Methodology in reporting diagnostic accuracy for Borrelia specific IgG and IgM antibodies. Tests often differ but performance of the test antigens may be quite similar anyway.

Ram  Dessau

Methodology in reporting diagnostic accuracy for Borrelia specific IgG and IgM antibodies. Tests often differ but performance of the test antigens may be quite similar anyway.

Unknown

Research output: Contribution to conference without publisher/journal › Poster › Research

Abstract

Methodology in reporting diagnostic accuracy for Borrelia specific IgG and IgM antibodies.
Tests often differ but performance of the test antigens may be quite similar anyway.
Dessau RB. Dept. of Clinical Microbiology, Slagelse Hospital Region Sjælland.

This a study of Diagnostic accuracy of Oxoid IDEIA and Diasorin Liaison IgG and IgM serumantibodies. A straight forward test comparison has been performed using 48 serumsamples from patients with Neuroborreliosis and 216 Danish blood donors as healthy controls. Using the instruction provided by the manufacturer the sensitivity of IDEIA/Liaison IgG was 44%/94% (p<0.001), IgM 99%/94% (p=0.02) and for IgG or IgM positive 81%/96%(p=0.02).
Conclusion 1: IDEIA was less sensitive in IgG and when IgG or IgM was combined.
The specificity of IDEIA/Liaison IgG was 99%/94% (p<0.001), IgM 99%/96% (p=0.12) and for IgG or IgM positive 98%/89% (p<0.001).
Conclusion 2: IDEIA was more specific in both IgG and IgM.
P-values were calculated by the mcNemar test.
However conclusions 1 and 2 are inadequate and possibly a misleading interpretation of assay performance. When just counting positive and negatives samples mainly differences in cutoff policy are measured, not the performance of test antigens.
The ROC curve shows the discriminatory power, allows for comparison of the differences and assesment of proposed cutoffs (see figure). Visual inspection of the figure is an important first step. It is quite evident that Liaison IgM is both less sensitive and specific compared to the other markers. When adding IgG and IgM together (a logistic regression model was used) the Liaison IgG appears quite similar to the combination of IDEIA IgG+IgM. The IDEIA IgG and IgM combined was sligthly more specific and sensitive than Liaison IgG (p<0.01, Mann-Whitney Rank sum tests). Liaison IgM does not contribute much diagnostic value as IgM positive samples are also IgG positive. It could be considered to adjust the cutoffs to increase sensitivity of the IDEIA assay and the specificity of the Liaison assay. The proposed intervals for the indeterminate results appear clinically irrelevant.

Conclusion: Differences in sensitivity and specificity were mainly related to choice of cutoff not the discriminatory ability of the assays. Reports on diagnostic accuracy for measurement of Borrelia specific antibody reactivity should include analysis of the quantitative data. Combination of IgG and IgM and the choice of cutoff should be discussed.

Translated title of the contribution	Metoder til evaluering af Borrelia serologi.
Original language	English
Publication date	1. Apr 2012
Number of pages	1
Publication status	Published - 1. Apr 2012
Externally published	Yes

Cite this

@conference{3cea774babae454491d5aa63947efe00,

title = "Methodology in reporting diagnostic accuracy for Borrelia specific IgG and IgM antibodies. Tests often differ but performance of the test antigens may be quite similar anyway.",

abstract = "Methodology in reporting diagnostic accuracy for Borrelia specific IgG and IgM antibodies. Tests often differ but performance of the test antigens may be quite similar anyway. Dessau RB. Dept. of Clinical Microbiology, Slagelse Hospital Region Sj{\ae}lland. This a study of Diagnostic accuracy of Oxoid IDEIA and Diasorin Liaison IgG and IgM serumantibodies. A straight forward test comparison has been performed using 48 serumsamples from patients with Neuroborreliosis and 216 Danish blood donors as healthy controls. Using the instruction provided by the manufacturer the sensitivity of IDEIA/Liaison IgG was 44%/94% (p<0.001), IgM 99%/94% (p=0.02) and for IgG or IgM positive 81%/96%(p=0.02). Conclusion 1: IDEIA was less sensitive in IgG and when IgG or IgM was combined. The specificity of IDEIA/Liaison IgG was 99%/94% (p<0.001), IgM 99%/96% (p=0.12) and for IgG or IgM positive 98%/89% (p<0.001). Conclusion 2: IDEIA was more specific in both IgG and IgM. P-values were calculated by the mcNemar test. However conclusions 1 and 2 are inadequate and possibly a misleading interpretation of assay performance. When just counting positive and negatives samples mainly differences in cutoff policy are measured, not the performance of test antigens. The ROC curve shows the discriminatory power, allows for comparison of the differences and assesment of proposed cutoffs (see figure). Visual inspection of the figure is an important first step. It is quite evident that Liaison IgM is both less sensitive and specific compared to the other markers. When adding IgG and IgM together (a logistic regression model was used) the Liaison IgG appears quite similar to the combination of IDEIA IgG+IgM. The IDEIA IgG and IgM combined was sligthly more specific and sensitive than Liaison IgG (p<0.01, Mann-Whitney Rank sum tests). Liaison IgM does not contribute much diagnostic value as IgM positive samples are also IgG positive. It could be considered to adjust the cutoffs to increase sensitivity of the IDEIA assay and the specificity of the Liaison assay. The proposed intervals for the indeterminate results appear clinically irrelevant. Conclusion: Differences in sensitivity and specificity were mainly related to choice of cutoff not the discriminatory ability of the assays. Reports on diagnostic accuracy for measurement of Borrelia specific antibody reactivity should include analysis of the quantitative data. Combination of IgG and IgM and the choice of cutoff should be discussed. ",

author = "Ram Dessau",

note = "Kongresopraget, men det er vel ikke streng peer review",

year = "2012",

month = apr,

day = "1",

language = "English",

}

TY - CONF

T1 - Methodology in reporting diagnostic accuracy for Borrelia specific IgG and IgM antibodies. Tests often differ but performance of the test antigens may be quite similar anyway.

