Osteoporosis-related nonvertebral fractures increase the risks for morbidity and mortality and add substantially to treatment costs. As an adjunct to controlled, clinical trials, observational studies based on information from large, integrated medical and pharmaceutical claims databases provide an opportunity to assess the "real-world" cost effectiveness of treatment options aimed at reducing osteoporotic fracture risk. Although they lack the stringent controls required to establish causation, these studies include a large treatment population that reflects current disease prevalence, prescribing practices, and treatment- cost patterns in the managed care setting. This article reviews data from key observational studies that used analyses of the large Protocare Sciences Integrated Medical and Pharmaceutical Claims Database, which includes claims and eligibility records for >3 million individuals covered annually under various public and private benefit plans, to explore the cost effectiveness of oral bisphosphonatesâ€”risedronate and alendronateâ€”in reducing the risk for nonvertebral fracture. In addition, the data from the Protocare database were used to examine the cost implications of differences in gastrointestinal tolerability in these agents.
(Am J Manag Care. 2006;12:S191-S198)
Osteoporosis is responsible for more than 1.5 million fractures annually, mostly (850 000) nonvertebral.1 Osteoporosis-related nonvertebral fractures spur disconcerting increases in the risks for disability and early mortality, especially in women. In 2001, 315 000 individuals in the United States >45 years of age were admitted to the hospital for hip fracture, the most common nonvertebral fracture type secondary to osteoporosis.2 Further, 20% of individuals who were ambulatory before a hip fracture required long-term care afterwards; and the mortality rate for individuals >50 years of age who experienced a hip fracture soared to almost 25% during the first year after a fracture. Not surprisingly, the direct costs related to hospital and nursing home care for osteoporotic hip fractures is substantial–$ 18 billion in 2002–and expected to grow in coming decades.
The results of long-term, placebo-controlled trials reveal potentially important differences among common oral treatments for osteoporosis in terms of nonvertebral fracture risk reduction (Figure 1).3-12
All oral agents–raloxifene and the bisphosphonates–yield at least some reduction in the relative risk for vertebral fracture. With risedronate treatment, for example, significant reductions are discerned within 6 months of treatment initiation.13 But, in the real-world, managed-care setting, what do these clinical trial results mean in terms of improved clinical and cost outcomes? To help answer that question, this article, using observational data from real-world, administrative claims databases, assessed the economic impact of different osteoporosis-treatment options, mainly bisphosphonates, in reducing fracture-related costs, as well as the costs associated with bisphosphonate-related gastrointestinal adverse events, providing a picture of the treatment costs relevant to a managed care setting.
Observational Database Studies
For a variety of reasons, no prospective head-to-head studies have been conducted to date comparing antiresorptive agents using fracture risk as the primary endpoint. Because the absolute risk of osteoporosis-related fracture is small in clinical trials, large study populations, perhaps in the tens of thousands, might be needed to provide adequate statistical power to discern significant differences in fracture rates between treatments. In addition, these trials would be time consuming and expensive for the sponsor. Moreover, about 80% of patients prescribed drugs for the management of osteoporosis in the real world would be ineligible for such clinical studies.14 The main reasons for ineligibility would likely include the presence of comorbidities, such as gastrointestinal disorders; the use of prior osteoporosis treatments; and current therapy with glucocorticoids, anticoagulants, or anticonvulsants. Yet, these excluded patients reflect a patient mix typically encountered in the clinical setting. As a consequence, the applicability of standard clinical trial results to the broader managed care population should be viewed with appropriate caution.
As an alternative, observational database research may represent the best remaining strategy for exploring the antifracture cost effectiveness of osteoporosis treatments. The benefits of observational database research include a large, relevant patient population, with copious longitudinal data that spans many types of health plans and physician specialties. Detection of treatment patterns that occur in the real-world managed care setting represents the chief potential benefit of this approach. Yet, at the same time, observational database research is fraught with important limitations–chiefly, absence of chart reviews to validate codes, inability to control extraneous variables or demonstrate causation, inability to assess the use of over-the-counter products, as well as the possible presence of selection bias.15 This article will review 2 studies conducted using a national insurance claims database assessing the incidence and economic impact of fractures and the gastrointestinal consequences of treatment.
