Resources utilisation and economic burden of percutaneous vertebroplasty or percutaneous kyphoplasty for treatment of osteoporotic vertebral compression fractures in China: a retrospective claim database study

Background Osteoporotic vertebral compression fractures (OVCF) is a common and often debilitating complication of osteoporosis, leading to significant morbidity and increased mortality. Percutaneous vertebroplasty (PVP) and Percutaneous kyphoplasty (PKP) are recommendable surgical treatments for OVCF. Objective To evaluate PVP/PKP utilisation and their related direct medical costs for OVCF treatment in China from the payer perspective. Methods A population-based medical claims database of a metropolitan city in China was analysed from the payer perspective, which included all inpatient claims from 01/01/2015 to 31/12/2017. All vertebral fractures patients that met the eligibility criteria (aged ≥50 years old, having vertebral fracture diagnosis, without unrelated diseases diagnoses such as tumour and scoliosis, received PVP/PKP) were deemed as OVCF patients. Baseline characteristics, surgery rate, length of stay in hospital, time to re-surgery, and costs (including costs per hospitalisation and annual costs) were described. Survival analysis function was used to estimate the re-surgery rate. Results Of the 50,686 patients with OVCF identified, 14,527 (28.66%) received a total number of 15,599 records of PVP/PKP surgeries from 2015 to 2017. Mean age was 75 at the first surgery captured in the database analysis period; females accounted for 79.54% of all cases. The median length of surgery stay was 9 days. Cumulative re-surgery rates were 1.22% in 30 days, 2.58% in 90 days, 3.61% in 183 days, 5.42% in 1 year, and 7.95% in 2 years. There was no significant difference in re-surgery rate between PVP and PKP (p = 0.3897). The median time to the re-surgery was 139 days. Mean costs per PVP/PKP-related hospitalisation were 35,906 CNY/5122 USD (34,195 CNY/4878USD for PVP, 44,414 CNY/6336 USD for PKP, p < 0.01). The overall costs of hospitalisation averaged 186.61 million CNY (26.62 million USD) per year in this metropolitan city. Conclusion From 2015 to 2017, nearly one-third of OVCF inpatients received PVP/PKP and the re-surgery rate was 7.95%. PVP/PKP procedures for OVCF place a high economic burden for both the healthcare system and patients. Early detection and treatment of patients with osteoporosis are critical in China.


Background
Osteoporosis places significant disease burden on patients, ranking the seventh among common chronic diseases with over 200 million people affected worldwide [1,2]. In China, the prevalence of osteoporosis has been rising over the past few decades in parallel with the aging population. The prevalence of osteoporosis in mainland China was approximately 13% for adults aged 20 and over during the period of 1980-2008 and rose to about 28% for people aged 15 and over between 2012 and 2015 [3,4] and the disease disproportionally affects females [4][5][6]. For most patients, osteoporosis is a silent and insidious disorder and often not diagnosed until fractures have occurred.
Osteoporotic vertebral compression fractures (OVCF) is a common and often debilitating complication of osteoporosis, leading to chronic back pain, insomnia, reduced activity, depression, and increased mortality [1]. Thirty to fifty percent of people over 50 years old could be affected by OVCF worldwide; in Europe, the incidence of OVCF was 570 per 100,000 males and 1070 per 100,000 females; in South Korea, the 5-year incidence of OVCF was 852 per 100,000 persons [7]. In Shanghai, a metropolitan city in China, 14.4% of people over 65 years old and 20.1% of people over 80 years old were suffering from OVCF and the risk of OVCF was significantly higher for females compared to their male counterparts (18.5% vs. 12.4%) [8]. Many countries around the world have studied the economic burden of OVCF. A report shows that the total annual costs of OVCF in the United  [9][10][11][12].
The OVCF treatment guidelines include both conservative and surgical interventions [1]. The conservative therapies entail initial bed rest, use of analgesics, external immobilization; while surgical treatments include percutaneous vertebroplasty (PVP) and percutaneous kyphoplasty (PKP). According to China's guidelines, for surgical treatments, PVP is recommended for patients with OVCF who are refractory to treatment with braces and medication, but not suitable for patients with extremely severe vertebral compression fractures that cannot establish working channels and merge with the lesions of the same site requiring surgical treatment, pedicle fractures, severe compression fractures; PKP is recommended for pain or kyphosis caused by osteoporotic compression fracture, but not for stable, cured, painless osteoporotic compression fracture, osteoporotic burst fracture [1].
In addition to the surgical treatment, anti-osteoporosis treatment is an integral part of the clinical management in order to fundamentally improve bone mass and strength, and reduce the risk of re-fracture [1]. In China, the recommended anti-osteoporosis therapy includes calcium and vitamin D supplements, anti-resorptive drugs, bone anabolic drugs, and traditional Chinese medicine [13]. However, in China, a survey showed that about half of osteoporotic patients are not diagnosed even if they had fractures; moreover, less than a quarter of the patients are receiving effective anti-osteoporosis drugs before fractures [14].
Previously, a number of studies have explored the clinical efficacy and costs of PVP/PKP in China based on hospital data. Evidence showed that PVP/PKP can effectively relieve fracture-related pain, and PVP/PKP is considered as the best choice for OVCF [1,[15][16][17] [18][19][20][21].
However, no studies so far have examined PVP/PKP utilisation using a large-scale population-based database in China. Therefore, the primary objective of this study is to estimate PVP/PKP utilisation, including surgery rate and re-surgery rates, and related direct costs in China through analysing a citywide inpatient claims database.

