Abstract
Background: Methotrexate (MTX) and cyclosporine (CYC) may adversely interact with common medications in patients with psoriasis. Objective: Our purpose was to investigate the prevalence and outcomes of MTX/CYC polypharmacy. Methods: We evaluated rates of events that may be associated with drug-related toxicity, health care resource utilization and costs for patients with psoriasis in the Ingenix® Impact National Managed Care Database (1999–2007) who were exposed or not exposed to potential drug-drug interactions. Results: Among 4,583 (57.6%) exposed and 3,372 (42.4%) nonexposed patients, nonsteroidal anti-inflammatory drugs and antibiotics were the most common drugs with potential interactions. The exposed patients had significantly greater risks of developing renal [adjusted odds ratio (OR): 2.58; p = 0.0145], gastrointestinal (OR: 1.36; p = 0.0197) and pulmonary events (OR: 1.20; p = 0.0470), and significantly greater health care resource utilization (e.g. OR for inpatient and emergency department visits: 1.47; p < 0.0001) and costs (adjusted incremental cost: USD 1,722; p < 0.0001). Conclusions: MTX/CYC polypharmacy is prevalent in patients with psoriasis and associated with significant risks.
Introduction
Psoriasis is a chronic inflammatory skin disease influenced by genetic and environmental characteristics and worsened by factors such as stress, drugs, tobacco and alcohol abuse [1,2,3]. Psoriasis can impair physical and psychological well-being and decrease work productivity [4].
Conventional oral systemic therapies, including methotrexate (MTX), cyclosporine (CYC) and oral retinoids, are often used to treat patients with moderate to severe psoriasis who are not responsive to topical therapies alone. These systemic immunosuppressant medications have risks for serious toxicities and adverse effects, including renal toxicity, hepatotoxicity, hematotoxicity and gastrointestinal toxicity [5]. Thus, patients treated with oral systemic therapies need to be monitored regularly with laboratory tests of blood cell counts and liver and renal function. Despite these toxicities, these inexpensive therapies, namely MTX and CYC, have been the mainstay for treating moderate to severe psoriasis for decades.
MTX and CYC may interact with a number of common medications and this is of particular concern for patients with psoriasis because psoriasis is associated with increased prevalence of several comorbidities [6,7,8,9,10]. Concomitant use of these drugs to control chronic conditions or comorbidities could alter or prolong the blood concentrations of MTX/CYC and enhance their toxicities, which can cause adverse effects and compromise the treatment efficacy.
Little is known about the clinical and economic ramifications of these potential drug-drug interactions in the real-world setting [11,12]. Other researchers have investigated the topic of comedication due to comorbidities in psoriasis patients and its interference with psoriasis treatments. However, to our knowledge, no studies on this topic have been conducted in a US psoriasis population of all severities, and none have investigated the clinical and economic ramifications of comedication. Given the seriousness of toxicities associated with systemic therapies and potential intensification of these toxicities by drug-drug interactions, this study used a large claims database to investigate the prevalence of real-world polypharmacy of oral conventional systemic therapies, in particular MTX and CYC, and drugs that may interact with them. In addition, the rates of potential adverse events were estimated, as well as the implications for health care resource utilization and costs associated with potential drug-drug interactions among psoriasis patients with conventional systemic therapies, MTX or CYC, as a group.
Methods
Data Source
The study used the Ingenix® Impact National Managed Care Database from 1999 to 2007. This database includes the medical and pharmacy history of >60 million patients from 46 health plans from all census regions of the USA and contains information on member eligibility, inpatient confinement, medical services utilization and pharmacy claims.
Study Population and Study Period
Patients were included in the study if they met the following criteria: (1) at least 1 diagnosis of psoriasis [based on International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) 696.1x codes] [13], (2) continuous insurance for at least 6 months before and after the first prescription fill of an oral conventional systemic therapy, including MTX or CYC (i.e. the index date and also the index medication), and (3) at least 18 years of age 6 months prior to the index date. Patients with both MTX and CYC on the index date and no other interactions were excluded because MTX and CYC are known to interact with each other and this analysis focuses on drug-drug interactions associated with comorbidity-related drugs instead of relationships between combinations of psoriasis therapies. However, the patients may have initiated treatment on only MTX or CYC and later on used combination therapy sometime during the study period. The baseline period included 6 months prior to the index date, and the patients were followed for 6 months after the index date (i.e. the study period) for the outcomes evaluation.
