Rheumatology Advance Access originally published online on May 17, 2008
Rheumatology 2008 47(7):1044-1050; doi:10.1093/rheumatology/ken141
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Treatment impact on estimated medical expenditure and job loss likelihood in rheumatoid arthritis: re-examining quality of life outcomes from a randomized placebo-controlled clinical trial with abatacept
1QualityMetric, Lincoln, RI and 2Bristol-Myers Squibb, Princeton, NJ, USA
Correspondence to: J. C. Cole, QualityMetric, Inc., 640 George Washington Highway, Suite 201, Lincoln, RI 02865-4207, USA. E-mail: jcole{at}qualitymetric.com
 |
Abstract |
|---|
 Top Abstract IntroductionParticipants and methodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
Objectives. Quality of life (QoL) improvement is important to demonstrate in RA clinical trials, but can be abstract. More meaningful measures of QoL include medical expenditure and job loss, aspects that have marked importance for RA patients, physicians and society. We re-examined previous positive QoL findings for abatacept over placebo by converting existing QoL measures into estimated medical expenditure and estimated likelihood of job loss.
Methods. Two double-blind, placebo-controlled, multicentre randomized clinical trials were undertaken: one for MTX failure (n = 652) and one for more severe anti-TNF failure patients (n = 391). Based on derived scores using previously published formulae, measures of monthly medical expenditure, current inability to work and job loss at 6 months, 1 yr and 2 yrs were analysed.
Results. Abatacept led to greater reduction in medical expenditure over time in MTX failure ($152 lower) and anti-TNF failure patients ($122 lower) compared with placebo at end-point. Likewise, significantly more reduction in likelihood for current and future job loss was achieved with abatacept compared with placebo, which has 25–64% greater likelihood.
Conclusions. QoL changes provided greater reduction in medical expenditure and likelihood of an inability to work. The strong effect sizes obtained for all significant analyses suggest that the results are clinically meaningful. Moreover, given the nature of the variables, results should also be meaningful for patients, physicians, employers and health care insurance entities. Limitations are discussed regarding using estimated outcomes rather than analysis of actual outcomes.
KEY WORDS: Medical expenditure, Job loss, Quality of life, Rheumatoid arthritis, Abatacept
|
Introduction |
|---|
|
TopAbstractIntroductionParticipants and methodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
Rheumatoid arthritis (RA) is a chronic, systemic and inflammatory disorder that primarily involves body joints. This symmetrical disease often progresses from peripheral to more proximal joints, resulting in marked decrements in both clinical symptoms and patient-reported outcomes [1]. Whereas most studies of RA focus on disease activity and loss of function (e.g. patient-reported loss of physical functioning), patient participation in daily life is frequently not assessed [2] despite its being noted as the third domain that should be assessed in the World Health Organization's International Classification of Functioning, Disability, and Health [3]. Various studies have shown that RA patients have an increased risk of current and future unemployment, an important measure of participation, because of RA-specific disability [4]. Indeed, Burton et al. [5] reviewed previous literature on job loss due to RA, finding that a median of 66% of employed persons with RA would experience job loss. Precarious employment is just one of the factors that make medical expenditure increases for RA so important. As noted by Yelin et al. [6], the total medical expenditure costs for patients with rheumatic conditions were $321.8 billion in the US during 2003. To give scope to this number, it is the equivalent of 3% of the gross national product of the United States. Yelin [7] previously noted that ‘economists consider a 1% fall in Gross National Product as being what happens when you have a[n economic] depression’. With
1% of the world population suffering from RA [8], the financial impact on the workforce from RA can be quite substantial. Wolfe [9] has noted that the burden placed on patients, physicians and society demands that we find better ways to prevent negative outcomes, such as unemployment and increased medical costs. Because of the exposure of the Bone and Joint Decade, concerns about the individual and population burden of RA are now growing [6]. One of the nation's Healthy People 2010 arthritis objectives is to reduce the unemployment among RA patients.
Predicting the amount of medical expenditures RA patients incur each year helps us to better understand financial burden of RA for patients and their institutions used to finance health care. Measurement of medical expenditure has been assessed as part of the Medical Expenditure Panel Study (MEPS) [10], a survey conducted annually on a large representative sample of the outpatient civilian US adults. Given the large annual medical expenditure among RA patients (estimated at $9519 per patient, or $793 per month, in 2001) [11], understanding the impact of new RA treatments on medical expenditure enhances information about the benefit of these new RA treatments.