Methods
This study is a retrospective administrative claims database study to estimate the surgery/re-surgery rates and direct medical costs in patients with OVCF and receiving PVP/PKP surgery from the health payer perspective.

Study setting and data source
Insurance claim data is a commonly used large-scale database for healthcare study. In China, over 95% of the residents have been covered by basic medical insurance. All data in this study were de-identified and extracted from an anonymous metropolitan inpatient claims database from 01/01/2015 to 31/12/2017. All OVCF patients receiving PVP/PKP surgery during such a period would be followed up to their second PVP/PKP surgery or the end of 2017 whichever comes first. The date of first surgery (PVP/PKP) during study time frame would constitute the index date. Detailed information on patients included demographics, diagnosis, treatment, costs, and so on. Patients meeting all eligibility criteria detailed below were included in this study. Because only deidentified records were used in the analysis, ethical approval and informed consent were not required, which was consistent with the local medical research policy on using electronic health data.

Patient selection
Records with admission date from 01/01/2015 to 31/12/ 2017 were extracted. Inclusion criteria included aged ≥50 years old, recorded OVCF or vertebral fracture as admission diagnosis, and received PVP and/or PKP surgery. Patients diagnosed with unrelated diseases were excluded (tumour, spinal deformity, scoliosis, spondylitis, and disc herniation). Patients receiving PVP/PKP in cancer hospitals were also excluded. In the main analyses, a key assumption was that a vertebral fracture with PVP/ PKP without unrelated diseases was in fact osteoporotic, which was aligned with osteoporosis diagnosis criteria according to Chinese guideline [13]. Keywords used for patient selection was displayed in Appendix A.

Surgery and re-surgery status
All PVP/PKP surgery records were classified into three categories: PVP, PKP, and unclear. Due to lack of standardized nomination for surgery procedure, surgeries of PVP or PKP that could not be judged by information in the extracted database were marked as "unclear". Surgery status was examined in terms of surgery rate and Length of Stay (LOS) in hospitals. Surgery rate was calculated by dividing the patient number receiving PVP/ PKP by the total number of OVCF patients. Re-surgery status was explored in terms of the cumulative resurgery rate for 3 months, 6 months, 1 year and 2 years follow-up, and median length of the time interval to resurgery. The first surgical record captured in the database during the study period was assumed as the first surgical record of patient and ensuing PVP or PKP procedures were identified as re-surgery.

Direct medical costs
Direct medical costs of PVP/PKP among OVCF patients were summarized in terms of both cost per hospitalisation (per visit) and annual costs (all PVP/ PKP related costs in 1 year). Costs per hospitalisation were the average of total PVP/PKP surgery-related hospitalisation costs, and could be broken down to different categories, for example, medical fee, diagnosis fee, treatment fee, and so on. The annual costs were the average of the sum of all PVP/PKP surgeryrelated hospitalisation costs.

