Osteoporosis is a highly prevalent condition affecting more than one in five Canadian women over the age of 50 years and predisposes to low-trauma fractures.1, 2 There are significant health consequences to individuals who sustain fractures, such as pain, morbidity, and reduced quality of life.3, 4 Some fractures, most notably hip fractures, are also associated with significant excess mortality and loss of independence.5, 6
The economic impact of fractures is also noteworthy. Cost-of-illness (COI) analysis is used to evaluate the economic burden of illness in terms of health care resources (HCR) consumption and production losses.7, 8 Incidence-based COI is useful when considering fracture-prevention measures and postfracture management from a cost distribution perspective. Previous studies have shown that there are significant excess costs related to osteoporotic fractures. Hip fractures have been studied most extensively because they are easily identified in hospitalization records and typically produce the highest cost per fracture event.9, 10 Costs related to other frequent fractures also can be substantial, although this has been less well quantified.11–13 Furthermore, there is increasing interest in traumatic fractures because these are associated with costs in their own right and appear to be predictive of future low-trauma fractures among older individuals.14
Economic costs are highly sensitive to the health care environment.8 Therefore, COI studies generally are country-specific. Costs also may change over time with alterations in practice patterns and/or the introduction of new therapies. The COI for fractures in Canada was last studied in 1995–1996.9 The current analysis was undertaken to examine COI in a Canadian population (Province of Manitoba, population 1.2 million) for a range of fractures at typical sites associated with osteoporosis and to assess temporal trends over 10 years. The incremental costs of fracture (costs in the year following the fracture minus costs in the year before fracture) were compared with sex- and age-matched controls.
We used the Population Health Research Data Repository held at the Manitoba Center for Health Policy (MCHP Repository) for the Province of Manitoba, Canada, to identify incident fractures (April 1997–March 2007) and HCR costs (April 1996–March 2008).15 The study population consisted of all women and men age 50 years and older who were resident in the Province of Manitoba with continuous health care coverage during the period of interest. Microcosting and standard average costs were used to estimate costs (quantity × unit costs) in the year before (Pre) and after (Post) fracture, with all costs expressed in constant dollars (2006 Canadian dollars). Microcosting techniques were used for physician services, and prescription and average costs were used for hospital care, home care, and nursing homes.16 Although clinical practice evolved over the course of the study (eg, new medications), there were no fundamental changes to the health care system, access to hospitalization, or reimbursement for clinical services. The Manitoba health care system is typical of other Canadian provinces because it must comply with the Canada Health Act, Canada's federal legislation for publicly funded health care insurance. The act establishes criteria and conditions related to insured health services and extended health care services that the provinces and territories must fulfill to receive the full federal cash contribution under the Canada Health Transfer. The study was approved by the Research Ethics Board of the University of Manitoba, and data access was granted by Manitoba's Health Information and Privacy Committee.
The administrative health databases of the MCHP Repository, which are the sources used for this analysis, contain anonymized records for almost all contacts with the provincial health care system, including physicians, hospitals, home care, nursing homes, and pharmaceutical dispensations.17 The MCHP Repository covers virtually the entire population of the province because there are no premiums payable for coverage. It also records mortality and migration into and out of the province. Studies have found the administrative health data to have a high degree of reliability and validity.18–20
A hospital abstract is completed when a patient is discharged from an acute-care facility with diagnoses and procedures coded using the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM), prior to 2004 and the 10th Revision, Canadian version (ICD-10-CA) thereafter. Each record includes up to 16 diagnosis codes in the ICD-9-CM and up to 25 diagnosis codes in the ICD-10-CA. Physicians submit billing claims to the provincial Ministry of Health on almost all outpatient services, including outpatient departments, and most inpatient services; these claims contain a single three-digit ICD-9-CM code.
