|Home | About | Journals | Submit | Contact Us | Français|
The majority of the 700,000 osteoporotic vertebral compression fractures (VCFs) that occur annually in the United States affect women. The total treatment costs exceed $17 billion and approximate the total costs of breast cancer ($13 billion) and heart disease ($19 billion). Balloon-assisted percutaneous vertebral augmentation with bone cement (kyphoplasty) reportedly reduces VCF-related pain and accelerates return of independent functional mobility. Kyphoplasty may decrease overall cost of VCF treatment costs by reducing use of posttreatment medical resources.
We evaluated complications, mortality, posthospital disposition, and treatment costs of kyphoplasty compared with nonoperative treatment using the Nationwide Inpatient Sample database.
We identified 5766 VCFs (71% female) in patients 65 years of age or older with nonneoplastic VCF as the primary diagnosis in nonroutine hospital admissions; 15.3% underwent kyphoplasty. Demographic data, medical comorbidities, and fracture treatment type were recorded. Outcomes, including complications, mortality, posthospital disposition, and treatment costs, were compared for each treatment type.
Women were more likely to be treated with kyphoplasty than were men. Patients undergoing kyphoplasty had comorbidity indices equivalent to those treated nonoperatively. Kyphoplasty was associated with a greater likelihood of routine discharge to home (38.4% versus 21.0% for nonoperative treatment), a lower rate of discharge to skilled nursing (26.1% versus 34.8%) or other facilities (35.7% versus 47.1%), a complication rate equivalent to nonoperative treatment (1.7% versus 1.0%), and a lower rate of in-hospital mortality (0.3% versus 1.6%). Kyphoplasty was associated with higher cost of hospitalization (mean $37,231 versus $20,112).
Kyphoplasty for treatment of VCF in well-selected patients may accelerate the return of independent patient function as indicated by improved measures of hospital discharge. The initially higher cost of treatment may be offset by the reduced use of posthospital medical resources.
Level III, therapeutic study. See Guidelines for Authors for a complete description of levels of evidence.
Osteoporosis affects greater than 55% of women older than age 55 years and is characterized by decreased bone mineral density [44, 46, 50, 57] resulting in increased bone fragility and a higher susceptibility to fracture. In the United States, osteoporosis leads to over 700,000 vertebral compression fractures (VCFs) per year and affects women twice as often as men . Treatment of osteoporosis and associated fractures incurs a cost exceeding $17 billion each year [44, 46, 57] including $6 billion for the indirect value of lost productivity of the patient and caregivers .
Initial treatment of VCF includes activity modification and analgesic medications [48, 49]. Patients are often unable to tolerate activities of daily living and may require bed rest to control pain. The clinical manifestation of a VCF can lead to chronic pain, deformity, and disability [39, 41]. The pain and fracture kyphosis can compromise respiratory function [38, 40, 51]. Many patients sustain serious cardiovascular, musculoskeletal, metabolic, and immune complications related to immobility and bedrest [9, 43]. Patients typically affected by VCF often cannot tolerate the complications of nonoperative care. It is not uncommon for a patient to be admitted to a hospital for treatment, discharged, and readmitted to treat complications with medical resources used at each stage . The result can be a downward-spiral of complications, functional decline, and a higher risk of death as a result of the VCF [6, 7, 12, 32, 38, 40, 51].
Treatment that accelerates the return of patient function can potentially reduce both the medical risks of VCF as well as the economic burden of the disease. Techniques of vertebral augmentation have been developed to treat VCFs refractory to nonoperative care. Percutaneous vertebroplasty was introduced in 1987 initially as a treatment for aggressive vertebral hemangiomas  and was later modified as kyphoplasty (Medtronic Sofamor Danek, Memphis, TN) . Both involve pedicle cannulation and injection of polymethylmethacrylate (PMMA) bone cement into the fracture [2, 14, 40, 41]. In kyphoplasty, an inflatable bone tamp is used to prepare a confined space for PMMA injection . Both techniques reportedly relieve fracture pain and improve functional outcome at both short- and long-term followup [2, 14, 25, 35, 40, 41]. The procedure is not without risk, however. Cement extrusion from the vertebra reported in up to 30% of vertebroplasty  has been improved to 0–8.6% by using the bone tamp in kyphoplasty [40, 56, 58]. Additionally, the initial cost of kyphoplasty may be higher than nonoperative care (including implanted PMMA and disposable instrumentation), but may be offset by reduced use of medical resources after hospital discharge. No large-scale, nationwide studies have been performed to confirm the effect of kyphoplasty on patient function or economic burden of VCF treatment.
