Supplemental Digital Content is Available in the Text. High-cost users among adults with back pain account for approximately 50% of healthcare spending and are associated with older age, lower income, comorbidities, and fair/poor general health.
Keywords: Back pain, Healthcare costs, High-cost users, Population-based cohort study, Canadian Community Health Survey
Abstract
Some patients with back pain contribute disproportionately to high healthcare costs; however, characteristics of high-cost users with back pain are not well defined. We described high-cost healthcare users based on total costs among a population-based cohort of adults with back pain within the Ontario government's single-payer health system across sociodemographic, health, and behavioural characteristics. We conducted a population-based cohort study of Ontario adult (aged 18 years or older) respondents of the Canadian Community Health Survey (CCHS) with back pain (2003-2012), linked to administrative data (n = 36,605; weighted n = 2,076,937, representative of Ontario). Respondents were ranked based on gradients of total healthcare costs (top 1%, top 2%-5%, top 6%-50%, and bottom 50%) for 1 year following the CCHS survey, with high-cost users as top 5%. We used multinomial logistic regression to investigate characteristics associated with the 4 cost groups. Top 5% of cost users accounted for 49% ($4 billion CAD) of total healthcare spending, with inpatient hospital care as the largest contributing service type (approximately 40% of costs). Top 5% high-cost users were more likely aged 65 years or older (ORtop1% = 16.6; ORtop2-5% = 44.2), with lower income (ORtop1% = 3.6; ORtop 2-5% = 1.8), chronic disease(s) (ORtop1% = 3.8; ORtop2-5% = 1.6), Aggregated Diagnosis Groups measuring comorbidities (ORtop1% = 25.4; ORtop2-5% = 13.9), and fair/poor self-rated general health (ORtop1% = 6.7; ORtop2-5% = 4.6) compared with bottom 50% users. High-cost users tended to be current/former smokers, obese, and report fair/poor mental health. High-cost users (based on total costs) among adults with back pain account for nearly half of all healthcare spending over a 1-year period and are associated with older age, lower income, comorbidities, and fair/poor general health. Findings identify characteristics associated with a high-risk group for back pain to inform healthcare and public health strategies that target upstream determinants.
1. Introduction
Back pain is a major driver of disability, healthcare utilization, and costs.12,14,17 Across health systems, healthcare utilization and costs are concentrated within a small proportion of the population,10,30 and small subgroups account for a disproportionate amount of care for back pain.16,20,27,40 In the United States, two-thirds of patients with back pain had 1 to 2 back pain visits over 5 years, while approximately 10% had heavy utilization of imaging, medication, emergency department visits, and inpatient encounters (eg, surgery and ≥3 opioid prescriptions).40 However, high-cost users for back pain are not well defined concerning demographic, socioeconomic, health, and behavioural characteristics. Comprehensively characterizing high-cost users with back pain, particularly with upstream and potentially modifiable determinants, can inform healthcare and public health strategies to improve population health and healthcare sustainability.
High-cost users of healthcare services are defined as top 5% according to the Ontario Ministry of Health and Long-term Care (MOHLTC).25 Few studies have investigated factors associated with high-cost users (previous studies examining top 1-25%)4,9,19,22 among adults with back pain. Disability, depression, or poor health characteristics were associated with high costs in patients with back pain in Germany and the Netherlands.22 In the United States, pain persistence and depression were predictors of high healthcare costs in patients with back pain.9 In Norway and the Netherlands, pain severity, disability, depression, and worse health-related quality of life were associated with high-cost users among older adults (aged 55 years or older) with back pain.19 However, these studies have limited generalizability in the context of cost analyses not conducted in real-world settings (eg, highly selected samples),4,22 a focus on adults aged 55 years or older only,19 or members of 1 US health insurance plan only (not a publicly funded system).9 Previous studies focused on clinical or health-related characteristics, and studies are needed to comprehensively examine upstream determinants, including behavioural factors. To our knowledge, no population-based studies have characterized high-cost users among adults with back pain within Canada's universal public healthcare system.
Understanding high-cost users across upstream determinants enables a broader perspective on health status and resource use among people with back pain. This can help decision-makers (eg, governments and health professional associations) identify higher-risk groups and tailor public health strategies to target upstream determinants to mitigate risk. Our objective was to describe high-cost healthcare users (top 1%-5%) based on total healthcare spending among a population-based cohort of Ontario adults with back pain in a government single-payer health system across sociodemographic, health, and behavioural characteristics.
2. Methods
We conducted a population-based cohort study of Ontario adult respondents of the Canadian Community Health Survey (CCHS) with back pain (2003-2012). This study was reported using the Strengthening the Reporting of Observational Studies in Epidemiology Statement.41 This project has been approved by the Health Sciences Research Ethics Board at the University of Toronto (Reference #37424).
2.1. Study sample
We included all Ontario respondents (aged 18 years or older) of at least 1 of 5 CCHS cycles (2003-2012) with back pain. Back pain was based on responding “yes” to the question, “Do you have back problems, excluding fibromyalgia and arthritis?” We excluded respondents who could not be linked with administrative databases (linkage rates 81%-85%); had death date preceding survey date (0.01% excluded); appeared in multiple CCHS cycles (first cycle used; <1% excluded); or were ineligible for Ontario Health Insurance Plan (OHIP) for the observation window (0.1% excluded). Ontario is the largest province by population (approximately 14.6 million in 2019) and most ethnically diverse province in Canada.36
2.2. Data sources
Individual-level responses from the CCHS survey data were deterministically linked to healthcare utilization data from population-based health administrative databases in Ontario.29 These data sets were linked using unique encoded identifiers and analyzed at ICES. ICES is an independent, nonprofit research institute whose legal status under Ontario's health information privacy law allows it to collect and analyze health care and demographic data, without consent, for health system evaluation and improvement.
The CCHS is a cross-sectional survey administered by Statistics Canada that collects self-reported data on health determinants, outcomes, and healthcare use in Canada.38 The CCHS uses a multistage sampling survey design to target Canadians aged 12 years or older living in private dwellings and excludes persons living in institutions (eg, long-term care or complex continuing care facilities), full-time members of the Canadian Forces, and persons living on reserve and other First Nation settlements.38 We restricted to respondents aged 18 years or older to focus on adults with back pain. Detailed survey methodology is documented by Statistics Canada.37 The CCHS data are representative of 98% of the Canadian population aged 12 years or older living in private dwellings at national and provincial levels.38 All Ontario residents, including CCHS respondents, are covered by a single-payer public healthcare system (OHIP), and healthcare utilization data are tracked in administrative databases (used to generate costs for total healthcare spending). Most healthcare services are publicly funded through the government-run provincial health insurance plan (OHIP), including family physician and specialist visits and basic and emergency healthcare services.23
2.3. Independent variables
We assessed the following characteristics based on CCHS data at baseline: sociodemographic (self-reported): age (years), sex (female, male), location (urban, rural), household income (income quintiles), education (less than secondary, secondary graduate, and more than secondary), ethnicity (white, visible minority), and newcomer status (Canadian born, newcomer); behavioural (self-reported): smoking (current, former, and never smoker), alcohol consumption [light/never drinker: <3 drinks/week (male individuals) or <2 drinks/week (female individuals); moderate/heavy: ≥3 drinks/week (male individuals) or ≥2 drinks/week (female individuals)], physical activity (active, moderately active, and inactive), life stress (quite a bit/extreme, a bit, and not very/not at all); health (self-reported): self-perceived general health (excellent-good, fair/poor), self-perceived mental health (excellent-good, fair/poor), body mass index (obese ≥30 kg/m2), overweight 25.00 to 29.99 kg/m2, normal weight 18.50 to 24.99 kg/m2, underweight <18.50 kg/m2).
