SPINE Volume 39, Number 7, pp E450-E454 ©2014, Lippincott Williams & Wilkins
DEFORMITY
Utility of Early Postoperative Radiographs After Posterior Spinal Fusion for Adolescent Idiopathic Scoliosis Michael J. Pensak, MD,* Mark C. Lee, MD,† Jennifer A. Bayron, BS,‡ and Jeffrey D. Thomson, MD†
Study Design. Retrospective. Objective. Evaluate the ability of serial full-length spine radiographs to detect clinically significant implant-related (IR) and non–implant-related (NIR) radiographical abnormalities in the first 6 months after routine posterior spinal fusion for adolescent idiopathic scoliosis (AIS). Summary of Background Data. Patients with AIS are exposed to repeated doses of ionizing radiation during the course of their treatment with potential consequences for their long-term health. Postoperative algorithms for AIS often involve frequent standing plain radiographs during the first 6 months after surgery to detect IR and NIR abnormalities that may impact a patient’s clinical course. However, the actual clinical utility of such repeated spine radiographs has not been studied. Methods. Retrospective chart and radiographical review was conducted at a single institution for patients with AIS after posterior spinal fusion between 2007 and 2012. Radiographical abnormalities identified on full-length spine radiographs or additional imaging modalities in the first 6 postoperative months were grouped into IR or NIR findings. The findings were considered clinically significant if they resulted in a deviation from an anticipated postoperative course or additional interventions. Results. For 129 patients, 761 full-length spine radiographs were obtained in the first 6 postoperative months. Eight patients (11 radiographs) had IR or NIR abnormalities, with only 2 of these considered clinically significant. Seven of the remaining 121 were
From the *University of Connecticut Health Center, Farmington, CT; †Department of Orthopaedics, Connecticut Children’s Medical Center, Hartford, CT; and ‡University of Connecticut Medical School, Farmington, CT. Acknowledgment date: September 5, 2013. First revision date: November 12, 2013. Second revision date: December 18, 2013. Acceptance date: December 19, 2013. The device(s)/drug(s) that is/are the subject of this manuscript is/are not FDAapproved for this indication and is/are not commercially available in the United States. No funds were received in support of this work. Relevant financial activities outside the submitted work: board membership, payment for lecture. Address correspondence and reprint requests to Mark C. Lee, MD, Department of Orthopaedics, Connecticut Children’s Medical Center, 282 Washington Street, Hartford, CT 06106; E-mail: [email protected] DOI: 10.1097/BRS.0000000000000219
E450
identified to have IR or NIR abnormalities using other imaging modalities, with 2 considered clinically significant. The sensitivity and specificity of a full-length spine radiograph for detecting a clinically significant abnormality was 50% and 95%, respectively. Conclusion. Routine full-length spine radiographs used with high frequency in the first 6 months after posterior spinal fusion rarely detected a radiographical abnormality that resulted in a meaningful change to a patient’s clinical management. Blanket postoperative screening algorithms should be reconsidered to minimize patient radiation exposure. Key words: postoperative spine radiograph, scoliosis, posterior spine fusion, pleural effusion, hardware malposition, thoracoplasty, sensitivity, specificity, pedicle screw, hardware failure. Level of Evidence: 3 Spine 2014;39:E450–E454
S
tanding plain radiographs remain the “gold standard” for monitoring and clinical decision making in adolescent idiopathic scoliosis (AIS). However, the frequency of such radiographs in a patient’s treatment course often raises concern about the dangers of repeated exposure to ionizing radiation in a pediatric population. It has been estimated that a typical pediatric patient with scoliosis will receive approximately 22 plain radiographical examinations during a 3-year treatment period, with as many as 618 plain radiographs reported for a single patient.1,2 The BEIR VII National Academy of Science panel concluded that children are more sensitive to radiation than adults because they have more time to express a cancer and have more dividing cells upon which radiation acts.3 There is specific concern about the risk of breast cancer, as progressive scoliosis is seen most commonly in the female population and the dividing breast tissue of teenage girls is particularly sensitive to ionizing radiation.3,4 Full-length spine radiographs are often obtained at a high frequency in the first 6 months after posterior spinal fusion for AIS. The goal of these radiographs is to monitor for potential postoperative implant-related (IR) or non–implant-related (NIR) complications, as well as to assess for the adequacy of the deformity correction. However, the clinical value of these radiographs in the postoperative management of a patient after scoliosis surgery has not been studied. This work seeks to determine the ability of routine full-length plain radiographs
www.spinejournal.com
April 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE131062_LR E450
05/03/14 12:18 PM
DEFORMITY obtained during the first 6 months after scoliosis surgery to detect IR- and NIR-related abnormalities and whether such findings lead to a significant change in a patient’s overall clinical management.
