Report 1: Economic Burden of Illness in Canada, 2005–2008 – EBIC hospital care expenditures, 2004–2008
Report 1: EBIC Hospital Care Expenditures, 2004-2008
A hospital is an institution licensed or approved by a provincial/territorial government or operated by the Government of Canada in which patients are accommodated on the basis of medical/nursing need and are provided with continuing medical/nursing care and supporting diagnostic and therapeutic services.Footnote 6 Hospital expenditures include all costs of operating and maintaining both public and private hospitals in Canada: drugs dispensed in hospitals, medical supplies, therapeutic and diagnostic outpatient costs, administrative costs, some research costs, accommodation and meals for patients, maintenance of hospital facilities, and gross salaries and wages for all hospital staff (such as physicians on hospital payroll, nurses, technicians and medical students) (5).
Although the current edition of EBIC focuses on the years 2005-2008, the 2004 results are also presented, as the data required to produce 2004 estimates were available. EBIC 2004-2008 hospital care expenditures were estimated and distributed across diagnostic category/subcategory, sex, age group and province/territory for each year of analysis. This report describes the data sources and methods used to derive the 2004-2008 hospital care expenditure estimates. Additionally, it presents and discusses the results and the data and methods limitations.
2. Data Sources
To estimate EBIC hospital care expenditures, the following databases from CIHI were used: Discharge Abstract Database (DAD), Hospital Morbidity Database (HMDB), National Ambulatory Care Reporting System (NACRS), Hospital Mental Health Database (HMHDB), Canadian Management Information Systems Database (CMDB) and NHEX.
The DAD, HMDB, NACRS and HMHDB hold information on hospital separations (discharges, deaths, sign-outs and transfers). In these databases, the data collected on each discharge abstract includes coded diagnoses, coded intervention, patient demographic information and administrative information. The DAD is a national database (excluding Quebec) of information on all acute inpatient hospital separations for each fiscal year (8-13).Footnote 7 Additionally, the DAD contains information on day surgeries for most provinces/territories and some data on chronic, rehabilitation and psychiatric hospital separations.Footnote 8 The HMDB is a national database that contains information on all acute inpatient hospital separations by fiscal year, similar to the DAD (14-19). However, the HMDB holds information on Quebec acute inpatient separations and excludes all day surgery records. The NACRS contains records of all Ontario ambulatory care separations (day surgery, emergency department, clinic and other ambulatory care), as well as some ambulatory care separations for a few other provinces/territories (20-25). The HMHDB contains information by fiscal year on all Ontario psychiatric hospital separations and on all separations from designated adult psychiatric beds in Ontario general hospitals; this information is partial for other provinces/territories (26-28).Footnote 9
The CMDB and NHEX hold hospital expenditure information. The CMDB provides public and private hospital financial information, such as total expenses and detailed inpatient and outpatient expenses incurred, by hospital and fiscal year (29,30). The NHEX supplies public and private hospital expenditure totals, as well as other expenditure totals (e.g. drug, physician), by province/territory and fiscal/calendar year (5).
In previous EBIC editions, variations of two costing methods were used to allocate hospital care expenditures by EBIC categories, the per diem method and the resource intensity weight (RIW) method. The former involves multiplying record-level length of stay (LOS) by a facility per diem (or cost per bed per day) to obtain a record-level cost (each record represents a hospital separation). Costs per diagnostic category/subcategory, sex, age group and province/territory are the sum of the costs per record within each category. The RIW method involves multiplying record-level RIWs by a facility-level cost per weighted case (CPWC) to obtain a record-level cost.Footnote 10,Footnote 11 Costs per diagnostic category/subcategory, sex, age group and province/territory are the sum of the costs per record within each category.
