One year outcomes and costs following a vertebral fracture

Transcription

1 Osteoporos Int (2005) 16: DOI /s x ORIGINAL ARTICLE One year outcomes and costs following a vertebral fracture R. Lindsay Æ R. T. Burge Æ D. M. Strauss Received: 18 June 2003 / Accepted: 1 March 2004 / Published online: 27 May 2004 Ó International Osteoporosis Foundation and National Osteoporosis Foundation 2004 Abstract Vertebral fractures are believed to be important predictors for future vertebral and other fractures, leading to at least a 4- to 5-fold increase in the risk of subsequent fractures. However, little is known about their associated near-term costs. The purpose of this study was to quantify the subsequent fracture and cost outcomes emanating from patients with an incident vertebral fracture. A probabilistic decision analysis model was developed to estimate the expected cost of all subsequent fractures. We ran Kaplan-Meier time-toevent models on placebo patients in risedronate s pivotal phase III clinical trial data to determine the cumulative incidence or probabilities of all fractures within one year of an incident vertebral fracture. Unit costs for health care payers in the USA and Sweden for vertebral, hip, other, and forearm/wrist fractures were multiplied by fracture probabilities to generate the expected costs of new fractures within one year of incident vertebral fractures. Our analysis found that that 26.1% of vertebral fracture patients with a mean age of 74 years refractured within 1 year (vertebral 17.4%; hip 3.6%; other 3.5%; forearm/wrist 1.6%). The calculated medical costs for those patients who refracture within 1 year was $5906 and 3670 for the USA and Sweden, respectively, while the weighted average cost across all patients (refracture and non-fracture) within a year of their incident fracture was $1541 (USA) and 958 (Sweden). These results suggest that therapies with R. Lindsay (&) Helen Hayes Hospital, West Haverstraw, N.Y., USA R. T. Burge Æ D. M. Strauss Department of Epidemiology and Pharmacoeconomics, Procter & Gamble Pharmaceuticals, Mason, Ohio, USA R. T. Burge School of Public Health, Boston University, Boston, Mass., USA proven, rapid efficacy may offer important economic value to healthcare payers, providers and patients. Keywords Costs Æ Osteoporosis Æ Outcomes Æ Refracture Æ Vertebral fractures Introduction Vertebral fractures are known to be important predictors for future vertebral and other fractures and may substantially increase the risk of subsequent fractures [1,2,3,4,5,6]. A summary of the literature suggests that a previous vertebral fracture increases the relative risk of new vertebral fractures between 3.9 and 11.1 times, depending on the number of pre-existing vertebral fractures, and the relative risk of hip, wrist, and other fractures increases between 1.7 and 3.8, 1.4 and 1.5, and 1.5 to 4.4 in women, respectively [7]. A recent study in the United Kingdom [6] reported a large increase in the risk of future fractures at different skeletal sites following an initial fracture event. The link between an incident and subsequent fracture was not restricted to a few skeletal sites, but existed across different sites. The excess risk of further fracturing after a vertebral fracture, for instance, was not limited to the hip, as humerus fractures were increased 3-fold and rib fractures 5-fold. Much less is understood about the associated costs following vertebral fractures, as only about 35 50% of vertebral fractures are clinically treated [9,10]. Furthermore, only about 1 in 12 elderly hospitalized women with radiographic evidence of a vertebral fracture are identified on the discharge abstract [11]. Although the costs of vertebral fracture events have been estimated [12,13,14, 15,16,17], none has shown the full, near-term cost impact of new fractures following vertebral fractures. Therefore, the purpose of this study is to estimate the expected short-term costs of new fractures emanating from elderly female patients with an incident vertebral fracture for health care payers in the USA and in Sweden.

