Sheilah Hogg-Johnson, PhD, Gabrielle van der Velde, DC, Linda J. Carroll, PhD, Lena W. Holm, DrMedSc, J. David Cassidy, DC, PhD, Jamie Guzman, MD, MSc, FRCP(C), Pierre Côté, DC, PhD, Scott Haldeman, DC, MD, PhD, Carlo Ammendolia, DC, PhD, Eugene Carragee, MD, FACS, Eric Hurwitz, DC, PhD, Margareta Nordin, PT, DrMedSc, Paul Peloso, MD, MSc,FRCP(C)
Institute of Environmental Medicine,
STUDY DESIGN: Best evidence synthesis.
OBJECTIVE: To undertake a best evidence synthesis on the burden and determinants of whiplash-associated disorders (WAD) after traffic collisions.
SUMMARY OF BACKGROUND DATA: Previous best evidence synthesis on WAD has noted a lack of evidence regarding incidence of and risk factors for WAD. Therefore there was a warrant of a reanalyze of this body of research.
METHODS: A systematic search of Medline was conducted. The reviewers looked for studies on neck pain and its associated disorders published 1980-2006. Each relevant study was independently and critically reviewed by rotating pairs of reviewers. Data from studies judged to have acceptable internal validity (scientifically admissible) were abstracted into evidence tables, and provide the body of the best evidence synthesis.
RESULTS: The authors found 32 scientifically admissible studies related to the burden and determinants of WAD. In the Western world, visits to emergency rooms due to WAD have increased over the past 30 years. The annual cumulative incidence of WAD differed substantially between countries. They found that occupant seat position and collision impact direction were associated with WAD in one study. Eliminating insurance payments for pain and suffering were associated with a lower incidence of WAD injury claims in one study. Younger ages and being a female were both associated with filing claims or seeking care for WAD, although the evidence is not consistent. Preliminary evidence suggested that headrests/car seats, aimed to limiting head extension during rear-end collisions had a preventive effect on reporting WAD, especially in females.
CONCLUSION: WAD after traffic collisions affects many people. Despite many years of research, the evidence regarding risk factors for WAD is sparse but seems to include personal, societal, and environmental factors. More research including, well-defined studies with accurate denominators for calculating risk, and better consideration of confounding factors, are needed.
From the Full-Text Article:
Main Findings on the Burden of Neck Pain
In this best evidence synthesis on the epidemiology of neck pain we found that neck pain, like other musculoskeletal conditions, is common in the general population. Although neck pain is common, when we start to qualify it (i.e., by duration, by frequency, by intensity, by whether health care was sought) we see the typical “iceberg of burden” that has been reported for other musculoskeletal conditions. That means we observe many cases of some pain, but fewer cases of any significant duration, fewer cases that lead to utilization of the health care system, and fewer still that are disabling.
Neck pain is experienced by people of all ages, including children and adolescents. [27, 53, 97, 98, 100, 102-110, 112] There is consistent evidence that previous neck pain or trauma is predictive of both incident and prevalent neck pain, suggesting that neck pain often follows an episodic course similar to low back pain.
Risk Factors and Factors Associated With Neck Pain
A variety of risk factors and associated factors have been considered, with key findings highlighted here. Here we compare our findings with those in other chapters of this best evidence synthesis.
The evidence on whether age is a risk factor for incident neck pain is equivocal. However, as with other musculoskeletal conditions,  most data show that prevalence of neck pain increases with older age, peaking in the middle years and declining in later life. This should be interpreted in light of findings from general population studies that younger persons with neck pain have a better prognosis,  and it may be this factor, rather than differences in incidence rates, which “drives” the relationship between age and prevalent neck pain.
There was consistent evidence that neck pain coexists with other health problems, including other musculoskeletal complaints like low back pain, headache and poorer self-rated health. There was also generally consistent evidence that neck pain is both predicted by and coexists with different types of psychological health conditions. Of note is that our best evidence synthesis of prognostic factors for neck pain in the general population also identifies health problems, musculoskeletal complaints and poor psychological health as prognostic of poor outcome. 
