Diagnostic accuracy of artery peak velocity variation measured by bedside real-time ultrasound for prediction of fluid responsiveness: a Meta-analysis / 中华危重病急救医学

ABSTRACT

Objective:To evaluate the diagnostic value of ultrasonic measurement of artery peak velocity variation (ΔVpeak) on predicting fluid responsiveness in critically ill patients.Methods:Databases of PubMed, Embase, Cochrane Library, SinoMed, Wanfang, CNKI and VIP were retrieved from the establishment of the database to November 2019. The retrieval literatures were about the research of ΔVpeak used to judge fluid responsiveness. According to the inclusion and exclusion criteria, the relevant literatures were screened by two researchers, and the data of the included literatures were extracted. The quality of literatures was evaluated by quality assessment of diagnostic accuracy studies (QUADAS). Meta Dics 1.4 software was used to analyze the literatures that met the quality standard by Meta-analysis. The pooled sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR) and diagnostic odds ratio ( DOR) were calculated. The summary receiver operating characteristic (SROC) curve was drawn and the area under SROC curve (AUC) was calculated. The χ 2 test and Spearman correlation coefficient were used to analyze heterogeneity, and Deek test was used to analyze publication bias. Results:A total of 1 854 patients were enrolled in 31 studies, including 11 domestic studies and 20 foreign studies. Using 14 items of QUADAS to evaluate the quality of literatures, it was found that the enrolled literatures were all in Grade A, indicating that the overall quality of literatures was high. The scattered distribution of SROC curve was not "shoulder arm shape", and Spearman correlation coefficient was 0.062 ( P = 0.710), so there was no threshold effect. The heterogeneity test showed that I2 = 57.2% ( P = 0.001), indicating that there was a certain degree of heterogeneity among the studies, and the source of heterogeneity was non threshold effect. Meta regression analysis showed that the reason for heterogeneity was the method of volume load test [ DOR = 3.87, 95% confidence interval (95% CI) was 1.56-9.57, P = 0.004 8]. According to the results of heterogeneity analysis, there was no significant heterogeneity ( I2 = 10.6, P = 0.288 5) among the studies after removing the samples of the passive leg raising (PLR) instead of the volume expansion. A meta-analysis was done with random effects model. The results showed that the pooled DOR was 23.85 (95% CI was 17.57 to 32.37), pooled sensitivity was 0.82 (95% CI was 0.80 to 0.85), pooled specificity was 0.83 (95% CI was 0.80 to 0.85), pooled PLR was 4.17 (95% CI was 3.58 to 4.86), and pooled NLR was 0.22 (95% CI was 0.18 to 0.28). The AUC was 0.901 2 (95% CI was 0.88 to 0.93), and Q index was 0.832 5. The results of Deek funnel plot showed that there was no published bias in all the studies ( P = 0.19). Conclusions:Ultrasonic measurement of ΔVpeak has a high value in predicting fluid responsiveness. It is a reliable parameter for the evaluation of shock, critical illness and surgical operation population who need to monitor the fluid responsiveness.