Two-year cost comparison of vertebroplasty and kyphoplasty for the treatment of vertebral compression fractures: are initial surgical costs misleading?
- First Online:
- Cite this article as:
- Ong, K.L., Lau, E., Kemner, J.E. et al. Osteoporos Int (2013) 24: 1437. doi:10.1007/s00198-012-2100-0
- 507 Downloads
The costs for treating kypho- and vertebroplasty patients were evaluated at up to 2 years postsurgery. There were no significant differences in adjusted costs in the first 9 months postsurgery, but kyphoplasty patients were associated with significantly lower adjusted treatment costs by 6.8–7.9 % in the remaining periods through 2 years postsurgery.
Vertebral augmentation has been shown to be safe and effective for treating vertebral compression fractures. Comparative cost studies of initial treatment costs for kypho- and vertebroplasty have been mixed. The purpose of our study was to compare the costs for treating kypho- and vertebroplasty patients at up to 2 years postsurgery.
Vertebroplasty and kyphoplasty patients diagnosed with pathologic or closed lumbar/thoracic vertebral fractures were identified from the 5 % sample of the Medicare dataset (2006–2009). The final study cohort with at least 2 years follow-up comprised of 1,609 vertebroplasty and 2,878 kyphoplasty patients. The cumulative treatment costs (adjusted to June 2011 US$) were determined from the payer perspective. Differences in costs and length of stay were assessed by generalized linear mixed model regression, adjusting for covariates.
The average adjusted costs for vertebroplasty patients within the first quarter and the first 2 years postsurgery were $14,585 [95 % confidence interval (CI), $14,109–15,078] and $44,496 (95 % CI, $42,763–46,299), respectively. The corresponding average adjusted costs for kyphoplasty patients were $15,117 (95 % CI, $14,752–15,491) and $41,339 (95 % CI, $40,154–42,560). There were no significant differences in adjusted costs in the first 9 months postsurgery, but kyphoplasty patients were associated with significantly lower adjusted treatment costs by 6.8–7.9 % in the remaining periods through 2 years postsurgery.
Our present study addresses some of the limitations in previous comparative cost studies of vertebroplasty and kyphoplasty. The higher adjusted costs for vertebroplasty patients than kyphoplasty patients by 1 year following the surgery reflect greater utilization of medical resources.
KeywordsCost Kyphoplasty Medicare Spine fracture Vertebroplasty
Vertebral compression fractures (VCFs) represent a significant burden to patients, the health care system, and society . VCFs are a leading cause of disability and morbidity, adversely affecting the quality of life, daily function, and survival of the patients [2, 3]. VCFs can be treated by nonsurgical conservative care or by vertebral augmentation that includes vertebroplasty and balloon kyphoplasty. Both surgical procedures have been in clinical use for over a decade and involve stabilizing a vertebral fracture by injection of bone cement. The distinction between the approaches is that balloon kyphoplasty requires the use of a bone tamp to create an initial cavity for the cement.
Vertebral augmentation has been shown to be safe and effective [4, 5, 6]. In 2007, the American Society of Interventional and Therapeutic Neuroradiology, Society of Interventional Radiology, American Association of Neurological Surgeons/Congress of Neurological Surgeons, and American Society of Spine Radiology issued a position statement that vertebroplasty and kyphoplasty were established, safe, and effective therapies for treating painful VCFs. However, there has been recent controversy over the surgical management of VCFs, due largely to mixed evidence reported in the literature for vertebroplasty. Two randomized trials [7, 8] raised questions about the efficacy of vertebroplasty, finding it to be no more effective than sham procedures in terms of improvement in disability and pain scores at up to 6 months posttreatment. Other studies have observed improved outcomes in their patient cohorts. For example, in a randomized trial of 202 patients, significantly more pain relief was reported for vertebroplasty patients than for those with nonsurgical treatment up to 1 year . The effectiveness of kyphoplasty was also compared to nonsurgical therapy in a randomized controlled trial of 300 patients by Boonen et al.  and Wardlaw et al.  Based on the 2 year results , the improvement in function, quality of life, and pain relief for kyphoplasty patients over nonsurgically managed patients were statistically significant when averaged across 24 months with the reduction in back pain remained statistically significant at all time-points.