A variation of the RIW method was used to estimate hospital care expenditures for EBIC 1998, whereas the per diem method was used for EBIC 2000 (3,4). The RIW method is considered a superior method for estimating hospital care expenditures since it does not assume a homogeneous patient population within a given hospital. When the per diem method is used, two patients in the same hospital with the same LOS would be assigned the same cost, when in reality their resource utilization may be very different. Several factors are considered in CIHI’s calculation of RIW values for DAD acute inpatient records: case mix group, age factor, comorbidity factor, a number of flagged interventions factor, intervention event factor, out-of-hospital intervention factor and possible interactions (32).Footnote 12
The EBIC costing method used to estimate 2004-2008 hospital care expenditures varied as a result of the differences in data availability by hospital type/service. However, the RIW or other weighting method, such as weighted LOS, was used when available. All RIW, CPWC and per diem fields were calculated and provided by CIHI.Footnote 13 For further information on these calculations, please consult the appropriate CIHI documentation (30-32).
An EBIC database for each year of analysis (2004-2008) was created to house total hospital care expenditures by diagnostic category/subcategory, sex, age group and province/territory.Footnote 14,Footnote 15,Footnote 16,Footnote 17 For all hospital types/services and years of analysis, all record-level costs were attributed to the health condition most responsible for the hospital stay. The most responsible health condition was coded in CIHI’s hospital databases using the International Classification of Diseases (ICD) coding; depending on the database either version 9 or 10 was used (33,34). Please consult the EBIC diagnostic category table (Appendix C) which illustrates how costs were grouped into the EBIC diagnostic categories/subcategories using ICD codes.Footnote 18 Although EBIC hospital care expenditure totals are not available by hospital/service type, the sections below (3.1-3.4) detail the method used for each hospital/service type, which often differed because of differences in data sources and availability.
3.1 Acute Inpatient Hospital Care
Acute inpatient 2004-2008 hospital care expenditures for all provinces/territories (except Quebec) were estimated using data from the DAD, employing the RIW method (RIW*CPWC).Footnote 19,Footnote 20,Footnote 21 Acute inpatient expenditures for Quebec were estimated using data from the HMDB, employing the per diem method (per diem*LOS), as RIWs and CPWCs were not available.Footnote 22 However, although record-level LOS was available for Quebec in the years of analysis, facility and provincial per diems were not. Therefore, the weighted average per diem for acute inpatient discharges of other provinces was used as a proxy for a Quebec per diem in the years 2005-2008. Per diems were not provided for any province in the 2004 hospital databases. Therefore, a 2004 Quebec per diem was estimated using a linear regression of the weighted average of per diems from 2005-2008, after adjusting for inflation.Footnote 23
Acute inpatient hospital care expenditures for general hospital designated psychiatric beds were estimated for Ontario in 2006-2008 using information from the HMHDB, employing the per diem method.Footnote 24,Footnote 25 It was not possible to estimate these expenditures for Ontario in 2004 and 2005 or for the other provinces/territories in 2004-2008.
3.2 Ambulatory Hospital Care
3.2.1 Day Surgery
Day surgery expenditures were estimated using information in the DAD and NACRS, employing the RIW method.Footnote 26 In the years 2004-2008, Quebec and Alberta did not submit day surgery information to these databases, thus their expenditures had to be estimated using different methods.Footnote 27 The cost distribution of all DAD and NACRS day surgery records, after adjustment for population, was used to distribute day surgery expenditure totals for Quebec and Alberta across EBIC categories.Footnote 28,Footnote 29 These totals were obtained by multiplying the DAD and NACRS day surgery cost per capita by the population of the respective province.Footnote 30
3.2.2 Emergency, Clinic and Other Ambulatory Care
