SCSmeta: The Split Component Synthesis function for meta-analysis of diagnostic test accuracy studies

Description

The SCSmeta function implements the robust inverse variance heterogeneity (IVhet) model for meta-analysis of diagnostic test accuracy studies^[1,2]. The SCS method starts off with the meta-analysis of the diagnostic odds ratios (DOR) and then splits the DOR into component measures (sensitivity, specificity, positive and negative likelihood ratio and AUC).

SCSmeta has the form SCSmeta=function(tp, fp, fn, tn). So you need to have a list of values for the number of true positive (i.e. tp), false positive (fp), false negative (fn) and true negative (tn) results for each of the studies that will be used in the meta-analysis.

An example

Here we present an example using data from 31 studies that assessed procalcitonin as a diagnostic marker for sepsis in critically ill patients^[3]. Let’s load the data. A way to do this is by using the following code:

library(readxl)
example <- read_excel("/the_path_of_the_data/PCT and sepsis.xlsx")

Let’s see how the data look:

Now we need to get the lists of tp, fp, fn and tn in the SCSmeta function and run it. You can simply do this by writing:

SCSmeta(example$tp,example$fp, example$fn, example$tn)

Following its execution, the SCSmeta function produces the following.

It first gives the point estimates and associated 95% confidence intervals for the diagnostic odds ratio (DOR), the pooled sensitivity (Se) and specifcity (Sp), the area under the curve (AUC) and the positive and negative likelihood ratio (pLR and nLR, respectively).

## [1] "The DOR is 7.566 ( 4.933 ;  11.607 )"
## [1] "The AUC is 0.733 ( 0.69 ;  0.773 )"
## [1] "The Se is 0.724 ( 0.653 ;  0.784 )"
## [1] "The Sp is 0.743 ( 0.674 ;  0.802 )"
## [1] "The pLR is 2.815 ( 2.073 ;  3.823 )"
## [1] "The nLR is 0.372 ( 0.276 ;  0.502 )"

Finally, it also produces the summary ROC plot:

The red line corresponds to the sROC plot, while the dotted lines represent upper and lower 95% confidence intervals for the sROC plot. The blue rhombus corresponds to the pooled point estimate for the \((Se_{pool}, 1-Sp_{pool})\) pair under the SCS method, while the circles represent the actual \((Se_i, 1-Sp_i)\) pairs from each study. Circle size is proportional to the weight of each study in the DOR meta-analysis using the IVhet estimator^[1].

Acknowledgements

This work was supported by Program Grant #NPRP10-0129-170274 from the Qatar National Research Fund (a member of Qatar Foundation). The findings herein reflect the work and are solely the responsibility of the authors.

Analyze your own data

You can use this Shiny app to upload your data and get results using the IVHet estimator. You need to provide a “.csv” file, organized in the format of the example above. The columns to be analyzed must be named tp,fp,fn,tn (it doesn’t matter if additional columns exist). If you wish to change the color of elements in your graph you can dowload the code and run it locally on your computer (if you have R-RStudio installed). The option to change color was removed from the online version because some users played for hours and this costs. For any questions or suggestions please do not hesitate to send an email.

References

[1] Doi SA, Barendregt JJ, Khan S, Thalib L, Williams GM. Advances in the meta-analysis of heterogeneous clinical trials I: The inverse variance heterogeneity model. Contemp Clin Trials 2015; 45(Pt A):130-8.

[2] Furuya-Kanamori L, Kostoulas P, Doi SA. A new method for synthesizing test accuracy data outperformed the bivariate method. J Clin Epidemiol. 2020.