L ower back pain is one of the most common presenting complaints in

Transcription

1 William R. Remus1 2 Glen Strome1 Frank L. Zwemer, Jr.2 Received April 17, 1997; accepted after revision August5, tmallinckrodt Institute of Radiology, Washington University Medical School, 216 S. Kingshighway, St. Louis, MO Address correspondence to W. R. Remus. 2Department of Internal Medicine, Washington University Medical School, St Louis, MO AJR 1998;17O: X/98/ American Roentgen Ray Society Use of Lumbosacral Spine Radiographs in a Level II Emergency Department OBJECTIVE. We studied indications for lumbosacral spine radiographs in an emergency department setting. SUBJECTS AND METHODS. Clinicians completed a lumbosacral spine request form before obtaining lumbosacral radiographs on 482 patients who were examined in a level II emergency department. The clinicians detailed the indications for the examination. their suspected clinical diagnoses. and the expected effect of the lumbosacral spine study on management. In addition. the duration of the patients symptoms was recorded as well as their age and sex. These data were compared with the actual radiographic findings as interpreted by boardcertified musculoskeletal radiologists. RESULTS. Major indications for lumbosacral radiographs were lower back pain (92%) and trauma (36% ). Patient expectation and medicolegal concerns. related either to insurance documentation or to physician litigation. were cited in 42% ofcases. A neurologic deficit was present in 88% of patients, and a history of neoplasm was elicited in 4%. Strain (56%) and fracture (20% ) were the iltost commonly suspected clinical diagnoses. Radiographs showed either normal findings or spondylosis in 86% of cases. Fractures. of which 10 were definitely acute, were identified in 55 patients. No fracture required decompression. Neoplastic involvement was found in seven patients. all of whom had histories of osseous meta.stases. Neurologic deficit, present in 37 patients. did not correlate with abnormalities seen on lumbosacral radiographs. CONCLUSION. As with studies of lumbosacral spine radiographs obtained in outpatient settings. our data from a level II emergency department support the use of lumbosacral spine radiographs for patients with a history of trauma. even if relatively minor. in elderly patients and in patients with lower back pain who have a history of neoplasm. Lumbosacral radiographs obtained for an isolated complaint of lower back pain generally provide no clinically useful information. Similarly. lumbosacral radiographs obtained for patients with isolated neurologic abnormalities are unrevealing. Such patients are better examined (although not necessarily at the time ofemergency department evaluation) with techniques such as MR imaging that show soft-tissue lesions. L ower back pain is one of the most common presenting complaints in the emergency department setting. Although evaluation of this problem often includes plain films. the role of lumbosacral spine radiographs remains controversial f Numerous authors have suggested that spine radiographs are not necessary in the workup of every patient who presents with lower back pain. In fact. the routine ordering of radiographs fbr lower back pain has been cited as emblematic of the overuse of imaging procedures [5). Even so. lumbosacral spine radiography remains one of the more commonly ordered studies in the emergency department. Other disadvantages of routine ordering of lumbar spine radiographs include the high gonadal dose of radiation, the poor correlation between radiologic findings and symptoms, and the high cost:benefit ratio [6-8]. A longterm Swedish study suggested that the mcidence of unexpected findings on radiographs of the lumbosacral spine is exceedingly low, approximately one in 2500 for patients younger than 50 years old [1]. To address the low yield of spine radiographs in patients with lower back pain, some authors [3, 9] have advocated the use of selection criteria based on clinical variables. These criteria were developed for use in outpatient settings. AJR:17O, February

