Non-overt disseminated intravascular coagulation (DIC) in pregnancy: a new scoring system for the identification of patients at risk for obstetrical hemorrhage requiring blood transfusion

Abstract

Background:

Non-overt disseminated intravascular coagulation (DIC) is a subclinical hemostatic dysfunction that has not yet reached the decompensation stage. The detection of pregnant patients at this stage may assist in the identification of those who will develop severe obstetrical hemorrhage, as it is one of the leading causes for preventable maternal mortality. Currently, non-overt DIC is diagnosed by a scoring system originally generated by the International Society on Thrombosis and Hemostasis (ISTH), which is based on non-pregnant patients, and does not address the physiologic changes of the hemostatic system during pregnancy.

Objectives:

1) To develop a pregnancy-specific non-overt DIC score, 2) to determine the diagnostic performance of this score in detecting women at risk for obstetrical hemorrhage requiring blood products transfusion, and 3) to compare it to the existing ISTH non-overt DIC score.

Study design:

This retrospective study has longitudinal and cross-sectional components and includes three steps: 1) characterization of the longitudinal changes in the components of modified ISTH non-overt DIC scores, including these parameters—fibrinogen, antithrombin, protein C, prothrombin time (PT), platelets, thrombin-antithrombin (TAT) complex, and D-dimer—during gestation in a group of normal pregnancies (n=50); 2) development of a pregnancy-specific non-overt DIC score in a cross-sectional design of high-risk (n=152) and control (n=50) pregnancies, based on the predictive performance of each analyte for the detection of women at risk for obstetrical hemorrhage requiring blood products transfusion and a logistic regression model; and 3) comparison between the diagnostic performance of the pregnancy-specific non-overt DIC score and the modified ISTH non-overt DIC score to detect, upon admission, women who are at increased risk for subsequent development of obstetrical hemorrhage requiring blood products transfusion.

Results:

1) The study cohort included 202 patients, of which 21 (10%) had obstetrical hemorrhage that required blood products transfusion and were considered to have non-overt DIC; 2) using the non-pregnant ISTH non-overt DIC score, 92% of the patients had a D-dimer concentration above the 0.5 mg/L threshold, and only 2% were identified to have a low fibrinogen concentration (<100 mg/dl), and this scoring system was unable to identify any of the patients with non-overt DIC based on the suggested cutoff of a score of ≥5; 3) the parameters included in the pregnancy-specific non-overt DIC score were selected based on their contribution to the performance of the model for the prediction of women at risk for obstetrical hemorrhage requiring blood products transfusion; as a result, we excluded the PT difference parameter from the score and the TAT complex concentration was added; and 4) a pregnancy-specific non-overt DIC score of ≥3 had a sensitivity of 71.4% and a specificity of 77.9% to identify patients at risk for obstetrical hemorrhage requiring blood products transfusion.

Conclusion:

1) We propose a pregnancy-specific non-overt DIC score adjusted for the physiologic changes in the hemostatic system during gestation; and 2) the pregnancy-specific non-overt DIC score can be a useful tool for the identification of patients at risk for obstetrical hemorrhage requiring blood products transfusion.

Introduction

Disseminated intravascular coagulation (DIC) is a life-threatening complication affecting 0.03% to 0.35% of pregnancies (1, 2), and this condition can be divided into two phases: non-overt and overt (3, 4). “Overt DIC” designates a decompensated hemostatic state that may lead to thrombosis of small- and medium-sized vessels and bleeding due to the consumption of platelets and coagulation factors and to subsequent organ dysfunction and maternal death (4–7).

Non-overt DIC is characterized by a compensated over-activation of the hemostatic system without organ failure or thrombosis of small- and medium-sized vessels (1–6, 8–29). The International Society on Thrombosis and Hemostasis (ISTH) developed a scoring algorithm for non-overt DIC in non-pregnant patients (7, 30–32). Wada et al. (33) proposed a modified non-overt DIC diagnostic criteria that was able to correctly predict the development of overt DIC within one week in 43/44 patients (97.7%) who were asymptomatic upon admission. Moreover, a retrospective study examined the effectiveness of early hemostatic intervention at the pre-DIC stage and at least 80% of cases showed remission, while only 8% progressed and developed overt DIC (3). Obstetrical hemorrhage is a leading cause of maternal morbidity and mortality (34–41), and recent reports suggest that about 50% of these cases are preventable (35–37, 42–46). Thus, early identification of patients at risk for obstetrical bleeding that requires blood products transfusion can be a key step in reducing, at least in part, the rate of preventable maternal deaths due to hemorrhage.

The key challenge in the management of DIC in pregnancy is early diagnosis, a serious concern that led to the development of diagnostic scores (2, 6, 31). To date, the hemostatic scoring systems used to identify patients at risk for the development of DIC are based on the coagulation parameters in the non-pregnant state. However, the physiologic changes in the hemostatic system during pregnancy (42, 47–49) may affect the diagnostic performance of the respective scoring systems. These changes may include an increased concentration of D-dimer (50), fibrinogen (2), factors VII, VIII, IX, X, and XII (2, 42, 51–55), and/or decreased prothrombin time (PT) difference (2). Indeed, the ISTH overt DIC score had a lower performance than that of the pregnancy-specific DIC score for the identification of pregnant patients with DIC (2, 47).