Nonvertebral Fracture Risk Reduction: Risedronate, Alendronate, and Calcitonin Compared
In a 2004 study, Watts and colleagues investigated the effectiveness of risedronate and alendronate, relative to nasal calcitonin and each other, in reducing the risk of nonvertebral fractures.16 Used as a data source in this study, the Protocare Sciences Integrated Medical and Pharmaceutical Claims Database includes claims and eligibility records for > 3 million individuals covered annually under various public and private benefit plans.
The subset of patients captured in this study included men and women =45 years of age who required a new prescription for nasal calcitonin, alendronate (5 mg or 10 mg daily, and 35 mg or 70 mg weekly), or risedronate (5 mg daily or 30 mg weekly) between July 1, 2000, and December 31, 2001. A total of 7081 patients were included, with 774 prescribed nasal calcitonin, 1000 risedronate, and 5307 alendronate, a distribution reflecting prescribing patterns at the time of the study.
In the 6-month pretreatment period, individual fracture histories were examined; patients with =1 fractures of the hip, vertebrae, or wrist were considered to have a prior fragility fracture. In the post-treatment period an assessment of the incidence of nonvertebral fractures–clavicle, leg, wrist, hip, pelvis, and humerus–was performed after 6 and 12 months of treatment. An analysis of baseline characteristics, however, discerned several statistical differences among groups (Table 1).
Most notably, significantly fewer patients receiving calcitonin used hormone therapy when compared with those receiving risedronate or alendronate. Further, a significant increase in the mean number of concomitant medications and the mean number of physician visits emerged in the calcitonin group versus the bisphosphonate groups.
The 6-month results revealed, versus calcitonin, relative risk reductions for nonvertebral fracture of 26% and 69% for alendronate and risedronate, respectively (Figure 2). Similar findings were seen in the 12-month analysis with relative risk reductions of 25% and 75%, respectively, for alendronate and risedronate compared with calcitonin. Moreover, risedronate treatment yielded relative risk reduction of 54% and 59% after 6 months and 12 months, respectively, versus alendronate. Findings from the Watts study, an observational analysis of "real-world" data from an actual large claims database, suggest that risedronate treatment can be more effective than either calcitonin or alendronate in reducing the risk for nonvertebral fractures early in treatment.
These findings were consistent with those from controlled clinical trials, in which comparable reductions surfaced in the risk for osteoporotic nonvertebral fractures with risedronate after 3 years' therapy (59%) and with alendronate after up to 4 years' therapy (32%-40%).5,13
From a managed care perspective, however, a central question remains unanswered: What is the economic benefit of reducing nonvertebral fracture risk with risedronate versus calcitonin or alendronate? Brixner and colleagues addressed this issue in a study that estimated direct medical costs for nonvertebral fractures in the first year of therapy in patients treated with risedronate, alendronate, or calcitonin using cost data from the Protocare Integrated Medical and Pharmaceutical Claims Database.17
Current Procedural Terminology (CPT)
This 12-month economic evaluation (from the start of new therapy) included 5024 patients: 656 receiving nasal calcitonin, 652 risedronate, and 3716 alendronate. The Medicare fee schedule for fracture-related payments, based on selected diagnosis-related group (DRG) and codes, was used to standardize costs. When the 12-month nonvertebral fracture reduction data from the Watts study (Figure 2) were applied to nonvertebral fracture costs gleaned from the Medicare fee schedule, substantial differences emerged in the annual nonvertebral fracture-related cost per patient per year for calcitonin, alendronate, and risedronate treatment, with risedronate treatment yielding significantly reduced costs (Figure 3).
Table 2 provides a further breakdown of the data, showing individual nonvertebral fracture sites and the per-patient, per-year costs for these fractures in the groups treated with calcitonin, alendronate, or risedronate.
These results suggest that for patients at least 45 years of age initiating treatment for osteoporosis, the use of risedronate can lower medical costs related to nonvertebral fractures during the first year of therapy. It seems reasonable to speculate that this cost benefit may be greatest in higher risk patients.