A computerized record of all outpatient drugs dispensed from pharmacies has been available since April 1, 1995. With approximately 20 million transactions annually, this system captures information about pharmaceutical use in real time for all Manitoba residents. The pharmacy database is accurate both for capture of drugs dispensed and for most prescription details.21 Each prescription record contains the date of dispensation; an exact identification of the dispensed drug, including substance, strength, route, and dosage form; the number of doses provided; the anticipated duration of the prescription in days; and a code for prescribing physician and dispensing pharmacy. All drugs are classified according to the Anatomical Therapeutic Chemical (ATC) System of the WHO (www.whocc.no/atc/).
Files containing data on nursing home residences (in Manitoba termed personal care homes [PCHs]) and home care services are also maintained for Manitoba residents. The nursing home file contains records of the number of days individuals are resident in a nursing home in Manitoba. The file containing records of the opening date and closing date of all home care episodes provides the number of days of care. These data, along with per-diem average cost information, were used to assign these costs to individuals.
We defined as cases all men or women 50 years of age and older with continuous residence in Manitoba (2 years of coverage prior to fracture date and no gaps in health care coverage of more than 90 days) whose hospitalization records and/or physician billing claims contained an incident fracture diagnosis in the period from April 1, 1996, to March 31, 2006. The date of earliest qualifying fracture code was considered the index date. Individuals living in nursing home facilities at the time of the fracture were included.
Each subject's longitudinal health service record was assessed for the presence of ICD-9-CM or ICD-10-CA fracture codes, excluding craniofacial, hand, and foot fractures that were not associated with trauma (ICD-9-CM E800-E879 and E890-E999 or ICD-10-CA V01–V99). Fracture sites of interest (collectively referred to as major osteoporotic) were those of the spine, humerus, wrist, and hip (proximal femur). Since spine fractures are difficult to identify in administrative databases,22 we were only able to ascertain fractures that necessitated medical attention. An incident fracture was identified if there was one hospitalization or two physician visits (within 3 months) with the relevant diagnosis code, provided that this was preceded by a 6-month period (washout) without any codes for the same diagnosis. To identify an incident wrist fracture, we also required a site-specific physician claim for reduction, fixation, or casting within 3 months. For an incident hip fracture, we required the presence of a hospitalization with a hip fracture diagnosis code as well as a physician claim for fracture reduction or fixation within 3 months of the index date. Additional nontrauma fractures of interest (collectively referred to as other osteoporotic) included ribs/sternum, shoulder and arm, pelvis, clavicle, patella, tibia/fibula, ankle, trunk not otherwise specified (NOS), and scapula, as well as forearm and femur fractures that did not meet the previous case definitions. We did not exclude pathologic fractures because they represent a small proportion of all fractures. Their exclusion could lead to underestimation of the fracture burden owing to osteoporosis.14, 23 Simultaneously occurring fractures at different sites were included in the analysis if they did not result from high trauma and were assigned randomly to a category from among the fracture sites sustained by the individual. A detailed list of the codes used for each type of fracture is available from the authors.
Three controls for each case were selected and matched on a fiscal-year basis by sex and birth date (within 5 years) and were assigned the same index dates as the matched cases. Controls also were required to have continuous health insurance coverage starting 5 years prior to the case's index fracture date and to be free of fracture diagnosis codes during that period. Control selection was done without replacement within a fiscal year and with replacement for each new fiscal year. The index date was taken to be the date assigned to the matched control.
Direct medical costs were tabulated according to the following general categories: prescription drugs dispensed through the provincial pharmacy system, physician billings, hospitalizations, nursing home admissions, and home care services. Standard (or average) costs are assigned for inpatient hospital care and day surgery, home care services, and personal care home residence. Microcosts (or user-specific costs) are assigned for physician services and prescription drugs. These health care costing methods are in accordance with national guidelines and are well established at MCHP for analyzing health care costs in Manitoba.24–27
The physician services database records the vast majority of encounters an individual has with a physician, both in the physician's office and in other settings such as hospitals and nursing homes. Claims for each service provided include a code used to determine the fee that is paid to the physician. Most physicians who are not paid on a fee-for-service basis submit “shadow claims,” and these were included in computing physician costs using the equivalent fee-for-service value. There is no reimbursement for vertebral augmentation following spinal fracture in Manitoba; therefore, this procedure is rarely performed and cannot be identified reliably. To determine the cost of physician services for an individual, we summed the fees paid to all physicians for that individual during the given period of time. Physician costs then were adjusted to 2006 dollars using the consumer price index for Manitoba health services from the Canadian Socio-Economic Information Management System (CANSIM).