AU - Dessau, Ram

N1 - Kongresopraget, men det er vel ikke streng peer review

PY - 2012/4/1

Y1 - 2012/4/1

N2 - Methodology in reporting diagnostic accuracy for Borrelia specific IgG and IgM antibodies. Tests often differ but performance of the test antigens may be quite similar anyway. Dessau RB. Dept. of Clinical Microbiology, Slagelse Hospital Region Sjælland. This a study of Diagnostic accuracy of Oxoid IDEIA and Diasorin Liaison IgG and IgM serumantibodies. A straight forward test comparison has been performed using 48 serumsamples from patients with Neuroborreliosis and 216 Danish blood donors as healthy controls. Using the instruction provided by the manufacturer the sensitivity of IDEIA/Liaison IgG was 44%/94% (p<0.001), IgM 99%/94% (p=0.02) and for IgG or IgM positive 81%/96%(p=0.02). Conclusion 1: IDEIA was less sensitive in IgG and when IgG or IgM was combined. The specificity of IDEIA/Liaison IgG was 99%/94% (p<0.001), IgM 99%/96% (p=0.12) and for IgG or IgM positive 98%/89% (p<0.001). Conclusion 2: IDEIA was more specific in both IgG and IgM. P-values were calculated by the mcNemar test. However conclusions 1 and 2 are inadequate and possibly a misleading interpretation of assay performance. When just counting positive and negatives samples mainly differences in cutoff policy are measured, not the performance of test antigens. The ROC curve shows the discriminatory power, allows for comparison of the differences and assesment of proposed cutoffs (see figure). Visual inspection of the figure is an important first step. It is quite evident that Liaison IgM is both less sensitive and specific compared to the other markers. When adding IgG and IgM together (a logistic regression model was used) the Liaison IgG appears quite similar to the combination of IDEIA IgG+IgM. The IDEIA IgG and IgM combined was sligthly more specific and sensitive than Liaison IgG (p<0.01, Mann-Whitney Rank sum tests). Liaison IgM does not contribute much diagnostic value as IgM positive samples are also IgG positive. It could be considered to adjust the cutoffs to increase sensitivity of the IDEIA assay and the specificity of the Liaison assay. The proposed intervals for the indeterminate results appear clinically irrelevant. Conclusion: Differences in sensitivity and specificity were mainly related to choice of cutoff not the discriminatory ability of the assays. Reports on diagnostic accuracy for measurement of Borrelia specific antibody reactivity should include analysis of the quantitative data. Combination of IgG and IgM and the choice of cutoff should be discussed.

AB - Methodology in reporting diagnostic accuracy for Borrelia specific IgG and IgM antibodies. Tests often differ but performance of the test antigens may be quite similar anyway. Dessau RB. Dept. of Clinical Microbiology, Slagelse Hospital Region Sjælland. This a study of Diagnostic accuracy of Oxoid IDEIA and Diasorin Liaison IgG and IgM serumantibodies. A straight forward test comparison has been performed using 48 serumsamples from patients with Neuroborreliosis and 216 Danish blood donors as healthy controls. Using the instruction provided by the manufacturer the sensitivity of IDEIA/Liaison IgG was 44%/94% (p<0.001), IgM 99%/94% (p=0.02) and for IgG or IgM positive 81%/96%(p=0.02). Conclusion 1: IDEIA was less sensitive in IgG and when IgG or IgM was combined. The specificity of IDEIA/Liaison IgG was 99%/94% (p<0.001), IgM 99%/96% (p=0.12) and for IgG or IgM positive 98%/89% (p<0.001). Conclusion 2: IDEIA was more specific in both IgG and IgM. P-values were calculated by the mcNemar test. However conclusions 1 and 2 are inadequate and possibly a misleading interpretation of assay performance. When just counting positive and negatives samples mainly differences in cutoff policy are measured, not the performance of test antigens. The ROC curve shows the discriminatory power, allows for comparison of the differences and assesment of proposed cutoffs (see figure). Visual inspection of the figure is an important first step. It is quite evident that Liaison IgM is both less sensitive and specific compared to the other markers. When adding IgG and IgM together (a logistic regression model was used) the Liaison IgG appears quite similar to the combination of IDEIA IgG+IgM. The IDEIA IgG and IgM combined was sligthly more specific and sensitive than Liaison IgG (p<0.01, Mann-Whitney Rank sum tests). Liaison IgM does not contribute much diagnostic value as IgM positive samples are also IgG positive. It could be considered to adjust the cutoffs to increase sensitivity of the IDEIA assay and the specificity of the Liaison assay. The proposed intervals for the indeterminate results appear clinically irrelevant. Conclusion: Differences in sensitivity and specificity were mainly related to choice of cutoff not the discriminatory ability of the assays. Reports on diagnostic accuracy for measurement of Borrelia specific antibody reactivity should include analysis of the quantitative data. Combination of IgG and IgM and the choice of cutoff should be discussed.

M3 - Poster

ER -