Life stress was measured based on the CCHS question, “Thinking about the amount of stress in your life, would you say that most days are: not at all stressful, not very stressful, a bit stressful, quite a bit stressful, or extremely stressful?” Physical activity was measured using total daily energy expenditure values (kcal/kg/day) based on the Physical Activity Index. Self-perceived general health was measured based on the CCHS question, “In general, would you say your health is…excellent, very good, good, fair, or poor?” This question refers to the perception of a person's health in general and is considered an indicator of overall health status.1,7,39 Self-perceived mental health was measured using the CCHS question, “In general, would you say your mental health is…excellent, very good, good, fair or poor?” This question has been validated as a measure of general mental health.21
We measured prior healthcare utilization and comorbidities by calculating ACG System Aggregated Diagnosis Groups (ADGs) using The Johns Hopkins ACG System version 10.0.1 (Johns Hopkins HealthCare, LLC; https://www.hopkinsacg.org/) using 2-year lookback window from the CCHS interview date. The Johns Hopkins ADGs provide a numeric method for grouping diagnostic codes similar for severity and likelihood of persistence and has been validated in the Ontario adult population.2 We used administrative database algorithms to ascertain comorbidities (diabetes, hypertension, congestive heart failure, chronic obstructive pulmonary disease, dementia, stroke, and coronary artery disease).8,11,18,35,43
2.4. Outcome: high-cost users
We defined high-cost healthcare users as top 5% of users based on total costs, according to the definition provided by the Ontario MOHLTC.25 We calculated total healthcare costs based on healthcare utilization for 1 year following interview date and ranked individuals according to gradients of total cost within the cohort. We ranked according to 1, 2 to 5, 6 to 50, and lower 50th percentiles to build 4 cost-ranking groups, with high-cost users as top 5%. Healthcare costs represent total healthcare spending, adjusted to 2018 Canadian dollars (CAD), using a person-centred costing approach developed at ICES and linked to administrative databases (Appendix 1, http://links.lww.com/PAIN/C17).44 This uses an algorithm to estimate person-level costs based on healthcare visits covered by the Ontario MOHLTC (healthcare-payer perspective). We focused on total healthcare spending rather than back pain–specific costs; back pain visits identified using billing codes in administrative data are prone to misclassification bias.45 Studies used this methodology to compute costs attributed to conditions including back pain, diabetes, and general population.5,30,33,44,46,47
Costing methodology computes cumulative individual-level healthcare costs for all publicly funded health system encounters over time. The methodology focuses on the formal component of direct healthcare costs and excludes copayments, costs associated with caregivers, private insurance, overheads and capital expenditures, and community-level services where an individual's health card number is not tracked. Indirect costs related to the provision of healthcare services (eg, transportation) were excluded. Comprehensive healthcare costs were available for all major healthcare sectors from 2003 and onwards, including inpatient hospitalizations, physician visits, imaging, complex continuing care, long-term care, home services, assistive devices, and pharmaceuticals. Other healthcare utilization (eg, physiotherapy, chiropractic treatment, and massage therapy) paid through extended health insurance, workers' compensation, auto-insurance, or out-of-pocket were also excluded.
2.5. Analysis
We used multinomial logistic regression to quantify the association between sociodemographic, health-related, and behavioural factors and gradients of total costs (4 cost-ranking groups: top 1%, top 2%-5%, top 6%-50%, and bottom 50%). We built multinomial models to measure the crude association between each independent variable and cost gradients and associations when accounting for age. We built a multivariable model with all stated factors as independent variables. We estimated odds for each of top 1%, top 2%-5%, and top 6%-50% groups compared with bottom 50% as the reference group. We selected a nominal multinomial regression model because we hypothesized differential associations between stated factors and cost gradients.
We tested for the proportional odds assumption to assess the appropriateness of treating the outcome as categorical compared with ordinal. A P < 0.05 indicates that the proportional odds assumption was violated (response should be treated as ordinal).6
We applied sampling and bootstrap weights provided by Statistics Canada to adjust for the complex survey design of the CCHS and produce population estimates for Ontario.26,34 We used a pooled approach to combine CCHS cycles for increased sample size and statistical power.42 All healthcare costs were adjusted to 2018 Canadian dollars, and the annual exchange rate for 1 Canadian dollar was $0.77 USD in 2018.3 Analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC).
We conducted a sensitivity analysis to assess the impact of potential misclassification of high-cost users in adults with back pain, using alternative definition of high-cost users as top 10% instead of top 5%. We also conducted univariable and multivariable models stratified by sex to examine sex-specific characteristics associated with high-cost users.
3. Results
There were 168,074 respondents of CCHS cycles 2003 to 2012 linked to administrative data (Appendix 2, http://links.lww.com/PAIN/C17). After applying exclusions [primarily younger than 18 years of age, appeared in multiple CCHS cycles (kept first cycle), or no self-reported back pain], our cohort included 36,605 adult respondents with self-reported back pain (weighted n = 2,076,937).
The overall sample was 31% aged 35 to 39 years, 30% aged 50 to 64 years, and 21% aged 65 years or older, and 54% were female (Table 1). Compared with bottom 50% cost users, top 1 to 5% high-cost users had a higher proportion of those aged 65 years or older, had less than secondary education, had lowest income quintile, were physically inactive, experienced chronic comorbidities, had higher mean ADG score, were obese, and exhibited fair/poor general health.
Table 1.
Weighted* distribution of demographic, socioeconomic, health and behavioural characteristics across total healthcare expenditure categories using the Canadian Community Health Survey cycles 2003 to 2012 for adults with back pain in Ontario, Canada.†‡
| Characteristic | Overall % (95% CI) |
Top 1% of total costs % (95% CI) |
Top 2%-5% of total costs % (95% CI) |
Top 6%-50% of total costs % (95% CI) |
Bottom 50% of total costs % (95% CI) |
|---|---|---|---|---|---|
| Total population | N = 2,076,937 | N = 20,185 | N = 81,318 | N = 937,800 | N = 1,037,634 |
| Demographics and socioeconomics | |||||
| Sex (female) | 54.0 (53.2-54.8) | 54.7 (46.1-63.3) | 57.7 (53.7-61.7) | 61.7 (60.5-62.8) | 46.7 (45.5-47.9) |
| Age group (y) | |||||
| 18-34 | 17.6 (16.9-18.2) | 6.3 (1.8-10.9) | 4.5 (2.5-6.6) | 10.9 (10.1-11.7) | 24.8 (23.8-25.9) |
| 35-49 | 30.6 (29.8-31.5) | 11.7 (6.1-17.3) | 11.8 (9.2-14.5) | 22.6 (21.4-23.8) | 39.7 (38.4-41.0) |
| 50-64 | 30.3 (29.5-31.2) | 31.4 (22.4-40.4) | 29.3 (25.4-33.3) | 30.7 (29.5-32.0) | 30.1 (28.9-31.2) |
| ≥65 | 21.5 (20.9-22.0) | 50.6 (42.0-59.2) | 54.3 (50.3-58.4) | 35.8 (34.8-36.8) | 5.4 (4.9-5.8) |
| Location of residence | |||||
| Rural | 16.6 (16.0-17.2) | 14.9 (10.2-19.7) | 17.0 (14.5-19.4) | 15.5 (14.8-16.3) | 17.6 (16.7-18.4) |
| Ethnicity | |||||
| White | 83.9 (83.0-84.7) | 86.5 (79.2-93.7) | 88.4 (84.6-92.2) | 84.4 (83.2-85.6) | 83.0 (81.7-84.2) |
| Visible minority | 16.1 (15.3-17.0) | 13.6 (6.3-20.8) | 11.6 (7.8-15.4) | 15.6 (14.5-16.8) | 17.0 (15.7-18.3) |
| Newcomer status | |||||
| Canadian born | 70.8 (69.8-71.7) | 69.1 (60.2-77.9) | 68.8 (64.4-73.2) | 67.9 (66.5-69.3) | 73.5 (72.2-74.9) |