2 79 Fig. 1 Decision analysis model of 1-year outcomes for vertebral fracture patients. Decision analysis by TreeAge (DATA), Version [18] software used to create the decision tree and to conduct all cost estimates Materials and methods A probabilistic decision tree model was developed to estimate the expected cost of all subsequent fractures emanating from a vertebral fracture event (Fig. 1). The model combines the probability of fractures and the cost of fractures to generate an expected or predicted cost outcome per patient. Calculation of 1-year re-fracture probabilities The probabilities of subsequent fracture outcomes (hip, vertebral, forearm/wrist, other) occurring within 1 year of an incident vertebral fracture event were calculated using patients who received placebo in the risedronate Phase III clinical trials: VERT-NA [19], VERT-MN [20], and HIP [21]. A recent study [22] used the same data and identified 381 placebo patients whose postbaseline vertebral status could be determined, and who suffered an incident vertebral fracture. The percentage of these patients suffering a new vertebral fracture within 1 year was estimated at 19.2%. We extended the analyses done in that previous study [22] to obtain estimates of the cumulative incidence or percentage of these same 381 patients incurring a new vertebral and/or non-vertebral fracture (i.e. hip, forearm/wrist, other), within 1 year of the incident vertebral fracture. These percentages of patients with new fractures, by fracture type, were applied as probabilities in the analysis to estimate costs. The mean age of these patients was 74 years. The relative risk of a new vertebral fracture increased by the number of prevalent vertebral fractures at baseline. Details on the baseline characteristics of the study population have been reported elsewhere [22]. In our analysis, we assigned the 381 patients into the following mutually exclusive fracture categories based on fracture events within 1 year of their first incident vertebral fracture: (1) patients with a hip/pelvis fracture; (2) patients not in (1) above and had a humerus/leg fracture; (3) patients not in (1) or (2) above and had a wrist fracture; (4) patients not included in (1) through (3) above and who had a new incident vertebral fracture; and (5) patients without a new vertebral or a non-vertebral fracture (i.e. no new fractures). The assignment of patients into the five fracture categories was done on a hierarchical basis according to severity of fracture so as to appropriately capture the expected associated costs of fracture events. For instance, a patient suffering both a hip fracture and a forearm/wrist fracture within 1 year of the index vertebral fracture would be assigned to the hip fracture category due to its greater severity, resource intensity and cost impact. Separate Kaplan-Meier (K-M) analyses were done for the time to the hip/pelvis, vertebral, wrist and humerus/leg fractures. An analysis was also done for time to either a vertebral or any of the above non-vertebral fractures within 1 year of the first incident vertebral fracture. For patients who fractured within 1 year, the time to event was the time to the specific fracture of interest from the time of the first incident vertebral fracture. Patients who did not fracture within 1 year of the first incident vertebral fracture, were censored at either (a) 12 months, or (b) the time from first incident vertebral fracture to last vertebral study visit if this elapsed time was less than 12 months. The results of these analyses are presented in Table 1. These 1-year fracture incidence values or percentages, Table 1 Summary of patients fracturing again within 1 year of their first incident vertebral fracture Fracture type K-M estimates a Frequency (%) 95% confidence interval Percentage Lower Upper Hip/pelvis 7 (1.8) Humerus/leg 7 (1.8) Vertebral 33 (8.7) Wrist 3 (0.8) Overall b No vertebral, no non-vertebral 331 (86.9) a The Kaplan-Meier estimate is the estimate of the cumulative proportion of patients who fractured, within 1 year of their first incident vertebral fracture. For patients whose non-vertebral fracture occurred after the last visit at which the status of vertebral fractures was determined, the time to non-vertebral fracture was used as the time of event or censoring b Overall estimate is the result of summing the individual Kaplan- Meier estimates for hip/pelvis, vertebral, wrist, and humerus/leg

