Objectives To compare the onset and trajectory of multimorbidity between individuals with and without rheumatoid arthritis (RA).
Methods A matched, retrospective cohort study was completed in a large, US commercial insurance database (MarketScan) from 2006 to 2015. Using validated algorithms, patients with RA (overall and incident) were age-matched and sex-matched to patients without RA. Diagnostic codes for 44 preidentified chronic conditions were selected to determine the presence (≥2 conditions) and burden (count) of multimorbidity. Cross-sectional comparisons were completed using the overall RA cohort and conditional logistic and negative binomial regression models. Trajectories of multimorbidity were assessed within the incident RA subcohort using generalised estimating equations.
Results The overall cohort (n=277 782) and incident subcohort (n=61 124) were female predominant (76.5%, 74.1%) with a mean age of 55.6 years and 54.5 years, respectively. The cross-sectional prevalence (OR 2.29, 95% CI 2.25 to 2.34) and burden (ratio of conditions 1.68, 95% CI 1.66 to 1.70) of multimorbidity were significantly higher in RA than non-RA in the overall cohort. Within the incident RA cohort, patients with RA had more chronic conditions than non-RA (β 1.13, 95% CI 1.10 to 1.17), and the rate of accruing chronic conditions was significantly higher in RA compared with non-RA (RA × follow-up year, β 0.21, 95% CI 0.20 to 0.21, p<0.001). Results were similar when including the pre-RA period and in several sensitivity analyses.
Conclusions Multimorbidity is highly prevalent in RA and progresses more rapidly in patients with RA than in patients without RA during and immediately following RA onset. Therefore, multimorbidity should be aggressively identified and targeted early in the RA disease course.
- outcome assessment
- health care
Statistics from Altmetric.com
What is already known about this subject?
Multimorbidity is a growing public health problem with an ageing population and rising rates of chronic conditions.
While select comorbid conditions are known to complicate rheumatoid arthritis (RA), the timing of onset and rate of accruing multimorbidity in RA is unknown.
What does this study add?
Using a large, commercial insurance database in the current treatment era, we have shown that multimorbidity is significantly more common in patients with RA.
Multimorbidity occurs early in the RA disease course and progresses more rapidly than in patients without RA.
How might this impact on clinical practice or future developments?
Multimorbidity should be targeted early in the RA disease course to prevent progression and achieve better long-term patient outcomes.
Rheumatoid arthritis (RA) is a systemic autoimmune disease that predisposes to physical impairment and premature mortality.1–3 While extra-articular involvement is well recognised to complicate the RA disease course,4 links between RA and other chronic diseases, including osteoporosis,5 cardiovascular disease,6 malignancy7 and mental health disorders,8 have also been identified. Additionally, therapies used to treat RA may have unintended consequences that predispose to the development of chronic diseases. For example, glucocorticoid use is associated with numerous adverse effects including bone loss, elevated blood glucose and blood pressure, and the development of cataracts and glaucoma, among other potential toxicities.9
The study of chronic conditions occurring in individuals with RA has primarily focused on select conditions co-occurring with RA, under the framework and terminology of ‘comorbidity’. Multimorbidity, which considers the burden of conditions to the patient rather than to an index condition, has become the preferred terminology to describe the co-occurrence of multiple chronic conditions in the general population and rheumatic diseases.10 Multimorbidity has been the subject of only limited investigation in RA to date. Initial studies in RA have shown lower rates of biological disease-modifying antirheumatic drug (DMARD) use and poorer response to treatment among those with multimorbidity,11–13 as well as multimorbidity contributing to excess mortality occurring in RA.14
The world’s population is ageing, with the WHO estimating the number of individuals aged 65 years or older to increase from 524 million in 2010 to 1.5 billion in 2050.15 As a result of this increased longevity and a rising frequency of chronic disease risk factors, chronic disease prevalence is projected to rise steadily with over 170 million individuals estimated to have at least one chronic condition by 2030 in the USA alone.16 Accompanying the growing prevalence of chronic diseases is the development of multiple chronic conditions, constituting multimorbidity. In 2014, over 40% of US adults had multiple chronic conditions.17 The consequences of multimorbidity include death and disability, reduced quality of life, and increased healthcare utilisation and costs.18 Thus, multimorbidity is a critically important public health concern that needs to be aggressively targeted. This is especially true in RA, a disease perhaps of accelerated ageing19 20 that portends poor long-term outcomes1–3 and carries an enormous economic impact.21
Targeting multimorbidity with interventions requires understanding its onset and rate of progression.The purpose of this study was to compare the burden and trajectory of multimorbidity between individuals with and without RA. We hypothesised that the burden of multimorbidity and rate of accruing chronic conditions would be greater in RA.