MATERIALS AND METHODS After institutional review board approval by Connecticut Children’s Medical Center, a comprehensive review of that institution’s medical records between September 2007 and January 2012 was performed. Patients were included in the study if they were: (1) diagnosed and treated for AIS, (2) between 9 and 18 years of age at the time of diagnosis, (3) managed with a posterior spinal fusion for their AIS using a hybrid construct (screws, wires, hooks) with or without thoracoplasties, and (4) followed for a minimum of 6 months. Patients were excluded if there was a diagnosis of a nonidiopathic etiology for their scoliosis (neuromuscular, congenital, or syndrome associated) or if they had any condition that required additional serial radiation exposure in the postoperative period, such as malignancy or fracture. Patient demographic data were gathered from the charts (sex, age at surgery), as well as visit frequency in the postoperative period. All patients participated in a uniform follow-up algorithm after discharge: each patient was seen at 6 weeks from discharge, 3 months from surgery, 6 months from surgery, and 1 year from surgery with posteroanterior (PA) and lateral radiographs obtained at each visit. Scoliosis curve characteristics (Cobb angles of the major curve and Lenke classification) were determined from preoperative radiographs. All postoperative full-length plain radiographs were evaluated, with each view tabulated separately, and examined for evidence of radiographical abnormality as well as changes in postoperative curve characteristics. A change in Cobb measurements for the instrumented or noninstrumented curves of 10° or more in the course of follow-up was thought to be significant, given existing information on interobserver and intraobserver reliability for Cobb measures in scoliosis.5,6 All radiographs in the study were reviewed by institution radiologists per routine and the reports were also evaluated. Findings on imaging studies were grouped into “implant-related” (IR) findings, such as implant failure or malposition, and “non– implant-related” (NIR) findings, such as pleural effusion or pneumothorax. Inpatient and outpatient charts were further reviewed for the patient’s clinical course to determine if the patient was found to have IR- or NIR-related abnormalities on additional non–spine imaging studies, such as chest radiographs or computed tomographic (CT) scans. Charts were further investigated to determine if an identified radiographical abnormality was clinically significant. A clinically significant radiographical abnormality is defined as a finding on imaging studies that results in a deviation from the anticipated postoperative course requiring additional interventions such as unplanned follow-up studies, braces, medications, blood transfusions, drainage catheter placements, bedside procedures, interventional radiological procedures, or reoperation. The definition is adapted from the Accordion Severity Grading System of Surgical Complications.7 Spine
Utility of Postoperative Spine Radiographs • Pensak et al
Means were provided for continuous variables. Sensitivity and specificity analysis was performed for the full-length spine radiographs obtained in the postoperative period with the “gold standard” reference of advanced imaging studies or non–spine-related plain radiographs. Odds ratio was calculated for patients with thoracoplasty procedures to quantify the likelihood of a clinically significant radiographical finding in this subgroup.
RESULTS One hundred twenty-nine patients, 91 females and 38 males, with an average age at the time of surgery of 14.4 years (range, 10–20) met our inclusion criteria. The average major curve Cobb angle was 59° and the most frequent Lenke curve classification was type 2. The average number of levels fused was 10 (range, 4–14). Seven patients received thoracoplasties. The mean curve correction of the largest Cobb angle was 38.5°. No patient demonstrated a significant change in the Cobb measures of the instrumented or noninstrumented segments when comparing postoperative radiographs during 6 months. For the study cohort, a total of 761 spine radiographs were obtained in the first 6 months after surgery, with an average of 5.9 (range, 2–11) per patient. There were 411 PA radiographs (54%) and 350 lateral (46%) radiographs. The average number of postoperative visits was 3.3 (Table 1). One hundred twenty-one patients had normal spine radiographs (750 radiographs), whereas 8 patients had abnormal
TABLE 1. Patient Characteristics* N
129
Age at surgery (range), (yr)
14.4 (10–20)
Sex, n (%) Female
91 (71)
Male
38 (29)
Lenke curve type, n (%) 1
39 (30)
2
48 (37)
3
7 (5)
4
6 (5)
5
19 (15)
6
10 (8)
Major curve Cobb angle (range), (°) Total no. of patients with thoracoplasties No. of ribs resected (range)
59 (41–98) 7 5 (4–6)
No. of levels fused (range)
10 (4–14)
No. of spine radiographs in the first 6 mo postoperatively (range)
5.9 (2–11)
No. of visits in the first 6 mo postoperatively (range)
3.3 (1–7)
*Average values provided unless otherwise stated.
www.spinejournal.com
E451
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE131062_LR E451
05/03/14 12:18 PM
DEFORMITY
Utility of Postoperative Spine Radiographs • Pensak et al
Figure 1. Implant-related and non–implant-related abnormalities identified in the patient population, grouped according to findings that did or did not result in a significant change in clinical management.