3 80 based on the K-M estimates, were entered into the decision tree model as probabilities of new fractures to generate the expected 1-year costs for vertebral fracture patients in the base case scenario. The 95% lower and upper confidence limits were used for sensitivity analyses. Of the 50 patients reporting fractures within 1 year of their first incident vertebral fracture, four suffered both vertebral and non-vertebral fractures. It should be noted that the estimate of 17.3% for vertebral fractures within 1 year is different from the previous estimate of 19.2% [22]. The difference is due to the fact that in our analyses, three of the 381 patients had both hip/pelvis and vertebral fractures and were assigned to the hip/pelvis fracture category rather than the vertebral fracture group according to the hierarchical decision rules based on fracture severity. However, using the present model, if the order of assignment to categories 1 and 4 was reversed and if the time to vertebral fracture was used as time to event, an estimate of 19.2% was obtained. Unit costs of fractures The expected unit costs to USA and Sweden health care payers for four major fracture categories hip, vertebral, forearm/wrist and other are presented in Table 2 below. For USA costs the Medical Care Services component of the Urban Consumer Price Index (CPI), Bureau of Labor Statistics, US Department of Labor, was used to create price inflation factors that were used to update all costs into year 2004 dollar terms. A brief discussion about the sources for each cost component for each facture type is provided in the following sections. Table 2 Unit costs by fracture type for women aged Cost component Fracture type Hip Vertebral a Forearm/Wrist b Other c Inpatient care $20,441 $488 $788 $1480 Outpatient care $3,615 $156 $464 $282 Long-term care d $12,071 $92 $176 $589 USA total $36,127 $736 $1428 $2351 Sweden total e 164,861 SEK 7476 SEK 20,654 SEK 33,970 SEK ( 17,958) ( 814) ( 2250) ( 3700) a Costs for the USA and Sweden are weighted averages between clinical and non-clinical fractures after applying the clinical fracture probability of 23%. For the USA, all costs associated with hospitalized cases were multiplied by 22.5% to reflect the inpatient hospital proportion of total clinical cases b USA inpatient costs were multiplied by 10% to reflect the inpatient proportion of total cases c USA inpatient costs were multiplied by 16.5% to reflect the inpatient proportion of total cases d USA nursing home costs for hip fracture patients include partial year and full year patient stays e SEK converted to euros ( ) using exchange rate to 1 =9.18 SEK (February 2004) Hip fractures: USA The unit costs of inpatient care included included hospital facility, inpatient physician, emergency room, short-stay/rehabilitation hospital admissions, and acute inpatient hospital readmission costs. Published estimates for the costs of inpatient physician services [17], emergency room services [23], short-stay/rehabilitation hospital admissions [23] and acute inpatient hospital readmissions [23] were updated to year 2004 dollar costs ($2851, $451, $1245, and $739, respectively). The costs for acute inpatient hospital facility care were estimated from the Nationwide Inpatient Sample (NIS) 1997 database, which is a large US hospital database with admissions from over 1000 short-term acute care hospitals [24]. A cost-to-charge ratio of 0.61 [25] was multiplied to average charges to obtain the cost estimate. The 1997 estimate was then updated to $15,155 to reflect year 2004 dollars. All costs for outpatient care were obtained from the NOF Working Group study [23], which included the costs for home health care, nonmedical home care, post-discharge physician visits, and disability-related costs. All costs were converted into 2004 dollars for a total of $3615. Long-term care costs were obtained from the NOF Working Group study [23] and were updated from $7193 in 1992 to $12,071 in Vertebral fractures: USA The cumulative incidence rates for vertebral fractures in Table 1 were based on evaluable radiographs of the lateral spine, and included a mix of clinical (23%) and non-clinical (77%) fractures [22]. Thus, the unit cost of a radiographic vertebral fracture used in the analysis is a weighted average of the costs of clinical and non-clinical vertebral fractures (where non-clinical fractures are assumed to have no associated cost). When estimating the costs of a clinical fracture, we used a weighted average between hospitalized and non-hospitalized patients. The literature suggests that about 20 25% of clinical cases are hospitalized [16,26,27,28,29]; thus, we applied a median proportion of 22.5% for costs of hospital cases and 77.5% for costs of non-hospitalized cases to calculate the weighted average cost, where the weighted average cost=(0.23) [(0.225) (costs of hospital cases)+(0.775) (costs of non-hospital cases)]. Inpatient facility costs were estimated by taking published estimates [12] on hospital charges for women aged and multiplying by a cost-to-charge ratio of 0.61 [25]. These costs were then updated to $8149 for the year The expected costs for inpatient facility care for radiographic fractures was $422 (= $8149). For inpatient physician costs, we updated the costs of physician services from the literature [17] using prices from the Medicare Fee Schedule for 2004, resulting in a cost of $1286. The expected inpatient physician service costs were $66 (= $1286). Total inpatient

