Kan Hong, Nagar Saurabh, Patel Jeetvan, Wallace Daniel J, Molta Charles, Chang David J
GlaxoSmithKline, Research Triangle Park, North Carolina.
GlaxoSmithKline, Research Triangle Park, North Carolina.
Clin Ther. 2016 Mar;38(3):610-24. doi: 10.1016/j.clinthera.2016.01.016. Epub 2016 Feb 19.
The treatment of systemic lupus erythematosus (SLE) is complex, with a wide range of drugs commonly prescribed. The aims of this study were to identify longitudinal treatment patterns in patients with incident SLE and to estimate the associations of treatment patterns with clinical and economic outcomes.
This retrospective, observational cohort study used a US managed care claims database to identify patients with newly diagnosed SLE and 4-year treatment follow-up. Patients were aged ≥ 18 years, with continuous medical and pharmacy benefits for 12 months before and 48 months after the index date (first medical claim with a diagnosis of SLE). Longitudinal treatment patterns were grouped using a k-means cluster analysis. Therapies were included in the cluster analysis if the mean number of prescriptions in each year was ≥ 0.05. Clinical and economic outcomes were compared across clusters using multivariate regression analyses.
Data from 1611 patients with incident SLE were analyzed (91.4% women; mean [SD] age, 44.5 [9.5] years; 56.2% managed primarily by a specialist). Hydroxychloroquine and corticosteroids were the most commonly prescribed therapies; methotrexate, azathioprine, and mycophenolate mofetil also met the criteria for inclusion in the cluster analysis. Ten treatment clusters were identified; the most common was minimally treated patients (42.8%). Hydroxychloroquine monotherapy, corticosteroid monotherapy, and corticosteroid/hydroxychloroquine combination therapy were received by 34.0%, 11.2%, and 7.8% of patients, respectively. Methotrexate or azathioprine with a corticosteroid/hydroxychloroquine were received by 4.2% of patients. Changes in therapy, except discontinuations, were rare. Compared with the minimally treated cluster, those that received corticosteroid monotherapy (mean dose, >12.0 mg/d) had poorer clinical and economic outcomes; the hydroxychloroquine-monotherapy cluster had similar or better outcomes; and patients who received a corticosteroid/hydroxychloroquine with or without methotrexate or azathioprine demonstrated outcomes that were poorer but that appeared better than those with corticosteroid monotherapy. SLE-related visits with a nonspecialist were common (~45%) and remained unchanged over time despite better clinical and economic outcomes associated with specialist visits.
This study utilized cluster analysis, an unsupervised machine-learning method, to systematically discern treatment patterns over 4 years and to estimate outcomes associated with the identified treatment patterns. The results suggest that minimal treatment is the most common approach in patients with newly diagnosed SLE. Clinical and economic outcomes are poorest with corticosteroid monotherapy but may improve with the addition of hydroxychloroquine and/or an immunosuppressive agent. A large proportion of SLE care is provided by nonspecialists despite the potential benefits of involving a specialist.
系统性红斑狼疮（SLE）的治疗较为复杂，通常会开具多种药物。本研究的目的是确定初发SLE患者的纵向治疗模式，并评估治疗模式与临床和经济结局之间的关联。
这项回顾性观察队列研究使用美国管理式医疗索赔数据库来识别新诊断的SLE患者，并进行4年的治疗随访。患者年龄≥18岁，在索引日期（首次诊断为SLE的医疗索赔）前12个月和后48个月享有连续的医疗和药房福利。使用k均值聚类分析对纵向治疗模式进行分组。如果每年的处方平均数≥0.05，则将疗法纳入聚类分析。使用多变量回归分析比较各聚类之间的临床和经济结局。
分析了1611例初发SLE患者的数据（91.4%为女性；平均[标准差]年龄为44.5[9.5]岁；56.2%主要由专科医生管理）。羟氯喹和皮质类固醇是最常用的疗法；甲氨蝶呤、硫唑嘌呤和霉酚酸酯也符合纳入聚类分析的标准。确定了10种治疗聚类；最常见的是治疗最少的患者（42.8%）。分别有34.0%、11.2%和7.8%的患者接受羟氯喹单药治疗、皮质类固醇单药治疗以及皮质类固醇/羟氯喹联合治疗。4.2%的患者接受甲氨蝶呤或硫唑嘌呤与皮质类固醇/羟氯喹联合治疗。除停药外疗法的改变很少见。与治疗最少的聚类相比，接受皮质类固醇单药治疗（平均剂量，>12.0mg/d）的患者临床和经济结局较差；羟氯喹单药治疗聚类的结局相似或更好；接受皮质类固醇/羟氯喹联合治疗且联合或不联合甲氨蝶呤或硫唑嘌呤的患者结局较差，但似乎比皮质类固醇单药治疗的患者要好。与非专科医生的SLE相关就诊很常见（约45%），尽管专科就诊与更好的临床和经济结局相关，但随着时间推移保持不变。
本研究利用聚类分析这一无监督机器学习方法，系统地识别4年期间的治疗模式，并评估与所确定治疗模式相关的结局。结果表明，最少治疗是新诊断SLE患者最常见的治疗方法。皮质类固醇单药治疗的临床和经济结局最差，但加用羟氯喹和/或免疫抑制剂可能会改善。尽管专科医生参与可能有益，但很大一部分SLE护理是由非专科医生提供的。