Knowledge of the likelihood of RA patients losing their job within 6 months or more or their inability to work (at the time of measurement) provides information on major life events. Based on current QoL measures, one can predict who may lose their job at some future time point as well as reductions in the odds of future job loss from treatment [12]. For example, in a study of 732 RA patients [13], 29% had stopped working because of RA after 5 yrs and work disability at 5 yrs was significantly predicted by use of QoL measures. From disease diagnosis, 22.2% of RA patients are unable to work and 31.5% are unable to work after 10 yrs [14].
Abatacept is a new biologic treatment for RA, the first in a new class of agents for the treatment of RA that selectively modulates a specific costimulatory signal required for full T-cell activation [15]. Abatacept's efficacy on clinical outcomes has been demonstrated during Phase III clinical trials in RA patients with inadequate MTX response [16] as well as for patients who have been unresponsive to anti-TNF treatment [17], a group previously considered to have run out of treatment options [17]. Moreover, substantial QoL benefits have been demonstrated for abatacept on both of these samples [18, 19]. In both studies, significant improvements were found on physical functioning, fatigue, all eight domains of the SF-36 [20], as well as the SF-36 physical and mental component summaries (PCS and MCS, respectively). Improvement rate was faster for abatacept than for placebo on the QoL measures, and the improvements from abatacept related to normal levels of QoL on many domains.
Abatacept previously has been demonstrated to lead to significant improvement in clinical symptoms and health-related QoL (HRQoL; i.e. SF-36 and the HAQ). Because direct measure of QoL decrements and restoration are indicative of changes to other meaningful factors in one's life [21], and because abatacept has demonstrated QoL benefits across a marked breadth of direct QoL measures [19], the current study was undertaken to expand the domains of QoL potentially influenced by abatacept to the previously mentioned major life outcomes: medical expenditure, job loss and inability to work.
Based on estimates derived from the Phase III clinical trial data, these hypotheses were examined using both MTX-unresponsive patients and patients with previous unresponsiveness to anti-TNF therapy.
|
Participants and methods |
|---|
|
TopAbstractIntroductionParticipants and methodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
Participants
Analyses in the current study were conducted separately for two Phase III clinical trials: AIM (Abatacept in Inadequate responders to Methotrexate) [16] examined the efficacy and safety of abatacept treatment in patients with RA with an inadequate response to MTX and ATTAIN (Abatacept Trial in Treatment of Anti-TNF Inadequate responders) [17] examined the effect of abatacept in anti-TNF inadequate responders. Both clinical trials were approved by institutional review boards and independent ethics committees. Both studies were carried out in accordance with the ethical principles of the Declaration of Helsinki. All participants provided written informed consent prior to randomization.
MTX failure sample (AIM)
A total of 652 patients with RA and inadequate response to MTX participated in the AIM trial [16]. A 2 : 1 treatment to placebo sampling was undertaken, resulting in ‘MTX failure abatacept’ group with n = 433 vs MTX and placebo (‘MTX failure placebo’) group with n = 219. Patients received abatacept or placebo at monthly infusion with the background of MTX therapy. Study design has been described elsewhere [16].
Anti-TNF failure (ATTAIN) sample
A total of 391 patients with RA and inadequate response to anti-TNF and MTX participated in the ATTAIN trial [17]. Again, a 2 : 1 treatment to placebo sampling was undertaken, resulting in ‘anti-TNF failure abatacept’ group with n = 258 vs MTX and placebo (‘anti-TNF failure placebo’) group with n = 133. Study design has been described elsewhere [17].
Measures
SF-36 health survey
HRQoL was measured in the AIM and ATTAIN studies using the Medical Outcomes Study (MOS) SF-36. The SF-36 has a well-documented history as a psychometrically appropriate measure of HRQoL [22, 23]. Analyses in the current study used SF-36 composite summary measures, PCS and MCS measures. Standardized scores with a mean of 50 and an S.D. of 10 in the general US population were used for PCS and MCS using norm-based methods; higher scores indicate better HRQoL.
