Validation of Self-Reported Rheumatoid Arthritis Using Medicare Claims: A Nationally Representative Longitudinal Study of Older Adults

Survey self-report of rheumatoid arthritis and treatments versus specialist clinician confirmation

OBJECTIVE

To assess agreement between patient survey report and physician recorded arthritic conditions and medication use in order to validate population-based epidemiologic approaches to auto-immune arthritic conditions.

METHODS

Rheumatologists in the U.S. Appalachian region recruited men 50 years or older with a confirmed rheumatoid arthritis (RA) diagnosis. For each participating patient, the treating specialist completed a brief chart abstraction that included rheumatic diagnosis and corresponding treatment. Patients participated in a telephone interview using the same questionnaire as applied in a larger random digit dial survey that queried arthritis diagnosis and treatment. We assessed patient-clinician agreement with the Prevalence Adjusted and Biased Adjusted Kappa (PABAK) statistic.

RESULTS

We included 36 patient-clinician dyads in this analysis. All clinicians and patients concurred in the RA diagnosis (PABAK = 1). For concomitant systemic lupus and scleroderma, we observed generally concordant responses (PABAK 0.89 and 1, respectively). For medication use, for hydroxychloroquine or sulfasalazine was associated with the lowest PABAK (0.39), intermediate values for methotrexate and for the "other conventional synthetic DMARDs" category (0.67), and with the highest agreement PABAK value for the "biologic DMARD or JAK 2 inhibitor" category (0.89).

CONCLUSION

Survey-based self-report of RA offers a useful approach in epidemiological investigation. This is particularly relevant to population-based approaches to autoimmune arthritis related to occupational and environmental factors.

A novel method to monitor rheumatoid arthritis prevalence using hospital and medication databases

BACKGROUND

Most estimates of rheumatoid arthritis (RA) prevalence, including all official figures in Australia and many other countries, are based on self-report. Self-report has been shown to overestimate RA, but the 'gold standard' of reviewing individual medical records is costly, time-consuming and impractical for large-scale research and population monitoring. This study provides an algorithm to estimate RA cases using administrative data that can be adjusted for use in multiple contexts to provide the first approximate RA cohort in Australia that does not rely on self-report.

METHODS

Survey data on self-reported RA and medications from 25 467 respondents of the Australian Longitudinal Study on Women's Health (ALSWH) were linked with data from the national medication reimbursement database, hospital and emergency department (ED) episodes, and Medicare Benefits codes. RA prevalence was calculated for self-reported RA, self-reported RA medications, dispensed RA medications, and hospital/ED RA presentations. Linked data were used to exclude individuals with confounding autoimmune conditions.

RESULTS

Of 25 467 survey respondents, 1367 (5·4%) women self-reported disease. Of the 26 840 women with hospital or ED presentations, 292 (1·1%) received ICD-10 codes for RA. There were 1038 (2·8%) cases by the medication database definition, and 294 cases (1·5%) by the self-reported medication definition. After excluding individuals with other rheumatic conditions, prevalence was 3·9% for self-reported RA, 1·9% based on the medication database definition and 0·5% by self-reported medication definition. This confirms the overestimation of RA based on self-reporting.

CONCLUSIONS

We provide an algorithm for identifying individuals with RA, which could be used for population studies and monitoring RA in Australia and, with adjustments, internationally. Its balance of accuracy and practicality will be useful for health service planning using relatively easily accessible input data.

Rheumatoid arthritis, quantitative parenchymal lung features, and mortality among smokers

OBJECTIVES

There have been limited investigations of the prevalence and mortality impact of quantitative computed tomography (QCT) parenchymal lung features in rheumatoid arthritis (RA). We examined the cross-sectional prevalence and mortality associations of QCT features, comparing RA and non-RA participants.

METHODS

We identified participants with and without RA in COPDGene, a multicentre cohort study of current or former smokers. Using a k-nearest neighbor quantifier, high resolution CT chest scans were scored for percentage of normal lung, interstitial changes, and emphysema. We examined associations between QCT features and RA using multivariable linear regression. After dichotomizing participants at the 75th percentile for each QCT feature among non-RA participants, we investigated mortality associations by RA/non-RA status and quartile 4 vs quartiles 1-3 of QCT features using Cox regression. We assessed for statistical interactions between RA and QCT features.

RESULTS

We identified 82 RA cases and 8820 non-RA comparators. In multivariable linear regression, RA was associated with higher percentage of interstitial changes (β = 1.7 ± 0.5, p= 0.0008) but not emphysema (β = 1.3 ± 1.7, p= 0.44). Participants with RA and >75th percentile of emphysema had significantly higher mortality than non-RA participants (HR 5.86, 95%CI 3.75-9.13) as well as RA participants (HR 5.56, 95%CI 2.71-11.38) with ≤75th percentile of emphysema. There were statistical interactions between RA and emphysema for mortality (multiplicative p= 0.014; attributable proportion 0.53, 95%CI 0.30-0.70).