Study Cohorts
The study population was separated into 2 mutually exclusive groups. A list of drugs that could potentially interact with MTX or CYC was first created using their respective product labels [6,7]. All drugs that could potentially interact with MTX or CYC and lead to an adverse event were included. However, drugs known to decrease MTX or CYC concentration were not included. Patients who were taking any drug that could interact with MTX or CYC 30 days before or 30 days after starting MTX or CYC were included in the exposed cohort. The nonexposed cohort included patients without any potential drug-drug interaction 30 days prior to the index date and 6 months following the index date.
Sensitivity Analysis
In a sensitivity analysis, the exposed cohort was restricted to a subgroup of patients with potential drug-drug interactions within 30 days before and after the index date. This subgroup was intended to define potential drug-drug interactions with a more stringent criterion by selecting patients who were taking a potentially interactive drug and refilled their drug prescription even after the initiation of MTX or CYC.
Statistical Analyses
Patient Characteristics. Patient characteristics were reported for both the exposed and the nonexposed cohorts. χ2 tests were performed for categorical variables and Wilcoxon rank sum tests were performed for continuous variables to test differences between the exposed and the nonexposed cohorts.
Potential Adverse Events. The prevalence of relevant events that may be associated with drug-related toxicities, including pulmonary, renal, hematologic, gastrointestinal and hepatic events, were identified using diagnosis codes from all medical claims (e.g. inpatient admissions, outpatient visits and emergency room visits) (online suppl. table 1, for all supplementary materials see www.karger.com?doi=10.1159/000275198). Psoriasis patients with any of these events at baseline were excluded. The study period was truncated at the first occurrence of a relevant event. Odds ratios (ORs) from logistic regressions were reported.
Health Care Resource Utilization. The utilization measures included inpatient admissions, emergency department (ED) visits, emergency care (inpatient and ED), outpatient visits and other medical services (e.g. laboratory, radiology or other ancillary services).
Incidence rate ratios (IRRs) and ORs were estimated between cohorts using generalized linear model (GLM) regression models with a log link function and negative binomial distribution for IRRs and logistic regression models for ORs.
Health Care Costs. Health care costs were adjusted for inflation and expressed in USD 2007. The cost components included pharmacy, inpatient, ED, outpatient and other medical service costs.
Incremental costs between the cohorts were reported using GLM models with a log link and a γ-distribution or 2-part models, where the first part is a logistic model with a binomial distribution and the second part is a GLM model with a log link and a γ-distribution, for cost components with a portion of 0 values >5%. The p values were estimated using a nonparametric bootstrap resampling technique of 500 iterations.
Study events, health care resource utilization and cost were compared between the exposed and nonexposed cohorts during the 6-month study period. In addition, appropriate multivariate regression models controlled for age, sex, comorbidities with a prevalence >2%, health care resource utilization and psoriasis treatment during the baseline period. Furthermore, ORs of potential adverse events were also adjusted for the number of other concomitant medications used during the 30 days before or 30 days after starting MTX or CYC.
Results
Sample Selection
Figure 1 shows the sample selection process. The study included 7,955 psoriasis patients who were taking either CYC or MTX; 3,372 patients (42.4%) comprised the nonexposed cohort and 4,583 (57.6%) comprised the exposed cohort. Among the exposed cohort, 3,629 patients (79.2%) initiated on MTX and 954 patients (20.8%) on CYC.
Sample selection process and study population. a Patients with drug-drug interactions between 30 days after the index date and the end of the study period were included. The index date was defined as the first prescription fill of MTX or CYC. b The nonexposed cohort included patients without drug-drug interactions 30 days prior to the index date or 6 months after the index date. c Drug-drug interactions were defined as a combination of MTX or CYC with any of the following drugs: for MTX, nonsteroidal anti-inflammatory drugs, trimethoprim, salicylates, phenylbutazone, sulfonamides, penicillins, azathioprine, retinol, tretinoin, isotretinoin, alitretinoin, etretinate, acitretin, tazarotene, bexarotene, adapalene, theophylline, chloramphenicol, phenytoin, tetracycline or cyclosporine; for CYC, nonsteroidal anti-inflammatory drugs, gentamicin, tobramycin, vancomycin, trimethoprim, clarithromycin, erythromycin, quinupristin, melphalan, amphotericin, ketoconazole, fluconazole, itraconazole, diclofenac, naproxen, sulindac, colchicine, cimetidine, ranitidine, tacrolimus, diltiazem, nicardipine, verapamil, methylprednisolone, allopurinol, bromocriptine, danazol, metoclopramide, amiodarone, digoxin, lovastatin, methotrexate, HIV medications, potassium-sparing diuretics, prednisolone or rifabutin. d The exposed cohort included patients with potential drug-drug interactions 30 days prior to the index date or 30 days after the index date.