2 Remus et al. In the emergency department setting, it is unclear if these same criteria are applicable. Because patient follow-up cannot be guaranteed, one might expect that clinicians would perform a more aggressive evaluation of lower back-related complaints. To further examine which patients with lower back complaints in this setting might benefit from lumbosacral spine radiographs, we prospectively collected data on patients for whom emergency department physicians had requested lumbosacral spine radiographs. In particular, we evaluated the reasons clinicians gave for ordering lumbosacral spine radiographs, their suspected clinical diagnoses, and the intended effect of these studies on patient management. Subjects and Methods All patients receiving lumbosacral spine radiographs in our emergency department between October 1994 and December 1995 were entered in the study. The emergency department cares for approximately 26,000 cases per year and has a level II designation. Because of this designation. the incidence of high-velocity trauma in our patients is low. Also, because ours is a teaching institution, patients are often seen initially by house staff. However, all cases are concurrently reviewed by an attending physician. Four hundred eighty-two patients (314 women and 168 men) with an average age of 56 ± 2 1 years (age range, years) were enrolled in the study. Every attempt was made to ensure complete data collection and, to our knowledge, all patients receiving lumbosacral spine radiography during the study period were included. Clinicians were required to submit a completed questionnaire as a condition for obtaining lumbosacral spine radiographs. Thus, all responses on the questionnaires predate viewing of the radiographs. The questionnaire was designed as a checklist to facilitate its completion. However, an opportunity was also provided for the clinician to include suspected clinical diagnoses that were not specified on the questionnaire. The attending staff was instructed on the use of the questionnaire by one of the authors who was available to answer questions and arbitrate problems. Because multiple rotating house staff work in the emergency department, the staff received no formal training on questionnaire completion but were supervised by the attending staff physician. The first section ofthe questionnaire asks the clinician to identify the indications for ordering lumbosacral spine radiographs. These indications (Table I ) are ones commonly encountered in the The clinicians were asked to specify a suspected clinical diagnosis on the questionnaire. Only one response was allowed in this section to limit the response to the diagnosis of greatest suspicion. The clinicians were also asked to identify the proposed effect(s) of obtaining the radiographic study on patient management, including confirmation of clinical diagnosis, patient reassurance, whether to admit or further evaluate Ofl an outpatient basis. and medicolegal issues (more than one choice was possible). Responses to the questionnaires were anonymous. However, clinicians were asked to record their level of training and their specialty. Because this study evaluates the ability of lumbosacral spine radiographs to aid in clinically important diagnoses. the charts of all patients with a clinically important radiologic diagnosis (e.g.. neoplasm or fracture) were reviewed for additional clinical information. Charts of patients whose lumbosacral spine studies showed normal findings or spondylosis were not reviewed. The routine lumbosacral spine examination at our institution includes anteroposterior. lateral. hilateral posterior oblique, and coned-down lateral views. All examinations are interpreted by boardcertified radiologists who specialize in musculoskeletal radiology. Official radiography report5 were used as the source of the recorded radiographic diagnoses. The radiologic diagnosis of greatest acuity was used as the official diagnosis when multiple diagnoses were described in the report. In all cases. the radiologist was unaware of the information contained in the questionnaire but not necessarily of the clinical history. For this study. spondylosis was defined as any combination of intervertebral disk space narrowing. hypertrophic spurring, and facet osteoarthritis. Fractures were assessed ftr age by the interpreting radiologists and were reviewed by the atithors for concordance. A fracture was deemed acute if it showed sharp angulation of the vertebral cortical margins and no productive bone or callus surrounding the fracture margins. Similarly, chronic fractures were identified by either the presence of smooth margins and productive bone or callus around the fracture, comparison with previous radiographs. or clinical history showing the fracture to be preexisting to the current emergency department visit. Fractures for which these findings were inconclusive were classified as being of indeterminate age. The data were entered into a computer and tabulated for analysis. Statistical testing was carried out using the chi-square or Fisher s exact test, as appropriate. P values of less than.05 were accepted as evidence to reject the null hypothesis. (927e) and trauma (36%) were the most frequent reasons for requesting a lumbosacral senes. The combination of patient expectation and medicolegal