spine radiographs (11 radiographs), yielding a normal radiograph rate of 98.6%. Of the 8 patients, 1 IR and 7 NIR abnormalities were identified (Figure 1). Two of the 8 patients with abnormal radiographs required a significant change in their clinical management. The first patient had a mild lumbar pedicle screw back out and was placed into a brace. Subsequent radiographs 1 year later showed no further change in position. The second patient had an infected pleural effusion that required readmission to the hospital, drainage with a pigtail catheter and intravenous antibiotics. Interestingly, although this latter patient had a mild pulmonary abnormality on full-length spine radiographs at the time of discharge, this was not recognized by the surgical team. A large pleural effusion was later identified on chest radiographs obtained after the patient returned to the hospital with complaints of shortness of breath. The 6 patients that required no significant change in their clinical management exhibited only mild pulmonary findings on standing spine radiographs. All 7 pulmonary abnormalities were identified on the PA full-length spine radiograph, whereas the IR complication was identified on the lateral spine radiograph. Of the 121 patients with normal spine radiographs, 7 patients were subsequently found to have radiographical abnormalities (3 IR, 4 NIR) from additional studies prompted by the patient’s clinical course. The additional studies consisted of chest and abdominal plain radiographs as well as CT scan. Two of these 7 patients required a significant change in their clinical management. The first patient had subtle neuromonitoring changes intraoperatively and complaints of radicular pain in the perioperative period, although no discrete IR abnormality was identified on a supine full-length spine radiograph. A CT scan demonstrated a medial breach of a thoracic pedicle screw and the patient required a return to the operating room for revision of the implant. The second patient had worsening abdominal pain in the postoperative period and underwent examination with a plain radiographical abdominal E452
series that revealed free air in the abdomen, although standing full-length spine radiographs were negative. Exploration by the general surgery team revealed a duodenal perforation through a pre-existing ulcer that required repair (Figure 2). Two of the 5 patients with no significant change in clinical management had IR abnormalities. The first patient underwent a chest CT scan in the emergency department shortly after discharge for evaluation of acute shortness of breath. A prominent thoracic pedicle screw tip was noted, but did not threaten any vascular or pulmonary anatomy and was therefore maintained. The second patient awoke from surgery with diffuse, bilateral numbness distal to the ankles despite normal neuromonitoring studies intraoperatively. A CT scan was obtained to evaluate implant position and identified a mild lateral breach of a lumbar pedicle screw. The lower extremity paresthesias were attributed to an epidural placed intraoperatively for pain control and resolved completely once the epidural had been removed. The remaining 3 patients had mild pleural effusions identified on chest radiograph obtained for shortness of breath that required no additional intervention. The calculated specificity of a full-length spine radiograph in detecting a clinically significant radiographical abnormality was 95%, whereas the sensitivity was 50%. Overall positive predictive value of a full-length spine radiograph for a clinically significant finding was 25%, whereas the negative predictive value was 98%. When evaluating the subgroup of 7 patients with thoracoplasties, 2 had a clinically significant radiographical abnormality. One was noted on full-length spine radiographs (the patient with the pleural effusion that later required drainage) and the other was noted on subsequent studies (the patient with the free air in the abdomen). The odds ratio of having a clinically significant radiographical finding after thoracoplasty compared with the group of patients without thoracoplasty was 24.
www.spinejournal.com
April 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE131062_LR E452
05/03/14 12:18 PM
DEFORMITY
Utility of Postoperative Spine Radiographs • Pensak et al
Figure 2. A, Anteroposterior postoperative day 6 standing plain radiograph showing a posterior spinal fusion construct from T4 to L3 with multilevel rightsided thoracoplasties. B, Anteroposterior upright abdominal radiograph obtained 14 hours after the standing full-length plain radiograph for worsening abdominal pain. The white arrows outline a pocket of free air in the abdomen from a pre-existing duodenal ulcer that had perforated.
DISCUSSION A protocol-based approach for managing patients is relied upon heavily by most surgeons to make uniform the delivery of care and to simplify the postoperative management of patients undergoing complex surgical procedures. However, these protocols should be critically evaluated after their implementation to determine if there is a tangible clinical benefit in light of potential risk to the patient and cost to the health system. The acquisition of routine, high-frequency postoperative full-length radiographs after surgery for scoliosis merits such investigation. It is unclear whether such repeated exposure to ionizing radiation adds meaningfully to the clinical management of pediatric patients in the first 6 months after surgery for AIS. From the data presented, the full-length spine radiography is a poor screening test for identifying clinically significant abnormalities in the early postoperative period, with a sensitivity of only 50%. Even when a full-length spine radiograph is abnormal, the low positive predictive value of 25% suggests that the test is poor at reporting a finding that will be of clinical significance. Although specificity of the full-length spine radiographs is high at 98%, suggesting good ability to detect the complete absence of a clinically significant abnormality, it is difficult to justify the use of 750 otherwise unnecessary plain radiographical studies in this pediatric patient population for such reassurance, especially when the overall prevalence of clinically significant abnormalities was low (4/129 = 3%). This study underscores the idea that clinical signs and symptoms can more reliably lead to a clinically significant radiographical finding than a blanket approach of serial radiographs. This is best highlighted by the one patient who was discharged from the hospital with a pulmonary infiltrate on spine radiographs not initially identified by the surgical team, Spine
