- Email: [email protected]
Projecting the direct cost burden of osteoarthritis in Canada using a microsimulation model
Osteoarthritis and Cartilage 23 (2015) 1654e1663
Projecting the direct cost burden of osteoarthritis in Canada using a microsimulation model B. Sharif y *, J. Kopec z, N. Bansback z, M.M. Rahman x k, W.M. Flanagan ¶, H. Wong z, P. Fines ¶, A. Anis z y Department of Community Health Sciences, University of Calgary, Calgary, Canada z School of Population and Public Health, University of British Columbia, Vancouver, Canada x Arthritis Research Centre of Canada, Richmond, BC, Canada k Department of Applied Statistics, East West University, Dhaka, Bangladesh ¶ Health Analysis Division, Statistics Canada, Ottawa, Canada
a r t i c l e i n f o
s u m m a r y
Article history: Received 29 December 2014 Accepted 26 May 2015
Objectives: To estimate the future direct cost of OA in Canada using a population-based health microsimulation model of osteoarthritis (POHEM-OA). Methods: We used administrative health data from the province of British Columbia (BC), Canada, a survey of a random sample of BC residents diagnosed with OA (Ministry of Health of BC data), Canadian Institute of Health Information (CIHI) cost data and literature estimates to populate a microsimulation model. Cost components associated with pharmacological and non-pharmacological treatments, total joint replacement (TJR) surgery, as well as use of hospital resources and management of complications arising from the treatment of osteoarthritis (OA) were included. Future costs were then simulated using the POHEM-OA model to construct profiles for each adult Canadian. Results: From 2010 to 2031, as the prevalence of OA is projected to increase from 13.8% to 18.6%, the total direct cost of OA is projected to increase from $2.9 billion to $7.6 billion, an almost 2.6-fold increase (in 2010 $CAD). From the highest to the lowest, the cost components that will constitute the total direct cost of OA in 2031 are hospitalization cost ($2.9 billion), outpatient services ($1.2 billion), alternative care and out-of-pocket cost categories ($1.2 billion), drugs ($1 billion), rehabilitation ($0.7 billion) and side-effect of drugs ($0.6 billion). Conclusions: Projecting the future trends in the cost of OA enables policy makers to anticipate the significant shifts in its distribution of burden in the future. © 2015 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.
Keywords: Cost of osteoarthritis Direct cost Simulation modeling Microsimulation
Introduction The direct costs associated with Osteoarthritis (OA) are significant compared to other musculoskeletal diseases due to high prevalence of disease and high use of associated resources1,2. A patient with OA, even in its early stages, consumes health care
* Address correspondence and reprint requests to: B. Sharif, Department of Community Health Sciences, Faculty of Medicine, University of Calgary, Room 3C64, Health Research Innovation Centre (HRIC), 3280 Hospital Drive NW, Calgary, AB T2N 4Z6, Canada. Tel.: 1-587-7038747; fax: 1-403-210-9574. E-mail addresses: [email protected] (B. Sharif), jkopec@ arthritisresearch.ca (J. Kopec), [email protected] (N. Bansback), rahman102@ gmail.com (M.M. Rahman), [email protected] (W.M. Flanagan), hubert. [email protected] (H. Wong), [email protected] (P. Fines), aslam.anis@ubc. ca (A. Anis).
resources at almost double the rate of the general population without OA3. At the same time, OA's burden is escalating at a fast rate due to the considerable rise in obesity and population aging4,5. Estimating the direct cost of OA and its future trends has posed a challenge to health economics researchers6,7. The majority of past national cost-of-illness studies (COI) have utilized top-down and prevalence-based methodologies that are defined according to health expenditure data and use of cross-sectional data1,7. These types of studies have been criticized because of inconsistencies in their estimates, their lack of flexibility to be used in economic evaluation studies, and their lack of ability to project cost in time and across sub-populations6,8. For example, direct cost estimates of OA in the US have been reported to have a 40-fold variation across different studies6, between 13 and 185 billion dollars (2010 $US)5,9. In Canada, the total direct cost of OA without out-of-pocket costs
http://dx.doi.org/10.1016/j.joca.2015.05.029 1063-4584/© 2015 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.
B. Sharif et al. / Osteoarthritis and Cartilage 23 (2015) 1654e1663
was estimated to be between 1.2 and 4.2 billion dollars (2010 $CAD) according to different studies6,10. Incident-based or ‘bottom-up’ studies can be more accurate but due to a lack of appropriate administrative data11, very few national incident based COI studies have been conducted1. Simulation modeling techniques and specifically microsimulation (MSM) can be used to project health indices across heterogeneous populations. Such models have been used in various chronic conditions12 such as osteoporosis13 and diabetes14. The capability of these types of models in projecting population demographics and disease risk factors in addition to their flexibility in evaluating the effect of policies on disease burden makes MSM an attractive epidemiological and economic modeling approach15. In this study, we use the Population Health Microsimulation Model of Osteoarthritis (POHEM-OA)16, a validated individual-level, continuous-time simulation model to perform a COI study for projecting the future direct costs of OA in Canada for the period of 2010e2031. Methods The POHEM-OA model is based on a validated individual-level simulation model developed by Statistics Canada for several chronic conditions17. These models utilize an array of large databases for demographic, immigration, mortality and morbidity statistics (Appendix A1). The POHEM-OA simulation model used in this study is a discrete-event simulation model that generates the probable time to OA-related events16. These continuous time variables are generated according to the appropriate hazard distributions for events such as OA diagnoses or hip/knee total joint replacement (TJR) surgery17. Fig. 1 shows the diagram of the POHEM-OA model.
1655
The POHEM-OA model uses the Canadian Community Health Survey (CCHS) in 2001 to sample individuals for its initial population18. The main risk factors for the OA diagnosis in POHEM-OA are age, sex and body mass index (BMI). The model simulates an individual's life history in terms of OA-related events through interrelated stochastic processes; at each calendar year, the events are assigned to individuals according to their characteristics, e.g., OA diagnosis is determined based on age, sex and BMI of each individual16 and BMI is modeled as an auto-regression model that includes age, sex, income, education, region, and previous BMI. The progression of OA in POHEM-OA is modeled in terms of patient flow in the healthcare system using BC administrative data. After an OA diagnosis, the first OA progression stage modeled is a visit to the orthopedic surgeon (OS). There may have already been several physician visits (GP, or specialist) by the time of the first OS visit, while most cases of OA never visit an OS. After the (OS) visit, patients may be hospitalized for a short stay and undergo an inpatient procedure or they may be assigned for total knee or hip replacement surgery (TJR). Further details of the POHEM-OA model are discussed elsewhere16. Description of the cost algorithm We developed the cost module of POHEM-OA that consists of input parameters representing the per-patient and per patient-year unit costs of the resources utilized by OA patients. In this cost module, each individual is assigned a cost profile that keeps track of the values of that individual's characteristics as they change throughout the simulation. To capture the heterogeneity among OA populations and account for the variation in their resource utilization, four types of cost factors have been defined so that
Fig. 1. POHEM-OA: A population-based microsimulation model of osteoarthritis *. The rectangle in the left represents the socio-demographic characteristics of individuals in the model. Cylinders inside the figure are data sets used for estimating model parameters; NPHS 1994e2004: National Population Health Survey, a longitudinal household survey by Statistics Canada (ref*) ** CCHS-2001: Canadian Community Health Survey (54); ** *BCLHD: British Columbia Population-based administrative data (same as PDBC) (84), PharmaNet is linked to BCLHD and has drugs-related data for all BC population; HRQL: Health-related Quality of life.
1656
B. Sharif et al. / Osteoarthritis and Cartilage 23 (2015) 1654e1663
whenever a value of a cost factor changes, the cost profile of a patient gets updated. Cost factors include age categories, sex, OA stage, and time in each stage. OA stage is defined by the OA-related events with relation to the patient's flow in the healthcare system: stage 1 between OA diagnosis and OS visit; stage 2 between OS visit and hip/knee TJR; stage 3 between primary hip/knee TJR and revision TJR; stage 4 after revision hip/knee TJR. The time in each of these disease stages is categorized as 0e2 years, 2e5 years and 5 years or more. Cost components The following cost components are included in our analysis: 1) physician visits and other outpatient procedures, 2) hospitalization and rehabilitation, 3) medications and 4) side-effects of OA drugs. Details of each cost components and their corresponding data sources are shown in Tables I and II. All costs were translated into 2010 $CAD using the overall Consumer Price Index (CPI) according to Statistics Canada rates19. The total cost of OA in future years was discounted back to 2010 Canadian dollars20. Data sources The main data source for estimating the input parameters of this study was the Population Data BC21e23, a pan-provincial population health data service that provides access to the health care services' databases for all BC residents registered with the province's publicly funded universal insurance program. This includes hospitalization episodes, physician visits, tests and other outpatient services and drugs.