There has also been mixed findings in the literature regarding the cost effectiveness of vertebral augmentation. A subset of a randomized controlled trial of 63 patients in Sweden  did not find that kyphoplasty was cost-effective compared to standard medical treatment. In contrast, kyphoplasty was found to be cost-effective for treating patients with hospitalized VCFs in the UK setting . Another study  reported a cost-effectiveness ratio for vertebroplasty that was also within the UK threshold for the willingness to pay for a quality adjusted life year. In addition, there are few direct comparisons of the costs of kypho- and vertebroplasty. The in- and outpatient costs have been found to be greater for kyphoplasty than vertebroplasty , based on an administrative claims dataset. On the other hand, others have reported, using national claims data, that the hospital charges for both procedures were comparable . These studies are limited by their focus on initial treatment costs but do not take into account potential differences in the cost of follow-up care or treating subsequent complications, or costs outside of the inpatient or outpatient facilities, such as skilled nursing or rehabilitation facilities. Because of potential for pain relief, kypho- and vertebroplasty have also been performed on patients with cancer-related VCFs [16, 17, 18]. The presence of cancer may be associated with significantly greater morbidity and utilization of healthcare resources. However, it is unclear what impact this may have on the relative treatment costs for cancer patients undergoing different spine fracture treatments.
Therefore, the objectives of our study were to (1) evaluate the cost to the payer for treating kypho- and vertebroplasty patients, including the cost of follow-up care and (2) compare the costs between kypho- and vertebroplasty patients, accounting for case mix such as pre-existing diagnosis of cancer. We hypothesized that the comparative costs for treating kypho- and vertebroplasty patients would be similar when longer term cost of care is considered. To address this hypothesis, we utilized longitudinal Medicare claims data from the Center for Medicare and Medicaid Services (CMS). This study considered cumulative costs to Medicare from all sources for up to 2 years following vertebral augmentation.
The 5 % Medicare claims data from January 1, 2005 and December 31, 2009 was used for this study. This file is a random sample compiled from claims from beneficiaries whose health insurance claims number contains the requisite digits in the sampling plan developed by the Centers for Medicare and Medicaid Services. Medicare is the largest nationwide insurance program that provides hospital insurance (Part A) or medical insurance (Part B) to the elderly. The beneficiaries include people age 65 or older, under age 65 with certain disabilities, and those with End-Stage Renal Disease or Amyotrophic Lateral Sclerosis (also known as Lou Gehrig’s disease). The 5 % systematic sample reflects the claims generated by approximately two million Medicare enrollees. The Medicare dataset is advantageous in that it is population-based, has a large sample size, and allows linkage of beneficiaries across years for longitudinal studies. This database has been used previously to compare the mortality rates for VCF patients and a matched control population .
Vertebro- and kyphoplasty patients who were diagnosed with a pathologic vertebral fracture or closed fracture of the lumbar or thoracic vertebra without spinal cord injury were identified from the 2006–2009 Medicare dataset. The 2005 data were used as a screening year for comorbidities for 2006 patients (described later). The VCF diagnoses were based on International Classification of Diseases, 9th revision, Clinical Modification (ICD-9-CM) diagnosis codes of 733.13, 805.2, or 805.4. Vertebroplasty patients were identified using ICD-9-CM code 81.65 or Current Procedural Terminology, 4th edition (CPT-4) codes 22520, 22521, and 22522. Kyphoplasty patients were identified using ICD-9-CM code 81.66 or CPT-4 codes 22523, 22524, and 22525.
Patients younger than 65, those enrolled in health maintenance organizations, those with <12 months of claim history prior to vertebral augmentation, and those with both vertebro- and kyphoplasty on the same claim were excluded from the analysis. Patients younger than 65 years who enroll in Medicare typically qualify for benefits as a result of their unique health condition in terms of disability or end-stage renal disease and therefore were excluded due to the possible confounding effects. Patients who received their Medicare health benefits through a health maintenance organization were also excluded because their healthcare expenses were not submitted to the CMS for payment and, therefore, claims from these beneficiaries were not available from the database. We excluded patients with <1 year of claim history prior to the vertebro- or kyphoplasty because the 1-year look-back period was used to establish the preoperative baseline general health status for each patient. Patients with kypho- and vertebroplasty on the same claim were excluded because they could not be uniquely stratified to one procedure group and may encounter confounding effects. Of the 5,489 vertebroplasty and 9,660 kyphoplasty patients with vertebral fracture identified during 2006–2009, 954 vertebroplasty (17.4 %) and 1,200 kyphoplasty patients (12.4 %) were excluded using the aforementioned criteria, leaving a study population of 4,535 vertebroplasty and 8,460 kyphoplasty patients. An additional 2,926 vertebroplasty and 5,582 kyphoplasty patients were not considered in this study because they did not have a full 2-year follow-up, which would have prohibited the collection of the cost of care for these patients over the same duration of follow-up. The final study cohort comprised of 1,609 vertebroplasty and 2,878 kyphoplasty patients.