Hospital costs were based on the provincial average direct cost per weighted case ($2,953.45 in 2006).26 Resource Intensity Weights (RIWs) are the relative case weights for case-mix groups (CMGs) used to measure the intensity of resource use (relative cost) associated with different diagnostic procedures, surgical procedures, and demographic characteristics of an individual.28, 29 CMGs are patient classification algorithms developed by the Canadian Institute for Health Information (CIHI), based on the most responsible diagnosis, that are used to group and describe types of inpatients discharged from acute-care hospitals in homogeneous groups. RIWs are assigned according to the CMG to which an individual is classified, as well as their age, health status, and discharge status and are based on microcosting. The Day Procedure Group (DPG) is a classification system for ambulatory care provided in hospitals, most commonly surgical procedures that can be performed without the need for an overnight stay. Patients are assigned a code according to the principal procedure recorded on the patient abstract. Patients within the same DPG category will be similar in terms of resource utilization and clinical episodes. An RIW is assigned to each DPG reflecting the average relative resource requirements for the type of procedure. Total direct hospital costs for each individual were calculated using the RIW and DPG weight employing the following formula: Cost = $2,953.45 × sum(RIW, DPG weight). Hospital costs do not include physician services, which are calculated separately, as described earlier. They also do not include costs that cannot be assigned directly to patient care (eg, hospital administration and physical plant, including electricity and heating) and also do not include any building capital costs (eg, mortgages or other capital-related costs). Hospital records do not include visits to ambulatory clinics, including emergency department care, and therefore will underestimate the total costs of hospitalization in some cases.
The prescription drug cost for an individual is the total expenditure for all dispensed outpatient prescriptions for the given period of time. Inpatient drug use is not captured in this system, but the total hospital costs include the cost of drugs administered while in hospital and comprise an unknown proportion of the total hospital-related costs. A similar situation arises in some nursing homes where drugs are obtained from a hospital pharmacy, but the great majority of nursing homes (73% to 81% in a given year) obtain their drugs through retail pharmacies. A limitation of the pharmacy database is the lack of records for nonprescription (over-the-counter) medication use. Drug costs then were adjusted to 2006 dollars using the consumer price index for Manitoba prescribed medications from the Canadian Socio-Economic Information Management System (CANSIM).
The average per diem for residence in a nursing home was calculated by dividing total expenditures for these facilities for a year by the total number of resident-days ($137.33 for 2006). The cost for an individual is determined by multiplying this mean cost by the number of days that person was a nursing home resident. A limitation of this approach is that it does not take into account the level of care required for a given individual. A similar approach was used for home care services, where the average per diem was calculated by dividing total expenditures for the home care program by the total number of days individuals had an open file ($29.15 for 2006). The cost for an individual is determined by multiplying this mean cost by the number of days the person had an open home care file. A limitation of the home care data is that they do not contain information about the frequency or intensity of home care services; that is, we do not know whether a person is receiving weekly homemaker services or daily nursing or rehabilitation care.
Frequency distributions were used to describe the ages of cases and controls for each fracture site. Incremental costs were calculated as the difference between costs during the year after the index date minus costs during the year prior to the index date; the median of these differences was computed. The median was used to conduct descriptive analyses because the distributions of costs were highly skewed, and the mean is sensitive to the presence of extreme observations in the tails of a distribution. Spearman rank correlation coefficients were used to examine the association between age and total incremental costs. Population total incremental costs were calculated by summing all costs for all incident fracture cases during the 10 years of fracture data.
