Medicolegal causation analysis of a lumbar spine fracture following a low speed rear impact traffic crash

Manuscript type : Case Report | Article Date : 2015/02/20

  • Autors

    1. Freeman Michael D.
  • Abstract

    Epidemiologic study indicates that low-speed traffic crashes are associated with a relatively low risk of significant injury. This fact is often used to defend legal actions in cases in which it is alleged that a significant injury has resulted from a low speed crash. Injury causation following a low-speed collision can only be assessed by comparing the risk of injury from the crash to the risk of the injury occurring at the same point in time, but absent the exposure to the crash. As a practical matter in a clinical setting, this means that a history of close temporal proximity between the collision and the onset of symptoms is the strongest indication of causality, irrespective of a low injury risk associated with the collision. In the present case study an acute superior endplate fracture and bilateral facet fracture was observed in the lower lumbar spine in a 49-year-old female occupant of a vehicle that was struck from behind at low speed. Injuries to the patient’s neck and shoulder were ultimately diagnosed as intervertebral disk derangements and glenoid labrum tear, respectively. As a result of the injuries the patient underwent four spine surgeries and an arthroscopic procedure on her shoulder. Various experts for the insurer asserted that the injuries and subsequent need for treatment were unrelated to the collision, in part based on the low risk of such injuries from a low speed crash. The methodologic and logical errors committed by the insurer’s experts in arriving at these conclusions are discussed.
  • Description

    INTRODUCTION

    The assessment of injury causation following a low speed traffic crash is often a contentious issue, for several reasons. The most prominent reason is related to the fact that the majority of low speed collisions consist of aligned rear impact crashes [1], and that the occupant most likely to be injured in such a collision is in the vehicle that is struck from the rear [2]. Thus, the occupant most likely to be injured is the one who is least likely to be at fault for the collision. The legal claim for compensation by the injured party triggers a response from the insurer of the at-fault driver, often accompanied by expert opinions disputing the cause of the injuries. A common approach to the defense of low speed crash-related injury claims is based in an analysis of the crash severity and associated biomechanical properties of the crash [3]. Expert opinions stemming from such an analysis are typically based on the precept that the low severity of the crash indicates a low risk of significant injury, which is in turn used as corresponding or “counterfactual” evidence that there is a high probability that the injury did not occur in the crash [4]. Another form of defense expert opinion is based on an analysis of the medical evidence, including MRI and other imaging. The lack of clear evidence of acute traumatic injury, such as fracture, hemorrhage, or bony edema, is used as a basis from which to infer that there has been no injury-producing trauma, and thus the symptoms must be related to pre-existing conditions, such as age-related degenerative changes. Injury causation assessment in low speed collisions can be challenging. The kind of traumatic forces resulting from lower speed crashes are “intermediate;” they are greater than the trivial level of trauma seen in daily activities but are not at a level at which injury occurs in the majority of occupants. An additional difficulty stems from the fact that the types of injuries commonly associated with lower speed crashes, ranging from short-lived spinal strain injuries to chronic widespread pain and symptomatic derangement of intervertebral disks [5, 6], are the same types of conditions that can have an insidious onset, or be triggered by a trivial trauma, and thus have a low specificity for a traumatic etiology. The present case study concerns an unusual case of a low speed rear impact collision that resulted in a lumbar vertebral body fracture, as well as other significant injuries.