CONCLUSIONS

Using machine learning-derived QCT data in a cohort of smokers, RA was associated with higher percentage of interstitial changes. The combination of RA and emphysema conferred >5-fold higher mortality.

Agreement between self-reports and statutory health insurance claims data on healthcare utilization in patients with mental disorders

BACKGROUND

Data on resource use are frequently required for healthcare assessments. Studies on healthcare utilization (HCU) in individuals with mental disorders have analyzed both self-reports and administrative data. Source of data may affect the quality of analysis and compromise the accuracy of results. We sought to ascertain the degree of agreement between self-reports and statutory health insurance (SHI) fund claims data from patients with mental disorders.

METHODS

Claims data from six German SHI and self-reports were obtained along with a cost-effectiveness analysis performed as a part of a controlled prospective multicenter cohort study conducted in 18 psychiatric hospitals in Germany (PsychCare), including patients with pre-defined psychiatric disorders. Self-reports were collected using the German adaption of the Client Sociodemographic and Service Receipt Inventory (CSSRI) questionnaire with a 6-month recall period. Data linkage was performed using a unique pseudonymized identifier. Missing responses were coded as non-use for all analyses. HCU was calculated for inpatient and outpatient care, day-care services, home treatment, and pharmaceuticals. Concordance was measured using Cohen's Kappa (κ) and intraclass correlation coefficient (ICC). Regression approaches were used to investigate the effect of independent variables on the agreements.

RESULTS

In total 274 participants (mean age 47.8 [SD = 14.2] years; 47.08% women) were included in the analysis. No significant differences were observed between the linked and unlinked patients in terms of baseline characteristics. Total agreements values were 63.9% (κ = 0.03; PABAK = 0.28) for outpatient contacts, 69.3% (κ = 0.25; PABAK = 0.39) for medication use, 81.0% (κ = 0.56; PABAK = 0.62) for inpatient days and 86.1% (κ = 0.67; PABAK = 0.72) for day-care services. There was varied quantitative agreement between data sources, with the poorest agreement for outpatient care (ICC [95% CI] = 0.22 [0.10-0.33]) and the best for psychiatric day-care services (ICC [95% CI] = 0.72 [0.66-0.78]). Marital status and time since first treatment positively affected the chance of agreement on utilization of outpatient services.

CONCLUSIONS

Although there were high levels of absolute agreement, the measures of concordance between administrative records and self-reports were generally minimal to moderate. Healthcare investigations should consider using linked or at least different data sources to estimate HCU for specific utilization areas, where unbiased information can be expected.

TRIAL REGISTRATION

This study was part of the multi-center controlled PsychCare trial (German Clinical Trials Register No. DRKS00022535; Date of registration: 2020-10-02).

Validation of serostatus of rheumatoid arthritis using ICD-10 codes in administrative claims data

PURPOSE

To determine the accuracy of International Classification of Diseases- Tenth Revision (ICD-10) diagnosis codes for rheumatoid arthritis (RA) serostatus using a U.S. claims database (Optum Clinformatics Data Mart, Optum) and to compare the results to a previous validation study performed in IBM Marketscan Research Database (sensitivity 73%, positive predictive value, PPV, 84%).

METHODS

In Optum (01/01/2016-03/31/2020) linked with laboratory results, we selected RA patients based on ≥2 ICD-10 diagnosis codes for RA (M05 or M06) and at least one dispensing of RA treatments. We included individuals with at least one laboratory result for rheumatoid factor (RF) or anti-cyclic citrullinated peptide (CCP) performed 365 days prior to and including the cohort entry date. An individual was "seropositive" if at least one of the 2 diagnosis codes used to define RA status was M05. "Seronegative" patients were required to have only M06. Secondary analyses were performed using subsets of M05 and M06 diagnosis codes. We calculated the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and kappa of M05 and M06 against the prespecified reference standard laboratory data.

RESULTS

We identified 14 490 adult RA patients who had at least 1 RF or anti-CCP result. The number of patients identified for each reference standard definition ranged from 3315 (reference standard definition: high + anti-CCP) to 13 636 (any + RF). PPV for seropositive RA, M05, was 77.1%. The PPV of M06 for seronegative RA was 61.6%. When we applied more restricted definitions of M05 and M06, the PPV for seropositive RA increased to 79.2%. The PPV for seronegative RA also notably increased to 89.5%.

CONCLUSION

ICD-10 codes (M05 and M06) can help identify RA serostatus in claims data, but their limitations should be acknowledged. The PPVs for seropositive and seronegative RA found in the Optum database were lower than those found in MarketScan, perhaps related to database variability or differing patient characteristics and clinical practice. When more restricted definitions of M05 and M06 were used, the PPVs for seropositive and seronegative RA improved to 79.2% and 89.5%, respectively.