GI Tolerability and Cost Implications.
Bisphosphonates have been linked to the development of upper gastrointestinal (GI) events, from nausea to ulceration, that can limit the usefulness of these products and increase treatment costs.18 Yet, in standard clinical studies, it is difficult to discern the real-world risk for bisphosphonate-related GI side effects, since high-risk individuals, such as those with active or previous GI disease or those taking antisecretory drugs, are typically excluded. For instance, 2 clinical studies compared the gastric ulcer-inducing potential of risedronate (5 mg) and alendronate (10 mg) in healthy postmenopausal women with no evidence of GI lesions at baseline.19,20 In these 2-week studies, endoscopic examinations, performed on days 8 and 15 of therapy, revealed that, compared with alendronate, risedronate reduced the risk for gastric ulceration by 50% to 70%, with the absolute incidence of gastric ulceration about 12% with alendronate and 4% to 6% with risedronate. But do these clinical trial findings correspond with the real-world findings that would reflect the population-based managed healthcare experience with these products?
To examine the differences in GI events between alendronate and risedronate, Miller and colleagues conducted a retrospective cohort study using the Protocare database.21 The study included men and women at least 65 years of age with new risedronate (5 mg) and alendronate (5 or 10 mg daily and 35 or 70 mg weekly) prescriptions between November 1, 2000, and May 31, 2002, a 19- month capture period. Excluded were patients with a diagnosis of Paget's disease, patients taking risedronate 30 mg or alendronate 40 mg daily (presumptive Paget's disease patients), as well as patients who switched products during the treatment period.
International Statistical Classification
of Diseases, 9th Revision
This study comprised 2 phases: the pretreatment period (6-month period before initiating bisphosphonate therapy) and the treatment period (4 months after the initiation of bisphosphonate therapy). During both phases, GI events, prescription GI medication use (H2 receptor blocker, proton pump inhibitor [PPI], and/or cytoprotective agents), and prescription nonsteroidal anti-inflammatory drug (NSAID)/salicylate use were evaluated. During the pretreatment period, a broad list of () diagnosis codes were used to distinguish individuals who had preexisting GI conditions and were thus predisposed to develop GI problems after initiating bisphosphonate therapy. In the treatment period, a more restricted code list was used to define GI events, including only those conditions that could possibly be associated with bisphosphonates.
Two methods were used to examine GI events: assessing those recorded as a primary diagnosis only, and those recorded as a primary diagnosis and up to 4 secondary diagnoses for a particular office visit or hospitalization. A total of 6120 mostly female (93%) patients with a mean age of 75 years initiated new bisphosphonate therapy: 865 receiving risedronate and 5255, alendronate (daily and weekly regimen groups combined because of similarity in the risk for GI events). In terms of sex, age, concomitant medication use, and mean number of hospitalizations and specialist visits, no significant differences were detected between the alendronate and risedronate groups.
During the pretreatment phase, GI-related events were significantly (= .02) higher in the group that later received risedronate (13.8%), when compared with the group that went on to receive alendronate (11.0%), alendronate patients being 23% less likely to experience a GI event pretreatment. During this phase, the prevalence of prescription GI medication use was similar in the risedronate (19.0%) and the alendronate (18.7%) groups. After treatment, GI events, based on a primary diagnosis code, occurred in 8.2% and 5.5% of the alendronate and risedronate patients, respectively (Figure 4). During the first 4 months of treatment, the prevalence of GI medication use was similar for both groups, about 17%.
After adjusting for age, sex, preexisting GI conditions, and the number of concomitant medications used, patients treated with alendronate demonstrated a statistically significant (= .03) 44% higher risk of GI events than patients treated with risedronate. Moreover, in patients with no history of GI events, the risk for GI events during the treatment period was a significant (= .03) 49% higher with alendronate than with risedronate therapy, a relative increase in GI events similar in magnitude to that seen in a randomized endoscopy trial of risedronate versus alendronate.19 It should also be noted that the adjusted relative risk for GI events was a significant (= .05) 57% higher with alendronate versus risedronate in patients with a known history of GI events. No significant differences were noted between groups in the use of NSAIDs or GI medications during the treatment period. These observational findings from a real-world healthcare database appear in step with results of randomized controlled trials and suggest that, over a 4-month period, alendronate may be associated with a significantly higher risk for GI events than risedronate in Medicare-eligible patients =65 years of age.
