Skip to main content
. Author manuscript; available in PMC: 2020 Oct 15.
Published in final edited form as: Stat Med. 2019 Jul 17;38(23):4503–4518. doi: 10.1002/sim.8310

Table 1:

Application of the Times Method, Marker Method, Lin (2000) Cox model, and Case-Only Sign Test to the Pseudo-AMP Trial Data Seta.

Approach (95% CI) No. Cases in Analysis (n = 64) Target Parameter Estimate of the Parameter (95% CI) Two-sided P-value
Times 58 (91%) 12μ^Tctrl 27.98 (27.96, 28.01)
Method μ^Tcase 32.17 (28.69, 35.65)
μ^T 4.19 (0.70, 7.67) 0.018
Repeat 30 (47%) 12μ^Tctrl 32.27 (32.24, 32.30)
Excluding μ^Tcase 39.83 (35.29, 44.37)
+++ FP μ^T 7.56 (3.03, 12.10) 0.001
Marker 60 (94%) μ^Sctrl 3.40 (3.33, 3.47)
Method μ^Scase 3.17 (2.96, 3.39)
μ^S 0.23 (0.003, 0.45) 0.047
Repeat 31 (48%) μ^Sctrl 3.16 (3.10, 3.23)
Excluding μ^Scase 2.73 (2.45, 3.02)
+++ FP μ^S 0.43 (0.14, 0.72) 0.004
Cox Model 60 (94%) Hazard Ratio 0.82 (0.66, 1.02) 0.071
Per Incr. Conc.
Repeat 31 (48%) Hazard Ratio 0.61 (0.48, 0.77) 0.002
Excluding Per Incr. Conc.
+++ FP
Case-Only 55 (86%) Prob. Infected 0.64 (0.496, 0.76) 0.058
Sign Test Second-Half
Repeat 29 (45%) Prob. Infected 0.69 (0.49, 0.85) 0.061
Excluding Second-Half
+++ FP
a

The four methods are implemented as described for the simulation study. Target parameters for controls are modified for the Times and Marker methods, and the null value for the Cox model was modified as described in Section 5.4.