Table 2.

Meta-regression analysis for identifying potential sources of heterogeneity in the diagnostic performance of ¹⁸F-FDG PET/CT for the detection of peritoneal carcinomatosis

		Univariate analysis				Multivariate analysis
Variables	No of studies	Sensitivity	p-value	Specificity	p-value	LRT χ2	p-value
Study location
Western countries	11	0.86 (0.76−0.95)	0.40	0.91 (0.88−0.94)	0.10	0.99	0.61
others	3	0.91 (0.79−1.00)		0.93 (0.89−0.98)
Study year
Before 2010	3	0.88 (0.72−1.00)	0.69	0.91 (0.88−0.94)	0.03	0.78	0.68
After or during 2010	11	0.87 (0.78−0.96)		0.95 (0.89−1.00)
Study design
Prospective	1	0.93 (0.78−1.00)	0.86	0.85 (0.69−1.00)	0.06	1.42	0.49
Retrospective	13	0.86 (0.78−0.94)		0.92 (0.89−0.95)
Image analysis
PB	12	0.87 (0.79−0.96)	0.21	0.91 (0.88−0.94)	0.10	0.43	0.81
LB	2	0.84 (0.61−1.00)		0.93 (0.88−0.98)
Cohort homogeneity
Homogenous	6	0.84 (0.70−0.98)	0.21	0.89 (0.84−0.94)	0.00	2.31	0.31
Heterogeneous	8	0.89 (0.80−0.98)		0.93 (0.90−0.96)

¹⁸F-FDG, fludeoxyglucose; LRT, likelihood ratio test; PET, positron emission tomography; LB, lesion-based analysis; PB, patient-based analysis.