危重患者无创血压测量的准确性及对临床决策的影响

Accuracy of noninvasive blood pressure measurement in critically ill patients and and its impact on clinical decision-making

目的:研究危重患者无创血压测量的准确性及对临床决策的影响。方法:采用前瞻性观察性研究方法,纳入2021年3月至12月南京鼓楼医院ICU已建立有创动脉血压监测的133例危重患者,入室后1 h内同时测量无创血压和有创血压,记录无创与有创收缩压(systolic blood pressure,SBP)、舒张压(diastolic blood pressure,DBP)、平均动脉压(mean arterial pressure,MAP)。将患者分为应用血管活性药物的休克组和未应用血管活性药物的非休克组。应用Spearman、Bland-Altman相关性分析、组内相关系数(ICC)验证两种测量方法的一致性,采用误差网格法分析无创血压对临床决策的影响,采用多因素Logistic回归分析影响无创血压测量准确性的危险因素。结果:①Spearman相关性结果显示,非休克组患者,有创与无创血压呈显著正相关(SBP有创-SBP无创、DBP有创-DBP无创、MAP有创-MAP无创r分别为0.69、0.72、0.70,P均<0.001);休克组患者,SBP有创-SBP无创呈一定正相关(r=0.34,P<0.05),但DBP有创-DBP无创、MAP有创-MAP无创无明显相关性(P均>0.05)。②简单线性回归分析显示,非休克组患者,有创与无创血压呈正相关(SBP有创-SBP无创、DBP有创-DBP无创、MAP有创-MAP无创相关系数R 2分别为0.56、0.58、0.43,P均<0.05);休克组患者,有创与无创SBP、MAP成一定正相关(SBP有创-SBP无创、MAP有创-MAP无创相关系数R 2分别为0.20、0.10,P均<0.05),但DBP有创-DBP无创线性回归方程无统计学意义(P>0.05)。③Bland-Altman结果显示,非休克组与休克组患者,有创与无创血压在平均差值区间±10 mmHg的数据点占比均不满足美国医疗仪器促进协会(AAMI)无创血压测量标准,一致性差(SBP有创-SBP无创、DBP有创-DBP无创、MAP有创-MAP无创在平均差值区间±10 mmHg的数据点占比:非休克组患者分别为63%、73%、76%,休克组患者分别为35%、50%、52%,P均<0.05)。④ICC结果显示,非休克组患者,DBP有创-DBP无创一致性较好(绝对一致性系数0.76>0.75,P<0.001);休克组患者,SBP有创-SBP无创、MAP有创-MAP无创一致性均较差(绝对一致性系数0.33、0.31<0.4,P均<0.05),DBP有创-DBP无创无显著一致性(P>0.05)。⑤误差网格分析结果显示,休克组患者,无创与有创SBP配对测量数据点在A~E区占比分别为88.7%、6.5%、4.8%、0%和0%,无创与有创MAP配对测量数据点的占比分别为61.3%、38.7%、0%、0%和0%;非休克组患者,无创与有创SBP配对测量数据点在A~E区占比分别为95.8%、4.2%、0%、0%和0%,无创与有创MAP配对测量数据点的占比分别为97.2%、1.4%、1.4%、0%和0%。⑥多因素Logistic回归结果显示,休克是导致无创与有创SBP、DBP、MAP差值大于10 mmHg的危险因素(P<0.05),年龄是导致无创与有创SBP、MAP差值大于10 mmHg的危险因素(P<0.05),心率是导致无创与有创SBP差值大于10 mmHg的危险因素(P<0.05),是无创与有创DBP、MAP差值大于10 mmHg的保护因素(P<0.05)。结论:未休克的危重患者,无创血压测量准确性较好,以无创MAP指导临床决策较可靠,伴有休克的危重患者应用无创血压代替有创血压测量需谨慎。

Objective:To investigate the accuracy of noninvasive blood pressure measurement and the impact on clinical decision-making in critically ill patients.Methods:Using a prospective observational study,a total of 133 critically ill patients with established invasive blood pressure monitoring in the ICU of Nanjing Drum tower Hospital from March to December 2021 were included.Non-invasive and invasive blood pressure were measured and recorded simultaneously within 1 h after admission.The patients were divided into shock group with vasoactive drugs and non-shock group without vasoactive drugs.Spearman,Bland-Altman correlation analysis and interclass correlation coefficient(ICC)were applied to verify the consistency of the two monitoring methods,and the error grid method was used to analyze the effect of non-invasive blood pressure on guiding clinical decision-making.Results:①Spearman correlation results showed that in non-shock patients,there were significant positive correlation between invasive and the non-invasive blood pressure(r of SBP,DBP,MAP were 0.69,0.72,0.70,all P<0.001).In shock patients,there was a significant positive correlation between SBP invasive-SBP noninvasive(r=0.34,P<0.05),while DBP invasive-DBP noninvasive,MAP invasive-MAP noninvasive without significant correlation(both P>0.05).②Simple linear regression analysis revealed that in non-shock patients,there was a significant positive correlation between invasive and non-invasive blood pressure(R^(2) of SBP,DBP,MAP were 0.56,0.58,0.43,all P<0.05).In shock patients,invasive and non-invasive SBP and MAP were significantly positively correlated(R^(2) of SBP,MAP were 0.20,0.10,P<0.05).DBP invasive and DBP noninvasive linear regression equation was not statistically significant(P>0.05).③The Bland-Altman analysis results showed that patients in the non-shock and shock groups,the proportion of data points between invasive and noninvasive blood pressure in the mean difference interval±10 mmHg did not meet the non-invasive blood pressure measurement criteria of AAMI.Poor consistency(SBP invasive-SBP noninvasive,DBP invasive-DBP noninvasive,MAP invasive-MAP noninvasive data points in the average difference range of 10 mmHg:63%,73% and 76% of patients in non-shock group,and 35%,50% and 52% in the shock group,respectively,all P<0.05).④The ICCs showed that in non-shock group patients,there was a significant positive correlation between SBP invasive and SBP noninvasive(the absolute consistency was 0.76>0.75,P<0.001).In shock group patients,there were bad consistency coefficients between SBP invasive and SBP noninvasive,MAP invasive and MAP noninvasive(the absolute consistencies were 0.33,0.31<0.4,both P<0.001),and the non-correlation between DBP invasive and DBP noninvasive(P>0.05).⑤Error grid analysis results showed that in the shock group patients,the proportions of SBP measurements in the A to E hazard areas were 88.7%,6.5%,4.8%,0%,and 0%.The proportions of the MAP measurements were 61.3%,38.7%,0%,0%,and 0%.In the non-shock group patients,the proportions of SBP measurements in the A to E hazard zone were 95.8%,4.2%,0%,0%,and 0%.The proportions of MAP measurements were 97.2%,1.4%,1.4%,0%,and 0%.⑥The multivariate Logistic regression showed that shock was a risk factor for the difference between non-invasive and invasive SBP,DBP and MAP greater than 10 mmHg(P<0.05),age was a risk factor for the difference between non-invasive and invasive SBP and MAP greater than 10 mmHg(P<0.05);heart rate was a risk factor for the difference between non-invasive and invasive SBP greater than 10 mmHg(P<0.05),and was a protective factor for the difference between noninvasive and invasive DBP and MAP greater than 10 mmHg(P<0.05).Conclusion:For critically ill patients with non-shock,the accuracy of non-invasive blood pressure measurement is better,and is more reliable to guide clinical decision-making with non-invasive MAP.It should be cautious when using non-invasive blood pressure instead of invasive blood pressure measurement in critically ill patients with shock.