CAOG Papers
ajog.org
Transvaginal cervical length and amniotic fluid index: can it predict delivery latency following preterm premature rupture of membranes? Suwan Mehra, MD; Erol Amon, MD; Sarah Hopkins, PhD; Jeffrey Gavard, PhD; Jaye Shyken, MD OBJECTIVE: We sought to determine whether transvaginal cervical
length (TVCL), amniotic fluid index (AFI), or a combination of both can predict delivery latency within 7 days in women presenting with preterm premature rupture of membranes (PPROM). STUDY DESIGN: This was a prospective observational study of
TVCL measurements in 106 singleton pregnancies with PPROM between 23-33 weeks. Delivery latency was defined as the period (in days) from the initial TVCL after PPROM to delivery of the infant, with our primary outcome being delivery within 7 days of TVCL. The independent predictability of significant characteristics for delivery within 7 days was determined using multiple logistic regression. Sensitivity, specificity, and predictive values were used to examine whether the presence of a short TVCL, AFI, or a combination of both affected the risk of delivery within 7 days. RESULTS: Delivery within 7 days occurred in 51/106 (48%) of pregnancies. Median duration (interquartile range) from PPROM to delivery and TVCL to delivery was 8 days (4.0e16.0) and 8 days (3.0e15.0), respectively. Using multiple regression TVCL as a continuous variable (odds ratio, 0.65; 95% confidence interval, 0.44e0.97; P < .05), AFI 5 cm (odds ratio, 4.69; 95% confidence
interval, 1.58e13.93; P < .01) were determined to be independent predictors of delivery within 7 days. In all, 42 women (40%) had a TVCL 2 cm, while 62 (59%) had AFI 5 cm. A total of 26 women (25%) had a combination of both TVCL 2 cm and AFI 5 cm, while 28 women (27%) had neither characteristic. The predictive value of delivery within 7 days for a TVCL 2 cm was 62%, and for an AFI 5 cm was 58%. Having a combination of low TVCL and low AFI did not increase the predictive value of delivery within 7 days (58%). In contrast, only 3 of 27 women (11%) with neither characteristic delivered within 7 days. The predictive value of delivery >7 days for TVCL >2 cm alone was 61%. This predictive value changed when analyzed in conjunction with an AFI 5 cm and >5 cm at 42% and 89%, respectively. CONCLUSION: A shorter TVCL and an AFI 5 cm independently
predict delivery within 7 days in women presenting with PPROM. The combination of an AFI >5 cm and TVCL >2 cm greatly improved the potential to remain undelivered at 7 days following cervical length assessment. These findings may be helpful for counseling and optimizing maternal and neonatal care in women with PPROM. Key words: amniotic fluid index, cervical length, labor latency, preterm birth, preterm premature rupture of membranes
Cite this article as: Mehra S, Amon E, Hopkins S, et al. Transvaginal cervical length and amniotic fluid index: can it predict delivery latency following preterm premature rupture of membranes? Am J Obstet Gynecol 2015;212:x.e1-x.
P
reterm premature rupture of membranes (PPROM) is a breach of the chorioamniotic membrane prior to the onset of labor at 1 digital examination following PPROM. Following study enrollment women were excluded for a variety of reasons (Figure 1). All women were hospitalized and placed on modified bed rest. TVCL was performed within 72 hours of admission using the CLEAR guidelines.22 Measurements of the TVCL were taken after visualizing the endocervical canal in its entirety for 3-5 minutes, with an empty maternal bladder. Calipers were placed where the anterior and posterior walls of the cervix were sonographically opposed and the shortest technically best measurements were used. The presence of funneling was noted. AFI was recorded at the time of the TVCL measurement. Prophylactic antibiotics
1.e2 American Journal of Obstetrics & Gynecology MONTH 2015 FLA 5.2.0 DTD YMOB10223_proof 27 January 2015 3:25 pm ce
CAOG Papers
