Screening of Hyperlipidemia among Patients with Rheumatoid Arthritis in the United States

Abstract

Objective:

To determine the proportion of primary lipid screening among patients with rheumatoid arthritis (RA) and compare it with patients with diabetes mellitus (DM) and patients with neither RA nor DM. To determine whether primary lipid screening varied according to the healthcare provider (rheumatologist vs. non-rheumatologist).

Methods:

We analyzed claims data from United States’ private and public health plans from 2006–2010. Eligibility requirements included: 1) continuous medical and pharmacy coverage for ≥12 months (baseline period), and 2) ≥2 physician diagnoses and relevant medications to define 4 disease categories: RA, DM, RA and DM, or neither condition. Among the 330,695 eligible participants, we calculated the proportion with a lipid profile ordered during the 2 years following baseline. Time-varying Cox proportional hazard models were used to determine the probability of hyperlipidemia screening in participants with RA according to provider specialty.

Results:

Over half the patients were 41–71 years old. Among patients with RA (n=12,182), DM (n=62,834), RA and DM (n=1,082), and neither condition (n=167,811), the proportion screened for hyperlipidemia was 37%, 60%, 55%, and 41%, respectively. Patients with RA who visited a rheumatologist and a non-rheumatology clinician during follow-up had a 55% (95% CI [1.36, 1.78]) higher screening probability than those who only visited a rheumatologist.

Conclusion:

Primary lipid screening was suboptimal among RA patients. It was also lower than patients with DM and minimally different from the general population. Screening was higher for RA patients receiving care from both a rheumatologist and a non-rheumatologist (e.g. primary care physician).

Introduction

Rheumatoid arthritis (RA) is associated with an increased risk for cardiovascular disease (CVD), which involves disease-specific mechanisms that are both similar and different from CVD in the general population.^1,2 Compared with age- and gender-adjusted individuals without RA, patients with RA have more CVD events and higher mortality.³ Indeed, a large meta-analysis of 24 cohorts showed a 50% higher risk for CVD mortality in patients with RA compared to those without RA.⁴ Evidence suggests that both traditional and non-traditional risk factors play a role in the development of CVD in patients with RA.^{5, 6} Traditional risk factors such as hyperlipidemia, hypertension, diabetes mellitus (DM), cigarette smoking, and obesity are also associated with increased risk of CVD in the RA population.⁷ Systemic inflammation was also associated with increased risk for CVD events and mortality among patients with RA. High erythrocyte sedimentation rate, high C-reactive protein levels, and other RA-related biomarkers have been found to be associated with an increased risk of myocardial infarction (MI) in patients with RA.^1,8

Statins are a major focus of CVD risk reduction in the general population, but to determine if these are indicated, a lipid profile is required to estimate 10-year CVD risk. In patients with RA, statins have been associated with a reduction in cardiovascular risk.^9–11 Additionally, it has been reported that, compared with patients with RA who continued statins, those who discontinued this medication had a 67% increased risk for MI.^12,13 Despite the known higher risk for CVD events and mortality, screening for hyperlipidemia among patients with RA have been observed to be low.^14–16

Previous studies using Medicare data among patients older than 65 years of age showed that primary lipid screening occurred in only 45% of patients with RA over a 3-year period from 2004–2006.¹⁵ The pattern of primary lipid screening among individuals with RA who are younger than 65 years of age is still unclear. Hence, the first goal of our study was to determine the proportion of primary lipid screening among patients with RA who are older than 40 years of age and compare these results to the primary lipid screening for patients with neither RA nor DM and patients with DM only. The second goal was to determine whether the proportion of patients with RA who received primary lipid screening varied based on the specialty of the health care provider encountered (rheumatologist vs. non-rheumatologist).

