Journal of Orthopaedic & Sports Physical Therapy Official Publication of the Orthopaedic and Sports Physical Therapy Sections of the American Physical Therapy Association The Use of Patient Symptoms to Screen for Serious Back Problems Kathryn E. Roach, PhD, PT' Mark Brown, M D, PhD * Elizabeth Ricker, MS, P T ~ Peter Ahenburger, MS, PT lanet Tompkins, RN5 L ow back pain is a broad diagnosis encompassing a wide variety of disorders which differ greatly in etiology, prognosis, and appropriate management. The actual cause of low back pain can range from musculoligamentous injuries to anatomic abnormalities or even underlying systemic diseases (2). It is estimated that some form of low back pain will affect 80% of all Americans at some time during their lives (5). Of this group, most will recover from the initial episode but will suffer frequent recurrences during their lifetime. Patients with low back pain now have direct access to physical therapy without a physician's referral in 22 states. Even when a patient is referred by a physician, the physical therapist must remain alert to the possibility that the patient may require medical care outside the realm of physical therapy. Physical therapists must be able to screen low back pain patients to identify serious low back problems that require additional diagnostic evaluation and treatment by a physician. Screening is a procedure used to identify groups of individuals who would benefit from some type of medical attention. The Commission on Chronic Illness (1) defined screening as ". . . the presumptive identification of unrecognized disease Even when a patient is referred by a physician, the physical therapist must remain alert to the possibility that the patient may require medical care outside the realm of physical therapy. Physical therapists must be able to screen low back pain patients to identify those who have serious low back problems which require additional diagnostic evaluation and treatment by a physician. It is important for physical therapists to know which symptoms and signs or combination of symptoms and signs best indicate the likelihood of a serious problem. The purpose of this study was to test the sensitivity and specificity of low back pain symptoms in distinguishing individuals with a benign low back problem from those requiring surgical or medical intervention. Demographic and clinical data were collected retrospectively from a standardized low back pain questionnaire located in the medical records of 174 low back pain patients. Patients were classified as having a benign low back problem (N = 4 1) or a serious low back problem (N = 133) based on surgical findings or long term follow-up. Some individual symptoms had high specificity, but none had high sensitivity. To improve sensitivity while attempting to maintain moderate specificity, a number of symptoms were considered in parallel. The highest combination of sensitivity (.87) and specificity (.SO)was obtained by combining in parallel the symptoms of unable to sleep, awakened and unable to fall back to sleep, medication required to sleep, and pain worsened by walking. Key Words: screening, low back pain, clinical decision-making 'Assistant Professor, Division of Physical Therapy, Department of Orthopaedics and Rehabilitation, University of Miami School of Medicine, 5915 Ponce de Leon Blvd., Plumer Building, 5th Floor, Coral Gables, FL 33146 'Professor and Chair, Department of Orthopaedics and Rehabilitation, University of Miami School of Medicine, Coral Gables, FL 'Staff Physical Therapist, St. loseph's Care Center, Phoenix, AZ 4Staff Physical Therapist, Novacare Valley of the Sun Rehabilitation Hospital, Glendale, AZ SClinical Research Nurse, Department of Orthopaedics and Rehabilitation, University of Miami School of Medicine, Coral Gables, FL or defect by the application of tests, examinations, or other procedures which can be applied rapidly to sort out apparently well persons who probably have the disease from those who probably do not. A screening test is not intended to be diagnostic. Persons with positive or suspicious findings must be referred to their physicians for diagnosis and necessary treatment." In the case of low back pain, screening is done to identify patients with a high probability of having low back problems that are serious and will require additional diagnostic and treatment procedures that are outside the realm of physical therapy practice. Since physical therapists do not generally employ radiologic or laboratory testing, screening by physical therapists for serious low back pain must be based primarily on patient symptoms and signs. Therefore, it is important for physical therVolume 21 e Number 1 e January 1995 e JOSPT RESEARCH apists to know which symptoms, signs, or combination of symptoms and signs best indicate the likelihood of a serious problem. The validity of a screening test is determined by its sensitivity and specificity. Sensitivity is the ability of a test to correctly identify subjects who have disease, and specificity is defined as the ability of a test to correctly identify those who do not have disease (4,6). Although a number of physical therapy texts propose screening criteria to identify low back pain patients who have serious underlying disorders, these sources fail to provide information on the sensitivity or specificity of the criteria they advocate (3,7). Deyo et a1 review addressed the accuracy of various symptoms and signs in distinguishing between benign and serious