4 81 costs were the sum of inpatient facility and inpatient physician service costs ($488=$422+$66). The costs of outpatient care included outpatient services and disability care for hospitalized patients after discharge; and for non-hospitalized patients. The costs of outpatient services were $598 for hospitalized patients, post-hospital discharge; and $442 for non-hospitalized patients [15]. Disability-related costs were taken from the NOF Working group study [23] and were updated to $201 for the year The weighted average outpatient cost for vertebral fractures was $156 =(0.23) [0.225 ($598+$201)+(0.775) ($442+$201)]}. Long-term care costs for vertebral fractures of $1785 were calculated by multiplying the ratio of long-term care to inpatient care costs (21.9%) from the literature [14], by the inpatient care costs for vertebral fractures ($8149); or $1785 (=0.219 $8149). The expected cost of long-term care for vertebral fractures used in the model was $92 [=0.23 (0.225 $1785)]. Forearm/wrist fractures: USA The unit cost of a forearm/wrist fracture was based on estimates from the NIS 1997 database and from the literature [14,23]. Mean hospital charges were multiplied by a cost-to-charge ratio of 0.61 to arrive at costs of $7880 after being updated to year 2004 dollar terms. Hospital inpatient costs were multiplied by 10% to yield $788 to reflect the proportion of all forearm/wrist fractures admitted to an inpatient hospital [16]. The costs of outpatient care included the costs of outpatient physician services and the costs of disability-related care. Outpatient physician service costs were estimated by multiplying the ratio of outpatient care to inpatient care costs [14] by the cost of inpatient care for wrist fractures, or $400=0.508 $788. Disability related costs were obtained from the NOF Working Group study [23] and were $64 after updating to year 2004 dollar terms. Longterm care costs were derived by taking the national proportion of long-term care to inpatient care costs (22.4%; [14]) and multiplying by the inpatient care estimates to yield $176 (=0.224 $788). Other fractures: USA Similar to forearm/wrist fractures, the unit costs of inpatient facility care for other fractures were obtained from an analysis of the NIS 1997 inpatient hospital database [24]. Mean charges were multiplied by a costto-charge ratio of 0.61 [25] and updated to the year The mean cost of $8970 was then multiplied by 16.5% to arrive at a cost of $1480 (=0.165 $8970) in order to adjust for the proportion of cases requiring a hospital stay, where we assumed that the other fracture cases would be hospitalized at a proportion between those of forearm/wrist (10%) and vertebral fractures (22.5%). Expected outpatient costs and long-term care costs were calculated using the same method as for forearm/wrist fractures, where the national proportion of outpatient costs to acute inpatient care costs (19.1%) and long-term care to inpatient costs (39.8%) [11] were multiplied by the inpatient cost estimate of $1480 for other fractures from the NIS database [24]. Expected outpatient costs were $282 (=0.191 $1480) and expected long-term care costs were $589 (=0.398 $1480). Unit costs of fractures: Sweden The unit costs of fracture for Sweden were obtained from the published literature [30,31]. The direct cost of hip fracture for women aged 75 was 164,861 SEK ( 17,958) [30], while the direct costs of vertebral, wrist and other (shoulder) were 32,504 SEK ( 3541), 20,654 SEK ( 2250), and 33,970 SEK ( 3700), respectively. For vertebral fracture unit costs, the 23% clinical proportion was applied to yield the expected cost of a radiographic fracture of 7476 SEK ( 814) for use in the analysis. Results Predicted costs: base case The base case results are shown in Table 3. The overall predicted cost of a patient with an incident vertebral Table 3 Predicted costs of subsequent fractures in women with an incident vertebral fracture Fracture type Probability USA Sweden Overall a Conditional b Unit cost Predicted cost Unit cost Predicted cost Hip 3.6% 13.9% $36,127 $ , Vertebral 17.3% 66.5% $736 $ Forearm/wrist 1.6% 6.1% $1428 $ Other 3.5% 13.5% $2351 $ Total: new fractures 26.1% 100% $ Total: no new fractures 73.9% $0 $0 0 0 Overall $ a Overall probability based on percentages or cumulative incidence values form K-M survival functions b Conditional probabilities calculated by dividing overall probability for each fracture by the total probability of a new fracture (26.1%)

