A total of 9170 pregnant women who gave birth were identified in the database, of whom 305 experienced PPH and nobody died in hospital. Among women with PPH, 125 gave birth by caesarean section, 163 were through spontaneous vaginal delivery and 17 were through forceps-assisted vaginal delivery. Among women who suffered PPH, 5 underwent hysterectomy and 13 received uterine artery embolization. Women who underwent hysterectomy were all caesarean-scarred uterus with placenta accreta. Of 13 women who underwent uterine artery embolization, 8 had cesarean-scarred uterus with placenta accreta, 2 had twins, 1 had uterine atony of lower uterine segment after vaginal delivery, 1 had breech birth, 1 had severe preeclampsia. Among women with PPH, 30 had fetal macrosomia, 25 had twins, 95 had placental accreta, 66 had placenta previa, 6 had placental abruption, 6 had low-lying placenta, 13 had chorioamnionitis, 34 had breech birth, 16 had intrahepatic cholestasis of pregnancy and 14 had no complications.
Among 305 PPH, 69 received blood transfusion with 24 received more than 4 red cell packs and 45 received no more than 4 red cell packs. Among 24 with severe PPH, 11 had caesarean-scarred uterus with placenta accreta, 7 had total placenta previa, 1 had uterine atony of lower uterine segment after vaginal delivery, 3 had hepatitis B virus, 2 had placenta accreta. Patients in PPH group had a higher incidence of complications than those in the control group, including fetal macrosomia, twins, placenta accreta, placenta previa, caesarean-scarred uterus, placental abruption, chorioamnionitis, and intrahepatic cholestasis of pregnancy. The following values were higher in PPH group than in control group: parity, spontaneous vaginal delivery, forceps-assisted vaginal delivery, the amount of postpartum bleeding, peak SI, peak SI in vaginal delivery, peak SI in cesarean delivery, SBP at admission, HR of peak SI. The following values were lower in PPH group than in control group: gestational age at delivery, cesarean delivery, Hb before delivery, Hb 24 hours later after delivery. No significant differences were found in the following values between the PPH and control groups: mean maternal age, weight before delivery, body mass index before delivery, SI at admission, HR at admission, SBP of peak SI (Table 1).
Table1. Demographic and clinical characteristics, postpartum outcomes, laboratory indexes of patients who experienced PPH or not.
|
|
PPH group
(n=305)
|
Control group
(n=305)
|
P
|
|
Mean maternal age(years)
|
29.8±4.2
|
29.8±3.8
|
0.984
|
|
Weight before delivery(Kg)
|
67.9±8.4
|
66.6±8.6
|
0.051
|
|
Body mass index before delivery(kg/m2)
|
26.5±2.8
|
26.5±6.6
|
0.987
|
|
Parity (n, %)
|
|
|
|
|
P1
|
178(58)
|
203(67)
|
0.037
|
|
≥P2
|
127(42)
|
102(33)
|
|
Gestational age at delivery(week)
|
38.2±2.4
|
39.0±1.3
|
0.000
|
Mode of delivery (n, %)
|
|
|
|
Spontaneous vaginal delivery
|
163(53)
|
110(36)
|
0.000
|
Forceps-assisted vaginal delivery
|
17(6)
|
0(0)
|
|
Cesarean delivery
|
125(41)
|
195(64)
|
|
The amount of postpartum bleeding
|
1116.5±654.9
|
429.9±126.0
|
0.000
|
Hb before delivery(g/L)
|
117.8±13.9
|
120.7±13.1
|
0.009
|
Hb 24 hours later after delivery(g/L)
|
89.8±15.6
|
110.3±14.3
|
0.000
|
Peak SI
|
0.9±0.2
|
0.8±0.1
|
0.000
|
Peak SI in vaginal delivery
|
0.9±0.2
|
0.8±0.1
|
0.000
|
Peak SI in cesarean delivery
|
0.9±0.1
|
0.8±0.1
|
0.000
|
HR at admission
|
84.7±7.2
|
84.4±9.0
|
0.620
|
SBP at admission
|
114.6±9.3
|
112.6±9.2
|
0.007
|
SI at admission(HR/SBP at admission)
|
0.7±0.1
|
0.8±0.1
|
0.254
|
HR of peak SI
|
95.4±14.5
|
84.6±7.9
|
0.000
|
SBP of peak SI
|
106.2±9.2
|
107.3±8.1
|
0.137
|
PPH: postpartum hemorrhage, SI: shock index, HR: heart rate, SBP: systolic blood pressure. Hb: hemoglobin. Values shown are mean ± SD, n (%).