low back problems. Although patient signs elicited during the physical examination almost certainly have an important role in screening for serious low back pain, this study was limited to an examination of patient symptoms. The purpose of this study was to determine the sensitivity and specificity of patient-reported symptoms in distinguishing individuals with benign low back problems from those who require surgical or other medical intervention outside the realm of physical therapy. STUDY quate number of subjects within each of the categories of serious low back problems, these categories were intentionally oversampled so that the final cohort included 133 subjects with serious low back problems and 41 subjects with benign low back problems. Subjects whose histories involved possible substance abuse or litigation or any psychogenic behavior noted during evaluation were excluded due to potential symptom magnification. Although these s u b jects may have had an organic problem, symptom magnification during self-reporting would make the validity of the data questionable (9). Procedure It is imporfant for physical therapists to know which symptoms, signs, or combination of symptoms and signs best indicate the likelihood of a serious problem. Deyo et al (2) compiled findings from a number of sources on the sensitivity and specificity of various elements of the patient history and physical examination in diagnosing serious low back disorders such as cancer, spinal stenosis, herniated disc, and compression fracture. They found that items such as previous history of cancer, unexplained weight loss, and failure to improve with a month of therapy all had specificities above .90 for the detection of cancer. Unfortunately, none of these items had a sensitivity above .35, indicating that many low back patients with cancer did not report these symptoms. None of the studies reported in the JOSPT Volume 21 Number 1 January 1995 METHODS Design A retrospective medical record review was conducted to determine the low back pain symptoms and demographic characteristics of a cohort of patients who presented to a tertiary referral outpatient orthopaedic clinic of a large teaching hospital for treatment of low back pain. Each subject's low back diagnosis was recorded and classified as either benign or serious (Table 1). All diagnoses were confirmed surgically or by long-term follow-up. Subjects The medical records of 174 subjects were selected for this study. Records were selected for inclusion in this study based on the availability of a completed intake questionnaire and a confirmed low back pain diagnosis. In order to include an ade- The medical records of all s u b jects contained a standardized, selfadministered intake questionnaire which was completed at the time of the first clinic visit. The questionnaire included demographic and selfreported symptoms, such as sleep disturbances related to low back pain, changes in bowel and bladder function, and presence of low Sack pain when performing activities such as walking, driving, pushing, sitting, or standing. Information from this standardized intake questionnaire was transferred, without modification or interpretation, to a data collection form which wa.. identified only by a unique code number. Demographic information and patient final diagne sis were abstracted from the medical record. Data Analysis Student's t test and chi square statistics were calculated to compare Benign back pain Low back pain that has a benign, self-limiting course not requiring surgery or other aggressive medical treatment. Serious back pain Low back pain attributable to spinal abnormalities requiring surgery or low back pain secondary to another underlying disorder. This category includes spinal stenosis or herniated disk requiring surgery, osteoporotic fracture, tumor, and infection. TABLE 1. Low back pain classifications. 3 RESEARCH STUDY the demographic characteristics of serious low back problem subjects to those with benign low back problems. The sensitivity and specificity of individual symptoms in distinguishing serious low back problems from benign low back problems were calculated. Sensitivity is defined as the p r e portion of individuals who have a disease and test positive for that disease using a screening instrument (4). For the purposes of this study, the sensitivity of an item indicated the proportion of subjects with serious low back problems who reported the presence of a symptom. Sensitivity was calculated by dividing the number of patients with serious low Specificity was the probability of responding negatively to an item if a subject had a benign low back problem instead of a serious low back problem. back problems who reported a specific symptom by the total number of patients with serious low back pain. Therefore, the higher the sensitivity of the symptom, the lower the p r e portion of subjects with serious low back problems who were classified as having benign low back problems (false negatives) based on the presence of that symptom (Table 2). It is important to note that sensitivity only provides information about patients with serious low back problems. Specificity is defined as the proportion of individuals who do not have a disease and test negative for that disease using a screening instrument (4). Therefore, in this study, specificity Serious Low Back Problem Benign Low Back Problem Total Symptom present (test positive) (a) True positive 0 False positive (a + b) All subjects who test positive Symptom