| Newcomer | 29.2 (28.3-30.2) | 31.0 (22.1-39.9) | 31.2 (26.9-35.6) | 32.1 (30.8-33.5) | 26.5 (25.1-27.8) |
| Education | |||||
| Less than secondary | 18.8 (18.1-19.5) | 33.7 (25.9-41.6) | 35.2 (31.2-39.1) | 23.3 (22.3-24.4) | 13.1 (12.2-14.0) |
| Secondary graduate | 18.6 (17.9-19.3) | 14.8 (10.2-19.5) | 17.1 (14.1-20.1) | 17.9 (16.9-18.9) | 19.3 (18.3-20.4) |
| More than secondary | 62.7 (61.8-63.6) | 51.5 (43.2-59.7) | 47.7 (43.5-52.0) | 58.8 (57.5-60.1) | 67.6 (66.3-68.9) |
| Household income quintiles | |||||
| 1 (lowest) | 18.8 (18.0-19.6) | 40.8 (32.1-49.4) | 30.5 (26.8-34.3) | 23.7 (22.4-24.9) | 13.2 (12.2-14.2) |
| 2 | 16.8 (16.1-17.5) | 22.8 (14.2-31.3) | 21.5 (18.2-24.8) | 18.6 (17.6-19.7) | 14.7 (13.7-15.7) |
| 3 | 20.0 (19.3-20.8) | 16.6 (11.4-21.8) | 19.4 (16.1-22.7) | 19.8 (18.8-20.9) | 20.3 (19.3-21.4) |
| 4 | 21.9 (21.2-22.7) | 13.7 (6.8-20.5) | 16.9 (13.1-20.7) | 20.0 (18.9-21.1) | 24.2 (23.1-25.3) |
| 5 (highest) | 22.4 (21.6-23.2) | 6.2 (3.0-9.4) | 11.7 (8.8-14.7) | 17.8 (16.8-18.9) | 27.6 (26.4-28.9) |
| Health status | |||||
| Has chronic disease | 44.6 (43.7-45.4) | 79.7 (72.3-87.1) | 74.0 (69.9-78.1) | 59.2 (57.9-60.6) | 28.3 (27.2-29.5) |
| Mean ADG score | 7.3 (7.1-7.5) | 21.6 (19.2-24.0) | 19.1 (17.8-20.4) | 10.8 (10.5-11.1) | 3.0 (2.8-3.2) |
| Body mass index (kg/m2) | |||||
| Obese (≥30.00) | 22.1 (21.3-22.8) | 26.0 (17.9-34.0) | 28.7 (24.8-32.6) | 26.1 (24.9-27.3) | 18.0 (17.1-18.9) |
| Overweight (25.00-29.99) | 36.4 (35.5-37.3) | 35.8 (26.6-44.9) | 32.7 (28.5-36.9) | 37.0 (35.6-38.4) | 36.2 (34.9-37.4) |
| Normal weight (18.50-24.99) | 39.4 (38.5-40.3) | 34.0 (26.0-42.0) | 36.2 (32.2-40.1) | 34.9 (33.6-36.3) | 43.5 (42.2-44.9) |
| Underweight (<18.50) | 2.2 (1.9-2.5) | 4.2 (1.7-6.8) | 2.5 (1.6-3.3) | 2.0 (1.7-2.3) | 2.4 (1.9-2.8) |
| Self-perceived general health | |||||
| Excellent/very good/good | 75.6 (74.9-76.3) | 33.7 (25.6-41.7) | 39.4 (35.6-43.2) | 66.7 (65.5-67.9) | 87.3 (86.4-88.2) |
| Fair/poor | 24.4 (23.7-25.2) | 66.4 (58.3-74.4) | 60.6 (56.8-64.4) | 33.3 (32.1-34.5) | 12.7 (11.9-13.6) |
| Self-perceived mental health | |||||
| Excellent/very good/good | 89.8 (89.2-90.3) | 84.2 (78.9-89.6) | 83.5 (80.8-86.3) | 86.7 (85.8-87.7) | 93.0 (92.3-93.6) |
| Fair/poor | 10.3 (9.7-10.8) | 15.8 (10.4-21.1) | 16.5 (13.7-19.2) | 13.3 (12.3-14.2) | 7.0 (6.4-7.7) |
| Health behaviour | |||||
| Smoking status | |||||
| Current smoker | 26.1 (25.3-26.9) | 25.0 (18.1-32.0) | 24.9 (21.2-28.7) | 22.6 (21.5-23.7) | 29.4 (28.2-30.6) |
| Former smoker | 28.0 (27.2-28.8) | 32.0 (24.0-39.9) | 33.2 (29.5-36.9) | 31.8 (30.5-33.0) | 24.2 (23.1-25.2) |
| Never smoker | 45.9 (45.0-46.8) | 43.0 (34.0-52.0) | 41.9 (37.5-46.3) | 45.6 (44.3-47.0) | 46.5 (45.1-47.8) |
| Physical activity | |||||
| Active | 21.0 (20.2-21.7) | 13.8 (8.5-19.1) | 14.5 (11.3-17.6) | 18.4 (17.3-19.4) | 23.8 (22.7-24.9) |
| Moderately active | 23.7 (22.9-24.5) | 18.8 (11.3-26.3) | 15.7 (12.6-18.9) | 22.5 (21.4-23.6) | 25.4 (24.2-26.6) |
| Inactive | 55.4 (54.4-56.3) | 67.4 (59.1-75.8) | 69.8 (65.8-73.9) | 59.1 (57.9-60.4) | 50.8 (49.4-52.1) |
| Alcohol consumption | |||||
| Moderate/heavy drinker | 29.2 (28.3-30.0) | 12.4 (7.8-16.9) | 20.6 (17.0-24.3) | 23.5 (22.4-24.6) | 35.2 (33.9-36.5) |
| Light/never drinker | 70.9 (70.0-71.7) | 87.6 (83.1-92.2) | 79.4 (75.7-83.1) | 76.5 (75.4-77.6) | 64.8 (63.5-66.1) |
| Life stress | |||||
| Quite a bit/extreme stress | 31.2 (30.4-32.1) | 35.9 (27.8-44.0) | 32.4 (28.5-36.4) | 31.4 (30.2-32.7) | 30.9 (29.6-32.1) |
| A bit of stress | 33.4 (32.6-34.3) | 26.5 (18.9-34.0) | 33.1 (29.2-37.1) | 32.2 (31.0-33.5) | 34.7 (33.4-35.9) |
| Not very/not at all | 35.4 (34.6-36.2) | 37.6 (29.4-45.9) | 34.4 (30.8-38.1) | 36.4 (35.2-37.5) | 34.5 (33.3-35.7) |
Data were derived from the Ontario components of Canadian Community Health Survey (2003-2012) linked to health administrative databases. All estimates were weighted using Canadian Community Health Survey sampling weights to provide population estimates.
Percentages represent “percent responded.”
Healthcare expenditure ($CAD) calculated for the year following Canadian Community Health Survey interview.
ADGs, Aggregated Diagnosis Groups.
3.1. Distribution of total healthcare spending
High-cost users accounted for the greatest proportion of total healthcare spending among adults with back pain in Ontario. Top 1% accounted for 21% of total healthcare spending, corresponding to $1.8 billion CAD in healthcare costs. Top 2%-5% accounted for 28% of total healthcare spending, corresponding to $2.3 billion CAD in healthcare costs. Overall, combined top 1%-5% of cost users accounted for a total of 49% of total healthcare spending (∼$4 billion CAD). By contrast, bottom 50% accounted for <5% of total spending (∼$400 million CAD). When stratified by healthcare services, the largest contributor for top 1% was inpatient hospital care, accounting for 41% of total costs (Fig. 1). For top 2%-5%, the largest contributor was inpatient hospital care at 39% of total costs. Bottom 50% group did not incur any costs for inpatient hospital care, and expenditures were mainly physician services (68%). By contrast, physician services accounted for 11% and 18% of total costs for top 1% and top 2%-5%, respectively.
Figure 1.
Distribution of total healthcare spending in adults with back pain in Ontario, Canada. (A) The proportion of total healthcare spending incurred by each user group. (B) Average (per-person) spending for total costs across service types by user group.
3.2. Average (per-person) expenditure
For the overall weighted sample, the average per-person spending was $3966 CAD (95% CI: 3795-4137). The average spending was $398 CAD (95% CI: 390-405) for bottom 50% (Table 2), while top 1% incurred on average $87,014 CAD (95% CI: 81,142-92,886). Highest average costs for top 1% corresponded to inpatient hospital care (average per-person spending: $35,275 CAD), followed by physician services ($9672 CAD per-person). Highest average costs for top 2%-5% corresponded to inpatient hospital care ($10996 CAD per-person), followed by physician services ($5102 CAD per-person). Bottom 50% group had physician services as the largest contributor to costs ($9270 CAD per person).
Table 2.
Average (per-person) expenditure across healthcare service types for the weighted* sample of adults with back pain in Ontario using the Canadian Community Health Survey cycles 2003 to 2012†.
| Service type | Overall, $CAD (95% CI) | Top 1%, $CAD (95% CI) | Top 2-5%, $CAD (95% CI) | Top 6-50%, $CAD (95% CI) | Bottom 50%, $CAD (95% CI) |
|---|---|---|---|---|---|
| Total population | N = 2,076,937 | N = 20,185 | N = 81,318 | N = 937,800 | N = 1,037,634 |
| Physician services | 1028 (996-1060) | 9547 (8566-10528) | 4980 (4724-5236) | 1408 (1373-1442) | 209 (204-215) |
| Rehabilitation | 80 (62-99) | 4590 (3100-6081) | 915 (604-1226) | 0 | 0 |
| Long-term care | 42 (27-57) | 2099 (925-3274) | 526 (274-778) | 2 (0-4) | 0 |
| Emergency department | 151 (144-157) | 1613 (1396-1829) | 867 (775-960) | 190 (182-199) | 30 (28-33) |
| Nonphysician services‡ | 30 (28-32) | 104 (51-158) | 91 (62-119) | 42 (39-45) | 13 (12-14) |
| Ontario Drug Benefit§ | 558 (523-593) | 5619 (3232-8007) | 2855 (2574-3135) | 853 (815-891) | 13 (12-15) |
| Complex continuing care | 44 (26-62) | 3586 (1717-5456) | 230 (117-343) | 1 (0-2) | 0 |
| Physician services – capitation‖ | 83 (81-85) | 125 (99-150) | 122 (111-133) | 102 (99-106) | 61 (59-64) |
| Day surgery | 163 (153-174) | 982 (470-1494) | 744 (611-877) | 274 (259-290) | 1 (1-2) |
| Inpatient hospital care | 953 (864-1042) | 35,275 (29080-41470) | 10,996 (10368-11624) | 398 (366-429) | 0 |
| Outpatient hospital care | 261 (244-279) | 4170 (3134-5206) | 1594 (1456-1731) | 342 (325-360) | 8 (6-9) |
| Home care | 246 (218-274) | 6344 (4521-8166) | 2894 (2482-3306) | 157 (139-175) | 0 |
| All services | 3966 (3795-4137) | 87,014 (81142-92886) | 28,356 (27560-29152) | 4012 (3930-4094) | 398 (390-405) |
Data were derived from the Ontario components of Canadian Community Health Survey (2003-2012) linked to health administrative databases. All estimates were weighted using Canadian Community Health Survey sampling weights to provide population estimates.
