Background To explore the national inpatient trends, regional variations, associated diagnoses, and outcomes of vertebral augmentation (vertebroplasty and kyphoplasty) in the USA from 2004 to 2017.
Methods Data from the National Inpatient Sample were used to study hospitalization records for percutaneous vertebroplasty and kyphoplasty. Longitudinal projections of trends and outcomes, including mortality, post-procedural complications, length of stay, disposition, and total hospital charges were analyzed.
Results Following a period of decreased utilization from 2008 to 2012, hospitalizations for vertebroplasty and kyphoplasty plateaued after 2013. Total hospital charges and overall financial burden of hospitalizations for vertebroplasty and kyphoplasty increased to a peak of $1.9 billion (range $1.7–$2.2 billion) in 2017. Overall, 8% of procedures were performed in patients with a history of malignancy. In multivariable modeling, lung cancer (adjusted OR (aOR) 2.6 (range 1.4–5.1)) and prostate cancer (aOR 3.4 (range 1.2–9.4)) were associated with a higher risk of mortality. The New England region had the lowest frequency of routine disposition (14.1±1.1%) and the lowest average hospital charges ($47 885±$1351). In contrast, 34.0±0.8% had routine disposition in the West Central South region, and average hospital charges were as high as $99 836±$2259 in the Pacific region. The Mountain region had the lowest number of procedures (5365±272) and the highest mortality rate (1.2±0.3%).
Conclusion National inpatient trends of vertebroplasty and kyphoplasty utilization remained stable after a period of decline from 2008 to 2012, while the financial burden of hospitalizations increased. Despite recent improvements in outcomes, significant regional variations persisted across the USA.
Data availability statement
Data are available upon reasonable request. Data from the Healthcare Cost and Utilization Project of the Agency for Healthcare Research and Quality, United States Department of Health and Human Service.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Percutaneous vertebral augmentation interventions, including vertebroplasty and kyphoplasty, are minimally invasive procedures for the treatment of symptomatic vertebral compression deformities using bone cement or an augmenting device (eg, balloon-assisted), respectively.1 2 Vertebroplasty was developed in the late 1980s, and introduced to clinical practice in the 1990s. Balloon-assisted vertebroplasty (kyphoplasty) was subsequently developed as a modification of vertebroplasty in an effort to restore vertebral body height in addition to stabilizing the vertebral column.3 4 Though osteoporotic fractures are the most common underlying diagnoses, malignant vertebral compressions may also benefit from vertebroplasty and kyphoplasty.4 The decision of whether to pursue vertebroplasty or kyphoplasty and their relative efficacy is controversial in the literature.3–5
In 2009, three distinct randomized controlled trials (RCTs) were published demonstrating no significant difference between vertebroplasty and sham surgery or conservative management.6–8 Following the publication of these studies, the American Academy of Orthopedic Surgeons (AAOS) and the American College of Radiology (ACR) guidelines were updated, limiting recommendations for vertebroplasty and kyphoplasty.9 10 Analysis of Medicare claims data from 2002 to 2014 demonstrated decreased utilization of vertebroplasty, likely influenced by RCTs from 2009 as well11 . In the years following, several studies demonstrated the safety and efficacy of vertebroplasty and kyphoplasty compared with conservative management, including open-label RCTs like Vertos II12 and FREE.13 These findings were followed by multiple double-blind, placebo controlled RCTs including VAPOUR14 and Vertos IV,15 showing the superiority of vertebroplasty over placebo treatment and sham procedure in long-term pain control of patients with acute osteoporotic spinal fractures. A recent review by De Leacy et al 16 summarized the evidence from available RCTs and highlighted the need for population-based, outcome-related and economic studies. Nonetheless, data regarding recent changes in vertebroplasty and kyphoplasty utilization are scarce. Moreover, regional variations and underlying diagnoses of patients receiving vertebroplasty and kyphoplasty have not been systematically studied to date.
The purpose of our study is to explore the national inpatient trends, regional variations, associated diagnoses, and outcomes of vertebral augmentation (vertebroplasty and kyphoplasty) in the USA from 2004 to 2017.