Patients and Methods:

Participants:

All patients were 41– 85 years of age. This retrospective study used a U.S. commercial health plan and public health plans (Medicare and Medicaid) claims data from 2006 to 2010 as a single dataset, a “multi-payer claims database”.¹⁷ This dataset included beneficiaries from four main regions in the U.S.: Northeast, South, West and Midwest. The proportion of beneficiaries by insurance was as follows: 32% were commercially insured, 64% were enrolled in Medicare, and 4% in Medicaid. Eligible participants were required to: 1) have 12 months of continuous medical and pharmacy coverage within the dataset (baseline period) to ensure a complete claims history and to characterize participants’ characteristics; and 2) to satisfy additional criteria that enabled patients to be placed into one of four mutually exclusive cohorts: patients with RA only, patients with DM, patients with both RA and DM, or patients with neither condition. RA was defined based on the presence of at least two International Classification of Diseases, Ninth Revision (ICD-9) codes for RA (714.xx) associated with an encounter with a physician plus at least one filled prescription for a disease-modifying anti-rheumatic drug (DMARD) (e.g., methotrexate, sulfasalazine, hydroxychloroquine, biological therapies, and leflunomide) during baseline.^18,19 The positive predictive value of this definition is >85% to identify RA compared to medical record review.²⁰ We excluded patients with inflammatory arthritis other than RA (e.g., psoriatic arthritis and ankylosing spondylitis), lupus, Sjogren’s syndrome, malignancy, or human immunodeficiency virus (HIV) infection by ICD-9 diagnoses from all four cohorts.

Patients with DM were identified by at least two physician diagnosis codes for DM (ICD9 250.xx) or a filled prescription for DM-specific medication during baseline.²¹ The cohort of patients with neither RA nor DM with commercial insurance, Medicare, or Medicaid was described hereafter as the “general population”. The “general population” comparator reflected individuals without physician diagnosis codes or medications for either RA or DM and were a random sample of insured individuals in the U.S, derived from the same “all-payer claims database”.^17,22 The cohort of patients with RA and DM were required to meet criteria for both RA and DM separately. Patients who had partial evidence for either RA and/or diabetes (e.g. only 1 physician diagnosis code) did not contribute person-time to the analysis until they met the criteria above. Thus, membership in the four disease cohorts was classified in a time-varying fashion.

We excluded in all four cohorts, patients with prevalent inpatient or outpatient MI, stroke, or coronary heart disease (CHD), those with a lipid profile tested and/or use of statins during the 12-month baseline. Follow-up began after the baseline year. We only included participants with 2-year follow-up data consistently available after the baseline year. We conducted a sub-group analysis for the second goal of this study that included only patients with RA regardless of having comorbid diabetes. The inclusion criteria for this sub-group of patients with RA were patients with an encounter with either a rheumatologist and/or “non-rheumatology practitioner”, defined as internal medicine physicians, family medicine physicians, nurse practitioners (NP), or physician assistants (PA). All of these patients also had two ICD-9 diagnosis codes for RA and DMARD medications. The institutional review board at the University of Alabama at Birmingham approved this project.

Variables

The variables for this analysis included age, gender, and hypertension (determined by ICD-9 codes). We had pharmacy data on RA medications, statins, and other non-statin lipid-lowering therapy. We identified physician specialties through evaluation and management codes associated with claims for services within the database and classified the types of outpatient physician evaluation and management (E/M) encounters into E/M with a rheumatologist, or E/M with another physician or provider. Non-rheumatology practitioners of interest focused on specialties commonly providing primary care, including internal medicine physicians, family medicine physicians, NPs, and PAs. Although NPs and PAs may have provided primary care services, they may have actually been attached to a rheumatology clinic. This data source could not differentiate their practice setting. Evaluation and management by other physician specialties (e.g. cardiologist) was not examined in this analysis.

Outcome

The outcome for the primary objective of this study was to determine the proportion of participants who were screened for hyperlipidemia in the cohorts of RA, DM, both RA and DM, and the general population. The outcome for the second objective was the likelihood of a patients with RA being screened for lipids, based on co-management (defined as patients been evaluated or followed) by a rheumatologist and a non-rheumatology practitioner (e.g. primary care) versus management by only a rheumatologist. This study did not determine whether there was communication or coordinated care between these practitioners but only that patients were evaluated in an ambulatory setting (i.e. had outpatient visits).

