Department of Clinical Physiology, Viborg Regional Hospital, Viborg, Denmark; Department of Health Science and Technology, Aalborg University, Aalborg, Denmark.
Department of Clinical Physiology, Viborg Regional Hospital, Viborg, Denmark.
Eur J Vasc Endovasc Surg. 2014 Mar;47(3):311-8. doi: 10.1016/j.ejvs.2013.11.016. Epub 2014 Jan 17.
To assess the intra- and inter-observer variation in laser Doppler flowmetry curve reading for measurement of toe and ankle pressures.
A prospective single blinded diagnostic accuracy study was conducted on 200 patients with known or suspected peripheral arterial disease (PAD), with a total of 760 curve sets produced. The first curve reading for this study was performed by laboratory technologists blinded to clinical clues and previous readings at least 3 months after the primary data sampling. The pressure curves were later reassessed following another period of at least 3 months. Observer agreement in diagnostic classification according to TASC-II criteria was quantified using Cohen's kappa. Reliability was quantified using intra-class correlation coefficients, coefficients of variance, and Bland-Altman analysis.
The overall agreement in diagnostic classification (PAD/not PAD) was 173/200 (87%) for intra-observer (κ = .858) and 175/200 (88%) for inter-observer data (κ = .787). Reliability analysis confirmed excellent correlation for both intra- and inter-observer data (ICC all ≥.931). The coefficients of variance ranged from 2.27% to 6.44% for intra-observer and 2.39% to 8.42% for inter-observer data. Subgroup analysis showed lower observer-variation for reading of toe pressures in patients with diabetes and/or chronic kidney disease than patients not diagnosed with these conditions. Bland-Altman plots showed higher variation in toe pressure readings than ankle pressure readings.
This study shows substantial intra- and inter-observer agreement in diagnostic classification and reading of absolute pressures when using laboratory technologists as observers. The study emphasises that observer variation for curve reading is an important factor concerning the overall reproducibility of the method. Our data suggest diabetes and chronic kidney disease have an influence on toe pressure reproducibility.
评估激光多普勒流量测量曲线读数在测量足趾和踝关节压力中的内部和观察者间变异性。
对 200 例已知或疑似外周动脉疾病（PAD）的患者进行了前瞻性单盲诊断准确性研究，共产生了 760 个曲线集。本研究的第一次曲线读数由实验室技师进行，他们对临床线索和首次数据采样至少 3 个月后的先前读数均不知情。在至少 3 个月后再次重新评估压力曲线。使用 Cohen's kappa 定量评估根据 TASC-II 标准的诊断分类观察者间一致性。使用组内相关系数、变异系数和 Bland-Altman 分析定量评估可靠性。
在诊断分类（PAD/非 PAD）方面，观察者内的总体一致性为 173/200（87%）（κ=.858），观察者间数据的一致性为 175/200（88%）（κ=.787）。可靠性分析证实了观察者内和观察者间数据的相关性均非常好（ICC 均≥.931）。变异系数范围为观察者内的 2.27%至 6.44%，观察者间的 2.39%至 8.42%。亚组分析显示，患有糖尿病和/或慢性肾脏病的患者的足趾压力读数的观察者变异较小，而未被诊断出患有这些疾病的患者的观察者变异较大。Bland-Altman 图显示，足趾压力读数的变异性高于踝部压力读数。
本研究表明，当使用实验室技师作为观察者时，在诊断分类和绝对压力读数方面具有很大的观察者内和观察者间一致性。该研究强调，曲线读数的观察者变异性是该方法整体可重复性的一个重要因素。我们的数据表明，糖尿病和慢性肾脏病会影响足趾压力的可重复性。
Zotero 插件