Meltzer Andrew J, Graham Ashley, Connolly Peter H, Karwowski John K, Bush Harry L, Frazier Peter I, Schneider Darren B
Division of Vascular and Endovascular Surgery, Weill Cornell Medical College, New York-Presbyterian Hospital, New York, NY 10065, USA.
Ann Vasc Surg. 2013 Jan;27(1):53-61. doi: 10.1016/j.avsg.2012.05.002. Epub 2012 Sep 12.
We apply an innovative and novel analytic approach, based on reliability engineering (RE) principles frequently used to characterize the behavior of manufactured products, to examine outcomes after peripheral endovascular intervention. We hypothesized that this would allow for improved prediction of outcome after peripheral endovascular intervention, specifically with regard to identification of risk factors for early failure.
Patients undergoing infrainguinal endovascular intervention for chronic lower-extremity ischemia from 2005 to 2010 were identified in a prospectively maintained database. The primary outcome of failure was defined as patency loss detected by duplex ultrasonography, with or without clinical failure. Analysis included univariate and multivariate Cox regression models, as well as RE-based analysis including product life-cycle models and Weibull failure plots. Early failures were distinguished using the RE principle of "basic rating life," and multivariate models identified independent risk factors for early failure.
From 2005 to 2010, 434 primary endovascular peripheral interventions were performed for claudication (51.8%), rest pain (16.8%), or tissue loss (31.3%). Fifty-five percent of patients were aged ≥75 years; 57% were men. Failure was noted after 159 (36.6%) interventions during a mean follow-up of 18 months (range, 0-71 months). Using multivariate (Cox) regression analysis, rest pain and tissue loss were independent predictors of patency loss, with hazard ratios of 2.5 (95% confidence interval, 1.6-4.1; P < 0.001) and 3.2 (95% confidence interval, 2.0-5.2, P < 0.001), respectively. The distribution of failure times for both claudication and critical limb ischemia fit distinct Weibull plots, with different characteristics: interventions for claudication demonstrated an increasing failure rate (β = 1.22, θ = 13.46, mean time to failure = 12.603 months, index of fit = 0.99037, R(2) = 0.98084), whereas interventions for critical limb ischemia demonstrated a decreasing failure rate, suggesting the predominance of early failures (β = 0.7395, θ = 6.8, mean time to failure = 8.2, index of fit = 0.99391, R(2) = 0.98786). By 3.1 months, 10% of interventions failed. This point (90% reliability) was identified as the basic rating life. Using multivariate analysis of failure data, independent predictors of early failure (before 3.1 months) included tissue loss, long lesion length, chronic total occlusions, heart failure, and end-stage renal disease.
Application of a RE framework to the assessment of clinical outcomes after peripheral interventions is feasible, and potentially more informative than traditional techniques. Conceptualization of interventions as "products" permits application of product life-cycle models that allow for empiric definition of "early failure" may facilitate comparative effectiveness analysis and enable the development of individualized surveillance programs after endovascular interventions.
我们应用一种创新的分析方法，该方法基于常用于描述制成品行为的可靠性工程（RE）原理，来研究外周血管腔内介入治疗后的结果。我们假设这将有助于改善对外周血管腔内介入治疗后结果的预测，特别是在识别早期失败的危险因素方面。
在一个前瞻性维护的数据库中识别出2005年至2010年因慢性下肢缺血接受股动脉以下血管腔内介入治疗的患者。失败的主要结局定义为通过双功超声检测到的通畅性丧失，无论有无临床失败。分析包括单变量和多变量Cox回归模型，以及基于RE的分析，包括产品生命周期模型和威布尔失效图。使用“基本额定寿命”的RE原理区分早期失败，多变量模型识别早期失败的独立危险因素。
2005年至2010年，共进行了434例主要的外周血管腔内介入治疗，其中间歇性跛行患者占51.8%，静息痛患者占16.8%，组织缺损患者占31.3%。55%的患者年龄≥75岁；57%为男性。在平均18个月（范围0 - 71个月）的随访期间，159例（36.6%）介入治疗后出现失败。使用多变量（Cox）回归分析，静息痛和组织缺损是通畅性丧失的独立预测因素，风险比分别为2.5（95%置信区间，1.6 - 4.1；P < 0.001）和3.2（95%置信区间，2.0 - 5.2，P < 0.001）。间歇性跛行和严重肢体缺血的失败时间分布符合不同的威布尔图，具有不同特征：间歇性跛行的介入治疗显示失败率增加（β = 1.22，θ = 13.46，平均失败时间 = 12.603个月，拟合指数 = 0.99037，R(2) = 0.98084），而严重肢体缺血的介入治疗显示失败率降低，表明早期失败占主导（β = 0.7395，θ = 6.8，平均失败时间 = 8.2，拟合指数 = 0.99391，R(2) = 0.98786）。到3.1个月时，10%的介入治疗失败。这一点（90%可靠性）被确定为基本额定寿命。使用失败数据的多变量分析，早期失败（3.1个月之前）的独立预测因素包括组织缺损、病变长度长、慢性完全闭塞、心力衰竭和终末期肾病。
将RE框架应用于外周介入治疗后临床结果的评估是可行的，并且可能比传统技术提供更多信息。将介入治疗概念化为“产品”允许应用产品生命周期模型，该模型允许对“早期失败”进行经验性定义，可能有助于比较有效性分析，并能够制定血管腔内介入治疗后的个性化监测方案。