Statistical Analysis

We used descriptive statistics to examine the baseline characteristics of the patients in this study. We determined the proportion of primary lipid screening during the 2-year follow-up period in each condition. We used Chi-squared tests to determine differences in the proportion of patients with a lipid profile between the different cohorts. We used Cox proportional hazard ratios to determine the likelihood of lipid screening only among patients with RA regardless of having comorbid DM based on visiting a non-rheumatology practitioner, a rheumatologist, or both (sub-group analysis for patients with RA). The type of provider visit (rheumatologist or non-rheumatologist practitioner) was time variant while we controlled for other demographic and RA-related variables at baseline (non-variant). The Cox model allowed for time-varying evaluation of co-management between rheumatologist and non-rheumatologist providers such that patients could be referred to a primary care physician after the start of follow-up and be correctly classified over time. This type of analysis allowed for a more accurate categorization of the main exposure (co-management between rheumatologist and non-rheumatology providers) for the longitudinal analysis.

Results:

Overall, 243,909 participants met the eligibility criteria for the four disease-specific groups: 12,182 patients had RA only, 62,834 had DM only, 1,082 had RA and DM (reflecting 8.2% of all RA patients), and 167,811 had neither condition (Figure 1). As part of cohort selection, 27% of patients with RA only, 25% of patients with DM only, and 22% of patients from the general population, were excluded due to baseline use of statin therapy, or lipid screening. Over half of the patients were 41–70 years of age. The age distribution by disease is presented in Table 1, along with other demographic and clinical patient characteristics. The prevalence of hypertension was similar between patients with RA and patients with neither RA nor DM (40% and 39%, respectively) and similar between those who had both RA and DM and only DM (70% and 79%, respectively). Table 1 also describes the pattern of provider visits (rheumatologist or non-rheumatology practitioner) during the 12-month baseline period.

Figure 1:

Construction of patient cohorts.

Table 1:

Baseline characteristics of study participants.

	Neither	DM only	RA only	RA and DM
Total Number	167,811	62,834	12,182	1,082
Demographics
Age 41–50, %	14	8	11	8
Age 51–60, %	15	15	18	18
Age 61–70, %	23	30	30	33
Age 71–85, %	48	47	42	42
Female, %	60	59	82	80
Caucasian, %	65	66	78	66
Black, %	12	16	9	16
Hispanic, %	4	5	4	6
Asian, %	3	3	1	2
Other race, %*	16	11	9	10
Clinical
Hypertension, %	39	79	41	70
Charlson comorbidities index, %
0	76	0.8	0.3	0.1
1–2	21	80	90	52
≥3	3	19	9	48
Inpatient stay, any, %	8	15	14	23
Physician visits
Rheumatologist visits, %
≤1	98.3	99.1	44.8	51.5
2–4	1.5	0.7	43.4	38.4
>5	0.2	0.1	11.8	10.2
Other physicians visits, %
≤1	58.3	28.0	42.6	29.6
2–4	26.9	34.7	36.9	34.2
>5	14.8	37.3	20.5	36.2
RA medications
Methotrexate monotherapy, %	0	0	27	30
TNF inhibitor, %	0	0	26	18
Non-TNF inhibitor biologic, %	0	0	5	5
MTX combination, %	0	0	13	14

^*Unknown and other race combined

Lipid screening

Among the patients with RA, DM, RA and DM, and neither condition, 37%, 60%, 55%, and 41%, respectively, were screened over the two-year follow-up period (RA versus neither, p < 0.0001; RA versus DM only, p < 0.0001, Figure 2).

Figure 2:

Proportion of Patients with Primary Screening for Low Density Lipoprotein (LDL) by Different Diseases during the 2 years of follow-up.

Screening for lipids based on type of providers visits

Table 2 describes the rheumatologist or non-rheumatology practitioner encounters for patients with only RA. Twenty-two percent of the patients with RA saw only a rheumatologist and 56% visited both a non-rheumatology practitioner and a rheumatologist during the 12-month baseline.