ajog.org used included ampicillin 2 g intravenously every 6 hours and azithromycin 500 mg intravenously daily for 2 days, followed by oral amoxicillin 250 mg every 8 hours and azithromycin 500 mg daily for 5 days. Two doses of 12 mg betamethasone were given intramuscularly, 24 hours apart. Tocolysis was generally administered during transport. Additional digital examinations were prohibited without visual evidence of cervical change. Expectant management was followed until 34þ weeks’ gestation. Our primary endpoint was a latency period within 7 days from performance of the TVCL at admission. To provide a more meaningful risk assessment for clinical purposes, TVCL and AFI were analyzed as dichotomous variables as 2.0 and >2.0 cm for TVCL, and 5 and >5 cm for AFI. Sensitivity, specificity, and predictive values were used to examine whether the presence of TVCL, AFI, or a combination of both characteristics affected the risk of delivery within 7 days. Comparative analyses were undertaken to determine whether other variables affected latency. Demographic, medical, obstetrical, sonographic, and delivery variables were recorded such as GA at PPROM, history of PPROM or preterm delivery, tobacco and drug use, history of cervical procedures, visual cervical dilation at admission, presence of vaginal bleeding, digital examination performed prior to admission, and presence or absence of funneling at the TVCL assessment. We based our preliminary sample size estimates on previously published PPROM data from Tsoi et al.17 In that study of women with a CL 2.0 cm, 76% delivered within 7 days compared to only 29% of women with a TVCL >2.0 cm. To detect a similar magnitude of difference with a power of 0.8, alpha 7 days using c2, Fisher exact test, and independent Students t test for continuous variables that were normally
TABLE 1
Demographic/medical/obstetric data by delivery latency at 1 week Demographic
£7 d (n [ 51)
>7 d (n [ 55)
Maternal age, y
25.0
(22.0e32.0)
25.0
(20.0e30.0)
Nulliparous
25
49.0
24
43.6
Caucasian
26
51.0
33
60.0
African American
23
45.1
21
38.2
2
3.9
1
1.8
P value 1.00 .58
Race
Other
.58
Body mass index, kg/m2
29.3
(24.5e33.8)
27.8
(23.3e35.2)
.21
Smoking
21
41.2
19
34.5
.48
Illicit drug use
10
19.6
3
5.5
< .05
Private
11
21.6
13
24.1
.10
Medicaid
33
64.7
25
46.3
Self-pay
7
13.7
16
29.6
10
19.6
5
9.1
.12
Pregnancy with PPROM
12
23.5
2
3.6
< .01
Preterm birth
17
33.3
6
10.9
< .01
Cervical incompetence
3
5.9
2
3.6
.67
LEEP or cone biopsy
5
9.8
4
7.3
.74
Insurance
Medical history Asthma Obstetrical history
Data are expressed as median (interquartile range) for continuous variables, and as number and percentage for categorical variables. Insurance was unknown for 1 woman. LEEP, loop electrosurgical excision procedure; PPROM, preterm premature rupture of membranes. Mehra. Cervical length and AFI: can it predict delivery latency PPROM? Am J Obstet Gynecol 2015.
distributed. Kolmogorov-Smirnov test was used to assess normality of data and Mann-Whitney U used for the nonparametric data. Cox regression was used to compare the relationships of latency from PPROM to delivery and from CL to delivery. The independent predictability of statistically significant univariate characteristics on latency from initial TVCL was examined by multiple logistic regression. Sensitivity, specificity, and predictive value were calculated for TVCL (2.0 cm and >2.0 cm) and AFI (5.0 cm and >5.0 cm) in relation to the latency period of 7 days. Positive predictive value (PPV) was defined as the probability for delivery within 7 days from the TVCL. Negative predictive value (NPV) was defined as the probability of remaining pregnant >7
days. A P value < .05 was used to denote statistical significance. All analyses were performed using software (SPSS, Version 21.0; IBM Corp, Armonk, NY).
R ESULTS In all, 129 singleton women with suspected PPROM consented for the study. Figure 1 illustrates the reasons for postconsent exclusions: 106 were included for final analysis. Subjects were between 18-41 years of age, with approximately 90% of cases occurring 7 days of latency from the TVCL measurement are summarized in Tables 1-3. CL and AFI were each significantly associated with delivery within 7 days. The median TVCL was significantly shorter and an AFI (5 cm) was significantly more frequent in women who delivered 7 days. A latent period 7 days was also significantly associated with a later GA at TVCL, prior preterm birth, prior PPROM, illicit drug use, and uterine contractions.