We also used the British Columbia Ministry of Health's OA (MOH-OA) survey24 to estimate the input parameters for alternative care categories, including physiotherapy, chiropractic, and massage therapy (Table A14). Trends for the number of knee and hip TJRs and their cost estimates were derived from the Canadian Joint Replacement Registry (CJRR) data25 and Centre for Health Information (CIHI) hospitalization cost data in 201126, respectively. The St. Paul's Hospital cost model27 and the CIHI hospitalization cost database26 were used to calculate hospitalization cost for procedures other than hip and knee TJR. Price indexes for the average hospitalization cost reported by Cost per Weighted Case (CPWC) across Canada28, along with those for drugs29, physicians and orthopaedic surgeons30,31 for the years 2003e2010 were taken from the CIHI Price Index Reports, while the overall economy CPI was used from the Statistics Canada report19. The Clinical Research Ethics Board at the University of British Columbia provided ethics approval for this study. Derivation of cost parameters Physician, drugs and hospitalization (other than hip/knee TJR) We have used Medical Services Plan (MSP) database of Population Data BC from 1986/7 to 2003 to calculate the outpatient unit costs of OA patients21. We have included physician's visits and ambulatory services including x-rays, MRI, blood work, other diagnostic tests and other types of services. All physician visits for OA and OA-related inpatient procedures were included. For estimating the drug cost, all OA-related drugs were categorized into four types: acetaminophen, non-steroidal anti-inflammatory drugs (NSAIDs), coxibs and opioids. A stratified random sample of
Table I Direct cost categories of OA and inflation rates implemented in the POHEM-OA direct cost module for publically funded cost components Cost category
Cost component
Hospitalization and Hip/knee TJR surgery inpatient costs
Other in-patient procedures for OA (nonhip/knee TJR Surgery)
Physician and outpatient care costs Drugs
* y z x k ¶ # ** yy
Parameter
Data sources for cost estimation (year)
Hazard rates of surgery by age and sex Average unit cost of surgery
PDBC* (1987e2003)
Number of surgeries per year and future trend in surgery rates Average patient-year unit cost
CIHIeHPy (2010)
Cost component's inflation index (annual inflation rate %)
Data sources for inflation (year) [Reference]
1. Orthopedic surgeon fee-for 1. CIHI-PSFz (2005e2010) 2. Refer to [45] service price index (2.1%) 2. Surgical Implant inflation 3. CIHI-HPI (2007e2011)x rate (9.8%) 3. Hospital employee wage index (3.6 %)
CJRRk(1996e2010)
PDBC (2003)
Average procedure costs CIHI-HPI (2010) Number of procedures per year St. Paul's hospital¶ (2006) Physician visits and medical Average patient-year unit costs PDBC (2003) tests Prescription drugs
Average patient-year unit costs PDBC (2003)
Over-the-counter drugs
Average patient-year unit costs NPHSyy (2006)
Hospital employee wage index CIHI-HPI (2007e2011) (3.6%)
Physician fee-for service price index (3.2%)
NPDB- PPI (2003e2008)#
Prescription-based generic and CIHI-DER** (2009e2012) brand name drugs price index (0.9%) Over-the-counter generic and CIHI-DER (2009e2012) brand name drugs price index (1.6%)
PDBC: Population data BC from 1987-200328. CIHI-HP: Canadian Institute of Health Information hospitalization cost data in 201031. CIHI-PSF: CIHI report on physician and surgeon fees from 2005-201035. CIHI-HPI: CIHI report on hospital price index33 for full time hospital employee wage index from 2007 to 2011. CJRR: Canadian Joint Replacement Registry data for number of surgeries from 1996-201030. St. Paul's hospital unit cost model in 200632. NPDB-PPI: National physician data base reported by CIHI on physician fee-for-service price index from 2003-200836. CIHI-DER: CIHI-Drug Expenses Report34. NPHS: National Population Health Survey used for estimating over-the-counter drugs cost for OA (unpublished data).
B. Sharif et al. / Osteoarthritis and Cartilage 23 (2015) 1654e1663
1657
Table II Direct cost categories of OA and inflation rates implemented in the POHEM-OA direct cost module for rehabilitation and out-of-pocket cost components Cost category
Cost component
Parameter
Healthcare system paid Probability for home/hospital discharge strategy rehabilitation and home care costs Unit costs for each discharge strategy Out-of-pocket-cost Average patient-year unit costs during rehab Probability and unit costs Formal caregiver, Formal caregiver and other costs transportation and community costs Alternative care Physical therapy, Utilization rates costs Chiropractic's, Per-service Fees acupuncture and other complementary care services Side effect of drug's Cardiovascular disease Utilization of drug use Probability of side effects for (CVD's), Stroke, each type of drug Gastrointestinal (GI), Dyspepsia Life time cost for CVD and stroke, Incidence-based cost for GI and dyspepsia
Rehabilitation and home care costs (after hip/knee surgery)
* y z x k ¶
Data sources for the base-year cost estimation (year)
Cost component inflation index Data source for inflation (annual inflation rate %) (year) [Reference]
Oldmeadow et al. (2000)33
Hospital employee wage index (3.6 %) Consumer Price Index (CPI) for overall economy Consumer Price Index (CPI) for overall economy Consumer Price index (CPI) for overall economy
CIHI-HPI (2007e2011)*
MOH-OAz (2007) MSPx (2003)
e Physician fee-for service price index (3.2%)
e NPDB- PPI (2003 e2008)k
PDBC¶ (2003) Literature review (2010) [unpublished] Refer to45
e e
e e
Coyte et al. (2001)32 45
Refer to
Gupta et al. (2005)34
CANSIM (2005e2011)y CANSIM (2005e2011) CANSIM (2005e2011)
Consumer Price Index (CPI) for Consumer Price Index overall economy (CPI) for overall economy
CIHI-HPI: CIHI Report on Hospital Price Index33 for full time hospital employee wage index from 2007 to 2011. CPI for overall economy was calculated based on an exponential extrapolation from historical data (2005e2011) from Statistics Canada ‘s CANSIM CPI Report26. MOH-OA: British Columbia Ministry of Health's OA survey29. MSP fees in 2003: Medical Services Plan fees from PDBC28. NPDB-PPI: National physician data base reported by CIHI on physician fee-for-service price index from 2003-200835. PDBC: Population data BC from 1987-200328.
100,000 individuals was generated from PharmaNet database from Population Data BC23 to estimate input parameters associated with the per patient-year OA drug costs. We have included different OArelated drug types and assumed that all prescription-based drugs are included in the healthcare-system paid cost categories for OA (Table A5). For the over-the-counter drug costs, we used the utilization rates and costs for the over-the-counter drug categories from National Population Health survey for acetaminophen, some NSAIDS and some opioids (unpublished data). For hospital procedures (inpatient procedures) other than hip/knee surgeries, up to 25 diagnosis codes and a maximum of 12 procedure codes, along with the type of procedure (primary, revision or emergency), were found for each admission date on hospital discharge summaries22. Details of methods for estimating input parameters are provided in Appendix A1. Hip/knee TJR surgery costs A separate analysis was performed to estimate per-patient hip/ knee TJR costs. To this end, we needed to estimate the costs of both a knee/hip TJR and the number of TJR events in the simulation model. For the age-specific unit cost of a hip/knee TJR we used CIHI hospitalization data26. The overall cost of a TJR equals the weighted average of the three types of TJR as defined in CIHI hospitalization data i.e., unilateral/bilateral, hip/knee, and with/without infection. The procedure cost was calculated according to different cost components of the hip/knee TJR surgery. To estimate the future trend in terms of the number of TRJ surgeries, we performed a regression analysis and projected the total number of surgeries according to historical data from CJRR using a linear regression method. We used the number of surgeries estimated from the CJRR data for the years 2000e2010, extrapolated them into the future and then calibrated the model by adjusting the hazard ratio of hip/knee TJR surgery by age and sex until the results from the model agreed with the extrapolated CJRR data25. Details of method for projecting number of surgeries are provided in Appendix A2. The costs for hip/knee TJRs were also extrapolated for the next 20 years, separately for primary and
revision surgeries and for different age categories. There was a change in the cost of TJR surgery over time, due to both changes in the average length of stay (LOS) (which has decreased within the last decade) and price inflation for CPWC reported in the CIHI Hospital Price Index Report28. For this purpose, we projected the average LOS for knee and hip replacement according to historical data from the CJRR26 using an exponential model (Appendix A2). Alternative care costs A major component of the out-of-pocket cost of OA patients is associated with visits to alternative care professionals including physiotherapists, chiropractors and complementary care professionals24. To estimate and project these costs in POHEM-OA, we have used the MOH-OA dataset and developed a regression model to estimate the probability and annual number of visits to alternative care professionals according to patients' age, sex and OA stage (Appendix A1). The unit costs of visits to each alternative care provider were then extracted from MSP dataset in 2003 and the physician's price index rates were applied for projecting the rates from 2010 to 203130. Rehabilitation, out-of pocket, and side-effect of OA-related drugs After the hip/knee TJR surgery, patients undergo different rehabilitation routes depending on the disease severity and other patient characteristics. For this, we have developed a separate discharge destination model adapted from Coyte et al.32 and Oldmeadow et al.33. In addition to the health-system paid cost during and after the hip/knee TJR, patients need to pay out-of-pocket for some services including medical aid devices and formal caregiver costs. We have included those costs in the model according to the results from Gupta et al.34. The last cost component included in our model is the cost associated with the side-effect of OA drugs. We have included four types of side-effects of OA drugs including stroke, cardiovascular diseases (CVD), serious gastrointestinal (GI) complications, and dyspepsia. The lifetime cost of stroke and CVD and the unit cost for GI complications and dyspepsia were extracted from the literature. Details are provided in Appendix A1.
1658
B. Sharif et al. / Osteoarthritis and Cartilage 23 (2015) 1654e1663
Internal validation We have investigated the integrity of the results of our model through two types of internal validation techniques: 1) Assigning zero unit costs to physician, drugs, and hospital procedures (other than hip/knee TJR surgery) and 2) Setting the probability of eventbased cost categories to zero for hip/knee TJR surgery, rehabilitation, formal caregiver, side effect of drug costs and alternative care costs. Further details of the internal validation are discussed in Appendix A2. Results According to the POHEM model16, the size of the adult Canadian population 20 years of age and older is projected to increase from 26.3 million in 2010 to 32.2 million in 2031 (22.5% increase). During the same period, the OA population is forecast to rise from 3.6 million to almost 6 million (64% increase). OA prevalence for females is projected to increase from 0.16 to 0.21, while male prevalence is projected to increase from 0.11 to 0.16 throughout the study period. As shown in Fig. 2, we estimate that the total direct cost of OA will rise from $2.9 billion dollars in 2010 to $7.6 billion in 2031, an almost 2.6-fold increase in total OA-related direct costs. The increase in the total direct cost reflects increase in prevalence and utilization rates of the healthcare resources in addition to the inflation rates for all healthcare resources. According to our results, the extra direct cost associated solely with inflation in 2031 is projected to be $2.4 billion, of which $1.8 billion is related to health care system costs and $600 million to out-of-pocket costs. Total direct cost by sex and age category We estimate the direct costs of OA for females to be $0.71 billion and $1.85 billion higher than those for males in 2010 and 2031, respectively. As shown in Fig. 3, the share of female cost in both out-of-pocket and healthcare system costs was higher than that of males. The out-of-pocket cost for females and males associated with OA in 2010 was $0.52 billion and $0.14 billion, respectively, and increased to over $1.12 and $0.47 billion in 2031. Female OA patients between 60 and 70 years of age constituted the highest-cost group in 2010 (Fig. 4). However, the costs for
patients in other age categories and the demography of the Canadian population tend to change significantly over the future decades to a degree that, according to our results, OA patients aged between 70 and 80 years will constitute the highest-cost age category in 2031. For females, the change in the highest-cost age category occurs in 2020 and for males in 2026. As shown in Fig. 4, between 2025 and 2030, the costs for those aged 70e80 and 80e90 will increase at the highest rate compared to other age groups. Total direct cost by cost component Fig. 5 represents the out-of-pocket and healthcare system share of the total direct cost of OA and the cost components associated with each in 2010 and 2031. According to the POHEM-OA projections, from 2010 to 2031, the hospitalization costs of hip/knee TJR surgeries would increase from $720 million dollars to almost $2.8 billion dollars, while the overall costs for physicians, and prescription drugs would increase from about $773 million to $1.9 billion. At the same time, the cost of side effects from OA drugs rises from $316 million to $665 million and out-of-pocket costs of patients including alternative care, over the counter drugs and other out-of-pocket costs increase from $842 million to $1.92 billion during the same period (Fig. 5). While the dollar amounts of all the cost categories are projected to increase during the study period, each cost category's share of the total direct cost has shown a more complex trend, as shown in Fig. 6. According to our model, the share of the overall costs for physician, OTC drugs, and prescription drugs in addition to alternative care and formal caregiver costs is expected to decrease while the share of hospitalization costs is expected to increase. The share of rehabilitation and side effect costs is projected to remain stable since 2010. Discussion In this study, we used an array of population-based data sources, OA surveys and cost reports to estimate input parameters of a validated individual-level simulation model (POHEM-OA) and then project the costs of OA over the next 20 years from 2010 to 2031. According to the POHEM-OA's projection, the overall direct cost of OA will increase by almost 160%, from $2.9 billion ($842 million out-of-pocket cost and $2.06 healthcare system expenditures) to
Fig. 2. Sex-specific and total direct cost of OA from 2010 to 2031. *All costs are discounted back to 2010 $CAD using discount rates and are presented in billion dollars.
