Appendix ESensitivity Analyses for Meta-Analyses Involving Multiple Readers per Study

Publication Details

Some comparative accuracy studies reported data separately for different readers. Our primary analyses discussed in the main report only used data from reader 1 for each such study. This appendix contains the results of sensitivity analysis of this choice, for two meta-analyses:

  • MDCT versus MRI for diagnosis (a seven-study meta-analysis in which four of the seven studies reported multiple readers separately). Of the four studies, two reported three readers each, and two reported two readers each. Thus, we performed 35 sensitivity analyses ((2x3x2x3)-1 primary analysis).
  • MDCT versus MRI for assessment of metastases (a five-study meta-analysis in which one of the five studies reported three readers separately. Thus, we performed 2 sensitivity analyses (3-1 primary analysis).

Sensitivity Analysis of MDCT vs. MRI for Diagnosis

The primary analysis yielded estimates for MDCT of 89% for sensitivity and 90% for specificity, whereas the estimates for MRI were 89% for sensitivity and 89% for specificity. The table below lists the results of the 35 sensitivity analyses; all analysis provided estimates that were very similar to the primary analysis.

Table E-1. Sensitivity analysis of MDCT vs. MRI for diagnosis.

Table E-1

Sensitivity analysis of MDCT vs. MRI for diagnosis.

Sensitivity Analysis of MDCT vs. MRI for Assessment of Metastases

The primary analysis yielded estimates for MDCT of 48% for sensitivity and 90% for specificity, whereas the estimates for MRI were 50% for sensitivity and 95% for specificity. The table below lists the results of the two sensitivity analyses; both provided estimates that were very similar to the primary analysis.

Table E-2. Sensitivity analysis of MDCT vs. MRI for assessment of metastases.

Table E-2

Sensitivity analysis of MDCT vs. MRI for assessment of metastases.

Sensitivity Analysis of MDCT vs. MRI for Assessment of Metastases

The primary analysis yielded estimates for MDCT of 48% for sensitivity and 90% for specificity, whereas the estimates for MRI were 50% for sensitivity and 95% for specificity. The table below lists the results of the two sensitivity analyses; both provided estimates that were very similar to the primary analysis.

Table E-3. Sensitivity analysis of MDCT vs. MRI for assessment of metastases.

Table E-3

Sensitivity analysis of MDCT vs. MRI for assessment of metastases.