Table 2:

Baseline^* pattern of physician visits among patients with RA^**

	All RA	Non-rheumatology practitioner*** Only	Rheum Only	Both Non-rheumatology practitioner & Rheum
Patients, n	8,606	1,934	1,872	4,800
Average number of outpatient visits during 12-month baseline period, mean (Standard Deviation)	8.4 (5.6)	8.0 (5.4)	5.6 (4.0)	9.8 (5.8)
≤1 visits, N (%)	187 (2.2)	55 (2.8)	131 (7.0)	1 (0.0)
2–4 visits, N (%)	1,927 (22.4)	483 (25.0)	799 (42.7)	645 (13.4)
>5 visits, N (%)	6,492 (75.4)	1,396 (72.2)	942 (50.3)	4,154 (86.5)

^*Baseline was defined as the documentation of non-rheumatology practitioner/rheumatologist visits and Disease Modifying anti-Rheumatic Drug (DMARD) prescription, no LDL tests during the 12-month baseline, and no statin or lipid-lowering medication use before the first LDL test during follow-up.

^**RA population differs from that in table 1 in that patients must have had RA diagnoses from a rheumatologist or specialties that typically provide primary care e.g. family practice, internal medicine, nurse practitioner, or physician assistant

^***Non-rheumatology practitioner consisted of a visit with either an internal medicine doctor, a family medicine doctor, a nurse practitioner, or a physician assistant in the outpatient setting. Note that this dataset did not distinguish if the nurse practitioner or the physician assistant were attached to a rheumatology clinic.

In a multivariable-adjusted model including age, sex, race, comorbidities, and RA medications, the likelihood of hyperlipidemia screening was 55% higher for patients who visited both a rheumatologist and a non-rheumatology practitioner during the 2-year follow-up than for those who only visited a rheumatologist (Table 3). Hyperlipidemia screening was 21% higher for patients who only visited a non-rheumatology practitioner than for those who only visited a rheumatologist.

Table 3:

Probability of Screening for Hyperlipidemia among Patients with RA.

Variable	Multivariable Hazard Ratio (95% Confidence Interval)
Physician Visit
Rheumatologist	Referent
Non-rheumatology practitioner*	1.21 (1.03, 1.41)
Non-rheumatology practitioner and Rheumatologist	1.55 (1.36, 1.78)
41–50 years old	0.86 (0.74, 1.00)
51–60 years old	Referent
61–70 years old	0.96 (0.86, 1.08)
71–85 years old	0.74 (0.66, 0.83)
Male vs. Female	0.94 (0.85, 1.05)
White	Referent
Black	1.14 (0.99, 1.31)
Other	1.12 (0.99, 1.26)
Charlson**
0	Referent
1–2	0.74 (0.31, 1.77)
≥3	0.68 (0.28, 1.65)
Diabetes Mellitus	1.48 (1.26, 1.74)
Hypertension	1.02 (0.94, 1.11)
TNF biologic	1.09 (0.98, 1.22)
non-TNF biologic	1.03 (0.86, 1.24)
MTX combination	1.07 (0.93, 1.22)
MTX monotherapy	1.06 (0.95, 1.19)

TNF = tumor necrosis factor alpha inhibitor; MTX = Methotrexate; MTX combination = Methotrexate combined with either non-biologic disease modifying anti-rheumatic drug, or with TNF biologic or non-TNF biologic.

^*includes internal medicine, family medicine, nurse practitioner, and physician assistant

^**Charlson scoring in this model did not include rheumatoid arthritis

Discussion:

Our study identified a low frequency of primary lipid screening over 2 years among patients with RA. If we consider the 27% of patients with RA that were screened or treated for hyperlipidemia at baseline (and thus excluded from the cohort sample), slightly less than two thirds of all patients with only RA (64%) identified in this dataset were screened. If we consider the 25% of patients with DM only that were excluded at baseline because they were already screened or on treatment for hyperlipidemia, most patients with DM only were screened for hyperlipidemia in this dataset. Patients with RA visited rheumatologists more frequently than they visited a non-rheumatology practitioner. Co-management with one of these non-rheumatology practitioners increased the likelihood of a patient being screened for hyperlipidemia. Compared with other studies that also studied hyperlipidemia screening in patients with RA in the U.S., our study includes a population-based sample of patients between the ages of 40–85, whereas previous studies were limited to 65 years of age and older, only insured by Medicare, or even focused on secondary CVD prevention.^14,15