Medical expenditure
Measurement of medical expenditure has been conducted as part of a nationally representative survey of health care utilization and expenditures for non-institutionalized civilians in the US, called the Medical Expenditure Panel Study (MEPS) [10]. Calculation of predicted monthly medical expenditure was derived from a formula based on the work from Fleishman et al. [24] based on the MEPS. As defined in MEPS, expenditures refer to direct payments for medical care, including payments from the patient, private insurance, Medicare, Medicaid and other sources. Expenses included are prescription medications, hospital inpatient stays, home health visits, medical supplies (including vision and hearing aids) and visits to office-based providers, hospital, clinical, emergency rooms and dental providers. Over-the-counter medications and most alternative care are not included in expenses. Unlike many other studies based on claims, MEPS data also include utilization of health services by persons in capitated programmes. Given the marked impact RA has on SF-36 scores, abatacept's ability to significantly and substantially improve SF-36-reported QoL as well as the appreciable predictive power of the SF-36 for medical expenditure [24, 25], it is worth studying the relationship between SF-36 improvements and medical expenditure reduction. This analysis would be based on available annual medical care costs for persons with RA estimated at $9519 per patient (i.e. $793 per month) in 2001 [11].
Based on their examination of predictive regression models on the 2000 MEPS data [10], Fleishman et al. [24] determined that a generalized Poisson regression model with PCS and MCS values had substantial predictive ability for medical expenditure when analysed by gender and age cohorts. Therefore, the current study estimated 30-day medical expenditure for each time point by applying a set of regression weights from Fleishman et al. for each person's age, gender and time-specific PCS and MCS scores (technically, age and gender interacted, and this was reflected in our model). Although the Fleishman et al.'s study was based on the SF-12, correlations between the SF-12 PCS and MCS scores with the SF-36 PCS and MCS scores (respectively) used with the current clinical data sets range from the 0.94s to the 0.98s [21]; any differences should be quite negligible.
Job loss and inability to work
Using data from the MOS [26], questions related to job loss (at 6 months, 1 yr and 2 yrs) and an inability to work were collected. Based on techniques described elsewhere [12], these outcome variables were regressed onto PCS, MCS, age and gender. The logistic regression terms from PCS and MCS were then converted to odds ratios (ORs). OR (i.e. logistic) curves were generated as a function of PCS and MCS scores centered on the population mean, such that OR (mean) = 1. The curves were then applied to the MTX failure and anti-TNF failure populations to estimate OR for these occupational outcomes during the course of the trials. The analyses to determine the weights have been provided as part of the current study.
Protocol
MTX failure (AIM) and anti-TNF failure (ATTAIN) studies were multisite, double-blind, randomized, placebo-controlled clinical trials. Abatacept was given by intravenous infusion in a fixed dose of 10 mg/kg or placebo. Medication was administered on days 1, 15, 29 and every 28 days thereafter. MTX failure (AIM) study continued for 12 months after baseline whereas anti-TNF failure (ATTAIN) study continued for 6 months after baseline. SF-36 questionnaires were completed on days 1, 29, 85, 169 and in the MTX failure (AIM) study only, on day 365.
Data analysis
Figure 1 attempts to clarify the flow of calculations conducted for both sets of outcomes (i.e. medical expenditure and job loss/inability to work). For each outcome, the first phase is the formula derivation wherein a large set of data was used to determine what weights are applied to PCS and MCS in order to derive an appropriate estimate of the outcome. For example, the MEPS database was used previously [24] to determine weight to apply to PCS, MCS, age and gender in order to obtain an estimated monthly medical expenditure. In the second phase (weight application) noted in Fig. 1, weights obtained from the formula derivation phase are then applied to data in the clinical trial databases (AIM and ATTAIN) in order to obtain estimates on the two sets of outcomes for the clinical trial patients. Finally, Phase 3 of Fig. 1 shows that planned statistical analyses (noted next) for each outcome in the clinical trail databases were conducted based on the estimates obtained in the weight application phase.
|
Prior to statistical analysis of the estimated outcomes, group differences were examined to ensure that baseline differences did not augment findings. Many of the relevant clinical, demographic and HRQoL differences have been examined elsewhere [18, 19]; no differences on any pertinent baseline variables were found to differ between treatment and placebo in either MTX failure (AIM) or anti-TNF failure (ATTAIN) samples. Examination of baseline differences on estimated medical expenses and ORs for job loss and inability to work were examined using t-tests.