Healthcare expenditure ($CAD) calculated for the year following Canadian Community Health Survey interview.
Nonphysician services include optometrists, physiotherapists, etc. for covered individuals (Ontarians aged 65 years or older, those with specific chronic diseases, and those in specific government assistance programs).
Individuals eligible for Ontario Drug Benefit include Ontarians aged 65 years or older, those receiving Ontario Works (a financial assistance program), those on the Ontario Disability Support Program, or those living in long-term care.
Physicians in the capitation system work in family health teams/organizations and receive a flat rate fee for each patient served regardless of the number or type of services provided.
3.3. Proportion of cost-rank groups using each healthcare service
Nearly all (99.6%-100%) adults with back pain in top 1%-50% cost groups saw a physician (noncapitation physical services) during the follow-up year, compared with 80% of the bottom 50% group (Table 3). Less than 2% of the bottom 50% group accessed rehabilitation, long-term care, complex continuing care, day surgery, hospital care, or home care in the follow-up year. By contrast, a considerable proportion of the top 1% group accessed rehabilitation (17%), long-term care (13%), complex-continuing care (9%), inpatient hospital care (88%), outpatient hospital care (83%), day surgery (34%), or home care (78%). The top 2%-5% group accessed these services similar to top 1% group but to a lesser extent.
Table 3.
Proportion of the weighted* Canadian Community Health Survey sample, according to total cost-rank groups, using each healthcare service type, using the Canadian Community Health Survey cycles 2003 to 2012†.
| Proportion of the population or cost-rank group using healthcare sector (%) | |||||
|---|---|---|---|---|---|
| Sector | Overall % (95% CI) |
Top 1% % (95% CI) |
Top 2-5% % (95% CI) |
Top 6-50% % (95% CI) |
Bottom 50% % (95% CI) |
| Total population | N = 2,076,937 | N = 20,185 | N = 81,318 | N = 937,800 | N = 1,037,634 |
| Physician services | 90.0 (89.4-90.5) | 100 | 99.9 (99.8-100.0) | 99.6 (99.4-99.7) | 80.3 (79.3-81.4) |
| Rehabilitation | 0.5 (0.4-0.6) | 17.2 (12.6-21.9) | 7.8 (4.8-10.8) | 0 | 0 |
| Long-term care | 0.3 (0.2-0.4) | 12.9 (7.0-18.7) | 3.2 (2.0-4.3) | 0.1 (0.0-0.1) | 0 |
| Emergency department | 26.9 (26.1-27.7) | 87.6 (81.6-93.7) | 69.2 (65.2-73.3) | 36.5 (35.3-37.8) | 13.7 (12.8-14.5) |
| Nonphysician services‡ | 24.5 (23.9-25.2) | 33.3 (25.6-41.0) | 39.2 (35.4-43.0) | 34.4 (33.2-35.6) | 14.3 (13.5-15.1) |
| Ontario drug benefit§ | 31.2 (30.5-32.0) | 90.4 (86.1-94.6) | 78.7 (75.0-82.3) | 51.6 (50.3-52.9) | 8.0 (7.4-8.6) |
| Complex continuing care | 0.2 (0.1-0.2) | 9.4 (5.9-12.8) | 2.2 (1.2-3.1) | 0 | 0 |
| Physician services—capitation‖ | 59.8 (58.9-60.6) | 66.6 (58.4-74.8) | 67.2 (63.8-70.6) | 65.3 (64.0-66.5) | 54.0 (52.7-55.3) |
| Day surgery | 12.9 (12.3-13.5) | 33.8 (24.4-43.3) | 35.0 (31.1-39.0) | 24.4 (23.3-25.5) | 0.4 (0.3-0.5) |
| Inpatient hospital care | 7.9 (7.5-8.4) | 88.2 (82.6-93.8) | 81.5 (78.5-84.4) | 8.6 (8.0-9.3) | 0 |
| Outpatient hospital care | 20.8 (20.1-21.6) | 83.3 (77.8-88.7) | 69.2 (65.9-72.6) | 36.2 (34.9-37.5) | 1.9 (1.6-2.3) |
| Home care | 5.7 (5.4-6.1) | 78.4 (72.2-84.6) | 56.5 (52.5-60.5) | 6.1 (5.6-6.6) | 0 |
| Any service | 96.4 (96.1-96.7) | 100 | 100 | 100 | 92.8 (92.1-93.4) |
Data were derived from the Ontario components of Canadian Community Health Survey (2003-2012) linked to health administrative databases. All estimates were weighted using Canadian Community Health Survey sampling weights to provide population estimates.
Healthcare expenditure ($CAD) calculated for the year following Canadian Community Health Survey interview.
Nonphysician services include optometrists, physiotherapists, etc. for covered individuals (Ontarians aged 65 years or older, those with specific chronic diseases, and those in specific government assistance programs).
Individuals eligible for Ontario Drug Benefit include Ontarians aged 65 years or older, those receiving Ontario Works (a financial assistance program), those on the Ontario Disability Support Program, or those living in long-term care.
Physicians in the capitation system work in family health teams/organizations and receive a flat rate fee for each patient served regardless of the number or type of services provided.
3.4. Factors associated with high-cost users based on total costs
In multinomial regression models, the proportional odds assumption was tested and found to be violated (P < 0.05). We treated the outcome of cost users as categorical rather than ordered.
3.4.1. Demographic/socioeconomic characteristics
Female adults (versus male adults) had 1.25 times higher odds of being in top 6%-50% than in bottom 50% in the multivariable model (95% CI: 1.12-1.38) (Table 4). Across sociodemographic factors, age had the strongest association with high-cost users. In the multivariable model, individuals aged 65 years or older (versus 18-34 years of age) had at least 16-fold higher odds of being in top 1%-5% than in bottom 50% (ORtop1% = 16.55, ORtop2-5% = 44.20). Those in the lowest income quintile had at least 1.8-fold higher odds of being in top 1%-5% than in bottom 50% users (ORtop1% = 3.64, ORtop2-5% = 1.82). We observed similar associations for the top 6%-50% group concerning older age and lowest income quintile, but effects were reduced compared with high-cost users. No associations were observed for rurality, ethnicity, newcomer status, or education.
Table 4.
