For patients newly diagnosed with rheumatoid arthritis (RA), rapid initiation of 1 or more disease-modifying antirheumatic drugs (DMARDs), usually a conventional synthetic DMARD (csDMARD) such as methotrexate, is now regarded as the standard of care.1 However, the threshold for initiating more costly biologic DMARD therapy is not as clear cut and has been influenced by sociodemographic, physician, and insurance coverage factors. To better understand these factors while eliminating confounding from differences in insurance coverage, Tatangelo et al2 examined sources of variation in the time to initiation of biologic DMARDs among a cohort of patients newly diagnosed with RA who were 67 years or older and had identical health and prescription medication coverage. Using linked administrative data from a universal, single-payer health care setting (ie, Ontario, Canada, the country’s most populous province), Tatangelo et al2 simultaneously assessed patient-level, prescriber-level, and region-level variation in the time from first csDMARD prescription to first biologic DMARD prescription. Unlike Medicare Part D in the United States, where coverage is administered by a variety of private insurers offering different plans, identical prescription drug coverage is provided by the provincial government to all residents of Ontario who are older than 65 years. In the study by Tatangelo et al,2 17 672 patients were followed up between 2002 and 2015, of whom 719 (4.1%) were prescribed a biologic DMARD. Patient-level factors associated with initiation of biologic DMARDs during follow-up included age, sex, immigrant status, disease duration, and distance from the patient’s own prescriber or from any rheumatologist.2 Controlling for patient-level characteristics, prescribers who graduated from medical school more recently or who practiced in areas with a greater supply of rheumatologists were more likely to initiate biologic DMARDs, while those practicing in rural areas were less likely.2 After adjusting for patient and prescriber characteristics as well as calendar year, Tatangelo et al2 found that 65% of the total variance in time to initiation of biologic DMARDs was associated with physician-level factors (ie, prescribing preferences), operationalized as the annual percentage of a physician’s biologic-naive patients with RA who initiated biologic DMARDs. A further 4.6% was owing to variance between regions and 30.4% to underlying variance.
A distinctive feature of the report by Tatangelo et al2 was the emphasis on longer time to initiation of biologic therapy as an ideal outcome. There is certainly a need to identify and address true instances of undertreatment in which biologic DMARDs are clinically warranted but delayed for nonmedical reasons. Indeed, Tatangelo et al2 acknowledged that some of the patient-level sociodemographic factors that were associated with longer time to initiation of DMARDs (eg, older age, birthplace outside Canada, living a greater distance from a rheumatologist) may be indicative of disparities in access or provision of care. For example, they noted that the age gradient (hazard ratio per 5-year increase in age, 0.66; 95% CI, 0.62-0.71), which was observed even within this sample of older patients with an interquartile age range of 70 to 79 years, may reflect an ongoing bias against prescribing biologic DMARDs to elderly patients, despite a lack of evidence that older patients have a higher risk of infections or other adverse events.2 But it is also important to recognize the consequences of overtreatment. The authors2 convey that the decision to initiate a biologic DMARD should not be taken lightly because it puts patients on a more complex and uncertain treatment pathway and increases costs for patients and health care payers. The out-of-pocket costs of biologic DMARDs for the patients in this study would have been minimal, with an annual deductible of up to $100.00 CDN (US$75.82) and a maximum copayment of $6.11 CDN (US$4.63) per prescription,3 regardless of drug choice. However, in some settings, including the US Medicare program (a comparable population to that studied by Tatangelo et al2), the out-of-pocket costs for biologic DMARDs, especially self-administered ones, are considerably higher than those for csDMARDs.4 There may be downstream effects and opportunity costs as well if payers increase premiums or taxes or make cuts to other areas of the health care system to accommodate the added expenditures.
The analysis by Tatangelo et al2 was restricted to individuals who were 67 years or older when they met the definition for incident RA and received a first prescription for csDMARDs, and the findings may not be generalizable to younger patients. As is typical in analyses of administrative data, the authors did not have access to standardized measures of disease activity, severity, or functional disability, such as the Disease Activity Score 28 or the Health Assessment Questionnaire, which could have been used to adjust for disease severity. Nor did they adjust for proxies of disease severity available from administrative data, such as the number of RA-related physician visits or magnitude of csDMARD use (ie, monotherapy vs combination therapy). Thus, it is possible that patients in certain subgroups (eg, women seeing a high-preference prescriber) had greater disease severity that warranted earlier initiation of biologic DMARDs. Interestingly, in a US registry–based study of patients with RA who had different sources of health insurance,5 physician preference was a strong determinant of initiation of biologic DMARD before and after clinical factors were added to models containing only variables available from administrative data. Regardless, while lacking clinical measures, the analysis by Tatangelo et al2 used an otherwise rich source of longitudinal, individual-level data that covered virtually all older patients in the province. The sample size was large (ie, >17 000 patients), representing approximately 19% of patients with RA in Canada, and enabled the identification of the aforementioned age gradient in initiation of biologic DMARD within the elderly population. Furthermore, there was a simultaneous, time-varying assessment of patient-level, prescriber-level, and region-level factors using multilevel models that reflected real-life practices and behaviors. Patients were nested within prescribers and prescribers within regions, but patients were permitted to change prescribers after their first csDMARD prescription, and patients and prescribers were permitted to move regions.
The findings in the study by Tatangelo et al2 regarding physician preferences are consistent with findings from Sweden,6 where drug coverage is also publicly funded. Furthermore, the study by Tatangelo et al2 provides clinicians and policy makers with empirical evidence in a North American setting of substantial variation in the time to initiation of biologic DMARD, which was not explained by differences in the presence or extent of insurance coverage. While initiation of biologic DMARDs decreased over time in this longitudinal analysis, rheumatologists’ preferences increased, at least on an aggregate level. This highlights the need for additional research, likely with more granular data, to elucidate the mechanisms behind the determinants in this population-level analysis and to uncover others. In the end, 30.4% of the overall difference in biologic initiation was owing to underlying (ie, unexplained) variance. The fact that the quantity-adjusted supply of rheumatologists, a new measure that reflects regional supply in relation to demand for care, was positively associated with initiation of biologic DMARDs might suggest that peer effects7 contribute to prescribing preferences. Examining the influence of additional prescriber-level attributes, such as practice size and setting, may be helpful. At the same time, patient-level factors should not be forgotten. Among the strongest factors in this analysis was immigrant status, with those born outside Canada being 41% less likely to initiate a biologic DMARD during follow-up (hazard ratio, 0.59; 95% CI, 0.42-0.84).2 While its mechanism and implications warrant further investigation, this finding is novel and compelling, and it could become increasingly important. Understanding the association of physician preferences with RA treatment decisions, along with observable patient-level, prescriber-level, and region-level characteristics, should ultimately help optimize the delivery of care for this complex and disabling condition.
Published: December 6, 2019. doi:10.1001/jamanetworkopen.2019.17065
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2019 McCormick N. JAMA Network Open.
Corresponding Author: Natalie McCormick, PhD, Clinical Epidemiology Program, Division of Rheumatology, Allergy, and Immunology, Massachusetts General Hospital, 100 Cambridge St, Ste 1600, Boston, MA 02114 (email@example.com).
Conflict of Interest Disclosures: Dr McCormick reported receiving a fellowship award from the Canadian Institutes of Health Research.
McCormick N. Which Patients With Rheumatoid Arthritis Will Start Biologics, How Soon, and Why—Much to Learn From a Universal Coverage Setting. JAMA Netw Open. 2019;2(12):e1917065. doi:10.1001/jamanetworkopen.2019.17065
Customize your JAMA Network experience by selecting one or more topics from the list below.