B. Sharif et al. / Osteoarthritis and Cartilage 23 (2015) 1654e1663
1659
Fig. 3. Sex-specific healthcare system and out-of-pocket cost of OA. * All costs are in 2010 $CAD billion dollars.
Fig. 4. Total direct cost of OA by age category for males and females*. * Costs are in 2010 $CAD; age categories are as the followings: 0 to 49, 50 to 59, 60 to 69, 70 to 79, 80 to 89 and 90þ.
$7.6 billion ($1.92 billion out-of-pocket and $5.7 billion healthcare system expenditures) in 2010 $CAD. The increase in total cost of OA is due to increase in prevalence and the average cost of the disease. According to our results, the average cost of OA increased from $805 in 2010 ($234 out-of-pocket and $571 healthcare system expenditures) to $1266 in 2031 ($320 out-of-pocket and $950 healthcare system expenditures) in 2010 $CAD. At the same time, the total number of people with OA increased from 3.6 million to almost 6 million during the study period. The average cost of OA has shown to be moderate, if not high, compared to other chronic conditions1. For example, based on our results, the average healthcare system expenditures of OA ($571) is relatively lower compared to other chronic diseases such as CVD ($801) and cancer ($715) as estimated using administrative data in Canada in $201035. On the other hand, prevalence of OA is
high and it is increasing with a fast rate. Consequently, the combination of moderate average cost and high and growing prevalence, makes OA an important health policy concern1. We also described how the proportion of total cost of OA would change across different cost components during the study period. The share of hospitalization cost was the highest among all cost components and increased from 25% to 37%, while physician visit, drug and out-of-pocket cost constituted the next highest proportion, respectively. Almost 95% of the hospitalization cost was due to hip and knee TJR surgeries and 5% due to other hospital procedures for OA patients. The increase in the cost of hip/knee TJR surgeries was not only due to increase in the number of cases, but also due to the high average cost because of high inflation rate for procedure and prosthesis cost specific to this type of surgery. According to our results, although the LOS after hip and knee TJRs will be decreasing
1660
B. Sharif et al. / Osteoarthritis and Cartilage 23 (2015) 1654e1663
Fig. 5. Total direct cost of OA by cost categories in 2010 and 2031: out-of-pocket and healthcare system's costs. *Out-of-pocket share of the rehabilitation cost includes rehabilitation and home care cost that is paid by patients and includes formal caregiver, home remodelling due to the disease and transportation cost during the year after the hip/ knee TJR surgery39; ** Healthcare system's share of the rehabilitation and home care cost that is paid by the government after the hip/knee TJR.; ^ Other out-of-pocket costs includes formal caregiver cost paid by the patients and community care services such as meals on wheels, transportation and home remodelling that are all paid before the hip/knee TJR surgery.
Fig. 6. Cost components' share of the total direct cost from 2010 to 2031. *Hospitalization costs include those related to hip/knee total joint replacements (TJR's) and other OArelated hospital procedures such as arthroscopy, shoulder TJR; ** Alternative care costs include those related to physiotherapy visits, chiropractor visits, and other alternative care professionals; Formal caregiver cost includes paid caregiver cost in addition to community services cost such as home re-modelling due to the disease and transportation as listed in Gupta et al.39.
from 2010 to 2031, the average cost of TJR would still be increasing mostly due to its very high inflation rate. The increasing number of TJR's is due to an increase in prevalence of OA and also due to policy decisions to increase capacity for performing these types of surgeries36. While TJR has shown to be highly cost-effective for those with severe OA37, recent studies have shown an increasing rate of TJR's for those in less advanced stages who may not benefit from the procedure due to higher failure rates and lower satisfaction38. Future studies using population health models such as POHEM-OA need to assess the budgetary implications of different appropriateness criteria to limit the TJR for those who do not benefit39, while provide means to encourage those who benefit the most from this procedure40. With respect to age groups, our model suggests that those with 70e80 years of age will constitute the highest cost population in 2031. This underlines an important implication for evaluating
policies for OA care and management targeted at younger age groups (e.g., 55e65 years in 2015) with mild OA in order to reduce the burden among elderly with severe OA in the future (e.g., 70e80 in 2031)41. This is highlighted by the OARSI conservative OA care and management guidelines for use of none-surgical care such as weight reduction and muscle strengthening exercises in early and mild stages of OA42. Our study was not free of limitations. On top of the health-caresector inflation, the utilization rates for hospitalization, drugs, and physicians have been also increasing during the past decade43. In this study, we only implemented the increase in hospitalization rates for the hip/knee TJR surgeries that was due to increase in both incidence and policy-related decisions to increase the number of surgeries. Other cost drivers, such as changes in the population age and BMI distribution or increases in life expectancy of the population, were also implemented in the POHEM-OA model16.
B. Sharif et al. / Osteoarthritis and Cartilage 23 (2015) 1654e1663
However, we did not include potential per-capita increases in drugs and physician utilization in this study. While physician utilization per capita (measured as total physician expenditure in constant prices) have increased by around 25% from 1998 to 200843, the increase in analgesics drugs was reported to be small in the same interval43. In terms of the drugs included in our model, we did not include newer analgesics, e.g., duloxetine, and intra-articular injections of hyaluronic acid. We also did not include any drug without conclusive evidence of effectiveness for general OA population, e.g., glucosamine44, and possible future drugs including disease-modifying OA drug (DMOAD's). Future studies using simulation models such as POHEM-OA can be used to address the cost-effectiveness of such treatments45. Additional OA-related cost components not included are nursing homes research costs and rare side-effects such as chronic kidney failure, hepatic failure, or addiction46. One of the strength of our COI approach was use of a comprehensive cost listings of all resources used in addition to consideration of heterogeneity of the population that reflected the intensity of the resource used8. Traditional COI studies have been the target of criticism since they use single data sources and they are not capable of projecting the cost into the future8. According to the Canadian Health Expenditure reports in 1994, the total direct OA cost in Canada was projected to reach around $1.1e$1.6 billion dollars in 201047, which is lower than our estimates. This is due to the fact that none of the previous cost projections of OA included both the BMI and demographic shifts among the Canadian population16. Our result was close to another previous Canadian study using a population database (rather than sample surveys) for the overall direct cost of OA in 2006 ($2.1 billion in $2010 CAD)10. However, it should be noted that none of the previous studies included out-of-pocket costs for patients with OA. While recent COI studies have used new techniques to estimate incremental costs by performing regression-based analysis to control for relevant comorbidities, there are limitations in these methodologies that have caused large variations in the estimates of total and average cost of OA8. For example, while several studies, consistent with our results, have reported hospitalization costs as the most expensive category of the total medical cost of OA5,6,10, a Canadian study reported average cost of physician and outpatient procedures to be slightly higher than the cost of hospitalization11. Our results for the distribution of total cost of OA across different cost categories, however, conforms to the recent study by Losina et al.36, describing lifetime costs of knee OA in US. Losina et al.36 have shown that surgery and post-surgical cost is within 33e67% of all the lifetime direct cost of knee OA across different eligibility criteria for surgery defined based on radiographic stages. Estimating and projecting the cost of OA has been a methodological challenge in health economics literature when trying to account for the complex interrelation of time varying risk factors such as demographics and BMI shifts among the population48,49. In this study, we performed a COI study with the use of MSM modeling to project changes in the distribution of direct cost of OA across different sub-populations and cost components. Constant escalation of OA costs leads to growing demands for more money to be devoted to arthritis care specifically among high cost subpopulations including females and those over the age of 70 as identified in this study. Given the magnitude of OA costs, small decrease in risk factors or cost would help attenuate rising healthcare expenditures in the near future. Author contributions Conceived and designed the analysis: BS JK NB MR WMF PF HW AA. Performed.
1661
Simulation and analysis: BS WMF PF. Analyzed the data: BS MR NB. Wrote the paper: BS JK NB MR WMF PF HW AA. Approved the final draft: BS JK NB MR WMF PF HW AA. Conflict of interest No financial support or other benefits from commercial sources were received for the work reported in this manuscript. None of the authors has financial interests for potential conflict of interest or the appearance of a conflict of interest with regard to the work. Acknowledgements Authors would like to thank Canadian Arthritis Network (CAN), the Canadian Institutes of Health Research (CIHR), FRN# 92246, and Dr Linda Li for their help and support. The views expressed in this paper are solely those of the authors and do not reflect those of Statistics Canada. All inferences, opinions, and conclusions drawn in this manuscript are those of the authors, and do not reflect the opinions or policies of the Data Steward(s). Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.joca.2015.05.029. References 1. Yelin E. The economics of osteoarthritis. In: Brandt K, Doherty M, Lohmander LS, Eds. Osteoarthritis. New York: Oxford University Press; 1998:23e30. 2. Bitton R. The economic burden of osteoarthritis. Am J Manag Care 2009 Sep;15(8 Suppl):S230e5. 3. Dibonaventura MD, Gupta S, McDonald M, Sadosky A, Pettitt D, Silverman S. Impact of self-rated osteoarthritis severity in an employed population: cross-sectional analysis of data from the national health and wellness survey. Health Qual Life Outcomes 2012 Mar:10e30. 4. Murray CJ, Vos T, Lozano R, Naghavi M, Flaxman AD, Michaud C, et al. Disability-adjusted life years (DALYs) for 291 diseases and injuries in 21 regions, 1990-2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 2012;380(9859):2197e223. 5. Yelin E, Murphy L, Cisternas MG, Foreman AJ, Pasta DJHC. Medical care expenditures and earnings losses among persons with arthritis and other rheumatic conditions in 2003, and comparisons with 1997. Arthritis Rheum 2007;56(5): 1397e407. 6. Maetzel A, Li LC, Pencharz J. The economic burden associated with osteoarthritis, rheumatoid arthritis, and hypertension: a comparative study. Ann Rheum Dis 2004;63:395e401. 7. Xie F, Thumboo J, Li S-C. True difference or something else? Problems in cost of osteoarthritis studies. Semin Arthritis Rheum 2007;37(2):127e32. 8. Akobundu E, Ju J, Blatt L, Mullins CD. Cost-of-illness studies: a review of current methods. Pharmacoeconomics 2006;24(9): 869e90. 9. Murphy LB, Helmick CG, Cisternas MG, Yellin EH. Estimating medical costs attributable to osteoarthritis in the US population: comment on the article by Kotlarz et al. Arthritis Rheum 2010;62(8):2566e7. 10. Anis AH, Zhang W, Bansback N, Guh DP, Amarsi Z, Birmingham CL. Obesity and overweight in Canada: an updated cost-of-illness study. Obes Rev 2010;11(1):31e40.
