RESULTS OF BEST EVIDENCE SYNTHESIS
Risk
Very little is known about the epidemiology of Whiplash-Associated Disorders (WAD). We did not find any population-based studies that provide estimates of the risk of WAD. Some studies deal with other clinical entities such as cervical disc disease, 52 chronic neck pain 59 or cervical trauma, but do not provide enough detailed information to assess whether the studied patients were suffering from WAD. Another limitation of most accepted studies is that the population under study is seldom the true population at risk of presenting with the clinical entity of interest. Therefore, it is seldom possible to compute risk estimates or to compare groups of people having different characteristics. Overall, we have very little information about the true determinants of the risk for WAD.
The only population-based study on the frequency of WAD comes from insurance data provided by the Folksam Company in Sweden.57 This study of all rear-seat occupants of vehicles involved in a collision during a 19-month period is population based, as it included both damage-only and injury-producing collisions. The paper does not give incidence rates. However, considering all neck injuries reported by people who answered the mail questionnaire and the denominator provided, the rate of neck injury in rear-seat occupants can be estimated to be 1.52% in children up to the age of 14 years and 3.53% in adults. Unfortunately, the paper does not provide the response rate of the questionnaire survey. Another study by Gustafsson et al36 is also based on Folksam data, but deals only with injuries in children. Although numbers of soft-tissue injuries resulting from all car collisions are provided, no rates can be computed.
In another population-based study, Björnstig et al 5 included soft tissue injuries from causes other than whiplash. In this study of all injuries that occurred in one catchment area of northern Sweden during a one-year period, the incidence of neck injuries was 1/1000 inhabitants and occurred more frequently in young males. However, because the report does not provide any information on the number and characteristics of people involved in collisions or other potentially injuring events, we cannot conclude that this information applies to the risk of suffering whiplash injury only. The over representation of young males probably reflects their higher risk of being involved in a collision. Finally, in a study of 886 patients who suffered head trauma, Neifeld et al 72 suggested that neck injuries were more likely to occur in association with head trauma. However, this study looked at major cervical trauma only. Unfortunately, no other reviewed study provides acceptable information on who is injured, when and where, in terms of risk.
Wearing of a seat belt at the time of the collision has been extensively studied as a risk factor for whiplash or soft-tissue injury of the neck. Several studies of the effect of implementing a seat belt law on the patterns of injury suggest that the frequency of WAD is higher after the law has been in effect. 2, 95, 106, 112 Only one study95 estimates both the number of people at risk and the people injured in one geographic area during similar 12-week periods, three years before and one year after the law was implemented in France. Although the study was not a very robust time-series, it provides some evidence that the law increased the incidence of soft-tissue neck injuries. A further limitation of this study is that it only dealt with a regional aspect of the impact of the French law. The study by Bodiwala et al, 7 although accepted for its information on prevalence of neck injuries, was not population based. Other studies reviewed were not based on the total population at risk.
Headrests have also been extensively studied, especially in the biomechanical literature, 109 which was not addressed in our best evidence synthesis. The only accepted study that provides information about the effect of headrests is that by Nygren et al. 75 Although the true population at risk cannot be deduced, there is evidence that the risk of WAD varies with the make of vehicle and whether there is a headrest or not. The role of the type of headrest in this study is inconclusive.
Several studies suggest the type of collision is a risk factor for soft-tissue neck injuries. For instance, whiplash injuries may be more likely to occur in rear-end collisions. 5 Although this seems obvious and compatible with biomechanical models of neck injuries, there are no studies based on an acceptable denominator, thus precluding estimation of relative risks. Therefore, it is difficult to assess the true effect of circumstances supposedly associated with an increase of the frequency of rear-end collisions, such as the use of daytime running lights on the risk of WAD. 21
Diagnosis
As part of its effort to summarize the Consensus of the Task Force, the Methods and Reporting Secretariat met to construct a grid on Diagnostic Methods. The full Task Force had already agreed on the rows (enumerating important diagnostic tests to be considered), the columns (classifying the patients by category of WAD) and the chronologic stage in the natural history of the problem. To our astonishment a diligent attempt to complete the grid resulted in no entries (see Table 5.3.2.1 in the Official Report).