Sample selection process and study population. a Patients with drug-drug interactions between 30 days after the index date and the end of the study period were included. The index date was defined as the first prescription fill of MTX or CYC. b The nonexposed cohort included patients without drug-drug interactions 30 days prior to the index date or 6 months after the index date. c Drug-drug interactions were defined as a combination of MTX or CYC with any of the following drugs: for MTX, nonsteroidal anti-inflammatory drugs, trimethoprim, salicylates, phenylbutazone, sulfonamides, penicillins, azathioprine, retinol, tretinoin, isotretinoin, alitretinoin, etretinate, acitretin, tazarotene, bexarotene, adapalene, theophylline, chloramphenicol, phenytoin, tetracycline or cyclosporine; for CYC, nonsteroidal anti-inflammatory drugs, gentamicin, tobramycin, vancomycin, trimethoprim, clarithromycin, erythromycin, quinupristin, melphalan, amphotericin, ketoconazole, fluconazole, itraconazole, diclofenac, naproxen, sulindac, colchicine, cimetidine, ranitidine, tacrolimus, diltiazem, nicardipine, verapamil, methylprednisolone, allopurinol, bromocriptine, danazol, metoclopramide, amiodarone, digoxin, lovastatin, methotrexate, HIV medications, potassium-sparing diuretics, prednisolone or rifabutin. d The exposed cohort included patients with potential drug-drug interactions 30 days prior to the index date or 30 days after the index date.
The 2 study cohorts exhibited different characteristics at baseline (table 1). The patients in the exposed cohort were older (51.0 vs. 47.6 years; p < 0.0001) and a greater percentage were women (56 vs. 52%; p < 0.0001). The prevalence of comorbidities was generally greater in the exposed cohort and the 4 most prevalent comorbidities were hypertension (30 vs. 16%; p < 0.0001), psoriatic arthritis (35 vs. 14%; p < 0.0001), hyperlipidemia (26 vs.19%, p < 0.0001) and rheumatoid arthritis (24 vs.10%; p < 0.0001). The patients in both cohorts had similar histories of topical therapy use during the baseline period, but more patients in the exposed cohort used retinoids (15 vs. 7%; p < 0.0001).
Among the drugs with potential interactions with MTX or CYC, nonsteroidal anti-inflammatory drugs (NSAIDs) and antibiotics were the 2 most common drug classes prescribed to patients who were taking MTX or CYC. The prevalence for each of the studied potential drug-drug interactions is shown in supplementary table 2.
Adverse Event Rates
The rate of occurrence of relevant adverse events during the study period was greater in the exposed than in the nonexposed cohort (tables 2 and 3). Compared with the nonexposed cohort, the exposed cohort initiating on CYC had significantly greater rates of gastrointestinal events (OR: 2.83; p < 0.0001) and pulmonary events (OR: 1.62; p = 0.0017). The renal event rates were also greater in the exposed cohort initiating on CYC than in the nonexposed cohort (OR: 2.39; p = 0.1186). Similarly, the exposed cohort initiating on MTX had significantly greater rates of renal events (OR: 4.33; p = 0.0061), hematologic events (OR: 1.56; p = 0.0014) and pulmonary events (OR: 1.43; p = 0.0005) compared with the nonexposed cohort.
When looking at all the study patients together, the most frequent events in the exposed and nonexposed cohorts were pulmonary (11 vs. 8%), hematologic (5 vs. 4%) and gastrointestinal events (5 vs. 3%). After adjusting for covariates at baseline, 3 types of adverse event remained significant. Renal events were more than twice as likely to occur in the exposed than in the nonexposed cohort (OR: 2.58; p = 0.0145). The likelihood of developing a gastrointestinal event was 36% greater in the exposed cohort than in the nonexposed cohort (OR: 1.36; p = 0.0197), and the likelihood of developing a pulmonary event was 20% greater (OR: 1.20; p = 0.0470). The differences between the cohorts for rates of hepatic and hematologic events were not statistically significant (table 3).
Health Care Resource Utilization
Utilization of health care resources was significantly greater among the exposed than the nonexposed patients (table 4). After controlling for confounding factors, the rates of inpatient and ED visits were 61 and 31% greater, respectively, for the exposed cohort compared with the nonexposed one (inpatient IRR: 1.61, p < 0.0001; ED IRR: 1.31, p < 0.0022). The adjusted odds of having an urgent care event during the study period were also significantly greater among the exposed patients (OR: 1.47; p < 0.0001).
