For Manuscript Submission, Check or Review Login please go to Submission Websites List.
For the academic login, please select your country in the dropdown list. You will be redirected to verify your credentials.
Cardiovascular Disease and Risk Factors among Psoriasis Patients in Two US Healthcare Databases, 2001–2002Kimball A.B.a · Robinson, Jr. D.b · Wu Y.b · Guzzo C.c · Yeilding N.c · Paramore C.d · Fraeman K.d · Bala M.b
aHarvard Medical School, Boston, Mass.; bJohnson & Johnson Pharmaceutical Services, and cCentocor Research and Development, Centocor, Inc., Malvern, Pa.; dUnited BioSource Corp (UBC), Bethesda, Md., USA Corresponding Author
Don Robinson, Jr, MSPH
Johnson & Johnson Pharmaceutical Services
200 Great Valley Parkway
Malvern, PA 19355 (USA)
Tel. +1 610 240 8468, Fax +1 610 651 6717, E-Mail firstname.lastname@example.org
Background: Cardiovascular diseases or risk factors (CVDR) seem to be more common in psoriasis patients than in the general population. Objective: We assessed the relationship of psoriasis with CVDR by analysis of healthcare claims data using a cross-sectional, prevalence-based study design. Patients and Methods: The IMS Health and MarketScan® claims databases were used to identify adults with psoriasis diagnostic codes. Non-psoriasis controls were matched 3:1 based on age, gender, census region and previous medical insurance coverage. Odds ratios evaluated the relative prevalence of CVDR, and Mantel-Haenszel confidence intervals were estimated. Results: CVDR prevalence was generally higher in psoriasis patients than controls in both datasets. Odds ratios for atherosclerosis, congestive heart failure, type 2 diabetes, and peripheral vascular disease were ≧1.20 for psoriasis patients. Elevated disease severity was associated with a higher rate of CVDR, but varied somewhat by dataset and condition. Conclusions: Elevated CVDR rates were found in psoriasis patients compared with controls. This pattern merits further examination.
© 2008 S. Karger AG, Basel
Several studies have demonstrated that cardiovascular diseases and their associated risk factors are more common in patients with psoriasis than in the general population [1,2,3]. The cause of this elevated risk is unclear. Some authors suggest that the profound psychological impact of psoriasis may drive risky behaviors such as obesity and smoking, and thereby directly increase cardiovascular risk . Others have suggested that there may be some intrinsic associated risk: elevated lipids have been documented in psoriasis patients at the time of their initial diagnosis when compared to non-psoriasis controls that were matched for BMI status, as well as other demographic, clinical, and lifestyle characteristics .
New evidence has led to the hypothesis that psoriasis is a systemic inflammatory state that confers increased cardiovascular risk above and beyond traditional risk factors. Several cytokines that have been identified as important mediators of psoriasis, such as interleukins-1, 4, 6, 8, 12 and tumor necrosis factor-α, have also been identified in metabolic syndrome, a chronic inflammatory state associated with obesity [6,7,8,9,10,11,12]. Further supporting this concept is the work of Prodanowich et al. , demonstrating that treatment of psoriasis patients with methotrexate (MTX) substantially decreases the risk of cardiovascular events, especially when supplemented with folate.
One limitation of the early epidemiologic studies examining cardiac risk in psoriasis patients is that they generally sampled hospitalized or specialty clinic patients, and were therefore likely biased toward patient populations with more severe disease [1, 14, 15]. Since most psoriasis patients have mild disease, generalizing from the inpatient population may overestimate cardiovascular risk. This limitation is important because several recent studies using healthcare databases have demonstrated that higher disease severity is associated with higher cardiovascular risk, as has been demonstrated in the arthritis population [3, 16, 17]. Data from a large dataset of ambulatory patients from the United Kingdom (UK), the General Practice Resource Database (GPRD), confirmed the elevated prevalence of cardiovascular risk factors and acute myocardial infarction in a general population of psoriasis patients. Consistent with other recent healthcare database studies, they found that the risk was increased in those who were receiving systemic therapy, a marker they used as a surrogate for disease severity [11, 18].
This study was designed to examine the risk of cardiovascular diseases and their related risk factors in the US population by using two large health plan datasets containing medical, as well as prescription claim data. Obesity is more common in the US than in the UK , and patterns of medical care, as well as treatment, differ across the countries. This in turn may lead to differences in the relationships between psoriasis treatment, disease severity, and cardiac risk. Using these databases, we were able to examine the medications prescribed for the treatment of cardiovascular diseases or risk factors (CVDR), some of which have implications for the management of psoriasis and its comorbidities.