The treatment costs were determined for each surgical group from the payer perspective, i.e., in terms of Medicare payment. The cumulative Medicare payment for each patient was identified for up to 2 years following the vertebro- or kyphoplasty procedure. The cumulative costs were derived from all sources, including those from the inpatient, outpatient, physician/carrier, skilled nursing facility, home health agency, hospice, and durable medical equipment claims files. Pharmaceutical benefit costs were not available. All costs were adjusted to 2011 (June) US dollars . Resource utilization, in terms of length of stay, was also computed and compared for the vertebro- and kyphoplasty patients who had inpatient procedures.
The differences in costs and length of stay were assessed by generalized linear mixed model regression. The regression models were used to evaluate the joint effect of multiple covariates, which included gender, age, race/ethnicity, patient health status (Charlson score), socioeconomic status (Medicare buy-in status), year of surgery, census region (Northeast, Midwest, South, and West), and type of treatment (kypho- and vertebroplasty). The Medicare buy-in status was an identifier of patients whose Medicare premiums and deductibles were subsidized by the state due to their financial status and was used as a proxy for the patient’s socioeconomic status. Baseline general health status and specific comorbidities for each patient were evaluated based on disease diagnoses from inpatient and outpatient claims during the 12 months prior to the index VCF. To account for the general health status of each patient prior to their VCF, the composite Charlson comorbidity index (Dartmouth–Manitoba version ) was calculated. In addition to the covariates described previously, the regression model for treatment costs also included site of service (inpatient or outpatient) and presence of any cancer diagnosis as covariates. Cancer patients were those diagnosed with ICD-9-CM codes 140-176, 179-208, 210-239, or V10 in the 12 months prior to the surgery.
A lognormal distribution with an identity link function was used for the length of stay regression model. The length of stay analysis only applied to patients treated in a hospital; thus, there was no “0” value for length of stay. Examination of the empirical distribution of the length of stay showed a characteristic right skewed set of values. Thus, even though length of stay was discrete (e.g., 3, 4, and 5 days), the lognormal distribution was a better fit than the Poisson or the normal distributions. The generalized chi-square per degree of freedom for the length of stay models of patients with and without cancer diagnosis were 0.803 and 0.739, respectively. For the cost of the surgery, a generalized mixed model was used with a Gamma distribution and a log-link function. The Gamma distribution is a fairly flexible distribution that can accommodate continuous data with characteristic skewed behavior typical of cost data. The Gamma distribution has been studied or utilized in a number of medical cost studies and found to provide better fit to such data than symmetric (e.g., normal) or discrete (e.g., Poisson or negative binomial) distributions . The goodness of fit for the model of cumulative costs at 2 years was 0.79.