concerns accounted for 201 responses (42%). A neurologic deficit was present in 37 patients (8%) and a history of neoplasm in 20 (4%). In the 40 cases in which pain was not an indication listed for the lumbosacral spine series, trauma was the most common indication (63%), followed by insurance-related medicolegal issues (35%), neurologic deficit ( 18%), and patient expectation ( I 8%). Table 2 lists the clinicians suspected diagnoses. Clinically. strain (56%) was the most commonly suspected diagnosis, followed by fracture (20%), herniated disk (13%), and neoplasni (4%). Of the 272 patients clinically suspected of having strain, one (0.4%) ultimately received a radiographic diagnosis of acute fracture and eight patients (3%) had fractures of indeterminate age. Of the 94 patients suspected of having an acute fracture, eight (9%) had radiographic diagnoses of acute fracture and 1 1 ( 12%) had fractures of indeterminate age. Overwhelmingly, confirmation of the suspected diagnosis (ii = 355, 74%) and patient reassurance (ii = 157, 33%) were the major motivations for obtaining the lumbosacral series. Medicolegal concerns, related either to insurance issues (ii = 84, 17%) or to potential litigation directed at the emergency department physician (ii = 41, 9%), accounted for 26% of responses. An impact on the decision to admit the patient was expected in 50 patients ( 10%). Table 3 gives the final radiographic diagnoses. Most common among these were inical Indications for Z.%g tphs in 4 atlents emergency department setting, including both complaint-related indications and relative indications such as patient expectation and the threat of medicolegal action. Clinicians could choose more than one indication, accounting for more responses in this Results The frequency of the various indications for lumbosacral spine radiographs (from which the section than patients enrolled in the study. The dura- clinicians could choose as many as were appro-,,,,,,..ians cou.j choose,,,.,,ations for each tion ofthe patient s complaint was also recorded. priate) is shown in Table 1. Lower back pain patient. Hence, total responses exceed AJR:170, February 1998

3 Use of Spine Radiographs in Emergency Department studies showing spondylosis (48%) and stud- The single younger patient, 33 years old, was ies showing normal findings (38%). Fifty- found to have mild compression of the T12 five fractures (I 1%) were identified. Of these, and LI vertebral bodies. On review, these 10 (18%) were definitely acute, 24 (44%) compressions more likely represented congen- (including three patients with pars interartic- ital transitional vertebrae. ularis defects) were of indeterminate age, and 21 (38%) were definitely old. Neoplastic involvement of the lumbar spine was diagnosed in seven (2%) of the 482 patients. Trauma A history of trauma was not significantly associated with a radiographic diagnosis of fracture in our sample (p =.69). We believe it is unlikely that this failure to show an association between fracture and trauma history is related to sample size. Instead, this absence of correlation more likely results from the fact that none of the fractures occurring in this sample was related to high-impact trauma. Chart review of the 173 patients with a history of trauma showed that in all cases the trauma was relatively minor (e.g., fall onto buttocks from a standing height or twisting injury). In six ofthe 10 patients with acute fracture, no trauma history existed, and these fractures were attributed to osteoporosis. In the remaining four patients with acute fractures, the trauma was related to a fall. All of the acute fractures in this sample occurred in patients who were at least 64 years old. Thus, in this population, it appears that osteoporosis was as frequently responsible for fractures as was trauma. Although all of the trauma patients had symptom durations of less than I week, this duration is not significantly disproportionate compared with the rest ofthe sample (p =.10). Symptom duration in the patients with a radiographic diagnosis of fracture of indeterminate age was also less than I week in 17 of 24 patients. Of the remaining patients with a radiographic diagnosis of fracture of indeterminate age, symptom duration was less than I month in three of 24, greater than 1 month in three of 24, and unspecified in one patient. Thus, most patients (79%) with acute or subacute fractures, even low-impact fractures as predominate in this sample, will likely seek medical care within a week of injury. Excluding patients with a history of trauma or neoplasm, six (3%) and nine (4%) of the remaining 236 patients in our sample had acute Neop!asm Neoplasm was suspected clinically in 20 patients. This