    CASE REPORT

    The patient was a 49-year-old previously healthy female restrained driver of a 2006 Volkswagen Golf that was struck in the rear by a 2003 Chevrolet full-sized pickup truck. The only damage observed in the Volkswagen was to the license plate frame, and there was no apparent damage to the pickup truck. The patient had an immediate onset of neck and back pain, and could not get out of the vehicle because of complaints of pressure and pain in the low back, as well as pain radiating into both legs. The patient was immobilized by emergency medical service providers and taken to the hospital for further evaluation. Plain radiographs taken at the hospital demonstrated an anterior superior vertebral body fracture at L5 (Figure 1), an injury that was later confirmed as acute by MRI (Figure 2), which demonstrated edema of the adjacent marrow. Serial MRI examinations demonstrated progressive healing of the fracture over the following 2.5 years (Figures 3-5). Because of persisting symptoms in the neck, right shoulder, and lumbar spine, the patient ultimately underwent 5 surgeries, including 1) an instrumented fusion for mechanical instability associated with bilateral facet fractures (confirmed intra-operatively) at L4-5, 2) removal of the L4-5 hardware for suspected pseudoarthrosis and/ or loose hardware, 3) a C3-4 anterior cervical discectomy and fusion with allograft, 4) a C5-6 discectomy and fusion with an intervertebral implant, and 5) an arthroscopic decompression of the subacromial space and repair of the glenoid labrum of the right shoulder. The patient made a legal claim against the insurer for her injuries and treatment following the collision, and the insurer obtained expert opinions that indicated the following: 1. The collision-related speed change of the Volkswagen was no more than 6 km/h (from an engineering expert); 2. A collision related speed change of 6 km/h has been tolerated more than 500 times in experimental volunteer crash tests, and therefore the risk of injury to the patient in the crash was less than 1 in 500 (from a biomechanical engineering expert). The expert also opined that there was “no mechanism for injury” in the crash; 3. There was no fracture of the L5 superior endplate; the finding was a pre-existing Schmorl’s node (from a radiologist); 4. All of the surgeries were performed on degenerative conditions that pre-existed the crash (from an orthopedic surgeon).