but readmitted for management of a large pleural effusion. In addition, the study suggests that operative details significantly impact the likelihood of a clinically significant radiographical abnormality in the first 6 months after spine fusion. A patient with a thoracoplasty procedure during posterior spine fusion, a procedure with a known pulmonary complication rate approaching 25%, was 24 times more likely in this study than a patient without a thoracoplasty to have a clinically significant radiographical finding.8,9 If serial screening radiographs are to be obtained, their yield will be greater if applied in patients with surgical approaches that have a recognized higher rate of specific complications. Limitations to this investigation were several. First, the findings of the study are applicable only to the specific population of patients undergoing posterior spinal fusion for AIS. Patients undergoing additional anterior procedures or complex spine osteotomies may benefit from serial, routine radiographs as IR and NIR complications are potentially greater. Second, the definition of a “clinically significant” abnormality is debatable. The study defined a clinically significant radiographical abnormality as any finding that altered the routine postoperative clinical course of a patient. Therefore, the definition of clinical significance will depend on the surgeon or institution-specific concept of what is “routine” for the patient with a postoperative spine problem. Finally, this investigation lacked the use of any validated algorithms for assessing radiographs. We reviewed all radiographs independently and all radiographical interpretations from the radiologists. However, it is difficult to know whether there is some observer variability when identifying abnormalities on radiographs, particularly subtle pulmonary findings. Strengths of this study included the large number of patients evaluated, stringent inclusion criteria, and a uniform, pre-existing office follow-up algorithm that minimized www.spinejournal.com
E453
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE131062_LR E453
05/03/14 12:18 PM
DEFORMITY the practice variation between surgeons so that more accurate follow-up information could be garnered. The study also addresses a clinical question that is paramount for both patient safety and cost of medical practice. A simple change in clinical practice, such as decreasing the frequency of postoperative radiographs in the first 6 months after surgery for this patient population, may lead to improvements in both patient safety and cost parameters. A reasonable schedule for obtaining postoperative radiographs in patients with AIS may be the following: PA and lateral radiographs prior to hospital discharge, chest radiograph for all patients with thoracoplasty as an inpatient, PA and lateral radiograph at 2 years and/or prior to final discharge from the office.
CONCLUSION Routine postoperative spine radiographs in the first 6 months after surgery rarely identified clinically relevant IR and/or NIR abnormalities in patients undergoing surgery for AIS. In addition, such radiographs often failed to detect IR and NIR abnormalities that were made evident on additional imaging studies prompted by the patient’s clinical course and physical examination. Protocols for obtaining routine, high-frequency, serial full-length spine radiographs in the first 6 months after an uncomplicated posterior spinal fusion for AIS should be re-examined to potentially minimize the ionizing radiation exposure for this patient population.
➢ Key Points Full-length spine radiographs in the first 6 months after posterior spine fusion for AIS have a low specificity for clinically significant radiographical abnormalities. Clinical signs and symptoms can more efficiently identify patients who will have a clinically significant radiographical finding.
E454
Utility of Postoperative Spine Radiographs • Pensak et al
Patients with thoracoplasty were 24 times more likely than patients who did not have thoracoplasty to have a clinically significant radiographical abnormality in the first 6 months after posterior spine fusion. Routine, full-length spine radiographs were obtained at a high frequency in the first 6 months after posterior spine fusion rarely identified clinically significant radiographical abnormalities.
References
1. Hoffman DA, Lonstein JE, Morin MM, et al. Breast cancer in women with scoliosis exposed to multiple diagnostic x-rays. J Natl Cancer Inst 1989;81:1307–12. 2. Doody MM, Lonstein JE, Stovall M, et al. Breast cancer mortality after diagnostic radiography: findings from the US Scoliosis Cohort Study. Spine (Phila Pa 1976) 2000;25:2052–63. 3. Bier VII US National Academy of Science, Report by the Committee on the Biological Effects of Ionizing Radiation. Washington, DC: National Academy of Science/National Research Council; 2005. 4. Howe GR, McLaughlin J. Breast cancer mortality between 1950 and 1987 after exposure to fractionated moderate-dose-rate ionizing radiation in the Canadian fluoroscopy cohort study and a comparison with breast cancer mortality in the atomic bomb survivors study. Radiat Res 1996;145:694–707. 5. Carman DL, Browne RH, Birch JG. Measurement of scoliosis and kyphosis radiographs. Intraobserver and interobserver variation. J Bone Joint Surg Am 1990;72:328–33. 6. Morrissy RT, Goldsmith GS, Hall EC, et al. Measurement of the Cobb angle on radiographs of patients who have scoliosis. Evaluation of intrinsic error. J Bone Joint Surg Am 1990;72:320–7. 7. Strasberg SM, Linehan DC, Hawkins WG. The accordion severity grading system of surgical complications. Ann Surg 2009;250: 177–86. 8. Chunguang Z, Yueming S, Limin L, et al. Convex short-length rib resection in thoracic adolescent idiopathic scoliosis. J Pediatr Orthop 2011;31:757–63. 9. Soultanis K, Pyrovolou N, Karamitros A, et al. The use of thoracoplasty in the surgical treatment of idiopathic scoliosis. Stud Health Technol Inform 2006;123:327–33.