Medical expenditure
A repeated-measures model was used to examine significant group (abatacept vs placebo) by time interactions, testing if the groups change at different rates over time. Specifically, we used a multivariate analysis of variance (MANOVA) approach to control the assumption of sphericity found in traditional repeated measures ANOVA (sphericity is an assumption that presumes that the variance on the outcome is the same at each time point and the correlations between the time points are all of equal value). Any significant baseline differences on medical expenditure between the groups were controlled by covarying baseline values in the MANOVA model. Additional MANOVA assumption checks were conducted and corrected as needed [27]. Finally, effect size estimates were provided as a measure of how much variance was explained (specifically, we used Cohen's d).
Job loss and inability to work
A total of eight logistic regression models were developed, one each to assess the relationship between PCS or MCS score and each of the four MOS outcomes (i.e. job loss by 6 months, 1 yr and 2 yrs, and inability to work due to health at baseline). Table 1 presents summary statistics for the outcome variables used in these analyses. The patient's age and gender were put in all models to control for any possible effects of age and gender, as these demographic variables can contribute to differences in PCS and MCS scores [23]. Please note that the intervals of 6 months, 1 yr and 2 yrs are fixed based on the MOS database.
|
Only logistic regression models that had statistically significant model fit (examined with the
2 test) and significant regression terms (examined with the Wald test) were used to derive ORs. Using the coefficients derived from the logistic regressions, ORs were calculated based on the relative risk of the MOS outcomes as a function of the initial PCS or MCS score, as compared with the general population mean. OR curves were then examined with the MTX failure (AIM) and anti-TNF failure (ATTAIN) data, estimating the relative risk for these major life outcomes for all patients in both samples. Finally, comparison of the difference in the relative risk between the treatment and placebo group was conducted using an ANOVA where the dependent variable was the OR at end-point and the independent variable group (abatacept vs placebo). Any significant baseline differences were controlled by entering the baseline OR as a covariate; normality and homogeneity of variance assumptions were examined prior to ANOVA calculations. |
Results |
|---|
|
TopAbstractIntroductionParticipants and methodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
Table 2 provides a summary of baseline comparisons between the abatacept and placebo groups [once for MTX failure (AIM) and once for anti-TNF failure (ATTAIN) data] on medical expenditure and the eight ORs (job loss at 6 months, 1 yr and 2 yr and inability to work, each of which was derived via PCS and then MCS). These results showed no significant difference at baseline between the abatacept and placebo groups on most outcomes, except on estimated medical expenditure in the anti-TNF failure (ATTAIN) sample.
|
Medical expenditure
A review of the statistical assumptions found no marked violations, allowing for straightforward interpretation of the results. MANCOVA models were run to adjust for any differences at baseline. For the MTX failure (AIM) data, baseline medical expenditure was controlled to begin at $614 per 30 days. The overall group x time interaction was significant (MANCOVA result of overall effect F [3,518] = 2.84, P = 0.04), with significantly more medical expenditure noted in placebo beginning at the first post-baseline assessment (day 29) and continuing to be present for each of the remaining time points with marked differences at days 169 and 365 (Fig. 2, left panel). Moreover, the effect of this interaction was nearly medium in size, with a Cohen's d = 0.40. Moreover, by day 365 the MTX failure abatacept group was incurring an estimated $394 in 30-day medical expenses (down by $220 dollars from baseline) whereas MTX failure placebo was incurring an estimated $462 in 30-day medical expenses (down by $152 dollars from baseline). Given the covariance process used, these means are based on the marginal means in the model and not on the actual calculated estimates.