suspicion was confirmed radiographically in three of these patients. Radiographs showed vertebral metastases in four additional patients for whom no suspicion of malignancy was indicated on the patient s survey form. However, chart review showed that these patients all had a previous diagnosis of osseous metastatic disease. All seven patients receiving a radiographic diagnosis of neoplasm were at least 60 years old. Neurologic Deficit Thirty-seven patients who had neurologic deficits consistent with a lumbar plexus distribution also had lumbosacral spine radiographs. Clinically, a herniated nucleus pulposus was suspected as the cause ofthe deficit in 17, metastatic disease in seven, strain in six, fracture in four, and spinal stenosis or spondylosis in the remaining three patients. Actual radiographic diagnoses showed spondylosis in 24, metastatic disease in three, chronic fractures in two, a fracture of indeterminate age in one, and normal radiographic findings in seven. Overall, 63 patients had a suspected clinical diagnosis of a herniated nucleus pulposus. Of these, 17 (27%) had a deficit on neurologic examination. In addition, 30 patients had herniated nucleus pulposus listed as the suspected radiographic diagnosis, but only five of these had a deficit on neurologic examination. Medicolegal When the lumbosacral examination was requested for medicolegal reasons pertaining to possible malpractice action, the radiographs showed either normal findings or spondylosis in 38 (93%) of 41 patients. For the remaining three patients, studies showed chronic fracture, spina bifida, and scoliosis. When insurance documentation was the indication for the lumbosacral series, 80 (95%) of 84 showed either normal findings or spondylosis. Three of these lumbosacral series showed chronic compression fractures and one showed a pars in- Discussion Prior studies [3, 4, 9-12] have proposed criteria for requesting lumbosacral spine radiographs in outpatient clinic-but not emergency department-settings. The criteria advocated by most authorities include obtaining lumbosacral spine series when the patient is older than 50 years or younger than 20 years or in the presence of a focal motor deficit, recent major trauma, a history of unexplained weight loss, a history of cancer, fever (temperature 37.8#{176}C),a history of drug abuse, or a history of corticosteroid use. Because 96% of repeated lumbosacral studies provide no clinically useful new information, Deyo and Diehl [9] recommended not obtaining additional lumbosacral radiographs with each new episode of lower back pain. Scavone et al. [4) recommended limiting lumbosacral series to patients with major trauma, cases of tumor evaluation for comparison with bone scan, and postmenopausal women with minor trauma who are likely to have osteoporosis. Dillane et al. [13] pointed out that most of the spontaneous episodes of lower back pain are self-limiting and that in more than 80% of nil RadIographiC _ses In 482 PatIent. and indeterminate-age fractures, respectively. terarticularis defect the age of which was With the exception of two patients with pars uncertain. Similarly, when the studies were interarticularis defects, these patients all had ordered because of patient expectation in the compression fractures. All but one of these absence of trauma, only three fractures of in- fractures were in patients older than 60 years, determinate age were discovered. The other and all received symptomatic treatment only. 41 cases showed no clinically acute disorder. AJR:170, February

4 Remus et al. cases a cause is never found. Thus, these re- The fact that identification of a fracture did likely to have disease than those that were searchers suggest that lower back pain should not alter treatment of patients who had neither excluded from imaging. Second. analytical be treated symptomatically over the first 2 weeks of an attack without any lumbosacral trauma nor a history of neoplasm suggests that the lumbosacral series was unnecessary in data show that this type of selection bias does not significantly affect analysis radiography. Similarly, a recent study by these cases. These results also coincide with Another limitation is that our sample patients. Elam et al. [14] concluded that because most lower back pain resolves with minimal intervention, emergency physicians should refrain from obtaining diagnostic imaging studies and specialist consultation. Our data, although gathered specifically to examine the use of lumbosacral series in the emergency department. shed some light on the above recommendations. As happened in multiple other studies, most of the lumbosacral series (86%) obtained in our sample showed either normal findings or