www.spinejournal.com
April 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE131062_LR E454
05/03/14 12:18 PM
DEFORMITY
Utility of Early Postoperative Radiographs After Posterior Spinal Fusion for Adolescent Idiopathic Scoliosis Michael J. Pensak, MD,* Mark C. Lee, MD,† Jennifer A. Bayron, BS,‡ and Jeffrey D. Thomson, MD†
Study Design. Retrospective. Objective. Evaluate the ability of serial full-length spine radiographs to detect clinically significant implant-related (IR) and non–implant-related (NIR) radiographical abnormalities in the first 6 months after routine posterior spinal fusion for adolescent idiopathic scoliosis (AIS). Summary of Background Data. Patients with AIS are exposed to repeated doses of ionizing radiation during the course of their treatment with potential consequences for their long-term health. Postoperative algorithms for AIS often involve frequent standing plain radiographs during the first 6 months after surgery to detect IR and NIR abnormalities that may impact a patient’s clinical course. However, the actual clinical utility of such repeated spine radiographs has not been studied. Methods. Retrospective chart and radiographical review was conducted at a single institution for patients with AIS after posterior spinal fusion between 2007 and 2012. Radiographical abnormalities identified on full-length spine radiographs or additional imaging modalities in the first 6 postoperative months were grouped into IR or NIR findings. The findings were considered clinically significant if they resulted in a deviation from an anticipated postoperative course or additional interventions. Results. For 129 patients, 761 full-length spine radiographs were obtained in the first 6 postoperative months. Eight patients (11 radiographs) had IR or NIR abnormalities, with only 2 of these considered clinically significant. Seven of the remaining 121 were
From the *University of Connecticut Health Center, Farmington, CT; †Department of Orthopaedics, Connecticut Children’s Medical Center, Hartford, CT; and ‡University of Connecticut Medical School, Farmington, CT. Acknowledgment date: September 5, 2013. First revision date: November 12, 2013. Second revision date: December 18, 2013. Acceptance date: December 19, 2013. The device(s)/drug(s) that is/are the subject of this manuscript is/are not FDAapproved for this indication and is/are not commercially available in the United States. No funds were received in support of this work. Relevant financial activities outside the submitted work: board membership, payment for lecture. Address correspondence and reprint requests to Mark C. Lee, MD, Department of Orthopaedics, Connecticut Children’s Medical Center, 282 Washington Street, Hartford, CT 06106; E-mail: [email protected] DOI: 10.1097/BRS.0000000000000219
E450
identified to have IR or NIR abnormalities using other imaging modalities, with 2 considered clinically significant. The sensitivity and specificity of a full-length spine radiograph for detecting a clinically significant abnormality was 50% and 95%, respectively. Conclusion. Routine full-length spine radiographs used with high frequency in the first 6 months after posterior spinal fusion rarely detected a radiographical abnormality that resulted in a meaningful change to a patient’s clinical management. Blanket postoperative screening algorithms should be reconsidered to minimize patient radiation exposure. Key words: postoperative spine radiograph, scoliosis, posterior spine fusion, pleural effusion, hardware malposition, thoracoplasty, sensitivity, specificity, pedicle screw, hardware failure. Level of Evidence: 3 Spine 2014;39:E450–E454
S
tanding plain radiographs remain the “gold standard” for monitoring and clinical decision making in adolescent idiopathic scoliosis (AIS). However, the frequency of such radiographs in a patient’s treatment course often raises concern about the dangers of repeated exposure to ionizing radiation in a pediatric population. It has been estimated that a typical pediatric patient with scoliosis will receive approximately 22 plain radiographical examinations during a 3-year treatment period, with as many as 618 plain radiographs reported for a single patient.1,2 The BEIR VII National Academy of Science panel concluded that children are more sensitive to radiation than adults because they have more time to express a cancer and have more dividing cells upon which radiation acts.3 There is specific concern about the risk of breast cancer, as progressive scoliosis is seen most commonly in the female population and the dividing breast tissue of teenage girls is particularly sensitive to ionizing radiation.3,4 Full-length spine radiographs are often obtained at a high frequency in the first 6 months after posterior spinal fusion for AIS. The goal of these radiographs is to monitor for potential postoperative implant-related (IR) or non–implant-related (NIR) complications, as well as to assess for the adequacy of the deformity correction. However, the clinical value of these radiographs in the postoperative management of a patient after scoliosis surgery has not been studied. This work seeks to determine the ability of routine full-length plain radiographs
www.spinejournal.com
April 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE131062_LR E450
05/03/14 12:18 PM
DEFORMITY obtained during the first 6 months after scoliosis surgery to detect IR- and NIR-related abnormalities and whether such findings lead to a significant change in a patient’s overall clinical management.
