| Objective:To study the consistency of ethanol concentration measurement results by dual-column dual-FID headspace gas chromatography(dual-column dual detection method),to explore the application value of different consistency evaluation methods in the field of forensic toxicological analysis,to establish a consistency evaluation system for dual-column and dual-FID detection results,to explore how to select the calculated value for the final quantification of ethanol,and to provide theoretical basis and experimental data for the forensic identification of suspected "drunk driving" Theoretical basis and experimental data were provided for the accurate quantitative determination of ethanol content in forensic cases.Methods:1.blood samples from actual cases sent to our center by traffic police during2017-2018 were selected for the study,and AC in blood was detected using a two-column double detection method,and the results were divided into FID1 group and FID2 group according to different capillary columns and FID detectors,and the two-sample phase difference and RSD were calculated.paired t-test and Pearson correlation analysis were performed using SPSS 20.0,respectively.ICC intra-group correlation coefficients.pearson correlation was performed with 0.8～1.0 very strong correlation,0.6～0.8 strong correlation,0.4～0.6 moderate correlation,0.2～0.4 weak correlation,and 0.0～0.2 very weak correlation or no correlation.r<0.4 for ICC coefficient was poorly consistent,0.4～0.75 for average consistency,and r>0.75 for good consistency.Statistically significant differences were considered at P < 0.05.The ATE/LER area plot analysis was done by SPSS20.0,and the scatter plot was drawn by using the FID-1 measurement as x and FID-2 measurement as y of the same sample,and the ATE area was divided based on RSD≤10%.The LER region was divided on the basis of RSD≤15%.Judgment criteria: more than 95% of the points fall in the ATE region(white area);no points fall in the LER region(black area);less than5% of the points fall in the intermediate region(gray area).Bland-Altman plot analysis method: data behavior judgment,the difference will be done respectively normality test,proportional bias test,chi-square test.The data behavior was poor as non-standard data,and the concentration gradient was used to group the FID1 group and FID2 group into low concentration group(0-100mg/100 m L),medium concentration group(100-200mg/100 m L),and high concentration group(200mg/100 m L or more),and the difference analysis was performed to determine the source of poor data behavior.The data ratios of the experimental groups were then transformed and analyzed.Med Calc 15.6 was used to perform calculations and plot Bland-Altman plots,using the mean value of the same sample FID-1 and FID-2 assay results as the x-axis and the ratio of assay results as the y-axis,with x ± 1.96 s of the ratio as the 95% limits of agreement(Lo A)and ± 1.71 SE as the confidence interval CI.95% of the scatter points fell within the Lo A CI.The results of BAC were evaluated for consistency by combining the above methods.To explore the value of different evaluation methods in the field of forensic science.2.20 cases of ethanol samples with the concentration of 80mg/100 m L were prepared,and the AC of the experimentally prepared samples was detected by double-column double-check method,and the results were divided into FID1 group and FID2 group according to different capillary columns and FID detectors,and the consistency of the AC results measured by double-column double-check method was verified by ICC,ATE/LER plots and Bland-Altman plots analysis.The Bland-Altman plot was used ±8mg/100 m L as the professional limit,and the maximum value of the difference within Lo A was used for comparison,and the consistency was good within the professional limit3.The ethanol concentration results of the experimental sample of 80mg/100 m L were taken as the object of study,and two adjacent injections were taken as parallel samples of the same sample,and the results of FID1 were recorded as FID1-1 and FID2 as FID2-1 according to the first injection,and the results of FID1 were recorded as FID1-2 and FID2 as FID2-2 according to the second injection.the mean values of each group were calculated : Group a(FID1-1+FID1-2 mean),group b(FID2-1+FID2-2 mean),group c(FID1-1+FID2-1 mean),group d(FID1-2 and FID2-2 mean),and group e(overall mean)were analyzed by ICC,ATE/LER plots,and Bland-Altman for agreement among the groups.The median of each group was also compared with the limit of 80 mg/100 m L to explore the selection of the final quantitative results of the samples.Results:1.actual case blood samples