absent (test negative) (c) False negative (dl True negative (C + d) All subjects who test negative Total (a + C) All subjects with a serious back problem (b + d) All subjects with a benign back problem (atbtctd) All subjects Sensitivity = True positive + all subjects with a serious back problem. Specificity = True negative + all subjects with a benign back problem. TABLE 2. Sensitivity and specificify. was the probability of responding negatively to an item if a subject had a benign low back problem instead of a serious low back problem. Specificity was calculated by dividing the number of subjects with benign low back problems who did not report a symptom by the total number of s u b jects with benign low back problems. Specificity only provides information about subjects with benign low back problems. A test with high specificity would not classify benign low back problems as serious low back p r o b lems (false positive). An ideal screening test has both a high sensitivity and a high specificity. As a strategy to improve sensitivity, various combinations of symp toms were considered in parallel. Screening in parallel entails the decision to classify a subject as screening positive if that subject reports any one of a group of symptoms (4). The sensitivity and specificity of various combinations of symptoms were also calculated. For screening in parallel, sensitivity was calculated by dividing the number of patients with serious low back problems who reported any one of the group of symptoms by the total number of subjects with serious low back pain. Specificity was calculated by dividing the number of s u b jects with benign low back problems who did not report any of the group of symptoms by the total number of subjects with benign low back p r o b lems. Data analysis was conducted using the SAS program (version 6.01, SAS Institute, Cary, NC) (8) on a VAX mainframe computer. RESULTS Subjects in the benign low back problem and serious low back p r o b lem groups were similar in height, weight, and gender distribution. However, the mean age of the serious low back problem group was 54.38 years compared with 42.68 years for the benign low back problem group (Table 3). The specificity and sensitivity calculated for each item are presented in Table 4. Items 4, 5, 6, 8, 9, and 10 demonstrated high specificity but low sensitivity, while items 3, 7, 11, and 13 demonstrated moderate sensitivity with moderate to low specificity. Items with moderate sensitivity and specificity, such as 7, generated a fairly large number of false positive and false negative findings (Table 5). No individual symptoms had high sensitivity. Combining symptoms in parallel increased sensitivity but substantially decreased specificity (Table 6). Combination 7 had the highest sensitivity but had very low specificity. Combination 5 had the highest combined sensitivity and specificity. This combination produced very few false negative findings (Table 7). Volume 21 Number 1 January 1995 JOSPT -... -.- - .-.--- ---. -.. -. ." .. . . . .- Characteristic Serious Low Back Problem ( N = 133) Age Weight Height Proportion male 54.38 157.1 1 66.77 50.25 . . -. . .- . .. . . . . Benign Low Back Problem ( N = 41) - -. . .. - .- . * Student's t test. t Chi square. TABLE 3. Subject demographic characteristics. Symptom Sensitii Specificity TABLE 4. Sensitivity and specificity of individual symptoms. Serious Low Back Problem Benign Low Back Problem Symptom present (test positive) 99 16 True positive False positive Symptom absent (test negative) False negative 33 24 True negative Total . .. .S T U D Y , . . .. DISCUSSION p value 1. Pain worse in AM 2. Pain worse in PM 3. Pain is constant 4. Unable to urinate 5. Can't hold urine 6. Currently smoke 7. Walking aggravates pain 8. Pain wakes from sleep 9. Need sleep medications 10. Unable to sleep 11. Sitting aggravates pain 12. Driving aggravates pain 13. Bending aggravates pain RESEARCH - - .- .- . .- . . -. . - Although the ideal screening test would be 100% sensitive and 100% specific, sensitivity and specificity are usually inversely related. This principle is demonstrated by the findings of this study. A number of symptoms had very high specificity but low sensitivity. If there must be a tradeaff between the sensitivity and specificity of a screening test, then it is important to emphasize the characteristic that is most important for a given clinical problem. The relative importance of sensitivity and specificity may differ depending on the disease or condition involved. If the diagnostic tests required to confirm a diagnosis The combination of symvtoms that includes the three sleep disturbance symptoms and pain worsened by walking has a sensitivity of .87. Sensitivity: 99/132 = 2 5 . Specificity: 24/40 = .60. TABLE 5. Sensitivity and specificity o f walking aggravates low back pain. SPP~W 1. Sleep problems* 2. Sleep problems* Sensitivity Specificity .50 .75 -85 .27 .76 .53 .77 .20 .87 .50 .92 .27 .98 .05 Pain worse when sitting 3. Sleep problems* Pain constant 4. Sleep problems* Pain worse when bending 5. Sleep problems* Pain worse when walking 6. Sleep problems* Pain worse when walking Pain constant 7. Sleep problems* Pain worse when walking Pain constant Pain worse when bending * Sleep problems consist oi: pain wakes trom sleep, need sleep medications, or unable to sleep. TABLE 6. Sensitivity and specificity of symptoms combined in parallel. JOSPT Volume 21 Number 1 January 1995 are very painful or very expensive, and the disease is not