194.
31. McLean S, Diatchenko L, Lee M, et al. Catechol O-methyltransferase haplotype predicts immediate musculoskeletal neck pain and psychological symptoms after motor vehicle collision. J Pain 2011;12:101–7.

whiplash-associated disorders; prediction; prognosis

© 2011 Lippincott Williams & Wilkins, Inc.