Statistical methods
Descriptive analysis was employed to characterize patient profiles at their first inpatient care. Continuous variables were presented as mean, standard deviation, median, inter-quartile range whereas categorical variables were displayed in frequencies and percentages. A re-surgery function was developed to illustrate the probability of re-surgery with the time interval between the first and second surgery. Patients without re-surgery were censored at the end of 2017. Kaplan-Meier method was used to establish the re-surgery function. Based on the function, cumulative re-surgery rates and their 95% confidence intervals were estimated at different time points, specifically, 3 months, 6 months, 1 year, and 2 years. The calculation of the time interval to re-surgery is shown in the following formula: Non-parametric statistical tests were used for the costs, days (Mann-Whitney Test for two groups, and Kruskal-Wallis Test for three or more groups) and resurgery rate (Log-rank Test and Wilcoxon Test) comparison, α = 0.05 was used as the significant level for all comparisons.
All analyses were conducted using Stata SE 14.

Subgroup analyses
Two subgroup analyses were performed on age and sex groups because they have a significant effect on the incidence of osteoporosis and vertebral fractures. Sex was divided into two categories: male and female; age was divided into five categories: 50-59, 60-69, 70-79, 80-89, 90+. LOS, re-surgery rates and costs were counted and compared between subgroups.

Sensitivity analysis
A key assumption of the main analysis was that all patients who met the eligibility criteria were identified as OVCF patient even if they did not have an OVCF diagnosis. To test the robustness of the results, a sensitivity analysis was performed focusing on patients who had established OVCF diagnosis. All the statistical analyses in the main analysis were implemented in the sensitivity analysis, except for the annual costs.

Study population
Extracted raw data included 8

Surgery and re-surgery status
Of all OVCF patients, the total surgery rate was 28.66%, of which PVP was 17.30% and PKP was 6.03% (Table 2). Median LOS in the hospital was 9 days, of which PVP was 9 days and PKP was 10 days (Table 3). Among all the patients during the study period, the re-surgery rate was 1.22% in 30 days, 2.58% in 90 days, 3.61% in 183 days, 5.42% in 1 year, and 7.95% in 2 years (Table 4; Figs. 2 and 3). Median time interval to resurgery was 139 days (Table 5). There was no significant difference in the re-surgery rate between PVP and PKP, regardless of the type of statistical tests (Log-rank: p = 0.3897; Wilcoxon: p = 0.1829).

Direct medical costs
The annual total costs attributable to PVP/PKP related hospitalisation from 2015 to 2017 were 153 million CNY   Table 8).

Subgroup analyses
Median LOS was 9 days for both male and female. However, the statistical test showed that the difference in LOS distribution between male and female was significant (p < 0.01). Median LOS was 8,8,9,10,10 for each age groups, respectively. The difference among age groups was also statistically significant (p < 0.01). (Supplementary Table 1).
The 2-year cumulative re-surgery rates were 6.92% for male and 8.20% for female, respectively. Patients in the 50-59 age group had the lowest 2-year cumulative re-surgery rate (4.48%), while patients in 80-89 age groups had the highest (10.22%). The statistical difference was not significant (p = 0.192) between sex but significant among age groups (p < 0.01). (Supplementary Table 2).
The average hospital costs of the male were significantly higher than that of female [37,950CNY (5414USD) vs 35,383CNY (5048USD), p < 0.01]. There were also significant differences in the average costs of hospitalisation for different age groups (p < 0.01), with patients in 50-59 age group having the highest costs  Table 3 and 4).
Median LOS in hospitals was 9 days, of which PVP was 8 days and PKP was 10 days (Supplementary Table 6). The re-surgery rate was 0.93% in 30 days, 2.10% in 90 days, 2.75% in 183 days, 3.63% in 1 year, and 5.51% in 2 years Supplementary Table 7). There was no difference in re-surgery rate between the PVP and PKP (Supplementary Table 8). The median time interval to re-surgery was 97 days (Supplementary Table 9). Total costs per hospitalisation averaged 34,561 CNY (4930 USD) (Supplementary Table 10).
In addition, outliers of costs and time interval were jointly checked. In 14,527 patients, 28 patients had time  interval to re-surgery less than 7 days; only 2 of them had total costs less than 13,842CNY/1975USD (the 5th percentile of total costs distribution); 13 of them had total costs less than 33,416CNY/4767USD (the median of total costs). Re-surgery and first surgery of these patients can be considered as one treatment. However, due to the very small patient number, it does not affect the overall results.