We therefore evaluated these parameters using a nationwide database presuming kyphoplasty performed in a population at high risk for medical complications of treatment would lead to (1) improved functional outcome as measured by a higher rate of routine hospital discharge to home; (2) an equivalent or lower rate of complications; and (3) a reduction in the use of medical resources after hospital discharge that could potentially offset the expectedly higher initial cost of surgical intervention.
The Nationwide Impatient Sample (NIS) is a database maintained and published by the Healthcare Cost and Utilization Project of the Agency for Healthcare Research and Quality. This database has been used to evaluate use trends and outcomes for a variety of procedures in spine surgery [1, 5, 10, 15, 20, 26, 28, 33, 36, 47, 52, 59] as well as the impact of comorbidities on the outcome of spine procedures [3, 11, 17, 18]. Such a database allows a larger patient sample to be amassed in relatively short period of time and for a lower cost than could be achieved by even a large multicenter trial; they also can offer insight into the actual use, complications, and outcomes of a procedure in “real life” and without the restrictive criteria imposed for inclusion in controlled trials. The database includes hospital discharge information for approximately eight million encounters from over 1000 hospitals and represents a sample of 20% of all nonfederally funded hospitals nationwide. The database includes many variables, including demographic, diagnostic, procedural, length of stay, payor, and total charge information. Individual encounters are given a unique identifying number but maintain no link to the patient’s personal identifying information. Users of the database, therefore, cannot identify individual patients whose information is included. We used the 2005 edition of the NIS to identify all encounters in patients 65 years of age or older with a primary diagnosis code for thoracic or lumbar vertebral compression fracture (International Classification of Diseases, 9th Revision, Clinical Modification [ICD-9-CM] codes 805.2 for closed thoracic and 805.4 for closed lumbar fractures). We included encounters representing emergent or urgent admissions only and excluded those patients admitted for elective fracture treatment. We excluded all VCF identified as pathologic fractures resulting from primary or metastatic tumors of the spine. We also excluded all fractures that underwent other surgical intervention such as vertebroplasty, spinal decompression, or fusion. These exclusions therefore identified a sample of patients admitted for treatment of VCF for whom no definitive treatment had likely been electively planned. The patients were separated into two groups: those who were treated with kyphoplasty and those who were treated nonoperatively.
We identified 5766 admissions for treatment of thoracic or lumbar VCF in patients 65 years of age or older who met the inclusion criteria (Table 1). The mean age of the patients was 81.1 years. The sample was comprised predominantly of women (71.3%) and fractures of the lumbar spine (62.9%). A large majority of admissions occurred in hospitals located in large metropolitan areas (79.0%). Kyphoplasty was performed in 15.3% of all admissions. In 66.2% of patients, the Deyo-modified Charlson Comorbidity Index  was equal to zero or one. The mean age of patients treated with kyphoplasty was 1.3 years younger than those treated without kyphoplasty (nonoperative group, Table 1). The percentage of patients in each comorbidity category was similar between those treated operatively and nonoperatively. Women were more likely to be treated with kyphoplasty than were men. Fractures of the thoracic spine were more likely to be treated with kyphoplasty than were fractures of the lumbar spine. Patients admitted to hospitals located in large metropolitan areas were more likely to be treated with kyphoplasty than were patients admitted to hospitals in nonmetropolitan areas. The NIS contains a variable that tabulates all procedures performed during a patient’s hospital admission. Patients on whom kyphoplasty was performed underwent an average of 1.5 more procedures than did patients treated nonoperatively (Table 1).
We assessed the outcome of treatment by first determining the rate of routine discharge (discharge to home), discharge to skilled nursing facilities and other facilities, and posthospital use of home care for each treatment group. We assessed the complication rate of each treatment type by evaluating the NIS for in-hospital mortality and specific complications, including infection, pneumonia, venous thrombosis, and pressure ulcer. Additionally, general overall complications were determined for each treatment group by evaluating each encounter for the presence of Diagnosis Clinical Classification System (DXCCS) code 238, “Complication of surgical procedure or medical care.” This classification system is a tool used by the NIS for grouping ICD-9-CM codes into comorbidity categories . For example, ICD-9-CM codes for all cancers of the head and neck are grouped into DXCCS Code 11, “Cancer of head and neck.” The DXCCS code 238 groups all reported complications of surgical procedures or medical care. The economic impact of treatment was evaluated by determining the total length of stay and total charge of hospitalization for each encounter.