We also contemplated changing the Grid to show the diagnostic tests in the rows and attributes such as predictive value, sensitivity, specificity and acceptability in the columns. These are the attributes that are essential to a scientific and clinically relevant approach to the evaluation of diagnostic tests. We were also unable to find any accepted papers that allowed one single judgement regarding such characteristics of diagnostic tests.
This highlights a substantial gap in a very large area to which we have given a high order of priority, namely the diagnostic tests that provide information essential to cost-effective management of patients sustaining WAD. Accuracy and reliability must be shown and the predictive value of the tests must be confirmed before the acceptability of these diagnostic methods can be documented.
We did accept some studies on diagnosis of WAD. These only address the use of X rays, magnetic resonance imaging (MRI), measurement of cervical range of motion, strength, and neurobehavioral tests. We did not find any accepted articles on other potentially relevant diagnostic procedures.*
Only one paper6 addressed relevant aspects of the use of MRI of the cervical spine. Although the study did not deal specifically with WAD patients, it demonstrates important information about the specificity of the technique. Multiplane magnetic resonance scans were done on 63 asymptomatic volunteers and interpreted independently by three neuroradiologists. The scans were interpreted as demonstrating an abnormality (herniated disc, bulging disc, foraminal stenosis) in as many as 19% of the asymptomatic subjects; the frequency of abnormalities was 28% in subjects aged 40 or older. Consequently, as MRI becomes more popular in the assessment of soft-tissue injuries, it can be expected that its use in patients with WAD will be impeded by this high proportion of false-positive findings.
A single accepted study 76 has addressed behavioral findings in WAD patients. This study compared the results of an array of neuropsychological tests in 34 "whiplash cases" with chronic symptoms, examined six to 12 months after the initial injury, and 21 non-hospitalized patients with chronic pain in the neck and arm but no history of trauma. The groups did not differ regarding gender, age and education. There were no clinically nor statistically significant differences in the results of the neuropsychological tests. The study, however, does not provide any comparative information on subjects without whiplash or with other sources of acute neck pain. Therefore, we cannot draw conclusions concerning the potential usefulness of neurobehavioral tests in the evaluation of WAD patients.
Accepted papers on history or physical examination dealt with only two aspects: the development of a neck disability index and the development of rules or algorithms for predicting the presence of bony-tissue injuries. No accepted studies have dealt with the use of the medical history or physical examination to make a diagnosis of WAD.
The neck disability index (NDI) has been developed to help classify patients with WAD. 115 The NDI includes ratings of pain intensity, personal care, lifting, reading, headaches, concentration, work, driving, sleeping and recreation. Its reliability and validity have been assessed in only 48 subjects, 70% of whom had sustained whiplash injury within the past four to six weeks, and two within one half year. This study suggests good test-retest reliability, good internal consistency and construct validity. The authors show good concurrent validity, as assessed by correlations of the index and a visual analogue scale rating improvement in activity. Reliability and validity of the NDI needs further study in larger samples of patients, and an assessment of the validity of the index for predicting recovery from whiplash injury.
Seven studies provide information on the value of the history or the physical exam to predict the presence of bony-tissue injury, 43, 47, 53, 68, 72, 87 including one study in children. 48 The objective of all these studies was to classify patients seen in an emergency room or a trauma center into those who do and do not need cervical X-ray examination. All of the studies were made in selected populations that did not represent the spectrum of patients relevant to the study of whiplash injury; specifically, the samples were skewed toward the most severe patients. These studies indicate that a prediction algorithm with high sensitivity but low specificity can be achieved. Two of them suggest that X rays are not necessary in patients who are alert and do not complain of neck symptoms. 53, 68 Further, Neifeld et al 72 and Roberge et al 87 suggest that X rays are not necessary in patients who are alert and have no neck tenderness on physical examination, even if they have cervical pain. On the other hand, the study by Jacobs and Schwartz 47 suggests that clinical prediction is not sufficient to rule out the presence of bony tissue injury.