We did not identify any evidence demonstrating that disc degeneration is a risk factor for neck pain. Longitudinal studies are required to demonstrate disc degeneration as a risk factor, and there were no such studies judged scientifically admissible. This is an important finding to note, given the existing body of literature based on the assumption that persistent and disabling neck pain is associated with cervical degenerative changes. [115-118 ]The lack of evidence identified here is consistent with findings reported elsewhere in the Neck Pain Task Force report around the use of diagnostic imaging.  Nordin et al  also concluded there is “no evidence that common degenerative changes on cervical MRI are strongly correlated with neck pain symptoms” and that “common degenerative changes in the cervical spine identified by MRI are at best fair to moderately reproducible.”
Modifiable Risk Factors and Prevention of Neck Pain
Most of the identified literature addressed nonmodifiable factors such as gender, a history of neck pain and genetics. Poor psychological health was identified as both a risk factor and an associated factor for neck and low back pain,  although to date, there is no evidence that treating psychological conditions will lead to reductions in neck pain or other musculoskeletal complaints.
Smoking was investigated as a risk factor or associated factor in several studies, [17, 22, 47, 62, 71, 81, 84, 102, 108, 109] with many studies reporting no statistically significant relationship with neck pain. However, in 5 studies [22, 62, 81, 84, 102] low-magnitude relationships between smoking and neck pain were identified and reported. Exposure to second-hand smoke during childhood was also identified as a risk factor for neck pain later in life.  Thus, attempts to reduce smoking in general may have potential benefits for preventing neck pain.
The evidence on exercise and physical activity varied. Many studies showed no association, although most of these relied on a single summary self-report measure of activity participation. One study with a more thorough assessment of activity participation suggested that certain levels of participation provided a protective effect against neck pain.  Although this study must be interpreted with caution (because of the possibility of prevalence-incidence bias), their findings generally support evidence cited elsewhere in this Neck Pain Task Force report; that exercise was found to be an effective intervention for neck pain,  and that workers who exercised had a better prognosis for recovery from neck pain,  although preliminary evidence suggests that general physical activity is not associated with prognosis in the general population. 
The State of the Literature
Our search identified 469 articles that were relevant to our investigation of neck pain epidemiology, and 249 were judged scientifically admissible and 101 of these studies related to the burden and determinants of neck pain in the general population. These articles drew on results from 86 different groups of study participants conducted in countries around the world.
The accepted studies varied widely in their design:
Most of the scientifically admissible studies (52 of 86) were cross-sectional and provided evidence about prevalence and/or factors associated with neck pain.
There were also 27 cohort studies, although for our purposes, 3 of these were analyzed in a cross-sectional manner. [56, 97, 103, 104] Of the remaining 24 cohorts, 5 produced estimates of incidence rates only, and 19 examined risk factors for neck pain or injury. Of these 19 studies, 13 were Phase I investigations; there were only 2 Phase II and 4 Phase III investigations.
In addition, there were 3 Phase II and one Phase III case-control studies which examined risk factors for neck pain or neck injury.
Finally, there were 3 twin studies examining the inheritability of neck pain.
Cross-sectional studies investigating prevalence are useful starting points for understanding the burden of neck pain. But in order to advance this field of study, we need more high-quality cohort studies and case-control studies to identify and investigate both risk and protective factors. Assembling cohorts of people before any episodes of neck pain may prove difficult or inefficient, given the reports of high prevalence seen in children and adolescents. [27, 53, 97, 98, 100, 102-110, 112] However, the evidence points to an episodic, recurrent course for neck pain,12, 17, 34, 56, 62, 63, 78, 81, 114 much like other musculoskeletal conditions. [122-131] Therefore, studies integrating investigations into risk and prognosis are likely to be most fruitful.