The concept of non-overt DIC is currently not in clinical use in obstetrics and the diagnostic performance of the ISTH non-overt DIC score during pregnancy was not tested. Therefore, we aimed to address the following objectives: 1) to develop a pregnancy-specific non-overt DIC score and 2) to determine the diagnostic performance of the pregnancy-specific non-overt DIC score in detecting women at risk for obstetrical hemorrhage requiring blood products transfusion and to compare it to the existing ISTH non-overt DIC score.

Materials and Methods

Study design and population:

This study included 202 singleton pregnant women assigned to the following study groups: 1) women with a normal pregnancy (n=50) who were followed up longitudinally to determine the changes of the hemostatic parameters included in the non-overt DIC score during pregnancy. Each patient has six blood samples collected at the following gestational-age intervals: 8 to 16 weeks, 16^+1/7 to 24 weeks, 24^+1/7 to 28 weeks, 28^+1/7 to 34 weeks, 34^+1/7 to 37 weeks, and >37 weeks, the sample collected at admission to delivery was used for the cross sectional analysis as well; and 2) women at risk for obstetrical hemorrhage (n=152);. The cross-sectional study consisted of the following diagnoses: uncomplicated normal pregnancies (n=50), preeclampsia (n=46), small for gestational age (SGA) (n=48), acute abruption (n=19), chronic abruption (n=6), preterm prelabor rupture of membranes (preterm PROM) with vaginal bleeding (n=11), recurrent vaginal bleeding (n=7), and HELLP (hemolysis, elevated liver enzymes, and low platelet count) syndrome (n=15).

Exclusion criteria:

1) multiple gestation; 2) known major fetal anomaly or chromosomal abnormality; 3) serious maternal medical illness (e.g., renal insufficiency, congestive heart disease, chronic respiratory insufficiency, asthma requiring systemic steroids, active hepatitis); and 4) patient requirement for anti-platelet or non-steroidal anti-inflammatory drugs.

Participants were recruited from the Center for Advanced Obstetrical Care and Research of the Perinatology Research Branch of the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD), National Institutes of Health, U. S. Department of Health and Human Services; Wayne State University School of Medicine; and the Detroit Medical Center (Detroit, Michigan, USA). All patients provided written informed consent prior to the collection and use of samples for research purposes under the protocols approved by the Institutional Review Boards of Wayne State University and NICHD.

Clinical Definitions:

A normal control pregnancy was defined as the delivery of a term neonate (≥37 weeks) of appropriate birthweight for gestational age without any medical, surgical or obstetrical complications. Labor was defined as the progressive cervical dilation accompanied by regular uterine contractions leading to delivery (56). Spontaneous rupture of membranes was defined as a spontaneous rupture of the chorioamniotic membranes prior to the onset of labor, confirmed with sterile speculum examination and a combination of pooling, ferning, and Nitrazine tests (57, 58). Preterm delivery was defined as any delivery that occurred <37 weeks of gestation (57). Preterm PROM is the rupture of the fetal membranes prior to the onset of labor <37 weeks of gestation (57, 58). Preeclampsia was diagnosed as the new onset of elevated systolic blood pressure ≥140 mm Hg or diastolic blood pressure ≥90 mm Hg after 20 weeks of gestation accompanied by proteinuria (59–63). HELLP syndrome was diagnosed given evidence of hemolysis, elevated liver enzymes, and a low platelet count (59–63). Placental abruption is the separation of the placenta from the uterine wall as diagnosed by painful vaginal bleeding, uterine tenderness or hypertonicity, and a retroplacental hematoma on the placental surface, based on sonographic or pathologic analysis (64, 65). An appropriate-for-gestational-age neonate was defined as having a birthweight between the 10^th and 90^th percentiles for gestational age; an SGA neonate was defined as having a birthweight less than the 10^th percentile for gestational age (66). Estimated blood loss data were retrieved from the clinical blood loss estimation in the patient’s medical record, documented by a physician or midwife. Blood products transfusion refers to the administration of intra-partum or postpartum intravenous blood products including whole blood, packed red blood cells, platelets, fresh-frozen plasma, and/or cryoprecipitate.

Laboratory testing:

Hemostatic markers included in the analysis:

we measured in all study participants, the hemostatic markers included in the modified ISTH non-overt DIC scores proposed by Wada et al. (33): platelet count, prothrombin time (PT) difference, D-dimer, antithrombin III, fibrinogen, thrombin-antithrombin (TAT) complexes, and protein C. Detailed information regarding laboratory testing of the included hemostatic markers can be found in Supplementary Table S1.

The determination of TAT complex was carried out using the Enzygnost® TAT Micro Kit (Siemens Healthcare Diagnostics Products GmbH, Marburg, Germany), according to manufacturer’s recommendations. The sensitivity of the assay was 1.354 mg/L, and the intra- and inter-coefficients of variation were 3.948% and 3.947%, respectively.

The in vitro measurement of fibrinogen was detected by utilizing the Human Fibrinogen SimpleStep ELISA® Kit (Abcam, Cambridge, MA, USA). The sensitivity of the assay was 0.486ng/mL, and the intra- and inter-coefficients of variation were 3.621% and 4.671%, respectively.

Protein C activity was measured with the CHROMOGENIX COAMATIC® Protein C Kit (Instrumentation Laboratory, Bedford, MA, USA). The sensitivity of the assay was 2.265%, and the intra- and inter-coefficients of variation were 2.037% and 11.817%, respectively.