Using significant variables from univariate analyses a stepwise multiple logistic regression model was performed (Table 4). We examined screening parameters for TVCL and AFI as dichotomous variables for the prediction of the latent period of 1 week from TVCL (Table 5). In all, 42 women (40%) had a TVCL 2 cm while 62 (59%) had an AFI 5 cm. Of 105 women, 26 (25%) had a combination of both TVCL 2 cm and AFI 5 cm, while 27 of 105 women (26%) had neither characteristic. Having a combination of low TVCL and low AFI did not increase the PPV of delivery within 7 days (58% for low TVCL/low
TABLE 2
Clinical and sonographic characteristics by delivery latency at 1 week £7 d (n [ 51)
Characteristic
>7 d (n [ 55)
P value
Maternal transports
36
70.6
43
79.6
Gestational age at PPROM, wk
31.4
(29.4e33.0)
28.7
(26.9e31.0)
< .001
Gestational age at first cervical length, wk
31.6
(29.4e33.1)
28.9
(27.0e31.3)
< .001
1.0
(1.0e2.0)
1.0
(1.0e2.0)
1.00
Duration from PPROM to first cervical length, d
.28
Digital cervical examinations following PPROM 0
31
60.8
38
69.1
1
20
39.2
17
30.9
Cervical dilation at admission, cm
1.0
.37
(0.0e2.0)
1.0
(0e1.0)
.21
Vaginal bleeding prior to PPROM
10
19.6
7
12.7
.34
Uterine contractions any time before PPROM
18
35.3
10
18.2
< .05
Uterine contractions immediately before PPROM
14
27.5
5
9.1
< .05
Uterine contractions after PPROM
27
52.9
14
25.9
< .01
Positive group-B streptococcus culture
12
24.0
15
27.3
.70
Positive gonococcus or chlamydia DNA
2
3.9
3
5.6
1.00
Cervical length, cm
2.1
(0.9e3.1)
3.0
(1.9e3.7)
< .01
Amniotic fluid index, cm
3.5
(1.3e5.3)
5.2
(2.7e8.4)
< .05
Funneling
8
16.0
13
23.6
.33
Cephalic
37
72.5
37
67.3
.41
Breech
13
25.5
14
25.5
Transverse
0
0.0
03
5.5
Funic
1
2.0
1
1.8
Fetal presentation at first ultrasound
Data are expressed as median (interquartile range) for continuous variables, and as number and percentage for categorical variables. Cervical dilation at admission was unknown for 2 women. Uterine contractions after PPROM, group-B streptococcus, and funneling were unknown for 1 woman. PPROM, preterm premature rupture of membranes. Mehra. Cervical length and AFI: can it predict delivery latency PPROM? Am J Obstet Gynecol 2015.
1.e4 American Journal of Obstetrics & Gynecology MONTH 2015 FLA 5.2.0 DTD YMOB10223_proof 27 January 2015 3:25 pm ce
CAOG Papers
ajog.org
TABLE 3
Delivery characteristics by delivery latency at 1 week £7 d (n [ 51)
Characteristic Gestational age at delivery, wk
>7 d (n [ 55)
31.9
Birth weight, g
1685
(29.4e33.9) (1415e2116)
31.5 1723
P value (29.3e33.5) (1348e2183)
.69 .28
Duration from PPROM to delivery, d
4.0
(3.0e6.0)
16.0
(11.0e22.3)
< .001
Duration from first cervical length to delivery, d
3.0
(2.0e4.0)
15.0
(10.0e20.0)
< .001
Cord prolapse after PPROM
2
3.9
2
3.7
39
76.5
41
75.9
Indicated delivery
9
17.6
6
11.1
Induced for 34 wk
3
5.9
7
13.0
Vaginal
33
64.7
32
59.3
Cesarean
18
35.3
22
40.7
8
15.7
12
22.7
.39
Histological chorioamnionitis
31
70.5
37
80.4
.27
Funisitis
17
38.6
25
54.3
.14
1
2.0
3
5.0
.62
1.00
Delivery Spontaneous delivery
.34
Mode of delivery
Clinical chorioamnionitis
Endometritis
.57
Data are expressed as median (interquartile range) for continuous variables, and as number and percentage for categorical variables. Delivery outcomes were unknown for 1 woman. Placental findings were unknown for 16 women. PPROM, preterm premature rupture of membranes. Mehra. Cervical length and AFI: can it predict delivery latency PPROM? Am J Obstet Gynecol 2015.