1662
B. Sharif et al. / Osteoarthritis and Cartilage 23 (2015) 1654e1663
11. Tarride JE, Haq M, O'Reilly DJ, Bowen JM, Xie F, Dolovich L, et al. The excess burden of osteoarthritis in the province of Ontario. Can Arthritis Rheum 2012;64(4):1153e61. 12. Rutter CM, Zaslavsky AM, Feuer EJ. Dynamic microsimulation models for health outcomes: a review. Med Decis Mak 2011;31(1):10e8. re O, Richy F, Gathon HJ, 13. Hiligsmann M, Ethgen O, Bruye Reginster JY. Development and validation of a Markov microsimulation model for the economic evaluation of treatments in osteoporosis. Value Health 2009;12(5):678e96. 14. The Mount Hood 4 Modeling Group. Computer modeling of diabetes and its complications: a report on the Fourth Mount Hood Challenge Meeting. Diabetes Care 2007;30(6):1638e46. 15. Vanness DJ, Tosteson AN, Gabriel SE, Melton 3rd LJ. The need for microsimulation to evaluate osteoporosis interventions. Osteoporos Int 2005 Apr;16(4):353e8. 16. Kopec JA, Sayre EC, Flanagan WM, Fines P, Cibere J, Rahman MM, et al. Development of a population-based microsimulation model of osteoarthritis in Canada. Osteoarthritis Cartilage 2010 Mar;18(3):303e11. 17. Zucchelli E, Jones AM, Rice N. The evaluation of health policies through microsimulation methods. HEDG working paper 10/ 03. University of York [cited 2015 Feb 01]. Available from:, http://www.york.ac.uk/media/economics/documents/herc/ wp/10_03.pdf; 2010. 18. Statistics Canada. Canadian community health survey (CCHS). Microdata File User Guide. Ottawa, Ontario: Health Statistics Division, Statistics Canada [cited 2015 April 10]. Available from:, http://www.statcan.gc.ca/eng/survey/household/3226; 2001. 19. Statistics Canada. The consumer price index. Ottawa, Canada [cited 2015 April 10]; Available from: http://www.statcan.gc. ca/tables-tableaux/sum-som/l01/cst01/econ46a-eng.htm 20. Boardman AE, Moore AM, Vining AR. The social discount rate for Canada based on future growth in consumption. Can Public Policy 2010;36(3):325e43. 21. British Columbia Ministry of Health [creator]. Medical services plan (MSP) payment information file. V2. Population Data BC [publisher]. Data Extract. MOH (2011) [cited 2015 April 10]. Available from: http://www.popdata.bc.ca/data; 2011. 22. British Columbia Ministry of Health [creator]. Discha Rge abstract database (Hospital separations). V2. Population Data BC [publisher]. Data Extract. MOH (2011) [cited 2015 April 10]. Available from: http://www.popdata.bc.ca/data; 2011. 23. British Columbia Ministry of Health [creator]. Data extract. Data Stewardship Committee (2011). PharmaNet. V2. BC Ministry of Health [publisher] [cited 2015 April 10]. Available from: http://www.popdata.bc.ca/data; 2011. 24. British Columbia Osteoarthritis Survey, Vancouver, Canada 2008 [cited 2015 Jan 10]. Available from: http://arthritis.rehab. med.ubc.ca/files/2011/08/BCOASurvey.pdf. 25. Canadian Institute of health Information (CIHI). Canadian joint replacement registry (CJRR). CIHI [cited 2015 Jan 10]. Available from: http://www.cihi.ca/CIHI-ext-portal/xlsx/internet/stats_ cjrr2011_clinicaldata_en; 2011. 26. Canadian Institute of health Information (CIHI). CMG client tables 2011. DAD Acute inpatient data [cited 2015 Jan 10]. Available from: https://secure.cihi.ca/estore/productFamily. htm?pf¼PFC1627&lang¼en&media¼0; 2012. 27. Borsa J, Anis A. The cost of hospital care in Canada: a comparison of two alternatives. Healthc Manage Forum 2005;18(1):19e27. 28. Canadian Institute for Health Information. Hospital cost drivers technical report; what factors have determined hospital expenditure trends in Canada?. Ottawa, Canada [cited
29.
30.
31.
32.
33.
34.
35.
36.
37. 38.
39.
40.
41.
42.
43.
2015 Jan 10]. Available from:, http://www.cihi.ca/cihi-extportal/pdf/internet/hospital_costdriver_tech_en; 2010. Canadian Institute for Health Information. Drivers of prescription drug spending in Canada. Ottawa, Canada [cited 2015 Jan 10]. Available from:, http://www.cihi.ca/CIHI-ext-portal/ pdf/internet/DRUG_SPEND_DRIVERS_EN; 2012. Canadian Institute for Health information. Health care cost drivers: Physician Expenditure d Technical report spending and health workforce, Ottawa, Canada, 2009 [cited 2015 Jan 10]. Available from: http://www.cihi.ca/cihi-ext-portal/pdf/ internet/health_costdriver_phys_tech_en. Canadian Institute for Health Information. Average payment per physician report, Fee-for-service physicians in Canada 2004e2005. Ottawa, Canada: National Physician Database [cited 2015 Jan 10]. Available from: https://secure.cihi.ca/free_ products/app_average_payment_per_physician_report_2004_ e.pdf; 2006. Coyte PC, Young W, Croxford R. Costs and outcomes associated with alternative discharge strategies following joint replacement surgery: analysis of an observational study using a propensity score. J Health Econ 2000;19(6):907e29. Oldmeadow LB, Mcburney H, Robertson VJ. Predicting risk of extended inpatient rehabilitation after hip or knee arthroplasty. J Arthroplasty 2003;18(6):775e9. Gupta S, Hawker GA, Laporte A, Croxford R, Coyte PC. The economic burden of disabling hip and knee osteoarthritis (OA) from the perspective of individuals living with this condition. Rheumatol (Oxford) 2005;44(12):1531e7. Patra J, Popova S, Rehm J, Bondy S, Flint R, Giesbrecht N. Economic cost of chronic disease in Canada 1995e2003 prepared for the Ontario Chronic Disease Prevention Alliance and the Ontario Public Health Association [cited 2015 March 10]. Available from: http://www.ocdpa.on.ca/sites/default/files/ publications/OCDPA_EconomicCosts.pdf; March 2007. Losina E, Paltiel AD, Weinstein AM, Yelin E, Hunter DJ, Chen SP, et al. Lifetime medical costs of knee osteoarthritis management in the United States: impact of extending indications for total knee arthroplasty. Arthritis Care Res(Hoboken) 2015;67(2):203e15. Katz JN. Total joint replacement in osteoarthritis. Best Pract Res Clin Rheumatol 2006;20(1):145e53. Ghomrawi HM, Alexiades M, Pavlov H, Nam D, Endo Y, Mandl LA, et al. Evaluation of two appropriateness criteria for total knee replacement. Arthritis Care Res (Hoboken) 2014;66(11):1749e53. Riddle DL, Jiranek WA, Hayes CW. Use of a validated algorithm to judge the appropriateness of total knee arthroplasty in the United States: a multicenter longitudinal cohort study. Arthritis Rheumatol 2014;66(8):2134e43. Hawker GA, Wright JG, Glazier RH, Coyte PC, Harvey B, Williams JI, et al. The effect of education and income on need and willingness to undergo total joint arthroplasty. Arthritis Rheumatol 2002;46(12):3331e9. Gruber WH, Hunter DJ. Transforming osteoarthritis care in an era of health care reform. Clin Geriatr Med 2010;26(3): 433e44. Zhang W, Moskowitz RW, Nuki G, Abramson S, Altman RD, Arden N, et al. OARSI recommendations for the management of hip and knee osteoarthritis, Part II: OARSI evidence-based, expert consensus guidelines. Osteoarthritis Cartilage 2008;16(2):137e62. Canadian Institute of Health Information (CIHI). Health care cost drivers: The facts. Ottawa, Canada [cited 2015 March 10]. Available from:, https://secure.cihi.ca/free_products/health_ care_cost_drivers_the_facts_en.pdf; October 2011.
B. Sharif et al. / Osteoarthritis and Cartilage 23 (2015) 1654e1663
44. Clegg DO, Reda DJ, Harris CL, Clegg DO, Reda DJ, Harris CL, et al. Glucosamine, chondroitin sulfate, and the two in combination for painful knee osteoarthritis. N Engl J Med 2006;354(8): 795e8. 45. Losina E, Daigle ME, Suter LG, Hunter DJ, Solomon DH, Walensky RP, et al. Disease-modifying drugs for knee osteoarthritis: can they be cost-effective? Osteoarthritis Cartilage 2013;21(5):655e67. 46. Zhang W, Moskowitz RW, Nuki G, Abramson S, Altman RD, Arden N, et al. OARSI recommendations for the management of hip and knee osteoarthritis, Part I: critical appraisal of existing treatment guidelines and systematic review of current research evidence. Osteoarthritis Cartilage 2007 Sep;15(9): 981e1000.
1663
47. Patra J, Popova S, Rehm J, Bondy S, Flint R, Giesbrecht N. Economic cost of chronic disease in Canada 1995e2003. Health Canada1e37 [cited 2015 April 01]. Available from: http://ocdpa.ca/sites/default/files/publications/OCDPA_ EconomicCosts.pdf; 2007(March). 48. Statistics Canada. The Life paths model [cited 2015 March 10]. Available from: http://www.statcan.gc.ca/eng/ microsimulation/lifepaths/overview/. 49. Direct cost of osteoarthritis in Canada : an Application of Microsimulation Modeling with Uncertainty Analysis- Sharif Behnam, University of British Columbia, PhD thesis, [cited 2015 April 01]. Available from: https://circle.ubc.ca/handle/ 2429/46614.