    DISCUSSION

    The subject case presented an unusual combination of conditions, in that an intermediate trauma was closely associated with injuries with a high degree of specificity for a traumatic etiology; an acute superior endplate fracture of L5 (confirmed with serial imaging studies) and a bilateral facet fracture at the same level (observed intra-operatively). Outside of an injury litigation setting, there may have never been a need for a determination of the cause of the patient’s post-collision diagnoses and need for treatment. All of the physicians providing care for the patient gave the opinion that the cause of the spine and extremity conditions requiring surgery was the crash that preceded the symptoms. In comparison with the opinions provided by the 4 experts for the insurer, the causation opinions of the treating physicians appeared imprecise and haphazard. When asked, the clinicians were unable to give an opinion regarding the severity of the collision, other than noting that it appeared to be at low speed, and they were unable to provide any imaging evidence that the fractures and intervertebral disk and shoulder injuries did not pre-exist the crash in some form. All of the treating physicians conceded that there were degenerative changes in the patient’s spinal disks and shoulder that pre-existed the crash. Inherent in the opinions of the experts for the insurer and the criticisms of the treating physicians’ causation opinion are a lack of both a systematic approach to causation and a logically reasoned assessment of the evidence. The following discussion is an explanation of this conclusion. Injury Causation Using a Comparative Risk Approach Causal assessment of crash-related injuries in individuals (specific causation) follows the general precepts originally set forth by Bradford-Hill in 1965, described as 9 “viewpoints” or criteria for assessing whether an observed association is causal [7]. These criteria have been modified and adapted to a wide range of general and specific causal assessments, including for crash-related injuries to the spine and spinal disks [8, 9]. These criteria can be grouped into the following categories: Plausibility The threshold for a causation determination is the plausibility of the association. This simply means that it first it must be determined that it is possible for the injury to result from the collision. Plausibility of injury is a very low threshold to meet for virtually all crashes; even an injury that only occurs in 1 in 100,000 occupants is still plausible. Further, there is no reliable evidence that absolute minimum load thresholds are established for most types of injuries [10]. As an illustration of this fact, injuries to the intervertebral disks have been described from forces that are substantially less than those of a minimal damage crash; including riding a roller coaster, sneezing, and from therapeutic manipulation [11-13]. Even femur fractures have been documented as occurring in traffic crashes for which the biomechanical probability of such fracture was deemed to be 0% [14]. Plausibility is a tautological construct; by this it is meant that for a given crash any injury is either plausible (possible), or implausible (impossible). The conclusion that an injury is improbable (unlikely) does not make the injury implausible. Implausibility is typically only going to be clearly established when a relationship violates a basic biological principle. As an example, leukemia is not caused by mechanical trauma, regardless of the proximity between the discovery of the disease and a predicate trauma. Because of these concepts, for virtually all crash related injuries (regardless of crash severity), plausibility is established by the fact that there is a lack of reliable evidence of implausibility of the relationship [9]. Temporal Association Once plausibility is established then the temporal proximity between the crash and the first indication of the injury can be assessed from the history and medical records. As a general principle, the closer the symptom onset is to the collision in time the more likely it is that the crash is the cause of the injuries underlying the symptoms. Clinical reasoning and experience is used to assess when the latency between a crash and the onset of symptoms is too great to relate the two, but caution should be exercised in such assessments. As an example, it is difficult to imagine an injury that could result from a crash that would not manifest symptomatically in some fashion until 3 months after the collision. On the other hand, merely asserting that all injuries must be apparent within the first 72 hours etc. of a traffic crash with no basis in epidemiologic study or general clinical consensus is not a valid basis for rejecting an apparent causal relationship. Competing Contemporaneous Cause of Symptoms/ Injury The pre-crash condition of the individual must be assessed in order to determine if the patient’s condition was such that the occurrence of symptoms of injury were likely to have occurred at the same point in time regardless of the occurrence of the crash. If this is the case then the collision is not causally related to the post-crash diagnoses. On the other hand, if there was little or no indication that patient was likely to spontaneously develop the same symptoms at the same point in time in the absence of the collision, then the crash becomes the most probable and proximate cause of the injury (assuming a reasonable temporal association). Relative/ Comparative Risk of Relationship Temporal proximity between a crash and the onset of symptoms is the strongest indication of causality, even when the risk of injury from the collision is thought to be quite low, as often occurs when the collision severity is minor. The following comparative risk example illustrates this principle [15]. If a patient presents with persisting symptoms of a cervical disk injury beginning the day of a minimal damage rear impact collision, it can be grossly estimated that the risk of such an injury is no less than 1 in 100 [16]. In