For the years of analysis, the NACRS contained complete reporting of emergency department, clinic and other ambulatory care visits for Ontario and partial reporting for certain other provinces/territories.Footnote 31 Therefore, Ontario’s cost distribution for ambulatory care (excluding day surgery), after adjustment for population, was used to distribute CMDB ambulatory care expenditure totals (excluding day surgery costs) across EBIC categories for all provinces/territories.Footnote 32,Footnote 33 CMDB emergency, clinic and other ambulatory care expenditure totals were not available for Quebec and Nunavut; these expenditure totals were estimated by multiplying an ambulatory care (excluding day surgery) cost per capita by the population of the respective province.Footnote 34
3.3 Psychiatric Hospital Care
As of April 2006, it has been mandatory for Ontario to report all information on psychiatric hospital stays to the Ontario Mental Health Reporting System (OMHRS); all OMHRS closed records (discharges) are also included in the HMHDB. As well, the HMHDB contains partial reporting of psychiatric hospital separations for other provinces/territories. RIWs were not available for HMHDB psychiatric hospital separations. Instead, CIHI maintains and updates the System for Classification of In-Patient Psychiatry (SCIPP) grouping and weighting methodology for Ontario mental health data (OMHRS data) (40).Footnote 35 Additionally, CIHI produces SCIPP Weighted Patient Days (SWPDs) for Ontario mental health data. SWPDs weight a patient’s LOS according to resource utilization.Footnote 36 To estimate EBIC 2006-2008 psychiatric hospital care expenditures, Ontario’s SWPD distribution for psychiatric hospital separations, after adjustment for population, was used to distribute CMDB psychiatric hospital expenditure totals across categories for all provinces/territories.Footnote 37,Footnote 38,Footnote 39
For the years 2004 and 2005, Ontario did not have mandatory reporting of psychiatric hospital separations. Thus, for these years, Ontario’s 2006 SWPD distribution for psychiatric hospital separations, after adjustment for population, was used to distribute CMDB psychiatric hospital expenditure totals across categories for all provinces/territories.Footnote 33,Footnote 34,Footnote 40
3.4 Chronic and Rehabilitation Hospital Care
The EBIC 1998 methods used to distribute chronic and rehabilitation hospital care expenditure totals across EBIC categories were adopted, in the absence of the Continuing Care Reporting System (CCRS) and the National Rehabilitation Reporting System (NRS) data.Footnote 41 For each year of analysis, the cost distribution of all DAD acute inpatient discharges with a length of stay equal to 100 days or more, after adjustment for population, was used to distribute CMDB chronic and rehabilitation hospital expenditure totals across EBIC categories for all provinces/territories.Footnote 42,Footnote 43,Footnote 44,Footnote 45
4. ResultsFootnote 46
4.1 Expenditures by Hospital Type/Service
In the years 2004-2008, acute inpatient and ambulatory hospital care expenditures accounted for, on average, 88.7% of total hospital care expenditures. Across the same years of analysis, psychiatric, chronic and rehabilitation hospital care expenditures accounted for much smaller percentages, on average 5.2%, 5.2% and 0.9% respectively.
4.2 Expenditures by Diagnostic Category
Table 4 illustrates EBIC hospital care expenditures by diagnostic category for the years 2004-2008. The five diagnostic categories with the largest expenditures in 2008 were neuropsychiatric conditions ($5.5 billion, 11.2%), cardiovascular diseases ($5.1 billion, 10.3%), injuries ($3.4 billion, 6.9%), digestive diseases ($2.8 billion, 5.8%) and malignant neoplasms ($2.3 billion, 4.7%); this is consistent across all years of analysis. Together, the costs for these five diagnostic categories represented almost 40% of total hospital care expenditures. EBIC unattributable hospital care expenditures are defined as total NHEX hospital expenditures minus total EBIC hospital care expenditures distributed across categories. As shown in Table 4, the unattributable percentage of EBIC 2008 hospital care expenditures was 20.8% ($10.2 billion).
4.3 Expenditures by Diagnostic Category and Sex
Table 5 illustrates EBIC 2008 hospital care expenditures by diagnostic category and sex. In 2008, 48.7% ($19.1 billion) and 51.3% ($20.0 billion) of expenditures were attributable to males and females respectively. The three diagnostic categories with the largest expenditure for males were neuropsychiatric conditions ($2.9 billion), cardiovascular diseases ($2.9 billion) and injuries ($1.7 billion). For females these were neuropsychiatric conditions ($2.6 billion), cardiovascular diseases ($2.1 billion) and injuries ($1.7 billion).