Projecting the direct cost burden of osteoarthritis in Canada using a microsimulation model B. Sharif y *, J. Kopec z, N. Bansback z, M.M. Rahman x k, W.M. Flanagan ¶, H. Wong z, P. Fines ¶, A. Anis z y Department of Community Health Sciences, University of Calgary, Calgary, Canada z School of Population and Public Health, University of British Columbia, Vancouver, Canada x Arthritis Research Centre of Canada, Richmond, BC, Canada k Department of Applied Statistics, East West University, Dhaka, Bangladesh ¶ Health Analysis Division, Statistics Canada, Ottawa, Canada
a r t i c l e i n f o
s u m m a r y
Article history: Received 29 December 2014 Accepted 26 May 2015
Objectives: To estimate the future direct cost of OA in Canada using a population-based health microsimulation model of osteoarthritis (POHEM-OA). Methods: We used administrative health data from the province of British Columbia (BC), Canada, a survey of a random sample of BC residents diagnosed with OA (Ministry of Health of BC data), Canadian Institute of Health Information (CIHI) cost data and literature estimates to populate a microsimulation model. Cost components associated with pharmacological and non-pharmacological treatments, total joint replacement (TJR) surgery, as well as use of hospital resources and management of complications arising from the treatment of osteoarthritis (OA) were included. Future costs were then simulated using the POHEM-OA model to construct profiles for each adult Canadian. Results: From 2010 to 2031, as the prevalence of OA is projected to increase from 13.8% to 18.6%, the total direct cost of OA is projected to increase from $2.9 billion to $7.6 billion, an almost 2.6-fold increase (in 2010 $CAD). From the highest to the lowest, the cost components that will constitute the total direct cost of OA in 2031 are hospitalization cost ($2.9 billion), outpatient services ($1.2 billion), alternative care and out-of-pocket cost categories ($1.2 billion), drugs ($1 billion), rehabilitation ($0.7 billion) and side-effect of drugs ($0.6 billion). Conclusions: Projecting the future trends in the cost of OA enables policy makers to anticipate the significant shifts in its distribution of burden in the future. © 2015 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.
Keywords: Cost of osteoarthritis Direct cost Simulation modeling Microsimulation
Introduction The direct costs associated with Osteoarthritis (OA) are significant compared to other musculoskeletal diseases due to high prevalence of disease and high use of associated resources1,2. A patient with OA, even in its early stages, consumes health care
* Address correspondence and reprint requests to: B. Sharif, Department of Community Health Sciences, Faculty of Medicine, University of Calgary, Room 3C64, Health Research Innovation Centre (HRIC), 3280 Hospital Drive NW, Calgary, AB T2N 4Z6, Canada. Tel.: 1-587-7038747; fax: 1-403-210-9574. E-mail addresses: [email protected] (B. Sharif), jkopec@ arthritisresearch.ca (J. Kopec), [email protected] (N. Bansback), rahman102@ gmail.com (M.M. Rahman), [email protected] (W.M. Flanagan), hubert. [email protected] (H. Wong), [email protected] (P. Fines), aslam.anis@ubc. ca (A. Anis).
resources at almost double the rate of the general population without OA3. At the same time, OA's burden is escalating at a fast rate due to the considerable rise in obesity and population aging4,5. Estimating the direct cost of OA and its future trends has posed a challenge to health economics researchers6,7. The majority of past national cost-of-illness studies (COI) have utilized top-down and prevalence-based methodologies that are defined according to health expenditure data and use of cross-sectional data1,7. These types of studies have been criticized because of inconsistencies in their estimates, their lack of flexibility to be used in economic evaluation studies, and their lack of ability to project cost in time and across sub-populations6,8. For example, direct cost estimates of OA in the US have been reported to have a 40-fold variation across different studies6, between 13 and 185 billion dollars (2010 $US)5,9. In Canada, the total direct cost of OA without out-of-pocket costs
http://dx.doi.org/10.1016/j.joca.2015.05.029 1063-4584/© 2015 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.
B. Sharif et al. / Osteoarthritis and Cartilage 23 (2015) 1654e1663
was estimated to be between 1.2 and 4.2 billion dollars (2010 $CAD) according to different studies6,10. Incident-based or ‘bottom-up’ studies can be more accurate but due to a lack of appropriate administrative data11, very few national incident based COI studies have been conducted1. Simulation modeling techniques and specifically microsimulation (MSM) can be used to project health indices across heterogeneous populations. Such models have been used in various chronic conditions12 such as osteoporosis13 and diabetes14. The capability of these types of models in projecting population demographics and disease risk factors in addition to their flexibility in evaluating the effect of policies on disease burden makes MSM an attractive epidemiological and economic modeling approach15. In this study, we use the Population Health Microsimulation Model of Osteoarthritis (POHEM-OA)16, a validated individual-level, continuous-time simulation model to perform a COI study for projecting the future direct costs of OA in Canada for the period of 2010e2031. Methods The POHEM-OA model is based on a validated individual-level simulation model developed by Statistics Canada for several chronic conditions17. These models utilize an array of large databases for demographic, immigration, mortality and morbidity statistics (Appendix A1). The POHEM-OA simulation model used in this study is a discrete-event simulation model that generates the probable time to OA-related events16. These continuous time variables are generated according to the appropriate hazard distributions for events such as OA diagnoses or hip/knee total joint replacement (TJR) surgery17. Fig. 1 shows the diagram of the POHEM-OA model.
1655
The POHEM-OA model uses the Canadian Community Health Survey (CCHS) in 2001 to sample individuals for its initial population18. The main risk factors for the OA diagnosis in POHEM-OA are age, sex and body mass index (BMI). The model simulates an individual's life history in terms of OA-related events through interrelated stochastic processes; at each calendar year, the events are assigned to individuals according to their characteristics, e.g., OA diagnosis is determined based on age, sex and BMI of each individual16 and BMI is modeled as an auto-regression model that includes age, sex, income, education, region, and previous BMI. The progression of OA in POHEM-OA is modeled in terms of patient flow in the healthcare system using BC administrative data. After an OA diagnosis, the first OA progression stage modeled is a visit to the orthopedic surgeon (OS). There may have already been several physician visits (GP, or specialist) by the time of the first OS visit, while most cases of OA never visit an OS. After the (OS) visit, patients may be hospitalized for a short stay and undergo an inpatient procedure or they may be assigned for total knee or hip replacement surgery (TJR). Further details of the POHEM-OA model are discussed elsewhere16. Description of the cost algorithm We developed the cost module of POHEM-OA that consists of input parameters representing the per-patient and per patient-year unit costs of the resources utilized by OA patients. In this cost module, each individual is assigned a cost profile that keeps track of the values of that individual's characteristics as they change throughout the simulation. To capture the heterogeneity among OA populations and account for the variation in their resource utilization, four types of cost factors have been defined so that
Fig. 1. POHEM-OA: A population-based microsimulation model of osteoarthritis *. The rectangle in the left represents the socio-demographic characteristics of individuals in the model. Cylinders inside the figure are data sets used for estimating model parameters; NPHS 1994e2004: National Population Health Survey, a longitudinal household survey by Statistics Canada (ref*) ** CCHS-2001: Canadian Community Health Survey (54); ** *BCLHD: British Columbia Population-based administrative data (same as PDBC) (84), PharmaNet is linked to BCLHD and has drugs-related data for all BC population; HRQL: Health-related Quality of life.
1656
B. Sharif et al. / Osteoarthritis and Cartilage 23 (2015) 1654e1663
whenever a value of a cost factor changes, the cost profile of a patient gets updated. Cost factors include age categories, sex, OA stage, and time in each stage. OA stage is defined by the OA-related events with relation to the patient's flow in the healthcare system: stage 1 between OA diagnosis and OS visit; stage 2 between OS visit and hip/knee TJR; stage 3 between primary hip/knee TJR and revision TJR; stage 4 after revision hip/knee TJR. The time in each of these disease stages is categorized as 0e2 years, 2e5 years and 5 years or more. Cost components The following cost components are included in our analysis: 1) physician visits and other outpatient procedures, 2) hospitalization and rehabilitation, 3) medications and 4) side-effects of OA drugs. Details of each cost components and their corresponding data sources are shown in Tables I and II. All costs were translated into 2010 $CAD using the overall Consumer Price Index (CPI) according to Statistics Canada rates19. The total cost of OA in future years was discounted back to 2010 Canadian dollars20. Data sources The main data source for estimating the input parameters of this study was the Population Data BC21e23, a pan-provincial population health data service that provides access to the health care services' databases for all BC residents registered with the province's publicly funded universal insurance program. This includes hospitalization episodes, physician visits, tests and other outpatient services and drugs.
We also used the British Columbia Ministry of Health's OA (MOH-OA) survey24 to estimate the input parameters for alternative care categories, including physiotherapy, chiropractic, and massage therapy (Table A14). Trends for the number of knee and hip TJRs and their cost estimates were derived from the Canadian Joint Replacement Registry (CJRR) data25 and Centre for Health Information (CIHI) hospitalization cost data in 201126, respectively. The St. Paul's Hospital cost model27 and the CIHI hospitalization cost database26 were used to calculate hospitalization cost for procedures other than hip and knee TJR. Price indexes for the average hospitalization cost reported by Cost per Weighted Case (CPWC) across Canada28, along with those for drugs29, physicians and orthopaedic surgeons30,31 for the years 2003e2010 were taken from the CIHI Price Index Reports, while the overall economy CPI was used from the Statistics Canada report19. The Clinical Research Ethics Board at the University of British Columbia provided ethics approval for this study. Derivation of cost parameters Physician, drugs and hospitalization (other than hip/knee TJR) We have used Medical Services Plan (MSP) database of Population Data BC from 1986/7 to 2003 to calculate the outpatient unit costs of OA patients21. We have included physician's visits and ambulatory services including x-rays, MRI, blood work, other diagnostic tests and other types of services. All physician visits for OA and OA-related inpatient procedures were included. For estimating the drug cost, all OA-related drugs were categorized into four types: acetaminophen, non-steroidal anti-inflammatory drugs (NSAIDs), coxibs and opioids. A stratified random sample of
Table I Direct cost categories of OA and inflation rates implemented in the POHEM-OA direct cost module for publically funded cost components Cost category
Cost component
Hospitalization and Hip/knee TJR surgery inpatient costs
Other in-patient procedures for OA (nonhip/knee TJR Surgery)
Physician and outpatient care costs Drugs
* y z x k ¶ # ** yy
Parameter
Data sources for cost estimation (year)
Hazard rates of surgery by age and sex Average unit cost of surgery
PDBC* (1987e2003)
Number of surgeries per year and future trend in surgery rates Average patient-year unit cost
CIHIeHPy (2010)
Cost component's inflation index (annual inflation rate %)
Data sources for inflation (year) [Reference]
1. Orthopedic surgeon fee-for 1. CIHI-PSFz (2005e2010) 2. Refer to [45] service price index (2.1%) 2. Surgical Implant inflation 3. CIHI-HPI (2007e2011)x rate (9.8%) 3. Hospital employee wage index (3.6 %)
CJRRk(1996e2010)
PDBC (2003)
Average procedure costs CIHI-HPI (2010) Number of procedures per year St. Paul's hospital¶ (2006) Physician visits and medical Average patient-year unit costs PDBC (2003) tests Prescription drugs
Average patient-year unit costs PDBC (2003)
Over-the-counter drugs
Average patient-year unit costs NPHSyy (2006)
Hospital employee wage index CIHI-HPI (2007e2011) (3.6%)
Physician fee-for service price index (3.2%)
NPDB- PPI (2003e2008)#
Prescription-based generic and CIHI-DER** (2009e2012) brand name drugs price index (0.9%) Over-the-counter generic and CIHI-DER (2009e2012) brand name drugs price index (1.6%)
PDBC: Population data BC from 1987-200328. CIHI-HP: Canadian Institute of Health Information hospitalization cost data in 201031. CIHI-PSF: CIHI report on physician and surgeon fees from 2005-201035. CIHI-HPI: CIHI report on hospital price index33 for full time hospital employee wage index from 2007 to 2011. CJRR: Canadian Joint Replacement Registry data for number of surgeries from 1996-201030. St. Paul's hospital unit cost model in 200632. NPDB-PPI: National physician data base reported by CIHI on physician fee-for-service price index from 2003-200836. CIHI-DER: CIHI-Drug Expenses Report34. NPHS: National Population Health Survey used for estimating over-the-counter drugs cost for OA (unpublished data).