Health Care Costs
The exposed cohort experienced significantly greater health care costs for all categories of medical service considered (table 5). During the 6-month study period, the average total cost for pharmacy and medical services per patient was 59% greater for the exposed cohort compared with the nonexposed cohort (USD 7,960 vs. 5,021, p < 0.0001). The adjusted incremental total costs for the exposed cohort were USD 1,722, 28% greater compared with the nonexposed cohort’s total costs (p < 0.0001). The adjusted incremental pharmacy (USD 711, p < 0.0001) and outpatient costs (USD 539, p < 0.0001) comprised a substantial portion of the total cost differential.
Sensitivity Analysis
The sensitivity analysis included a subgroup of 2,704 patients in the exposed cohort who refilled their prescriptions for potentially interactive drugs after the initiation of MTX or CYC. Compared with the nonexposed cohort, this subgroup had an almost 3-fold greater risk of developing a renal event (OR: 2.82; p = 0.0138) and a 38% greater likelihood of developing a gastrointestinal event (OR: 1.38; p = 0.0377; table 6).
The sensitivity analysis also demonstrated greater total incremental health care costs for the exposed cohort compared with the nonexposed cohort (table 7). The adjusted incremental total costs were USD 1,905 greater per person over a 6-month period in the exposed than in the nonexposed cohort (p < 0.0001). This value represents USD 400 more per year than the original analysis, in which the total incremental costs for the exposed cohort were USD 1,722 (p < 0.0001).
Discussion
This study investigated the prevalence of potential drug-drug interactions between the conventional systemic therapies, MTX and CYC, and various drugs typically taken by patients with psoriasis. In addition, we evaluated the outcomes of potential drug-drug interactions, such as adverse events, healthcare resource utilization and costs, by comparing the patients exposed to drug-drug interactions with those who were not exposed. The study did not aim to determine which immunosuppressant was better in terms of safety or tolerability. The study looked at all events that may be caused by toxicities related to these 2 oral conventional systemic therapies. These analyses were performed using a large, US managed-care database that included, but was not limited to, patients with psoriasis.
The results demonstrated that more than half of the patients treated with MTX or CYC also received prescriptions for interacting drugs at the time of MTX/CYC initiation. Compared with the patients who were not taking interacting drugs, these patients experienced significantly greater risks of pulmonary, renal and gastrointestinal events and incurred substantially greater costs for resource utilization and treatment. A sensitivity analysis that used more stringent inclusion criteria for selecting exposed patients was confirmatory, with results that were robust and more pronounced than for the initial exposed cohort, especially for renal events.
Even at recommended dosages of MTX or CYC, there is a risk of renal or gastrointestinal toxicity when MTX or CYC are coadministered with compounds that increase their blood concentration and/or decrease their clearance [15,16]. In this study, the exposed patients experienced greater risks of these particular adverse events, consistent with the known clinical consequences of those drug-drug interactions. Furthermore, coadministration of drugs that may interact with MTX or CYC and that could be deemed high-risk interactions, based on information available in the product labels [6,7], was relatively frequent.
The most commonly used drugs with potential to interact with both MTX and CYC in the exposed cohort were NSAIDs and sulfonamides. NSAIDs are often used to treat acute or chronic pain and inflammatory conditions, such as rheumatoid arthritis and psoriatic arthritis. The interaction of NSAIDs with MTX elevates and the prolongs serum concentrations of MTX and increases the risk of aplastic anemia, bone marrow suppression and gastrointestinal toxicity [6,17]. Concomitant administration of CYC and NSAIDs can be associated with a more pronounced decrease in renal function than that seen with either agent alone and is linked to renal toxicity [7,18]. When taken alone, NSAIDs may lead to renal or gastrointestinal effects, particularly in patients with pre-existing chronic renal disease or with decreased effective circulating fluid volume [19,20,21]. Most guidelines for psoriasis treatment do not recommend nonbiologic systemic therapies, especially high doses of MTX, for patients who are already at risk of renal and gastrointestinal toxicity events or who have uncontrolled hypertension [6,8,22].