Table 2 described SI values corresponding to postpartum blood loss at each stage between PPH group and control group: the average SI values of blood loss were 0.80, 0.87,0.92, 0.95 and 1.06, respectively, corresponding to blood loss in༜500ml in control group and 500-1000mL, 1000-1500ml and > 2000ml in PPH group. The SI value corresponding to 500-1000ml blood loss in the control group was 0.75, which referred to patients with 500-1000ml blood loss after caesarean section.
Table 2
SI distribution corresponding to postpartum bleeding at different stage in PPH group and control group
PPH
|
PPH group (\(\stackrel{-}{x}\))
|
Control group (\(\stackrel{-}{x}\))
|
<500ml
|
|
0.80 (0.65~1.07)
|
[500,1000) ml
|
0.87 (0.68~1.18)
|
0.75 (0.62~0.87)
|
[1000,1500) ml
|
0.92 (0.62~1.49)
|
|
[1500,2000) ml
|
0.95 (0.72~1.49)
|
|
>2000 ml
|
1.06 (0.78~1.68)
|
|
PPH: postpartum hemorrhage |
SI was inferior to clinical estimation of postpartum bleeding based on drapes weighing method in detecting PPH (coefficient of correlation, 0.351 vs 0.759) (Figure a and figure b). Figure a showed a fitted curve with the maximum SI plotted against total blood loss calculated by hemoglobinometry for the PPH group and control group, depicting the distribution. The coefficient of determination (R2) was 0.1239 and the coefficient of correlation was 0.351, indicating a low correlation between maximum SI and blood loss.
To determine whether SI is better at detecting PPH than other vital signs including HR and SBP, we evaluated the performance of vital signs based on AUROC. Table 3 summarized the AUROC values with 95% CI for each vital sign to detect PPH (blood loss ༞1,000 mL and ༞1,500 mL). Regarding PPH (༞1,000 and ༞1,500 mL), SI had the highest AUROC of 0.646 (0.590-0.700) and 0.665 (0.582-0.748), respectively. SI was significantly higher than the AUROC of HR (p<0.0001 and p=0.0001, respectively) and SBP (p<0.05 and p༜0.05, respectively). The cutoff points of SI for PPH (༞1,000 and ༞1,500 mL) were 0.87 (sensitivity, 61.88%; specificity ,66.21%) and 0.87 (sensitivity, 75.6%; specificity, 56.2%), respectively. The cutoff points of HR for PPH (༞1,000 and ༞1,500 mL) were 91 bpm (sensitivity, 57.5%; specificity, 61.4%) and 90 bpm (sensitivity, 62.2%; specificity, 52.3%), respectively.
Table 3
AUROC values with 95% confidence interval of vital signs to detect postpartum hemorrhage
Vital sign
|
AUROC (95% CI) for PPH (༞1,000 mL)
|
AUROC (95% CI) for PPH (༞1,500 mL)
|
peak SI
|
0.646 (0.590-0.700)
|
0.665 (0.582-0.748)
|
HR of peak SI
|
0.540 (0.474-0.605)
|
0.574 (0.486-0.662)
|
SBP of peak SI
|
0.580 (0.561-0.605)
|
0.607 (0.561-0.643)
|
PPH: postpartum hemorrhage, SI: shock index, HR: heart rate, SBP: systolic blood pressure. |
According to our data, the mean peak SI in surgical interventions was 0.93±0.19, the mean peak SI in blood transfusion was 0.98±0.24. Age, BMI, delivery mode, fetal macrosomia, placenta accreta, placenta previa, low-lying placenta, HR of peak SI and caesarean-scarred uterus were associated with surgical interventions. Logistic regression showed delivery mode (caesarean delivery) and placenta previa to be independent risk factors for surgical interventions (Table 4).