|
From baseline, reduction in medical expenditure for abatacept was markedly faster than placebo. Indeed, by the next assessment (day 28) placebo had more medical expenditure than abatacept. Baseline medical expenditure was controlled to begin at $693 per 30 days. The overall group x time interaction was significant (F[1,355] = 38.27, P < 0.001), with significantly more medical expenditure noted in placebo beginning at the first post-baseline assessment (day 29) and continuing to be present for each of the remaining time points with marked differences at days 85 and 169 (Fig. 2, right panel). Moreover, the effect of this interaction was between medium and large, with a Cohen's d = 0.67. By day 169 the anti-TNF failure abatacept group was incurring an estimated $527 in 30-day medical expenses (a reduction $166 in monthly medication expenses from baseline or $1992 per year) whereas anti-TNF failure placebo was incurring an estimated $672 in 30-day medical expenses (down by just $24 dollars from baseline). Unlike the MTX failure (AIM) placebo group, which flattens in medical expenditure, the more severely diseased anti-TNF failure placebo group surged in medical expenditure after an initial decrease. In sum, abatacept saw a decrease of more than 6.9 times that of anti-TNF failure placebo on monthly medical expenditure.
Job loss and inability to work
A total of 841 participants in the MOS database had information on 6-month job loss (11% indicating that they had lost their job since initial interview), whereas 789 had 1-yr job loss data (15% had lost their job by then) and 770 had 2-yr job loss data (16% had lost their job by then). An inability to work at the initial MOS interview was noted in 25% of the 2195 interviewees with such data. Based on these data, logistic regression analyses were conducted and results of the pertinent unstandardized regression weights examined (available on request from the corresponding author). Lower PCS scores were related to a greater likelihood of having lost a job or an inability to work compared with lower MCS scores (i.e. negative coefficients were larger for PCS than for MCS).
The unstandardized coefficients from the logistic regression models were then used to estimate ORs for each person at each time point, once for the PCS weight and once for the MCS. These ORs are a means of measuring the relative risk of each of the four major life events as a function of one's PCS or MCS scores at a given time point. Upon calculation of the OR estimate for each of the four outcomes, an ANOVA was conducted to determine group (abatacept vs placebo) differences on the end-point ORs [once for PCS-based ORs and once for MCS-based ORs analysed with MTX failure (AIM) and then anti-TNF failure (ATTAIN) samples]. Sufficient ANOVA assumptions were met, and there was no need to control for baseline differences given the similarity of baseline values for each analysis.
Results of the ANOVAs are displayed in Table 3. In both the MTX failure (AIM) and anti-TNF failure (ATTAIN) samples, the PCS-based ORs for the major life events were significantly different at end-point, noting significantly less likelihood of occurrence in the abatacept groups than in the placebo groups. A similar confluence of significance was demonstrated in the less-severe MTX failure (AIM) data for the PCS-derived ORs: ORs for experiencing the respective major life events were significantly lower for abatacept patients than for placebo patients at end-point. Similar results were found in PCS-derived estimates from ATTAIN, although differences in MCS-based estimates were not statistically significant. For interpretation's sake, examine the MTX failure (AIM) MCS-derived results in Table 3 for 6-month job loss. At end-point, estimates indicate that the placebo group was 64% more likely (i.e. mean OR = 1.64) to have job loss than the general population whereas the MTX failure (AIM) abatacept group was only 39% (i.e. OR = 1.39) more likely than the general population, or a 25% (i.e. OR = 1.25) difference in likelihood of experiencing a job loss in 6 months compared with placebo.
|
|
Discussion |
|---|
|
TopAbstractIntroductionParticipants and methodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
QoL improvement is important to demonstrate in RA clinical trials, but can be abstract [28]. The current article provides a more palpable interpretation of what QoL changes mean in terms of medical expenditure and job loss (both current and predicted). Considered with previous results [18, 19] related to abatacept's impact on QoL, the current results provide a meaningful rationale for abatacept's use in RA patients. Because medical expenditure, job loss and inability to work data were not addressed directly, the best interpretation of the current results is that they provide another interpretation to existing knowledge. Evidence has already been published regarding the diverse benefits on the SF-36 that the abatacept provides over time (along with physical functioning and fatigue). Given that the expanded QoL estimates from the current results were derived from the previously examined SF-36 data, the current results provide a new way in which to interpret existing knowledge. We knew to expect significant differences, but now we can place dollar amounts and likelihood estimates on these expected changes. For example, in addition to providing significantly greater improvement on physical and mental composites, as derived from the SF-36, we can also now say that such benefits would translate into large reductions in medical expenditures and significantly greater reductions in the likelihood of future job loss and inability to work. Given the richness of previous databases, current results linking the SF-36 to medical expenditure and major life events are not only possible to embellish our current knowledge, but also provide specificity appropriate enough to examine particular information such as changes in dollars spent. For example, in a study with a sample of nearly 8000 RA patients collected just a few years before the current data, the average monthly medical expenses were estimated at $793 [11] and Yelin et al. [6] indicated that those suffering from rheumatic diseases have a mean monthly medical expenditure of $581.50 (in 2003). Prior to treatment, the anti-TNF failure sample had an average (marginal mean) monthly medical expenditure of $696 (S.D. = $263), demonstrating that the current estimates are not markedly different, falling between the currently reported estimates.