B. Sharif et al. / Osteoarthritis and Cartilage 23 (2015) 1654e1663
1657
Table II Direct cost categories of OA and inflation rates implemented in the POHEM-OA direct cost module for rehabilitation and out-of-pocket cost components Cost category
Cost component
Parameter
Healthcare system paid Probability for home/hospital discharge strategy rehabilitation and home care costs Unit costs for each discharge strategy Out-of-pocket-cost Average patient-year unit costs during rehab Probability and unit costs Formal caregiver, Formal caregiver and other costs transportation and community costs Alternative care Physical therapy, Utilization rates costs Chiropractic's, Per-service Fees acupuncture and other complementary care services Side effect of drug's Cardiovascular disease Utilization of drug use Probability of side effects for (CVD's), Stroke, each type of drug Gastrointestinal (GI), Dyspepsia Life time cost for CVD and stroke, Incidence-based cost for GI and dyspepsia
Rehabilitation and home care costs (after hip/knee surgery)
* y z x k ¶
Data sources for the base-year cost estimation (year)
Cost component inflation index Data source for inflation (annual inflation rate %) (year) [Reference]
Oldmeadow et al. (2000)33
Hospital employee wage index (3.6 %) Consumer Price Index (CPI) for overall economy Consumer Price Index (CPI) for overall economy Consumer Price index (CPI) for overall economy
CIHI-HPI (2007e2011)*
MOH-OAz (2007) MSPx (2003)
e Physician fee-for service price index (3.2%)
e NPDB- PPI (2003 e2008)k
PDBC¶ (2003) Literature review (2010) [unpublished] Refer to45
e e
e e
Coyte et al. (2001)32 45
Refer to
Gupta et al. (2005)34
CANSIM (2005e2011)y CANSIM (2005e2011) CANSIM (2005e2011)
Consumer Price Index (CPI) for Consumer Price Index overall economy (CPI) for overall economy
CIHI-HPI: CIHI Report on Hospital Price Index33 for full time hospital employee wage index from 2007 to 2011. CPI for overall economy was calculated based on an exponential extrapolation from historical data (2005e2011) from Statistics Canada ‘s CANSIM CPI Report26. MOH-OA: British Columbia Ministry of Health's OA survey29. MSP fees in 2003: Medical Services Plan fees from PDBC28. NPDB-PPI: National physician data base reported by CIHI on physician fee-for-service price index from 2003-200835. PDBC: Population data BC from 1987-200328.
100,000 individuals was generated from PharmaNet database from Population Data BC23 to estimate input parameters associated with the per patient-year OA drug costs. We have included different OArelated drug types and assumed that all prescription-based drugs are included in the healthcare-system paid cost categories for OA (Table A5). For the over-the-counter drug costs, we used the utilization rates and costs for the over-the-counter drug categories from National Population Health survey for acetaminophen, some NSAIDS and some opioids (unpublished data). For hospital procedures (inpatient procedures) other than hip/knee surgeries, up to 25 diagnosis codes and a maximum of 12 procedure codes, along with the type of procedure (primary, revision or emergency), were found for each admission date on hospital discharge summaries22. Details of methods for estimating input parameters are provided in Appendix A1. Hip/knee TJR surgery costs A separate analysis was performed to estimate per-patient hip/ knee TJR costs. To this end, we needed to estimate the costs of both a knee/hip TJR and the number of TJR events in the simulation model. For the age-specific unit cost of a hip/knee TJR we used CIHI hospitalization data26. The overall cost of a TJR equals the weighted average of the three types of TJR as defined in CIHI hospitalization data i.e., unilateral/bilateral, hip/knee, and with/without infection. The procedure cost was calculated according to different cost components of the hip/knee TJR surgery. To estimate the future trend in terms of the number of TRJ surgeries, we performed a regression analysis and projected the total number of surgeries according to historical data from CJRR using a linear regression method. We used the number of surgeries estimated from the CJRR data for the years 2000e2010, extrapolated them into the future and then calibrated the model by adjusting the hazard ratio of hip/knee TJR surgery by age and sex until the results from the model agreed with the extrapolated CJRR data25. Details of method for projecting number of surgeries are provided in Appendix A2. The costs for hip/knee TJRs were also extrapolated for the next 20 years, separately for primary and
revision surgeries and for different age categories. There was a change in the cost of TJR surgery over time, due to both changes in the average length of stay (LOS) (which has decreased within the last decade) and price inflation for CPWC reported in the CIHI Hospital Price Index Report28. For this purpose, we projected the average LOS for knee and hip replacement according to historical data from the CJRR26 using an exponential model (Appendix A2). Alternative care costs A major component of the out-of-pocket cost of OA patients is associated with visits to alternative care professionals including physiotherapists, chiropractors and complementary care professionals24. To estimate and project these costs in POHEM-OA, we have used the MOH-OA dataset and developed a regression model to estimate the probability and annual number of visits to alternative care professionals according to patients' age, sex and OA stage (Appendix A1). The unit costs of visits to each alternative care provider were then extracted from MSP dataset in 2003 and the physician's price index rates were applied for projecting the rates from 2010 to 203130. Rehabilitation, out-of pocket, and side-effect of OA-related drugs After the hip/knee TJR surgery, patients undergo different rehabilitation routes depending on the disease severity and other patient characteristics. For this, we have developed a separate discharge destination model adapted from Coyte et al.32 and Oldmeadow et al.33. In addition to the health-system paid cost during and after the hip/knee TJR, patients need to pay out-of-pocket for some services including medical aid devices and formal caregiver costs. We have included those costs in the model according to the results from Gupta et al.34. The last cost component included in our model is the cost associated with the side-effect of OA drugs. We have included four types of side-effects of OA drugs including stroke, cardiovascular diseases (CVD), serious gastrointestinal (GI) complications, and dyspepsia. The lifetime cost of stroke and CVD and the unit cost for GI complications and dyspepsia were extracted from the literature. Details are provided in Appendix A1.
1658
B. Sharif et al. / Osteoarthritis and Cartilage 23 (2015) 1654e1663
Internal validation We have investigated the integrity of the results of our model through two types of internal validation techniques: 1) Assigning zero unit costs to physician, drugs, and hospital procedures (other than hip/knee TJR surgery) and 2) Setting the probability of eventbased cost categories to zero for hip/knee TJR surgery, rehabilitation, formal caregiver, side effect of drug costs and alternative care costs. Further details of the internal validation are discussed in Appendix A2. Results According to the POHEM model16, the size of the adult Canadian population 20 years of age and older is projected to increase from 26.3 million in 2010 to 32.2 million in 2031 (22.5% increase). During the same period, the OA population is forecast to rise from 3.6 million to almost 6 million (64% increase). OA prevalence for females is projected to increase from 0.16 to 0.21, while male prevalence is projected to increase from 0.11 to 0.16 throughout the study period. As shown in Fig. 2, we estimate that the total direct cost of OA will rise from $2.9 billion dollars in 2010 to $7.6 billion in 2031, an almost 2.6-fold increase in total OA-related direct costs. The increase in the total direct cost reflects increase in prevalence and utilization rates of the healthcare resources in addition to the inflation rates for all healthcare resources. According to our results, the extra direct cost associated solely with inflation in 2031 is projected to be $2.4 billion, of which $1.8 billion is related to health care system costs and $600 million to out-of-pocket costs. Total direct cost by sex and age category We estimate the direct costs of OA for females to be $0.71 billion and $1.85 billion higher than those for males in 2010 and 2031, respectively. As shown in Fig. 3, the share of female cost in both out-of-pocket and healthcare system costs was higher than that of males. The out-of-pocket cost for females and males associated with OA in 2010 was $0.52 billion and $0.14 billion, respectively, and increased to over $1.12 and $0.47 billion in 2031. Female OA patients between 60 and 70 years of age constituted the highest-cost group in 2010 (Fig. 4). However, the costs for
patients in other age categories and the demography of the Canadian population tend to change significantly over the future decades to a degree that, according to our results, OA patients aged between 70 and 80 years will constitute the highest-cost age category in 2031. For females, the change in the highest-cost age category occurs in 2020 and for males in 2026. As shown in Fig. 4, between 2025 and 2030, the costs for those aged 70e80 and 80e90 will increase at the highest rate compared to other age groups. Total direct cost by cost component Fig. 5 represents the out-of-pocket and healthcare system share of the total direct cost of OA and the cost components associated with each in 2010 and 2031. According to the POHEM-OA projections, from 2010 to 2031, the hospitalization costs of hip/knee TJR surgeries would increase from $720 million dollars to almost $2.8 billion dollars, while the overall costs for physicians, and prescription drugs would increase from about $773 million to $1.9 billion. At the same time, the cost of side effects from OA drugs rises from $316 million to $665 million and out-of-pocket costs of patients including alternative care, over the counter drugs and other out-of-pocket costs increase from $842 million to $1.92 billion during the same period (Fig. 5). While the dollar amounts of all the cost categories are projected to increase during the study period, each cost category's share of the total direct cost has shown a more complex trend, as shown in Fig. 6. According to our model, the share of the overall costs for physician, OTC drugs, and prescription drugs in addition to alternative care and formal caregiver costs is expected to decrease while the share of hospitalization costs is expected to increase. The share of rehabilitation and side effect costs is projected to remain stable since 2010. Discussion In this study, we used an array of population-based data sources, OA surveys and cost reports to estimate input parameters of a validated individual-level simulation model (POHEM-OA) and then project the costs of OA over the next 20 years from 2010 to 2031. According to the POHEM-OA's projection, the overall direct cost of OA will increase by almost 160%, from $2.9 billion ($842 million out-of-pocket cost and $2.06 healthcare system expenditures) to
Fig. 2. Sex-specific and total direct cost of OA from 2010 to 2031. *All costs are discounted back to 2010 $CAD using discount rates and are presented in billion dollars.