A cross-sectional, prevalence-based study design was used for this investigation. Adult patients, defined as at least 17 years of age in 2001, were selected from two US healthcare claims databases: the IMS Health Integrated Claims Database (IMS Health) and the MarketScan® Commercial Claims and Encounters Database (MarketScan). Patients had to be continuously enrolled with medical and drug benefits from January 1, 2001 to December 31, 2002, but an enrollment gap of up to 60 days was allowed. Patients with psoriasis were identified as having at least 1 medical service claim at any time between January 1, 2001 and December 31, 2002, with an International Classification of Diseases, 9th Revision (ICD-9) code of 696.0 or 696.1 within any diagnostic field. Psoriatic arthritis was considered a clinical subtype of psoriasis. Patients without psoriasis, who also required at least 1 medical service claim during the 2-year study period, were randomly selected as controls from the initial study population and frequency matched at a 3:1 ratio to each psoriasis patient based on age (±2 years), gender (exact), region (exact), and length of prebaseline medical insurance coverage (exact to the month). Matching on the duration of prebaseline medical insurance ensured a similar likelihood for chronic conditions between cases and controls . No psoriasis ICD-9 codes, as defined above, were allowed in the treatment record of the control group.
Two datasets were used for validation purposes, where findings in one were considered robust if confirmed in the other. The initial MarketScan population was about 6 million patients during the observation period (2001–2002). The datasets had some representation of the elderly through a few retiree and Medicare health plans that offered medical and drug coverage. During this time period, most MarketScan patients had healthcare benefits provided through self-insured employers, many being Fortune 500 firms. About 11 million patients were originally available for study with IMS Health, where healthcare coverage was provided by mostly commercial plans of various types. Both datasets were mainly midwestern and southern in their regional characteristics and have been often used for epidemiology and cost of illness studies. Additional details on these data sources have been previously published .
Cardiovascular conditions and their risk factors were selected based on the public health importance of these comorbidities [3, 11, 13, 16, 18]. Cardiovascular diseases and risk factors were identified through ICD-9 codes commonly used in epidemiologic literature (table 1). To ensure maximum sensitivity to CVDR comorbidities, only 1 medical service claim with the appropriate ICD-9 code was required within any diagnostic field. All CVDR events that occurred at any point over the 2-year observation period were captured for analysis. Documentation of CVDR drug therapies was based on National Drug Code (NDC) identifiers found on the prescribed drug claims as well as on Healthcare Common Procedure Coding System (HCPCS) codes. As with diagnostic data, a prescription drug claim that occurred at any point in the 2-year observation period was included in pharmacotherapy analysis.
To explore disease severity patterns, patients were grouped into 2 levels: Mild and Severe. Severe patients received at least 1 systemictreatment during the time period while everyone else was classified as Mild (table 2) [11, 16]. CVDR rates were compared between the psoriasis and matched control cohorts by using the odds ratio (OR) statistic. The Mantel-Haenszel method was used to estimate the 95% confidence interval (CI) for the OR between the psoriasis and matched control population . A single programmer who was experienced with both datasets wrote a common code for data table construction and conducted quality assurance on the data reported. A second programmer confirmed the coding logic and consistency. The study was exempt from Institutional Review Board (IRB) processes because the patient data in both datasets were de-identified and fully compliant with the Health Insurance Portability and Accountability Act (HIPAA). Version 8.2 of the SAS system (SAS Institute, Cary, N.C., USA) was used to generate all statistical analyses.
Several sensitivity analyses were conducted. One sensitivity analysis raised the psoriasis case identification criteria from all patients with at least 1 medical service claim to all patients with at least 1 inpatient or at least 2 ambulatory face-to-face encounters [21, 23, 24]. New controls were generated for this analysis using the same matching criteria as used for the main population model. A second sensitivity analysis compared CVDR rates using different utilization thresholds within the nonsystemic and systemic psoriasis treatment categories. This sensitivity analysis used patients from the main population model, but varied the analytical approach. Based on the observed distribution of utilization data, 2 psoriasis treatment thresholds were chosen for this analysis: 3 or more claims and 8 or more claims, when contrasting CVDR rates with controls.
Psoriasis and control patients were very similar in demographic profile as a result of the matching program, with no difference in age and gender. Some variability occurred in regional representation, particularly with MarketScan. Psoriasis patients in both datasets had demographic profiles consistent with the literature from multi-site observational studies on ambulatory psoriasis patients, although they tended to be older, depending on the dataset [25,26,27,28]. Patients in IMS Health were younger than those from MarketScan (table 3), and in both datasets the patients with psoriasis were older and more likely to be male than in the overall population from which they were chosen . The prevalence of psoriasis was similar across the datasets: 0.79% in MarketScan and 0.89% in IMS Health (table 3).
