A total of 378 pregnant women were admitted to ICU, 352 with an ICU stay at least 24 hours were included, and 26 with an ICU stay of less than 24 hours were excluded. There were 290 women (82.4%) developed serious complications, which we defined as the critical group, while the other 62 women were defined as non-critical group. Of these, two patients died, one from a pulmonary hypertension crisis and one from pulmonary embolism. In general, the most frequent causes of ICU admission were hypertensive disorders of pregnancy (108/352, 30.7%), followed by cardiovascular disease (94/352, 26.7%), obstetric hemorrhage (61/352, 17.3%) and infection (22/352, 6.3%). As is shown in Table 1, there was no difference in age between women with critically ill or not. However, gestational age in group of critically ill were shorter than the group of not critically ill (33.13 ± 0.267 vs. 35.79 ± 0.47, p < 0.001). Both the length of ICU stay (3.503 ± 0.314 vs. 1.758 ± 0.994, p < 0.001) and total length of stay (9.724 ± 0.379 vs. 6.984 ± 0.319, p < 0.001) were longer in women with critically ill. Women experiencing the primary outcome were more likely than those who did not to have a hypertension (33.8% vs. 16.1%, p < 0.001) and obstetric hemorrhage (19.7% vs. 6.5%, p < 0.001). There were no significant differences between the two groups of patients with cardiovascular disease and suspected sepsis. Among the 94 cases with heart disease, 63 cases had a pregnancy risk rating of mWHO III or higher.
In general, MOEWS and ICNARC OEWS in critically ill patients were generally higher than not critically ill (MOEWS 7.56 ± 3.26 vs. 3.31 ± 1.56, p < 0.001; ICNARC OEWS 5.16 ± 3.30 vs. 2.6 ± 1.41, p < 0.001). The trigger frequency of MEOWS Chart and MEWT in critical group were higher than non-critical group (MEOWS Chart 83.4% vs. 51.6%, p < 0.001; MEWT 44.8% vs. 1.6%, p < 0.001).
Table 1
Comparison of clinical characteristics and parameters between critical and non-critical women.
|
|
Critical(n = 290)
|
Non-critical(n = 62)
|
P value
|
Age (years)
|
31.94 ± 0.31
|
31.19 ± 0.52
|
0.326
|
Gestational age(weeks)
|
33.13 ± 0.27
|
35.79 ± 0.47
|
< 0.001
|
Length of ICU stay(days)
|
3.50 ± 0.31
|
1.758 ± 0.99
|
< 0.001
|
Total length of stay(days )
|
9.72 ± 0.38
|
6.98 ± 0.32
|
< 0.001
|
Hypertensive disorders (n, %)
|
98(33.8%)
|
10(16.1%)
|
0.004
|
Cardiovascular disease (n, %)
|
64(22.1%)
|
30(48.4%)
|
0.175
|
Pregnancy risk grade ≥ mWHO III(n)
|
63
|
0
|
|
Obstetric hemorrhage (n, %)
|
57(19.70%)
|
4(6.5%)
|
0.007
|
Suspected sepsis (n, %)
|
17(5.9%)
|
5(8.1%)
|
0.341
|
MOEWS
|
7.56 ± 3.26
|
3.31 ± 1.56
|
< 0.001
|
ICNARC OEWS
|
5.16 ± 3.30
|
2.6 ± 1.41
|
< 0.001
|
MEOWS Chart trigger (n, %)
|
242(83.4%)
|
32(51.6%)
|
< 0.001
|
MEWT trigger (n, %)
|
130(44.8%)
|
1(1.6%)
|
< 0.001
|
The AUROC of MOEWS was 0.91(95% CI, 0.87–0.95, p < 0.001) had a sensitivity of 96.9% and specificity of 83.9% with PPV 96.6% and NPV 85.2%. AUROC of ICNARC OEWS was 0.765 (95% CI, 0.71–0.82, p < 0.001) had a sensitivity of 66.6% and specificity of 75.8% with PPV 93% and NPV 32%. Whereas the AUROC of MEOWS Chart was 0.657 (95% CI, 0.58–0.74, p < 0.001) had a sensitivity of 83.4% and specificity of 48.4% with PPV 88% and NPV 38%. Otherwise, the AUROC of MEWT was 0.716(95% CI, 0.66–0.77, p < 0.001) had a sensitivity of 44.8% and specificity of 98.4% with PPV 99% and NPV 28%. Details and the AUROC curves were shown in Table 2 and Fig. 1.
Table 2
AUCs of four scoring systems for prediction of maternal morbidity.