Antithrombin III activity was measured using the CHROMOGENIX COAMATIC® Antithrombin Kit (Instrumentation Laboratory). The sensitivity of the assay was 24.891%, and the intra- and inter-coefficients of variation were 3.374% and 7.616%, respectively.

Platelet, D-dimer, and prothrombin time assays were performed by the Hematology Laboratory of the Detroit Medical Center according to its standardized protocols.

Statistical analysis

Longitudinal analysis

For patients with a normal uncomplicated pregnancy (n=50), blood samples were collected between 10 and 41 weeks of gestation. The total number of blood samples was 300, representing 6 blood samples per patient. A quantile regression method for longitudinal data was applied to study the changes in accord with the gestational weeks for each coagulation parameter.

Cross-sectional analysis

Cross-sectional analysis was performed on blood samples collected from 202 patients (50 controls and 152 with a complicated pregnancy) upon admission to the Labor and Delivery Unit. The Wilcoxon signed-rank test was applied for the cross-sectional analysis when testing the significance of the median difference in the estimated blood loss between the groups of patients with a vaginal delivery or a cesarean delivery and the groups of patients with or without a blood products transfusion. The diagnostic cutoff for estimated blood loss was based on the analysis of the Receiver Operating Characteristic (ROC) curves in determining the association between estimated blood loss and blood products transfusion. The ROC analyses were performed separately for vaginal and cesarean deliveries. The pregnancy-based new diagnostic cut-off values for each coagulation parameter were determined according to the results of the ROC curve analysis indicating an association between the coagulation parameter and blood products transfusion.

Results

Demographics and clinical characteristics

Demographic and clinical characteristics of the study population are presented in Table 1. Of the 202 women included in the study, the majority were African American (95%); median maternal age was 24 years [interquartile range (IQR): 15–29]; the median body mass index was 27.8 kg/m² (IQR: 15.6–33.85); 37% were nulliparous; and 21% self-reported as smokers. The median gestational age at blood sample collection was 36.1 weeks (IQR: 31.7–37.6).

Table 1:

Maternal characteristics in normal and complicated pregnancies

	Group
	Normal pregnancy (n=50)	Preeclampsia (n=46)	Acute abruption (n=19)	Chronic abruption (n=6)	HELLP syndrome (n=15)	PPROM with vaginal bleeding (n=11)	Recurrent vaginal bleeding (n=7)	SGA (n=48)
Maternal age (years)	22.5 (20.0–27.75)	26.0 (22.25–28.7)	21.5 (19.5–29.0)	23.5 (20.0–27.8)	23.0 (19.5–31.0)	24.0 (22.0–31.0)	29.0 (21.5–32.5)	24.0 (21.0–32.5)
BMI	29.3 (24.5–33.8)	32.5 (26.8–38.1)	22.1 (19.1–28.3)	24.6 (22.7–32.7)	24.1 (22.3–35.5)	25.5 (20.6–31.0)	25.7 (23.9–31.1)	25.5 (20.6–31.0)
Smoking status	1(2%)	10(21.7%)	8(42%)	1(16.7%)	2(13.3%)	2(18%)	0(0%)	19(39.6%)
African American	47(94%)	45(97.8%)	18(89.5%)	6(100%)	13(86.7%)	10(91%)	6(85.7%)	48(100%)
Previous preeclampsia	0(0%)	13(28.3%)	0(0%)	0(0%)	1(6.7%)	0(0%)	0(0%)	0(0%)
Nulliparity	19(38%)	17(37%)	0(0%)	0(0%)	5(33.3%)	0(0%)	0(0%)	0(0%)
Cesarean delivery rate	10(20%)	23(50%)	17(84%)	3(50%)	14(93.3%)	5(55%)	4(57.1%)	17(35.4%)
Estimated blood loss (ml)	300 (250–388)	425 (262–900)	800 (600–950)	450 (263–675)	800 (650–1100)	700 (500–900)	1000 (350–1100)	300 (200–700)
Patients requiring blood transfusion	1(2%)	1(2.2%)	3(15.8%)	4(66.7%)	6(40%)	2(18%)	1(14.3%)	3(6.2%)

Data is presented as median (minimum – maximum)

BMI, body mass index; HELLP, hemolysis, elevated liver enzymes, low platelet count; PPROM, preterm prelabor rupture of the membranes; SGA, small for gestational age.

Estimation of blood loss

The estimated blood loss at delivery was recorded for each patient included in the study. The median estimated blood loss differed between women who had either a vaginal [median 250mL (IQR: 200–300mL)] or a cesarean [median 800mL (IQR: 600–1000mL)] delivery (p<0.001). A significantly higher estimated blood loss was reported for patients who required blood products transfusion compared to those who did not (p value <0.001). Based on the results of the ROC analysis, the best estimated maternal blood loss cut-off values that detected patients with obstetrical hemorrhage who required blood products transfusion were ≥ 600 mL in women who delivered vaginally and ≥ 1200mL in those who had a cesarean delivery.

Overall, 10% (21/210) of patients included in the cross-sectional study had an obstetrical hemorrhage requiring blood products transfusion. This rate was lower in the normal pregnancy group than in the group at high risk for obstetrical hemorrhage [2% (1/50)] vs. 13.2% (20/152), respectively; p = 0.025]. Figure 1 presents the distribution of patients who had an obstetrical hemorrhage requiring blood products transfusion associated with the various pregnancy complications. The highest proportion of such cases was among patients with HELLP syndrome [40% (6/15)] and those with chronic abruption [26.7% (4/15)]. Women who had HELLP syndrome had a lower platelet count than other study groups (supplementary Figure A). However, the DIC score of patients with HELLP syndrome was higher than other groups, except for patients with chronic placental abruption and recurrent vaginal bleeding, which did not reach significant difference (Supplementary Figure B). Three of the six patients with HELLP syndrome who required blood products transfusion received also platelets.