Sulfonamides are also frequently used to treat patients with psoriasis and can potentially interact with MTX or CYC. Sulfonamides are synthetic antimicrobial agents and are also applied in the treatment of inflammatory bowel disease. Sulfonamides can diminish renal tubular transport of MTX, thus increasing MTX concentrations and the risk of toxicities, including renal toxicity, hepatotoxicity and bone marrow suppression [23,24,25]. Penicillin is often used as treatment for guttate psoriasis, which is usually a streptococcal (bacterial) infection that may be responsible for chronic plaque psoriasis [26,27,28].
The patients in the exposed cohort experienced significantly greater health care costs than those in the nonexposed cohort. The baseline characteristics showed that the exposed patients had greater prior resource use and more extensive comorbidity profiles compared with the nonexposed cohort. To control for selection bias, we also controlled for baseline healthcare resource utilization (inpatient, ED, outpatient), a comprehensive list of comorbidities and history of psoriasis treatments prior to initiation of MTX/CYC systemic therapy.
Because of the risk of toxicities with conventional systemic therapies, their use requires comprehensive and regular laboratory monitoring, which can increase the total costs [29,30]. A lack of efficacy with these immunosuppressant agents may lead to dose titration or use of combination therapies [31,32]. Poor response and issues related to drug safety and tolerability may result in either worsening of the disease or an adverse effect and potential hospitalization.
Although claims data are a rich source for examining health care utilization and costs, there are some inherent limitations in any study conducted with claims data. This investigation is based on administrative claims data in which patients’ psoriasis severity, adverse event severity and comorbidities could not be directly observed. Therefore, despite the comprehensive list of covariates used in the study, selection bias may still remain. In addition, pharmacy claims data do not provide information on whether drugs dispensed were actually taken according to prescribed instructions. Thus, the study determined potential drug-drug interactions under the assumption that patients took their medicine as prescribed. This limitation was overcome to a certain extent in the sensitivity analysis, in which stricter criteria were used. Another drawback is that the study was limited to prescribed medication and did not considerer over-the-counter products that could also be associated with drug-drug interactions; thus, the findings may have underestimated the impact of drug-drug interactions.
In addition, with claims data, it is not possible to be certain that an adverse event is related to a drug-drug interaction. The adverse event rates reported in this study were based on the emergence of specific ICD-9 diagnosis codes in patients’ medical claims that were used in the definition of potential drug-related toxicities. These rates represent all patients who experienced the studied toxicity during the follow-up period in the case and control cohorts. Patients with the studied conditions at baseline were excluded from the analysis. Based on the temporal relationship between concomitant drug use and the emergence of these ICD-9 codes, one could infer that at least some of the diagnoses were due to drug interactions. However, the conditions identified by the ICD-9 diagnosis codes may not have solely resulted from drug use. For example, conditions may have developed irrespective of a drug interaction with MTX or CYC because of patients’ lifestyle factors (e.g. tobacco use), environmental factors or family histories; information about such factors was not available in the claims data.
To minimize the impact of factors that could not be controlled for in our analysis of claims data, the same definition of toxicity was applied to both cohorts. The results, therefore, show that in a relative sense, exposed patients were more likely to experience a studied toxicity than nonexposed patients. The study findings are based on relative risks. Events that were not directly related to drug toxicities should have been as frequent in both cohorts and, therefore, would not have had an impact on the relative risk.
Previous studies have shown that patients with psoriasis have an increased prevalence of multiple comorbidities [9,10]. This study found that, overall, 50% of the patients treated with conventional oral systemic therapies (MTX or CYC) may experience potential drug-drug interactions due to comedication with interacting drugs. The total impact of these interactions (increased toxicity risk, healthcare resource use and overall costs) is substantial. Caution should be exercised when prescribing conventional systemic therapies to patients with psoriasis because of the likelihood and potential effects of interactions with concurrent medication use. Both the drug and the patient’s unique characteristics need to be considered when making treatment decisions to promote optimal outcomes, even if that means selecting more costly therapies. As has been previously demonstrated, certain agents may have safer treatment profiles and promote better adherence that ultimately offset a psoriasis patient’s total cost of care [33].
Acknowledgment
The authors thank Arbor Communications, Inc., Ann Arbor, Mich., USA, for editorial assistance on behalf of Abbott Laboratories. This analysis was supported by Abbott Laboratories.
Disclosures
J.-H.S. has served as a consultant for Abbott Laboratories and has participated in continuing medical education events supported by unrestricted educational grants from Abbott Laboratories. The work reported here was performed under contract for Abbott Laboratories by Analysis Group. A.P.Y. and E.Q.W., A.G. and D.L.-V. are employees of Analysis Group. S.R.G., Y.B. and P.M.M. are employees of Abbott Laboratories.