life-threatening or there is little benefit from early detection, then high specificity may be important. On the other hand, if failure to identify a patient at j e o p ardy prolongs suffering and risks death or disability, then high sensitivity may be critical. Although it may be argued that the diagnostic work-up for a serious low back problem is expensive, a stronger case can be made for the importance of identifying and referring patients who have low back problems that cannot be managed by physical therapy alone. The combination of symptoms that includes the three sleep distur- RESEARCH STUDY Sensitivity: 1 l5/l33 = .86. Specificity: 2 1/41 = .5 1. Symptoms considered in parallel: I ) walking aggravates pain, 2) pain wakes from sleep, 3) need sleep medications, and 4) unable to sleep. TABLE 7. Sensitivity and specificity when screening in parallel. bance symptoms and pain worsened by walking has a sensitivity of .87. This means that 87% of individuals with serious low back problems would report at least one of these symp toms. This same combination of symptoms had a specificity of .50. This indicates that 50% of the patients with a benign low back p r o b lem would report none of these symptoms. Unfortunately, it is also true that 50% of patients with benign low back problems would report at least one of these symptoms. A clinician involved in screening low back pain patients might also want to know what proportion of the patients who screen positive actually have a serious low back problem. The proportion of patients who actually have the prob lem, out of all those who have tested positive, is known as the positive predictive value of a screening test (4,6). The positive predictive value depends on both the sensitivity of the test and the prevalence of serious low back problems in the population of low back pain pz+ tients being screened. Because this study oversampled serious low back problems, the prevalence of serious low back prob lems in this .sample did not reflect the prevalence of serious low back pain in a typical clinic population. Therefore, the positive predictive power of the symp toms and combination of symptoms could not be accurately estimated. Addii tional prospective studies in a variety of settingx will be required to clarifjr this point This was a retrospective study and was, therefore, limited to the information available in the standardized intake questionnaire. For that reason, this study only dealt with patient symptoms. Although it was possible to correctly classify a large proportion of patients as having either benign or serious low back problems using only self-reported symptoms, it is possible that the additional information from a physical examination would have improved the classification of patients as either serious or benign. Screening based on symp toms should be viewed as a first step in identifying patients with serious low back problems. Additional studies will be required to determine which aspects of the physical examination, in combination with the patient's self-reported symptoms, will produce the highest levels of sensitivity and specificity in identifying patients who require assistance outside the realm of physical therapy. CONCLUSION The results of this study demonstrate a number of important findings. First, there are a number of symptoms that are highly specific for serious low back problems, since very few patients with benign back p r o b lems report these symptoms. Unfortunately, these same symptoms have low sensitivity, indicating a large number of patients with serious back problems do not report these symp toms either. Second, by considering a number of symptoms together, it is possible to increase sensitivity to the point that very few patients with serious back problems are not identified. However, the cost for this increased sensitivity is an increase in the number of patients with benign low back problems who also screen positive. Therefore, the decision to refer such a patient for additional diagnostic work must be based on the entire history of the patient's illness, including previous diagnostic work, the physical examination, and the results of the screening test. lOSlT REFERENCES Commission on Chronic Illness: Chronic Illness in the United States (Vol I), pp 45. Cambridge, MS: Commonwealth Fund, Harvard University Press, 1957 Deyo R, Rainville J, Kent D: What can the history and physical examination tell us about low back pain? JAMA 268(6):760-765, 1992 Goodman CC, Snyder TEK: Differential Diagnosis in Physical Therapy: Musculoskeletal and Systemic Conditions, Philadelphia: W.B. Saunders Company, 1990 Hennekens C, Buring J: Epidemiology in medicine. In: Mayrent S (ed), Epidemiology in Disease Control, pp 327335. Boston: Little, Brown, and Co., 1987 Kelsey J, White A: Epidemiology and impact of low-back pain. Spine 5(2): 133- 142, 1980 Mausner JS, Kramer S: Screening in the detection of disease. In: Mausner and Baum (eds), Epidemiology: An Introductory Text, pp 2 14-237. Philadelphia: W.B. Saunders Company, 1985 Randall T, McMahon K: Screening for musculoskeletal system disease. In: Boissonnault WG (ed), Examination in Physical Therapy Practice: Screening for Medical Disease, pp 199-236. New York: Churchill Livingstone, Inc., 1 99 1 SAS Institute Inc: SASbTAT User's Guide, version 6, (4th Ed, Vol 1 and 2), Cary, NC: SAS Institute Inc., 1987 Waddell G: Occupational low-back pain, illness behavior, and disability. Spine 16(6):683- 685, 1991 Volume 21 Number 1 January 199.5 JOSPT