The five diagnostic categories with the largest difference in cost distribution across the sexes were other neoplasms (36.2% male, 63.8% female), genitourinary diseases (39.6% male, 60.4% female), endocrine disorders (41.0% male, 59.0% female), cardiovascular diseases (57.9% male, 42.1% female) and nutritional deficiencies (42.2% male, 57.8% female).Footnote 47 Estimation of unattributable hospital care expenditures by sex was not possible.
4.4 Expenditures by Diagnostic Category and Age Group
Figure 14 illustrates EBIC 2008 hospital care expenditures for each age group. Individuals aged 0-14 years incurred the lowest percentage of hospital care expenditures (8.1%) and those aged 75+ years the highest (29.4%). Additionally, individuals aged 55 years and older accounted for approximately 60% of total EBIC hospital care expenditures. Estimation of unattributable hospital care expenditures by age group was not possible.
Figure 15 illustrates EBIC 2008 hospital care expenditures by diagnostic category and age group for the five most costly diagnostic categories. Expenditures were highest for individuals aged 75+ years, except in the neuropsychiatric conditions category where individuals aged 35-54 years (31.1%) accounted for the highest expenditures. Expenditures for cardiovascular diseases and malignant neoplasms increased with age; individuals aged 75+ years accounted for 44.7% and 31.3% of expenditures in the cardiovascular diseases and malignant neoplasms categories respectively. Finally, individuals aged 35-54 years together with those aged 75+ years accounted for 50% of hospital care expenditures in both the injuries and digestive diseases categories.
5. Discussion and Limitations
5.1 Comparison Across EBIC Categories and Years
Comparisons of EBIC 2004-2008 hospital care expenditures across provinces/territories should be made with caution. Each province/territory maintained different levels of reporting, which resulted in varying levels of unattributable costs. For example, province A may have unattributable costs of 30% and province B may have unattributable costs of 18%. Therefore, a lower per capita cost for a diagnostic category in one particular province could be a reflection of a higher unattributable cost. Additionally, the unattributable costs for one province could vary across years. The per diem method was used to estimate costs of Quebec acute inpatient separations, which may have resulted in higher estimated costs relative to other provinces.Footnote 48 Additionally, province/territory of residence was used to assign hospital care costs by geographic category; however, this field was unavailable for Quebec hospital stays.Footnote 49 Therefore, all Quebec hospital separation costs were assigned to Quebec using the province of occurrence field; this may have resulted in higher hospital costs for the province.
Previous editions of EBIC used different data sources and methods to estimate hospital care expenditures. A variation of the RIW method was used to estimate hospital care expenditures for EBIC 1998 and the per diem method was used for EBIC 2000 (3,4). Although a variation of the RIW method was used in EBIC 1998, comparisons between EBIC 1998 and EBIC 2004-2008 hospital care expenditures should be made with extreme caution. First, EBIC 2004-2008 and EBIC 1998 grouped costs by ICD code; however, different groupings were used. The diagnostic grouping tables of both editions should be compared before any attempt is made to compare costs by diagnostic category. For example, cardiovascular diseases were coded in the same way in both editions, although infectious and parasitic diseases were not. Second, there were differences in the RIW method used in the two editions. For EBIC 2004-2008, costing information was available by facility (in most cases), from which a CPWC or per diem was estimated. The availability of costing information for EBIC 1998 was much more limited. EBIC 1998 distributed hospital type expenditure totals (not at the facility level) across categories using an RIW or LOS distribution. These hospital type expenditure totals were estimated using the per diem method, specifically, by multiplying an average per diem for all hospital types by the total number of beds for a particular hospital type. Thus, the same per diem cost was assigned for all hospital types. Furthermore, in EBIC 2004-2008, the NACRS was used to distribute ambulatory care costs for Canada instead of the Ambulatory Care Classification System (ACCS), which was used in EBIC 1998. Additionally, in EBIC 2004-2008 weighted cost and weighted LOS distributions were used to distribute chronic/rehabilitation and psychiatric hospital expenditures across EBIC categories, instead of a simple LOS distribution, as used in EBIC 1998. Finally, there are different percentages of unattributable costs in the two editions.