Discussion
This study was a retrospective longitudinal study aiming at measuring the burden of PVP/PKP due to OVCF in  China from the payer perspective. To our knowledge, our study was the first to estimate the utilisation of PVP/PKP procedures and their costs for OVCF treatment in China by analysing real-world city-wide claims data. We noted the utilisation of PVP/PKP was high for OVCF patients with 28.66% PVP/PKP surgery rate and 7.95% 2-year cumulative re-surgery rate. Median LOS in hospitals was 9 days, and median time interval to resurgery was 139 days. Hospitalisation costs due to PVP/ PKP were high: per hospitalisation cost averaged 35,906 CNY (5122 USD), and annual hospitalisation costs in the metropolitan city totalled 187 million CNY (27 million USD). It brought a significant burden to both medical insurance institution and patients. Subgroup analyses and sensitivity analysis indicated that men had shorter hospital stays but higher hospital costs than women. In addition, there was no difference in reoperation rates between men and women. With the increase of age, both LOS and the re-surgery rates increased (the sudden drop in the resurgery rate in 90+ age group might be due to the impact of sample size, mortality, etc.). However, the costs of hospitalisation for the 50-59 age group were the highest, which emphasized the value of early intervention for osteoporosis.
Overall, the results of this study were comparable with those of previous studies. In this study, the number of female OVCF patients was about 2.3 times that of male patients, which is consistent with the higher risk of OVCF in females in other studies [1,4,8,22]. The surgery rate of PVP/PKP in OVCF patients in our study was also similar to the finding of another study at 23.1% [23]. A comparison of re-surgery rate among studies was not conducted due to the lack of studies on PVP/PKP re-surgery rates.
Whether it is PVP or PKP, surgery-related inpatient costs are high. In contrast, taking a short-term view, PVP seems to be better than PKP because there was no significant difference in the re-surgery rate between PVP and PKP, but the costs of PVP were nearly 30% lower than that of PKP. Another clinical study in China also recommended PVP because the clinical result (pain relief) differed little but the costs of PKP were higher [24]. However, due to the lack of confounder control in this study, we should be more cautious about asserting a causal relationship between surgery types and the outcomes. The costs associated with vertebral fractures in many countries are well documented: in America, the costs per surgery-related hospitalisation were 7805 USD (54, [28]. One reason is that patients with PKP surgery use fewer drugs and pay lower maintenance costs after surgery. PKP significantly reduced 6.8-7.9% treatment costs during the 2-year post-surgery periods in America and reduced 33% painkiller costs during the 4-year postsurgery periods in German [27,28]. The surgical sequelae and its burden would be an interesting point for future research.
Aside from surgery intervention, conservative treatment for OVCF is widely employed in China. Conservative treatment is recommended for mild/chronic OVCF, while severe/acute OVCF is treated by surgery [29]. Anticoagulant therapy of low molecular weight heparin calcium injection can be given during bed rest; nonsteroidal anti-inflammatory analgesics are mainly used to relieve acute pain; anti-osteoporosis medication, such as alendronate sodium, and complex calcium carbonate vitamin D tablets can also be employed [29,30]. In terms of treatment effect, for pain relief, the short-term effect of surgery is better than that of conservative treatment, but no difference was observed in terms of longterm effect [31][32][33][34][35]. Recently, a study indicated that conservative treatment with anabolic drug could achieve comparative outcomes than PVP/PKP in treating acute OVCF [36]. However, from the perspective of recovery  of vertebral stability and vertebral height, surgery is superior to conservative treatment [34]. Although successful PVP/PKP treatment for OVCF can effectively alleviate pains and other symptoms, the procedures are not free of untoward effects. Studies have explored that the PVP/PKP may accelerate local bone absorption due to bone cement, thereby increasing the risk of recurrent fracture of the surgical vertebra [37][38][39][40]. In addition, studies also reported that additional stress of adjacent vertebrae caused by the cement augmentation and cement leakage are important factors in causing new adjacent vertebral fractures after PVP/PKP [41][42][43][44].
