Figures 2 and 3 show the distributions of, respectively, PCHOL and P/SCHOL values in each of the patient groups. All the transudates had PCHOL values below 55 mg/dl, a threshold misclassifying 8/67 (11.9 percent) of neoplastic exudates, 4/65 (6 percent) of tuberculous exudates, and 5/56 (8.9 percent) of the miscellaneous group as transudates. A P/SCHOL threshold of 0.3 misclassifies 8/65 (12.3 percent) of transudates as exudates and 9/67 (13.4 percent) of neoplastic exudates, 3/65 (4.6 percent) of tuberculous exudates, and 2/56 (3.5 percent) of the miscellaneous exudates as transudates. Table 3 shows that PCHOL afforded the fewest errors in distinguishing transudates from exudates, 6.7 percent, a misclassification rate that was significantly different from those of PLDH (p<0.001) and P/SLDH (p<0.01) but not from those of P/SCHOL, P/SPROT, and the three criteria of Light et al used in battery.
Table 4 lists the indices characterizing the capacity of the parameters studied in this work to diagnose exudates, together with those reported by Light et al. PCHOL had a sensitivity of 91 percent, a specificity of 100 percent, a PPV of 1 (K) percent, an NPV of 79 percent, and an efficiency of 93.2 percent. These results are better than those of any other parameters studied.

Figure 2. Pleural fluid cholesterol levels (PCHOL) (mg/dl) in the different groups studied.

Figure 3. Ratios of pleural fluid cholesterol to serum cholesterol levels (P/SCHOL) in the different groups studied.

Table 3 —Number of Misclassified Pleural Effusions in Each Group for Every Parameter Studied

Table 4— Sensibility, Specificity, Positive Predictive Value (PPV), Negative Predictive Value (NPV), and Efficiency for Every Parameter Studied