The distribution of costs violated the assumption of normality, which underlies ordinary least-squares regression. In addition, many negative incremental costs were encountered, which precluded a simple log transformation. Therefore, quantile regression analysis was performed using simplex optimization with resampling to generate confidence limits. The 50th percentile (median), 25th percentile, and 75th percentile of total costs and total incremental costs were estimated for cases and controls. Cases and controls were matched for sex and age but differed in terms of their prefracture health status and health care utilization. Therefore, the covariates included in the quantile regression analyses were group (ie, case versus control), total costs in the prior year, number of ambulatory diagnostic groups (ADGs) as a measure of health status, and prior nursing home residence status. Separate models were fit to the data for males and females. ADGs represent 32 comorbidity clusters of ICD-9-CM diagnostic codes from the Johns Hopkins Ambulatory Care Group System.30 The number of ADGs was categorized as none (reference category), 1 to 2, 3 to 5, and 6 or more. The number of ADGs has been shown to be a strong predictor of osteoporotic fractures in a previous analysis.31 Prior nursing home status was a dichotomous variable. In a subsequent set of analyses, index year was added as a covariate. These analyses were used to estimate the annualized change in median incremental total costs over time. Statistical significance was assessed using α = 0.05. Descriptive and inferential analyses were performed using SAS (Version 9.1; SAS Institute, Cary, NC, USA).
Table 1 summarizes the study populations. Cases with incident fractures were well matched to the nonfracture controls. We identified 33,887 fracture cases (22,953 women and 10,934 men) during the study period, with 101,661 controls (68,859 women and 32,802 men). All fracture types were more common in women than in men, except for traumatic fracture, which showed a slight excess among men. A total of 16,966 (50.1%) cases were classified as major osteoporotic fracture cases, 15,997 (47.2%) were classified as other osteoporotic fractures, and 924 (2.7%) were traumatic fractures. Among the major osteoporotic fractures, fractures of the forearm were the most common (n = 6295), followed by hip (n = 5217), humerus (n = 3216), and spine (n = 2128). All hip fractures and traumatic fracture patients were hospitalized after the fracture. For other fracture types, this ranged from 31.9% (wrist fracture) to 69.8% (clinical spine fracture).
Table 1. Demographic Characteristics of Fracture Cases and Nonfracture Controls
Among fracture cases, the proportion living in nursing homes on the index date increased with older age for women (0.2% for ages 50 to 59, 34.6% for ages 90 + ) and men (0.3% for ages 50 to 59, 35.0% for ages 90 + ). Home care use also increased with older age for women (3.0% for ages 50 to 59, 25.4% for ages 90 + ) and men (2.0% for ages 50 to 59, 29.1% for ages 90 + ). Multiple comorbidities (ADG score of 6 or more) were more common in older women (29.0% for ages 50 to 59, 42.3% for ages 90 + ) and men (20.2% for ages 50 to 59, 46.9% for ages 90 + ). These characteristics showed little change over the time period of the study (data not shown).
Individual fracture costs
Median total (unadjusted) costs are summarized in Table 2 for the fracture cases and nonfracture controls. During the year prior to the index fracture date (prefracture), total costs and individual health care costs were consistently higher among fracture cases than their matched controls. The highest total prefracture ratios were for hip fractures (5.7-fold for women and 7.6-fold for men). The smallest differences between cases and controls in prefracture period costs were for wrist fractures and traumatic fractures (ratios 1.3 to 1.6). Total case/control cost ratios were higher in the year after fracture (postfracture) compared with the year before for all fracture definitions and both sexes, ranging from a minimum of 2.5 for women with wrist fractures to a maximum of 15.1 for men with hip fractures.
Table 2. Median Total Health Care Costs According to Fracture Type
Note: All costs are expressed in 2006 constant dollars (Canadian). Δ = median (post − pre); this number is not equivalent to (median post − median pre).
The median incremental health care costs (Δ = median of postfracture minus prefracture) increased substantially for all fracture definitions in both women and men. The highest median incremental costs were related to hip fractures ($16,171 in women and $13,111 for men). Spine fractures also were associated with important incremental health care costs ($8,345 in women and $6,267 in men). Traumatic fractures were slightly less costly ($5,062 in women and $5,484 in men). Wrist fractures were associated with the lowest median incremental health care costs ($663 in women and $764 in men). All fracture types showed increasing incremental costs in relation to older age (p < 0.001), as shown in Fig. 1.