comparison, for a middle-aged individual with an unremarkable prior history of symptomatic cervical spine problems, the annual risk of the spontaneous onset of persisting cervical spine pain is less than 1% [17]. Thus, in rough terms, the risk of a persisting injury from the collision is more than 1 in 100, and the chance that the same condition would have started on the same day as the collision is less than 1 in 3,650 (1% [1 in 100] annual risk divided by 365 days); a ratio of nearly 37 to 1 favoring the crash as the cause of the injury. If the symptoms indicative of injury had been evident within an hour of the collision, the risk from the crash would stay the same (1 in 100), but the competing risk of spontaneous onset of symptoms would decrease commensurate with the time, to less than 1 in 87,600 (1% annual risk divided by 8,760 hours in a year). In this scenario the crash would be favored as the cause of the injury by nearly 900 times. Case Study Injury Causation Analysis The comparative risk approach described herein is not necessary for the assessment of the vertebral fractures observed in the patient described in the case study. This is due to the high degree of specificity of fracture of a health bone for a traumatic event, and the exceedingly close temporal relationship between the symptom onset indicative of fracture and the only known traumatic event affecting the patient, which was the collision. Thus, despite the low injury risk of the collision, the competing risk of the injury occurring spontaneously at the same time as the crash would have been nearly zero. This approach to causation is based on “counterfactual” reasoning, where the numerator risk is known to be small and difficult to define, but the denominator risk can be defined as exceedingly remote [15]. By the use of this approach the investigated cause can be deemed the only reasonably plausible cause because the only alternative explanation for the injury, which is the spontaneous occurrence of the fracture at the same time as the collision, is essentially implausible. With regard to the non-fracture injuries diagnosed in the patient’s cervical spine and shoulder following the collision, these injuries could be analyzed using the comparative risk approach described earlier. Thus, the close temporal relationship between the collision and the onset of symptoms and the low probability that any pre-existing age-related degenerative change would spontaneously convert to a symptomatic condition at the same time as the collision is weighed against the minor severity of the collision. Added to this calculus is the reasonable assumption that if the collision had the capacity to cause a spinal fracture in a healthy lumbar vertebra, then it had the capacity to cause symptomatic disk herniation in the cervical spine, as well as a symptomatic internal derangement of the shoulder. It is reasonable to assume that these injuries could occur in the patient at a lower load than would be required to cause a spinal fracture. Therefore, if the collision had the capacity to cause the fracture it also had the capacity to cause the observed intervertebral disk and glenoid labrum injuries, inter alia. The biomechanical loading of a rear impact collision on both the cervical spine and the shoulder, which includes shear, compression, and rotational forces [18], combined with the lack of any evidence of implausibility of the injury, serve as a competent explanation for the observed injuries. An additional application of the counterfactual approach to causation is to assess the assumptions necessary to accept the competing explanation for injury (that the injuries occurred spontaneously) in preference to the explanation provide by the collision. In order to accept that the injury occurred spontaneously one must first accept that the patient was so exceptionally fragile that ordinary daily forces could cause her injury, a reasonable possibility. In order to accept this explanation preferentially over the collision as the cause of the injuries, however, it would have to be assumed that the patient’s fragility did not make her more susceptible to injury from the collision, which is a nonsensical conclusion. A rejection of such fallacious reasoning results in a rejection of the spontaneous occurrence explanation for the observed injury, separate from any comparative risk assessment. Analysis of opinions by insurer experts The following analysis demonstrates the lack of relevancy of the opinions provided by the insurer’s experts regarding the cause of the patient’s injuries: 1. The collision-related speed change of the Volkswagen was no more than 6 km/h; Discussion: Irrespective of the accuracy of the crash severity analysis, the injury risk of the collision cannot be concluded to be zero for the reasons discussed earlier in this paper. Thus, the only purpose of the reconstruction of the collision severity in this context (i.e. in aid of the defense against the claim that the fracture and other injuries were caused by the crash) is to conclude that the injury risk of the collision to the general population is low, a point that is readily conceded to and accounted for in a systematic approach to an injury causation assessment. In the circumstances of the facts of the case study, the reconstructed speed change of the collision serves little to no purpose in an injury causation analysis. 