5 82 Table 4 One-way sensitivity analysis on fracture probabilities and unit costs for the USA Variable Model input value Predicted cost fracture over the following 12 months is $1541 in the USA, and 958 in Sweden. The estimates are a weighted average of fracture patients (26.1%) and non-fracture patients (73.9%). Also, the expected cost estimates exclude costs associated with the index vertebral fracture episode, and all patients were assumed to have no osteoporosis therapy costs. Sensitivity analysis Low High Low High Probability a of Hip fracture 0.99% 6.25% $591 $2492 Vertebral fracture 11.97% 22.74% $1502 $1581 Forearm/wrist fracture 0.00% 3.36% $1519 $1567 Other fracture 0.96% 6.10% $1481 $1602 Unit costs b of Hip fracture $27,091 $45,159 $1214 $1868 Vertebral fracture $552 $921 $1509 $1573 Forearm/wrist fracture $1071 $1785 $1536 $1547 Other fracture $1763 $2939 $1521 $1562 a The low and high input values for fracture probabilities are based on the 95% confidence intervals b For unit costs, +25% or 25% from the mean values were used We assessed the robustness of our predicted cost estimates in the base case by conducting one-way and multiway variable sensitivity analysis. Sensitivity analyses are important in outcomes studies because many parameters often are combined from a variety of sources, thus increasing overall uncertainty in the predicted results. In Tables 4 and 5, the one-way sensitivity analysis results are shown when the lower and upper 95% confidence intervals around the 1-year fracture probabilities were applied to the model. Changes in the probability of a hip fracture have the greatest impact on the expected cost of new fractures, ranging from $591 to $2492 for the USA, and 486 to 1430 in Sweden (Table 5). Changes in other fracture probabilities have minimal effect on the model s expected cost results. A similar pattern was found for unit costs of fracture. Increases or decreases of 25% in unit costs had a smaller effect compared to changes in probabilities. The largest impact was found for changes in unit cost of hip fracture, where the predicted cost from the model for the USA was $ for low and high unit costs, respectively; and in Sweden. Table 6 summarizes the results from the multi-way sensitivity analyses. Under each scenario, all variables indicated are changed simultaneously. For example, under the lower 95% confidence interval scenario, the low probability value is applied to all fracture probabilities. The results are consistent with the one-way sensitivity analyses, as changes in fracture probabilities have the greatest impact on the model s base case expected cost. Under the low probability scenario, expected costs per patient fall to $468 in the USA and 311 in Sweden, or almost one-third of the base case results. If high-end probabilities are used, predicted costs rises to $2617 in the USA and 1609 in Sweden. The table shows the impact of our hierarchical ranking of fracture patients. If we apply the 19.2% probability from Lindsay et al. [22], the probability of hip fracture must be reduced and the overall predicted cost falls to $995 in the USA and 695 in Sweden. Next, the results from simultaneously changing all unit costs of fractures are presented, where the predicted costs decrease or increase consistent with unit fracture cost changes, as expected, where low unit costs decrease the base case estimates to $1156 (USA) and 718 (Sweden), while high unit costs increase the base case values to $1927 (USA) and 1197 (Sweden). Lastly, we show the predicted results if incremental fracture costs are used in place of direct Table 6 Multi-way sensitivity analysis Table 5 One-way sensitivity analysis on fracture probabilities and unit costs for Sweden Variable Model input value Predicted cost Low High Low High Probability a of Hip fracture 0.99% 6.25% Vertebral fracture 11.97% 22.74% Forearm/wrist fracture 0.00% 3.36% Other fracture 0.96% 6.10% Unit costs b of Hip fracture 13,469 22, Vertebral fracture Forearm/wrist fracture Other fracture a The low and high input values for fracture probabilities are based on the 95% confidence intervals b For unit costs, +25% or 25% from the mean values were used Scenario Predicted costs Fracture Fracture USA Sweden probabilities unit costs Base Base $ Lower 95% confidence Base $ intervals Upper 95% confidence Base $ intervals Vertebral fracture Base $ probability=19.2% a Base Low $ Base High $ Base Incremental b $ a The probability of hip fracture is reduced to 2.07% in this analysis b USA incremental costs [32], updated to year 2004 dollars, were $18,864 (hip), $755 (vertebral), $2732 (forearm/wrist), and $2475 (other). Sweden incremental costs [33], updated to year 2004, were 10,878 (hip), 277 (vertebral), 1363 (forearm/wrist), and 2241 (other)