Adjusted fracture costs
The results of the quantile regression analysis performed to adjust for baseline differences between fracture cases and nonfracture controls are summarized in Table 3. The findings generally were consistent with the unadjusted cost data. Based on the median (50th percentile), the most costly individual fracture was hip fracture (women $20,129, 95% CI $19,770–$20,488; men $19,330, 95% CI $18,362–$20,297). The next most costly fractures again were spine fracture, followed by traumatic fracture. The least costly individual fracture was wrist fracture (women $693, 95% CI $646–$741; men $1041, 95% CI $838–$1,243). Estimates of the 25th and 75th percentiles followed the same ranking.
Table 3. Adjusted Total Incremental Costs in Fracture Cases Versus Nonfracture Controls (95% CIs)
50th Percentile (median)
Note: Results from quantile regression adjusted for total costs in the prior year, number of ADGs, and prior nursing home residence status. All costs are expressed in 2006 constant dollars (Canadian).
Trends in fracture costs
Unadjusted total incremental costs expressed in constant 2006 Canadian dollars were studied in relation to fiscal year for fractures that occurred from 1997–1998 to 2006–2007 (Fig. 2), and the adjusted annualized change in median costs was computed using quantile regression (Table 4). Total median incremental costs showed a small but statistically significant increase over time for some fracture types. The largest increase in women was for hip fracture (median of $13 per year, 95% CI $6–$21, p < 0.001), and no change was seen for traumatic fractures. The largest increase in men was for humerus fracture (median of $11 per year, 95% CI $3–$19, p = 0.007), whereas no change in the median was seen for hip fractures, spine fractures, or traumatic fractures.
Table 4. Trends in Adjusted Median Costs for Fracture Cases From 1997–1998 to 2006–2007
Annual change in median (50th percentile) cost (95% CI). Results from quantile regression adjusted for group (case versus control), total costs in the prior year, number of ADGs, and prior nursing home residence status. All costs are expressed in 2006 constant dollars (Canadian).
Table 5 shows that global population costs related to all fracture types exceeded $245 million in women and $108 million in men during the 10 years. Those aged 80 to 89 years were responsible for the largest proportion of the global population costs: for women $108 million (44.0%) and for men $38 million (35.4%). Hip fractures were responsible for the largest proportion of the costs in the age groups 80 to 89 years and 90+ years, but the group of other fractures was more important prior to age 80 in both women and men. Trauma fractures were responsible for 9.0% of costs in women aged 50 to 59 years and 19.0% of costs for men aged 50 to 59 years, but this proportion declined for older subgroups.
Table 5. Contribution of Different Fracture Types to Population Total Incremental Costs by Age Subgroup
Note: All costs are expressed in 2006 constant dollars (Canadian). Figures in parentheses are percent of row total.
Figure 3 depicts the global impact of fracture-related costs during the study period for all ages combined. Major osteoporotic fractures were responsible for 59.5% of fracture-related costs in women (33.2% from hip fractures alone) and 53.9% in men (29.4% from hip fractures alone). However, other osteoporotic fractures also made a large contribution to global costs (38.4% in women and 39.0% in men).
Hospital costs were responsible for the majority of incremental health care costs in all situations (Table 6). For hip fractures, this reached 75.7% of total incremental costs for women and 89.3% for men. For hip fractures, the next most costly components of care were related to nursing homes (16.6% of total for women and 8.2% of total for men), with the remainder divided between home care and physician costs. In all situations, outpatient drug costs made up only a very small component of total incremental costs (<2%). Outpatient drug costs actually decreased for some fracture types (such as hip fractures) likely because medications that one normally would take (and pay for) while at home were administered in hospital or in nursing homes and therefore are not captured in the pharmacy database.
Table 6. Relative Contribution of Different Health Care Costs to Total Costs According to Fracture Type
Δ = median (post − pre); this number is not equivalent to (median post − median pre).