Although statistically significant differences between treatment and placebo groups were expected (given their derivation from the PCS and MCS, which have been shown elsewhere to be significantly more improved for abatacept than placebo), the degree of differences between the groups were of interest. These degrees are discussed for medical expenditure and job loss/inability to work.
Abatacept samples in the MTX failure and anti-TNF failure studies showed significant and large decreases on medical expenditure compared with placebo. In fact, the medical expenditure trends over time were sensitive to the differences between the MTX failure sample and the more severe anti-TNF failure samples. When examining Fig. 2 (MTX failure), one can see that medical expenditure for the placebo groups drops until about day 85, then becomes relatively stable through the rest of the year of assessment. However, Fig. 2 (anti-TNF failure) shows that the placebo group has only a slight reduction in medical costs through day 85, thereafter increasing in expenditure to where only a $24 monthly medical expenditure reduction was ultimately obtained for the anti-TNF failure placebo group.
Abatacept also provided large improvements on the major life events over placebo. Based on PCS, ORs were substantially lower in both samples for 6-month, 1-yr and 2-yr job loss predictions, as well as for inability to work. Based on MCS, the results held as for PCS, except that anti-TNF failure abatacept ORs were not quite significantly lower than placebo ORs. The lack of MCS-based results is not altogether surprising, and is arguably less directly related to the disease of RA than the PCS-based ORs. A litany of results has demonstrated larger improvement on PCS than on MCR for RA patients treated with anti-TNF medications [29]. Although abatacept treatment in anti-TNF failure patients has been shown to lead to significant and important QoL improvement on mental domains [19], those results also showed markedly larger results for PCS-based results than for MCS-based results. Moreover, as PCS-based results reflect one of the key symptom areas noted by the WHO [3] and ACR [30], PCS-based ORs are the most logical to use with RA.
There may be a few other reasons why the MCS-based estimates were not significant in the ATTAIN sample. First, a comparison between the MCS-derived ORs from the MTX failure and anti-TNF failure databases (Table 3) shows that a mean OR difference of 0.11 on anti-TNF failure (2-yr job loss) was not sufficient to achieve significance (P = 0.10) whereas a mean OR difference of 0.13 (1-yr job loss) and 0.14 (2-yr job loss) was sufficient in the MTX failure sample. Not only was it sufficient, but the P-values were markedly lower than 0.05. The marked difference in P-values for OR differences that were relatively similar suggests that power in the anti-TNF failure sample may be insufficient to detect the differences found to be significant in the MTX failure study. As the MTX failure sample has two-thirds more sample size, power may have played an important role in this lack of finding. Another consideration is that MCS-derived ORs lead to estimates that simply were not really different from one another at end-point between abatacept and placebo groups. Previous QoL results on the anti-TNF failure data [19] demonstrated significantly greater improvements on MCS for abatacept than for control. Nevertheless, the effect size for MCS was markedly less than that for PCS; MCS obtained a small–medium effect size (Cohen's d = 0.41) whereas PCS obtained a large effect size (Cohen's d = 0.87). Therefore, there is an overall trend that anti-TNF failure participants received somewhat less MCS-based benefit than they did PCS-based benefit. It may be that patients with a longer history of treatment failure simply need more time to demonstrate greater improvement on MCS, which would then translate into more improved MCS-derived ORs for major life events as conducted herein. Given that MTX failure data were collected over 1 yr and anti-TNF failure data were collected over just 6 months, the difference in time could be at least partially responsible for the lack of difference in the anti-TNF failure group.