6 83 medical costs. For the USA, we apply incremental costs from the literature [32] updated to year 2004 dollars, after imposing the 23% clinical vertebral fracture rate. Incremental fracture costs of hip and vertebral fractures for Sweden were taken from a European study [33], while we imputed the incremental costs of forearm/wrist and other by applying the ratios of direct forearm/wrist and other fracture costs to direct hip fracture costs, respectively, to the incremental hip fracture cost estimate. The incremental costs for Sweden were updated to the year 2004 using the consumer price index for Sweden (Statistics Sweden). The results show lower predicted costs compared to our low unit cost scenarios, where the USA results are $945 and the Sweden results are 543. Discussion Several studies have shown vertebral fractures to elevate the relative risk of new fractures, but none has reported the expected cost impact from subsequent fractures within a single year. The analysis presented here used a decision analysis model to estimate the expected costs of fracture outcomes within 1 year of an incident vertebral fracture. One-year fracture probabilities for hip, vertebral, forearm/wrist and other fractures were estimated using Kaplan-Meier time to event models on placebo patients with incident vertebral fractures in the risedronate Phase III studies (VERT-NA [19], VERT-MN [20], and HIP [21]). The incidence of a new fracture within a year was 26.1%. The literature provides estimates of the increased relative risk of new fractures following a vertebral fracture; however, most of those studies do not account for the probability of multiple types of new fractures. That is, only risks for specific fracture types are captured rather than all fracture types within a similar time frame. By using a combination of studies conducted on a wide variety of patient populations with different underlying risk profiles, the refracture probability for any given type of fracture may be biased. Studies that have estimated subsequent risks of multiple fracture types following a vertebral fracture include those by Black et al. [3], Melton et al. [6] and van Staa et al. [8]. The latter study estimated risk of new fractures only for those different from the index type. In the study by Black et al., the Study of Osteoporotic Fractures (SOF) database was used to estimate the risk of all types of new fractures but they do not report 1-year refracture probabilities. Melton et al. [6] conducted an observational study on a cohort of residents in Rochester, Minn., USA. Vertebral fractures were limited to compression fractures of the spine between vertebra T1 and L5. Subsequent vertebral fractures were counted only if they occurred in a different thoracic or lumbar vertebra. Therefore, instead of using probabilities from the literature, which may lack 1-year precision, we estimated 1-year fracture probabilities among a population of patients with incident vertebral fractures and known risk profiles under well-controlled clinical trials. Unit costs of each fracture type and 1-year fracture probabilities were entered into the model and the expected cost within 1 year of an incident vertebral fracture event was $5906 for patients who re-fracture, $1541 across all patients (i.e. weighted average of those who refracture and those who do not) in the USA. Comparable figures for Sweden were 3670 and 958. These results were most sensitive to changes in the probability and unit cost of a hip fracture, but generally robust to changes in other parameters. There are some important limitations to this analysis that require discussion. In this study, a causal association is assumed between the index vertebral fracture and subsequent fractures. Although we cannot ascertain the exact causal relationship between these intertemporal fracture events, the probability of new fractures estimated here is consistent with evidence from the literature regarding vertebral fractures as being important risk