More evidence on an abominable pairing: atrial fibrillation and kidney disease.
More evidence on an abominable pairing: atrial fibrillation and kidney disease.
复制标题
更多关于令人厌恶的配对的证据：心房颤动和肾脏疾病。
DOI：
10.1161/circulationaha.112.000640
复制
发表时间：
2013
期刊：
影响因子：
37.8
通讯作者：
Winkelmayer,WolfgangC
中科院分区：
文献类型：
--
作者：
Winkelmayer,WolfgangC
Winkelmayer房颤和ESRD发病率561条件，生物特征参数包括血压，血清血红蛋白和试纸蛋白尿测量，以及几种心血管药物的使用。超过206000例eGFR< 60 mL/min/1.73 m2但无终末期肾病且无既往诊断的房颤的患者被确定并纳入队列。未校正的分析显示，房颤患者的时间与终末期肾病的发生率增加18%相关，增幅虽小，但具有显著性。然而，对测量的潜在混杂因素（其中一些是时间依赖性的）进行调整后，得出的相关性估计值几乎是4倍，表明在考虑其他相关因素时，诊断房颤后花费的人-时间与终末期肾病的发生率高出67%相关。尽管这一发现在定性和定量上与Watanabe等人的研究一致8（在校正分析中，房颤与eGFR终点发生率增加80%和蛋白尿终点发生率增加116%相关），通过多变量校正消除的向零的强混杂是惊人的，不寻常的，如果在多变量校正后，暴露个体中与更高风险相关的未校正估计值增加，这通常意味着，在某些或所有观察到的潜力方面，协会的混淆因素。这意味着eGFR降低和房颤的患者（或更确切地说，花费的人-时间）必须比没有房颤的患者（或花费的人-时间）更健康。鉴于我们对房颤风险因素的了解，这似乎有点违反直觉，尽管我们了解到研究队列中年龄较小与房颤事件风险增加相关。(This可能是信息审查的指示，因为更有可能死亡的慢性肾脏疾病老年患者经历房颤的机会更少，死亡和可能从医疗保健系统迁出是竞争风险。）在一项具有时间不变暴露的队列研究中，人们通常可以从典型的表1中收集暴露和未暴露个体之间的任何差异。不幸的是，在随时间变化的暴露研究中无法获得此类信息，就像这里的情况一样。然而，从仅包括单个或更多限制性潜在混杂因素集的模型中提供额外的关联估计值，以了解哪些因素（共同）在未调整的比较中导致这种相对较大的关联低估，这将是有用的。eGFR降低的房颤患者终末期肾病发生率增加的可能解释是什么？正如作者所承认的，残余混杂肯定是一种可能性，但考虑到已观察到的特征的混杂方向，似乎不需要担心仅确定存在的合并症严重程度的缺失信息。然而，一个可能的候选因素是肾功能下降率，它已演变为许多心血管结局的独立风险因素，与eGFR水平无关。9，10可以想象
Winkelmayer Atrial Fibrillation and ESRD Incidence 561 conditions, biometric parameters including blood pressure, serum hemoglobin, and dipstick proteinuria measurements, and use of several cardiovascular medications. More than 206 000 individuals with eGFR< 60 mL/min/1.73 m2 but without end-stage renal disease and free of previously diagnosed atrial fibrillation were identified and included in the cohort. The unadjusted analysis showed that person-time spent with atrial fibrillation was associated with a small but significant 18% increment in the rate of incident end-stage renal disease. However, adjustment for the measured potential confounders, some of which were time-dependent, yielded an almost 4-times larger estimate of association, suggesting that person-time spent after diagnosis of atrial fibrillation was associated with a 67% higher rate of end-stage renal disease when taking into account other relevant factors. Although this finding is qualitatively and quantitatively consistent with the study by Watanabe et al8 (in which atrial fibrillation was associated with an increased incidence of the eGFR end point by 80% and the proteinuria end point by 116% in adjusted analyses), the strong confounding toward the null, removed by multivariable adjustment, is striking, unusual, and should require further consideration.If an unadjusted estimate of an association toward higher risk among exposed individuals increases after multivariable adjustment, it usually means that exposed individuals were systematically healthier compared with unexposed individuals on some or all of the observed potential confounders of the association. This means that patients (or rather, person-time spent) with reduced eGFR and atrial fibrillation must have been healthier than patients (or person-time spent) without atrial fibrillation. This seems a bit counterintuitive given what we know about risk factors for atrial fibrillation, although we learn that younger age was associated with an increased risk of incident atrial fibrillation in the study cohort.(This may perhaps be an indication of informative censoring in that older patients with chronic kidney disease who are more likely to die have less opportunity to experience atrial fibrillation, death and, perhaps, outmigration from the health care system being competing risks.) In a cohort study with time-invariant exposure one can usually glean any differences between exposed and unexposed individuals from a typical table 1. Unfortunately, such information is not available in studies with time-varying exposure, as is the case here. It would have been useful, however, to provide additional estimates of the association from models that included only single or more restricted sets of potential confounders to understand which factors (jointly) drove this relatively large underestimation of the association in unadjusted comparison. What are the possible explanations for this increased incidence of end-stage renal disease in patients with reduced eGFR who developed atrial fibrillation? Residual confounding is certainly a possibility, as acknowledged by the authors, but given the direction of the confounding by already observed characteristics, it does not seem that one needs to be concerned about missing information on severity of comorbidities whose presence was only ascertained. One possible candidate, however, is rate of decline of kidney function, which has evolved as an independent risk factor for a number of cardiovascular outcomes, independent of the level of eGFR. 9, 10 It is conceivable