Accuracy estimates from sensitivity and subgroup analyses of the included studies by different study characteristics†
| Down (trisomy 21) | Edwards (trisomy 18) | Patau (trisomy 13) | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Variables | N | SN (95% CI) | SP (95% CI) | n | SN (95% CI) | SP (95% CI) | n | SN (95% CI) | SP (95% CI) |
| All studies | 40 | 0.993 (0.989 to 0.996) | 0.999 (0.999 to 1.000) | 33 | 0.974 (0.958 to 0.984) | 0.999 (0.999 to 1.000) | 24 | 0.974 (0.861 to 0.996) | 1.000 (0.999 to 1.000) |
| Sensitivity analyses | |||||||||
| Excluding outliers‡ | 37 | 0.993 (0.989 to 0.996) | 1.000 (0.999 to 1.000) | 32 | 0.977 (0.961 to 0.986) | 0.999 (0.999 to 1.000) | 22 | 0.977 (0.818 to 0.998) | 1.000 (0.999 to 1.000) |
| Test failures | |||||||||
| Assuming all+ve | 40 | 0.997 (0.990 to 0.999) | 0.981 (0.972 to 0.988) | 33 | 0.973 (0.956 to 0.983) | 0.983 (0.974 to 0.990) | 24 | 0.979 (0.873 to 0.997) | 0.981 (0.966 to 0.989) |
| Assuming all−ve | 40 | 0.962 (0.948 to 0.973) | 1.000 (0.999 to 1.000) | 33 | 0.942 (0.913 to 0.962) | 0.999 (0.999 to 1.000) | 24 | 0.885 (0.796 to 0.939) | 1.000 (0.999 to 1.000) |
| Intention to diagnosis | 40 | 0.976 (0.959 to 0.986) | 0.981 (0.972 to 0.989) | 33 | 0.958 (0.927 to 0.976) | 0.983 (0.973 to 0.990) | 24 | 0.903 (0.811 to 0.953) | 0.981 (0.966 to 0.989) |
| Assuming all+ve | 40 | 0.994 (0.989 to 0.997) | 0.999 (0.999 to 1.000) | 33 | 0.974 (0.958 to 0.985) | 0.999 (0.999 to 1.000) | 24 | 0.974 (0.863 to 0.996) | 1.000 (0.999 to 1.000) |
| Assuming all−ve | 40 | 0.993 (0.987 to 0.996) | 0.999 (0.999 to 1.000) | 33 | 0.970 (0.945 to 0.984) | 0.999 (0.999 to 1.000) | 24 | 0.976 (0.855 to 0.996) | 1.000 (0.999 to 1.000) |
| Intention to diagnosis | 40 | 0.993 (0.988 to 0.996) | 0.999 (0.999 to 1.000) | 33 | 0.972 (0.950 to 0.985) | 0.999 (0.999 to 1.000) | 24 | 0.976 (0.855 to 0.996) | 1.000 (0.999 to 1.000) |
| Subgroup analyses | |||||||||
| Study design | |||||||||
| Cohort | 5 | 0.932 (0.853 to 0.971) | 0.999 (0.996 to 1.000) | 4 | 0.868 (0.591 to 0.968) | 0.998 (0.994 to 0.999) | 3 | 0.851 (0.498 to 0.971) | 0.999 (0.995 to 1.000) |
| Others | 35 | 0.976 (0.963 to 0.985) | 0.998 (0.997 to 0.999) | 29 | 0.941 (0.914 to 0.960) | 0.998 (0.997 to 0.999) | 21 | 0.970 (0.852 to 0.994) | 1.000 (0.999 to 1.000) |
| Population risk | |||||||||
| General | 6 | 0.959 (0.874 to 0.987) | 0.999 (0.998 to 1.000) | 4 | 0.865 (0.627 to 0.961) | 0.998 (0.997 to 0.999) | 4 | 0.775 (0.135 to 0.987)§ | 1.000 (0.999 to 1.000) |
