Study population and setting
This retrospective cohort study included 539 Chinese Han patients with inpatient records at the Renmin Hospital of Wuhan University COVID-19 unit from January 11 to April 1, 2020 – including 36 pregnant women – and 36 age-matched pregnant women without COVID-19 or any respiratory symptoms, fever, or other complications who delivered at the same hospital from September 20 to October 30, 2019. They extracted clinical, laboratory (drawn at admission and discharge and, in patients with more than mild disease, with each change in clinical status), and outcome (stratified to mild, moderate, severe, and critical COVID-19) data from the electronic medical record. They described patient characteristics by COVID-19 outcomes and compared pregnant women with COVID-19 to pregnant women without COVID-19 and 82 non-pregnant women with COVID-19 (aged 20-40 years) by COVID-19 outcomes with medians and interquartile ranges or frequency and percentages by groups. They assessed differences by groups with Mann-Whitney U, Kruskal-Wallis H, or Chi-squared tests and used a Bonferroni correction to correct for multiple comparisons. They then used generalized estimating equations to assess how laboratory values changed depending on COVID-19 outcomes and random forests to predict COVID-19 outcomes (mild or moderate vs. severe or critical) based on baseline laboratory values and assessed model discrimination with area under the receiver operating curve (AUC) values.
Summary of Main Findings
Of the 36 pregnant women with COVID-19 (median age 29 years), 7 had mild cases whereas 29 had moderate cases. Most were in their third trimester (25) and 3 reported comorbidities — as compared to a median age of 31 years and no reported comorbidities among the matched 36 pregnant women from 2019. Pregnant women with COVID-19 showed similar immune markers (lymphocyte count, white blood cell count, neutrophil-to-lymphocyte ratio, C3+, CD4+, CD8+, and CD16+56+ cell counts, and IL-6/10) but different coagulation and fibrinolysis markers (higher fibrinogen and D-dimer, lower prothrombin time and activated partial thromboplastin time) as compared with nonpregnant women with severe or critical COVID-19. When compared to nonpregnant women with mild or moderate COVID-19, pregnant women with COVID-19 had increased white blood cell counts, neutrophil-to-lymphocyte ratios, D-dimer, and fibrinogen and decreased lymphocyte counts and IL-4 that were more similar to findings in nonpregnant women with moderate or severe COVID-19. Pregnant women with and without COVID-19 had similar immune and coagulation/fibrinolysis laboratory values. Finally, they found that random forest models including only immune function markers was able to discriminate well between mild or moderate versus severe or critical COVID-19 in the total population with COVID-19 (AUC=0.801, 95% confidence interval (CI)=0.764, 0.838), but improved marginally with the addition of coagulation/fibrinolysis markers (AUC=0.815, 95% CI=0.779, 0.851) and with the addition of additional laboratory values and age (AUC=0.839, 95% CI=0.807, 0.872).
This study compares immune, coagulation, and fibrinolysis markers among pregnant women with COVID-19 to age-matched non-pregnant women with COVID-19 and pregnant women without COVID-19.
The authors did not report the predictive accuracy (how close the predicted outcome probability was to the real outcome) of the models, nor indices of model overfitting, making it difficult to interpret how their findings apply to patients outside of their study population. The random forest model on a sample of 36 pregnant women with COVID-19 and 36 age-matched pregnant women without COVID-19, lacked a sufficient sample size to meaningfully interpret the model outputs. Further, key aspects of the model were not described, including whether they used laboratory values as the outcomes and COVID-19 severity as the exposure (which makes interpretation difficult) or directly compared coefficients for laboratory values from three different models (with pregnant women with COVID-19 as the referent group in all three), each of which has limitations.
These data provide preliminary evidence that pregnancy-associated immune and coagulation/fibrinolysis cascade changes may contribute to biomarker differences among pregnant women with COVID-19 as compared to COVID-19-associated immune and coagulation/fibrinolysis cascade changes in age-matched non-pregnant women with COVID-19.
This review was posted on: 12 February 2021