Costs of Treatment in Patients With Moderate to Severe Plaque Psoriasis: Title and subTitle BreakEconomic Analysis in a Randomized Controlled Comparison of Methotrexate and Cyclosporine FREE

PSORIASIS IS a chronic disease for which only remissive, as opposed to curative, treatments are currently available. Methotrexate and cyclosporine are now the predominant systemic treatment modalities for patients with moderate to severe psoriasis. They have been used as such for the past 40 and 20 years, respectively. Therapy with methotrexate has been evaluated in noncomparative trials, reporting effective induction and maintenance of remission in patients with severe psoriasis.¹ In addition, there is strong evidence supporting the use of cyclosporine, and optimal dosage and treatment schemes for cyclosporine have been systematically assessed.²

In clinical decision making, the choice of treatment is primarily based on evidence of its effectiveness, adverse effects, safety, and (dis)comfort. Methotrexate and cyclosporine therapy have been shown to be effective, but their unit costs differ substantially. A distinct profile of adverse effects and possible differences in clinical course (induction of remission and relapse) may generate additional differences in medical resource utilization or have other economic implications. From the perspective of efficient allocation of health care resources, total costs associated with treatment should also be taken into account.³

We conducted an economic evaluation of these treatments as part of the first prospective randomized controlled trial, to our knowledge, designed to compare the effectiveness of methotrexate and cyclosporine treatment in patients with moderate to severe psoriasis.⁴ Assuming that methotrexate and cyclosporine are equally effective in their reduction of disease activity, the trial was designed to detect a potential clinically significant difference after 16 weeks of treatment. The economic evaluation studied direct and indirect costs during 16 weeks of treatment and 36 weeks of follow-up, producing estimates of 1-year cumulative total costs of treatment.

Details of the randomized controlled trial can be found elsewhere.⁴

Patients were eligible for the randomized controlled trial if they had moderate to severe psoriasis and no previous treatment with methotrexate or cyclosporine. A range of precautionary safety criteria was used.

During treatment, a strict regimen was followed, including increases and decreases in medication dosages related to insufficient effectiveness and safety criteria, respectively. After 16 weeks, oral treatment was discontinued and the use of subsequent (active) therapies was discouraged so that we could evaluate the duration of remission and relapse of disease activity.

Patients visited the outpatient clinic of the Academic Medical Center, University of Amsterdam, and were seen every other week during the first month and every 4 weeks in the subsequent 12 weeks of treatment and 36 weeks of follow-up. During these outpatient contacts, the following clinical, functional, and economic outcome measures were documented: severity of psoriasis (using the Psoriasis Area and Severity Index [PASI]⁵ ), symptoms and adverse effects (in terms of frequency and intensity and their relation to the study medication), health-related quality of life (using the Medical Outcomes Study 36-item Short-Form Health Survey [SF-36]⁶ ), and resource utilization. Prescriptions of other drugs related to adverse effects, trial medication, and UV-B phototherapy sessions during follow-up were documented in the medical record.

We used a bottom-up strategy to prospectively collect data on utilization of health care and other resources, in parallel to the clinical process. The duration of the study was 1 year, including treatment and follow-up. The cost analysis was set up from a societal perspective,³ implying that costs indirectly generated by health care utilization (traveling, informal care, and productivity loss) are also taken into account.

Costs were classified into 3 categories.^{7 - 8} Direct medical costs are generated by health care utilization (medication, diagnostic procedures, and visits to health care providers). In this case, direct medical costs were generated by outpatient visits, study medication, comedication and therapies applied during follow-up, diagnostic and laboratory tests, additional visits to health care providers, and out-of-pocket expenses. Direct nonmedical costs reflect utilization of non–health care resources generated by the condition and intervention (personal care of psoriasis and travel expenses). Indirect costs are associated with lost or impaired ability to work or to engage in leisure activities (productivity loss due to absence from work and time spent by patients or relatives in personal care).⁸

Resource data were collected prospectively. A detailed registration system was set up and incorporated into the clinical record form documenting additional health care utilization, absence from work, and time associated with treatment or personal care of psoriasis. Prescription of comedication and number of UV-B phototherapy sessions were extracted during a review of the clinical records. Absence from work was documented by using a standardized questionnaire.⁹ Existent administrative registration sources (hospital information systems) were used for validation or completion of prospective data.