Table 4
Univariate and logistic regression to identify markers associated with surgical interventions
Univariate regression
|
Value
|
p
|
OR
|
95%CI Lower
|
95%CI Upper
|
age
|
1.753
|
0.020
|
|
|
|
BMI
|
2.961
|
0.000
|
|
|
|
delivery mode
|
173.021
|
0.000
|
18.686
|
10.177
|
34.308
|
fetal macrosomia
|
6.658
|
0.010
|
0.265
|
0.090
|
0.781
|
placenta accreta
|
126.564
|
0.000
|
13.851
|
7.605
|
25.228
|
placenta previa
|
177.654
|
0.000
|
35.363
|
15.161
|
82.485
|
Cesarean-scarred uterus
|
30.418
|
0.000
|
6.985
|
3.132
|
15.577
|
Low-lying placenta
|
12.042
|
0.001
|
|
|
|
HR of peak SI
|
1.921
|
0.000
|
|
|
|
Logistic regression
|
B
|
p
|
OR
|
95%CI lower
|
95%CI upper
|
delivery mode
|
1.967
|
0.000
|
7.150
|
3.555
|
14.381
|
placenta previa
|
2.302
|
0.000
|
9.989
|
3.918
|
25.472
|
BMI: Body mass index, SI: shock index, HR: heart rate, CI: confidence interval, OR: odds ratio |
BMI, delivery mode, placenta accreta, placenta previa, caesarean-scarred uterus, and peak SI ≥ 1 and HR of peak SI were associated with blood transfusion. Four of these risk factors (BMI, delivery mode, placenta previa, peak SI ≥ 1) were independent risk factors for blood transfusion (Table 5).
Table 5
Univariate and logistic regression to identify markers associated with blood transfusion
Univariate regression
|
Value
|
p
|
OR
|
95%CI Lower
|
95%CI Upper
|
BMI
|
3.709
|
0.000
|
|
|
|
delivery mode
|
80.274
|
0.000
|
11.735
|
5.926
|
23.240
|
placenta accreta
|
63.355
|
0.000
|
9.969
|
4.964
|
20.018
|
placenta previa
|
120.971
|
0.000
|
15.316
|
7.998
|
29.330
|
Cesarean-scarred uterus
|
41.270
|
0.000
|
8.029
|
3.776
|
17.073
|
Peak SI≥1
|
12.999
|
0.000
|
2.805
|
1.564
|
5.030
|
HR of peak SI
|
2.158
|
0.000
|
|
|
|
Logistic regression
|
B
|
p
|
OR
|
95%CI lower
|
95%CI upper
|
BMI
|
-0.172
|
0.011
|
0.842
|
0.737
|
0.962
|
delivery mode
|
1.876
|
0
|
6.529
|
2.588
|
16.473
|
placenta previa
|
1.84
|
0
|
6.297
|
2.701
|
14.681
|
Peak SI≥1
|
1.638
|
0
|
5.143
|
2.331
|
11.345
|
BMI: Body mass index, HR: heart rate, SI: shock index. CI: confidence interval, OR: odds ratio |
The sensitivity, specificity, positive likelihood ratio and negative likelihood ratio, negative predictive value for SI ≥ 1 to predict blood transfusion were 0.39 (95%CI:0.28 -0.52), 0.81(95%CI: 0.76-0.86), 2.10 and 0.75, 82% respectively.