B. Sharif et al. / Osteoarthritis and Cartilage 23 (2015) 1654e1663
1659
Fig. 3. Sex-specific healthcare system and out-of-pocket cost of OA. * All costs are in 2010 $CAD billion dollars.
Fig. 4. Total direct cost of OA by age category for males and females*. * Costs are in 2010 $CAD; age categories are as the followings: 0 to 49, 50 to 59, 60 to 69, 70 to 79, 80 to 89 and 90þ.
$7.6 billion ($1.92 billion out-of-pocket and $5.7 billion healthcare system expenditures) in 2010 $CAD. The increase in total cost of OA is due to increase in prevalence and the average cost of the disease. According to our results, the average cost of OA increased from $805 in 2010 ($234 out-of-pocket and $571 healthcare system expenditures) to $1266 in 2031 ($320 out-of-pocket and $950 healthcare system expenditures) in 2010 $CAD. At the same time, the total number of people with OA increased from 3.6 million to almost 6 million during the study period. The average cost of OA has shown to be moderate, if not high, compared to other chronic conditions1. For example, based on our results, the average healthcare system expenditures of OA ($571) is relatively lower compared to other chronic diseases such as CVD ($801) and cancer ($715) as estimated using administrative data in Canada in $201035. On the other hand, prevalence of OA is
high and it is increasing with a fast rate. Consequently, the combination of moderate average cost and high and growing prevalence, makes OA an important health policy concern1. We also described how the proportion of total cost of OA would change across different cost components during the study period. The share of hospitalization cost was the highest among all cost components and increased from 25% to 37%, while physician visit, drug and out-of-pocket cost constituted the next highest proportion, respectively. Almost 95% of the hospitalization cost was due to hip and knee TJR surgeries and 5% due to other hospital procedures for OA patients. The increase in the cost of hip/knee TJR surgeries was not only due to increase in the number of cases, but also due to the high average cost because of high inflation rate for procedure and prosthesis cost specific to this type of surgery. According to our results, although the LOS after hip and knee TJRs will be decreasing
1660
B. Sharif et al. / Osteoarthritis and Cartilage 23 (2015) 1654e1663
Fig. 5. Total direct cost of OA by cost categories in 2010 and 2031: out-of-pocket and healthcare system's costs. *Out-of-pocket share of the rehabilitation cost includes rehabilitation and home care cost that is paid by patients and includes formal caregiver, home remodelling due to the disease and transportation cost during the year after the hip/ knee TJR surgery39; ** Healthcare system's share of the rehabilitation and home care cost that is paid by the government after the hip/knee TJR.; ^ Other out-of-pocket costs includes formal caregiver cost paid by the patients and community care services such as meals on wheels, transportation and home remodelling that are all paid before the hip/knee TJR surgery.
Fig. 6. Cost components' share of the total direct cost from 2010 to 2031. *Hospitalization costs include those related to hip/knee total joint replacements (TJR's) and other OArelated hospital procedures such as arthroscopy, shoulder TJR; ** Alternative care costs include those related to physiotherapy visits, chiropractor visits, and other alternative care professionals; Formal caregiver cost includes paid caregiver cost in addition to community services cost such as home re-modelling due to the disease and transportation as listed in Gupta et al.39.
from 2010 to 2031, the average cost of TJR would still be increasing mostly due to its very high inflation rate. The increasing number of TJR's is due to an increase in prevalence of OA and also due to policy decisions to increase capacity for performing these types of surgeries36. While TJR has shown to be highly cost-effective for those with severe OA37, recent studies have shown an increasing rate of TJR's for those in less advanced stages who may not benefit from the procedure due to higher failure rates and lower satisfaction38. Future studies using population health models such as POHEM-OA need to assess the budgetary implications of different appropriateness criteria to limit the TJR for those who do not benefit39, while provide means to encourage those who benefit the most from this procedure40. With respect to age groups, our model suggests that those with 70e80 years of age will constitute the highest cost population in 2031. This underlines an important implication for evaluating
policies for OA care and management targeted at younger age groups (e.g., 55e65 years in 2015) with mild OA in order to reduce the burden among elderly with severe OA in the future (e.g., 70e80 in 2031)41. This is highlighted by the OARSI conservative OA care and management guidelines for use of none-surgical care such as weight reduction and muscle strengthening exercises in early and mild stages of OA42. Our study was not free of limitations. On top of the health-caresector inflation, the utilization rates for hospitalization, drugs, and physicians have been also increasing during the past decade43. In this study, we only implemented the increase in hospitalization rates for the hip/knee TJR surgeries that was due to increase in both incidence and policy-related decisions to increase the number of surgeries. Other cost drivers, such as changes in the population age and BMI distribution or increases in life expectancy of the population, were also implemented in the POHEM-OA model16.
B. Sharif et al. / Osteoarthritis and Cartilage 23 (2015) 1654e1663
However, we did not include potential per-capita increases in drugs and physician utilization in this study. While physician utilization per capita (measured as total physician expenditure in constant prices) have increased by around 25% from 1998 to 200843, the increase in analgesics drugs was reported to be small in the same interval43. In terms of the drugs included in our model, we did not include newer analgesics, e.g., duloxetine, and intra-articular injections of hyaluronic acid. We also did not include any drug without conclusive evidence of effectiveness for general OA population, e.g., glucosamine44, and possible future drugs including disease-modifying OA drug (DMOAD's). Future studies using simulation models such as POHEM-OA can be used to address the cost-effectiveness of such treatments45. Additional OA-related cost components not included are nursing homes research costs and rare side-effects such as chronic kidney failure, hepatic failure, or addiction46. One of the strength of our COI approach was use of a comprehensive cost listings of all resources used in addition to consideration of heterogeneity of the population that reflected the intensity of the resource used8. Traditional COI studies have been the target of criticism since they use single data sources and they are not capable of projecting the cost into the future8. According to the Canadian Health Expenditure reports in 1994, the total direct OA cost in Canada was projected to reach around $1.1e$1.6 billion dollars in 201047, which is lower than our estimates. This is due to the fact that none of the previous cost projections of OA included both the BMI and demographic shifts among the Canadian population16. Our result was close to another previous Canadian study using a population database (rather than sample surveys) for the overall direct cost of OA in 2006 ($2.1 billion in $2010 CAD)10. However, it should be noted that none of the previous studies included out-of-pocket costs for patients with OA. While recent COI studies have used new techniques to estimate incremental costs by performing regression-based analysis to control for relevant comorbidities, there are limitations in these methodologies that have caused large variations in the estimates of total and average cost of OA8. For example, while several studies, consistent with our results, have reported hospitalization costs as the most expensive category of the total medical cost of OA5,6,10, a Canadian study reported average cost of physician and outpatient procedures to be slightly higher than the cost of hospitalization11. Our results for the distribution of total cost of OA across different cost categories, however, conforms to the recent study by Losina et al.36, describing lifetime costs of knee OA in US. Losina et al.36 have shown that surgery and post-surgical cost is within 33e67% of all the lifetime direct cost of knee OA across different eligibility criteria for surgery defined based on radiographic stages. Estimating and projecting the cost of OA has been a methodological challenge in health economics literature when trying to account for the complex interrelation of time varying risk factors such as demographics and BMI shifts among the population48,49. In this study, we performed a COI study with the use of MSM modeling to project changes in the distribution of direct cost of OA across different sub-populations and cost components. Constant escalation of OA costs leads to growing demands for more money to be devoted to arthritis care specifically among high cost subpopulations including females and those over the age of 70 as identified in this study. Given the magnitude of OA costs, small decrease in risk factors or cost would help attenuate rising healthcare expenditures in the near future. Author contributions Conceived and designed the analysis: BS JK NB MR WMF PF HW AA. Performed.
1661
Simulation and analysis: BS WMF PF. Analyzed the data: BS MR NB. Wrote the paper: BS JK NB MR WMF PF HW AA. Approved the final draft: BS JK NB MR WMF PF HW AA. Conflict of interest No financial support or other benefits from commercial sources were received for the work reported in this manuscript. None of the authors has financial interests for potential conflict of interest or the appearance of a conflict of interest with regard to the work. Acknowledgements Authors would like to thank Canadian Arthritis Network (CAN), the Canadian Institutes of Health Research (CIHR), FRN# 92246, and Dr Linda Li for their help and support. The views expressed in this paper are solely those of the authors and do not reflect those of Statistics Canada. All inferences, opinions, and conclusions drawn in this manuscript are those of the authors, and do not reflect the opinions or policies of the Data Steward(s). Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.joca.2015.05.029. References 1. Yelin E. The economics of osteoarthritis. In: Brandt K, Doherty M, Lohmander LS, Eds. Osteoarthritis. New York: Oxford University Press; 1998:23e30. 2. Bitton R. The economic burden of osteoarthritis. Am J Manag Care 2009 Sep;15(8 Suppl):S230e5. 3. Dibonaventura MD, Gupta S, McDonald M, Sadosky A, Pettitt D, Silverman S. Impact of self-rated osteoarthritis severity in an employed population: cross-sectional analysis of data from the national health and wellness survey. Health Qual Life Outcomes 2012 Mar:10e30. 4. Murray CJ, Vos T, Lozano R, Naghavi M, Flaxman AD, Michaud C, et al. Disability-adjusted life years (DALYs) for 291 diseases and injuries in 21 regions, 1990-2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 2012;380(9859):2197e223. 5. Yelin E, Murphy L, Cisternas MG, Foreman AJ, Pasta DJHC. Medical care expenditures and earnings losses among persons with arthritis and other rheumatic conditions in 2003, and comparisons with 1997. Arthritis Rheum 2007;56(5): 1397e407. 6. Maetzel A, Li LC, Pencharz J. The economic burden associated with osteoarthritis, rheumatoid arthritis, and hypertension: a comparative study. Ann Rheum Dis 2004;63:395e401. 7. Xie F, Thumboo J, Li S-C. True difference or something else? Problems in cost of osteoarthritis studies. Semin Arthritis Rheum 2007;37(2):127e32. 8. Akobundu E, Ju J, Blatt L, Mullins CD. Cost-of-illness studies: a review of current methods. Pharmacoeconomics 2006;24(9): 869e90. 9. Murphy LB, Helmick CG, Cisternas MG, Yellin EH. Estimating medical costs attributable to osteoarthritis in the US population: comment on the article by Kotlarz et al. Arthritis Rheum 2010;62(8):2566e7. 10. Anis AH, Zhang W, Bansback N, Guh DP, Amarsi Z, Birmingham CL. Obesity and overweight in Canada: an updated cost-of-illness study. Obes Rev 2010;11(1):31e40.