Kyphoplasty (n = 2,878) (%)
Vertebroplasty (n = 1,609) (%)
% with Medicare buy-in
Charlson index 0
% with cancer diagnosis
Length of stay (number of days)
With cancer diagnosis
Without cancer diagnosis
Average (standard deviation)
Average (standard deviation)
Statistical analysis of length of stay (with and without cancer diagnosis)
With cancer diagnosis
Without cancer diagnosis
Factor p value
95 % CI
Level p value
Factor p value
95 % CI
Level p value
BKP v VP
F v M
65-69 v 85+
70-74 v 85+
75-79 v 85+
80-84 v 85+
Black v White
Other v White
1-2 v 0
3-4 v 0
5+ v 0
No v Yes
Midwest v West
Northeast v West
South v West
Statistical analysis of cumulative costs at 2 years
Factor p value
95 % CI
Level p value
F vs. M
65–69 vs. 85+
70–74 vs. 85+
75–79 vs. 85+
80–84 vs. 85+
Black vs. white
Other vs. white
1–2 vs. 0
3–4 vs. 0
5+ vs. 0
No vs. yes
Yes vs. no
Midwest vs. West
Northeast vs. West
South vs. West
Site of service
Inpatient vs. outpatient
Average adjusted cumulative costs
Period (in years)
Average adjusted cost (95 % confidence interval)
Adjusted relative ratio (kyphoplasty vs. vertebroplasty)
Adjusted p value
With prior cancer diagnosis
Without prior cancer diagnosis
Average adjusted cumulative costs by service component
Period (in years)
For patients with prior cancer diagnosis, vertebroplasty patients had an unadjusted average treatment costs of $14,039 (±$503 standard error), compared with $18,804 (±$365) for kyphoplasty patients within the first quarter following surgery. By the end of 2 years following surgery, the difference decreased from $4,765 to $1,431, based on the unadjusted average costs of $45,612 (±$1,498) for vertebroplasty patients and $47,043 (±$1,038) for kyphoplasty patients. The corresponding median costs for vertebro- and kyphoplasty patients were $32,357 and $33,891, respectively, by 2 years. After adjusting for covariates, there were no significant differences (non-Bonferroni corrected) in adjusted costs in the first 9 months postsurgery (Table 5). However, in the remaining periods through 2 years postsurgery, kyphoplasty patients were associated with significantly lower adjusted treatment costs by 6.9–8.9 %. For example, by the end of 2 years, kyphoplasty had 8.1 % lower costs (p = 0.022). However, after accounting for multiple comparisons using Bonferroni correction for the level of significance, no significant differences between kypho- and vertebroplasty cohorts were found.
For patients with no prior cancer diagnosis, the difference in unadjusted costs for vertebro- and kyphoplasty patients decreased from $3,294 within the first quarter to $2,295 by 2 years following surgery. After adjusting for covariates, there were no significant differences (non-Bonferroni corrected) in adjusted costs in the first 6 months postsurgery (Table 5). However, in the remaining periods through 1.25 years postsurgery, kyphoplasty patients were associated with significantly lower adjusted treatment costs by 7.4–8.7 %. In the subsequent periods through 2 years postsurgery, no significant differences were found between both groups. After accounting for multiple comparisons using Bonferroni correction for the level of significance, only cumulative costs at the 1-year time point were found to be significantly lower for the kyphoplasty cohort.
The results of this study did not support the hypothesis that the cumulative 2-year posttreatment costs for kypho- and vertebroplasty patients were similar. Instead, for the overall cohort (i.e., those with and without prior cancer diagnosis), we observed significantly lower adjusted cumulative treatment costs in the kyphoplasty patient population at 1 year through 2 years postsurgery. In addition, hospitalized kyphoplasty patients had significantly shorter length of stay than vertebroplasty patients.
Previous studies have shown that the treatment costs for vertebral augmentation patients are higher than those undergoing nonsurgical management at the point of care, due to the cost of the surgical procedure, instruments, and anesthesia . After taking longer-term costs and outcomes, such as quality adjusted life years, into consideration, the cost effectiveness of vertebral augmentation has demonstrated mixed conclusions. A randomized controlled trial in Sweden  did not find that kyphoplasty was cost-effective compared to standard medical treatment. In contrast, a cost-effectiveness study by Strom et al.  estimated that kyphoplasty was cost-effective with a cost per QALY gained that fell within the threshold range in the UK for the willingness to pay for a QALY. Another study  reported a cost-effectiveness ratio for vertebroplasty that was also within the UK threshold. Based on a retrospective study of vertebroplasty and conservative care patients in an Italian setting, Masala et al. observed that vertebroplasty was cost-effective with regards to visual analogue scale, ambulation, and activities of daily living scales at 1 week, but no significant differences were found at 12 months .