We used data from the National Inpatient Sample (NIS) of the Healthcare Cost and Utilization Project (HCUP) and analyzed discharge records from 2004 to 2017. The International Classification of Diseases, Ninth Revision, Clinical Modification (ICD9-CM) codes were used to identify the diagnoses and procedures that were associated with each hospitalization record through the third quarter of 2015. The International Classification of Diseases, 10th Revision, Clinical Modification (ICD10-CM) codes were used to identify the diagnoses and procedures from the fourth quarter of 2015 through 2017, which corresponded to a change in the US administrative data. For each hospital admission, one primary procedure (principal procedure) and up to 24 secondary procedures were recorded. Likewise, for every hospitalization, one principal diagnosis and up to 29 secondary diagnosis codes were recorded. Procedure codes related to vertebroplasty and kyphoplasty were identified (online supplemental appendix 1), and hospitalizations with vertebroplasty or kyphoplasty as the primary procedure were used for subsequent analysis. Only de-identified data were used throughout the study and no Institutional Review Board approval was required or sought. The authors have no relevant conflict of interest with regard to this study.
Basic characteristics and associated diagnoses
Basic characteristics of the study population were classified into demographics (age, gender, race, and location), financial variables (household income and principal healthcare payer/insurance), and hospital characteristics. Clinical indications for each hospital stay were extracted based on the associated diagnoses. We followed the HCUP guidelines in providing estimates with limited data points. Only diagnoses with an incidence of 1% or greater were included in the final analysis. All of the extracted primary diagnoses were indicative of pathological fractures without granularity regarding the level of injury and presence of neoplastic disease. An all-inclusive evaluation of all secondary discharge diagnoses codes was undertaken. The ICD codes related to underlying malignancies with known risk of pathological vertebral fracture were mapped (online supplemental appendix 2), including multiple myeloma, lung cancer, breast cancer, prostate cancer, renal and urinary tract cancers, and gastrointestinal cancers.17 18 The number of records with an associated diagnosis of melanoma were limited and were therefore not included in the final analysis.
Due to complexities in extracting the underlying comorbidities from diagnosis codes in administrative data, we followed the recommendation from the Agency for Healthcare Research and Quality (AHRQ), and used HCUP’s Elixhauser Comorbidity Software (beta version) to identify the comorbidities associated with each hospitalization record. A format library was created, mapping diagnosis codes into comorbidity indicators, and excluding conditions that were related to complications or were related to the primary diagnosis. Next, the identified codes were used to create Elixhauser comorbidity variables, comprising 29 common comorbid conditions.19 The Elixhauser comorbidity measures have been studied extensively and shown to be associated with clinical outcomes, including mortality and readmissions.20 21
Multiple hospital characteristics were extracted, including region (Northeast, Midwest, South, or West), size (small, medium, or large; based on number of beds with cut-off points resulting in terciles that were balanced across region, location, and teaching status strata), proprietorship (governmental, private non-profit, or private investor), and teaching status (rural, urban non-teaching, or urban teaching). National data were regionally stratified across the nine US Census Bureau divisions, including New England (Maine, New Hampshire, Vermont, Massachusetts, Rhode Island, Connecticut), Middle Atlantic (New York, Pennsylvania, New Jersey), East North Central (Wisconsin, Michigan, Illinois, Indiana, Ohio), West North Central (Missouri, North Dakota, South Dakota, Nebraska, Kansas, Minnesota, Iowa), South Atlantic (Delaware, Maryland, District of Columbia, Virginia, West Virginia, North Carolina, South Carolina, Georgia, Florida), East South Central (Kentucky, Tennessee, Mississippi, Alabama), West South Central (Oklahoma, Texas, Arkansas, Louisiana), Mountain (Idaho, Montana, Wyoming, Nevada, Utah, Colorado, Arizona, New Mexico), and Pacific (Alaska, Washington, Oregon, California, Hawaii).