Weighted* unadjusted, age-adjusted, and adjusted odds ratio (and corresponding 95% confidence intervals) according to multinomial logistic regression (reference group = bottom 50% users) using the Canadian Community Health Survey cycles 2003 to 2012†.
| Characteristic | Top 1% | Top 2%-5% | Top 6%-50% | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Crude association | Association when accounting for age | Multivariable model‡ | Crude association | Association when accounting for age | Multivariable model‡ | Crude association | Association when accounting for age | Multivariable model‡ | |
| Demographics and socioeconomics | |||||||||
| Sex | |||||||||
| Female | 1.38 (0.97-1.95) | 1.37 (0.96-1.94) | 0.66 (0.42-1.04) | 1.56 (1.32-1.85) | 1.54 (1.29-1.85) | 0.90 (0.72-1.13) | 1.84 (1.71-1.97) | 1.83 (1.69-1.99) | 1.25 (1.12-1.38) |
| Male | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) |
| Age group (y) | |||||||||
| 18-34 | Ref (1.00) | — | Ref (1.00) | Ref (1.00) | — | Ref (1.00) | Ref (1.00) | — | Ref (1.00) |
| 35-49 | 1.16 (0.42-3.17) | — | 1.28 (0.35-4.70) | 1.63 (0.93-2.86) | — | 2.49 (1.37-4.53) | 1.30 (1.15-1.47) | — | 1.38 (1.18-1.62) |
| 50-64 | 4.09 (1.57-10.66) | — | 2.67 (0.81-8.81) | 5.35 (3.17-9.02) | — | 5.73 (3.21-10.23) | 2.33 (2.08-2.61) | — | 2.05 (1.73-2.42) |
| ≥65 | 36.79 (15.06-89.84) | — | 16.55 (4.57-59.99) | 55.38 (33.33-92.04) | — | 44.20 (24.40-80.07) | 15.17 (13.29-17.32) | — | 11.42 (9.33-13.97) |
| Location of residence | |||||||||
| Rural | 0.82 (0.56-1.21) | 0.71 (0.49-1.05) | 0.93 (0.60-1.44) | 0.96 (0.80-1.15) | 0.83 (0.68-1.00) | 1.08 (0.84-1.38) | 0.86 (0.80-0.93) | 0.78 (0.72-0.85) | 0.93 (0.84-1.03) |
| Urban | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) |
| Ethnicity | |||||||||
| White | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) |
| Visible minority | 0.76 (0.39-1.48) | 1.18 (0.60-2.31) | 0.71 (0.20-2.50) | 0.64 (0.43-0.96) | 1.01 (0.67-1.52) | 0.81 (0.50-1.32) | 0.90 (0.80-1.02) | 1.19 (1.04-1.36) | 0.96 (0.78-1.17) |
| Newcomer status | |||||||||
| Canadian born | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) |
| Newcomer | 1.25 (0.81-1.92) | 0.94 (0.60-1.46) | 0.64 (0.31-1.35) | 1.26 (1.01-1.57) | 0.94 (0.75-1.18) | 0.76 (0.57-1.01) | 1.31 (1.20-1.44) | 1.08 (0.98-1.20) | 0.90 (0.77-1.04) |
| Education | |||||||||
| Less than secondary | 3.38 (2.28-4.99) | 1.77 (1.16-2.69) | 0.69 (0.42-1.13) | 3.80 (3.10-4.64) | 1.91 (1.54-2.36) | 1.12 (0.84-1.50) | 2.04 (1.84-2.27) | 1.27 (1.13-1.44) | 0.99 (0.86-1.14) |
| Secondary graduate | 1.01 (0.67-1.51) | 0.91 (0.60-1.37) | 0.68 (0.38-1.20) | 1.25 (0.99-1.58) | 1.12 (0.88-1.43) | 0.89 (0.65-1.23) | 1.07 (0.97-1.17) | 1.00 (0.90-1.11) | 0.87 (0.76-0.99) |
| More than secondary | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) |
| Household income | |||||||||
| Quintile 1 (lowest) | 13.82 (7.93-24.09) | 11.55 (6.46-20.67) | 3.64 (1.86-7.12) | 5.46 (3.96-7.53) | 4.44 (3.22-6.14) | 1.82 (1.18-2.82) | 2.78 (2.44-3.17) | 2.45 (2.13-2.81) | 1.50 (1.26-1.78) |
| Quintile 2 | 6.93 (3.55-13.53) | 4.85 (2.31-10.16) | 2.38 (0.99-5.74) | 3.44 (2.48-4.78) | 2.32 (1.66-3.26) | 1.27 (0.99-5.74) | 1.97 (1.74-2.23) | 1.52 (1.32-1.76) | 1.11 (0.93-1.32) |
| Quintile 3 | 3.65 (2.03-6.55) | 2.86 (1.56-5.21) | 1.58 (0.79-3.18) | 2.25 (1.60-3.15) | 1.72 (1.22-2.42) | 1.01 (0.66-1.55) | 1.51 (1.35-1.70) | 1.28 (1.12-1.45) | 1.04 (0.88-1.22) |
| Quintile 4 | 2.52 (1.18-5.40) | 2.39 (1.10-5.20) | 1.32 (0.55-3.14) | 1.65 (1.13-2.39) | 1.54 (1.04-2.26) | 1.11 (0.70-1.79) | 1.28 (1.15-1.43) | 1.23 (1.09-1.39) | 1.03 (0.89-1.19) |
| Quintile 5 (highest) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) |
| Health status | |||||||||
| Has chronic disease | |||||||||
| Yes | 9.91 (6.12-16.05) | 4.95 (2.88-8.49) | 3.76 (1.99-7.11) | 7.19 (5.74-9.00) | 3.23 (2.55-4.08) | 1.61 (1.19-2.18) | 3.68 (3.38-3.99) | 2.23 (2.03-2.45) | 1.52 (1.34-1.72) |
| No | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) |
| ADG score (quintiles) | |||||||||
| Quintile 1 (lowest) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) |
| Quintile 2 | 1.97 (0.53-7.31) | 1.80 (0.48-6.70) | 3.49 (0.61-20.20) | 2.05 (1.26-3.33) | 1.86 (1.13-3.05) | 1.47 (0.81-2.69) | 2.23 (1.94-2.56) | 2.11 (1.82-2.45 | 1.98 (1.67-2.35) |
| Quintile 3 | 2.25 (0.76-6.63) | 1.91 (0.64-5.66) | 3.32 (0.79-14.01) | 3.73 (2.35-5.91) | 3.14 (1.96-5.01) | 1.87 (1.04-3.35) | 4.08 (3.56-4.68) | 3.68 (3.19-4.25) | 3.24 (2.74-3.83) |
| Quintile 4 | 5.38 (1.94-14.94) | 4.54 (1.63-12.63) | 5.99 (1.47-24.39) | 8.58 (5.47-13.44) | 7.16 (4.53-11.32) | 4.34 (2.48-7.61) | 6.98 (6.04-8.06) | 6.26 (5.37-7.31) | 5.42 (4.55-6.46) |
| Quintile 5 (highest) | 33.71 (12.39-91.70) | 24.64 (9.01-67.33) | 25.38 (6.60-97.52) | 35.38 (22.68-55.17) | 25.28 (15.95-40.08) | 13.91 (8.04-24.06) | 19.38 (16.75-22.42) | 15.74 (13.54-18.29) | 11.32 (9.43-13.59) |
| Body mass index (kg/m2) | |||||||||
| Obese (≥30.00) | 1.85 (1.17-2.93) | 1.82 (1.14-2.90) | 0.80 (0.47-1.36) | 1.93 (1.54-2.40) | 1.90 (1.50-2.40) | 1.13 (0.83-1.55) | 1.81 (1.65-1.99) | 1.79 (1.61-1.99) | 1.27 (1.11-1.45) |
| Overweight (25.00-29.99) | 1.27 (0.82-1.97) | 1.09 (0.70-1.70) | 0.95 (0.57-1.58) | 1.09 (0.88-1.35) | 0.93 (0.74-1.16) | 0.88 (0.68-1.14) | 1.27 (1.16-1.40) | 1.14 (1.03-1.27) | 1.11 (0.98-1.26) |
| Under weight (<18.50) | 2.30 (1.18-4.50) | 2.96 (1.47-5.95) | 1.24 (0.45-3.37) | 1.25 (0.82-1.91) | 1.68 (1.06-2.66) | 0.71 (0.37-1.34) | 1.06 (0.80-1.39) | 1.27 (0.94-1.72) | 0.73 (0.49-1.94) |
| Normal weight (18.50-24.99) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) |
| Self-perceived general health | |||||||||
| Excellent/very good/good | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) |
| Fair/poor | 13.51 (9.34-19.53) | 11.24 (7.81-16.16) | 6.71 (4.04-11.16) | 10.53 (8.85-12.53) | 8.64 (7.16-10.44) | 4.58 (3.58-5.86) | 3.42 (3.10-3.77) | 2.99 (2.69-3.33) | 1.75 (1.53-2.01) |
| Self-perceived mental health | |||||||||
| Excellent/very good/good | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) |
| Fair/poor | 2.48 (1.64-3.75) | 3.44 (2.23-5.33) | 0.96 (0.54-1.73) | 2.61 (2.09-3.26) | 3.76 (2.94-4.81) | 1.12 (0.81-1.56) | 2.02 (1.78-2.30) | 2.51 (2.18-2.89) | 1.41 (1.18-1.69) |
| Health behaviour | |||||||||
| Smoking status | |||||||||
| Current smoker | 0.92 (0.60-1.43) | 1.47 (0.93-2.33) | 1.09 (0.61-1.93) | 0.94 (0.74-1.19) | 1.55 (1.21-2.00) | 1.42 (1.04-1.93) | 0.78 (0.72-0.86) | 1.04 (0.94-1.14) | 1.07 (0.95-1.22) |
| Former smoker | 1.43 (0.94-2.18) | 1.08 (0.70-1.65) | 1.05 (0.63-1.75) | 1.52 (1.25-1.86) | 1.13 (0.92-1.40) | 1.28 (0.98-1.66) | 1.34 (1.22-1.46) | 1.09 (0.98-1.21) | 1.20 (1.05-1.36) |
| Never smoker | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) |
| Physical activity | |||||||||
| Active | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) |
| Moderately active | 1.28 (0.70-2.34) | 1.08 (0.59-1.98) | 0.96 (0.47-1.94) | 1.02 (0.73-1.42) | 0.84 (0.60-1.18) | 0.67 (0.46-0.98) | 1.15 (1.03-1.28) | 1.02 (0.90-1.15) | 0.82 (0.71-0.96) |
| Inactive | 2.30 (1.45-3.64) | 1.89 (1.20-3.00) | 0.92 (0.80-1.04) | 2.27 (1.73-2.97) | 1.82 (1.38-2.41) | 1.16 (0.84-1.59) | 1.51 (1.38-1.66) | 1.31 (1.18-1.46) | 0.92 (0.80-1.04) |
| Alcohol consumption | |||||||||
| Moderate/heavy drinker | 0.26 (0.17-0.39) | 0.26 (0.17-0.39) | 0.44 (0.28-0.71) | 0.48 (0.38-0.60) | 0.47 (0.37-0.59) | 0.81 (0.63-1.04) | 0.57 (0.52-0.62) | 0.56 (0.51-0.61) | 0.78 (0.69-0.87) |
| Light/never drinker | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) |
| Life stress | |||||||||
| Quite a bit/extreme stress | 1.07 (0.71-1.61) | 2.31 (1.50-3.57) | 1.46 (0.78-2.74) | 1.05 (0.86-1.29) | 2.41 (1.90-3.05) | 1.23 (0.92-1.65) | 0.97 (0.88-1.06) | 1.60 (1.44-1.77) | 1.12 (0.99-1.28) |
| A bit of stress | 0.70 (0.45-1.09) | 1.13 (0.71-1.79) | 0.88 (0.48-1.61) | 0.96 (0.78-1.17) | 1.58 (1.27-1.98) | 1.21 (0.93-1.17) | 0.88 (0.81-0.96) | 1.22 (1.11-1.34) | 1.46 (0.78-2.74) |