Study design and patient selection
We performed a matched, retrospective cohort study within the Truven MarketScan commercial claims and encounters database from 1 January 2006 to 30 September 2015. MarketScan is a US-wide database of commercially insured individuals with medical and pharmacy claims data that have been used extensively for rheumatic disease research.22–25 Patients and the public were not involved in this study.
We constructed two RA cohorts (overall RA and incident RA) that were matched (1:1) to patients without RA from 1 January 2006 to 31 December 2014. We required patients to have 12 months of continuous enrolment during our study window to be eligible for analyses. We used validated RA algorithms that required at least two RA diagnostic codes (International Classification of Diseases ninth edition, clinical modification (ICD-9-CM): 714.0, 714.1, 714.2 and 714.8) between 30 and 365 days apart, including at least one diagnostic code from a rheumatology provider, and a DMARD prescription. Similar algorithms have a positive predictive value (PPV) for RA >90%.26 Within this overall RA cohort, we identified a subcohort of incident patients with RA using an administrative algorithm requiring ≥12 months of continuous enrolment without RA diagnostic codes or DMARD prescription (PPV of 70%–80%).27 The date patients fulfilled the algorithm was considered the RA index date. We then selected patients without diagnostic codes for RA and matched them 1:1 with patients with RA on sex, year of birth and year entering the database during our study window. We required controls to be enrolled on the index date of the accompanying patient with RA and assigned the same index date. Patients were followed until disenrolment, death or end of the study observation period (30 September 2015 due to transition to ICD-10).
Chronic conditions and multimorbidity
In addition to using established comorbidity indices (see below), we manually assembled a list of 44 chronic conditions based on their prevalence and importance in the general population and RA, informed by prior studies including systematic reviews of multimorbidity.4 28–30 Diagnostic codes for these conditions were adapted from enhanced definitions for established comorbidity indices and the Healthcare Cost and Utilisation Project Clinical Classification Software codes (https://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp) (provided in online supplemental table S1). We queried these conditions from 1 January 2006 to 30 September 2015, a period using only ICD-9-CM codes, within inpatient and outpatient encounters. To minimise misclassification of these conditions (eg, unconfirmed or rule-out diagnoses), we required at least two diagnoses for these chronic conditions to be considered present, with the date of the second diagnostic code considered the date of onset. Once a condition occurred, we considered the condition prevalent throughout the remainder of follow-up.
We defined multimorbidity as the presence of at least two conditions from the aforementioned list. We did not include RA as one of the conditions, as this would inherently bias our results towards greater multimorbidity in RA. We also used a more stringent definition of multimorbidity, requiring the presence of at least three conditions from the list. The total count of chronic conditions present (possible range of 0–43, as two conditions were sex-specific) was considered to represent the burden of multimorbidity. To ensure results were not dependent on these definitions of multimorbidity, we also used established comorbidity indices. This included the Charlson-Deyo Comorbidity Index,31 which has been extensively used in health services research, and the Rheumatic Disease Comorbidity Index (RDCI),32 which has specifically been validated in individuals with rheumatic diseases.
We compared the cross-sectional prevalence of multimorbidity and individual chronic conditions between RA and non-RA in the overall cohort using conditional logistic regression models, conditioning on the matched pair. Comparisons of multimorbidity burden were completed using conditional negative binomial regression. In primary analyses, these comparisons were completed at the index date, while in secondary analyses we performed these comparisons at 1 year of follow-up to ensure all patients with RA had prevalent, rather than incident, disease.