|
|
AUCs
(95% CI)
|
Sensitivity
(95% CI)
|
Specificity
(95% CI)
|
PPV
(95% CI)
|
NPV
(95% CI)
|
P value
|
MOEWS
|
0.91
(0.87–0.95)
|
96.9%
(98–100)
|
83.9%
(72–92)
|
96.6%
(94–98)
|
85.2%
(74–93)
|
< 0.001
|
ICNARC OEWS
|
0.765
(0.71–0.82)
|
66.6%
(60–71)
|
75.8%
(63–86)
|
93%
(88–96)
|
33%
(25–41)
|
< 0.001
|
MEOWS Chart
|
0.657
(0.58–0.74)
|
83.4%
(78–87)
|
48.4%
(35–61)
|
88%
(84–92)
|
38%
(28–50)
|
< 0.001
|
MEWT
|
0.716
(0.66–0.77)
|
44.8%
(35–51)
|
98.4%
(91–100)
|
99%
(96–100)
|
28%
(22–34)
|
< 0.001
|
We then further explored the predictive efficacy of four scoring systems for pregnancy with hypertension, heart disease, hemorrhage and infection. AUCs of MOEWS, ICNARC OEWS, MEOWS Chart and MEWT in hypertension were 0.896(95%CI, 0.822–0.971, P < 0.001), 0.888(95%CI, 0.813–0.962, P < 0.001), 0.685(95%CI, 0.478–0.892, P > 0.05), and 0.777(95%CI, 0.647–0.906, P < 0.005) respectively. In patients with heart disease and hemorrhage, AUROC of MOEWS was 0.926 (95%CI, 0.871–0.98, P < 0.001) and 0.932 (95%CI, 0.871–0.98, P < 0.005), but AUCs of the other three scoring systems were all low with P values > 0.05. Besides, all the other three scoring systems can effectively discriminate patients with critical infection except MEOWS, which clarified that MOEWS was most effectively identify high-risk women than other systems. Specific results were shown in Fig. 2.
As was shown in Table 3, MOEWS had the highest triggering frequency of T, SBP and DBP, the incidence of critical illness was 88.16%, 88.28% and 96.52%, respectively. ICNARC OEWS had the highest triggering frequency of RR, but only 78.53% of patients are critically ill, indicating that its RR threshold setting is not reasonable. Moreover, the highest SPO2 triggering frequency was found in MEOWS chart with incidence of critically ill was 88.32%. In addition, the triggering frequency of oxygen therapy in MOEWS was 264 cases, of which 84.09% had serious complications, as for ICNARC OEWS, FIO2 of maintaining oxygen concentration with SPO2 > 96% triggered 325 cases, of which 82.77% were critical patients. In general, MEWT had the lowest trigger frequency of all parameters, despite the incidence of serious complications was all above 90%. Otherwise, the triggering frequency and incidence of critical illness of conscious state in the four scoring systems were consistent.
Table 3. Trigger frequency and incidence of critical illness of each parameter index in four scoring systems.
|
MOEWS
Frequency Critical
|
ICNARC OEWS
Frequency Critical
|
MEOWS Chart
Frequency Critical
|
MEWT
Frequency Critical
|
|
T
|
76
|
67(88.16%)
|
37
|
35(94.59%)
|
24
|
23(95.83%)
|
24
|
23(95.83%)
|
RR
|
108
|
102(94.6%)
|
312
|
245(78.53%)
|
107
|
102(95.33%)
|
37
|
37(100%)
|
HR
SBP
|
153
239
|
132(86.27%)
211(88.28%)
|
105
104
|
93(88.57%)
139(95.77%)
|
122
225
|
108(88.52%)
197(87.56%)
|
73
95
|
67(91.38%)
92(96.84%)
|
DBP
SPO2
|
115
138
|
111(96.52%)
122(88.41%)
|
113
|
109(96.46%)
|
113
197
|
109(96.46%)
174(88.32%)
|
66
86
|
65(98.48%)
82(95.35%)
|
FIO2
|
|
|
325
|
269(82.77%)
|
|
|
|
|
Oxygen therapy
|
264
|
222(84.09%)
|
|
|
|
|
|
|
Conscious level
|
5
|
5(100%)
|
5
|
5(100%)
|
5
|
5(100%)
|
5
|
5(100%)
|
Then, we analyzed the characteristics of various parameters between critically ill patients and non-critically ill patients, it was indicated that the RR, SBP and DBP of critically ill were higher than those of non-critically ill patients, while SPO2 was lower, with P < 0.05. Meanwhile, T and HR of critical group were generally higher than that of non-critical group, but the results were not statistically different. Additionally, the predictive ability of each parameter index to critically ill patients was further analyzed. It was shown that the AUCs of individual variables for predicting critically ill patients in order was DBP (0.626, P < 0.005), SBP (0.616, P < 0.005), RR (0.607, P < 0.05) and SPO2 (0.321, P < 0.001). Among them, temperature, HR and conscious level had poor prediction performance. Results could be seen in Table 4 and Fig. 3.
Table 4. Comparison of individual variables between critical and non-critical patients.
|
Critical
Mean Min Max
|
Non-critical
Mean Min Max
|
AUCs
(95% CI)
|
P value
|
T (℃)
|
37.17 ± 0.42
|
34
|
42
|
37.04 ± 0.6
|
36
|
39
|
0.553
(0.477–0.628)
|
0.191
|
RR(breaths/min)
|
20.76 ± 0.24
|
0
|
48
|
19.84 ± 0.14
|
16
|
23
|
0.607
(0.542–0.672)
|
0.008
|
HR(beats/min)
|
91.62 ± 1.59
|
0
|
224
|
85.42 ± 2.47
|
47
|
129
|
0.566
(0.494–0.638)
|
0.102
|
SBP (mmHg)
|
125.63 ± 2.09
|
0
|
239
|
109.21 ± 2.49
|
84
|
165
|
0.616
(0.553–0.679)
|
0.004
|
DBP (mmHg)
|
78.85 ± 1.39
|
0
|
148
|
67.32 ± 1.60
|
50
|
108
|
0.626
(0.564–0.689)
|
0.002
|
SPO2 (%)
|
92.1 ± 6.385
|
47
|
98
|
94.47 ± 01.32
|
91
|
99
|
0.321
(0.262–0.381)
|
0.000
|
Conscious level
|
Alert
|
|
|
Not alert
|
|
|
0.507
(0.428–0.585)
|
0.865
|