A study by Hoffman et al 43 also assessed criteria for ordering X rays. The authors suggest that X rays are not necessary for patients who are alert, are not intoxicated, have an isolated blunt trauma and have no neck tenderness on physical examination. However, the results of this study must be taken with some caution. The estimate of "no false negatives" is based on 27 patients with fractures, who had at least one of the following four characteristics: mid-line neck tenderness, evidence of intoxication, abnormal level of alertness, or several painful injuries elsewhere, and who would have had X rays using the decision criteria. However, this is a small number of fracture patients upon which to base a conclusion. Moreover, the study was carried out on 974 patients receiving cervical spine X rays for blunt trauma, but only 283 had a whiplash mechanism without a direct blow to the head or neck. The authors report that none of these latter patients had a fracture (0/283), but do not provide confidence intervals. This study also suggests that false-positive findings are frequent on X rays.
The study by Jaffe et al 48 was done in a select sample of 206 children including 59 (29%) with bony injury of the cervical spine. The authors propose indications for X rays, including history of neck trauma, neck pain, abnormal sensation, abnormal reflexes, limitation of neck mobility, neck tenderness, abnormal strength and abnormal mental status. However, the authors did not attain perfect sensitivity, even when the specificity was very low.
Other studies on the use of X rays in cervical spine injury patients did not address the issue of validity of diagnosis of these injuries. These studies, however, provide some useful information about the reliability and variability of X ray imaging or measurement. The study by Annis et al1 suggests that interpretation of X rays is dependent on the level of training of the interpreter. The study by Sistrom et al 96 illustrates that most of the variation in usual measures of soft tissue on cervical X rays is related to variations in age and gender of the subjects, rather than variations in symptomatology and objective findings. Studies by Templeton et al111 and Young et al 123 further suggest that there are a number of false-positive X-ray findings in WAD patients, both in the measurement of soft tissue and the laminar space. The study by Freemyer et al 99 suggests that there is no benefit of adding a supine oblique view to the three view series used in emergency rooms. The study by Dvorak et al, 20 although specifically done on whiplash patients, does not provide useful information about X rays; there are not enough details in the paper about the type of manifestations nor about the comparison with the control group of healthy adults to make conclusions on the predictive validity of range of motion assessment on X ray in whiplash patients. Another study by Jackson et al 46 suggests that the use of a marking system on X ray is a reliable way to assess the upper cervical spine; unfortunately the paper does not provide any information on the study population, thus precluding our using this information for WAD patients.**
Devices to measure range of motion of the cervical spine have been assessed by Foust et al, 30 Rheault et al 86 and Youdas et al. 122 All these studies were reliability assessments, carried out on healthy subjects 30 or a group with very few whiplash patients. 86, 122 The reliability varies considerably depending on the direction of movement. Furthermore, gender and age are major sources of variation in the measurements. None of these studies address issues of validity of the devices to predict clinical findings or the course of the disorder. Variation due to sources other than clinically relevant variables is also reported by Vernon et al, 116 in a study of the modified dynamometer to evaluate neck muscle strength in whiplash patients.
Barnsley et al 3 studied single versus comparative blocks for diagnosing zygapophysial joint pain in patients with chronic WAD. Subjects were randomly assigned to either short-acting (lignocaine) or long-acting (bupivacaine) injections on the first block and the complementary anesthetic on the second block. A concordant response was defined as shorter-duration relief from the lignocaine than from the bupivacaine. Only 34 to 45 subjects who had pain relief from the first block had a concordant response to the second injection. Because of the high proportion of discordant responses, this study suggests that positive response to a single nerve block is not adequately predictive of consistent response. Further evaluation of single versus comparative nerve blocks on substantially larger samples of subjects is needed.
A recent study by Helliwell et al 38 also suggests that a finding of a straight cervical spine on X ray is not specific to muscle spasm caused by pain.