To avoid OVCF, anti-osteoporosis therapy should be considered as primary prevention. Osteoporosis and resulting osteoporotic vertebral fractures typically develop silently with a long-time window from the initial decrease in bone density to the occurrence of OVCF. Use of anti-osteoporosis drugs can reduce the risk of fractures. For women with osteoporosis but without vertebral fractures, alendronate significantly reduced the first vertebral fracture by 44% [45]. Studies have shown that anti-osteoporosis treatment after fracture can reduce the risk of re-fracture by 40% within 3 years [46]. In this regard, anti-osteoporosis treatment should be considered as part of long-term treatment strategies to reduce the risk of fracture [47][48][49][50]. Although antiosteoporosis therapy plays a significant role in OVCF prevention, the current situation of drug use is not optimistic. This might be due to the low diagnosis rate of osteoporosis and vertebral fractures. What's more, there is a large gap in the diagnosis and prevention of osteoporosis and vertebral fractures among different level hospitals, especially in community hospitals where the diagnosis ability is poor. Although inadequate diagnosis and prevention of osteoporosis is consensus, related real-world studies are lacking. The insufficiency of osteoporosis diagnosis and prevention can only be indirectly understood from the situation of fracture patients. A study in mainland China showed only 13.9% of patients used anti-osteoporosis drugs before fractures [23]. Moreover, a retrospective study showed that in China, the rate of misdiagnosis of vertebral fractures was 54.27%; while in the patients with vertebral fractures  [51]. Compounding the problem is poor compliance with the medications as short-term medication intake has no evidentiary clinical benefits for fracture prevention [52][53][54]. In China, an analysis of medical insurance claims database from 2009 to 2010 showed that the adherence to bisphosphonate treatment was even worse with the mean Medication Possession Ratio (MPR) being 0.34, 0.22, and 0.15 at the 3rd, 6th, and 12th month over the follow-up period, respectively; moreover, only 2.1% patients were observed with high adherence (MPR > 0.8) during the 12-month follow-up [55]. The strength of our study stems from a welldefined study population which was population-based, i.e. residents of the metropolitan city covered by government health insurance were all included. The findings are likely to be more robust than those derived from a single hospital or hospitals from convenient sampling. Nevertheless, our study also has a number of limitations. On the one hand, some of the shortcomings of this study are due to the limitations of the database. Mortality and comorbidity are important in health research. However, our study was based on administrative claim database which was unable to support research about survival of patients in real world. Besides, claim database lacks detailed medical information in terms of comorbidity. In addition, researchers should be cautious about extrapolating findings to other regions or cities in China or to patients without health insurance because of variation in adopting surgical intervention for OVCF, health insurance coverage schemes and steep costs for self-paying patients. On the other hand, study design can lead to some estimation bias. First, we assumed all vertebral fracture cases meeting eligibility criteria were osteoporotic. This may include a few patients with non- osteoporotic vertebral fractures. In order to justify the assumption, we ran the sensitivity analysis among patients who had established OVCF diagnosis. The results were in line with those in the main analysis, which proves that our assumption is reliable. Second, there may be omissions in selecting PVP/PKP patients due to the lack of standardized procedure name, which would lead to the under-estimation of the surgery rate. Third, re-surgery rate calculation could also be underestimated due to the lack of knowledge of patients' history before 2015. In addition, because we only captured the records of hospitalisations during which surgery was performed, there are two circumstances of postoperative patient care: discharged at the same visit until full recovery; discharged shortly after surgery, home or transferred to other hospitals for postoperative care. In the latter case, since we were not able to make the linkage between surgery and rehabilitative stay, the LOS and costs of hospitalisation may be underestimated.

Conclusion
From 2015 to 2017, about a third of OVCF patients received PVP/PKP surgery and the 2-year cumulative resurgery rate reached 7.95%. PVP/PKP due to OVCF brought a high economic burden on China's healthcare system. Early detection and treatment of patients with osteoporosis to prevent OVCF are critical in China.