MATERIALS AND METHODS After institutional review board approval by Connecticut Children’s Medical Center, a comprehensive review of that institution’s medical records between September 2007 and January 2012 was performed. Patients were included in the study if they were: (1) diagnosed and treated for AIS, (2) between 9 and 18 years of age at the time of diagnosis, (3) managed with a posterior spinal fusion for their AIS using a hybrid construct (screws, wires, hooks) with or without thoracoplasties, and (4) followed for a minimum of 6 months. Patients were excluded if there was a diagnosis of a nonidiopathic etiology for their scoliosis (neuromuscular, congenital, or syndrome associated) or if they had any condition that required additional serial radiation exposure in the postoperative period, such as malignancy or fracture. Patient demographic data were gathered from the charts (sex, age at surgery), as well as visit frequency in the postoperative period. All patients participated in a uniform follow-up algorithm after discharge: each patient was seen at 6 weeks from discharge, 3 months from surgery, 6 months from surgery, and 1 year from surgery with posteroanterior (PA) and lateral radiographs obtained at each visit. Scoliosis curve characteristics (Cobb angles of the major curve and Lenke classification) were determined from preoperative radiographs. All postoperative full-length plain radiographs were evaluated, with each view tabulated separately, and examined for evidence of radiographical abnormality as well as changes in postoperative curve characteristics. A change in Cobb measurements for the instrumented or noninstrumented curves of 10° or more in the course of follow-up was thought to be significant, given existing information on interobserver and intraobserver reliability for Cobb measures in scoliosis.5,6 All radiographs in the study were reviewed by institution radiologists per routine and the reports were also evaluated. Findings on imaging studies were grouped into “implant-related” (IR) findings, such as implant failure or malposition, and “non– implant-related” (NIR) findings, such as pleural effusion or pneumothorax. Inpatient and outpatient charts were further reviewed for the patient’s clinical course to determine if the patient was found to have IR- or NIR-related abnormalities on additional non–spine imaging studies, such as chest radiographs or computed tomographic (CT) scans. Charts were further investigated to determine if an identified radiographical abnormality was clinically significant. A clinically significant radiographical abnormality is defined as a finding on imaging studies that results in a deviation from the anticipated postoperative course requiring additional interventions such as unplanned follow-up studies, braces, medications, blood transfusions, drainage catheter placements, bedside procedures, interventional radiological procedures, or reoperation. The definition is adapted from the Accordion Severity Grading System of Surgical Complications.7 Spine
Utility of Postoperative Spine Radiographs • Pensak et al
Means were provided for continuous variables. Sensitivity and specificity analysis was performed for the full-length spine radiographs obtained in the postoperative period with the “gold standard” reference of advanced imaging studies or non–spine-related plain radiographs. Odds ratio was calculated for patients with thoracoplasty procedures to quantify the likelihood of a clinically significant radiographical finding in this subgroup.
RESULTS One hundred twenty-nine patients, 91 females and 38 males, with an average age at the time of surgery of 14.4 years (range, 10–20) met our inclusion criteria. The average major curve Cobb angle was 59° and the most frequent Lenke curve classification was type 2. The average number of levels fused was 10 (range, 4–14). Seven patients received thoracoplasties. The mean curve correction of the largest Cobb angle was 38.5°. No patient demonstrated a significant change in the Cobb measures of the instrumented or noninstrumented segments when comparing postoperative radiographs during 6 months. For the study cohort, a total of 761 spine radiographs were obtained in the first 6 months after surgery, with an average of 5.9 (range, 2–11) per patient. There were 411 PA radiographs (54%) and 350 lateral (46%) radiographs. The average number of postoperative visits was 3.3 (Table 1). One hundred twenty-one patients had normal spine radiographs (750 radiographs), whereas 8 patients had abnormal
TABLE 1. Patient Characteristics* N
129
Age at surgery (range), (yr)
14.4 (10–20)
Sex, n (%) Female
91 (71)
Male
38 (29)
Lenke curve type, n (%) 1
39 (30)
2
48 (37)
3
7 (5)
4
6 (5)
5
19 (15)
6
10 (8)
Major curve Cobb angle (range), (°) Total no. of patients with thoracoplasties No. of ribs resected (range)
59 (41–98) 7 5 (4–6)
No. of levels fused (range)
10 (4–14)
No. of spine radiographs in the first 6 mo postoperatively (range)
5.9 (2–11)
No. of visits in the first 6 mo postoperatively (range)
3.3 (1–7)
*Average values provided unless otherwise stated.
www.spinejournal.com
E451
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE131062_LR E451
05/03/14 12:18 PM
DEFORMITY
Utility of Postoperative Spine Radiographs • Pensak et al
Figure 1. Implant-related and non–implant-related abnormalities identified in the patient population, grouped according to findings that did or did not result in a significant change in clinical management.