2. A collision related speed change of 6 km/h has been tolerated more than 500 times in experimental volunteer crash tests, and therefore the risk of injury to the patient in the crash was less than 1 in 500. This serves as an indication that was “no mechanism for injury” in the crash; Discussion: As mentioned above, the absolute risk of injury from the collision is of little use in a specific injury causation assessment. Further, the derivation of a less than 1 in 500 risk of injury in the general population based on the fact that 500 crash tests have not produced the same level of injury is fallacious in several regards. First, volunteers for crash tests are not representative of the range of injury susceptibility of the general population, as they are typically young, healthy, and robust males who are prepared for an impact. In contrast, Real world crashes involve occupants who are typically not prepared for the crash, often out of ideal seating position or in a rotated or awkward position, with prior history of injury or other health problems, and a variety of other factors that make them more susceptible to injury than a selected and prepared individual who is sitting in an ideal position in a vehicle waiting for an impact of a known severity [19]. The automotive engineering community has rejected the use of data gathered from the results of volunteer testing as a basis for setting minimum injury thresholds for these reasons [10]. To compound the error the biomechanical engineering expert failed to note that the 500 tests were conducted on fewer than 100 volunteers. The fact that 100 mostly male young volunteers were able to withstand a collision of similar severity as the subject crash without sustaining the same injury as the middle-aged female patient is not evidence of anything that is relevant to an investigation of the cause of the patient’s injuries. The information stands only as evidence that it is possible to be exposed to a 6 km/h rear impact collision without sustaining injury. Epidemiologic study indicates that approximately 75% of the general population exposed to a 6 km/h rear impact collision will not be injured to any degree [16]. This information is unhelpful in identifying the 25% of the population who is injured to some degree in such collisions. The expert’s conclusion that, because injury was unlikely in the collision there was “no mechanism of injury,” is likewise a result of fallacious reasoning. The conclusion is akin to stating that “injury was impossible in the collision” without any supporting information or data, except the erroneous conclusion that 500 volunteer tests serve as an indication that the risk of injury was less than 1 in 500. The transmutation from “injury is unlikely” to “injury is impossible” with no additional information is semantic rather than material. 3. There was no fracture of the L5 superior endplate; the finding was a pre-existing Schmorl’s node; Discussion: A Schmorl’s node is the result of the herniation of the nucleus pulposus of the intervertebral disk through the cartilaginous and bony vertebral endplate and into the vertebral body [20]. Most often the finding is incidental, asymptomatic, and is of a developmental rather than traumatic etiology. In come cases, however, Schmorl’s nodes result from an axial load on the vertebral body, and can be acute and painful [21]. The radiology expert’s conclusion that the Schmorl’s node must have pre-existed the collision was justified with the incorrect assertion that all Schmorl’s nodes are acquired atraumatically. A traumatically acquired endplate fracture can heal with the appearance of a Schmorl’s node, and therefore the two descriptions are the same entity described at different times. Further, the contemporaneous finding of vertebral body edema adjacent to the endplate fracture that resolved over time provided ample pathophysiologically consistent evidence of an acute and likely painful injury occurring near the time of the collision (Figures 2-5). Finally, the expert’s conclusion was at odds with the history of the injury, and further disproven by the intraoperative discovery of an incompletely healed facet fracture at the same vertebral level. 4. All of the surgeries were performed on degenerative conditions that pre-existed the crash. Discussion: This assertion is both semantically correct and meaningless with regard to the cause of the patient’s need for multiple surgeries. Age related degenerative changes in the spinal disks and joints are a common finding in both asymptomatic and symptomatic populations. In one study of an asymptomatic population of men and women with an average age of 48 years, MRI evidence of degenerative changes of the neck and low back was present in nearly 80% of the subjects [22]. While there is little to no correlation between degenerative changes of the spine and pain [23], such changes do make the intervertebral disks more susceptible to injury in the event of a sudden traumatic load [24]. Since degenerative changes can be asymptomatic or symptomatic, and only patients with symptoms will seek and undergo treatment, then it is the conjunction of both degeneration and symptoms that lead to a patient undergoing spine surgery or other treatments. The conclusion by the insurer’s orthopedic surgeon that the surgical treatment addressed only spinal or shoulder conditions that pre-existed the collision was thus fallacious; the surgical treatment the patient underwent addressed symptoms that did not pre-exist the collision. The patient was not symptomatic prior to the collision, and there was no indication that the pre-existing age appropriate and asymptomatic degenerative changes in her spine and shoulder were likely to convert to a symptomatic state at any time in the foreseeable future, absent a trauma. Given that there were no pre-crash MRI studies of either the spine or shoulder, there is no indication of whether the collision produced morphologic changes detectable on the post-collision scans. Absent the occasion when a traumatic injury results in acute findings detectable on MRI (edema, hemorrhage, fracture, etc.), such imaging has virtually no discriminatory power to differentiate between painful and non-painful tissue [25]. The greatest failing of this expert’s opinion is evident from the counterfactual perspective; if it is true that the only conditions in the patient that required surgery were those that were present prior to the collision, then it also must be true that had the collision not occurred the surgeries would have still been necessary at the same point in time. The prior discussion regarding the comparative risk of causation given the salient facts of the case study demonstrated the exceedingly low probability of this explanation.