Figure 1.

The distribution of patients who required blood product transfusion according to clinical criteria (blue) and who were identified by the pregnancy-specific non-overt DIC score (red).

What are the changes in the hemostatic parameters during normal pregnancy?

The changes in the individual hemostatic parameters during normal pregnancy are presented in Figure 2 and Table 2: 1) the maternal plasma activity of antithrombin III and protein C were relatively constant; 2) maternal plasma D-dimer and fibrinogen concentrations increased with advancing gestation. Of note, the upper percentiles of maternal plasma D-dimer concentration increased substantially after 34 weeks of gestation; 3) the median TAT complex concentrations increased until 24 weeks of gestation, subsequently remained relatively constant until term, and then increased again upon admission for delivery; 4) the median and mean platelet count decreased with advancing gestation; and 5) the median PT of patients with a normal pregnancy became increasingly shorter than that of the controls with advancing gestation, leading to increasingly negative PT differences (Figure 2 and Table 2).

Figure 2.

The distribution (5th, 10th, 50th, 90th and 95th percentiles) and changes in the concentrations of the different components of the non-overt DIC score with advancing gestation in the maternal plasma of women with a normal pregnancy (n=50): A) Thrombin-antithrombin III complex, B) fibrinogen, C) D-dimer, D) platelet count, E) antithrombin activity, and F) protein C activity.

Table 2:

The maternal plasma concentrations of the coagulation parameters included in the non-overt DIC score during gestation in women with a of the normal pregnancy (n=50)

Analyte	Gestational Age at Sample Collection	Median (IQR)	Mean ± SD
Fibrinogen (mg/dl)	8 to 16 weeks	344.37 (311.75, 406.99)	370.66 (114.86)
	16+1/7 to 24 weeks	347.49 (299.8, 432.04)	363.76 (85.2)
	24+1/7 to 28 weeks	335.6 (292.09, 383.72)	337.92 (68.9)
	28+1/7 to 34 weeks	321.87 (301.97, 392.51)	350.75 (75.97)
	34+1/7 to 37 weeks	389.24 (309.51, 452.39)	392.47 (99.33)
	>37 weeks (at admission to delivery)	376.1 (320.63, 431.59)	392.76 (95.77)
Platelets (×103/ul)	8 to 16 weeks	234.55 (198.07, 260.75)	234.93 (68.04)
	16+1/7 to 24 weeks	225 (199.2, 269.8)	228.64 (58.98)
	24+1/7 to 28 weeks	215.85 (182.53, 250.23)	221.86 (54.98)
	28+1/7 to 34 weeks	217.5 (187.4, 253.15)	225.93 (61.45)
	34+1/7 to 37 weeks	216.2 (169.05, 253.15)	215.99 (56.1)
	>37 weeks (at admission to delivery)	224.2 (165.35, 257.8)	214.59 (59.73)
Antithrombin (% of normal)	8 to 16 weeks	96.65 (92.99, 100.91)	97.87 (11.66)
	16+1/7 to 24 weeks	95.01 (88.25, 100.25)	95.61 (11.58)
	24+1/7 to 28 weeks	94.98 (89.58, 100.79)	95.27 (10.43)
	28+1/7 to 34 weeks	99.94 (90.56, 102.27)	97.8 (9.27)
	34+1/7 to 37 weeks	97.29 (92.08, 101.49)	97.19 (9.25)
	>37 weeks (at admission to delivery)	95.4 (88.13, 100.89)	95.65 (12.95)
Protein C (% of normal)	8 to 16 weeks	121.04 (86.59, 152.82)	143.61 (85.49)
	16+1/7 to 24 weeks	133.55 (104.45, 179.7)	166.6 (102.34)
	24+1/7 to 28 weeks	127.35 (108.3, 210.34)	175.74 (136.91)
	28+1/7 to 34 weeks	122.48 (108.36, 145.79)	156.14 (97.65)
	34+1/7 to 37 weeks	119.17 (95.9, 136.36)	124.07 (46.26)
	>37 weeks (at admission to delivery)	122.57 (101.99, 185.36)	147.2 (73.8)
PT difference (seconds)	8 to 16 weeks	−0.43 ( −0.7, 0.54)	0.01 (0.88)
	16+1/7 to 24 weeks	−0.65 ( −0.71, 0)	−0.3 (0.65)
	24+1/7 to 28 weeks	−0.68 ( −0.73, 0.16)	−0.03 (1.12)
	28+1/7 to 34 weeks	−0.68 ( −0.74, −0.11)	−0.1 (1.06)
	34+1/7 to 37 weeks	−0.71 ( −0.74, −0.68)	−0.53 (0.61)
	>37 weeks (at admission to delivery)	−0.65 ( −0.73, 0.05)	−0.12 (1.02)
D-dimer (mg/L)	8 to 16 weeks	0.64 (0.49, 0.99)	0.79 (0.45)
	16+1/7 to 24 weeks	0.85 (0.69, 1.16)	1.02 (0.48)
	24+1/7 to 28 weeks	1.13 (0.9, 1.71)	1.46 (0.91)
	28+1/7 to 34 weeks	1.26 (1.04, 1.86)	1.61 (0.94)
	34+1/7 to 37 weeks	1.52 (1.31, 2.19)	1.99 (1.2)
	>37 weeks (at admission to delivery)	1.81 (1.64, 2.71)	2.36 (1.31)
TAT Complexes (mg/L)	8 to 16 weeks	6.33 (4.82, 10.42)	121.16 (352.22)
	16+1/7 to 24 weeks	8.72 (6.5, 12.76)	98.26 (253.83)
	24+1/7 to 28 weeks	11.23 (8.04, 20.98)	190.85 (420.14)
	28+1/7 to 34 weeks	11.72 (9.05, 14.9)	158.16 (328.04)
	34+1/7 to 37 weeks	11.96 (10.6, 146.56)	190.68 (331.05)
	>37 weeks (at admission to delivery)	14.78 (11.95, 20.6)	264.51 (627.01)