| Not very/not at all stressed | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) | Ref (1.00) |
Data were derived from the Ontario components of Canadian Community Health Survey (2003-2012) linked to health administrative databases. All estimates were weighted using the Canadian Community Health Survey sampling weights to provide population estimates.
Healthcare expenditure ($CAD) calculated for the year following the Canadian Community Health Survey interview.
Multivariable model that includes all stated sociodemographic, health-related, and behavioural factors as independent variables.
ADGs, Aggregated Diagnosis Groups.
3.4.2. Health-related characteristics
In the multivariable model, individuals with chronic disease(s) had at least 1.6-fold higher odds of being in the top 1%-5% than in the bottom 50% group (ORtop1% = 3.76, ORtop2-5% = 1.61) (Table 4). Individuals in the highest ADG quintile had at least 14-fold higher odds of being in the top 1%-5% than in the bottom 50% group (ORtop1% = 25.38, ORtop2-5% = 13.91). Individuals with fair/poor general health (versus excellent-good) had more than 4.5-fold higher odds of being in the top 1%-50% than in the bottom 50% (ORtop1% = 6.71, ORtop2-5% = 4.58). Similar associations were observed for top 6%-50% users regarding chronic disease, highest ADG quintile, and fair/poor general health, but effects were reduced compared with those for high-cost users. High-cost users tended to report obesity or fair/poor mental health (top 6%-50%).
3.4.3. Behavioural characteristics
High-cost users tended to be current or former smokers (top 1%-50%) compared with those who never smoked (Table 4). In the multivariable model, heavy/moderate alcohol drinkers had lower odds of being in the top 1%-50% group than in the bottom 50% group (OR 0.44-0.81). No associations were observed for physical activity or stress.
3.5. Sensitivity analysis
When high-cost users were defined as top 10%, we observed similar associations with top 5% group across sociodemographic, health-related, and behavioural factors (Appendix 3, http://links.lww.com/PAIN/C17). Results were similar to the primary analysis when examining sex-specific characteristics associated with high-cost users, including associations with older age, lower income, comorbidities, and fair/poor general health among female and male adults (Appendix 4A and 4B, http://links.lww.com/PAIN/C17). A notable addition is that men with quite a bit/extreme stress also tended to be high-cost users.
4. Discussion
Among adults with back pain, top 5% of cost users accounted for 49% (∼$4 billion CAD) of total healthcare spending. Inpatient hospital care was the largest contributing health service to top 5% cost-user group, accounting for approximately 40% of total costs. Top 1%-5% high-cost user status was associated with age 65 years or older, lowest income quintile, chronic disease(s), highest ADG quintile, and fair/poor self-rated general health than the bottom 50% group. High-cost users tended to be current or former smokers, be obese, and report fair/poor mental health. Findings comprehensively characterize high-cost healthcare users (based on total costs) among adults with back pain within a government single-payer health system in the Canadian context.
Findings are consistent with those of prior studies. Studies in Europe reported that poor health characteristics (including poor physical health, disability, and poor health-related quality of life) were associated with high healthcare and societal costs in patients with back pain.4,19,22 In the United States, studies reported that comorbidities (eg, diabetes and depression) were associated with higher healthcare costs in patients with back pain.9,27 Advancing knowledge from previous studies, we examined factors beyond clinical and health-related characteristics to investigate sociodemographic and behavioural factors associated with high-cost users. We found that high-cost users tended to be aged 65 years or older, receive low income, smokers, obese, and have comorbidities and fair/poor health. This provides a broader perspective of high-cost users among adults with back pain, particularly on socioeconomic and behavioural determinants that may be upstream.
Our findings have potential implications. We found that older ages (65 years or older), comorbidities, smoking, and fair/poor self-perceived general health were associated with high costs among adults with back pain. These are consistent with risk factors of back pain in previous literature.24 Decision-makers, including governments and health professional associations, can tailor primary prevention strategies to target upstream determinants. Primary prevention strategies can address prevention or cessation of tobacco use, including support for schools and public health agencies to encourage youth and communities to stay tobacco-free. Primary prevention strategies may focus on poverty reduction strategies and support for low-income families, including addressing broader socioeconomic issues (eg, food security, affordable housing, education, and income/disability assistance).13 Decision-makers can tailor health programming and resource allocation to individuals of older ages (aged 65 years or older) and those with poor health status. Moreover, associations with self-rated general health and mental health highlight the importance of considering the patient perspective for healthcare use.
Future research may consider these findings to help identify populations at risk of becoming high-cost users among adults with back pain to inform prevention strategies. This is a phase 2 study in prognostic factor investigation with multivariable modelling used to identify determinants of high-cost users among adults with back pain.15 Future studies can take the direction of developing and validating a population risk prediction model for predicting high-cost users with back pain based on identified determinants. Rosella et al. developed and validated a population risk tool for predicting high-resource users in the general population within a single-payer health system.13,31,32 Given the substantial burden of back pain, a population risk prediction model specific to back pain may be informative in guiding prevention strategies. Of note, adults with moderate/heavy alcohol consumption were less likely to be high-cost users in our study. There are potential explanations for this observation, which is likely multifactorial. It is possible that there is confounding (eg, concerning socioeconomic status) distorting the relationship between alcohol consumption and high-cost users. There may be measurement error, including capturing some adults who do not consume alcohol due to health conditions. We checked mortality rates, and this is not likely an explanation in our study. Studies suggest that light-to-moderate alcohol consumption may be protective for some conditions, such as cardiovascular disease outcomes.28 Previous literature also found similar associations between alcohol consumption and healthcare costs.26 Future research in this area is warranted.
4.1. Strengths and limitations
Our study has strengths. We linked population-based survey and administrative data to build a population-based cohort of adults with self-reported back pain. The CCHS captures socioeconomic, health, and behavioural factors to more broadly characterize high-cost users among adults with back pain. The CCHS data are representative of 98% of community-dwelling Canadian population aged 12 years or older.38 Data linkage to administrative data allowed for measuring person-level direct costs to generate total healthcare spending within a government single-payer health system. These represent actual costs instead of cost projections or costing approaches limited by recall. We used multinomial logistic regression to further investigate trends across 4 cost-ranking groups.