Four main outcomes were specifically documented for every hospitalization record and analyzed in this study, including death, length of stay (days), disposition status, and total hospital charges (US dollars). Following HCUP’s categorization of disposition, “routine” disposition was defined as discharge to home or self-care. This contrasts with other forms of disposition including transfer to short-term hospital, transfer to skilled nursing facility or other/intermediate care facility, home healthcare, against medical advice, and unknown discharges. Post-procedural complications were defined as the fifth studied outcome. To define the complication variable, potential post-procedural diagnoses related to vertebroplasty and kyphoplasty were identified and mapped to corresponding ICD codes (details of the final 455 specific and non-specific codes that were selected are presented in online supplemental appendix 3). Only codes that were specifically designed to indicate post-procedural events were included in this group.
We used data from 2004 to 2017 for trend analysis, and pooled data from 2012 to 2017 to evaluate associated diagnoses and outcomes. Implemented in 2012, the NIS had a major change in methodology from a nationwide sample of US community hospitals (2011 and before) to a national sample of hospitalizations representing 97% of the US population (2012 and later). Non-federal, short-term general, and other specialty hospitals were included, while long-term acute care hospitals and hospital units of institutions were excluded from the NIS.22 Trends of hospitalizations and outcomes for vertebroplasty and kyphoplasty were studied using joinpoint trend analysis. Regression models with the minimum number of joinpoints were fitted and annual percentage changes (APCs) were calculated. Statistical significance of adding new joinpoints were assessed using a Monte Carlo Permutation method.23 Overall financial burden attributed to vertebroplasty and kyphoplasty hospitalizations were calculated using the number of hospitalizations and total hospital charges. Upper and lower (95th percentile) uncertainty intervals were calculated. Longitudinal trajectory of trends across years were depicted using cubic spline interpolation with joinpoints as the control points and confidence intervals derived by bootstrapping24 .
Next, data from 2012 to 2017, including a total of 42 937 066 hospitalization records, were screened. Each code used to map the procedures, associated diagnoses, comorbidities, and complications were reviewed by two authors. Potential disagreements were resolved by consulting the senior author. Hospitalizations with a primary procedure of vertebroplasty or kyphoplasty were identified (n=29 430 records) and weighted following the survey structure of the NIS. Estimates of outcomes were studied according to basic characteristics and associated diagnosis. Continuous variables were compared using Student’s t-test. Categorical variables were compared using the Mann-Whitney U test. We used multivariable regression modeling to study the association between underlying malignancy and outcomes, while adjusting for the effect of basic characteristics. Adjusted odds ratios (aOR) were calculated with non-malignant causes as the reference group. The length of stay and total hospital charges were categorized into binary variables for the purpose of modeling. Sensitivity analysis using different thresholds was performed to confirm the robustness of findings for different thresholds. Finally, regional variations in basic characteristics, associated diagnosis, and outcomes were calculated for the nine US Census Bureau regions. Estimates were categorized in pantiles for presentation.
Following HCUP recommendations and requirements for systematic analysis of the NIS data, and to calculate nationally representative estimates, a survey approach was utilized throughout the study, considering the sampling design, strata, and hospital clusters within the NIS structure. All estimates are reported along with standard errors of estimates (±SE) or with 95% uncertainty intervals in parentheses. PASW (v18, Chicago, IL) was used for data handling. Analysis was performed using SAS (v9.4. Cary, NC).
Analyzing hospitalization trends for vertebroplasty and kyphoplasty, we found three distinct periods starting with an initial phase of increasing utilization from 2004 to 2008 (table 1). After a period of decreasing trends from 2008 to 2011–2012 (annual change of −20.2% (range −34.3% to −12.4%) and −9.6% (range −32 .% to −0.6%] for vertebroplasty and kyphoplasty, respectively), a plateau state was reached and trends remained stable for the remainder of the study period (figure 1). In terms of outcomes, we observed improvements in mortality after vertebroplasty (0.9±0.2% to 0.4±0.2%, p<0.05), decreased frequency of routine disposition (32.5±2.7 to 25.8±1.5 for vertebroplasty, and 66.4±2.1 to 25.6±0.8 for kyphoplasty, p<0.05), and increasing total hospital charges ($29 545±$1798 to $65 914±$2165 for vertebroplasty, and $30 094±$1284 to $84 788±$1874 for kyphoplasty, p<0.05) during the study period (table 2). The financial burden of hospitalizations for vertebroplasty and kyphoplasty followed a similar pattern, increasing from $186 million (range $136–$242 million) in 2004 to $1.5 billion (range $1.2–$1.9 billion) in 2008, followed by a decrease to $1.2 billion (range $0.9–$1.5 billion) in 2011 and a steady annual increase afterwards to $1.9 billion (range $1.7–$2.2 billion) in 2017 (figure 2).