A necessary limitation to this study is that the outcomes analysed were not measured directly, but rather were estimated based on links between these outcomes and the SF-36 from large non-clinical samples. As noted previously, our current estimates appear to match expectations based on similar research with direct measures of the outcome variables. Nevertheless, estimated scores lack the randomness and around the prediction equation that true data do [31] thereby making the overall pattern of responses less varied than real data. Moreover, the databases from which the weights were derived were not exclusively comprised of RA patients. Although this lack of homogeneity may impact the exact calculations, the magnitude of difference between the treatment and control groups should remain stable, as the heterogeneity impacts both treatment and control groups.
|
Conclusions |
|---|
|
TopAbstractIntroductionParticipants and methodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
In addition to the marked benefit abatacept provides to physical functioning, fatigue and SF-36 based scores, we can now also interpret the expected differences of monthly medical expenditure and job loss/inability to work between abatacept and placebo groups. The reduction in monthly medical expenditure was $152 more for abatacept than placebo in the MTX failure group and $122 more for abatacept than placebo in the anti-TNF failure group. Additionally, abatacept led to between 25% and 64% more reduction in the likelihood of current and predicted job loss compared with placebo. The strong effect sizes obtained for all significant analyses suggest that the results are clinically meaningful. Moreover, given the nature of the variables, results should also be meaningful for patients, patient families, providers, employers and health care insurance entities.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionParticipants and methodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
Funding: This project was funded by Bristol-Myers Squibb.
Disclosure statement: QualityMetric is a consultant for Bristol-Myers Squibb (J.C.C., P.L. and G.V.W. work for QualityMetric). T.L. is an employee of and owns company stock in Bristol-Myers Squibb. R.M. is an employee and shareholder of Bristol-Myers Squibb. G.V.W. is a consultant for Bristol-Myers Squibb.
|
References |
|---|
|
TopAbstractIntroductionParticipants and methodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
- Cole JC, Motivala SJ, Khanna D, et al. Validation of single-factor structure and scoring protocol for the Health Assessment Questionnaire-Disability Index. Arthritis Rheum (2005) 53:536–42.[CrossRef][Web of Science][Medline]
- Hazes JM, Geuskens GA, Burdorf A. Work limitations in the outcome assessment of rheumatoid arthritis. J Rheumatol (2005) 32:980–2.
[Free Full Text] - WHO. International classification of functioning, disability, and health: ICF. (2001) (1 December 2007, date last accessed). http://www.whoint/classifications/icf/en/.
- Verstappen SM, Bijlsma JW, Verkleij H, et al. Overview of work disability in rheumatoid arthritis patients as observed in cross-sectional and longitudinal surveys. Arthritis Rheum (2004) 51:488–97.[CrossRef][Web of Science][Medline]
- Burton W, Morrison A, Maclean R, Ruderman E. Systematic review of studies of productivity loss due to rheumatoid arthritis. Occup Med (2006) 56:18–27.
[Abstract/Free Full Text] - Yelin E, Murphy L, Cisternas MG, et al. Medical care expenditures and earnings losses among persons with arthritis and other rheumatic conditions in 2003, and comparisons with 1997. Arthritis Rheum (2007) 56:1397–407.[CrossRef][Medline]
- Yelin E. The costs of rheumatoid arthritis: absolute, incremental, and marginal estimates. J Rheumatol (1996) 44(Suppl):47–51.
- Spector TD. Rheumatoid arthritis. Rheum Dis Clin North Am (1990) 16:513–37.[Web of Science][Medline]
- Wolfe F. The burden of rheumatoid arthritis. Am J Manage Care (1999) S859.
- Cohen JW, Monheit AC, Beauregard KM, et al. The Medical Expenditure Panel Survey: a national health information resource. Inquiry (1996) 33:373–89.[CrossRef][Web of Science][Medline]
- Michaud K, Messer J, Choi HK, Wolfe F. Direct medical costs and their predictors in patients with rheumatoid arthritis: a three-year study of 7,527 patients. Arthritis Rheum (2003) 48:2750–62.[CrossRef][Web of Science][Medline]
- Bjorner JB, Wallenstein GV, Martin MC, et al. Interpreting score differences in the SF-36 vitality scale: using clinical conditions and functional outcomes to define the minimally important difference. Curr Med Res Opin (2007) 23:731–9.[CrossRef][Web of Science][Medline]
- Young A, Dixey J, Kulinskaya E, et al. Which patients stop working because of rheumatoid arthritis? Results of five years’ follow up in 732 patients from the Early RA Study (ERAS). Ann Rheum Dis (2002) 61:335–40.