Recombinant Zoster Vaccine Uptake and Risk of Flares Among Older Adults with Immune-Mediated Inflammatory Diseases in the United States

OBJECTIVE

Persons with immune-mediated inflammatory diseases (IMIDs) are at increased risk of herpes zoster (HZ). In 2018, CDC recommended a highly efficacious vaccine, recombinant zoster vaccine (RZV), for prevention of HZ in immunocompetent ≥50 years-olds. We estimated RZV vaccination among adults age ≥50 years with IMIDs during 2018-2019 and examined possible vaccine-related flares following RZV.

METHODS

We identified a cohort of IMID patients using medical claims data from IBM® MarketScan® (age 50-64 years) and Centers for Medicare and Medicaid Services Medicare (age ≥65 years). Presumed flares were defined as hospitalization/emergency department visit for their respective IMIDs, or steroid treatment with a short-acting oral glucocorticoid or parenteral glucocorticoid injection. We conducted a self-controlled case series (SCCS) analysis to examine a temporal association between RZV and flares.

RESULTS

Among enrollees with IMIDs, 14.8% of 55,654 MarketScan and 43.2% of 160,545 Medicare enrollees received ≥1-dose of RZV during 2018-2019. Two-dose series completion was 76.6% in MarketScan and 85.4% in Medicare enrollees. In the SCCS analysis, 10% and 13% developed flares in the control as compared to 9% and 11-12% in the risk window following 1- or 2-doses of RZV among MarketScan and Medicare enrollees, respectively. We found no statistically significant increase in flares following RZV administration for any IMID in either age group following RZV dose-1 or dose-2.

CONCLUSIONS

We did not find an increase in presumed flares following RZV vaccination. Among adults aged ≥50 years with IMIDs, a substantial proportion received RZV compared to general zoster coverage estimates and series-completion was high. This article is protected by copyright. All rights reserved.

No increased risk of Alzheimer's disease among people with immune-mediated inflammatory diseases: findings from a longitudinal cohort study of U.S. older adults

OBJECTIVE

Immune-mediated inflammatory diseases (IMID) are characterized by systemic inflammation affecting the joints and bodily organs. Studies examining the association between individual IMIDs and the risk of Alzheimer's disease (AD) have yielded inconsistent findings. This study examines AD risk across a group of IMIDs in a large population-based sample of older adults.

METHODS

Data on a national sample of US adults over age 50 was drawn from the Health and Retirement Study (HRS) and linked Medicare claims from 2006 to 2014. IMIDs include rheumatoid arthritis, psoriatic arthritis, ankylosing spondylitis, Crohn's disease, ulcerative colitis, and related conditions. We identified IMIDs from 2006 to 2009 Medicare claims using International Classification of Diseases (ICD9-CM) codes. The date of incident AD was derived from Chronic Conditions Warehouse (CCW) identifiers. We examined the risk of AD from 2009 to 2014 using Cox proportional hazards models, both unadjusted and adjusted for age, gender, education, race, and the genetic risk factor APOE-e4.

RESULTS

One hundred seventy-one (6.02%) of the 2842 total HRS respondents with Medicare coverage and genetic data were classified with IMIDs. Over the subsequent 6 years, 9.36% of IMID patients developed AD compared to 8.57% of controls (unadjusted hazard ratio (HR): 1.09, 95% CI .66-1.81, p = 0.74). Adjusted HR 1.27 (95% CI 0.76-2.12, p = 0.35). Age (HR for 10-year increment 3.56, p < .001), less than high school education (HR 1.70, p = .007), and APOE-e4 (HR 2.61, p < .001 for one or two copies), were also statistically significant predictors of AD.

CONCLUSION

HRS respondents with common IMIDs do not have increased risk of Alzheimer's disease over a 6-year period.

No association between rheumatoid arthritis and cognitive impairment in a cross-sectional national sample of older U.S. adults

BACKGROUND

Studies suggest an increased prevalence of cognitive impairment (CI) among people with rheumatoid arthritis (RA). However, most prior studies have used convenience samples which are subject to selection biases or have failed to adjust for key confounding variables. We thus examined the association between CI and RA in a large national probability sample of older US adults.

METHODS

Data were from interviews with 4462 participants in the 2016 wave of the nationally representative U.S. Health and Retirement Study with linked Medicare claims. RA diagnoses were identified via a minimum of two ICD-9CM or ICD-10 codes in Medicare billing records during the prior 2 years. The Langa-Weir Classification was used to classify cognitive status as normal, cognitively impaired non-dementia (CIND), or dementia based on a brief neuropsychological battery for self-respondents and informant reports for proxy respondents. We compared the odds of CI between older adults with and without RA using logistic regression, adjusted for age, education, gender, and race.

RESULTS

Medicare records identified a 3.36% prevalence of RA (150/4462). While age, gender, education, and race independently predicted CI status, controlling for these covariates we found no difference in CI prevalence according to RA status (prevalent CI in 36.7% of older adults with RA vs. 34.0% without RA; adjusted OR = 1.08, 95% CI 0.74-1.59, p = .69).

CONCLUSION

There was no association between RA and CI in this national sample of older U.S. adults.