Prediction of Stillbirth: An Umbrella Review of Evaluation of Prognostic Variables

R. Townsend, F. G. Sileo, J. Allotey, J. Dodds, A. Heazell, L. Jorgensen, V. B. Kim, L. Magee, B. Mol, J. Sandall, G. C.S. Smith, B. Thilaganathan, P. Vondadelszen, S. Thangaratinam, A. Khalil

Research output: Contribution to journalEditorialOtherpeer-review


The global incidence of stillbirth remains high, despite decreases in maternal and neonatalmorbidity. Therefore, developing strategies to predict and prevent stillbirth is a priority. Reviewing known risk factors can help to identifywomen at high risk of this complication. The aim of this study was to identify and rate the risk factors of stillbirth that could be used to develop clinical models useful in predicting this complication. This study involved a literature search of MEDLINE, Embase, DARE, and Cochrane library from inception to November 2019. Included were reviews that assessed single variables that could accurately predict or were associated with stillbirth, as well as studies that evaluated predictive tests at any trimester of pregnancy. An assessment of the quality of the studies was conducted using the AMSTAR tool and Quality in Prognosis Studies tool to grade the strength of the evidence supporting an association between the variable and stillbirth. A total of 69 systematic reviews were included in the analysis, representing 64 individual variables. The most frequently reported variables associated with stillbirth were maternal age (35 years or older), obesity measures including body mass index (BMI), and maternal diabetes (n = 5, 6, and 5, respectively). Sickle cell disease had the strongest association among other maternal conditions (n = 1; relative risk [RR], 3.99; 95% confidence interval [CI], 2.63-6.04) with supporting evidence rated as "highly suggestive."Other variables with strong associations with stillbirth were previous stillbirth (n = 2; odds ratio [OR], 4.83; 95% CI, 3.77-6.18), previous preterm birth (n = 1; OR, 2.98; 95% CI, 2.05-4.34), and previous birth of a small-forgestational- age neonate (n = 1; OR, 6.00; 95%CI, 3.43-10.49).Maternal smoking was also strongly associated with stillbirth (n = 2), but caffeine and alcohol use were not. The strongest, "highly convincing"evidence of an association with stillbirth were for nulliparity, preexisting hypertension, and high maternal BMI. The best performing ultrasound test for stillbirth appeared to be uterine arteryDoppler, with pooled sensitivity of 65%(95% CI, 38%-85%) and specificity of 82% (95% CI, 72%-88%); the OR was 8.3 (95% CI, 3.0-22.4). PAPP-A and placental growth factor appeared to be the best performing biochemical tests for stillbirth. The sensitivity and specificity of PAPP (n = 2) for stillbirth was 15% and 95%, respectively (95% CI, 8%-26% and 95% CI, 95%-96%, respectively). The OR of placental growth factor (n = 2) was 49.2 (95% CI, 12.7-191). Although these tests had significant associations, the evidence was rated as weak. Risk factors for stillbirth identified from obstetric history and examination had the strongest evidence of association. These included nulliparity, preexisting hypertension, and high BMI. Strong associations were observed for ultrasound and biochemical markers, but the evidence was weak.

Original languageEnglish
Pages (from-to)315-317
Number of pages3
JournalObstetrical and Gynecological Survey
Issue number6
Publication statusPublished - Jun 2021

Cite this