FID-1 group AC mean concentration was145.90mg/100 m L.fid-2 group AC mean concentration result was 148.60mg/100 m L.relative difference maximum 5.73%,RSD maximum 4.05%.fid1 group and fid2 group two groups: paired t-test: t =-8.737(p<0.001);Pearson correlation coefficient:r = 0.997(P<0.001);intra-group correlation coefficient: ICC = 0.997(P<0.001);ATE/LER regional method showed that 100% of the points fell in the ATE region and no points in the LER region.Bland-Altman plots:(1)Data behavior:(1)regression line of difference vs.mean for experimental group: y=-0.4784-0.0151 x,P-value of slope b < 0.05 indicates proportional bias of difference;regression analysis of residuals vs.mean for difference pretest: y=-0.320+0.020 x,P-value of slope b < 0.05 indicates uneven variance of difference.K-Stest shows that the P-value > 0.05,the difference is in line with the normal distribution;(2)ratio analysis method experimental group data behavior: total concentration group ratio and mean regression analysis: y =0.987-0.00002779 x,the P value of slope b > 0.05 indicates that the ratio does not have proportional bias;residual absolute value and mean regression analysis: y =0.014+0.000024 x,the P-value of slope b >The K-S test showed that the P-value was>0.05 and the ratio was normally distributed.(2)Consistency analysis: 95% Lo A was(0.942～1.026)mg/100 m L,Lo A CI was(0.937～1.032),95.68%(155/162)of the points fell within Lo A,97.53%(158/162)of the points fell within Lo A CI.the maximum absolute value of the intra-Lo A difference was 10.21 g/ 100 m L,with an RSD of 3.27%.2.Analysis of the prepared 80 mg/100 m L experimental samples: ethanol concentration was(79.57 ± 1.65)mg/100 m L with an RSD of 2.07%;FID-2 group was(81.42 ± 1.33)mg/100 m L with an RSD of 1.63%.Intra-group correlation coefficient: ICC= 0.739(p<0.001);ATE/LER regional method showed that 100% of the points fell in the ATE region and none in the LER region.Bland-Altman analysis: data behavior judgment: regression line of difference vs.mean: y=-21.751+0.247 x,P-value of slope b > 0.05 indicates no proportional bias of difference;regression line of absolute value of difference residual vs.mean: y=-4.616+0.66 x,P-value of slope b > 0.05 indicates flush variance of difference.sw test shows P-value > 0.05 and the difference was normally distributed.lo A was(-3.974～0.260)mg/100 m L,Lo A CI was(-4.852～1.138),95%(19/20)of the points fell within the limits of agreement and the maximum absolute value of the difference within Lo A CI was 3.71 mg/100 m L,which was less than the professional limit ± 8mg/100 m L with an RSD of 3.28%.3.a,b,c,d,and e groups in the two-comparison consistency test,ICC coefficients were distributed between 0.781-0.950,with good correlation among the groups.a.In the ATE/LER region plot,all scatter points in the two-comparison between the groups were located in the ATE region,accounting for 100%.b.Bland-Altman plot,100% of the points in each group fell within the confidence interval of the consistency limits indicating good results Good consistency.The concentration distribution of each group was: group a mean ± standard deviation:(79.57 ± 1.57)mg/100 m L,median concentration 79.43 mg/100 m L,concentration range: 77.33-82.84 mg/100 m L;group b mean ± standard deviation(81.42 ± 1.17)mg/100 m L,median concentration 81.15 mg/100 m L,and concentration range 79.79-83.89mg/100 m L;group c mean±standard deviation:(80.75±1.47)mg/100 m L,median 80.75mg/100 m L,concentration range(78.57-83.61)mg/100 m L;group d mean±standard deviation:(80.23±1.34)mg/100 m L,concentration median79.99mg/100 m L,concentration range:(78.55-83.11)mg/100 m L;group e mean±standard deviation:(80.49±1.31)mg/100 m L,median concentration80.29mg/100 m L,concentration range: 78.56-83.36mg/100 m L.Compared with the limit value of 80mg/100 m L,30% of the points in group a were located above 80mg/100 m L,90% of the values in group b fell above80mg/100 m L,50% of the values in group c fell above 80mg/100 m L,50% of the values in group d fell above 80mg/100 m L,and 80% of the points in group e were above 80mg/100 m L.Conclusions:1.The results of FID1 and FID2 in the dual-column double-check method for the determination of ethanol concentration in blood samples and experimentally prepared samples were in good agreement.2.It is recommended to use Bland-Altman plot analysis,ATE/LER regional method and ICC as the consistency evaluation method for the detection of ethanol concentration in samples by dual-column duplex assay.3.In the two-column dual detection method,it is recommended to use the mean value of FID1 for the same sample continuously as the final quantitative result. |
PDF Full Text Request