The trajectory of multimorbidity burden in RA vs non-RA in the incident subcohort was assessed using generalised estimating equations with an interaction term between RA status and year of follow-up (to assess differences in the rate of accruing chronic conditions over time) and an autoregressive covariance matrix. The burden of multimorbidity (count of chronic conditions) was specified using a Gaussian distribution for clinical relevance. Skewness and residuals were similar to models generated using a negative binomial distribution (online supplemental figure 1), and observed means suggested a linear relationship between multimorbidity burden and RA status on the raw scale (online supplemental figure 2). In our primary approach, we censored individuals, but not the pair, who disenrolled from the insurance plan to maximise follow-up time. To account for differences that developed between patients with RA and patients without RA over follow-up periods, models included adjustments for age (updated at each year of follow-up) and sex. To investigate multimorbidity trajectory specifically during the period of RA onset, we performed secondary analyses restricting the sample to individuals with ≥3 years of observation before the index date (date classified as RA) and started follow-up at 2 years prior to the index date.
To ensure robustness of our results, we performed several sensitivity analyses. These were: (1) Removing chronic conditions that are known to be closely associated with RA or may be misclassified as RA (anaemia, osteoarthritis, fibromyalgia, interstitial lung disease, chronic back pain, gout, osteoporosis, inflammatory skin disorders), (2) Restricting our sample to patients with ≥1 year of follow-up, (3) Censoring the pair when one patient in the pair disenrolled, (4) Using a stricter 2-year period without RA diagnostic codes or DMARD prescription for incident RA,27 (5) Requiring only ≥1 ICD-9 code to be present for a condition, (6) Adjusting for multimorbidity burden at the index date, and (7) Removing ‘silent chronic conditions’ that could be subject to surveillance bias. We assessed adjustment for geographical region but this did not confound results and was not included in the final models (data not shown). Analyses were completed using SAS V.9.4. Data are available on reasonable request and ethical approval.
Of the >147 million individuals enrolled in MarketScan between 2006 and 2014, we identified 138 891 who fulfilled our eligibility criteria and the RA algorithm, including 30 562 with incident RA (figure 1). After matching (1:1), there were 277 782 patients in the overall cohort and 61 124 patients in the incident subcohort. Baseline characteristics of these patients are shown in table 1. The study sample was female predominant (76.5% overall and 74.1% incident subcohort) with a mean age of 55.6 years and 54.5 years (overall and incident). Time elapsed from entering the database to the index date was 1.5 years (SD 1.7) in the overall cohort and 3.5 years (SD 1.8) in the incident subcohort. Biologic DMARD use was significantly less frequent in the incident cohort (10.1%) compared with the overall cohort (24.6%).
Multimorbidity prevalence and burden
At baseline, 57.4% of RA and 40.8% of non-RA had at least one chronic condition with 33.9% and 21.1% being multimorbid, respectively. The odds of multimorbidity were 2.3-fold higher in RA than non-RA at baseline (conditional OR 2.29, 95% CI 2.25 to 2.34) (table 2). Similar odds of multimorbidity for RA versus non-RA were observed when at least three conditions was used to define multimorbidity or when requiring ≥1 year of follow-up (table 2). The prevalence of multimorbidity in patients with RA was 51.8% when ≥1 year of follow-up was mandated. Of the 44 chronic conditions, 39 were over-represented in RA (online supplemental table 2). The most over-represented chronic conditions in RA were interstitial lung disease (OR 12.62, 95% CI 10.54 to 15.11), fibromyalgia (OR 5.86, 95% CI 5.50 to 6.25), osteoarthritis (OR 5.16, 95% CI 4.98 to 5.35) and osteoporosis (OR 4.54, 95% CI 4.19 to 4.92).
Multimorbidity burden (count of chronic conditions) was significantly higher in RA than non-RA (ratio 1.68, 95% CI 1.66 to 1.70) (table 3). Use of the Charlson-Deyo Comorbidity Index (ratio 1.32, 95% CI 1.29 to 1.35) and RDCI (ratio 1.39, 95% CI 1.37 to 1.41) to measure multimorbidity burden also showed a higher burden of multimorbidity in RA. Similar findings were obtained when requiring ≥1 year of follow-up after RA diagnosis.