5.2 Data Limitations
There are several data limitations that may lead to a misrepresentation of hospital care expenditures across categories. First, the hospital databases capture hospital separations only by fiscal year.Footnote 50 These separations may not represent the distribution of expenditures across categories within a fiscal year, since certain individuals with costly health conditions may stay in hospital for longer than a fiscal year. If patients stayed in a hospital for longer than a fiscal year, all of their respective costs would be assigned to the year in which they were discharged. Furthermore, if some of these patients' costs fell in years other than the year of study, assigning all hospital costs for the stay to the year of study is not consistent with a prevalence-based approach.
Day surgery information was available for most provinces/territories.Footnote 51 However, complete information on emergency, clinic and other ambulatory care separations was available only for Ontario. Therefore, Ontario’s cost distribution of these services was used to distribute the CMDB provincial/territorial ambulatory care expenditure totals (excluding day surgery) across EBIC categories. Although Ontario represents approximately 38% of Canada’s population, Ontario’s burden of disease may not reflect that of other provinces/territories, even after adjustment for differences in population (sex and age specific). Specifically, it may misrepresent the burden of disease for jurisdictions with different disease-specific risk factor profiles, as well as those with different distributions of urban/rural, aboriginal and other minority populations. Furthermore, the CMDB ambulatory care expenditure totals (excluding day surgery) distributed across category included different hospital types/services than did the data used to distribute the totals. The CMDB expenditure totals included costs for poison and drug information services and excluded all private clinic costs, whereas the cost distribution was based on data that excluded information on Ontario poison and drug information services and included information on Ontario private clinic visits. It was not possible to obtain provincial/territorial CMDB hospital expenditure totals for private clinics. The inclusion of poison and drug information service costs in the distributions will likely have minimal effect on costs by category, as less than half of the provinces/territories have these services. For province/territories with poison and drug information services, costs are on average only 0.3% of ambulatory care expenditure totals (excluding day surgery).
For 2004-2008, information on psychiatric hospital separations was partial for most provinces/territories. However, this information was complete for the province of Ontario. Therefore, Ontario’s SCIPP weighted patient day distribution of psychiatric hospital separations was used to distribute CMDB psychiatric hospital expenditure totals across EBIC categories. Ontario’s weighted patient day distribution may not accurately represent that of other provinces/territories. Adjusting for the differences in sex- and age-specific populations attempts to account for changes in the number of discharges due to the differences in these populations. However, it is possible that the prevalence of certain diseases/disorders within the same sex and age group is different across jurisdictions.
Databases containing information on chronic (CCRS) and rehabilitation (NRS) hospital stays were not available for EBIC analyses. The cost distribution of acute inpatient DAD discharges with a LOS equal to 100 days or more was used to distribute total CMDB chronic and rehabilitation hospital expenditures across EBIC category; this method was also used in EBIC 1998. DAD patient separations with a LOS equal to 100 days or more may not reflect the characteristics (sex, age, diagnosis) of patients in chronic/rehabilitation hospitals. However, expenditures for these hospital types represent, on average, only approximately 6% of total EBIC hospital care expenditures in the years 2004-2008. Therefore, the effect on the overall EBIC estimates is likely small, unless certain diagnoses are a majority in these hospital types and not represented to the same proportion in the distribution. The accuracy of future EBIC estimates would likely increase with the inclusion of CCRS and NRS data.