Table 3 demonstrates the basic characteristics of the study population and the associated outcomes. Statistically significant variations in disposition and total hospital charges were observed for all demographic, financial, and hospital-related characteristics. Having more than three comorbidities was associated with unfavorable outcomes, including higher mortality (1.1±0.1 vs 0.4±<0.1, p<0.05), post-procedural complication (1.6±0.1 vs 1.0±0.1, p<0.05), lengthier stay (7.2±0.1 days vs 5.4±<0.1 days, p<0.05), lower routine disposition (22.1±0.4% vs 33.0±0.4%, p<0.05), and higher total hospital charges ($76 538±$807 vs $65175±$552, p<0.05). Detailed evaluation of individual comorbidities and outcomes is presented in online supplemental appendix 4.
Overall, 8.4% of procedures were performed in patients with a history of malignancy (figure 3). In multivariable modeling and compared with patients with no known malignancy, lung cancer (aOR 2.6 (range 1.4–5.1)) and prostate cancer (aOR 3.4 (range 1.2–9.4)) were associated with a higher risk of mortality (table 4). With the exception of prostate cancer (aOR 1.4 (range 0.9–2.0), which did not reach the threshold for statistical significance), all other malignancies were associated with lengthier hospital stays compared with hospitalizations with no malignancy (aOR for stays >8 days ranging from 1.6 (range 1.1–2.1) for gastrointestinal cancers to 2.3 (range 1.9–2.8) for multiple myeloma; p<0.05). Presence of underlying malignancy was associated with higher total hospital charges (aOR for hospital charges >$75 k ranging from 1.7 (range 1.2–2.4) for prostate cancer to 2.7 (range 2.2–3.2) for multiple myeloma; p<0.05) compared with hospitalizations with no record of malignancy.
We found significant regional variations in basic characteristics and comorbidities across the USA (table 5). Small, statistically significant variations in comorbidities, including higher frequency of comorbidities in the East North Central (44.7±0.7% having more than three comorbidities) and underlying malignancy in the Middle Atlantic (10.1±0.7%) regions, were observed. Of note, only 4.8±0.5% of hospitals in the East North Central region were private investor hospitals compared with up to 37.7±2.4% in the Mountain region (p<0.05). Likewise, the East North Central region had the highest frequency of hospitalizations with Medicare or Medicaid as the primary payer (89.3±0.4%). Finally, we found significant regional variations in the number of procedures and outcomes across the USA (table 6, figure 4). The South Atlantic region had the highest number of procedures (38 505±929 from 2012 to 2017), while the Mountain region had the fewest number of procedures (5365±272; p<0.05) and the highest mortality rate (up to 1.2±0.3% compared with 0.4±0.1% in the East South Central region; p<0.05). The New England region had the lowest rates of routine disposition (14.1±1.1% compared with up to 34.0±0.8% in the West South Central regions; p<0.05) and the lowest average hospital charge ($47 885±$1351), while the average hospital charge was more than double in the Pacific region ($99 836±£2259; p<0.05).
We analyzed the longitudinal trends of hospital stays for vertebroplasty and kyphoplasty in the USA, and found three distinct periods: an early era of increasing trends from 2004 to 2008, a period of decline from 2008 to 2011–2012, and a plateau from 2012 to 2017. The decreasing trends were likely associated with the publication of initial RCTs showing no significant benefit for vertebroplasty compared with conservative, non-surgical management in 2009.7 8 Several subsequent RCTs highlighted the improvements in outcomes following vertebroplasty and kyphoplasty in patients with osteoporotic12 13 25 26 and cancer-related27 28 compression fractures, which might explain the plateau period noted from 2013 to 2017. Previous limited analyses of the NIS29–31 highlighted the declining trends that were attributed to the publication of RCTs questioning the efficacy of vertebroplasty and kyphoplasty. Our study elucidated the second change in the longitudinal trajectory of vertebroplasty and kyphoplasty hospitalizations from 2013 to 2017, with an estimated number of hospitalizations for kyphoplasty that was not significantly different from the pre-2009 era. The observed increase in the utilization of vertebroplasty and kyphoplasty may highlight the effect of open-label RCTs like Vertos II and FREE12 13 as well as double-blind placebo controlled RCTs like VAPOUR14 and Vertos IV15 showing the safety, efficacy and sustained pain control after vertebroplasty and kyphoplasty. Moreover, prior analyses of the NIS were mostly limited to the pre-2012 period when NIS consisted of a sample of selected US community hospitals. The changes in the NIS design, survey structure, and sampling strategy that were implemented in 2012 allowed us to calculate nationally representative estimates that were also stratified by US regions. While the NIS provides all-payer nationally representative estimates of hospitalizations, claims-based studies may provide a superior assessment of utilization patterns in outpatient settings and ambulatory clinics. Moreover, long-term mortality benefits after discharge, follow-up assessments and readmissions may also be better tracked using claims-based datasets with linkage to other nationally representative surveys.32 Overall, the observed changes in utilization patterns32–34 and relative safety and efficacy of vertebroplasty and kyphoplasty35 were consistent with previous studies of Medicare beneficiaries.
While utilization trends have remained stable since 2012, we observed an increasing trend of hospital costs, resulting in a growing financial burden reaching peak of $1.9 billion (range $1.7–£2.2 billion) in 2017. Several studies from multiple countries and health settings underscored the cost-benefit ratio of kyphoplasty compared with conservative strategies,36–38 as it reduced hospital admissions, decreased the length of stay, and facilitated earlier discharge.39 40 Chen et al studied the Medicare population and found that despite higher hospital costs, patients receiving kyphoplasty had 20% lower mortality, shorter hospital stays and lower risk of complications including pneumonia and decubitus ulcers41 . Svedbom et al studied the cost-effectiveness of non-surgical management, vertebroplasty and kyphoplasty using a Markov simulation model. They concluded that kyphoplasty can be a cost-effective approach when the incremental benefits in mortality and health-related quality of life are taken into account37 . Future large-scale comparative studies are needed to further elucidate the factors differentiating the utilization pattern and outcomes for kyphoplasty versus vertebroplasty.
It is important to highlight that the calculated financial burden associated with inpatient stays for vertebroplasty and kyphoplasty includes all direct and indirect hospitalization costs related to vertebral compression fractures and do not represent procedure-related costs. In fact, the increasing costs associated with hospitalizations for vertebral augmentation interventions are in line with increasing costs of inpatient healthcare in the USA, derived by multiple factors including administrative costs.42 43 Health-related and economic policies should consider multiple factors including cost-effectiveness and relative improvements in outcomes of patients undergoing vertebroplasty and kyphoplasty versus conservative treatments. The influence of public healthcare policies on funding and utilization of vertebroplasty and kyphoplasty, and their subsequent impact on short- and long-term morbidity and mortality of patients with vertebral compression fractures, were shown to be considerable16 .
We found basic characteristics and underlying comorbidities to be associated with hospital-related outcomes, including length of stay, routine disposition, and total hospital charges. In line with previous literature,29 31 comorbidities contributed to a delay in care, increased length of stay, and increased cost of hospitalizations39 . Even in the presence of multiple comorbidities (>3), hospitalizations for vertebroplasty and kyphoplasty were associated with only a small risk of mortality (1.1±0.1%) and complications (1.6±0.1%). Our findings were in accordance with prior studies demonstrating the safety and efficacy of vertebroplasty and kyphoplasty in patients with osteoporotic and malignant fractures.12 13 27 44 Nevertheless, to the best of our knowledge, no prior study has made a direct comparison of the outcomes based on the presence of an underlying malignancy. Compared with patients with no known malignancy, presence of malignancy was associated with increased length of stay and higher hospital charges. Moreover, lung cancer (aOR 2.6 (range 1.4–5.1)) and prostate cancer (aOR 3.4 (range 1.2–9.4)) were associated with higher mortality.
Finally, we observed significant regional variations in basic characteristics, comorbidities, and outcomes. In the 2012–2017 period, the South Atlantic region had the highest number of hospitalizations for vertebroplasty and kyphoplasty (38 505±929), while the Mountain region had the lowest (5365±272). Interestingly, the Mountain region had the highest mortality rate (up to 1.2±0.3%), while the South Atlantic region had one of the lowest rates of mortality (0.5±0.1%). Though these patterns might suggest better outcomes in regions with higher volumes of cases, further evaluation based on hospital-specific volumes fell beyond the scope of our study. With regards to disposition patterns and hospital charges, the New England region had the lowest frequency of routine disposition (14.1±1.1%) and the lowest average total hospital charges ($47 885±$1351). Meanwhile, average hospital charges in the Pacific region was as high as $99 836±$2259.
Strengths of our study included the systematic and comprehensive review of available data, and an extensive, analytical approach in utilizing a large, nationally representative database. Identifying baseline comorbidities and post-procedural complications has been shown to be challenging and cumbersome in large-scale surveys based on administrative data45 . For comorbidities, we took a systematic approach and used the Elixhauser comorbidity index, which has been well validated in several prior studies across different fields.20 21 For complications, we used a comprehensive approach by identifying the potential post-procedural complications that were related to vertebroplasty and kyphoplasty. We performed an extensive review of all secondary codes, and mapped the relevant codes that were specifically designed to incur a post-procedural incident. With this approach, we avoided overestimating the complications by excluding secondary codes with no determined timestamp46 . Our study had several limitations related to the administrative nature of the NIS data. In terms of associated diagnoses, no time-to-event analysis was performed. Due to privacy concerns and the study design, NIS did not provide any longitudinal follow-up evaluations of patients or individual hospitals. Our assessment of hospital costs and regional variations were also bound to aggregate measures due to limited granularity of the NIS data.
In conclusion, our study of NIS data found that inpatient trends of vertebroplasty and kyphoplasty utilization remained stable after a period of decline from 2008 to 2012, while the financial burden of hospitalizations increased. The increase in the financial burden of hospitalizations for vertebroplasty and kyphoplasty is in line with increasing costs of inpatient healthcare in the USA, rather than procedure-related costs. Even in the presence of comorbidities, hospitalizations for vertebroplasty and kyphoplasty were associated with only a small risk of mortality and complications. Despite recent improvements in outcomes, significant regional variations persisted across the USA.
Data availability statement
Data are available upon reasonable request. Data from the Healthcare Cost and Utilization Project of the Agency for Healthcare Research and Quality, United States Department of Health and Human Service.
This study uses de-identified data from the National Inpatient Sample (NIS), part of the Healthcare Cost and Utilization Project (HCUP), Agency for Healthcare Research and Quality (AHRQ), United States Department of Health and Human Services. HCUP recommendations and protocols were followed throughout the study. No IRB was required/applicable.
Twitter @NimaHNejad, @alexjsolomonIR
Contributors All authors approved the final version of the article, including the authorship list. Conception and design: NH-N, CRB and CRW; analysis and interpretation of the data (NH-N, CRB, MK, and CRW); drafting of the article (NH-N, AJS and MAA); critical revision of the article for important intellectual content (NH-N, CRB, AJS, MAA, JAC, MK and CRW); final approval of the article (NH-N, CRB, AJS, MAA, JAC, MK and CRW).
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Disclaimer Unrelated to this project: Dr Carrino has received consulting fee from Pfizer, Inc, Covera, IAG, Image Biopsy Lab, and Simplify Medical, and is a member of scientific advisory board/other office of IAG. Dr Khan is consultant for Stryker Medical and Medwaves Avecure Medical Corporation. Dr Weiss has received research grants from Siemens Healthcare, Merit Medical, Medtronic, and BTG and is a consultant for BTG and Medtronic.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.