In the trajectory analyses using the incident RA subcohort, the mean follow-up was 2.0 (SD 1.8) years in RA and 1.8 (SD 1.8) years in non-RA. Patients with RA had a greater burden of multimorbidity at diagnosis and throughout follow-up (table 4 and figure 2A). The rate of accrual of chronic conditions was significantly higher over time in patients with RA relative to patients without RA (table 4; RA × time (years) β 0.21, 95% CI 0.20 to 0.21, p<0.001). Other factors associated with greater multimorbidity burden were female sex, older age, and a longer duration of follow-up. The greater burden of multimorbidity throughout follow-up and higher rate of accruing chronic conditions persisted when removing conditions closely related to RA or that may be misclassified as RA (table 4 and figure 2B; RA × time β 0.12, 95% CI 0.11 to 0.13, p<0.001). The accelerated accrual of chronic conditions over time was greater in RA when restricting to individuals with pre-RA data (table 4 and figure 2C; RA × time β 0.33, 95% CI 0.32 to 0.35, p<0.001). Among those with pre-RA data, chronic conditions developed at a significantly higher rate in RA versus non-RA after RA onset (RA status × post/pre-RA period β 0.67, 95% CI 0.63 to 0.71, p<0.001). All sensitivity analyses confirmed a greater burden of multimorbidity and a higher rate of accruing conditions in RA (online supplemental figure 3).
Given an ageing population and growing prevalence of chronic conditions, multimorbidity represents a major public health concern.15–17 In this study, we have evaluated the onset and trajectory of multimorbidity in individuals with RA in a large, US commercial claims database during the current treatment era with robust capture of medical care. We found a substantially higher prevalence and burden of multimorbidity in individuals with RA relative to those without RA. Importantly, we identified that the heightened burden of multimorbidity in RA appears to start early in the RA disease course or even during the pre-RA period. Our findings shed important light on the natural history of multimorbidity and will help inform the future development of preventive and/or therapeutic interventions aimed at reducing multimorbidity burden in this high-risk population.
RA is known to predispose to many chronic conditions and there are ongoing efforts to better understand multimorbidity in RA.33–35 In this study, we have demonstrated that multimorbidity is highly prevalent in RA. When requiring ≥1 year of postdiagnosis follow-up in our overall cohort, >51% of patients with RA were multimorbid. Moreover, the odds of multimorbidity were 2.3-fold to 2.5-fold higher in RA relative to patients without RA. Similarly, the burden of multimorbidity, operationalised as the count of chronic conditions, was more than 60% higher in RA. To avoid overestimation resulting from cohort construction, we did not consider RA to be a condition contributing to the definition of multimorbidity. Therefore, our results underestimate the true prevalence and burden of multimorbidity affecting patients with RA. Notably, the list of 44 chronic conditions we compiled from prior studies of multimorbidity in the general population and known RA-related conditions was more sensitive for assessing the burden of multimorbidity in RA than either the Charlson-Deyo Index or the RDCI. While the focus of our study was on multimorbidity as a whole, most individual chronic conditions were over-represented in RA (39 of 44 conditions), as previously reported.36 As expected, extra-articular manifestations (eg, interstitial lung disease) and other musculoskeletal conditions were the conditions most closely associated with RA.
In addition to demonstrating a higher multimorbidity burden in RA, trajectory analyses in incident RA illustrate that the rate of acquiring chronic conditions increases disproportionately compared with persons without RA. This finding supports our proposed hypothesis and was robust to several sensitivity analyses, including analyses that excluded conditions that may be directly related to RA or misclassified as RA. It is also consistent with results from a recent study evaluating postdiagnosis conditions as predictors of mortality within the Nurses’ Health Study where scores for the Multimorbidity Weighted Index increased more rapidly among women with RA than controls.14 A novel finding from our national study of both women and men is that even during a treatment era characterised by earlier RA diagnosis and DMARD initiation, progression of multimorbidity in RA outpaced the rate in patients without RA during the pre-RA period. There are many potential mechanisms for this accelerated progression of multimorbidity in RA. In addition to some chronic conditions being well-established extra-articular features of RA, others may result from the inflammatory processes (eg, cardiovascular disease) and/or disease burden (eg, mental health disorders) accompanying RA.37 Medications used to treat RA or manage RA symptoms may also contribute to the development of chronic conditions.9 Finally, the onset of RA results in an increase in healthcare encounters and utilisation that may contribute to increased chronic disease screening and identification.38 Because chronic conditions were frequent in our patients without RA and results were similar with adjustment for the number of chronic conditions at baseline as well as with the exclusion of ‘silent chronic conditions’, it is unlikely that heightened surveillance accounts for our findings.