Treatment patterns were consistent across the datasets, with a few exceptions. Between 18 and 27% of psoriasis patients, depending on the dataset, had no recorded use of any of the specified psoriasis treatments over the 2-year observation period. Between 59 and 67% of patients received only nonsystemic treatment (table 4). Psoriasis patients having a history of only nonsystemic treatment (65% [n = 13,394] and 57% [n = 14,578] of all psoriasis patients in MarketScan and IMS Health, respectively) were almost always using topical monotherapy (data not shown). Topical treatment was combined with ultraviolet (UV) treatment in 2–3% of patients across the datasets (data not shown). Among patients having a record for both systemic and nonsystemic treatments, the top 3 combinations were consistent across the data sources: topical/MTX (3%), topical/psoralen ultraviolet A (PUVA) (1%) and topical/retinoid (1%; data not shown). A severe form of psoriasis was found in approximately 15 and 14% of patients in the MarketScan and IMS Health databases, respectively, when use of systemic treatment was the marker for psoriasis severity (table 5).
A greater prevalence of CVDR was observed in the psoriasis group in both datasets when compared with controls (table 5; fig. 1). The ORs were similar across MarketScan and IMS Health, with the lower confidence limit generally >1.00. The highest ORs consistently meeting or exceeding 1.20 were for atherosclerosis, congestive heart failure (CHF), type 2 diabetes mellitus, and peripheral vascular disease (PVD).
Medications typically used to manage CVDR were also used more frequently among psoriasis patients than controls (fig. 2, 3). The magnitude of difference between the psoriasis and control groups in the drug-based analysis was similar to that found when using ICD-9-based medical claims. CVDR risk generally increased with disease severity, using psoriasis drug utilization as the indicator for severity (fig. 4, 5). In both datasets, higher disease severity raised the risk for CHF, type 2 diabetes mellitus, hypertension, and ischemic heart disease (fig. 4).
The case-selection sensitivity analysis (at least 1 inpatient or at least 2 ambulatory face-to-face encounters) produced similar CVDR patterns, although with lower ORs. In this analysis, a higher prevalence for atherosclerosis and CHF was found in patients with psoriasis than in controls in IMS Health (OR = 1.23 for both). However, atherosclerosis and CHF had CI lower limits that approached 1.00, while some other CVDR exceeded 1.00. The treatment utilization threshold analysis (3 or more, 8 or more) produced a stepped pattern of increasing CVDR risk that was similar to that found with the severity analysis (data not shown).
Consistent with previous research, this study found greater occurrence of CVDR among psoriasis patients when using data from 2 large US claims databases. While the increased CVDR occurrence in psoriasis patients is modest for any given disease or condition, the pattern is pervasive. The prevalence of CVDR rose with disease severity when using standard psoriasis treatments as surrogates for psoriasis severity. Other studies have found these relationships , but this is the largest investigation of CVDR prevalence among psoriasis patients that used US population-level datasets. The consistency of the CVDR patterns across the datasets is notable, as these information sources represent several million US patients from multiple regions of the country who were covered by a variety of health insurance plans. In a related paper using IMS Health data, Han et al.  found higher rates for CVDR among psoriatic arthritis patients except for hyperlipidemia than reported here among all psoriasis patients.
Our method for identifying patients with psoriasis, cardiovascular diseases and their associated risk factors depended on the reliability of medical claims coding and administrative processing within real-world treatment environments. Psoriasis prevalence rates in these datasets (0.79–0.89%) were fairly close to the lower range that appears in the literature (1–2%), although they were higher than the 0.5% found among Swedish military conscripts and lower than the 1.5% reported in the UK [8, 12,30,31,32]. The minor difference in prevalence across the datasets may result from the observed disparity in age and variation in coding practices, as well as differences in health benefits, particularly regarding access to dermatologists. Nonsystemic treatments were the main therapies for these patients across both datasets; the use of systemic treatment was quite low. The most frequent treatment pattern was the use of multiple nonsystemic treatments. When systemic treatment was administered to a patient it was almost always accompanied by nonsystemic treatment.
A number of limitations to these data need to be noted. A cross-sectional, prevalence-based study design was used in this investigation, which combines new and old cases. Consequently, the impact of disease duration in our analysis is unclear and causality cannot be established. Healthcare claim datasets lack information on many important patient attributes, such as height, weight, ethnicity, health behaviors (smoking, exercise, alcohol consumption, diet) and family medical history, all of which may function as unknown confounders in the analysis. No measure of disease severity was available. Laboratory data in both datasets were too infrequent to be reliably incorporated into this study. A healthy worker effect is quite likely in this population because it was comprised of stable employees and their family members.