The studies which comprise our best evidence synthesis on burden and determinants of neck pain in the general population showed geographically distinct patterns of distribution. All but 2 of the 15 sports-related studies were based in North America. The remaining studies were predominantly from northern European countries (Scandinavia, Netherlands and the U.K.). Some more varied geographical distribution appeared in the cross-sectional studies with studies from Australia, Asia and the Middle East, partially due to the Community Oriented Programme for Control of Rheumatic Disease (COPCORD), an initiative first planned by the International League of Associations of Rheumatology (ILAR) and the World Health Organization (WHO) in 1981 to assess the global burden of rheumatic diseases. [132, 133] Still, more than half of the cross-sectional studies were from northern Europe.
The importance of “case definition” in public health and clinical inquiry have been duly noted in the literature, [134-140] particularly for symptom-based conditions; this has also been true of other methodologic issues when it comes to measuring prevalence.113 In out exploration of neck pain epidemiology and risk, we saw that summarizing the evidence and making comparisons across populations and settings was especially challenging due to extensive between-study variations in the case definitions of “neck pain/neck injury.” This has led the Neck Pain Task Force to propose a framework for case definitions.  We found that case definitions and sampling frames used by researchers varied in many different ways:
Some studies only included specific conditions such as prolapsed disc [22, 42] or disc protrusion or herniation resulting in radicular syndrome.  Others included any self-reported pain in the neck region.
Some studies included only those people with neck pain who were presenting for health care; others surveyed general populations and generally yielded higher estimates of neck pain incidence or prevalence.
The actual location of symptoms also varied. Some investigators only considered pain felt in a very clearly delimited part of the body [24, 78]; others included “neck/ shoulder” pain within their case definition. Generally, the more extensive the area(s) of the body, the higher the prevalence of neck pain. 
There were also differences in the period under consideration for measures of prevalence (from point prevalence to lifetime) and also in the frequency, duration, and severity of symptoms captured by different case definitions, as previously described in a framework presented by Beaton et al  Some case definitions also specified pain with disability or interference with activities of daily living.
Generally, the fewer conditions placed on duration, period, frequency and intensity, the higher the incidence or prevalence of neck pain. We also noted that some studies were specifically focused on neck pain, and so only neck pain statistics were gathered, whereas other studies included information on a wide variety of musculoskeletal complaints in different parts of the body.
Typically, estimates of neck pain prevalence were lower in the broader investigations. The unified framework for case definitions proposed elsewhere in the Neck Pain Task Force report,  if used consistently by researchers, would provide a flexible and comprehensive set of approaches to case definition and facilitate future literature synthesis.
Limitations of Our Review
Our findings on the epidemiology and risk factors for neck pain in the general population are, of course, limited by the available literature. There were far fewer studies of incidence and risk in neck pain compared to studies looking at prevalence and associated factors. We found limited evidence on risk factors and even more limited evidence on “modifiable” risk factors. No admissible evidence was found for some categories of factors, such as societal level factors and cultural factors. For practical reasons, we restricted our literature search to articles appearing in Medline; we only searched for pub- lications in English, French and Swedish; and we limited publication to a specific time frame (between 1980 and 2005). Thus, it is possible that we might have missed relevant literature not indexed by these databases or which fell outside these restrictions. 
Neck pain is common in the adult general population, with typical 12-month prevalence estimates from 30% to 50%. Among children and adolescents, 12-month prevalence estimates range from 21% to 42%.
Neck pain which limits activities is less common, with 12-month prevalence estimates ranging from 2% to 11%.
There is no evidence to support the assumption that degenerative disc changes are a risk factor for neck pain without radiculopathy.
Poor psychological health is a risk factor for neck pain and is often associated with it.
Helmet use during activities such as bicycling, skiing, and hockey may reduce some types of injuries without increasing the risk of neck injury.