We have reported prefracture, postfracture, and incremental costs for all identifiable health care services incurred by women and men who sustained an incident fracture in Manitoba over a 10-year period. The greatest total median incremental costs were associated with hip fractures, followed by spine fractures, and the lowest costs were associated with wrist fractures. Median incremental costs for all fracture types increased in relation to older age. Hip fractures were the major source of costs in the very elderly (after age 80 years), but other osteoporotic fractures were more important prior to age 80. Costs showed a slight increase to no change depending on fracture sites over the 10 years used to define incident fractures (adjusted to 2006 Canadian dollars).
Although matching controlled for the effects of age and sex, it does not adjust for other factors that may affect health care costs, including comorbidities and baseline (prefracture) differences in health care utilization. The matched nonfracture cohort was drawn from the entire population and therefore would be expected to be generally healthier than the fracture cohort. This is supported by the finding that costs in the year prior to the fracture index date generally were lower among the controls than the cases for all nontraumatic fracture definitions. Examining incremental costs in the year after fracture compared with the year before fracture mitigates these factors because cases served as their own controls. In addition, we performed regression analysis to adjust for baseline differences in factors associated with osteoporosis and that are more frequent among fracture patients, thereby contributing to health care utilization and costs. Results from the regression analysis were similar to those obtained from the unadjusted incremental cost.
Our findings are broadly consistent with other fracture COI analyses. Studies have consistently shown hip fractures to be associated with the highest costs.10–12 A previous Canadian study of hip fractures in 504 patients aged 50 years and older that occurred in 1995–1996 estimated mean costs in the year after fracture to be $26,527 Canadian,9 similar to the current study. Spine fractures are difficult to study, and reports show more heterogeneity, with a wide range in cost estimates from no significant acute-care costs to US$14,977 in the first year after fracture for those aged 50 to 64 years.10, 12 Fractures other than hip and spine factures have been less thoroughly studied, but a large commercial insurance and Medicare claims–based analysis showed costs in the first year after fracture that were less than hip or spine fractures but still quite significant (US$9183 for ages 50 to 64 years, US$6,106 for ages 65 years and older).12 Because of their high prevalence, the authors found that the total first-year costs of nonhip, nonspine fractures was greater for those aged 50 to 64 years than for hip and spine fractures combined. Our study confirms this finding, with hip and spine fractures responsible for 18.5% of population costs in women aged 50 to 59 years and 24.9% for men aged 50 to 59 years, whereas other nontrauma fractures were responsible for 72.4% and 56.1%, respectively. Pike and colleagues11 found that nonhip, nonspine fractures were associated with greater aggregate incremental health care costs than hip fractures in the first year after fracture, of which work loss accounted for 29.5% of total costs per fracture employee, and such fractures also have reduced but substantial second-year costs.32 Pike and colleagues11 also noted that the second most expensive fractures (after hip fractures) were those involving multiple sites (incremental cost of US$9642), similar to our finding of high incremental costs in those defined as having traumatic fractures ($12,179 Canadian in women, $15,792 Canadian in men). Kilgore and colleagues13 looked at a wide range of fracture types between 1999 and 2005 in Medicare recipients aged 65 years and older. Except for hip fractures, costs attributable to the fracture were found to be a minority of the incremental postfracture costs. Our study parallels previous reports and also evaluates traumatic fractures and trends in costs over a long time span by type of health service. It is interesting that these costs have been increasing over time; the reason is unclear and requires confirmation and additional study.