Estimates of unit costs were based on calculations of real costs,^{10 - 11} Dutch pharmaceutical unit cost listings,¹² guideline prices,⁸ and tariffs.¹³ Productivity loss was estimated using the friction cost approach⁸ and the shadow-pricing method. The friction cost approach assumes that costs of lost productivity are limited to the time required to replace someone in the workforce (friction period); the shadow-pricing method imputes hypothetical incomes for lost productivity for unpaid activities (household and volunteer work). Time spent in personal care by relatives was also appraised as volunteer aid,⁸ whereas patients' own time was not separately appraised. Unit costs were set at the 1999 price level⁷ and were discounted by 3% annually.

Multiplying the respective volumes of resource use with their corresponding unit prices resulted in the associated total costs. Costs were calculated in the European currency (Euro) and converted to US dollars (exchange rate: €1 = US $1, December 31, 1999). Total costs were estimated for each period between 2 outpatient visits (including the latter) and aggregated to summarize 16 weeks of treatment and 36 weeks of follow-up.

Separate estimates were made for different cost categories. Protocol costs—costs associated with the study itself—were excluded from the analyses. Parallel to the clinical study, the comparison between treatments was according to the intention-to-treat principle.¹⁴ Although conservative management with respect to active treatment was propagated, patients who insisted were allowed to use some form of therapy (including systemic). In the case of systemic therapy, cost-specific data collection was discontinued, and cost estimates are based on calculations during the treatment period. Fluctuation over time was accounted for by imputing the average cost per 4-week episode during 16 weeks of treatment in the corresponding treatment group into the subsequent 4-week periods during follow-up.

Descriptive statistics of the aggregate cost analysis are presented: average cumulative costs per patient during treatment and follow-up and difference in the average between study groups. Thereby, geometric means of successive 4-week episodes are calculated for different cost categories and are accumulated during treatment and follow-up.

In sensitivity analyses, the robustness of the conclusions were evaluated for differences in the unit costs of cyclosporine and UV-B phototherapy. In addition to aggregated cost analyses, in which resource utilization and associated costs are accumulated per study group, the amount of variability in costs of treatment and follow-up across individual patients was estimated. For statistical reasons (nonnormal distribution of cost estimates of individual treatment paths), nonparametric statistical test results are reported.

Between October 13, 1998, and June 15, 2000, 85 patients were enrolled in the study. During the study, patients missed only a few outpatient visits, leading to almost complete data on clinical status and resource utilization.

Demographic and clinical characteristics of the study population are given in Table 1. At baseline, age and sex did not vary systematically between treatment groups, and neither did severity of psoriasis (PASI score) and quality of life (physical and mental health scores on the SF-36).

Thirteen patients did not complete the 16-week treatment because of reversible elevations in liver enzyme levels (12 patients taking methotrexate) or bilirubin levels (1 patient taking cyclosporine). Their use of resources was used in the analysis, in line with the intention-to-treat principle. After 16 weeks, the primary outcome measure of disease activity (PASI score) showed clear and comparable improvement in both groups: from a mean (SD) PASI score of 13.4 (3.6) to 5.0 (4.5) in the methotrexate group and from 14.0 (6.6) to 3.8 (3.0) in the cyclosporine group. Adjusted for baseline, the mean score was 1.3 PASI-units lower in the cyclosporine group (P = .09). The health-related quality-of-life (SF-36) scales showed only marginal improvements on average from baseline in either group (Table 1).

Results of the cost analysis are summarized in Table 2, reporting resource utilization and associated costs for both treatment groups in the respective study periods separately. During 16 weeks of treatment, costs per patient were generated by outpatient visits in both groups (of which 5 had been planned in the trial protocol), generating costs of approximately $490. Costs for medication were a substantial additional cost factor in the cyclosporine treatment group ($890 vs $16 in the methotrexate group) based on the intake of 77 tablets of 2.5 mg of methotrexate and 971 capsules of 25 mg of cyclosporine or the equivalent in different dosage units.