only spondylosis. The latter, particularly the components of facet osteoarthritis and hypertrophic spurring, has been shown to correlate poorly with lower back pain [7, 10, 15-18]. On the other hand, some have suggested that finding degenerative disk disease does correlate with symptoms [7, 19]. The overall poor correlation between back pain and spondylosis may explain why spondylosis was suspected as the clinical cause of complaints in only 4% of our patients, despite the high prevalence of spondylosis in both prior studies and our patient sample (48%). On the other hand, the poor correlation does not explain why spondylosis was suspected as the radiographic diagnosis in only 15% of our patients, particularly given the older mean age of the sample. A history of trauma did not correlate well with a radiographic diagnosis of fracture in our sample. As noted earlier, most of our reported fracture cases represented uncomplicated compression fractures in older, probably osteoporotic, patients. The fact that our emergency department has a level II designation and is directly adjacent to one with a level I designation accounts for the absence of highimpact trauma in our population. It seems likely that trauma of increased severity, as would be seen in a level I emergency department, would have correlated better with a radiographic diagnosis of fracture. Although at first analysis it appears that a duration of symptoms of less than 1 week and a patient age of older than 60 years were most predictive of the report of Scavone et al. [4j, who found no fractures in patients with minor trauma except in patients with osteoporosis. All seven ofthe patients with a radiographic diagnosis of vertebral metastasis had known primary carcinomas. These individuals accounted for 2% of the patients in the sample. slightly greater than rates reported in the literature [3, 20, 2 1]. In each case, the patient had previously diagnosed osseous metastatic disease before presentation in the emergency department. Deyo and Diehl 1201 found the diagnosis of metastatic disease correlated with older age (>50 years), a history of cancer. pain of greater than I month s duration. failure to improve with conservative measures, anemia, and elevated erythrocyte sedimentation rate. Thus, the available data suggest that discovery of vertebral metastasis on lumbosacral radiographs is extremely unlikely in the absence of implicating clinical data. Furthermore, most lumbosacral series showed no important diagnostic information in 33 (89%) of 37 patients with neurologic deficits. Three of the remaining patients were found to have metastases on radiography, but this diagnosis had been made before the current emergency department visit in all cases. One patient had a compression fracture of indeterminate age. These data are coincident with the observation that most patients who have neurologic deficits related to their lumbar plexuses either have soft-tissue abnormalities not visible on lumbosacral spine series or have spinal stenosis, a diagnosis for which lumbosacral radiographs are not sensitive. We found a similarly low yield of radiographic diagnoses when medicolegal reasons or patient reassurance was the only motivating factor for the lumbosacral spine series. These lumbosacral studies showed only three fractures of indeterminate age in the 87 patients who met these criteria. Limitations of our study include that no guarantee exists that clinicians ordered all the potential lumbosacral radiographs on patients with problems related to lower back pain. coming froni a level II emergency department, may not ditthr greatly from previous outpatient populations studied. Evaluation of cases from a level I trauma center may give different results. It is clear from our data and the studies that have preceded ours that lumbosacral series, whether obtained in patients seen in clinic or in an emergency department. yield clinically important diagnoses only occasionally. The Public Health Service estimated in 1973 that 3 million lumbosacral radiographs were obtamed yearly in the United States at a cost of $150 million to patients In addition. a single lumbar spine series gives radiation doses to the gonads equivalent to having a daily chest radiograph for more than 6 years (241. Furthermore. in 1982 Liang and Komaroff 161estimated that, given the prevalence of serious disease, $2072 would be spent on lumbosacral spine series few each day s suffering that was avoided. Despite these indications that lumbosacral spine radiography is infrequently beneficial, it is well documented that patients who come to clinic for lower back problems are more likely to be satisfied with theircare iftheirclinician obtains a lumbosacral