The diagnostic performance of the modified ISTH non-overt DIC score during pregnancy

We tested the diagnostic performance of the modified ISTH non-overt DIC score (33) for the identification of patients at risk for obstetrical hemorrhage requiring blood products transfusion. Based on the ISTH non-overt DIC cut-off for concentrations of fibrinogen (100 mg/dl) and D-dimer (>0.5 mg/L), only 2% of women were below the cut-off for fibrinogen and 92% were above the D-dimer cut-off (Table 3). None of the 21 patients who subsequently required blood products transfusion were identified by the ISTH non-overt DIC score.

Table 3:

Percentages of patients identified as having an abnormal coagulation test at admission by the hemostatic parameters included in the pregnancy-specific and ISTH non-overt DIC scores (n=202).

	Pregnancy-specific cut-off value	Percentage of patients identified by the pregnancy-specific criteria (%)	ISTH non-overt DIC cut-off value	Percentage of patients identified by ISTH criteria (%)
Fibrinogen (mg/dl)	≤ 284	21	<100	2
Antithrombin Activity (% of normal)	≤ 88	27	<80	11
Protein C Activity (% of normal)	≤ 108	41	<70	8
Platelets count (×103/ul)	≤ 100	7	<100	7
D-dimer (mg/L)	>2	42	>0.5	92
Thrombin-antithrombin III (mg/L)	≥ 19	29	NA	NA

^*NA not applicable

Construction of a pregnancy-specific non-overt DIC scoring system

To construct the pregnancy-specific non-overt DIC scoring system, we undertook two analytic approaches: 1) to define the cut-off values for each analyte included in the score based on its distribution in the normal pregnancy group (this approach yielded low diagnostic performance); and 2) to derive the cut-off value for each analyte from the ROC curves plotted to identify the concentration that had the best diagnostic performance in identifying patients who subsequently had obstetrical hemorrhage that required blood products transfusion. The cut-off values, the number of patients identified, and the diagnostic performance for each parameter included in the pregnancy-specific non-overt DIC scoring system are presented in Table 3.

Some of the parameters required additional consideration: 1) the platelet count cut-off value derived from the ROC curve was 194 × 10³/μL; however, this value was not clinically relevant and we then employed two cut-offs at 100,000 × 10³/μL and 50,000 × 10³/μL that had also been used in the ISTH non-overt DIC scores (7, 30–32); and 2) in the case of D-dimer concentrations, the cut-off value derived from ROC was 2 mg/L. However, the D-dimer concentration changes at a relatively faster rate after 34 weeks of gestation, especially at the higher percentiles (Figure 2). To account for these changes in D-dimer concentrations, we added an additional cut-off value at 4 mg/L (based on the 90th percentile for normal pregnancy at admission).

The pregnancy-specific non-overt DIC scores were derived by computing the sum of the weights across the coagulation parameter. The PT difference had a low predictive value for subsequent obstetrical hemorrhage requiring blood products transfusion and was excluded from our proposed scoring system for non-overt DIC during pregnancy. Maternal plasma thrombin antithrombin (TAT) complex concentrations improved the performance of the score and, therefore, were included in the pregnancy-specific non-overt DIC score. The template for scoring non-overt DIC in pregnancy is presented in Table 4. We assigned a value of 1 point when the concentrations were as follows: fibrinogen ≤ 284 mg/dl, antithrombin activity ≤ 88% of normal, protein C activity ≤ 108% of normal, TAT III ≥ 19 mg/L, platelet count > 50 and ≤100 × 103/μL, and D-dimer ≥2 and <4 mg/L. A value of 2 points was assigned for these parameters: a platelet count ≤ 50 × 103/μL or a D-dimer concentration ≥ 4 mg/L. Supplementary Table S2 displays the distribution of the patients according to their pregnancy-specific non-overt DIC scores. A score of 0 represents a normal coagulation status. A higher score, on the other hand, characterizes patients with a compensated abnormal coagulation status. The median non-over DIC score of women with HELLP syndrome was higher than that of controls (supplementary Figure B).

