A total of 83 studies were included. Over 70% met QUADAS criteria for avoidance of verification bias, independence of reference tests and blinding of assessment of index test. Studies scored poorly on descriptions of selection criteria and reference tests, blinding assessment of reference test and data availability.
Pre-eclampsia: There were 74 studies, of which 69 were cohort studies, with 79,547 women (range 28 to 16,808).
In low and unspecified risk patients, pre-eclampsia was best predicted in the second trimester by increased pulsatility index with notching (LR+ 7.5, 95% CI 5.4 to 10.2; LR- 0.59, 95% CI 0.47 to 0.71; one study), followed by bilateral notching (LR+ 6.5, 95% CI 4.3 to 8.7; LR- 0.61 95% CI 0.44 to 0.79; 17 studies).
In high-risk patients, pre-eclampsia was best predicted in the second trimester by unilateral notching (LR+ 20.2, 95% CI 7.5 to 29.5; LR- 0.17 95% CI 0.03 to 0.56; one study), or increased pulsatility index with notching (LR+ 21.0, 95% CI 5.5 to 80.5; LR- 0.82 95% CI 0.72 to 0.93; one study).
Using preventive treatment was not found to improve predictive performance. When using only high quality studies pre-eclampsia in low-risk women was best predicted in the second trimester by unilateral notching (LR+ 12.5, 95% CI 5.1 to 20.0; LR- 0.45 95% CI 0.09 to 0.80).
Intrauterine growth restriction: There were 61 studies, of which 57 were cohort studies, with 41,131 women (range 28 to 7,851).
In low-risk patients, intrauterine growth restriction was best predicted in the second trimester by increased pulsatility index with notching (LR+ 9.1, 95% CI 5.0 to 16.7; LR- 0.89 95% CI 0.85 to 0.93; one study).
In high-risk patients predictive performance was deemed to be low for all indices. Using preventive treatment was not found to improve predictive performance.
Results for other Doppler indices, severe outcomes and first-trimester assessment were also presented.