Prediction of miscarriage in women with viable intrauterine pregnancy—A systematic review and diagnostic accuracy meta-analysis

Abstract

Both ultrasound and biochemical markers either alone or in combination have been described in the literature for the prediction of miscarriage. We performed this systematic review and meta-analysis to determine the best combination of biochemical, ultrasound and demographic markers to predict miscarriage in women with viable intrauterine pregnancy. The electronic database search included Medline (1946-June 2017), Embase (1980-June 2017), CINAHL (1981-June 2017) and Cochrane library. Key MESH and Boolean terms were used for the search. Data extraction and collection was performed based on the eligibility criteria by two authors independently. Quality assessment of the individual studies was done using QUADAS 2 (Quality Assessment for Diagnostic Accuracy Studies-2: A Revised Tool) and statistical analysis performed using the Cochrane systematic review manager 5.3 and STATA vs.13.0. Due to the diversity of the combinations used for prediction in the included papers it was not possible to perform a meta-analysis on combination markers. Therefore, we proceeded to perform a meta-analysis on ultrasound markers alone to determine the best marker that can help to improve the diagnostic accuracy of predicting miscarriage in women with viable intrauterine pregnancy. The systematic review identified 18 eligible studies for the quantitative meta-analysis with a total of 5584 women. Among the ultrasound scan markers, fetal bradycardia (n=10 studies, n=1762 women) on hierarchical summary receiver operating characteristic showed sensitivity of 68.41%, specificity of 97.84%, positive likelihood ratio of 31.73 (indicating a large effect on increasing the probability of predicting miscarriage) and negative likelihood ratio of 0.32. In studies for women with threatened miscarriage (n=5 studies, n=771 women) fetal bradycardia showed further increase in sensitivity (84.18%) for miscarriage prediction. Although there is gestational age dependent variation in the fetal heart rate, a plot of fetal heart rate cut off level versus log diagnostic odds ratio showed that at ≤110 beat per minutes the diagnostic power to predict miscarriage is higher. Other markers of intra uterine hematoma, crown rump length and yolk sac had significantly decreased predictive value. Therefore in women with threatened miscarriage and presence of fetal bradycardia on ultrasound scan, there is a role for offering repeat ultrasound scan in a week to ten days interval.

First‐trimester intrauterine hematoma and pregnancy complications

“…When present, the size, location and content of the IUH did not impact on the risk of miscarriage. This finding is similar to that of a recent meta-analysis aiming to assess predictors of miscarriage in viable pregnancy, which also found that IUH was not associated with miscarriage 39 .…”

Which Ultrasonographic Characteristics Predict Miscarriage Risk?

“…Despite the study limitations, we believe the high specificity of bradycardia for prediction of miscarriage is a useful marker for emergency physicians to more accurately counsel patients, manage expectations, and ensure referral for repeated ultrasonography within 7 to 10 days. 4…”

Diagnosis and prediction of miscarriage: can we do better?

“…15 In a recent study of 61 miscarriages, abnormal size of gestational sac and yolk sac appeared as early as 6 weeks of gestation, followed by abnormal changes in fetal heart rate and CRL at 7 and 8 weeks. 14 Although subchorionic haematoma was found to be a predictor of miscarriage in a metaanalysis 11 and in the study by Wan et al, 7 a recent study on pregnancies with detectable fetal heartbeat did not concur with these findings. 10 A meta-analysis of 15 studies including 1263 women with threatened miscarriage found that serum CA 125 is the only serum marker that is useful in predicting outcome of a pregnancy with a viable fetus, whereas serum human chorionic gonadotropin and progesterone are not useful.…”

“…7,9,10 A meta-analysis of 18 eligible studies on ultrasound markers among 5584 women found that fetal bradycardia is the most significant marker, with a sensitivity of 84.2% in the prediction of miscarriage. 11 A more recent study found that the combination of low fetal heart rate and small CRL increases the risk of subsequent pregnancy loss, from 5.0% to 21%. 10 Because fetal heart rate varies with gestation, cut-offs for low fetal heart rate of ≤122, ≤123, and ≤158 beats per minute for gestational weeks 6, 7, and 8, respectively, have been proposed.…”

“…10 Other investigators have suggested a single fetal heart rate cut-off at ≤110 or 100 beats per minute to predict miscarriage. 11,12 Other ultrasonographic markers associated with miscarriage include a small difference between MSD and CRL, 13 and abnormal size of yolk sac. 14 Using three-dimensional ultrasonography, small gestational sac volume (below the 5th percentile) is associated with risk of miscarriage with odds ratio of 5.25.…”