spine radiographs (11 radiographs), yielding a normal radiograph rate of 98.6%. Of the 8 patients, 1 IR and 7 NIR abnormalities were identified (Figure 1). Two of the 8 patients with abnormal radiographs required a significant change in their clinical management. The first patient had a mild lumbar pedicle screw back out and was placed into a brace. Subsequent radiographs 1 year later showed no further change in position. The second patient had an infected pleural effusion that required readmission to the hospital, drainage with a pigtail catheter and intravenous antibiotics. Interestingly, although this latter patient had a mild pulmonary abnormality on full-length spine radiographs at the time of discharge, this was not recognized by the surgical team. A large pleural effusion was later identified on chest radiographs obtained after the patient returned to the hospital with complaints of shortness of breath. The 6 patients that required no significant change in their clinical management exhibited only mild pulmonary findings on standing spine radiographs. All 7 pulmonary abnormalities were identified on the PA full-length spine radiograph, whereas the IR complication was identified on the lateral spine radiograph. Of the 121 patients with normal spine radiographs, 7 patients were subsequently found to have radiographical abnormalities (3 IR, 4 NIR) from additional studies prompted by the patient’s clinical course. The additional studies consisted of chest and abdominal plain radiographs as well as CT scan. Two of these 7 patients required a significant change in their clinical management. The first patient had subtle neuromonitoring changes intraoperatively and complaints of radicular pain in the perioperative period, although no discrete IR abnormality was identified on a supine full-length spine radiograph. A CT scan demonstrated a medial breach of a thoracic pedicle screw and the patient required a return to the operating room for revision of the implant. The second patient had worsening abdominal pain in the postoperative period and underwent examination with a plain radiographical abdominal E452
series that revealed free air in the abdomen, although standing full-length spine radiographs were negative. Exploration by the general surgery team revealed a duodenal perforation through a pre-existing ulcer that required repair (Figure 2). Two of the 5 patients with no significant change in clinical management had IR abnormalities. The first patient underwent a chest CT scan in the emergency department shortly after discharge for evaluation of acute shortness of breath. A prominent thoracic pedicle screw tip was noted, but did not threaten any vascular or pulmonary anatomy and was therefore maintained. The second patient awoke from surgery with diffuse, bilateral numbness distal to the ankles despite normal neuromonitoring studies intraoperatively. A CT scan was obtained to evaluate implant position and identified a mild lateral breach of a lumbar pedicle screw. The lower extremity paresthesias were attributed to an epidural placed intraoperatively for pain control and resolved completely once the epidural had been removed. The remaining 3 patients had mild pleural effusions identified on chest radiograph obtained for shortness of breath that required no additional intervention. The calculated specificity of a full-length spine radiograph in detecting a clinically significant radiographical abnormality was 95%, whereas the sensitivity was 50%. Overall positive predictive value of a full-length spine radiograph for a clinically significant finding was 25%, whereas the negative predictive value was 98%. When evaluating the subgroup of 7 patients with thoracoplasties, 2 had a clinically significant radiographical abnormality. One was noted on full-length spine radiographs (the patient with the pleural effusion that later required drainage) and the other was noted on subsequent studies (the patient with the free air in the abdomen). The odds ratio of having a clinically significant radiographical finding after thoracoplasty compared with the group of patients without thoracoplasty was 24.
www.spinejournal.com
April 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE131062_LR E452
05/03/14 12:18 PM
DEFORMITY
Utility of Postoperative Spine Radiographs • Pensak et al
Figure 2. A, Anteroposterior postoperative day 6 standing plain radiograph showing a posterior spinal fusion construct from T4 to L3 with multilevel rightsided thoracoplasties. B, Anteroposterior upright abdominal radiograph obtained 14 hours after the standing full-length plain radiograph for worsening abdominal pain. The white arrows outline a pocket of free air in the abdomen from a pre-existing duodenal ulcer that had perforated.
DISCUSSION A protocol-based approach for managing patients is relied upon heavily by most surgeons to make uniform the delivery of care and to simplify the postoperative management of patients undergoing complex surgical procedures. However, these protocols should be critically evaluated after their implementation to determine if there is a tangible clinical benefit in light of potential risk to the patient and cost to the health system. The acquisition of routine, high-frequency postoperative full-length radiographs after surgery for scoliosis merits such investigation. It is unclear whether such repeated exposure to ionizing radiation adds meaningfully to the clinical management of pediatric patients in the first 6 months after surgery for AIS. From the data presented, the full-length spine radiography is a poor screening test for identifying clinically significant abnormalities in the early postoperative period, with a sensitivity of only 50%. Even when a full-length spine radiograph is abnormal, the low positive predictive value of 25% suggests that the test is poor at reporting a finding that will be of clinical significance. Although specificity of the full-length spine radiographs is high at 98%, suggesting good ability to detect the complete absence of a clinically significant abnormality, it is difficult to justify the use of 750 otherwise unnecessary plain radiographical studies in this pediatric patient population for such reassurance, especially when the overall prevalence of clinically significant abnormalities was low (4/129 = 3%). This study underscores the idea that clinical signs and symptoms can more reliably lead to a clinically significant radiographical finding than a blanket approach of serial radiographs. This is best highlighted by the one patient who was discharged from the hospital with a pulmonary infiltrate on spine radiographs not initially identified by the surgical team, Spine