Some healthcare systems that exist in Europe may provide advantages regarding CVD risk factor assessment in patients with RA. Recently, several published studies have found that European patients with RA are equally or more likely to be treated for hyperlipidemia and other modifiable risk factors, such as hypertension and smoking cessation, when compared with American patients with RA.^23,24 Many U.S. rheumatologists are still reluctant take responsibility to assess and mitigate (if needed) CVD risk for patients with RA.²⁵ In contrast, the European League Against Rheumatism (EULAR) recommendations for CVD risk management in inflammatory arthritis, and in particular in RA, have contributed to better management of CVD risk in patients with RA in some European countries.^26–28 Whereas EULAR emphasizes that CVD risk is the rheumatologist’s responsibility, our study, as well as others conducted in the United States, showed that coordinated care between rheumatologists and non-rheumatology practitioners increases the likelihood of primary lipid screening.^15,27

Our data suggests that rheumatologists are less likely than other non-rheumatology practitioners to initiate primary lipid screening in patients with RA. Two qualitative studies that evaluated possible reasons for such hesitation among physicians found that this inaction resulted from first, perceived role boundaries between specialty doctors, including rheumatologists, and primary care providers. Second, lack of familiarity with CVD treatment guidelines. Third, challenges in communication between physicians. Fourth, misalignment in the perceived responsibility of who should be in charge of screening and management of hyperlipidemia in RA patients.^25,29

Our study has several strengths, including a population-based sample that included not only patients with RA but patients with DM, a disease with a very high risk for CVD, and a random sample of the patients enrolled in similar health insurance programs. However, our “general population” cohort consisted of Medicare, Medicaid, and commercially insured individuals and did not include individuals that were uninsured. The benefit of an all-payer claims database like this for population based research has been previously described.^17,22 Regarding the limitations of this study, our results may not affect delivery of care in light of current CVD screening guidelines, since these recommend screening for hyperlipidemia every 5 years and our ascertainment period spanned only 3 years (1 year baseline, 2 year follow-up). However, EULAR CVD management guidelines for inflammatory arthritis from 2009, relevant at the time of the study, recommended annual CVD risk assessment.²⁸ Still, our results do serve as a useful starting point to assess the U.S. experience, as a springboard to inform future CVD management practices and interventions to mitigate CVD risk among patients with RA. The data source available did not have information to determine smoking status, body mass index, or familial CVD history. The specialty of the clinicians caring for these individuals may have been misclassified, as suggested by the observation that in these data, 22% of patients with RA did not visit a rheumatologist. Indeed, it is likely that some of the NPs and PAs were providing rheumatology-specific care (likely in collaboration with a rheumatologist), but NPs and PAs attached to a rheumatology clinic could not be distinguished from those attached to a primary care clinic. We also recognize that internists recently completing fellowship and transitioning to become rheumatologists may still be classified as internists in the health insurance claims data that we used. Finally, while we recognize that the RA and diabetes cohort assignments were derived from administrative data, we note that our definitions included a combination of two ICD-9 codes for 714.xx plus condition-specific medications, which makes misclassification less likely. However, it still may have missed some cases, particularly for under-recognized diseases or diabetes managed only with lifestyle modification.²⁰

In conclusion, reducing modifiable CVD risk factors should be a priority in patients with RA. Measures to achieve this goal must be implemented and may include defining specific roles for rheumatologists, non-rheumatology practitioners, and patients to determine who should be responsible for hyperlipidemia screening and treatment for patients with RA.

Significance:

• Despite the generally recognized increased risk for RA-associated cardiovascular disease (CVD), patients with RA in this population-based study were under-screened for primary hyperlipidemia.

• The proportion of patients with RA screened for primary hyperlipidemia was comparable to the general population and lower than that of patients with diabetes.

• Co-management, defined as patients managed both by rheumatologist and by specialists who predominantly provide primary care services, increased the likelihood of primary hyperlipidemia screening in patients with RA by 55%