Table 4:

Diagnostic performance of coagulation factors and the non-over DIC score in the identification of patients who will subsequently require blood products transfusion

	Optimal cut-off	Sensitivity	Specificity	LR+	LR-
Thrombin-antithrombin III (mg/L)	19.21	61.90	75.69	2.55	0.50
Fibrinogen (mg/dl)	283.81	57.14	83.98	3.57	0.51
D-Dimer (mg/L)	2.08	57.14	61.88	1.50	0.69
Antithrombin Activity (% of normal)	87.97	38.10	74.59	1.50	0.83
Protein C Activity (% of normal)	108.44	61.90	42.54	1.08	0.90
Platelet count(×103/ul)	194.00	47.62	63.54	1.31	0.82
ISTH non- overt DIC score	≥3	0	96.1	0	NA
Pregnancy specific non-overt DIC score	≥3	71.4	77.9	3.23	0.37

A pregnancy-specific non-overt DIC score of ≥ 3 at the time of admission to delivery identified 71.42% (15/21) of patients who had obstetrical hemorrhage requiring a transfusion of blood products. The distribution of the number of patients identified by the pregnancy-specific non-overt DIC score across the different high-risk subgroups is presented in Figure 1. Of note, in the control, preeclampsia, and recurrent vaginal bleeding groups, there was only 1 patient per group who required a blood transfusion, and these were correctly identified by their respective scores. Among the high-risk subgroups that comprised more than 1 patient requiring a blood transfusion, the pregnancy-specific non-overt DIC score diagnosed 83.3% (5/6) of patients in the HELLP syndrome group, 66.7% (2/3) of patients with acute abruption, 75% (3/4) of those with chronic abruption, but only 33.3% (1/3) in the SGA group (Figure 1).

The ROCs of the pregnancy-specific and the ISTH non-overt DIC scores that identified patients at risk for obstetrical hemorrhage requiring blood products transfusion and those with elevated blood loss are shown in Figure 3 and the diagnostic performance is presented in Table 5. Using the pregnancy-specific non-overt DIC criteria, a score of ≥ 3 identifying patients who would subsequently require blood products transfusion secondary to obstetrical hemorrhage had a 71.4% sensitivity, 77.9% specificity, 27.2% positive predictive value, 96.0% negative predictive value, a positive likelihood ratio of 3.23, and a negative likelihood ratio of 0.37. Application of the ISTH non-overt DIC criteria to a score ≥ 3 had sensitivity of 0% and a specificity of 96.1% for the identification of these patients (Table 5).

Figure 3.

Receiver operating characteristic (ROC) curves comparing the ISTH and pregnancy-specific non-overt DIC scores. (AUC: area under the curve).

Table 5.

Template for calculating the pregnancy-specific non-overt DIC score.

Analyte	Points			Score
	0	1	2
Fibrinogen (mg/dl)	> 284	≤ 284	-
Antithrombin Activity (% of normal)	> 88	≤ 88	-
Protein C Activity (% of normal)	> 108	≤ 108	-
Platelet count (×103/ul)	> 100	> 50 and ≤ 100	≤ 50
D-Dimer (mg/L)	< 2	≥ 2 and < 4	≥ 4
Thrombin-antithrombin III (mg/L)	< 19	≥ 19	-
Total score*

^*A score ≥ 3 fulfills the criteria of pregnancy-specific non-overt DIC.

Discussion

Principal findings of the study

1) The existing ISTH non-overt DIC scoring system, utilizing a cut of ≥ 5, was unable to identify any patient who had an obstetrical hemorrhage requiring blood products transfusion; 2) the pregnancy-specific non-overt DIC scoring system was constructed on the basis of the contribution of each hemostatic parameter as well as the exclusion of PT differences and the TAT complex concentrations introduced to the score; and 3) a pregnancy-specific non-overt DIC score of ≥ 3 points at admission to delivery identified patients at risk for obstetrical hemorrhage requiring blood products transfusion with a sensitivity of 71.4%, specificity of 77.9%, positive predictive value of 27.2%, and negative predictive value of 96.0%.

Obstetrical hemorrhage: Risk factors and clinical consequences

The incidence of postpartum hemorrhage in the general pregnant population ranges between 1% and 5% (67–70). Obstetrical hemorrhage can result from either acute bleeding events, such as a uterine rupture, postpartum hemorrhage, placenta previa, and/or a morbidly adherent placenta (71, 72), or from obstetrical complications that may present with mild or no overt bleeding, such as acute or chronic placental abruption, preeclampsia/HELLP syndrome, preterm PROM with vaginal bleeding, among other complications (73–75). The development of obstetrical hemorrhage that requires blood products transfusion is associated with increased maternal morbidity such as hysterectomy (76–79), transfusion reaction (38), admission to the intensive care unit (38, 78, 79), infectious morbidity and sepsis (76, 79), prolonged hospitalization (38, 76, 79–81), renal failure (78, 79), infertility (79, 81), and maternal and perinatal mortality (76, 78–80, 82, 83).

Obstetrical hemorrhage is the leading cause of preventable maternal death and near-miss events worldwide (34, 43, 84, 85), specifically in low-income countries (34, 35, 37). Despite advancement in the clinical management of obstetrical hemorrhage, 30% to 50% of direct maternal deaths are related to this condition, mainly in the immediate postpartum period (43, 46, 71, 86) and at least 50% of such deaths are preventable (36, 44, 45, 87). However, to date there is no established assessment or scoring method, nor a single laboratory test, that can identify the patients at risk for obstetrical hemorrhage who will subsequently require blood products transfusion.

How can we identify patients at risk for obstetrical hemorrhage?