[Abstract/Free Full Text] - Wolfe F, Hawley DJ. The longterm outcomes of rheumatoid arthritis: work disability: a prospective 18 year study of 823 patients. J Rheumatol (1998) 25:2108–17.[Web of Science][Medline]
- Emery P. The therapeutic potential of costimulatory blockade with CTLA4Ig in rheumatoid arthritis. Expert Opin Investig Drugs (2003) 12:673–81.[CrossRef][Web of Science][Medline]
- Kremer JM, Genant HK, Moreland LW, et al. Effects of abatacept in patients with methotrexate-resistant active rheumatoid arthritis: a randomized trial. Ann Intern Med (2006) 144:865–76.
[Abstract/Free Full Text] - Genovese MC, Becker JC, Schiff M, et al. Abatacept for rheumatoid arthritis refractory to tumor necrosis factor alpha inhibition. N Engl J Med (2005) 353:1114–23.
[Abstract/Free Full Text] - Russell AS, Wallenstein GV, Li T, et al. Abatacept improves both the physical and mental health of patients with rheumatoid arthritis who have inadequate response to methotrexate treatment. Ann Rheum Dis (2007) 66:189–94.
[Abstract/Free Full Text] - Westhovens R, Cole JC, Li T, et al. Improved health-related quality of life for rheumatoid arthritis patients treated with abatacept who have inadequate response to anti-TNF therapy in a double-blind, placebo-controlled, multicentre randomized clinical trial. Rheumatology (2006) 45:1238–46.
[Abstract/Free Full Text] - Ware JE Jr, Bjorner JB, Kosinski M. SF-36 health survey: manual and interpretation guide (2007) 2nd. Lincoln, RI: QualityMetric, Inc.
- Ware JE Jr, Kosinski M, Bjorner JB, et al. User's manual for the SF-36 health survey (Version 2) (2006) 2nd. Lincoln, RI: QualityMetric, Inc.
- McHorney CA, Ware JE Jr, Raczek AE. The MOS 36-item short-form health survey (SF-36): II. Psychometric and clinical tests of validity in measuring physical and mental health constructs. Med Care (1993) 31:247–63.[Web of Science][Medline]
- Ware JE Jr, Kosinski M. SF-36 physical & mental health summary scales: a manual for users of version 1 (2001) 2nd. Lincoln, RI: QualityMetric, Inc.
- Fleishman JA, Cohen JW, Manning WG, Kosinski M. Using the SF-12 health status measure to improve predictions of medical expenditures. Med Care (2006) 44:I54–63.[Medline]
- Hornbrook MC, Goodman MJ. Assessing relative health plan risk with the RAND-36 health survey. Inquiry (1995) 32:56–74.[Web of Science][Medline]
- Stewart AL, Greenfield S, Hays RD, et al. Functional status and well-being of patients with chronic conditions. Results from the Medical Outcomes Study. J Am Med Assoc (1989) 262:907–13.
[Abstract/Free Full Text] - Tabachnick BG, Fidell LS. Using multivariate statistics (2007) 5th. New York: Pearson Education.
- Ware JE Jr, Keller SD. Interpreting general health measures. Quality of life and pharmacoeonomics in clinical trials (1995) Philadelphia, PA: Lippincott-Raven Publishers.
- FDA, DHHS. Review of BLA Submission 99-0128 Infliximab 9REMICADE) for signs and symptoms of rheumatoid arthritis. (1999) (28 November 2007, date last accessed). http://www.fdagov/cder/biologics/review/inflcen111099r4.pdf.
- ACR Subcomittee on AR Guideline. Guidelines for the management of rheumatoid arthritis: 2002 update. Arthritis Rheum (2002) 46:328–46.[CrossRef][Web of Science][Medline]
- Iversen GR. Bayesian statistical inference (1984) Newbury Park, CA: Sage.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||