Skip to main content

Table 5 Summary meta-analysis results following application of model (2) with and without the imputed data included

From: Meta-analysis of test accuracy studies: an exploratory method for investigating the impact of missing thresholds

Without imputed data With imputed data
Threshold value, x No. studies with this threshold Summary estimate 95% CI Tau No. studies with this threshold Summary estimate 95% CI Tau
Lower Upper
Lower Upper
Sensitivity
0.13 1 0.897 0.756 0.961 0.000 1 0.897 0.756 0.961 0.000
0.14 2 0.910 0.841 0.951 0.000 3 0.955 0.801 0.991 1.147
0.15 5 0.944 0.901 0.969 0.067 6 0.937 0.909 0.957 0.001
0.16 3 0.960 0.831 0.991 0.850 6 0.925 0.890 0.950 0.091
0.17 3 0.909 0.860 0.942 0.000 5 0.911 0.879 0.935 0.000
0.18 3 0.873 0.816 0.914 0.000 5 0.894 0.860 0.920 0.000
0.19 4 0.902 0.851 0.936 0.000 6 0.889 0.855 0.917 0.096
0.2 6 0.875 0.828 0.910 0.234 8 0.886 0.839 0.921 0.312
0.21 3 0.892 0.834 0.931 0.000 7 0.882 0.848 0.909 0.000
0.22 1 0.983 0.782 0.999 0.000 5 0.899 0.761 0.961 0.669
0.23 1 0.950 0.786 0.990 0.000 5 0.870 0.766 0.932 0.534
0.24 2 0.877 0.575 0.974 0.973 5 0.850 0.756 0.912 0.473
0.25 2 0.953 0.832 0.988 0.000 5 0.850 0.759 0.910 0.442
0.28 2 0.818 0.531 0.947 0.837 5 0.818 0.715 0.890 0.473
0.3 4 0.938 0.829 0.979 0.975 7 0.893 0.773 0.954 1.076
0.31 1 0.883 0.713 0.959 0.000 5 0.780 0.689 0.851 0.362
0.32 1 0.850 0.675 0.939 0.000 5 0.781 0.687 0.853 0.363
0.35 1 0.833 0.562 0.951 0.000 4 0.733 0.641 0.809 0.296
0.39 2 0.662 0.530 0.772 0.320 4 0.699 0.607 0.778 0.278
0.4 2 0.720 0.650 0.780 0.000 3 0.724 0.661 0.779 0.000
0.45 1 0.700 0.436 0.876 0.000 3 0.691 0.627 0.748 0.000
0.49 1 0.842 0.774 0.893 0.000 2 0.657 0.590 0.718 0.000
0.5 2 0.844 0.433 0.975 1.230 2 0.844 0.433 0.975 1.230
Specificity
0.13 1 0.651 0.570 0.724 0.000 1 0.651 0.570 0.724 0.000
0.14 2 0.671 0.597 0.736 0.000 3 0.624 0.524 0.714 0.250
0.15 5 0.562 0.366 0.740 0.795 6 0.583 0.421 0.728 0.717
0.16 3 0.803 0.499 0.943 1.026 6 0.661 0.462 0.816 0.910
0.17 3 0.765 0.463 0.925 1.042 5 0.677 0.465 0.834 0.922
0.18 3 0.856 0.436 0.979 1.503 5 0.726 0.458 0.893 1.188
0.19 4 0.708 0.653 0.758 0.000 6 0.720 0.522 0.858 0.961
0.2 6 0.818 0.597 0.931 1.245 8 0.775 0.609 0.884 1.031
0.21 3 0.750 0.672 0.815 0.000 7 0.707 0.635 0.771 0.332
0.22 1 0.692 0.409 0.880 0.000 5 0.705 0.599 0.793 0.438
0.23 1 0.769 0.478 0.924 0.000 5 0.738 0.623 0.828 0.508
0.24 2 0.863 0.764 0.925 0.000 5 0.762 0.636 0.855 0.592
0.25 2 0.848 0.764 0.907 0.000 5 0.798 0.655 0.892 0.724
0.28 2 0.945 0.863 0.979 0.000 5 0.845 0.681 0.933 0.947
0.3 4 0.917 0.703 0.981 1.515 7 0.916 0.793 0.969 1.311
0.31 1 0.923 0.609 0.989 0.000 5 0.886 0.707 0.962 1.183
0.32 1 0.923 0.609 0.989 0.000 5 0.892 0.716 0.964 1.194
0.35 1 0.948 0.876 0.979 0.000 4 0.898 0.701 0.971 1.239
0.39 2 0.980 0.092 1.000 3.125 4 0.933 0.713 0.987 1.556
0.4 2 0.903 0.675 0.977 0.982 3 0.872 0.715 0.949 0.796
0.45 1 0.994 0.915 1.000 0.000 3 0.944 0.587 0.995 1.883
0.49 1 0.842 0.774 0.893 0.000 2 0.838 0.779 0.883 0.000
0.5 2 0.863 0.764 0.925 0.000 2 0.863 0.764 0.925 0.000
  1. NB All meta-analyses used model (2), as model (1) often poorly estimated ρ10t at +1 or -1.
Back to article page

Systematic Reviews

ISSN: 2046-4053