Statistical analysis was performed using STATA/IC Data Analysis and Statistical Software, Release 10 (StataCorp LP, College Station, TX). Descriptive statistics were used to measure dispersion of population variables, including gender, location of fracture, comorbid conditions, and occurrence of kyphoplasty. Student’s t-test was used to compare the mean ages and number of procedures of patients treated with kyphoplasty with those treated nonoperatively. Chi square analysis was used to compare the proportion of patients treated with kyphoplasty with those treated nonoperatively with respect to gender, location of fracture, comorbidity category, and hospital location. To evaluate the impact of kyphoplasty, Student’s t-test was used to compare the mean length of stay and total hospital charge of patients treated with kyphoplasty with those treated nonoperatively. Chi square analysis was used to compare the proportion of patients treated with kyphoplasty with those treated nonoperatively with respect to in-hospital mortality, complications, and posthospital disposition. Logistic regression models were created to evaluate the probability of discharge to home, skilled nursing or other facilities, use of home care, in-hospital mortality, and other complications. We controlled for patient comorbidity using the Deyo-modified Charlson Comorbidity Index [8, 19].
At the time of discharge, patients treated with kyphoplasty were more likely to be discharged home with no special accommodations (routine discharge: odds ratio 2.59, p < 0.001) or to home with home care (odds ratio 1.51, p < 0.001; Table 2). For nonroutine discharges, patients who were treated with kyphoplasty were less likely to be discharged to a skilled nursing facility or other facility (odds ratios 0.62 and 0.59, respectively, p < 0.001 for both).
The overall rate of reported complications for hospital treatment of VCF was 1.1% (Table 2). The overall rate of complications was similar for patients treated with kyphoplasty compared with those treated nonoperatively. Specific complications of treatment, including pressure ulcer, deep venous thrombosis, all causes of infection, and pneumonia, were equivalent when comparing the two groups. The rate of in-hospital mortality for patients treated with kyphoplasty was lower (odds ratio 0.52, p = 0.003) than that of patients treated nonoperatively.
Patients treated with kyphoplasty remained hospitalized for a mean of 0.7 days longer than patients treated nonoperatively (Table 2). The overall cost of hospital treatment for VCF was approximately $17,000 higher per patient treated with kyphoplasty than patients treated nonoperatively.
The impact of osteoporosis on aging patients, particularly women, is astounding and has been considered a “silent epidemic” worldwide . In the United States alone, over 700,000 VCFs occur annually. Percutaneous augmentation of vertebral fractures by vertebroplasty [2, 14, 42] or kyphoplasty [25, 35, 37, 40] has been used as a minimally invasive method of stabilizing such fractures and thereby reducing the associated pain of fracture motion in patients who have failed nonoperative care, including spinal bracing, activity modification, or narcotic analgesic medications. Many reports to date suggest that at both short- and long-term followup, well-selected patients can expect pain relief that limits narcotic requirements and enhances function in up to 90% of cases treated with kyphoplasty [16, 24, 27, 35, 40, 56, 58]. Recent randomized, controlled studies have suggested vertebral augmentation procedures are no more efficacious than sham placebo procedures [4, 34]. Both Buchbinder and colleagues  and Kallmes and colleagues  evaluated first-generation vertebroplasty for the treatment of painful VCF and reported no difference in their outcome measures. Kallmes and colleagues  noted a trend toward a higher proportion of patients with clinical improvement at 1 month in the vertebroplasty group. Notably, they recorded a 43% crossover of patients from sham to vertebroplasty compared with a 12% crossover from vertebroplasty to sham. This disparity casts substantial doubt over the validity of the conclusions of their intention-to-treat analysis of the efficacy of vertebroplasty. Additionally, neither study evaluated kyphoplasty for treatment of symptomatic VCF. It is unknown whether conclusions made regarding the outcome of vertebroplasty can be generalized to that of kyphoplasty as a result of the inherent differences between the procedures as well as the documented variability in the length of stay and hospital discharge outcomes between the two procedures . Buchbinder et al.  further commented that their results are contrary to many other reports of the efficacy of vertebroplasty. Additionally, the data reporting the efficacy of kyphoplasty have been recently synthesized by Taylor and colleagues  in a systematic review and meta-analysis. The authors concluded kyphoplasty is more effective than medical management of VCF. Despite these recent reports questioning the benefit of vertebral augmentation, the best evidence to date appears to support the use of kyphoplasty for treatment of symptomatic VCF [16, 24, 27, 35, 40, 56, 58]. No large-scale, nationwide studies have been performed to confirm the effect. We therefore examined the complications, mortality, posthospital disposition, and treatment costs using the NIS of the Healthcare Cost and Utilization Project of the Agency for Healthcare Research and Quality and specifically asked whether kyphoplasty would lead to (1) improved functional outcome as measured by a higher rate of routine hospital discharge to home; (2) an equivalent or lower rate of complications; and (3) a reduction in the use of medical resources after hospital discharge that could potentially offset the expectedly higher initial cost of surgical intervention.