In the current practice, patients at risk for obstetrical hemorrhage are identified on the basis of clinical and epidemiological characteristics, which define the overall risk of certain groups of patients but cannot ascertain the individual risk for bleeding in a specific parturient. These risk factors include obstetrical complications such as preeclampsia (5, 20, 75, 88–90), HELLP syndrome (20), acute and chronic abruption (5, 20, 81), vaginal bleeding (81), preterm PROM (91), SGA (92, 93), retained stillbirth (5, 20), septic abortion (5, 20), intrauterine infection (5, 20), amniotic fluid embolism (20, 94), and acute fatty liver in pregnancy (20) as well as some subsets of patients with preterm labor (2, 20, 28, 81, 95). Thus, our study aimed to generate a scoring system that is capable of identifying, especially among high-risk patients, those who would subsequently develop obstetrical hemorrhage requiring blood products transfusion. To achieve this goal, we introduced the concept of non-overt DIC as a tool for the identification of such patients.

What is non-overt DIC?

Non-overt DIC is a compensated hemostatic dysfunction, which is not clinically evident, and has not yet reached the stage of overt DIC (3–7). The ISTH introduced the concept of non-overt DIC by proposing a scoring system for the identification of patients at risk for subsequent development of DIC prior to its clinical onset. It has been proposed that “early identification of non-overt DIC can improve outcome by widening the therapeutic time window in critically ill patients at risk of transition to the severe form of DIC and death” (7). Indeed, Wada et al. (33) reported that among 613 patients at risk for the development of DIC, the modified non-overt DIC criteria correctly identified 43/44 patients who developed DIC within a week of admission. Moreover, patients with non-overt DIC had a significantly higher rate of mortality than those with a negative score (33). The concept of non-overt DIC is currently not used in obstetrics although it may be beneficial in the identification of patients at risk for obstetrical hemorrhage requiring blood products transfusion and of those who will gain from preventive hemostatic interventions.

The definition of non-overt DIC was based on the premise of the scientific ISTH sub-committee on DIC and published in 2001 (6). The authors define the criteria for non-overt DIC to “identify the presence of hemostatic dysfunction when it is not yet at the stage of frank decompensation” (6), advocating that global hemostatic tests, such as prothrombin time and partial thromboplastin time, lack the sensitivity to identify the subtle hemostatic changes of patients with non-overt DIC. Therefore, apart from the parameters of platelet count and prothrombin time, scientific ISTH sub-committee on DIC suggested to include in the non-overt DIC score not only tests to assess fibrin degradation but also more specialized coagulation testing for anticoagulation proteins as antithrombin and protein C, and for the TAT complex as a measure of thrombin generation in the score (6, 7, 30–32). The need for more specialized tests, especially those of thrombin generation, is even more prominent during pregnancy given the physiological procoagulant state of the mother (42, 51–53, 96). Indeed, the maternal plasma concentrations of most of the parameters included in the ISTH non-overt DIC score changes during gestation (2, 6, 28, 97–99), suggesting that the applicability of the ISTH non-overt DIC score in pregnancy is questionable.

What are the changes in the components of non-overt DIC during pregnancy?

To address the physiological changes in the hemostatic system occurring during pregnancy, we constructed the normal distribution of each component of the non-overt DIC scoring system in our obstetric population in accord with previous reports (2, 5, 50, 97, 98). In women included in the normal pregnancy group, the concentrations of maternal plasma TAT complex, D-dimer, and fibrinogen increased during gestation, those of antithrombin III and protein C were not affected by gestational age, and the platelet count gradually decreased.

Of interest among our study population, the PT difference between the test (patient’s value) and the control was negative throughout gestation and was not predictive for the need of blood products transfusion. This is most likely due to the physiologic prothrombotic changes in pregnancy that lead to a shorter prothrombin time in pregnant patients that, in turn, leads to a smaller prothrombin time difference between the results of the patients’ samples versus those of the laboratory controls (2, 95, 99). The clinical implication of this observation is that during pregnancy the PT difference will become within the “normal range” only after a substantial hemorrhage; thus, this parameter can serve as an important component of overt DIC during pregnancy(2). However, it is not sensitive enough to identify patients with subclinical hemostatic dysfunction.

D-dimer, an additional component of the non-pregnant ISTH non-overt DIC score, changes substantially during pregnancy. In the current study, the median concentration of D-dimer increased during pregnancy, reaching a value of 1.81 mg/L at admission to delivery, almost 4 times higher than the cut-off value proposed by the ISTH non-overt DIC score of 0.5 mg/L. This observation is supported by other investigators: 1) a large cohort study of 714 healthy pregnant Danish women reported that the 2.5^th percentile of D-dimer was 0.4 mg/L in the third trimester (50); and 2) other studies observed a gradual increase in D-dimer concentration and variance during the third trimester (100–102). In uncomplicated pregnancies, the 2.5^th and 97.5^th percentiles were 0.05 mg/L and 0.74 mg/L in the first trimester and increased to 0.13 mg/L and 1.7 mg/L in the third trimester, respectively (97, 98, 103). Therefore, applying the ISTH D-dimer cut-off on a normal pregnant population will result in misclassification of most patients, especially during the third trimester. Indeed, in our study, 92% of the women with a normal pregnancy were defined as having elevated D-dimer concentrations, suggesting that the ISTH non-overt DIC D-dimer cut-off may not be applicable to pregnant women. Therefore, we have modified this cut-off based on the ROC curve results and the distribution of D-dimer concentrations in our obstetrical population.