factors for subsequent fracture events. Another limitation concerns the generalizability of our results to other patient populations. Our patient sample had a mean age of 74, and over 80% had a prevalent vertebral fracture at baseline in the clinical trials before their index vertebral fracture. The probability of new fractures may differ among younger or older patient groups with different fracture histories. We also analyzed the costs only for the USA and for Sweden. Although the results from the two countries provide some frame of reference for other countries, the cost structures and reimbursement environments can differ significantly both within (especially in the USA) and across countries in North America and Europe. Therefore, extreme caution should be exercised when attempting to generalize these results to other geographies and health care systems. Any attempts to do so should apply local estimates of unit costs of fractures. Finally, we do not consider all relevant outcomes following a vertebral fracture, including the loss in patients health-related quality of life, costs of multiple fractures, and the economic impact of treating patients with osteoporosis therapies. For instance, recent research on a USA population suggests that on a scale from 1 (perfect health) to 0 (death), the health utility decrease from a hip fracture in the first year is 0.18 and 0.09 in subsequent years, while a vertebral fracture leads to a loss in health utility of 0.16 and 0.08 in the first and subsequent years, respectively [34]. In those women suffering a hip fracture who have a prevalent vertebral fracture, the first year loss in utility is Another study on a Swedish population reports that utility loss from morphometric vertebral fractures is 0.223, while clinical vertebral fractures lead to a utility loss of in the first year following the fracture [35]. Further research is needed to gain a better understanding of the longer term health-related quality of life consequences among women with prevalent vertebral fractures. In terms of economic benefits of treating these vertebral fracture patients, the potential cost reductions from

7 84 fewer fractures under therapy must be weighed against the added therapy cost, and then compared to the costs of not treating patients. For example, using the decision analytic model developed here, if vertebral fracture patients are treated with a therapy costing $700 annually ( 560), the threshold for experiencing overall cost savings (compared with not treating the patients) would be met if the therapy reduced fractures by 45% in 1 year in the USA, and by 58% in Sweden. Reductions in fractures greater than 45% in the USA and above 58% in Sweden would lead to larger net savings. The possible cost-effectiveness of using osteoporosis therapies would require assumptions on the efficacy of these therapies in preventing multiple fracture types within a 1-year period. Additional studies are necessary to confirm these assumptions. In conclusion, the results of this analysis suggest that patients with an incident vertebral fracture, if left untreated, will incur significant costs attributable to additional fractures within a single year. The use of therapies with proven efficacy within a single year, in terms of reduction in the relative risk of fractures, may reduce patient morbidity and minimize medical care system costs. Acknowledgements This study was supported by the Alliance for Better Bone Health (Aventis Pharma and Procter & Gamble Pharmaceuticals). Conflict of interest: No information supplied. References 1. Ross PD, Davis JW, Epstein RS, Wasnich RD (1991) Preexisting fractures and bone mass predict vertebral fracture incidence in women. Ann Int Med 114: Ross PD, Genant HK, Davis JW, Miller PD, Wasnich RD (1993) Predicting vertebral fracture incidence from prevalent fractures and bone mass density among non-black, osteoporotic women. Osteoporos Int 3: Black DM, Arden NK, Palermo L, Pearson M, Cummings SR (1999) Prevalent vertebral deformities predict hip fractures and new vertebral deformities but not wrist fractures. J Bone Miner Res 14: Kotowicz M, Melton LJ, Cooper C, Atkinson EJ, O Fallon WM, Riggs BL (1994) Risk of hip fracture in women with vertebral fracture. J Bone Miner Res 5: Johnell O, Kanis JA, Oden A, Sernbo I et al. (2004) Fracture risk following an osteoporotic fracture. Osteoporos Int 15: Melton LJ, Atkinson EJ, Cooper C, O Fallon W, Riggs BL (1999) Vertebral fractures predict subsequent fractures. Osteoporos Int 10: Klotzbuecher CM, Ross PD, Landsman PB, Abbott TA, Berger M (2000) Patients with prior fractures have an increased risk