series Our data, similar to those from outpatient studies, support limited use of lumbosacral radiographs in an emergency department only for patients with a history of significant trauma. patients with a combined history of both lower back pain and neoplasm, and, possibly. older patients (>60 years old) with minor trauma. Identification of a fracture did not change therapy in this last group. Our data suggest that lunibosacral series performed in patients with only a complaint of lower back pain or pain associated with a focal neurologic deficit are unlikely to yield clinically important information. Because most lower back pain is self-limiting, common sense dictates that lumbosacral radiography be reserved for patients whose pain fails to respond to a I -month trial of conservative therapy unless other compelling findings are present I 131. Patients with acute fracture, these findings are probably arti- However, we believe inclusion bias was un- focal neurologic deficit are better examined facts of the study. As might be expected in an likely to have significantly affected our analy- with an imaging technique capable of showing emergency department, we found that the short sis for two reasons. First, clinicians normally soft-tissue detail. such as MR imaging. al- duration of symptoms in fracture patients was act as a low-pass filter, excluding the patients though such examination should not usually not significantly different from the symptom duration in those without fractures. with the lowest probabilities of acute disease. As a result, the included cases are ones more be obtained at the time of the emergency department evaluation. 446 AJR:170, February 1998

5 Use of Spine Radiographs in Emergency Department References 1. Nachemson AL. The lumbar spine: orthopedic challenge.spine 1976;l : Deyo RA. Practice variations, treatment fads, rising disability: do we need a new clinical research paradigm? Spine 1993;l8: Rockey PH, Tompkins RK, Wood RW, Wolcott BW. The usefulness of x-ray examinations in the evaluation of patients with back pain. J Fan: Pratt 1978:7: Scavone JG, Latshaw RF, Rohrer GV. Use of lumbar spine films: statistical evaluation at a university hospital.jama 1981:246: Hall FM. Overutilization of radiological examinations. Radiology 1976:120: Liang M, Komaroff AL. Roentgenograms in primary care patients with acute low back pain: a cost effectiveness analysis. Arch Intern Med 1982; 142: Torgerson WR. Dotter WE. Comparative roentgenographic study of the asymptomatic and symptomatic lumbar spine. J Bone Joint Surg Am 1976;58-A: Botticelli MG. A cost-conscious approach to the evaluation of patients with low back pain. Hawaii MedJ 1986:45: Deyo RA, Diehl AK. Lumbar spine films in primary care: current use and effects of selective ordering criteria. J Gen Intern Med 1986:1: Deyo RA, Bigos SJ, Maravilla KR. Diagnostic imaging procedures for the lumbar spine. Ann Intern Med 1989:111: I 1. Spitzer WO, LeBlanc FE, Depui M. Scientific approach to the assessment and management of activity-related spinal disorders: a monograph for clinicians-report of the Quebec Task Force on Spinal Disorders. Spine 1987:12: Deyo RA, Rainville J, Kent D. What can the history and physical examination tell us about low back pain? JAMA 1992;268: Dillane JB, Fry I, Kalton G. Acute back syndrome: a study from general practice. Br Med J 1966:2: Elam KC, Cherkin DC, Deyo RA. How emergency physicians approach low back pain: choosing costly options. JEmerg Med 1995:13: Splithoff CA. Roentgenographic comparison of patients with and without backaches. JAMA 1953; 152: Fullenlove TM, Williams Al. Comparative roentgen findings in symptomatic and asymptomatic backs. Radiology 1957:68: Magora A, Schwarta A. Relation between the low back pain syndrome and x-ray findings. I. Degenerative findings. Scand J Rehabil Med 1976;8: I I 8. LaRocca H. Macnab I. Value of pre-employment radiographic assessment of the lumbar spine. Can MedAssoc J 1969:101: Rowe ML. Low back pain in industry: a position paper. J Occup Med 1969:1 1: Deyo RA, Diehl AK. Cancer as a cause of back pain: frequency, clinical presentation, and diagnostic strategies. J Gen Intern Med 1988;3: Kaplan DM, Knapp M. Romm Fi, Velez R. Low back pain and x-ray films of the lumbar spine: a prospective study in primary care. South Med J 1986;79: Begg CB, Greenes RA. Assessment of diagnostic tests when disease verification is subject to selection bias. Biometrics 1983:39: Connors JP. Summary remarks by Gitlin J. In: Summars report and proceedings of the Conference on Low Back X-Rays in Preemployment Physical Examinations. Tucson: American College of Radiology, Hall FM. Back pain and the radiologist. Radiology 1980:137: AJR:17O, February