As mentioned in section 5.1, both the absence of the province of residence field and the use of the per diem method for Quebec cost estimates may have resulted in higher acute inpatient and ambulatory care costs assigned to Quebec relative to other province/territories. Expenditure data for Quebec were largely unavailable for the years of analysis, as the province did not submit information to the CMDB. Although PHAC estimated Quebec acute inpatient and ambulatory care costs, costs for psychiatric, chronic and rehabilitation hospitals were not estimated. As these hospital types may operate in the province, lower costs for illnesses largely seen in such hospitals may have been assigned to Quebec relative to other provinces/territories. Also, Nunavut did not submit costing data to the CMDB. PHAC estimated acute inpatient and ambulatory hospital care expenditures for Nunavut but assumed the province did not have designated psychiatric, chronic and rehabilitation hospitals. Therefore, costs for these hospital types were not estimated.
General hospitals may have designated psychiatric, chronic and/or rehabilitation beds. However, it was not possible to estimate expenditures for these hospital bed separations, except for Ontario general hospital designated psychiatric beds in 2006-2008.Footnote 52 As a result, Ontario may have slightly higher costs for neuropsychiatric conditions costs for 2006-2008 relative to the other provinces/territories.
A small number of separations in the hospital databases had required fields with missing values. For example, 0.002% of acute inpatient records in 2008 were missing a value for the most responsible health condition, province, sex and/or age field; these records accounted for 0.005% of total 2008 acute inpatient expenditures. Given the small magnitude of records missing required fields no attempt was made to distribute these costs across category.
In 2004-2008, annual unattributable hospital care expenditures were, on average, 23%.Footnote 53 The presence of unattributable hospital care expenditures may misrepresent the true distribution of expenditures across EBIC category. If unattributable costs account for a large percentage of costs for a particular category and this is not reflected in the cost distribution of attributable costs, costs by category may be misrepresented. For example, if cardiovascular diseases represented 30% of unattributable costs in 2008 and now suddenly we could attribute all these costs, cardiovascular diseases could surpass neuropsychiatric conditions as the most costly diagnostic category. However, if the unattributable and attributable cost distributions were similar, then EBIC hospital care expenditures would reflect the true distribution of the economic burden.
5.3 Methodological Limitations
There are many health conditions that cause secondary health conditions, which themselves result in hospitalization. Secondary health conditions may also contribute to the development of more severe (primary) health conditions, which require hospitalization. Furthermore, secondary health conditions may increase hospital resource utilization and LOS even if they are not the main reason for the hospital visit. Comorbidity refers to the presence of one or more diseases/disorders in addition to a primary disease/disorder, as well as to the effect these secondary diseases/disorders may have. Examples of common comorbidities are diabetes and hypertension (high blood pressure). CIHI’s calculation of RIW values involved a comorbidity factor that considered certain comorbidities to increase hospital costs anywhere from 25% to 125% (31,32). Although RIWs were used to estimate record-level expenditures (for most hospital types/services), all EBIC hospital care expenditures were attributed solely to the most responsible health condition. Since EBIC hospital care expenditures were not attributed to comorbid conditions, costs may be underestimated for certain conditions. Each discharge abstract, in the majority of hospital databases, contained information on comorbidities. Future editions of EBIC would benefit from the development of methods to weight expenditures across primary and comorbid conditions.
EBIC 2004-2008 hospital care expenditures were estimated by diagnostic category/subcategory, sex, age group and province/territory. In 2008, these expenditures were attributed to the EBIC categories for 79.2% of total hospital care expenditures. The three diagnostic categories with the highest expenditures were neuropsychiatric conditions (11.2%), cardiovascular diseases (10.3%) and injuries (6.9%). Females accounted for just over half (51.3%) of 2008 hospital care expenditures. EBIC 2008 hospital care expenditures were lowest and highest for individuals aged 0-14 years (8.1%) and 75+ years (29.4%) respectively.
Given the changes in methods, it is not recommended that comparisons be made between 2004 and 2008 estimates and those from previous EBIC editions. The main limitation in the current edition was incomplete hospital separation data for certain hospital types/services. In these cases, distributions for provinces with complete hospital separation data (100% submissions) were used to distribute cost totals for the other provinces/territories with incomplete data. The availability of complete data for all provinces/territories and hospital types/services would likely increase the accuracy of future EBIC estimates.
- Date modified: