Abstract

Objective To examine the association between rheumatologist access, early treatment, and ongoing care of older-onset rheumatoid arthritis (RA) and healthcare utilization and costs following diagnosis.

Methods We analyzed data from a population-based inception cohort of individuals aged > 65 years with RA in Ontario, Canada, diagnosed between 2002 and 2014 with follow-up to 2019. We assessed 4 performance measures in the first 4 years following diagnosis, including access to rheumatology care, yearly follow-up, timely treatment, and ongoing treatment with a disease-modifying antirheumatic drug. We examined annual healthcare utilization, mean direct healthcare costs, and whether the performance measures were associated with costs in year 5.

Results A total of 13,293 individuals met inclusion criteria. The mean age was 73.7 (SD 5.7) years and 68% were female. Total mean direct healthcare cost per individual increased annually and was CAD $13,929 in year 5. All 4 performance measures were met for 35% of individuals. In multivariable analyses, costs for not meeting access to rheumatology care and timely treatment performance measures were 20% (95% CI 8-32) and 6% (95% CI 1-12) higher, respectively, than where those measures were met. The main driver of cost savings among individuals meeting all 4 performance measures were from lower complex continuing care, home care, and long-term care costs, as well as fewer hospitalizations and emergency visits.

Conclusion Access to rheumatologists for RA diagnosis, timely treatment, and ongoing care are associated with lower total healthcare costs at 5 years. Investments in improving access to care may be associated with long-term health system savings.

Key Indexing Terms:

Rheumatoid arthritis (RA) is a chronic inflammatory disease associated with high economic burden to patients and health systems.1,2 Beyond the numerous personal effects3,4 and indirect costs of RA5 incurred at the individual level, direct health system costs for RA are high.5 Medication costs and acute care services are responsible for most direct costs.5,6 Healthcare utilization and resulting health system costs may be affected by variations in the quality of care provided.7 Studies show an association between higher treatment adherence with lower rates of hospitalizations, emergency department (ED) visits, and healthcare costs.8-10 As early diagnosis and ongoing rheumatology care result in superior clinical outcomes11,12 and patients under the care of a rheumatologist are more likely to receive treatment and persist with therapy,13 understanding whether there is a potential to reduce costs through improved RA care is important.

Health system performance measures have been defined for many health conditions to evaluate health systems and physician practices to identify areas for improvement. Performance measures are usually developed based on a rigorous process, leveraging best practices from guidelines and expert opinion.14-18 These measures are frequently categorized according to Donabedian’s original description as relating to optimal structures, processes, or outcomes of care.19 In general, it is posited that optimal structures and processes of care should result in improved patient outcomes. What is less clear is whether improved adherence to performance measures also improves health system outcomes (eg, reduced hospitalizations and ED visits) and thus lowers healthcare costs. In Canada, system-level performance measures for RA have been developed to evaluate processes of care measuring access to rheumatologist care, timely treatment, as well as ongoing care.20 Although the measures have been rigorously developed and tested in multiple Canadian provinces,21-23 it remains to be demonstrated how meeting performance measures affects health system outcomes. In an aging population with those over 65 years of age incurring approximately 45% of all healthcare costs, it is particularly important to understand the impact of care quality on this segment of the population.24 The present study aims to determine whether adherence to RA performance measures (access to a rheumatologist, early treatment, and ongoing rheumatology care and treatment) over the first 4 years following diagnosis influences subsequent healthcare use and costs in those over age 65.

METHODS

Study design and setting. This was a retrospective population-based cohort study among those over age 65 evaluating the effect of adherence to established RA performance measures20 on health system outcomes. In Ontario, most healthcare services, including physician services, inpatient care, ambulatory clinics, ED, same day surgery, inpatient rehabilitation, complex continuing care, long-term care, and home care, are publicly funded. These services are paid through the provincial government-run health insurance plan (Ontario Health Insurance Plan [OHIP]). In addition, for those over the age of 65, the Ontario Drug Benefit (ODB) program covers the cost of medications (with a small copayment). Information on health care use is recorded in administrative datasets that can be linked to enable population-based evaluations of healthcare utilization and cost. We used these population-based administrative health datasets to develop an inception cohort of individuals with RA at least 66 years and older at diagnosis to allow at least 1 year for enrollment into the ODB program. For these individuals, drug coverage data were available, and they had at least 5 years of follow-up.

Data sources. The health administrative datasets used for this study are described in Supplementary Table S1 (available with the online version of this article).

In brief, we used the following datasets: the Discharge Abstract Database (for inpatient hospitalizations), the ODB database (for pharmacy dispensations), the National Ambulatory Care Reporting System (for ED visits, same day surgery events, cancer and dialysis clinics), the National Rehabilitation Reporting System (for inpatient rehabilitation stays), the Home Care Database (for all home care services), the Ontario Mental Health Reporting System (for inpatient mental health stays), the OHIP claims database (for physicians and publicly funded allied professional services and laboratory tests), the Continuing Care Reporting System (complex continuing care and long-term care), the Registered Persons Database (for patient sociodemographics), and the ICES Physician Database (for identifying physician specialty). These datasets were linked using unique encoded identifiers and analyzed at ICES (www.ices.on.ca). The use of data in this study was approved by a privacy impact assessment at ICES and was authorized under section 45 of Ontario’s Personal Health Information Protection Act, which does not require review by a research ethics board.

Study population. We used the Ontario Rheumatoid Arthritis Database (ORAD)25 for this study and identified individuals who had at least 3 physician claims with an RA diagnosis code (International Classification of Disease, 9th Revision, code 714) within 2 years with at least 1 claim by a rheumatologist.25,26 Cohort entry was defined as the first RA diagnosis code. Patients were required to be aged 66 years and older at the time of their first RA diagnosis code (ie, RA onset) and for RA onset to be between April 1, 2002, to March 31, 2014 (to permit at least 5 years of follow-up; a flow diagram of study inclusion is shown in Supplementary Figure S1, available with the online version of this article). We limited the evaluation to older-onset RA as complete medication data are only available for those aged 65 and older. Age 66 and older was used as an inclusion criterion to allow 1 year to transition to public drug benefits. Individuals were excluded if they were non-Ontario residents at the date of their first RA code, did not have OHIP eligibility in the 5-year period prior to their first RA code (to exclude patients with prevalent RA moving to Ontario), or had less than 5 years of follow-up (see Supplementary Figure S1 for cohort selection).

Performance measure adherence. Following cohort entry up to the beginning of year 5, individuals were characterized according to whether they accessed rheumatologist care (seen by a rheumatologist within 365 days of their first RA diagnosis code), had persistent rheumatologist follow-up (an annual rheumatologist visit), had at least 1 annual dispensation of a disease-modifying antirheumatic drug (DMARD), and had timely DMARD initiation (90 days prior to or within 14 days after the first rheumatologist visit to allow for those starting DMARDs prior to their first rheumatologist visit).20 DMARDs included conventional synthetic DMARDs, targeted synthetic DMARDs, biologic agents, as well as other immunosuppressives used to treat complications of RA (see Supplementary Table S2 for included DMARDs, available with the online version of this article). Measure performance was assessed dichotomously and reported individually or as a composite measure requiring all 4 measures to be met.

Health system outcome measures. Annual healthcare utilization and costs were determined for the cohort in the first 5 years after RA diagnosis. Healthcare utilization included the following: the number of primary care visits and those for a musculoskeletal (MSK)-related reason, the number of nonrheumatologist MSK specialist visits (including orthopedic surgeons, physiatrists, and physiotherapists), the number of rheumatologist visits, the number of joint imaging tests, the number of ED visits (including those for an MSK-related cause), the number of prescription drugs, and the number of inpatient hospitalizations. Specific diagnostic codes used for MSK-related diagnoses are shown in Supplementary Table S3 (available with the online version of this article).

We determined direct healthcare costs (health system payor perspective) using a validated ICES person-level costing algorithm.27 Costs were expressed in 2019 Canadian dollars. Costs for hospitalizations, ED visits, and same day surgery were calculated according to established Resource Intensity Weight (RIW) methodology.27 This involved multiplying the individual case RIW by the average provincial costs per weighted case (weighted by case mix groups).27 Costs for all prescription drugs were calculated on an individual basis. Outpatient costs for diagnostic or laboratory tests are identified in OHIP data. In-hospital laboratory and diagnostic costs are included in total hospital costs. Inpatient mental health costs were combined with inpatient and same day surgery hospital costs. Other costs, including those for nonphysician claims (eg, optometrists, physiotherapists), complex continuing care, long-term care, home care, cancer, dialysis clinics, and chemotherapy covered by the new drug funding program were included under the category “other” for reporting. Due to the lack of cost information prior to April 1, 2006, the cost of outpatient clinic, cancer clinic, and dialysis clinic attendance was estimated only for year 5.

Covariates. Individual characteristics were determined at cohort entry. Sociodemographic variables included age, sex, and rural location of residence (based on postal code and a community size of < 10,000 residents).28 Neighborhood income quintiles were defined using postal codes and the Statistics Canada Census as a proxy for socioeconomic status. Linear distance in kilometers between individuals meeting the case definition for RA (calculated using distance from the center of the patient’s postal code area) and their rheumatologist was determined, and individuals residing ≥ 100 km from the first rheumatologist seen were described as living in a remote location. Comorbidity was defined based on physician diagnosis and hospital discharge codes in the 3 years prior using validated algorithms29-34 and included coronary artery disease, severe cardiovascular disease (hospitalizations related to acute myocardial infarction, heart failure, percutaneous intervention, or coronary artery bypass grafting), diabetes, deep venous thromboembolism or pulmonary embolism, and acute and chronic renal disease (Supplementary Table S4, available with the online version of this article). The Johns Hopkins Adjusted Clinical Group ACG® System, version 11, was used to describe case-mix35 including a frailty indicator. Aggregated Diagnosis Groups (ADGs) derived from this system consist of 32 diagnosis clusters, each having similar clinical criteria and expected health resource needs. Individuals can have comorbidities belonging to between 0 and 32 ADGs. The number of ADGs was determined based on a 3-year lookback period prior to cohort entry. The number of full-time equivalent practice rheumatologists per 75,000 adults living in the patient’s healthcare region in the year of RA diagnosis was used to estimate rheumatologist supply.

Statistical analyses. We reported baseline cohort characteristics. Annual healthcare utilization and mean (SD) annualized costs were determined for the whole cohort, and for individuals where performances measures were met.

To evaluate whether meeting the performance measures (over their first full 4 years) is associated with lower health costs in year 5, multivariable linear regression models were used. Log costs were used as the dependent variable to account for positively skewed distribution and robust standard errors were used to further assure that the reported results are conservative. Accordingly, the adjusted associations with meeting performance measures were estimated in terms of mean relative (%) cost reduction or increase, together with model-based 95% CIs. The models were adjusted for potential confounders selected using both a priori knowledge and data-dependent criteria. Baseline characteristics were selected based on substantive subject area knowledge of their potential effect on healthcare costs and are shown in Table 1. All these variables were considered as potential confounders in the regression models. The final multivariable model was obtained using backward selection. Covariates with P < 0.05 were retained in the final model. The possibility of interactions among performance measures was specified a priori; statistically significant interactions between performance measures at the 0.05 level were retained in the adjusted model. All analyses were performed at ICES using SAS version 9.3 (SAS Institute).

Table 1.

Baseline characteristics of individuals aged ≥ 66 years with RA at cohort entry.

RESULTS

Cohort characteristics. There were 13,293 individuals meeting the case definition for RA in the cohort. The mean age was 73.7 (SD 5.7) years and 68% were female. Baseline characteristics of the cohort are shown in Table 1.

Performance measurement. Within the cohort, 12,493 (94%) saw a rheumatologist within 1 year, 8366 (63%) had at least 1 annual rheumatology visit, 7581 (57%) had annual DMARD dispensations during the 4 years following diagnosis, and 7975 (60%) were dispensed a DMARD within 14 days of a rheumatologist visit (or 90 days prior). Over the first 4 years following diagnosis, all 4 performance measures were met for approximately one-third of the cohort (35%), 3 measures were met for 24%, 2 were met for 22%, 1 was met for 18%, and in 1% of cases no measures were met.

Healthcare utilization. During the annual follow-up periods, apart from hospitalizations, ED visits, and pharmacy dispensations, annual outpatient primary care physician MSK encounters tended to decline with each additional year (Table 2). In contrast, visits to other MSK specialists did not decline over follow-up. Among 13,293 individuals with older-onset RA, there was an immediate attrition from rheumatology care. Half of the cohort continued to have joint imaging throughout the first 5 years following diagnosis. Where all 4 performance measures were met, these individuals had fewer ED visits and hospitalizations across all measurement years.

Table 2.

Annual healthcare utilization of older individuals with RA.

Healthcare costs. The annual mean total costs for individuals with older-onset RA increased over 5 years (Figure 1). The mean total direct costs per patient in year 5 was 2019 CAD $13,929 (SD 23,389; Table 3). By year 5, medication and inpatient costs comprised 22% and 21% of mean total costs, respectively. Other costs including those for continuing and long-term care, home care, cancer, and dialysis clinics were only estimable due to data availability in year 5 and comprised 30% of total costs.

Figure 1.

Mean total costs over the first 5 years for older individuals with rheumatoid arthritis (total cohort). Cost categories include clinician costs, laboratory testing, hospital, emergency, and prescription medications.

Table 3.

Annual direct (per person) mean (SD) healthcare costs in the fifth year following diagnosis (in 2019 Canadian dollars) for patients with older onset RA by performance measures met.

Association between performance measures and healthcare costs. Annual mean healthcare costs in year 5 were modestly lower among individuals where all 4 performance measures were met compared to those where they were not (CAD $13,643 vs $13,929; Table 3). This difference was driven by lower “other” costs, including from complex continuing and long-term care, home care, cancer, and dialysis clinics, as well as less frequent hospitalization and ED visits.

We examined whether measure adherence in the first 4 years following diagnosis was predictive of total costs in the fifth year following diagnosis (Table 4). In the fully adjusted analysis (adjusting for age, sex, income quintile, diagnosis period, ADG, frailty, presence of coronary artery disease, congestive heart failure, chronic obstructive pulmonary disease or asthma, diabetes, hypertension, and health region), access to rheumatology care and timely DMARD treatment were independently associated with lower costs, although yearly rheumatology care combined with DMARD treatment were associated with higher costs. Those who received access to rheumatology care within a year of diagnosis and timely treatment had lower costs of otherwise comparable people who did not receive access to care and timely treatment (84% [95% CI 76-92] and 94% [95% CI 90-99], respectively). As expected, there was an interaction observed between annual rheumatologist follow-up and annual DMARD treatment. Costs of individuals seeing a rheumatologist annually were 51% (95% CI 40-63) higher than the costs of those individuals who did not have annual DMARD dispensations and rheumatologist follow-ups. The costs for individuals who had annual DMARD dispensations were 71% (95% CI 55-88) higher than the costs of those who did not have annual rheumatologist follow-ups or DMARD dispensations. The costs for individuals with both annual rheumatologist follow-ups and treatment were 84% (95% CI 72-97) higher than the costs of those meeting neither. R2 for the adjusted model was 0.116, indicating that approximately 12% of the total variation in log total costs was explained by the factors included in the model.

Table 4.

Effect of performance measures on log total costs in year 5 among older individuals with RA.

DISCUSSION

The present study evaluates the effect of adherence to national performance measures on healthcare utilization and costs in an incident population-based cohort of individuals with older-onset RA. Importantly, our findings demonstrate the importance of access to specialist care and early treatment. We found that total healthcare costs for individuals with access to rheumatologist care within a year of diagnosis and timely initiation of DMARD treatment were reduced. Perhaps unsurprisingly, having an annual rheumatologist follow-up and receiving ongoing treatment with a DMARD were associated with higher costs in year 5, driven primarily by higher prescription and specialist visit costs.

The present study focused on whether adherence to performance measures was independently associated with healthcare costs. The performance measures represent a minimum standard of care and, although access to rheumatology care within a year of diagnosis and timely treatment were both associated with lower total costs, it is likely that more granular measures may be required to demonstrate larger differences in healthcare costs. For example, the proportion of individuals seen within 1 year of diagnosis represents a measure of rheumatologist access. It does not identify wait times for rheumatology care as these require referral data, which were not available.36 Other studies have demonstrated that delays in care are associated with worse outcomes,37 and that there is a window of opportunity such that diagnosis and treatment within the first 3 months is crucial in improving outcomes.38 Further, this study did not include individuals who were never evaluated by a rheumatologist. Such individuals may have competing health issues, mild disease, or challenges in access to care, and the healthcare costs associated with these reasons may vary and warrant further examination.

We have previously demonstrated the close link between retention in rheumatologist care and ongoing DMARD treatment.13 The purpose of the present evaluation was not to determine the effectiveness of DMARDs at reducing long-term healthcare costs, as this would require a more complex evaluation. We only assessed whether patients had at least 1 annual dispensation for a DMARD; this minimum standard does not address other important factors such as treatment adherence or whether treatment was tailored to control disease activity. Although annual DMARD treatment is also a recognized measure in the United States,39 it is likely that individuals with continuous treatment may have improved health outcomes.40 The health system benefits of DMARD treatment (including continuous treatment) have been examined previously. For example, in a retrospective cohort study using administrative data, in the 12-month period following initiation of etanercept for treatment of RA, there was a significant decline in overall health services utilization with on average 1 fewer office visit and a decrease of 10% in ED visits.41 High treatment adherence measured using the proportion of days covered had the lowest health service utilization compared to the least adherent treatment patterns.41 The type of treatment also clearly has a role in healthcare costs with biologic and targeted synthetic DMARDs incurring higher prescription medication costs compared to conventional synthetic DMARDs; however, the current DMARD performance measure does not distinguish between types of DMARDs. In a Canadian study of direct healthcare costs in Alberta using administrative data linked to a provincial pharmacovigilance database over 2 years,42 patients initiated on anti–tumor necrosis factor (anti-TNF) therapy who had a treatment response and were maintained on therapy had the lowest annual healthcare costs. The highest costs were seen in those switching biologic agents, followed by costs for conventional synthetic DMARD and those about to start an anti-TNF agent.42 In all treatment groups approximately 40% of healthcare costs could be attributed to RA but were lowest in the anti-TNF group.42

Our study found that “other” costs (a category including long-term care costs) along with inpatient and prescription drug costs accounted for the highest proportion of total costs. The distribution of cost components likely differs for younger individuals. Healthcare costs tend to vary among populations reflecting the proportion treated with biologic agents,43 comorbidity differences between populations affecting hospitalization and ED use, and potential differences in disease severity.6,44 Health system structure45 and insurance availability also affect costs.46 Biologic DMARDs continue to incur the highest public drug spending in Canada.47 Beyond medication expenses, health systems costs attributed to RA care may also be driven by subgroups including those with difficult-to-treat RA.44,48 Future studies are required to understand the rising costs of RA. However, the remarkably low percentage of older individuals with RA meeting performance measures calls for immediate attention to improve the delivery of RA care.

Although the present study has strengths, including the use of population-based data from a publicly funded healthcare system, there are limitations to discuss. As with all studies of administrative data, there is the potential for misclassification, although we used a validated case definition and all included individuals had a diagnosis made by a rheumatologist, which increased the specificity. We did not include individuals with RA who never saw a rheumatologist; this may have affected results, as the comparison group excludes patients who likely would have received even less care for their RA (ie, those not seen by a rheumatologist). Higher disease severity42 and functional disability41,49 may affect the need to see a rheumatologist and can be associated with increased direct costs; however, we did not have this information available. We have not determined serologic status or the presence of erosions, and these characteristics have been associated with more severe RA and greater healthcare utilization.50 Further, our evaluation is limited to older-onset RA (as there are limited medication data available on patients aged < 65 yrs) and results may not be generalizable to younger populations who likely experience lower healthcare utilization and different drug utilization. We also excluded individuals who died within the first 5 years; costs generally increase prior to death, but these individuals are not reflected in our cohort.

In conclusion, this study demonstrates that access to a rheumatologist and early treatment appear to be associated with lower total healthcare costs at 5 years. More granular measures of continuous treatment and further adjustment for confounders of disease activity at baseline may be needed to demonstrate health system benefits of ongoing rheumatologist care and DMARD treatment. Considering the high economic burden of RA, it is critical for healthcare planners to continue to prioritize the implementation of strategies to improve access to RA care.

ACKNOWLEDGMENT

The datasets were linked using unique, encoded patient and physician identifiers and are securely held and analyzed at ICES (www.ices.on.ca). ICES is a prescribed entity under section 45 of Ontario’s Personal Health Information Protection Act (PHIPA). Section 45 of PHIPA authorizes ICES to collect personal health information, without consent, for the purpose of analysis or compiling statistical information with respect to the management of, evaluation or monitoring of, the allocation of resources to or planning for all or part of the health system. The use of data in this study was authorized under section 45 of Ontario’s PHIPA (which does not require review by a Research Ethics Board) and approved by ICES’ Privacy and Legal Office. We thank IQVIA Solutions Canada for use of their Drug Information file. 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.

Footnotes

Funding for this study was provided by a Project Grant from the Canadian Institutes for Health Research (CIHR), PJT169194. CEHB receives support from an Arthritis Society Stars Career Development Award funded by the CIHR Institute of Musculoskeletal Health, STAR-19-0611/CIHR SI2-169745. DL holds the Mary Pack Chair in Rheumatology Research from the University of British Columbia and the Arthritis Society Canada. JAAZ is the BC Lupus Society Research Scholar and the Walter & Marilyn Booth Research Scholar. HX holds the Maureen and Milan Ilich/Merck Chair in Statistics for Arthritis and Musculoskeletal Diseases. JW receives support from the Arthritis Society Stars Career Development Award, STAR-19-0610. CB is a Canada Research Chair in Rheumatoid Arthritis and Autoimmune Diseases. DAM receives salary and research support in part by the Arthur J.E. Child Chair in Rheumatology and a Canada Research Chair in Health Systems and Services Research. 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. Parts of this material are based on data and information compiled and provided by the MOH and the Canadian Institute for Health Information. 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.

The authors declare no conflicts of interest relevant to this article.

Accepted for publication November 14, 2022.Copyright © 2023 by the Journal of Rheumatology

This is an Open Access article, which permits use, distribution, and reproduction, without modification, provided the original article is correctly cited and is not used for commercial purposes.