These results also reflect US population treatment patterns at a particular point in time, and confirmation of these findings in US data sources with stronger Western representation, as well as Medicaid patients, would be useful. Due to the proprietary source of the datasets, verification of ICD-9 coding by medical chart review could not be conducted. However, quality assurance tests are regularly conducted upon these data by the suppliers. The advice from several external experts with experience in using US healthcare claim datasets for psoriasis research (see Acknowledgements) suggests that there is a good true positive case identification rate when using psoriasis ICD-9 diagnostic codes, although the false negative rate among the control population is unknown. The 2-year observation period used in this study has been found to raise the sensitivity of case identification in a Medicare population for several other chronic conditions when using claims-based diagnostic codes . The increased precision of higher utilization thresholds for claims-based case identification needs to be balanced with selection effects since in our datasets, raising the utilization thresholds for case identification notably lowered the prevalence rate to a point well below published reports (data not shown).
In spite of their limitations, healthcare claims data have been widely used for epidemiologic research in psoriasis as well as other autoimmune diseases because they are readily available and contain useful clinical information drawn from large population samples [3, 11, 13,16,17,18, 23, 31]. The risk of an observation bias in this study seems small because the CVDR patterns reported here were also found, albeit muted, among the mild psoriasis patients. Mild psoriasis patients had medical utilization only slightly higher than controls. Similarly, these CVDR patterns continued when the medical utilization of the psoriasis controls was raised in the sensitivity analysis. Psoriasis and controls had the same level of medical utilization in the sensitivity analysis (data not shown).
It is difficult to untangle the complex relationships between disease severity, treatments, and patient characteristics in retrospective claims database studies because there is, appropriately, a relationship between these factors in daily medical practice. Prospective, longitudinal studies are needed to further explore pathogenic linkages between psoriasis and cardiovascular disease [29, 33]. Our knowledge of psoriasis pathogenesis, patient health, and treatment effectiveness would be greatly enriched if biomarkers, clinical assessments, and patient quality-of-life questionnaires were incorporated into such studies [4, 5, 32]. The results from this study may have implications for the clinical care of psoriasis patients if they are confirmed by prospective, longitudinal studies. If elevated CVDR risk is related to psoriasis pathogenesis, more aggressive management of cardiovascular disease and related risk factors may be warranted, especially for the more severely afflicted patients.
The authors appreciate the research insight provided by Steve Feldman, MD, PhD, of Wake Forest University; Harold Javitz, PhD, of SRI International; Robert Kirsner, MD, PhD, of the University of Miami; Adam Long, PhD, of Gordian Health Solutions, and Dave Margolis, MD, PhD, of the University of Pennsylvania. The data support provided by Debra Eisenberg, PhD, Ning Zhao, PhD, and Stephen Xu, MS, of Centocor was indispensable, as was the editorial and submission support provided by Kristin Ruley Sharples, PhD, Rebecca E. Clemente, PhD, Mary Ann Thomas, RN, BSN, and Mary Whitman, PhD, of Centocor. We gratefully recognize the clinical advice of Catherine Cabot, MD, and Larry Deckelbaum, MD, of Centocor, regarding cardiovascular pharmacotherapy.
Funding for this study was provided by Centocor, Inc., Malvern, Pa. Drs. Bala, Guzzo, Yeilding, Wu, and Mr. Robinson were employees of Centocor, Inc. at the time of manuscript preparation. Dr. Kimball has served as a consultant, investigator, a paid speaker for Centocor, and has a fellowship funded by Centocor. Mr. Paramore and Ms. Fraeman have served as research consultants and analysts for Centocor.
Don Robinson, Jr, MSPH
Johnson & Johnson Pharmaceutical Services
200 Great Valley Parkway
Malvern, PA 19355 (USA)
Tel. +1 610 240 8468, Fax +1 610 651 6717, E-Mail email@example.com
Copyright: All rights reserved. No part of this publication may be translated into other languages, reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying, recording, microcopying, or by any information storage and retrieval system, without permission in writing from the publisher or, in the case of photocopying, direct payment of a specified fee to the Copyright Clearance Center.
Drug Dosage: The authors and the publisher have exerted every effort to ensure that drug selection and dosage set forth in this text are in accord with current recommendations and practice at the time of publication. However, in view of ongoing research, changes in government regulations, and the constant flow of information relating to drug therapy and drug reactions, the reader is urged to check the package insert for each drug for any changes in indications and dosage and for added warnings and precautions. This is particularly important when the recommended agent is a new and/or infrequently employed drug.
Disclaimer: The statements, opinions and data contained in this publication are solely those of the individual authors and contributors and not of the publishers and the editor(s). The appearance of advertisements or/and product references in the publication is not a warranty, endorsement, or approval of the products or services advertised or of their effectiveness, quality or safety. The publisher and the editor(s) disclaim responsibility for any injury to persons or property resulting from any ideas, methods, instructions or products referred to in the content or advertisements.