Treatment with methotrexate induced an average of an additional 3 days of sick leave from work (resulting in lost productivity), generating more indirect costs ($469 vs $290 in the cyclosporine group). Modest differences in health care utilization and associated costs are related to more precautionary measures (radiography and liver echography) in the methotrexate group and more UV-B phototherapy in patients in this group who discontinued trial medication use. In addition, during treatment and follow-up, patients in the methotrexate group reported more time of their own and of their relatives spent in personal care of the psoriasis: approximately an additional 20 minutes and 10 minutes per week, respectively, than patients in the cyclosporine group.

The follow-up period showed more variability in utilization of health care and pharmacologic resources. Overall, patients in the cyclosporine group used substantially more topical therapy than and almost twice as much UV-B phototherapy as patients in the methotrexate group. On the other hand, the methotrexate group reported substantially more day care treatments than patients treated with cyclosporine. In addition, more time spent on personal care by patients and their relatives in the methotrexate group induced more nonmedical costs in the methotrexate group. Hospital visits for UV-B phototherapy generated additional traveling costs in the cyclosporine group.

We found a difference between treatment groups in average total costs for the treatment period of $521 per patient in favor of methotrexate (Table 3). During follow-up, the difference in average costs was $112 in favor of cyclosporine treatment. Consequently, taking into account all relevant costs, the 1-year difference was $409 for cyclosporine treatment vs methotrexate treatment, comparable to an additional 10% for cyclosporine treatment vs the total costs of treatment with methotrexate.

An accumulation of costs during treatment and follow-up, stratified by different types of costs, is presented in Figure 1. During treatment, cost differences were predominantly due to the higher unit costs of cyclosporine, health care utilization, and productivity loss by sick leave during methotrexate treatment. During follow-up, the profile of medical and other resources used is slightly different: in the cyclosporine group, more costs were associated with UV-B phototherapy, whereas in the methotrexate group, more costs were generated by (other) health care utilization. The respective additional total costs are comparable.

Estimates of individual costs were found to vary substantially across patients rather than between therapies. Distributions were skewed, with a few patients generating relatively high costs (Figure 2). The plot visualizes the distribution of total costs by treatment group for the treatment and follow-up periods separately. The treatment period shows a clear difference between treatments, with limited variation among most patients and a few outliers with high costs. The statistical test pointed to a significant difference during the treatment period (Kolmogorov-Smirnov = 3.645 [2-sample test]; P<.001). During follow-up, however, the box sizes reflect substantial individual variation in costs, with only marginal (and statistically nonsignificant) differences between treatment groups (Kolmogorov-Smirnov = 0.764 [2-sample test]; P = .60). Medians (25th-75th percentiles) of estimated costs of 1-year treatment and follow-up were $2574 ($1873-$4143) and $3498 ($2814-$5319) for methotrexate and cyclosporine, respectively.

In this article, we present data from the economic evaluation of 2 established systemic treatment modalities— methotrexate and cyclosporine—widely used in the treatment of moderate to severe psoriasis.⁴ Clinical outcomes showed comparable results and are more elaborately reported and discussed elsewhere.⁴ Quality of life as measured by the SF-36 also resulted in comparable outcomes yet showed only marginal improvement after 16 weeks of treatment, probably owing to the insensitivity of the SF-36 to disease-specific aspects of psoriasis. Given the considerable discrepancy in unit costs of methotrexate and cyclosporine, and the chronic character of psoriasis, substantial differences in costs associated with initial treatment or management of psoriasis during follow-up were expected.

Cyclosporine treatment was found to generate more costs during 16 weeks of treatment, mainly owing to higher direct medical costs. During follow-up, treatment patterns varied, with slightly higher total costs in the methotrexate group. However, in contrast with the cost difference suggested by the unit costs, 1 year after randomization, the difference between study groups was only 10% of the total costs. This is explained by the finding that costs associated with systemic treatment of psoriasis are not restricted to medication costs but also involve other costs directly and indirectly generated by utilization of health care resources.

