Figure 4 is a multi-panel graph, containing three panels of square plots. From left to right, the overall test performance or test accuracy of CHr, %HYPO, and sTfR were plotted in Panel A, B, and C of Figure 4, respectively. Each panel is a square plot that graphs a calculated value that is 100 percent minus the specificity of the study on the x-axis against the reported sensitivity on the y-axis. The positive and negative likelihood ratios of all included studies are calculated and plotted on the graph. Studies with a positive likelihood ratio of 10 or more, or a negative likelihood ratio of 0.1 or less, are considered as representing informative test results; the vertical and horizontal shaded areas on the plot represent areas of the graph where studies with these values would be located (respectively). We visual compared the locations of the ROC curves across panels to aid the indirect comparisons across studies. It appears that that curves for CHr (Panel A) and %HYPO (Panel B) are closer to the upper left hand corner (denoting perfect ability to predict response) than the curves for ferritin and TSAT (in Panels E and F of Figure 5, respectively), indicating better test performance. However, it appears that that there was no difference in the test performance between sTfR (Panel C) and ferritin or TSAT (Panels E and F of Figure 6, respectively).

Figure 4Indirect comparisons of the overall test accuracy of newer with classical markers of iron status (at baseline) to predict response to IV iron among adult HD CKD patients—CHr, %HYPO, sTfR

%HYPO=percent hypochromic red blood cells; CHr=content of hemoglobin in reticulocytes; CRI=corrected reticulocyte index; EPO= erythropoietin stimulating agents; Hb=hemoglobin; HD=hemodialysis; IV=intravenous; sTfR= soluble transferrin receptor; TSAT=transferrin saturation; ZPP=erythrocyte zinc protoporphyrin

Note: Each symbol represents one reference standard, and sensitivity/specificity pairs from the same study (using different cutoffs) are connected with lines. Each study was labeled by its first author’s last name (next to the corresponding symbol). Studies that fall in the shaded area to the left of the near vertical line have a positive likelihood ratio ≥ 10, and studies that fall in the shaded area above the near horizontal line have a negative likelihood ratio ≤ 0.1. Studies that reported LR+ ≥10 and LR- ≤0.1 were deemed to have adequate predictive ability of the marker’s test result for the response to IV iron.

From: Results

Cover of Biomarkers for Assessing and Managing Iron Deficiency Anemia in Late-Stage Chronic Kidney Disease
Biomarkers for Assessing and Managing Iron Deficiency Anemia in Late-Stage Chronic Kidney Disease [Internet].
Comparative Effectiveness Reviews, No. 83.
Chung M, Moorthy D, Hadar N, et al.

PubMed Health. A service of the National Library of Medicine, National Institutes of Health.