Our study has limitations. First, CCHS and administrative data were only linked for those who agreed to linkage (81%-85%). However, coverage rates of linkage between CCHS and administrative data are adequate for community-dwelling individuals.43 We also accounted for minor differences by applying survey weights provided by Statistics Canada, which adjust for nonparticipation in survey and linkage. Second, there is potential misclassification of high-cost users based on top 5% of total healthcare costs. This threshold was used in previous studies,22,30 and we examined top 10% in our sensitivity analysis, which produced similar results. There is also potential measurement error with self-reported data, such as the CCHS question on measuring life stress, which has not been validated. Third, healthcare expenditures in our study are limited to those covered by Ontario's universal health insurance plan. Ontario health insurance plan coverage excludes costs including prescription drugs (outside of those received in hospital), allied health care (eg, physiotherapy, chiropractic care, massage therapy), and assistive devices, except for certain eligible members (eg, aged 65 years or older, receiving government assistance, or with specific diseases). Other healthcare utilization paid through extended health insurance, workers' compensation, auto-insurance, or out-of-pocket were also excluded. Fourth, our sample consisted of adults who reported chronic back pain during the CCHS survey. We might have included people who only had a 1-time episode of back pain or those with minor pain/disability who never sought health services because of their back pain (thus, the reported healthcare utilization is related to other conditions).
5. Conclusion
High-cost users among adults with back pain accounted for nearly half of all total healthcare spending over a 1-year period in Ontario. This represents a high concentration of healthcare expenditures within a small proportion of the population with back pain. Among adults with back pain, high-cost users (based on total costs) are associated with older age, lower income, comorbidities, and fair/poor self-rated general health. High-cost users tended to be smokers, obese, and report fair/poor mental health. Findings inform healthcare and public health strategies that target identified upstream determinants in Canada. Decision-makers (governments, health professional associations) can tailor primary prevention and education on back pain to target identified upstream determinants, including public health programs for poverty reduction, tobacco use, education, and social services. Overall, a broader understanding of high-cost users informs policies, health programs, and resource allocation to improve population health in Ontario.
Conflict of interest statement
J.W. reports research grants from the Canadian Institutes of Health Research (CIHR), Social Sciences and Humanities Research Council (SSHRC), and Canadian Chiropractic Research Foundation (paid to university) and travel expenses from Eurospine for teaching and Chiropractic Academy of Research Leadership for research meetings outside the submitted work. P.C. reports research grants from CIHR, Canadian Chiropractic Research Foundation, and College of Chiropractors of British Columbia (paid to university), funding from World Health Organization (paid to university), payment for court testimony from Canadian Chiropractic Protective Association and NCMIC, and travel expenses from Eurospine for teaching and Sophiahemmet University for research collaboration outside the submitted work. The remaining authors declare no conflicts of interest.
Appendix A. Supplemental digital content
Supplemental digital content associated with this article can be found online at http://links.lww.com/PAIN/C17.
Supplementary Material
Acknowledgements
This study was supported by ICES, which is funded by an annual grant from the Ontario Ministry of Health (MOH) and the Ministry of Long-Term Care (MLTC). This document used data adapted from the Statistics Canada Postal CodeOM Conversion File, which is based on data licensed from Canada Post Corporation, and/or data adapted from the Ontario Ministry of Health Postal Code Conversion File, which contains data copied under license from Canada Post Corporation and Statistics Canada. Adapted from Statistics Canada, Canadian Community Health Survey, 2003 to 2012. This does not constitute an endorsement by Statistics Canada of this product. We thank IQVIA Solutions Canada Inc for use of their Drug Information File. Parts of this material are based on data and information compiled and provided by the Ontario Ministry of Health (MOH) and the Canadian Institute for Health Information (CIHI). The analyses, conclusions, opinions, and statements expressed herein are solely those of the authors and do not reflect those of the funding or data sources; no endorsement is intended or should be inferred.
Prior presentations: This research was presented at the Society for Epidemiologic Research (SER) Meeting on June 13 to 16, 2023 and the International Society of Physical and Rehabilitation Medicine (ISPRM) 2023 World Congress on June 4 to 8, 2023.
Dr. Jessica Wong was supported by a Banting Postdoctoral Fellowship awarded by the Canadian Institutes of Health Research (CIHR) and the Faculty Research Hours Program at the Canadian Memorial Chiropractic College. Dr. Laura Rosella is funded by a Tier 2 Canada Research Chair in Population Health Analytics. Dr. Andrea Tricco is funded by a Tier 2 Canada Research Chair in Knowledge Synthesis. Dr. Pierre Côté was funded by a Tier 2 Canada Research Chair in Disability Prevention and Rehabilitation.
Data availability: The datasets generated and analyzed during the current study are not publicly available due to data sharing agreements and privacy policies that prohibit ICES from sharing the dataset publicly. The dataset from this study is held securely in coded form at ICES. While legal data sharing agreements between ICES and data providers (eg, healthcare organizations and government) prohibit ICES from making the dataset publicly available, access may be granted to those who meet prespecified criteria for confidential access, available at www.ices.on.ca/DAS (email: das@ices.on.ca). The full data set creation plan and underlying analytic code are available from the authors upon request, understanding that the computer programs may rely upon coding templates or macros that are unique to ICES and are therefore either inaccessible or may require modification.
Author contributions: J.J.W.: conceptualization, methodology, formal analysis, and writing—original draft, review, and editing; P.C.: conceptualization, methodology, and writing—review and editing; A.C.T.: methodology and writing—review and editing; T.W.: data curation and writing—review and editing; L.C.R.: conceptualization, methodology, and writing—review and editing.
Footnotes
Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article.
Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's Web site (www.painjournalonline.com).
Contributor Information
Pierre Côté, Email: pierre.cote@ontariotechu.ca.
Andrea C. Tricco, Email: Andrea.Tricco@unityhealth.to.
Tristan Watson, Email: tristan.watson@mail.utoronto.ca.
Laura C. Rosella, Email: laura.rosella@utoronto.ca.
References
- [1].Association of Public Health Epidemiologists in Ontario. Self-rated Health. Available at: https://core.apheo.ca/index.php?pid=96. Accessed December 1, 2023. [Google Scholar]
- [2].Austin PC, Walraven C. The mortality risk score and the ADG score: two points-based scoring systems for the Johns Hopkins aggregated diagnosis groups to predict mortality in a general adult population cohort in Ontario, Canada. Med Care 2011;49:940–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [3].Bank of Canada. Annual Exchange Rates, 2018. Available at: https://www.bankofcanada.ca/rates/exchange/annual-average-exchange-rates/. Accessed September 28, 2020. [Google Scholar]
- [4].Becker A, Held H, Redaelli M, Strauch K, Chenot JF, Leonhardt C, Keller S, Baum E, Pfingsten M, Hildebrandt J, Basler HD, Kochen MM, Donner-Banzhoff N. Low back pain in primary care: costs of care and prediction of future health care utilization. Spine (Phila Pa 1976) 2010;35:1714–20. [DOI] [PubMed] [Google Scholar]
- [5].Bilandzic A, Rosella L. The cost of diabetes in Canada over 10 years: applying attributable health care costs to a diabetes incidence prediction model. Health Promot Chronic Dis Prev Can 2017;37:49–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [6].Brant R. Assessing proportionality in the proportional odds model for ordinal logistic regression. Biometrics 1990;46:1171–8. [PubMed] [Google Scholar]
- [7].Canadian Institute for Health Information. Perceived health, 2022. Available at: https://www.cihi.ca/en/indicators/perceived-health. Accessed December 10, 2023. [Google Scholar]