but readmitted for management of a large pleural effusion. In addition, the study suggests that operative details significantly impact the likelihood of a clinically significant radiographical abnormality in the first 6 months after spine fusion. A patient with a thoracoplasty procedure during posterior spine fusion, a procedure with a known pulmonary complication rate approaching 25%, was 24 times more likely in this study than a patient without a thoracoplasty to have a clinically significant radiographical finding.8,9 If serial screening radiographs are to be obtained, their yield will be greater if applied in patients with surgical approaches that have a recognized higher rate of specific complications. Limitations to this investigation were several. First, the findings of the study are applicable only to the specific population of patients undergoing posterior spinal fusion for AIS. Patients undergoing additional anterior procedures or complex spine osteotomies may benefit from serial, routine radiographs as IR and NIR complications are potentially greater. Second, the definition of a “clinically significant” abnormality is debatable. The study defined a clinically significant radiographical abnormality as any finding that altered the routine postoperative clinical course of a patient. Therefore, the definition of clinical significance will depend on the surgeon or institution-specific concept of what is “routine” for the patient with a postoperative spine problem. Finally, this investigation lacked the use of any validated algorithms for assessing radiographs. We reviewed all radiographs independently and all radiographical interpretations from the radiologists. However, it is difficult to know whether there is some observer variability when identifying abnormalities on radiographs, particularly subtle pulmonary findings. Strengths of this study included the large number of patients evaluated, stringent inclusion criteria, and a uniform, pre-existing office follow-up algorithm that minimized www.spinejournal.com
E453
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE131062_LR E453
05/03/14 12:18 PM
DEFORMITY the practice variation between surgeons so that more accurate follow-up information could be garnered. The study also addresses a clinical question that is paramount for both patient safety and cost of medical practice. A simple change in clinical practice, such as decreasing the frequency of postoperative radiographs in the first 6 months after surgery for this patient population, may lead to improvements in both patient safety and cost parameters. A reasonable schedule for obtaining postoperative radiographs in patients with AIS may be the following: PA and lateral radiographs prior to hospital discharge, chest radiograph for all patients with thoracoplasty as an inpatient, PA and lateral radiograph at 2 years and/or prior to final discharge from the office.
CONCLUSION Routine postoperative spine radiographs in the first 6 months after surgery rarely identified clinically relevant IR and/or NIR abnormalities in patients undergoing surgery for AIS. In addition, such radiographs often failed to detect IR and NIR abnormalities that were made evident on additional imaging studies prompted by the patient’s clinical course and physical examination. Protocols for obtaining routine, high-frequency, serial full-length spine radiographs in the first 6 months after an uncomplicated posterior spinal fusion for AIS should be re-examined to potentially minimize the ionizing radiation exposure for this patient population.
➢ Key Points Full-length spine radiographs in the first 6 months after posterior spine fusion for AIS have a low specificity for clinically significant radiographical abnormalities. Clinical signs and symptoms can more efficiently identify patients who will have a clinically significant radiographical finding.
E454
Utility of Postoperative Spine Radiographs • Pensak et al
Patients with thoracoplasty were 24 times more likely than patients who did not have thoracoplasty to have a clinically significant radiographical abnormality in the first 6 months after posterior spine fusion. Routine, full-length spine radiographs were obtained at a high frequency in the first 6 months after posterior spine fusion rarely identified clinically significant radiographical abnormalities.
References
1. Hoffman DA, Lonstein JE, Morin MM, et al. Breast cancer in women with scoliosis exposed to multiple diagnostic x-rays. J Natl Cancer Inst 1989;81:1307–12. 2. Doody MM, Lonstein JE, Stovall M, et al. Breast cancer mortality after diagnostic radiography: findings from the US Scoliosis Cohort Study. Spine (Phila Pa 1976) 2000;25:2052–63. 3. Bier VII US National Academy of Science, Report by the Committee on the Biological Effects of Ionizing Radiation. Washington, DC: National Academy of Science/National Research Council; 2005. 4. Howe GR, McLaughlin J. Breast cancer mortality between 1950 and 1987 after exposure to fractionated moderate-dose-rate ionizing radiation in the Canadian fluoroscopy cohort study and a comparison with breast cancer mortality in the atomic bomb survivors study. Radiat Res 1996;145:694–707. 5. Carman DL, Browne RH, Birch JG. Measurement of scoliosis and kyphosis radiographs. Intraobserver and interobserver variation. J Bone Joint Surg Am 1990;72:328–33. 6. Morrissy RT, Goldsmith GS, Hall EC, et al. Measurement of the Cobb angle on radiographs of patients who have scoliosis. Evaluation of intrinsic error. J Bone Joint Surg Am 1990;72:320–7. 7. Strasberg SM, Linehan DC, Hawkins WG. The accordion severity grading system of surgical complications. Ann Surg 2009;250: 177–86. 8. Chunguang Z, Yueming S, Limin L, et al. Convex short-length rib resection in thoracic adolescent idiopathic scoliosis. J Pediatr Orthop 2011;31:757–63. 9. Soultanis K, Pyrovolou N, Karamitros A, et al. The use of thoracoplasty in the surgical treatment of idiopathic scoliosis. Stud Health Technol Inform 2006;123:327–33.
www.spinejournal.com
April 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE131062_LR E454
05/03/14 12:18 PM