| High | 22 | 0.973 (0.951 to 0.985) | 0.997 (0.994 to 0.998) | 19 | 0.930 (0.892 to 0.955) | 0.997 (0.995 to 0.999) | 11 | 0.953 (0.864 to 0.985) | 0.999 (0.996 to 1.000) |
| Others | 12 | 0.974 (0.940 to 0.989) | 0.999 (0.998 to 0.999) | 10 | 0.958 (0.907 to 0.982) | 0.999 (0.999 to 1.000) | 9 | 0.988 (0.547 to 1.000) | 1.000 (0.999 to 1.000) |
| Population | |||||||||
| Others | 36 | 0.977 (0.965 to 0.985) | 0.998 (0.997 to 0.999) | 31 | 0.943 (0.917 to 0.960) | 0.998 (0.997 to 0.999) | 23 | 0.974 (0.861 to 0.996) | 1.000 (0.999 to 1.000) |
| Twins | 4 | 0.894 (0.750 to 0.960) | 0.996 (0.996 to 0.996) | 2 | 0.737 (0.202 to 0.969)§ | 0.998 (0.986 to 1.000) | 1* | ||
| First trimester | |||||||||
| 100% | 7 | 0.960 (0.887 to 0.987) | 0.999 (0.998 to 1.000) | 5 | 0.925 (0.814 to 0.972) | 0.998 (0.997 to 0.999) | 5 | 0.850 (0.770 to 0.906)§ | 0.999 (0.998 to 0.999) |
| Others | 33 | 0.973 (0.958 to 0.983) | 0.998 (0.997 to 0.999) | 28 | 0.939 (0.910 to 0.960) | 0.998 (0.997 to 0.999) | 19 | 0.966 (0.872 to 0.992) | 1.000 (0.999 to 1.000) |
| Test types | |||||||||
| DANSR | 9 | 0.958 (0.898 to 0.983) | 0.999 (0.997 to 1.000) | 6 | 0.948 (0.879 to 0.979) | 0.998 (0.996 to 0.999) | 3 | 0.606 (0.216 to 0.895) | 1.000 (0.998 to 1.000) |
| MPSS | 25 | 0.978 (0.963 to 0.987) | 0.998 (0.997 to 0.999) | 23 | 0.936 (0.899 to 0.960) | 0.998 (0.997 to 0.999) | 16 | 0.959 (0.989 to 0.991) | 1.000 (0.999 to 1.000) |
| SNP technology | 4 | 0.984 (0.937 to 0.996) | 0.998 (0.993 to 1.000) | 4 | 0.918 (0.751 to 0.976) | 0.998 (0.994 to 1.000) | 5 | 0.870 (0.647 to 0.960) | 0.998 (0.992 to 0.999) |
| Publication year | |||||||||
| 2007–2013 | 18 | 0.977 (0.958 to 0.988) | 0.998 (0.995 to 0.999) | 15 | 0.954 (0.919 to 0.975) | 0.998 (0.995 to 0.999) | 9 | 0.933 (0.799 to 0.980) | 0.999 (0.993 to 1.000) |
| 2014–2015 | 22 | 0.966 (0.939 to 0.981) | 0.999 (0.998 to 0.999) | 18 | 0.915 (0.853 to 0.952) | 0.996 (0.998 to 0.999) | 15 | 0.984 (0.770 to 0.999) | 1.000 (0.999 to 1.000) |
*Bivariate model inestimable for only one study in the subgroup.23
†Excluded studies with inestimable sensitivity (T21—Hall 2014; T18—Comas 2014, Hall 2014, Zhang (twins) 2015; T13—Sehnert 2011, Beamon 2014, Comas 2014, Bevilacqua 2015, Wax 2015, Zhang (twins) 2015).
‡Excluded outliers (T21—Dhallan 2007, Chiu 2011, Sparks 2012; T18—Chen 2011; T13—Chen 2011, Palomaki 2012).
‡p Value for subgroup differences <0.05 (statistically significant).
SN, sensitivity; SNP, single nucleotide polymorphism; SP, specificity.