Kane and colleagues, using the same Protocare database, explored the resource utilization and associated direct medical costs of GI events related to alendronate and risedronate treatment.22 In this study, the inclusion and exclusion criteria were similar to those in the previous study by Miller and colleagues, except that patients with a history of GI events were excluded to obtain "cleaner" cost estimates. The study included men and women at least 65 years of age with new risedronate (5 mg) or alendronate (5 or 10 mg daily and 35 or 70 mg weekly) prescriptions between November 1, 2000, and February 28, 2002, a 16-month capture period. The study comprised a 6- month pretreatment period and a 4-month treatment period. During the pretreatment period, GI events were defined primarily by codes, prescriptions for GI medications, and procedure codes. During the treatment period, GI-related direct medical costs, the primary outcome variable, were assessed using reimbursed amounts, in 2002 dollars, for GI-related outpatient costs (cost of hospital outpatient and physician office visits) and inpatient costs (cost of stay as hospital inpatient, or in an emergency department, surgical suite, or extended-care facility), GI medication costs (cost of prescriptions for H2 antagonists, PPIs, and cytoprotectives), as well as other GI-related costs.
This study included a total of 4259 predominantly female (94%) participants, mean age, 75 years, who received either risedronate (n = 623) or alendronate (n = 3636) treatment. No differences were detected between treatment groups in terms of age, sex, or previous exposure to bisphosphonates.
During the treatment period, GI events affected significantly (= .034) fewer patients treated with risedronate (3.4%) than with alendronate (5.4%). Total average GI-related direct medical costs per 1000 members per month was significantly (< .001) lower with risedronate ($2164) than with alendronate ($5996) treatment, irrespective of daily or weekly dosing.
Further, outpatient and inpatient costs (per 1000 members per month) were 2.6 and 6 times higher, respectively, with alendronate versus risedronate therapy, while the actual medication cost was similar between treatments (Table 3).
These cost differences may be attributable to a reduction in resource utilization detected with risedronate. When compared with alendronate, risedronate patients averaged about half the number of outpatient visits (8.8 vs 15.3 per 1000 members per month; = .027) and inpatient visits (2.4 vs 6.0 per 1000 members per month; = .016).
The results from this study, based on commercial insurance claims data across a variety of populations and health plans, imply that the appropriate use of bisphosphonate therapy can reduce the incidence and cost of bisphosphonate-related GI events.
The results of randomized clinical trials, because of their control of confounding variables, provide indispensable information related to the efficacy and tolerability of bisphosphonates in the management of osteoporosis-related fracture. Yet, because standard inclusion and exclusion criteria used by the studies filter out many patients who would be encountered in a typical clinical setting, their results should be viewed with circumspection in the managed care setting. Observational studies, using information from large, integrated, managed care claims databases, can complement and extend the results of randomized controlled trials by providing an important glimpse into real-world disease prevalence as well as drug usage and cost patterns. It should be kept in mind that observational studies, by their retrospective nature, are associated with important inherent limitations. For instance, they cannot establish causation or eliminate the influence of extraneous variables that may confound the results. Observational studies have, nonetheless, helped elucidate cost-effectiveness issues in the use of bisphosphonate therapy for osteoporosis-related fracture reduction.
In the managed care setting, the true cost effectiveness of bisphosphonate therapy should be measured not only in terms of the cost savings accruing from reductions in fracture risk but also by how well an agent reduces the risk for troublesome and expensive untoward effects, such as GI events. When these 2 factors are considered together, based on results from large observational studies using claims databases, risedronate appears to offer substantial cost benefits versus alendronate therapy.
Corresponding author: Diana Brixner, PhD, RPh, Associate Professor and Chair, Department of Pharmacotherapy, Outcomes Research Center, University of Utah, Salt Lake City, UT. E-mail: email@example.com.