1662
B. Sharif et al. / Osteoarthritis and Cartilage 23 (2015) 1654e1663
11. Tarride JE, Haq M, O'Reilly DJ, Bowen JM, Xie F, Dolovich L, et al. The excess burden of osteoarthritis in the province of Ontario. Can Arthritis Rheum 2012;64(4):1153e61. 12. Rutter CM, Zaslavsky AM, Feuer EJ. Dynamic microsimulation models for health outcomes: a review. Med Decis Mak 2011;31(1):10e8. re O, Richy F, Gathon HJ, 13. Hiligsmann M, Ethgen O, Bruye Reginster JY. Development and validation of a Markov microsimulation model for the economic evaluation of treatments in osteoporosis. Value Health 2009;12(5):678e96. 14. The Mount Hood 4 Modeling Group. Computer modeling of diabetes and its complications: a report on the Fourth Mount Hood Challenge Meeting. Diabetes Care 2007;30(6):1638e46. 15. Vanness DJ, Tosteson AN, Gabriel SE, Melton 3rd LJ. The need for microsimulation to evaluate osteoporosis interventions. Osteoporos Int 2005 Apr;16(4):353e8. 16. Kopec JA, Sayre EC, Flanagan WM, Fines P, Cibere J, Rahman MM, et al. Development of a population-based microsimulation model of osteoarthritis in Canada. Osteoarthritis Cartilage 2010 Mar;18(3):303e11. 17. Zucchelli E, Jones AM, Rice N. The evaluation of health policies through microsimulation methods. HEDG working paper 10/ 03. University of York [cited 2015 Feb 01]. Available from:, http://www.york.ac.uk/media/economics/documents/herc/ wp/10_03.pdf; 2010. 18. Statistics Canada. Canadian community health survey (CCHS). Microdata File User Guide. Ottawa, Ontario: Health Statistics Division, Statistics Canada [cited 2015 April 10]. Available from:, http://www.statcan.gc.ca/eng/survey/household/3226; 2001. 19. Statistics Canada. The consumer price index. Ottawa, Canada [cited 2015 April 10]; Available from: http://www.statcan.gc. ca/tables-tableaux/sum-som/l01/cst01/econ46a-eng.htm 20. Boardman AE, Moore AM, Vining AR. The social discount rate for Canada based on future growth in consumption. Can Public Policy 2010;36(3):325e43. 21. British Columbia Ministry of Health [creator]. Medical services plan (MSP) payment information file. V2. Population Data BC [publisher]. Data Extract. MOH (2011) [cited 2015 April 10]. Available from: http://www.popdata.bc.ca/data; 2011. 22. British Columbia Ministry of Health [creator]. Discha Rge abstract database (Hospital separations). V2. Population Data BC [publisher]. Data Extract. MOH (2011) [cited 2015 April 10]. Available from: http://www.popdata.bc.ca/data; 2011. 23. British Columbia Ministry of Health [creator]. Data extract. Data Stewardship Committee (2011). PharmaNet. V2. BC Ministry of Health [publisher] [cited 2015 April 10]. Available from: http://www.popdata.bc.ca/data; 2011. 24. British Columbia Osteoarthritis Survey, Vancouver, Canada 2008 [cited 2015 Jan 10]. Available from: http://arthritis.rehab. med.ubc.ca/files/2011/08/BCOASurvey.pdf. 25. Canadian Institute of health Information (CIHI). Canadian joint replacement registry (CJRR). CIHI [cited 2015 Jan 10]. Available from: http://www.cihi.ca/CIHI-ext-portal/xlsx/internet/stats_ cjrr2011_clinicaldata_en; 2011. 26. Canadian Institute of health Information (CIHI). CMG client tables 2011. DAD Acute inpatient data [cited 2015 Jan 10]. Available from: https://secure.cihi.ca/estore/productFamily. htm?pf¼PFC1627&lang¼en&media¼0; 2012. 27. Borsa J, Anis A. The cost of hospital care in Canada: a comparison of two alternatives. Healthc Manage Forum 2005;18(1):19e27. 28. Canadian Institute for Health Information. Hospital cost drivers technical report; what factors have determined hospital expenditure trends in Canada?. Ottawa, Canada [cited
29.
30.
31.
32.
33.
34.
35.
36.
37. 38.
39.
40.
41.
42.
43.
2015 Jan 10]. Available from:, http://www.cihi.ca/cihi-extportal/pdf/internet/hospital_costdriver_tech_en; 2010. Canadian Institute for Health Information. Drivers of prescription drug spending in Canada. Ottawa, Canada [cited 2015 Jan 10]. Available from:, http://www.cihi.ca/CIHI-ext-portal/ pdf/internet/DRUG_SPEND_DRIVERS_EN; 2012. Canadian Institute for Health information. Health care cost drivers: Physician Expenditure d Technical report spending and health workforce, Ottawa, Canada, 2009 [cited 2015 Jan 10]. Available from: http://www.cihi.ca/cihi-ext-portal/pdf/ internet/health_costdriver_phys_tech_en. Canadian Institute for Health Information. Average payment per physician report, Fee-for-service physicians in Canada 2004e2005. Ottawa, Canada: National Physician Database [cited 2015 Jan 10]. Available from: https://secure.cihi.ca/free_ products/app_average_payment_per_physician_report_2004_ e.pdf; 2006. Coyte PC, Young W, Croxford R. Costs and outcomes associated with alternative discharge strategies following joint replacement surgery: analysis of an observational study using a propensity score. J Health Econ 2000;19(6):907e29. Oldmeadow LB, Mcburney H, Robertson VJ. Predicting risk of extended inpatient rehabilitation after hip or knee arthroplasty. J Arthroplasty 2003;18(6):775e9. Gupta S, Hawker GA, Laporte A, Croxford R, Coyte PC. The economic burden of disabling hip and knee osteoarthritis (OA) from the perspective of individuals living with this condition. Rheumatol (Oxford) 2005;44(12):1531e7. Patra J, Popova S, Rehm J, Bondy S, Flint R, Giesbrecht N. Economic cost of chronic disease in Canada 1995e2003 prepared for the Ontario Chronic Disease Prevention Alliance and the Ontario Public Health Association [cited 2015 March 10]. Available from: http://www.ocdpa.on.ca/sites/default/files/ publications/OCDPA_EconomicCosts.pdf; March 2007. Losina E, Paltiel AD, Weinstein AM, Yelin E, Hunter DJ, Chen SP, et al. Lifetime medical costs of knee osteoarthritis management in the United States: impact of extending indications for total knee arthroplasty. Arthritis Care Res(Hoboken) 2015;67(2):203e15. Katz JN. Total joint replacement in osteoarthritis. Best Pract Res Clin Rheumatol 2006;20(1):145e53. Ghomrawi HM, Alexiades M, Pavlov H, Nam D, Endo Y, Mandl LA, et al. Evaluation of two appropriateness criteria for total knee replacement. Arthritis Care Res (Hoboken) 2014;66(11):1749e53. Riddle DL, Jiranek WA, Hayes CW. Use of a validated algorithm to judge the appropriateness of total knee arthroplasty in the United States: a multicenter longitudinal cohort study. Arthritis Rheumatol 2014;66(8):2134e43. Hawker GA, Wright JG, Glazier RH, Coyte PC, Harvey B, Williams JI, et al. The effect of education and income on need and willingness to undergo total joint arthroplasty. Arthritis Rheumatol 2002;46(12):3331e9. Gruber WH, Hunter DJ. Transforming osteoarthritis care in an era of health care reform. Clin Geriatr Med 2010;26(3): 433e44. Zhang W, Moskowitz RW, Nuki G, Abramson S, Altman RD, Arden N, et al. OARSI recommendations for the management of hip and knee osteoarthritis, Part II: OARSI evidence-based, expert consensus guidelines. Osteoarthritis Cartilage 2008;16(2):137e62. Canadian Institute of Health Information (CIHI). Health care cost drivers: The facts. Ottawa, Canada [cited 2015 March 10]. Available from:, https://secure.cihi.ca/free_products/health_ care_cost_drivers_the_facts_en.pdf; October 2011.
B. Sharif et al. / Osteoarthritis and Cartilage 23 (2015) 1654e1663
44. Clegg DO, Reda DJ, Harris CL, Clegg DO, Reda DJ, Harris CL, et al. Glucosamine, chondroitin sulfate, and the two in combination for painful knee osteoarthritis. N Engl J Med 2006;354(8): 795e8. 45. Losina E, Daigle ME, Suter LG, Hunter DJ, Solomon DH, Walensky RP, et al. Disease-modifying drugs for knee osteoarthritis: can they be cost-effective? Osteoarthritis Cartilage 2013;21(5):655e67. 46. Zhang W, Moskowitz RW, Nuki G, Abramson S, Altman RD, Arden N, et al. OARSI recommendations for the management of hip and knee osteoarthritis, Part I: critical appraisal of existing treatment guidelines and systematic review of current research evidence. Osteoarthritis Cartilage 2007 Sep;15(9): 981e1000.
1663
47. Patra J, Popova S, Rehm J, Bondy S, Flint R, Giesbrecht N. Economic cost of chronic disease in Canada 1995e2003. Health Canada1e37 [cited 2015 April 01]. Available from: http://ocdpa.ca/sites/default/files/publications/OCDPA_ EconomicCosts.pdf; 2007(March). 48. Statistics Canada. The Life paths model [cited 2015 March 10]. Available from: http://www.statcan.gc.ca/eng/ microsimulation/lifepaths/overview/. 49. Direct cost of osteoarthritis in Canada : an Application of Microsimulation Modeling with Uncertainty Analysis- Sharif Behnam, University of British Columbia, PhD thesis, [cited 2015 April 01]. Available from: https://circle.ubc.ca/handle/ 2429/46614.