Limitations of the study data sources are enumerated in the description of the methods. The apparent reduction in outpatient pharmacy drug costs observed after some fractures associated with hospitalization and nursing home transfer (eg, hip fractures) are expected. The cost of drugs administered while in hospital and in some nursing homes is part of the facility's global budget and therefore is included in the cost assigned for inpatient care. Drug costs for a minority of nursing home residents are not captured in this analysis. A standard costing approach was used for hospitalization, nursing home, and home care costs. A microcosting approach would be preferable in order to assign costs to individuals, but this is not feasible with our data sources. For hospital costs, we included only costs that can be related directly to an individual's care. Estimates of indirect hospital costs, including both building capital costs and administration costs, have not been developed for individual types of patients, so we were unable to include these in our analysis. Therefore, our estimates for fractures whose total costs are dominated by hospital stays (eg, hip fractures) may be best considered as lower-bound estimates. Clearly, not all incremental health care costs are related purely to fracture care and also reflect care required for comorbidities contributing to fracture risk, as well as unrelated conditions. This is supported by the effect of older age on increasing health care costs in the year prior to incident fracture. An incremental costing approach attenuates but does not eliminate this concern. Another limitation to this study is that our fracture cases are biased by fractures that come to medical attention. By definition, these will result in greater costs than fractures that are undiagnosed, which are also likely to be less severe. Spine fractures would be particularly susceptible to this spectrum bias. Our case definition for spine fractures is based on one hospitalization or two or more physician interactions in a 3-month period with a 6-month washout window in order to optimize diagnostic and temporal specificity for the acute event and minimize diagnostic contamination from prevalent fractures. Most spine fractures are not diagnosed clinically but still may have health consequences and economic implications.33 Misclassification of fractures as traumatic is possible because trauma codes may not be applied consistently. In particular, lack of a trauma code does not necessarily mean an absence of trauma, and coding is likely incomplete based on the proportion of trauma fractures reported in cohort studies.14 The overall proportion of costs attributable to trauma fractures therefore would be underestimated. Indirect costs (eg, lost wages and informal caregiver support) have not been included in this study because the focus is on the direct cost of care for the health care system. We also acknowledge that some of the costs in the year after fracture potentially could include costs related to a second fracture affecting either the same skeletal site (eg, a second vertebral fracture) or a second site (eg, hip fracture after a wrist fracture). Finally, there may be a slight underestimation in postfracture costs by not including costs beyond 1 year.34, 35
In conclusion, we have estimated direct health care costs in Canada associated with a variety of incident fracture definitions and shown large incremental costs (exceeding $245 million Canadian in women and $108 million Canadian in men for the study period) in overall health care utilization for the year after fracture compared with the year before fracture. Hip fractures were associated with the greatest incremental health care costs ($16,171 Canadian in women and $13,111 Canadian for men), whereas wrist fractures were associated with the lowest health care costs. Nontrauma fractures occurring at sites other than the major osteoporotic sites (ie, hip, spine, wrist, and humerus) were common and were responsible for over one-third of global fracture-related costs. This work provides evidence to support the suggestion that reducing the prevalence of osteoporosis-related fractures may translate into reduced costs across the system. Further research is needed to determine the impact that population-based changes may have on maintenance of optimal skeletal health and the cost of fracture care.
WDL received research grants from Merck Frosst, Amgen, and Genzyme and research honoraria and unrestricted educational grants from Sanofi-Aventis, Procter & Gamble, and Novartis and served on advisory boards for Genzyme, Novartis, and Amgen. CJM received a research grant from Amgen. LML received a research grant from Amgen. SNM served as a consultant to Procter & Gamble, Sanofi-Aventis, Amgen, Novartis, and Eli Lilly; served in the speakers' bureau for Procter & Gamble, Sanofi-Aventis, Amgen, and Novartis; and received a research grant from Amgen. PC received a research grant from Amgen. All the other authors state that they have no conflicts of interest.
We acknowledge the Manitoba Center for Health Policy for use of data contained in the Population Health Research Data Repository (HIPC Project No. 2008/2009–16). The results and conclusions are those of the authors, and no official endorsement by the Manitoba Center for Health Policy, Manitoba Health, or other data providers is intended or should be inferred. This work was funded through a research grant from Amgen Canada, Ltd. The funding source had no access to the data prior to publication, no input into the writing of the manuscript, and no input in the decision to publish the results.
Authors' roles: All authors contributed to the design of the study. MA, CJM, and WDL were responsible for the acquisition and initial analysis of the data. All authors participated in subsequent analyses and interpretation of the results. WDL drafted the initial manuscript, and all authors participated in the revisions and preparation of the final manuscript.