There is a constant debate regarding the cut-off of fibrinogen concentration, which confers a risk for bleeding during pregnancy. The ISTH proposed a cut-off value of <100 mg/dl that has identified only 2% of our pregnant patients as having reduced fibrinogen concentration. Erez et al. (2) determined the longitudinal changes in fibrinogen throughout gestation and observed that a fibrinogen concentration of <300 mg/dl is associated with a relative risk of 662.9 for DIC. This cut-off of the fibrinogen concentration is close to our cut-off value of 284 mg/dl, deduced from the ROC curve analysis. Similarly, Abbassi-Ghanavati et al. (97) reported that in the third trimester the 2.5^th and 97.5^th percentiles of fibrinogen concentration in uncomplicated pregnancies are 373 mg/dl and 619 mg/dl, respectively. By contrast, the Clark et al. (94) (based on expert opinion) proposed to use a fibrinogen concentration <200 mg/dl as a cut-off for DIC in pregnancy in women with amniotic fluid embolism had a sensitivity of 14.9% for the diagnosis of patients with DIC when tested in a cohort of pregnant women with DIC but without amniotic fluid emboli (104). Moreover, Charbit et al (105) reported at the onset of treatment of an acute episode of post-partum hemorrhage, fibrinogen concentrations ≤200 mg/dl was the only marker associated with the occurrence of severe PPH (a positive predictive value of 100%). However, our study examined the patients at time of admission to labor and delivery prior to the onset of obstetrical hemorrhage that requires blood products transfusion; thus, a fibrinogen concentration of ≤200 mg/dl may be relevant at time of the acute event but may be too low for the early detection of patients at risk for hemorrhage. Therefore, we would like to suggest that maternal plasma fibrinogen concentration <300 mg/dl can be an appropriate cut-off for patients at risk for obstetrical hemorrhage.

The contribution of the TAT complex concentration to the predictive performance of the pregnancy-specific non-overt DIC scoring is not surprising. Women with pregnancy complications who are at risk for obstetrical hemorrhage have elevated maternal plasma TAT complex concentrations (106–113). The higher fibrinogen concentrations observed during pregnancy (52, 114) compensate in many of the cases for the increased thrombin generation observed in obstetrical syndromes. When this is no longer the case, and the fibrinogen concentration decreases and can no longer compensate for the thrombin generation associated with the obstetrical complication, the risk for maternal hemorrhage increases. This may explain why maternal plasma TAT complex concentrations improved the predictive performance of the pregnancy-specific non-overt DIC score.

Non-overt DIC scoring during pregnancy

Our study is the first to suggest the implementation of non-overt DIC scoring during pregnancy to identify patients at risk for obstetrical hemorrhage requiring blood products transfusion. This proposal is important given that up to 50% of maternal deaths caused by obstetrical hemorrhage are preventable (34–37, 42–46), and this tool may assist the clinician in identifying such patients early in the course of the disease, allowing valuable time to implement preventive strategies. Our score is based on the adjustment of the ISTH non-overt DIC score to the physiological changes of the hemostatic system during pregnancy. A point of criticism that may rise is the utilization of coagulation tests that are not readily available globally. However, as described by Taylor et al. (6) in the publication of the scientific DIC ISTH sub-committee, the diagnosis of non-overt DIC using a standard scoring system is more complicated than that of overt DIC. This is apparent because global coagulation parameters are not sensitive enough and are of limited value in assessing the subtle change of the hemostatic system in the presence of non-overt DIC, especially during pregnancy. Therefore, the diagnosis of non-overt DIC requires the utilization of advanced coagulation tests (e.g., antithrombin, protein C, and TAT complex), that when incorporated into a diagnostic score can identify, for the first time, pregnant women at risk for obstetrical hemorrhage requiring blood products transfusion. This pregnancy-specific non-overt DIC score can also serve as a basis for future comparisons to point-of-care testing such as thromboelastography.

Strengths and limitations of the study

This is the first study to assess a pregnancy-specific scoring system for non-overt DIC. Our scoring template provides the means to predict excessive blood loss that will require blood products transfusion, bearing in mind that obstetrical hemorrhage is the number one cause of preventable maternal death. Moreover, this study shows the importance of the addition of the TAT complex to the pregnancy-specific non-overt DIC scoring system. We replicated findings that showed pregnancy-related prothrombotic physiologic changes in the group of normal pregnant women, specifically fibrinogen and D-dimer.

Our study mainly comprised one ethnic group (African American), consistent with our population. However, our results were similar to those reported in two independent cohorts from Denmark and Japan (2, 50, 103). A second limitation is the use of estimated blood loss as an outcome measure. Traditionally, estimated blood loss during labor has been considered inaccurate with low validity, (115), and a substantial amount of actual blood loss is significantly underestimated (116). To overcome this obstacle, we designed the desired outcome of our score as a combination of the estimated blood loss and the actual need for blood products transfusion. The measurement of coagulation proteins included in the ISTH and in the pregnancy specific non-overt DIC scores expensive and requires a special laboratory that are not available in low resources settings. However, this study is a proof of concept, in which we demonstrated that patients who are at risk for severe obstetrical hemorrhage that will require blood product transfusion may be identified already at admission to the labor and delivery unit. The results of our study can lay the foundation for the development of diagnostic modalities that could be used for the identification of women at risk for obstetrical hemorrhage requiring blood products transfusion also in low resource settings.

Conclusions

We present for the first time a pregnancy-specific non-overt DIC scoring system that can identify, among high-risk patients, the majority of those who will subsequently develop obstetrical hemorrhage requiring blood products transfusion. This scoring system has the potential to assist obstetricians to identify patients at risk for major obstetrical hemorrhage and to employ the necessary preventive measures.