    CONCLUSION

    In the present case study an unusual case of spinal fracture is described as a result of a low speed rear impact collision. A systematic approach to causation, based on the tenets of the Hill criteria, indicated that the only likely cause of the injuries diagnosed in the patient was the collision. Despite the undeniable evidence of the medical imaging demonstrating acute injury from the time of the collision and the lack of any plausible alternative explanation for the symptoms and diagnoses, the insurer was nonetheless able to produce experts who disputed the cause of the patient’s injuries. An examination of the methods and opinions of the insurers expert revealed that they were lacking in a logical, factual, and scientifically valid basis. The present case study serves as an illustration of the range of injuries potentially resulting from low speed traffic crashes, including spinal fracture, symptomatic disk derangement, and internal derangement of the shoulder. From a clinical judgment perspective, the fact that such injuries are relatively unusual in such collisions is largely meaningless when assessing the causation of the injuries. Temporal proximity between the collision and the onset of symptoms related to the diagnosed injury serves as the strongest index of causality for the clinician to rely on.

  • Reference

    1. 1.Styrke, J., et al., A 10-year incidence of acute whiplash injuries after road traffic crashes in a defined population in northern Sweden. PM R, 2012. 4(10): p. 739-47.
    2. 2.Croft, A.C., M.T. Haneline, and M.D. Freeman, Low speed frontal crashes and low speed rear crashes: is there a differential risk for injury? Annu Proc Assoc Adv Automot Med, 2002. 46: p. 79-91.
    3. 3.Walz, F.H. and M.H. Muser, Biomechanical assessment of soft tissue cervical spine disorders and expert opinion in low speed collisions. Accid Anal Prev, 2000. 32(2): p. 161-5.
    4. 4.Schmitt, K.U., et al., Whiplash injury: cases with a long period of sick leave need biomechanical assessment. Eur Spine J, 2003. 12(3): p. 247-54.
    5. 5.Curatolo, M., et al., The role of tissue damage in whiplash-associated disorders: discussion paper 1. Spine (Phila Pa 1976), 2011. 36(25 Suppl): p. S309-15.
    6. 6.Van Oosterwijck, J., et al., Evidence for central sensitization in chronic whiplash: a systematic literature review. Eur J Pain, 2013. 17(3): p. 299-312.
    7. 7.Hill, A.B., The Environment and Disease: Association or Causation? Proc R Soc Med, 1965. 58: p. 295-300.
    8. 8.Freeman, M.D., C.J. Centeno, and S.S. Kohles, A systematic approach to clinical determinations of causation in symptomatic spinal disk injury following motor vehicle crash trauma. PM R, 2009. 1(10): p. 951-6.
    9. 9.Freeman, M.D. and S.S. Kohles, An evaluation of applied biomechanics as an adjunct to systematic specific causation in forensic medicine. Wien Med Wochenschr, 2011. 161(19-20): p. 458-68.
    10. 10.Society of Automotive Engineers. Body Engineering Committee., Society of Automotive Engineers. Automotive Safety Committee., and Society of Automotive Engineers. Motor Vehicle Safety Systems Testing Committee., Human tolerance to impact conditions as related to motor vehicle design--SAE J885 FEB2011: report of the Body Engineering and Automotive Safety Committees, approved March 1964. 2011 ed. SAE information report. 2011, Warrendale, PA: Society of Automotive Engineers. 20 p.
    11. 11.Boucher, P. and S. Robidoux, Lumbar disc herniation and cauda equina syndrome following spinal manipulative therapy: a review of six court decisions in Canada. J Forensic Leg Med, 2014. 22: p. 159-69.
    12. 12.Sadanand, V., et al., Sudden quadriplegia after acute cervical disc herniation. Can J Neurol Sci, 2005. 32(3): p. 356-8.
    13. 13.Freeman, M.D., et al., Significant spinal injury resulting from low-level accelerations: a case series of roller coaster injuries. Arch Phys Med Rehabil, 2005. 86(11): p. 2126-30.
    14. 14.Tencer, A.F., et al., Femur fractures in relatively low speed frontal crashes: the possible role of muscle forces. Accid Anal Prev, 2002. 34(1): p. 1-11.
    15. 15.Freeman, M.D., Cahn P.J., Franklin F.A., Applied forensic epidemiology. Part 1: medical negligence. OA Epidemiology, 2014. 2(1): p. 11.
    16. 16.Freeman, M.D. Biomechanical, mechanical, and epidemiologic characteristics of low speed rear impact collisions. in 67th Annual Meeting of the American Academy of Forensic Science. 2014. Orlando, FL: American Academy of Forensic Sciences.
    17. 17.Cote, P., et al., The annual incidence and course of neck pain in the general population: a population-based cohort study. Pain, 2004. 112(3): p. 267-73.
    18. 18.Ivancic, P.C., M.M. Panjabi, and S. Ito, Cervical spine loads and intervertebral motions during whiplash. Traffic Inj Prev, 2006. 7(4): p. 389-99.
    19. 19.Freeman, M.D., et al., A review and methodologic critique of the literature refuting whiplash syndrome. Spine (Phila Pa 1976), 1999. 24(1): p. 86-96.
    20. 20.Mattei, T.A. and A.A. Rehman, Schmorl"s nodes: current pathophysiological, diagnostic, and therapeutic paradigms. Neurosurg Rev, 2014. 37(1): p. 39-46.
    21. 21.Kyere, K.A., et al., Schmorl"s nodes. Eur Spine J, 2012. 21(11): p. 2115-21.
    22. 22.Matsumoto, M., et al., Tandem age-related lumbar and cervical intervertebral disc changes in asymptomatic subjects. Eur Spine J, 2013. 22(4): p. 708-13.
    23. 23.Steffens, D., et al., Does magnetic resonance imaging predict future low back pain? A systematic review. Eur J Pain, 2014. 18(6): p. 755-65.
    24. 24.Brinckmann, P. and R.W. Porter, A laboratory model of lumbar disc protrusion. Fissure and fragment. Spine (Phila Pa 1976), 1994. 19(2): p. 228-35.
    25. 25.Hancock, M.J., et al., Systematic review of tests to identify the disc, SIJ or facet joint as the source of low back pain. Eur Spine J, 2007. 16(10): p. 1539-50.