The observation that multimorbidity occurs and progresses early in the disease course, or even preceding disease onset, has important implications for future strategies targeting multimorbidity in RA to improve long-term outcomes. The early RA period is typically characterised by establishing the diagnosis of RA, initiating DMARDs, monitoring disease activity and adjusting DMARD regimen following a treat-to-target approach.39 40 Other management considerations during this time include administration of vaccinations, adjunctive treatment modalities (physical and occupational therapy) and symptom management. While the rheumatologist may be focused on these important tasks, our findings illustrate the need for the early RA period to also include aggressive screening for, and management of, multimorbidity. Optimal care models for screening and managing multimorbidity in RA are not known and should be a focus of future research. Specifically, studies are needed to assess the existing patterns of screening for multimorbidity in patients with RA, which providers are performing these screenings and whether such methods are effectively identifying chronic conditions. In RA, co-management with a primary care physician improves screening for hyperlipidaemia.41 Alternative care delivery models that use case managers and multidisciplinary teams have been tested with heterogenous results in the general population.42
Limitations of this study include the inability to adjust for health behaviours and sociodemographics, which may result in unmeasured confounding. There may be misclassification of RA status, incident versus prevalent RA and chronic condition development. However, we used validated algorithms for RA and required the presence of at least two diagnostic codes for chronic conditions.26 27 The sample consisted of US individuals with commercial insurance and may not be generalisable outside of this setting. Because of the frequency of disenrolment from the commercial health plans, follow-up time was limited. Chronic conditions were considered independent, and future work will be needed to precisely characterise the interconnectedness of chronic conditions that defines multimorbidity in RA. The chosen ‘silent conditions’ may cause symptoms and are not exhaustive, but were selected as those most likely to be influenced by surveillance bias. Finally, while multimorbidity differentiates itself from comorbidity by not specifying an index condition, the study of multimorbidity in a specific population, such as RA, requires anchoring on the characteristic of that population.
In conclusion, in this large cohort study using a national commercial insurance database, we found a significantly higher burden of multimorbidity in RA compared with non-RA individuals. Our trajectory analyses demonstrate that multimorbidity onset occurs early in the RA disease course, or even precedes RA onset, and patients with RA experience an accelerated rate of accruing chronic conditions. Strategies aimed at managing multimorbidity to prevent its progression and complications will need to be delivered early in the RA disease course.
Disclaimer : This is a summary of a scientific article written by a medical professional (“the Original Article”). The Summary is written to assist non medically trained readers to understand general points of the Original Article. It is supplied “as is” without any warranty. You should note that the Original Article (and Summary) may not be fully relevant nor accurate as medical science is constantly changing and errors can occur. It is therefore very important that readers not rely on the content in the Summary and consult their medical professionals for all aspects of their health care and only rely on the Summary if directed to do so by their medical professional. Please view our full Website Terms and Conditions.
Copyright © 2021 BMJ Publishing Group Ltd & European League Against Rheumatism. Medical professionals may print copies for their and their patients and students non commercial use. Other individuals may print a single copy for their personal, non commercial use. For other uses please contact our Rights and Licensing Team.
Handling editor Josef S Smolen
Presented at This work was previously presented in abstract form at the 2019 American College of Rheumatology Annual Meeting: BRE, PR, FY, HS, KM, FX, JRC, TRM. Trajectory of multimorbidity in rheumatoid arthritis in a US commercial insurance claims database from 2006–2015. Arthritis Rheumatol 2019; 71 (suppl): Abstract 840.
Contributors BRE, TRM and JRC designed the study. FX and JRC were responsible for acquisition of data. BRE, PR, YY and HS analysed the data. All authors were responsible for interpretation of the data and for drafting, revising and approving the final submitted manuscript.
Funding This work was supported by the Rheumatology Research Foundation Scientist Development Award (BRE), Great Plains IDeA-CTR Scholars Award (BRE) and Patient-Centered Outcomes Research Institute (JRC). TRM is supported by the NIH/NIGMS (U54GM115458) and NIAAA (R25AA020818).
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Patient consent for publication Not required.
Ethics approval This study was reviewed by the institutional review boards at the University of Nebraska Medical Center and University of Alabama at Birmingham.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available upon reasonable request and ethical approval.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.