While the comparisons of costs between vertebral augmentation and nonsurgically managed patients are limited, even fewer studies have been published comparing costs following kypho- and vertebroplasty treatments. Using the Nationwide Inpatient Sample, a national database of hospital inpatients stays in the USA, Lad et al.  reported comparable mean hospital charges of $30,144 for kyphoplasty patients and $29,517 for vertebroplasty patients. In contrast, using cost data derived from hospital accounting systems or from converting charges to cost, Mehio et al.  reported greater adjusted inpatient costs and outpatient costs for kyphoplasty than vertebroplasty patients. However, these previous cost findings do not provide an accurate representation of the cost of care because they are only limited to those from the initial surgery. Longer-term cost of care, such as those arising from the differential rate of treating subsequent complications or nonroutine discharges, should be considered when comparing the economic impact of particular treatments. It has been suggested that routine discharge to home can be assumed to use fewer medical resources than admission to another facility, such as a skilled nursing facility or a rehabilitation facility, and decrease overall cost of care . Furthermore, these studies are also limited by the failure to account for costs in terms of the amounts that are reimbursed by the payers.
The higher adjusted costs for vertebroplasty patients than kyphoplasty patients after 1 year following the surgery reflect greater utilization of medical resources, likely to treat subsequent complications or seek additional medical attention. In a systematic review by Felder-Puig et al. , the authors found that both kypho- and vertebroplasty significantly reduce pain in the majority of patients and can lead to short-term and possibly long-term improvement of function. However, kyphoplasty was linked to fewer clinically relevant complications than vertebroplasty. In another review study, Lee et al.  conducted a meta-analysis of complication following vertebral augmentation. The investigators concluded that although the incidence of adverse events for both vertebroplasty and kyphoplasty were low, it appeared that vertebroplasty was associated with a significant increased rate of procedure-related complications and cement extravasation. Both vertebro- and kyphoplasty were also found to be effective in improving pain, functional disability, and quality of life, based on a prospective study of 28 vertebroplasty and 24 kyphoplasty patients . However, kyphoplasty was observed to provide greater degree of improvement, which was maintained over long-term follow-up. The kyphoplasty group experienced 15 % greater pain relief at 1 week postsurgery compared with the vertebroplasty group and continued to have greater pain relief than their vertebroplasty counterparts at the last follow-up (mean follow-up of 42 weeks). The researchers hypothesized that these differences may have arisen from the relative reduction in the kyphotic angle between both groups. Conversely, some have reported little difference in outcomes between kypho- and vertebroplasty patients [26, 27], while others have indicated that kyphoplasty may be associated with higher rates of adjacent level fractures .
Our present study addresses some of the previous limitations in other comparative cost studies of vertebro- and kyphoplasty [14, 15], which include the costs for follow-up care and adopting a payer’s perspective. However, our study has several limitations. Due to the administrative claims nature of the Medicare data, we were unable to evaluate certain clinical details and account for potential differences in the patient characteristics in both treatment groups. For example, we were not able to consider other confounding factors , such as obesity and smoking, even though the analysis accounted for potential differences in patient characteristics and pre-existing comorbidities in terms of the Charlson score. Because drug intervention data were not available in the dataset, we were unable to investigate the effects of drug intervention on the costs. The present study focused on costs following vertebral augmentation surgery, but did not explore other procedure related outcomes, such as complications. However, to the extent that the treatment arms of the present study experienced differential clinical outcomes following their surgical intervention, these outcomes are implicit in the cost analysis. The present study was based on the Medicare population aged 65 years and older. It is unclear if differences in costs between treatment modalities for the younger VCF patient population would still exist. It is also unclear if the findings from the present study will change with longer follow-up.
The controversy surrounding vertebral augmentation has led to some advocating that the surgical treatment options for VCFs be curtailed as a mechanism to fund comparative effectiveness research , with the intent that such research would help reduce costs and improve health care by objectively comparing the costs, outcomes, and cost-effectiveness of clinical interventions. The evidence from this study suggests that the treatment costs for kyphoplasty patients are significantly lower than those for vertebroplasty after 1 year through 2 years postsurgery. These results are expected to help provide additional data for the debate regarding the role of augmentation and, in particular, any differences between kypho- and vertebroplasty, for treating VCF patients.
This study was funded by a research grant from Medtronic Spinal and Biologics.
Conflicts of interest
Jason Kemner is an employee and stockholder of Medtronic, Inc.; Edmund Lau, Kevin Ong, and Steven Kurtz received institution research grant from Medtronic Inc for this study.