Spinal Fracture in Patients with Ankylosing Spondylitis Cohort Definition, Distribution of Injuries, and Hospital Outcomes

CERVICAL SPINE

Spinal Fracture in Patients With Ankylosing Spondylitis Cohort Definition, Distribution of Injuries, and Hospital Outcomes

Adam M. Lukasiewicz, MSc,Ã Daniel D. Bohl, MD,y Arya G. Varthi, MD,Ã Bryce A. Basques, MD,y Matthew L. Webb, AB,Ã Andre M. Samuel, BBA,Ã and Jonathan N. Grauer, MDÃ

occurred in 13.1% of patients. Patients were treated with fusion Study Design. A retrospective cohort. in 49.9% of cases. In-hospital adverse events occurred in 29.4% Objective. The aim of this study was to characterize spinal of patients, and 6.6% of patients died during their admission. fractures in patients with ankylosing spondylitis. Conclusion. More than 10% of patients had fractures in more Summary of Background Data. Patients with ankylosing than 1 region of the spine. There is a high risk of adverse events spondylitis are susceptible to fractures of the spinal column, in this population, and 6.6% of patients died during their even from minor trauma. However, the literature describing inpatient stay. These results provide clinicians with a better patients with ankylosing spondylitis and spinal fractures consists understanding of the distribution and the high morbidity and largely of case reports and small case series. The purpose of this mortality of fractures in the ankylosed spine. study is to better characterize fractures of the ankylosed spine, Key words: ankylosing spondylitis, distribution of injuries, including the patient population, locations of fracture, and national inpatient sample, spinal cord injury, spine fracture outcomes in a large, nationally representative sample. Level of Evidence: 3 Methods. All patients with diagnoses of both fracture of the Spine 2016;41:191–196 spinal column and ankylosing spondylitis admitted between 2005 and 2011 were identified in the National Inpatient Sample (NIS). Patient demographics, fracture regions, and complications were characterized with descriptive statistics. The associations between injury characteristics and outcomes were assessed using nkylosing spondylitis (AS) is a seronegative spon- Poisson regression. dyloarthropathy that primarily affects the joints Results. A total of 939 patients with ankylosing spondylitis and ligaments of the spinal column. The disease admitted with a spinal fracture were identified in NIS. The A usually develops around 25 years of age and affects men average age was 68.4 14.7 years, and 85% of patients were 1–3 Æ more than women. As the disease progresses, chronic male. Cervical fractures were the most common (53.0%), inflammation of the spinal joints, ligaments, and interverte- followed by thoracic (41.9%), lumbar (18.2%), and sacral bral discs typically results in a classic ‘‘bamboo spine’’ in (1.5%). Spinal cord injury was present in 27.5% of cervical which the fused spine acts as a continuous, rigid axial fractures, 16.0% of thoracic fractures, and 21.1% of cases support.4,5 overall. Fractures involving more than 1 region of the spine Patients with AS are at a higher risk of spinal fractures than the general population, and even very low energy injuries can cause spinal fractures among patients with this disease.6 The increased risk of spine fractures in the AS From the ÃDepartment of Orthopaedics and Rehabilitation, Yale School of population is generally thought to arise both from reduced Medicine, New Haven, CT; and yDepartment of Orthopedic Surgery, Rush University Medical Center, Chicago, IL vertebral bone quality and from amplified forces due to the 7–9 Acknowledgment date: May 12, 2015. First revision date: August 5, 2015. rigid spine acting as a lever. Acceptance date: August 11, 2015. Previous reports have found that the lower cervical spine The manuscript submitted does not contain information about medical is the most common site for fracture in the AS popu- device(s)/drug(s). No funds were received in support of this work. Relevant lation,10,11 and that there is an increased risk of noncontig- financial activities outside the submitted work include consultancy, expert 12–16 testimony, grants. uous fractures in the AS population. Spinal fractures in Address correspondence and reprint requests to Jonathan N. Grauer, MD, patients with AS are also often unstable, 3-column injuries Department of Orthopaedics and Rehabilitation, Yale School of Medicine, that are prone to displacement. Accordingly, the risk of 800 Howard Avenue, New Haven, CT 06510; E-mail: jonathan.grauer@ spinal cord injury (SCI) is thought to be higher in patients yale.edu with a spinal fracture in the context of AS, than the general 17 DOI: 10.1097/BRS.0000000000001190 spinal fracture population. Spine www.spinejournal.com 191 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. CERVICAL SPINE Spinal Fracture in Patients With Ankylosing Spondylitis Lukasiewicz et al

AS is an uncommon condition, and although patients risk when the outcome is relatively common and odds ratios with the condition are at a higher risk of spinal fracture, less are not easily interpretable. Patient comorbidities and age than one-fifth of patients have spinal fractures.18,19 As a were controlled for using CCI. Injuries were categorized as result, most of the existing literature on the spinal fractures cervical, thoracic, lumbar, or multiregional. All calculations in this patient population is limited to relatively small case were performed using Stata 13.1 (StataCorp, College series; a meta-analysis of all case series of patients with Station, TX). All tests were 2-tailed with significance at a spinal fracture and AS between 1980 and 2007 found a total equal to 0.05. of 345 patients. Studies included in this meta-analysis evaluated 4.5 patients on average, and the largest study included RESULTS 37.5,20 To better understand spinal fractures in patients with AS, Patient Characteristics we used a large, nationally representative cohort of patients In total, 939 patients with a diagnosis of AS admitted for a from the National Inpatient Sample (NIS) to characterize spinal fracture were identified in the NIS between 2005 and the demographics and injuries of AS patients with spinal 2011. Patient characteristics are summarized in Table 1. The fracture. In addition, we examined the hospital course of average age in this cohort was 68.4 14.7 years (mean these patients to describe and quantify the risk of mortality standard deviation), and 798 (85.0%)Æ patients were male. and general health adverse events. Æ The most common comorbid conditions in this population were hypertension (541 patients, 57.6%), uncompli- cated diabetes mellitus (199, 21.2%), fluid/electrolyte METHODS AND MATERIALS disturbances (195, 20.8%), chronic pulmonary disease A retrospective cohort of patients admitted to a hospital (158, 16.8%), and nutrient-deficiency anemia (147, between 2005 and 2011 for spinal fracture with a diagnosis 15.7%). Approximately 50% of patients had a CCI of 3 of AS was identified in the NIS.21 The NIS is a large or less. administrative database that captures a representative The distribution of fractures along the spine is depicted in sample of all inpatient hospital stays in the United States Figure 1 and Table 2. The cervical spine was the region most each year. Patients with AS were identified by an Inter- commonly fractured, with fractures in 498 patients national Classification of Diseases, Ninth Revision (ICD-9) (53.0%). Fractures in the thoracic spine were present in code of 720.0. Patients with a spinal fracture were identified 393 patients (41.9%), fractures in the lumbar spine in 171 using ICD-9 codes in the 805 and 806 ranges, and only patients (18.2%), and fractures in the sacrum in 14 patients patients with a spinal fracture listed as their primary or (1.5%). Fractures in multiple regions of the spine were secondary diagnosis were included. present in 123 patients (13.1%). Ten patients had fractures Patient comorbidities were extracted from predefined in all regions of the spine. SCI occurred in 198 patients variables in the NIS database, as well as ICD-9 diagnosis codes from the list of patient diagnoses. A modified Charl- son comorbidity index (CCI), a measure of patient comorbidity burden, was calculated for each patient as described TABLE 1. Demographics 22 previously. Length of stay was extracted directly from the Characteristic No. % NIS database. Spinal fractures were characterized by region (cervical, Total 939 100% thoracic, lumbar, and sacral) and by SCI according to ICD-9 Male 798 85.0% codes. Fractures in multiple regions of the spine were Age < identified by the presence of ICD-9 codes documenting 30 4 0.4% fractures in each of the involved regions. 30–40 22 2.3% Treatment details were discerned using ICD-9 procedure 40–50 94 10.0% codes. Treatments were grouped into spinal fusion and other 50–60 148 15.8% procedures. In the absence of a code indicating a surgical 60–70 185 19.7% procedure, the patient was assumed to have been managed 70–80 233 24.8% conservatively with a thoracolumbar brace, cervical collar, >80 253 26.9% halo, or other nonsurgical treatment. CCI 0 97 10% Demographics and spinal fracture distributions were 1 100 11% examined using descriptive statistics. The associations 2 148 16% between injury characteristics and length of stay were inves- 3 158 17% tigated with ordinary least squares multivariate regression. 4 188 20% The associations between injury characteristics and 5 108 12% mortality and adverse events during the inpatient stay were 6 140 15% assessed using Poisson regression with robust error var- þ iance.23 This approach allows for an estimation of relative CCI indicates Charlson Comorbidity Index.

192 www.spinejournal.com February 2016 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. CERVICAL SPINE Spinal Fracture in Patients With Ankylosing Spondylitis Lukasiewicz et al

53% Cervical Spine

6.5% Cervical and Thoracic Spine 41.9% 1.1% 1.3% Cervical, Thoracic Spine Cervical and Thoracic and Lumbar Spine 4.3% Lumbar Spine Thoracic and Lumbar Spine

18.2% Lumbar Spine

1.5% Sacrum

Figure 1. Distribution of spine fractures by region. The leftmost diagram shows the total proportions of fractures in each region. The remaining four diagrams show the proportion of each combination of multiregional fractures.

(21.2%). Two-thirds of SCI cases (137) were due to cervical Table 4 summarizes the results of the multivariate spine fractures, and one-third (63) were due to thoracic analysis for analyses that controlled for gender and CCI. spine fractures. Patients were treated with spinal fusion in Relative to patients with cervical fractures, patients with 469 cases (49.9%). thoracic fractures had reduced mortality (relative risk The in-hospital adverse events associated with the above- [RR] 0.43, p 0.014). The mortality associated with described fractures are summarized in Table 3. Notably, 62 lumbar¼ or multiple¼ region fractures was not significantly patients (6.6%) died. Regarding other adverse events, the different than that of cervical fractures. Patients with SCI most common were urinary tract infection (90 patients, were at a greater risk of mortality (RR 3.45, P < 0.001), 9.6%), intubation (80, 8.5%), acute kidney injury (66, adverse events (RR 2.18, P < 0.0001),¼ and longer length 7.0%), and pneumonia (59, 6.3%). Overall, 29.4% of of stay (6.3 days longer,¼ P < 0.0001). patients had an inpatient stay complicated by an adverse event. The mean length of hospital stay in this patient DISCUSSION population was 9.0 8.9 days, with an interquartile range Patients with AS are at a high risk for spinal fracture, and Æ of 4 to 11 (Fig. 2). often have unstable fractures with a high risk of SCI. As AS is a rare condition and spinal fracture is a relatively rare complication, the details of demographics, injuries, and outcomes in this population have not been well elucidated. TABLE 2. Fracture Patterns in the Ankylosed Spine

Fracture Type No. % TABLE 3. In-Hospital Adverse Events Region Adverse Event No. % Cervical 498 53.0% Thoracic 393 41.9% Death 62 6.6% Lumbar 171 18.2% Urinary tract infection 90 9.60% Sacral 14 1.5% Intubation 80 8.50% With spinal cord injury 198 21.1% Acute kidney injury 66 7.00% Cervical 137 14.6% Pneumonia 59 6.30% Thoracic 63 6.7% Cardiac arrest 17 1.80% Lumbar 0 0% Venous thromboembolism 17 1.80% Sacral 0 0% Myocardial infarction 14 1.50% Multiregion injuries 123 13.1% Surgical site infection 9 1.00% Cervical and thoracic 61 6.5% Peripheral nerve injury 3 0.30% Cervical and lumbar 12 1.3% Stroke 2 0.20% Thoracic and lumbar 40 4.3% Wound dehiscence 2 0.20% Cervical, thoracic, and lumbar 10 1.1% Any adverse event 276 29.4%

Spine www.spinejournal.com 193 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. CERVICAL SPINE Spinal Fracture in Patients With Ankylosing Spondylitis Lukasiewicz et al

double the previous estimates, possibly reflecting the

250 advanced disease state and poorer overall health in patients with AS complicated by spinal fracture. The high prevalence of pulmonary disease might be attributable to the pulmon-

200 ary involvement of AS, which is thought to reduce lung volumes through both decreased chest expansion and inflammatory changes to the lung tissue itself.32 Previous work has largely emphasized that the cervical 150 spine is the primary site of fracture in AS patients. In one

Count recent meta-analysis of published case series, 78% of fractures were in the cervical spine.20 Although the cervical 100 spine was the most common site of fracture in our study, cervical fractures were present in just slightly more than half of patients. The underlying cause of this mild discrepancy is 50 unclear. Clinicians need to appreciate that although cervical fractures are common and concerning, thoracic fractures were present in about two-fifths of this population, and 0 0 20 40 60 80 100 lumbar fractures were present in about one-fifth. Hence, the Length of Stay (Days) entire spine needs to be thoroughly evaluated in all AS Figure 2. Histogram of length of stay. patients with possible spinal fracture. Patients with isolated thoracic injuries were at a reduced Although the demographics of the population in this risk of mortality compared with patients with cervical study were generally similar to demographics in previous injuries. However, patients with fractures in multiple work using smaller case series, there were some differences. regions had the same risk of mortality as patients with The average age in this study of 68.4 years was higher than isolated cervical fractures. Although the risk of immediate the average age in most previously published case series mortality may not increase with multiple fracture sites, these examining this population. In those case series, the average additional fractures could cause substantial morbidity age was usually between 50 and 60 years.20,24–26 The reason outside the primary hospitalization. Future work should for this older than expected population is unclear and may establish the effect of multiple fractures beyond the initial reflect improving management of the disease course, thereby hospital stay. improving survival and delaying fractures.27,28 The pro- Although AS patients are known to occasionally have portion of female patients was roughly in the middle of multiple distinct spine fractures,5,16,29 these injuries have the ranges reported in previous case series.20,29 The pro- not been well described. In this study, more than 10% of portion of patients treated surgically in this study was also patients had an injury in more than 1 region of the spine. As similar to previous results showing that slightly more than ICD-9 coding only differentiates between broad regions of half of patients were treated operatively.20 the spine, distinct but more closely related fractures—for The comorbidities in this population were particularly instance, both upper and lower cervical vertebrae—could notable for the high prevalence of hypertension and pul- not be distinguished. Hence, the true proportion of patients monary disease. Individuals with AS are known to have with multiple, distinct fractures may well be higher. These greater rates of hypertension, and increased cardiovascular results reinforce the idea that clinicians need to have a high mortality compared with the general population, possibly suspicion of fracture throughout the spine in AS patients, reflecting both the general inflammatory component of AS even if a primary fracture site is found. and the limitations on activity with arthritis.30,31 However, The risk of SCI in this study was substantial, with more the prevalence of hypertension in this study was more than than one-fifth of patients having a neurological deficit.

TABLE 4. Injury Characteristics and Outcomes Mortality Adverse Events LOS Factor RR P RR P Days P þ Fracture region (compared with cervical) Thoracic 0.44 0.014 0.97 0.763 0.73 0.268 Lumbar 0.33 0.129 1.04 0.828 0.03 0.974 Multiple regions 1.03 0.932 1.15 0.316 0.35 0.686 Spinal cord injury 3.44 <0.0001 2.18 <0.0001 6.29 <0.0001 Bold indicates significance. LOS indicates length of stay.

194 www.spinejournal.com February 2016 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. CERVICAL SPINE Spinal Fracture in Patients With Ankylosing Spondylitis Lukasiewicz et al

However, the rate of SCI here is lower than that of most Sample between 2005 and 2011, this study previous studies, which have reported rates of SCI as high as examined the patterns of spinal fracture in 81% and 97%.5,20 This discrepancy may reflect the com- this population. paratively reduced proportion of cervical injury in our SCI occurred in 21.1% of patients, and 13.1% of study, as, sensibly, no SCI associated with lumbar and sacral patients had fractures in more than 1 region of injuries was reported. Among patients with cervical frac- the spine. tures, only 28% of patients had recorded SCI. This result In-hospital mortality in the population was 6.6%, could reflect poor capture of SCI in the NIS database, a focus and 29.4% of patients had an adverse event on the most clinically severe fractures in case reports, during the inpatient stay. or both. Although patients in this population with SCI will have poorer long-term functional outcomes, they are also at a greater risk of immediate mortality and morbidity, and they References stay longer in the hospital. Clinicians need to be aware of the 1. Braun J, Sieper J. Ankylosing spondylitis. Lancet 2007;369:1379– significant effect of SCI on immediate outcomes, and coun- 90. sel patients appropriately. 2. Feldtkeller E, Khan MA, van der Heijde D, et al. Age at disease The rate of complications in this study is in line with onset and diagnosis delay in HLA-B27 negative vs. positive 20 patients with ankylosing spondylitis. Rheumatol Int previous estimates of close to 30%. However, our study 2003;23:61–6. only included general health complications during the inpa- 3. Braun J, Bollow M, Remlinger G, et al. Prevalence of spondylar- tient stay, and therefore likely substantially underestimates thropathies in HLA-B27 positive and negative blood donors. Arthritis Rheum 1998;41:58–67. the true likelihood of adverse events in this patient popu- 4. Weinstein PR, Karpman RR, Gall EP, et al. Spinal cord injury, lation if the postoperative course after discharge were also spinal fracture, and spinal stenosis in ankylosing spondylitis. J considered. The mortality rate of 6.6% in this study was Neurosurg 1982;57:609–16. similar to most previous estimates, although less than rates 5. Einsiedel T, Schmelz A, Arand M, et al. Injuries of the cervical 4 spine in patients with ankylosing spondylitis: experience at two from some smaller case series. However, as with adverse trauma centers. J Neurosurg Spine 2006;5:33–45. events, only mortality during the primary inpatient stay was 6. Trent G, Armstrong GW, O’Neil J. Thoracolumbar fractures in captured, and the true mortality in the immediate postinjury ankylosing spondylitis. High-risk injuries. Clin Orthop Relat Res period is likely far higher. 1988;227:61–6. 7. Hitchon PW, From AM, Brenton MD, et al. Fractures of the There are several limitations to this study. Detailed data thoracolumbar spine complicating ankylosing spondylitis. J Neu- on neurological status, both before and after the injury, were rosurg 2002;97 (Spine 2):218–22. not available. Functional outcomes were also not available 8. Graham B, Van Peteghem PK. Fractures of the spine in ankylosing for this dataset. Although this study could evaluate patients spondylitis. Spine (Pa 1976) 1989;14:803–7. 9. Olerud C, Frost A, Bring J. Spinal fractures in patients with for fractures in multiple regions of the spine, multiple ankylosing spondylitis. Eur Spine J 1996;5:51–5. cervical, thoracic, or lumbar fractures would not have been 10. Menezes AH. Management of cervical spine injuries in patients detected. Furthermore, ICD-9 coding does not specify frac- with ankylosing spondylitis. J Neurosurg 1990;72:210–5. 11. Thumbikat P, Hariharan RP, Ravichandran G, et al. Spinal cord ture type or exact vertebral level, so these data were not injury in patients with ankylosing spondylitis: a 10-year review. available. As mentioned before, NIS data only cover the Spine (Pa 1976) 2007;32:2989–95. inpatient hospital stay. Hence, adverse events are likely 12. Osgood C, Abbasy M, Mathews T. Multiple spine fractures in higher in the immediate postinjury period. ankylosing spondylitis. J Trauma 1975;15:163–6. 13. Anwar F, Khayer A Al, Fraser M, et al. Concomitant cervical and In conclusion, patients with AS complicated by vertebral thoracic spinal fractures in ankylosing spondylitis: a case report fracture are at a high risk for inpatient complications and and review of literature. Inj Extra 2009;40:242–5. mortality. Although the cervical spine is the most common 14. Broom MJ, Raycroft JF. Complications of fractures of the cervical location for spinal fractures, fractures in the thoracolumbar spine in ankylosing spondylitis. Spine (Pa 1976) 1988;13:763–6. 15. Akman MN, Karatas M, Kilinc¸ S, et al. Double spinal cord injury spine are more common than previously reported. SCI in a patient with ankylosing spondylitis. Spinal Cord occurred in one-fifth of patients, and more than one-tenth 1999;37:305–7. of patients had a fracture in more than 1 region of the spine. 16. Fox MW, Onofrio BM, Kilgore JE. Neurological complications of Clinicians need to be vigilant for fractures throughout the ankylosing spondylitis. J Neurosurg 1993;78:871–8. 17. Alaranta H, Luoto S, Konttinen YT. Traumatic spinal cord injury spine when evaluating AS patients with suspected spinal as a complication to ankylosing spondylitis. An extended report. fractures. Clin Exp Rheumatol 2002;20:66–8. 18. Ghozlani I, Ghazi M, Nouijai A, et al. Prevalence and risk factors of osteoporosis and vertebral fractures in patients with ankylosing Key Points spondylitis. Bone 2009;44:772–6. 19. Ralston SH, Urquhart GD, Brzeski M, et al. Prevalence of vertebral compression fractures due to osteoporosis in ankylosing spondy- Patients with AS are highly susceptible to litis. BMJ 1990;300:563–5. spinal fractures. 20. Westerveld L, Verlaan JJ, Oner FC. Spinal fractures in patients Using 939 patients hospitalized for spinal fracture with ankylosing spinal disorders: a systematic review of the liter- in the setting of AS in the National Inpatient ature on treatment, neurological status and complications. Eur Spine J 2009;18:145–56. Spine www.spinejournal.com 195 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. CERVICAL SPINE Spinal Fracture in Patients With Ankylosing Spondylitis Lukasiewicz et al

21. Agency for Healthcare Research and Quality Healthcare Cost and 28. Wanders A, Heijde Van Der D, Landewe´ R, et al. Nonsteroidal Utilization Project (HCUP). Introduction to the HCUP Nation- antiinflammatory drugs reduce radiographic progression in wide Inpatient Sample. 2013. patients with ankylosing spondylitis: a randomized clinical trial. 22. Basques Ba, Bohl DD, Golinvaux NS, et al. Preoperative factors Arthritis Rheum 2005;52:1756–65. affecting length of stay after elective anterior cervical discectomy and 29. Caron T, Bransford R, Nguyen Q, et al. Spine fractures in patients fusion with and without corpectomy: a multivariate analysis of an with ankylosing spinal disorders. Spine (Pa 1976) 2010;35: academic center cohort. Spine (Phila Pa 1976) 2014;39:939–46. E458–64. 23. Zou G. A modified Poisson regression approach to prospective 30. Han C, Robinson DW, Hackett MV, et al. Cardiovascular disease studies with binary data. Am J Epidemiol 2004;159:702–6. and risk factors in patients with rheumatoid arthritis, psoriatic 24. Taggard DA, Traynelis VC. Management of cervical spinal frac- arthritis, and ankylosing spondylitis. J Rheumatol 2006;33: tures in ankylosing spondylitis with posterior fixation. Spine (Pa 2167–72. 1976) 2000;25:2035–9. 31. Peters MJ, Van Der Horst-Bruinsma IE, Dijkmans Ba, et al. 25. Hunter T, Dubo H. Spinal fractures complicating ankylosing Cardiovascular risk profile of patients with spondylarthropathies, spondylitis. Ann Intern Med 1978;88:546–9. particularly ankylosing spondylitis and psoriatic arthritis. Semin 26. Finkelstein J, Chapman J, Mirza S. Occult vertebral fractures in Arthritis Rheum 2004;34:585–92. ankylosing spondylitis. 1999;37:444–7. Spinal Cord 32. Feltelius N, Hedenstro¨m H, Hillerdal G, et al. Pulmonary 27. Braun J, Brandt J, Listing J, et al. Treatment of active ankylosing involvement in ankylosing spondylitis. Ann Rheum Dis 1986;45: spondylitis with infliximab: a randomised controlled multicentre 736–40. trial. Lancet 2002;359:1187–93.