Efficacy of biologic agents in improving the Health Assessme(3)

Efficacy of biologic agents in improving the Health Assessment Questionnaire (HAQ) score in established and early rheumatoid arthritis a meta-analysis with indirect comparisons

The PRISMA checklist was followed for reporting this meta-analysis. The protocol for this meta-analysis was not registered.

Data extraction and quality appraisalThe outcome measure chosen was im-provement in HAQ, which we defined as the difference in the mean change in HAQ (SD) for the biologic group compared to the mean change in HAQ formed on the following groups (estab-lished a priori): studies with established RA patients (i) failing DMARDs or (ii) failing anti-TNF at enrolment and (iii) patients with ERA. ERA trials were de-fined by the trial authors as symptoms <2 years. If a study had <10% prior exposure to a biologic agent, it was in-cluded in group (i). Re-analysis of the DMARD-failure subgroup excluding studies with patients exposed to biolog-1 golimumab and 2 infliximab), 4 on abatacept, 3 on rituximab and 4 on to-cilizumab. Nineteen trials reported on patients failing DMARDs (n=8115), 4 on anti-TNF failures (n=1694) and 5 on ERA (n=2492). There was no signifi-cant publication bias identified (Suppl. Fig. 2). Characteristics of the trials are presented in Table 1: mean age 47-56, disease duration 6 months to 13 years and baseline HAQ score of 1 to 1.9. (SD) HAQfor the control group ( HAQB-portant C). The minimally clinically im-was considered to be >0.22 (15). difference in HAQB- HAQC Absolute improvement in HAQ was chosen to allow for comparison with other meta-analyses, which more com-monly report change in HAQ and for ease of interpretation with respect to the minimally clinically important dif-ference, an important patient-centered outcome. If both 6 and 12 month data were available, we used 12 month data. In studies where median HAQ scores with interquartile range was reported, the median HAQ was set equal to the mean sumption of normal HAQ distribution HAQ score based on the as-in the study sample. Similarly, the SD of HAQ was calculated using: SD≈ IQR / 1.35. For studies, only report HAQ ing -on the table for distribution of the t-p-values, SD was imputed based statistic. When exact reported, we set the to the estimated reported value (pp-value to be equal -values were not pe.g. if = to 0.001).

was reported as <0.001, p was made Other extracted information included: study design (intervention administra-tion and doses, co-interventions, prior treatment, controls, length of follow-up, number of patients in each study arm, cross-over/escape and intention to treat) and baseline characteristics of patients (mean age, mean disease dura-tion, and baseline HAQ scores). Four reviewers collected data using a stand-ardised form. Quality assessment of studies was conducted using guidelines published by the Cochrane group (16), as well as, the Jadad score (17).Statistical analysis

Random-effects meta-analyses (DerSi-monian and Laird method) were per-

ics was conducted naïve and methotrexate-naïve patients post hoc. DMARD-were also biologic naïve. Some meth-otrexate-naïve studies included patients exposed to other DMARDs, such as sulfasalazine and hydroxychloroquine. For studies with multiple treatment arms of the same biologic (clinically used doses, co-interventions i.e. multiple or treatment arms were averaged together. comparators), the HAQ for the Heterogeneity was reported using the I2drugs compared to the control group . Indirect comparisons of the different were conducted using the Q-test based on analysis of variance and reported as a psignificant). -value (p-value <0.05 was considered sumed that the efficacy of each biologic Indirect comparisons as-was consistent across studies. Other subgroup analyses decided cluded: different clinically used doses, a priori in-co-intervention (DMARD control (placebo alone vs. none), placebo) and follow-up (6 months vvs. DMARD + 12 months). Meta-regression was per-s. formed using the method of moments including the following variables that were decided tion, baseline HAQ score, % cross-over a priori: disease dura-from control to intervention arms, year of publication and Jadad score.

Publication bias was assessed us-ing Funnel plots and the method. All statistical analyses were Trim and Fit performed using Comprehensive Meta-Analysis version 2 software.

RReview of the literature and included esultsstudiesSearch results and reasons for exclu-sion are summarised in Figure 1. A total reporting on anti-TNF agents (7 adali-of 28 studies were included: 17 mumab, 3 certolizumab, 4 etanercept, 335

Quality assessment using the Jadad score revealed 4 trials of poor quality (score <3). Three studies had a high risk of bias because incomplete data was not addressed. Exclusion of these trials did not change the results (data not shown). The majority of trials (n=19) had >20% cross-over from the control group to intervention groups and these studies used intention-to-treat analyses.

Efficacy of biologic agents at lowering HAQ in established RA patients failing DMARDs

Meta-analysis of trials reporting on patients failing DMARDs yielded a pooled difference in mean HAQ in biologics HAQcompared to control ( HAQ-B-(C95% I2=55%), with the upper limit of the ) of -0.22; 95%CI: -0.24, -0.20 the minimally clinically important dif-confidence interval not meeting ference (MCID) for change in HAQ (24) (Fig. 2). However, 4 trials had up to 10% of patients with a previous an-ti-TNF exposure. With these trials ex-cluded, the HAQ95%CI: -0.29, -0.22 (B- HAQI2C was -0.25; the MCID for change in HAQ. Because =60%), meeting of the significant heterogeneity, we analysed whether the type of biologic con--tributed to the heterogeneity.

Subgroup analysis of the different bio-logics revealed a significant difference in mean difference in change in HAQ (abatacept (-0.20; 95% CI:-0.28, -0.12; p<0.0001). The HAQB- HAQC for I2CI: -0.17, -0.05; =0%), and infliximab than the other anti-TNF agents with I2=0%) were lower (-0.11; 95%

HAQI2 HAQ=0% for all) (B- HAQC of -0.32 to -0.35; p<0.02) (Fig. 2). The (-0.20; 95% CI:-0.24, -0.17; B- HAQC for tocilizumab was lower compared to adalimumab I2=0%) and certolizumab (p<0.001) (Fig. 2).

百度搜索“77cn”或“免费范文网”即可找到本站免费阅读全部范文。收藏本站方便下次阅读，免费范文网，提供经典小说综合文库Efficacy of biologic agents in improving the Health Assessme(3)在线全文阅读。