- [8].Chen H, Kwong JC, Copes R, Villeneuve PJ, Goldberg MS, Ally SL, Weichenthal S, van Donkelaar A, Jerrett M, Martin RV, Brook JR, Kopp A, Burnett RT. Cohort profile: the Ontario population health and environment cohort (ONPHEC). Int J Epidemiol 2017;46:405–405j. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [9].Engel CC, von Korff M, Katon WJ. Back pain in primary care: predictors of high health-care costs. PAIN 1996;65:197–204. [DOI] [PubMed] [Google Scholar]
- [10].French EKE, Kelly E. Medical spending around the developed world. Fiscal Stud 2016;37:327–44. [Google Scholar]
- [11].Gershon AS, Wang C, Guan J, Vasilevska-Ristovska J, Cicutto L, To T. Identifying individuals with physcian diagnosed COPD in health administrative databases. COPD 2009;6:388–94. [DOI] [PubMed] [Google Scholar]
- [12].GBD 2015 Disease and Injury Incidence and Prevalence Collaborators. Global, regional, and national incidence, prevalence, and years lived with disability for 310 diseases and injuries, 1990-2015: a systematic analysis for the Global Burden of Disease Study 2015. Lancet 2016;388:1545–602. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [13].Government of Canada. Opportunity for all—Canada's first poverty reduction strategy, 2022. Available at: https://www.canada.ca/en/employment-social-development/programs/poverty-reduction/reports/strategy.html. Accessed January 3, 2022. [Google Scholar]
- [14].Hart LG, Deyo RA, Cherkin DC. Physician office visits for low back pain. Frequency, clinical evaluation, and treatment patterns from a U.S. national survey. Spine (Phila Pa 1976) 1995;20:11–9. [DOI] [PubMed] [Google Scholar]
- [15].Hayden JA, van der Windt DA, Cartwright JL, Côté P, Bombardier C. Assessing bias in studies of prognostic factors. Ann Intern Med 2013;158:280–6. [DOI] [PubMed] [Google Scholar]
- [16].Hirsch O, Strauch K, Held H, Redaelli M, Chenot JF, Leonhardt C, Keller S, Baum E, Pfingsten M, Hildebrandt J, Basler HD, Kochen MM, Donner-Banzhoff N, Becker A. Low back pain patient subgroups in primary care: pain characteristics, psychosocial determinants, and health care utilization. Clin J Pain 2014;30:1023–32. [DOI] [PubMed] [Google Scholar]
- [17].Hoy D, March L, Brooks P, Blyth F, Woolf A, Bain C, Williams G, Smith E, Vos T, Barendregt J, Murray C, Burstein R, Buchbinder R. The global burden of low back pain: estimates from the Global Burden of Disease 2010 study. Ann Rheum Dis 2014;73:968–74. [DOI] [PubMed] [Google Scholar]
- [18].Hux JE, Ivis F, Flintoft V, Bica A. Diabetes in Ontario: determination of prevalence and incidence using a validated administrative data algorithm. Diabetes Care 2002;25:512–6. [DOI] [PubMed] [Google Scholar]
- [19].Killingmo RM, Chiarotto A, van der Windt DA, Storheim K, Bierma-Zeinstra SMA, Småstuen MC, Zolic-Karlsson Z, Vigdal ØN, Koes BW, Grotle M. Modifiable prognostic factors of high costs related to healthcare utilization among older people seeking primary care due to back pain: an identification and replication study. BMC Health Serv Res 2022;22:793. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Licciardone JC. The epidemiology and medical management of low back pain during ambulatory medical care visits in the United States. Osteopath Med Prim Care 2008;2:11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [21].Mawani F, Gilmour H. Validation of self-rated mental health. Ottawa, ON: Statistics Canada, 2010. Available at: http://www.statcan.gc.ca/pub/82-003-x/2010003/article/11288-eng.pdf. Accessed February 10, 2018. [PubMed] [Google Scholar]
- [22].Mutubuki EN, Luitjens MA, Maas ET, Huygen F, Ostelo R, van Tulder MW, van Dongen JM. Predictive factors of high societal costs among chronic low back pain patients. Eur J Pain 2020;24:325–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [23].Ontario Ministry of Health and Long-term Care. Understanding health care in Ontario, 2012. Available at: http://www.health.gov.on.ca/en/ministry/hc_system/. Accessed April 1, 2019. [Google Scholar]
- [24].Parreira P, Maher CG, Steffens D, Hancock MJ, Ferreira ML. Risk factors for low back pain and sciatica: an umbrella review. Spine J 2018;18:1715–21. [DOI] [PubMed] [Google Scholar]
- [25].Rais S, Nazerian A, Ardal S, Chechulin Y, Bains N, Malikov K. High-cost users of Ontario's healthcare services. Healthc Pol 2013;9:44–51. [PMC free article] [PubMed] [Google Scholar]
- [26].Rao J, Wu C, Yue K. Some recent work on resampling methods for complex surveys. Survey methodology, statistics Canada. Cat 1992;18:209–17. [Google Scholar]
- [27].Ritzwoller DP, Crounse L, Shetterly S, Rublee D. The association of comorbidities, utilization and costs for patients identified with low back pain. BMC Musculoskelet Disord 2006;7:72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [28].Ronksley PE, Brien SE, Turner BJ, Mukamal KJ, Ghali WA. Association of alcohol consumption with selected cardiovascular disease outcomes: a systematic review and meta-analysis. BMJ 2011;342:d671. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [29].Roos LL, Wajda A. Record linkage strategies. Part I: estimating information and evaluating approaches. Methods Inf Med 1991;30:117–23. [PubMed] [Google Scholar]
- [30].Rosella LC, Fitzpatrick T, Wodchis WP, Calzavara A, Manson H, Goel V. High-cost health care users in Ontario, Canada: demographic, socio-economic, and health status characteristics. BMC Health Serv Res 2014;14:532. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [31].Rosella LC, Kornas K, Sarkar J, Fransoo R. External validation of a population-based prediction model for high healthcare resource use in adults. Healthcare (Basel) 2020;8:537. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [32].Rosella LC, Kornas K, Yao Z, Manuel DG, Bornbaum C, Fransoo R, Stukel T. Predicting high health care resource utilization in a single-payer public health care system: development and validation of the high resource user population risk tool. Med Care 2018;56:e61–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [33].Rosella LC, Lebenbaum M, Fitzpatrick T, O'Reilly D, Wang J, Booth GL, Stukel TA, Wodchis WP. Impact of diabetes on healthcare costs in a population-based cohort: a cost analysis. Diabet Med 2016;33:395–403. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [34].Rust K, Rao J. Variance estimation for complex surveys using replication techniques. Stat Methods Med Res 1996;5:283–310. [DOI] [PubMed] [Google Scholar]
- [35].Schultz SE, Rothwell DM, Chen Z, Tu K. Identifying cases of congestive heart failure from administrative data: a validation study using primary care patient records. Chronic Dis Inj Can 2013;33:160–6. [PubMed] [Google Scholar]
- [36].Statistics Canada. Canada at a glance 2018: population, 2018. Available at: https://www150.statcan.gc.ca/n1/pub/12-581-x/2018000/pop-eng.htm?HPA=1. Accessed April 1, 2019. [Google Scholar]
- [37].Statistics Canada. Canadian community health survey 2003: user guide for the public use microdata file. Ottawa, ON, Canada: Statistics Canada, 2005. (Catalogue no. 82M0013GPE). [Google Scholar]
- [38].Statistics Canada. Canadian community health survey (CCHS)—annual component. Available at: http://www23.statcan.gc.ca/imdb-bmdi/document/3226_D7_T9_V8-eng.htm. Accessed October 20, 2018. [Google Scholar]
- [39].Statistics Canada. Self-rated health, 2023. Available at: https://www160.statcan.gc.ca/health-sante/self-rated-health-sante-autoevaluee-eng.htm. Accessed December 1, 2023. [Google Scholar]
- [40].Stewart WF, Yan X, Boscarino JA, Maeng DD, Mardekian J, Sanchez RJ, Von Korff MR. Patterns of health care utilization for low back pain. J Pain Res 2015;8:523–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [41].STROBE Statement. Strengthening the reporting of observational studies in epidemiology: STROBE checklists, 2009. Available at: https://www.strobe-statement.org/?id=available-checklists. Accessed December 10, 2023. [Google Scholar]
- [42].Thomas S, Wannell B. Combining cycles of the Canadian community health survey. Health Rep 2009;20:53–8. [PubMed] [Google Scholar]
- [43].Tu K, Chen Z, Lipscombe LL; Canadian Hypertension Education Program Outcomes Research Taskforce. Prevalence and incidence of hypertension from 1995 to 2005: a population-based study. CMAJ 2008;178:1429–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [44].Wodchis W, Bushmeneva K, Nikitovic M, McKillop I. Guidelines on person level costing using administrative databases in Ontario. Toronto: Health System Performance Research Network, 2013. [Google Scholar]
- [45].Wong JJ, Côté P, Tricco AC, Watson T, Rosella LC. Assessing the validity of health administrative data compared to population health survey data for the measurement of low back pain. PAIN 2021;162:219–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [46].Wong JJ, Côté P, Tricco AC, Watson T, Rosella LC. Effect of back problems on healthcare utilization and costs in Ontario, Canada: a population-based matched cohort study. PAIN 2021;162:2521–31. [DOI] [PubMed] [Google Scholar]
- [47].Wong JJ, Côté P, Tricco AC, Watson T, Rosella LC. Joint effects of back pain and mental health conditions on healthcare utilization and costs in Ontario, Canada: a population-based cohort study. PAIN 2022;163:1892–904. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplemental digital content associated with this article can be found online at http://links.lww.com/PAIN/C17.