Despite the benefits of a database, studies using database information, including ours, are fraught with certain limitations. First, any database study in which ICD-9-CM codes are used in the selection process assumes accurate information was entered by the hospital coding staff. Because hospital reimbursement is often based on a diagnosis-related group fee schedule, it is possible coding could be selected in a manner such that the maximally reimbursed code is listed as the primary diagnosis for admission. Although such potential for variability can typically be equalized by the volume of patients included in the NIS, we followed precedent to include only encounters in which VCF was listed as the primary admission diagnosis [10, 15, 26]. Second, any research using the NIS database cannot assess traditionally supported and validated outcome measures such as 30-day and 1-year mortality, pain scores, and outcome questionnaires (SF-36, Oswestry Disability Index, and others). This disadvantage has been surmounted by using previously described surrogate measures of functional outcome including in-hospital complications [3, 5, 10, 11, 33, 52, 53], in-hospital mortality [1, 3, 10, 33, 53], and patient disposition at hospital discharge [10, 11, 33, 52, 53]. Third, evidence is mounting to suggest patients who undergo kyphoplasty experience up to 36% higher risk of subsequent VCF resulting from the increased biomechanical stiffness of the treated vertebral body [22, 29, 45]. The NIS database contains the information pertaining to a specific hospitalization for a given patient whose identification has been removed from the record. A given patient cannot be tracked into the future to assess the occurrence of subsequent fracture or long-term complications. Similarly, patients treated nonoperatively cannot be followed for posthospital discharge complications or readmissions. For these reasons, we chose to evaluate only the in-hospital, short-term results of treatment for a population at high risk for poor outcome of the VCF, knowing that both nonoperative and operative care could equally pose a substantial risk of complications specific to each treatment choice. Such drawbacks could only be overcome with a large-scale, nationwide, multicenter trial.
We studied a sample of patients at high risk for a poor medical outcome resulting from a VCF. The sample represents a population of patients admitted for hospital treatment of a VCF under nonelective conditions and without an elective treatment plan. Approximately 15% of the patients underwent kyphoplasty for treatment. Patients treated with kyphoplasty were of equivalent comorbidity as those treated nonoperatively. Although the age of patients treated with kyphoplasty was 1.3 years younger than those treated without, the clinical importance is most likely negligible as a result of the biologic variability of octogenarians as well as equivalence of comorbidity between the two groups.
Patients treated with kyphoplasty were approximately 2.5 times more likely to undergo routine discharge to home and approximately 40% less likely to use posthospital inpatient care (skilled nursing or other facility). Other authors have used hospital disposition as a measure of the impact of a procedure [10, 33, 36, 52] or of a comorbidity or complication [11, 53] on the functional status of a patient at the time of discharge. Compared with a patient discharged to a skilled nursing or rehabilitation facility, a patient undergoing routine discharge from a hospital can be assumed to require no or minimal special accommodations and, therefore, have a higher level of independent function. Our results suggest patients treated with kyphoplasty are afforded a higher level of independent function at hospital discharge than are patients treated nonoperatively. Additional evidence for this conclusion is suggested by the increased likelihood of a patient treated with kyphoplasty to be discharged to home with home care. Again, this suggests a patient discharged to home, albeit with the help of a home health worker, has a higher level of function than that of one discharged to a subacute inpatient care facility.
Compared with patients treated nonoperatively, patients treated with kyphoplasty have an equivalent rate of in-hospital complications and a lower rate of in hospital mortality. Other studies using the NIS database have considered the occurrence of in-hospital complications and mortality as a measure of the immediate and early outcome of a surgical intervention [1, 3, 5, 10, 17, 18, 33, 36, 52] or of the impact of a comorbidity or complication on outcome [11, 53]. The lower rate of in-hospital mortality in patients treated with kyphoplasty suggests that, in our patient sample, kyphoplasty affects the restoration of independent functional mobility in a manner that leads to a lower likelihood of developing a life-threatening complication of treatment. Although it is certainly plausible the ultimate cause of death in nonoperatively treated patients may have contraindicated surgery and simply have been undiagnosed and, therefore, not listed in hospital discharge coding (for example, massive pulmonary embolism or myocardial infarction with no clinical data or autopsy), it is equally plausible such conditions would have occurred with the same frequency in patients after kyphoplasty. Finally, patients who were treated with kyphoplasty had an average of 1.5 more procedures per hospital admission than patients treated nonoperatively. Despite this increase in the number of procedures, with the patient exposed to risks of complication with each unique procedure, the incidence of complications was still equivalent if a patient underwent kyphoplasty or nonoperative care.
The average inpatient charge for those undergoing kyphoplasty was approximately $17,000 more than that of nonoperative care. This value includes the cost of the surgical procedure as well as the implanted PMMA and disposable instruments used during the procedure. A portion of this value likely includes the additional 0.7 days of hospital admission as well as the additional testing required to medically optimize and “clear” a patient for surgery. The added initial inpatient cost is likely to be offset by the reduction in the cost of posthospital inpatient care associated with admission to skilled nursing and other facilities. As discussed, patients who undergo kyphoplasty appear to have a higher likelihood of avoiding the use of these posthospital inpatient facilities. Other authors have used the discharge status of a patient as a measure of the use of posthospital discharge resources [3, 10, 11, 52]. A routine discharge to home can be assumed to use fewer medical resources than an admission to another facility and therefore decrease the overall cost of care for the VCF. Furthermore, the treatment of VCF with kyphoplasty in a population at high risk for medical complications of treatment may also reduce the cycle of discharge, complications, and readmission to inpatient hospital care that we have seen in our own practices and others have anecdotally noted. Although discharge data serve well as surrogate measures of resource use, a true determination of the cost of posthospital care can be made only by following individual patients longitudinally in a large-scale, nationwide trial. No controlled trials have evaluated this parameter to our knowledge. Recently, however, Ström and colleagues evaluated the cost-effectiveness of kyphoplasty in the United Kingdom . Their model included estimates of the probability and cost of subsequent fractures in patients treated both nonoperatively and with kyphoplasty but did not appear to include estimates of the rate or cost of rehospitalization for nonoperatively treated fractures. They found kyphoplasty a cost-effective treatment for symptomatic VCF. If kyphoplasty is able to reduce the hospital readmission rate for patients with VCF, it is likely the initially higher cost of treatment will be offset even further.
In summary, we evaluated the immediate, short-term impact of kyphoplasty on hospital discharge status, in-hospital complications and death, and hospital cost for treatment of symptomatic VCF in a sample of the elderly population comprised mostly of women. We found kyphoplasty leads to a higher likelihood of routine hospital discharge with an equivalent rate of complications and lower rate of in-hospital mortality than nonoperative treatment for symptomatic VCF in a high-risk population. The data suggest kyphoplasty leads to more rapid return of independent functional mobility at a low risk of complications compared with nonoperative care. In contrast to recent reports questioning the efficacy of vertebral augmentation [4, 34], our results support the use of kyphoplasty and confirm the results of other reports [16, 24, 27, 35, 40, 56, 58] that kyphoplasty may accelerate the restoration of patient functional mobility. The improvement in hospital discharge parameters may offset the initially higher cost of treatment and lead to a reduction in the use of medical resources after hospital discharge.
Each author certifies that he or she has no commercial associations (eg, consultancies, stock ownership, equity interest, patent/licensing arrangements, etc) that might pose a conflict of interest in connection with the submitted article.
Each author certifies that his or her institution approved or waived approval for the human protocol for this investigation and that all investigations were conducted in conformity with ethical principles of research.
This work was performed at the Department of Orthopaedic Surgery, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, USA.