Relatively more patients had to discontinue methotrexate treatment. Consistent with the intention-to-treat approach in the analysis of the clinical outcome, follow-up resource utilization by these patients (eg, UV-B phototherapy) during the first 16 weeks is therefore summarized as costs for the treatment period. The reported costs associated with resource utilization over time, therefore, reflect real costs of daily clinical routine, not just costs associated with treatment according to the protocol.¹⁴

A recent study¹⁵ in patients with rheumatoid arthritis showed that supplementation of folates substantially reduces the incidence of hepatotoxicity during methotrexate treatment. It is likely that early discontinuation of treatment in the methotrexate group could have been avoided, achieving better clinical results at minimal additional efforts or costs.

During the study, not all patients could be contacted at each scheduled visit. The number of missing data was small, with no indication of systematic or informative missing data.

The observed differences in costs associated with study medication are highly affected by the unit costs of methotrexate and cyclosporine. If unit costs for cyclosporine decrease more than those for methotrexate, the margin in total costs will further decrease, leading to negligible differences.

The pattern in clinical outcome and resource utilization during follow-up reflects individualization in disease management, which could hardly be systematically related to the allocated initial treatment. Some patients could defer subsequent treatment, whereas others almost immediately restarted topical therapy. Series of UV-B phototherapy sessions and even renewed systemic medication were reported during follow-up for many participants (42%). The question arises whether this subsequent treatment is invoked by a substantial relapse in disease activity or by patients' preferences for maintenance of the achieved treatment results. A detailed assessment of the relation among disease activity, treatment, and related resource utilization requires more complex analyses and is beyond the scope of this article.

Which treatment provides the greatest health benefit for each dollar spent?¹⁶ According to recommended methods in this area, the efficiency of treatment alternatives is expressed as cost-effectiveness, that is, costs of achieving 1 unit of health outcome (eg, costs per year of life gained). Applying this approach to the evaluation of psoriasis treatment is less obvious, as conventional measures for health gains, such as life-years gained, have limited value in the evaluation of psoriasis treatment. So far, no appropriate alternative psoriasis-related measure is available for calculating cost-effectiveness. The PASI score, used in this trial as an expression of disease severity, has limited scale properties: a PASI score change from 40 to 35 is not likely to be equivalent to a similar 5-point PASI score change from 10 to 5. Its quantitative scale application in cost-effectiveness calculations is therefore questionable. The concept of psoriasis-free years as a measure of effectiveness operationalized in an economic simulation model¹⁷ is not appropriate in our study, as the aim was not to obtain periods with complete remission but rather to compare treatments in terms of onset and rate of remission and relapse induced by a standardized schedule.

A balanced appraisal of the different treatment strategies for moderate to severe psoriasis not only takes into account effectiveness or cost considerations but also considers adverse effects, safety of treatment, and discomfort for patients. In a model with a long-term horizon for cost-effective management of psoriasis, adverse effects as well as safety and convenience aspects need to be integrated. Patients may differ in their judgment of these aspects: individual tradeoffs between potential long-term safety risks and short-term relief or between practical application of oral therapy and more inconvenient topical treatment or frequent hospital visits (eg, for UV-B or psoralen–UV-A phototherapy).

Compliance behavior and disease management are important issues in psoriasis.¹⁸ Patients are "experts" by experience.¹⁹ Consequently, from the perspective of shared decision making, to select the appropriate treatment, physicians should assess and integrate individual patients' preferences and tradeoffs of relevant aspects of disease and treatment in the decision-making process.²⁰

Within the concept of rotational therapy for psoriasis, treatment selection aims at optimizing health risks and benefits, taking into account convenience and costs. The clinical outcomes of this randomized controlled trial showed comparable results of treatment with either methotrexate or cyclosporine. This economic analysis demonstrated that in contrast to the considerable difference in medication costs of systemic therapies, these costs make up for only a small fraction of the total treatment costs, with a small between-therapy margin after 1 year of treatment. Combining both results implies that economic arguments are most likely not decisive for individual patient decisions or for the development of guidelines for systemic therapy.

Corresponding author: Brent C. Opmeer, PhD, Department of Clinical Epidemiology and Biostatistics, Academic Medical Center, University of Amsterdam, PO Box 22660, 1100 DD Amsterdam, the Netherlands.

Accepted for publication June 25, 2003.

This study was supported by grant OG 97-009 from the Dutch Health Insurance Board, Diemen, the Netherlands.

We thank Hans B. Reitsma, MD, PhD, for his valuable critical comments on the manuscript.