of future fractures: a summary of the literature and statistical synthesis. J Bone Miner Res 15: van Staa TP, Leufkens HG, Cooper C (2002) Does a fracture at one site predict later fractures at other sites? A British cohort study. Osteoporos Int 13: Cooper C, Atkinson EJ, O Fallon W, Melton LJ (1992) Incidence of clinically diagnosed vertebral fractures: a populationbased study in Rochester, Minnesota: J Bone Miner Res 7: Clark AP, Schuttinga JA (1992) Targeted estrogen/progestogen replacement therapy for osteoporosis: calculation of health care cost savings. Osteoporos Int 2: Gehlbach SH, Bigelow C, Heimisdottir M, May S, Walker M, Kirkwood JR (2000) Recognition of vertebral fracture in a clinical setting. Osteoporos Int 11: Burge RT, Puleo E, Gehlbach S, Worley D, Klar J (2002) Inpatient hospital and post-acute care for vertebral fractures in women. Value-In-Health 5: Gehlbach SH, Burge RT, Puleo E, Klar J (2003) Hospital care of osteoporosis-related vertebral fractures. Osteoporos Int 14: Ray NF, Chan JK, Thamer M, Melton LJ (1997) Medical expenditures for the treatment of osteoporotic fractures in the United States in 1995: report from the National Osteoporosis Foundation. J Bone Miner Res 12: Phillips S, Fox N, Jacobs J, Wright W (1988) The Direct medical costs of osteoporosis for American women aged 45 and older, Bone 9: Chrischilles E, Shireman T, Wallace R (1994) Costs and health effects of osteoporotic fractures. Bone 15: US Congress Office of Technical Assistance (1994) Hip fracture outcomes in people age fifty and over: background paper. US Government Printing Office, Washington, D.C. 18. Decision Analysis by TreeAge (DATA), Version TreeAge Software, Inc. Williamstown, Mass., USA 19. Harris ST, Watts NB, Genant HK et al. (1999) Effects of risedronate treatment on vertebral and nonvertebral fractures in women with postmenopausal osteoporosis: a randomized controlled trial. JAMA 280: Reginster J, Minne HW, Sorensen OH et al. (2000) Randomized trial of the effects of risedronate on vertebral fractures in women with established postmenopausal osteoporosis. Vertebral Efficacy with Risedronate Therapy (VERT) Study Group. Osteoporos Int 11: McClung MR, Geusens P, Miller PD et al. (2001) Effect of risedronate on the risk of hip fracture in elderly women. N Engl J Med 344: Lindsay R, Silverman SL, Cooper C, Hanley DA, Barton I, Broy SB, Licata A, Benhamou L, Geusens P, Flowers K, Stracke H, Seeman E (2001) Risk of new vertebral fracture in the year following a fracture. JAMA 285: National Osteoporosis Foundation (1998) Osteoporosis: review of the evidence for prevention, diagnosis and treatment and cost-effective analysis. Osteoporos Int Supplement 4:s1 s Nationwide Inpatient Sample, Release 6 (1997) Healthcare Cost and Utilization Project, Agency for Healthcare Research and Quality, Rockville, Maryland, USA 25. Hoerger T, Downs K, Lakshmanan M et al. (1999) Healthcare use among US women aged 45 and older: total costs and costs for selected postmenopausal health risks. J Women s Health Gender-Based Med 8: Jacobsen SJ, Cooper C, Gottlieb MS et al. (1992) Hospitalization with vertebral fracture among the aged: a national population-based study, Epidemiology 3: Ankjaer-Jensen A, Johnell O (1996) Prevention of osteoporosis: cost-effectiveness of different pharmaceutical treatments. Osteoporos Int 6: Cooper C, O Neill T, Silman AJ (1993) European Vertebral Osteoporosis Study Group. The epidemiology of vertebral fractures. Bone 14:S Holbrook TL et al. (1984) The frequency of occurrence, impact, and cost of selected musculoskeletal conditions in the US. American Academy of Orthopedic Surgeons, Chicago 30. Zethraeus N, Stro mberg L, Jo nsson B, Svensson O, O hle n G (1997) The cost of a hip fracture. Estimates for 1709 patients in Sweden. Acta Orthop Scand 68: Zethraeus N, Borgstro m F, Johnell O, Kanis J, O nnby K, Jo nsson B (2002) Costs and quality of life associated with osteoporosis related fractures results from a Swedish survey. SSE/EFI Working Paper Series in Economics and Finance, No 512, October

8 Gabriel SE, Tosteson ANA, Leibson CL et al. (2002) Direct medical costs attributable to osteoporotic fractures. Osteoporos Int 13: De Laet C, van Hout BA, Burger H et al. (1999) Incremental cost of medical care after hip fracture and first vertebral fracture: the Rotterdam study. Osteoporos Int 10: Tosteson ANA, Gabriel SE, Grove MR et al. (2001) Impact of hip and vertebral fractures on quality-adjusted life years. Osteoporos Int 12: Kanis JA, Johnell O, Oden A, Borgstrom F, Zethraeus N, De Laet C, Jonsson B (2004) The risk of vertebral fractures in Sweden. Osteoporos Int 15:20 26