AFI compared to either characteristic alone, 62% for low TVCL, and 58% for low AFI) (Table 5). Only 3 of 27 women (11%) who had neither
characteristic delivered within 7 days. The NPV overall for TVCL >2 cm was 61%. The NPV changed when TVCL >2 cm was analyzed together with an AFI
TABLE 4
Final multiple logistic regression models predicting delivery latency for 104 women with PPROM £7 d from initial cervical length Characteristic
OR
95% CI
P value
Prior pregnancy with PPROM
10.62
1.84e61.45
< .01
Gestational age at cervical length, wk
1.35
1.12e1.63
< .01
Uterine contractions after PPROM
5.55
1.91e16.11
< .01
Cervical length, cm
0.65
0.44e0.97
< .05
Amniotic fluid index (5 cm)
4.69
1.58e13.93
< .01
Final multiple logistic regression model predicting delivery latency for each time period was generated from characteristics that were statistically significant in univariate group comparisons. Amniotic fluid index was entered as dichotomous categorical variable (5 cm and >5 cm). Final model was based on 104 women, since uterine contractions after PPROM and amniotic fluid index were unknown for 1 woman each. Only gestational age at cervical length (and not at PPROM) was included in model. CI, confidence interval; OR, odds ratio; PPROM, preterm premature rupture of membranes. Mehra. Cervical length and AFI: can it predict delivery latency PPROM? Am J Obstet Gynecol 2015.
5 cm (42%) and AFI >5 cm (89%) (Table 5). Tables 6 and 7 describe the analysis of these testing parameters stratified by a GA in women 30 and >30 weeks, respectively. Compared to the overall cohort the NPV of these tests were enhanced in women 30 weeks at PPROM. In women >30 weeks at PPROM the PPV of delivery within a week was enhanced. These findings further highlight the importance of GA as an independent risk factor for latency. Figure 2 shows a receiver operating characteristic (ROC) curve comparing the initial TVCL in women with AFI >5 cm for predicting latency >7 days with an area under the curve of 0.908. According to this ROC curve, a TVCL of 2.17 cm in the setting of an AFI >5 cm had a 79% sensitivity and 93% specificity in predicting latency >7 days. For practical clinical purposes and for comparability with other studies we
MONTH 2015 American Journal of Obstetrics & Gynecology FLA 5.2.0 DTD YMOB10223_proof 27 January 2015 3:25 pm ce
1.e5
CAOG Papers
ajog.org
TABLE 5
Predicting delivery latency within 7 days following PPROM by each factor Variable
Sensitivity
Specificity
PPV
NPV
TVCL 2 vs >2 cm (n ¼ 106)
26/51 ¼ 51%
39/55 ¼ 71%
26/42 ¼ 62%
39/64 ¼ 61%
AFI 5 vs >5 cm (n ¼ 105)
36/50 ¼ 72%
29/55 ¼ 53%
36/62 ¼ 58%
29/43 ¼ 67%
TVCL 2 cm in women with AFI 5 cm (n ¼ 62)
15/36 ¼ 42%
15/26 ¼ 58%
15/26 ¼ 58%
15/36 ¼ 42%
TVCL 2 cm in women with AFI >5 cm (n ¼ 43)
11/14 ¼ 76%
24/29 ¼ 83%
11/16 ¼ 69%
24/27 ¼ 89%
Prevalence of delivery 7 d from test. AFI, amniotic fluid index; NPV, negative predictive value; PPV, positive predictive value; TVCL, transvaginal cervical length. Mehra. Cervical length and AFI: can it predict delivery latency PPROM? Am J Obstet Gynecol 2015.
decided to use a TVCL of 2 cm. In the setting of an AFI >5 cm, a TVCL of 2 cm had a 79% sensitivity and 83% specificity.
C OMMENT The ability to predict the timing of delivery is helpful to both the patient and physician. To better counsel our patients, we sought to determine whether CL and amniotic fluid volume would independently predict latency in a period of 7 days. Interventions such as the administration of steroids, magnesium for neuroprotection, or the safe transfer to a tertiary center can be optimized with this information.23 This information may be particularly important in counseling women who would reject in-hospital management19 or those who leave against medical advice. Expectant management in women with PPROM improves neonatal survival by approximately 2% for each additional day of in utero maturation, with the optimal benefit between 2-27 weeks.24 The relationship of CL on latency following PPROM was demonstrated
in our cohort. We found a positive association between shorter CL and higher delivery rates within 7 days, even after adjustment for other confounding factors. Without knowledge of the results of the TVCL or AFI approximately 48% of women delivered within 7 days. With the knowledge of a TVCL 2 cm or an AFI 5 cm this delivery probability changed to 62% and 58%, respectively. For a better understanding of delivery latency within 7 days common factors have been identified. This may have important clinical implications as the optimal benefits of corticosteroid therapy are seen in women who deliver within this time frame. Using multiple logistic regression we demonstrated that these factors are the presence of uterine contractions after PPROM, history of PPROM, history of preterm birth, higher GA at TVCL (within 3 days of admission), lower CL, and AFI 5 cm. The knowledge of these additional risk factors for a shorter latency may be helpful in counseling and managing women with PPROM. The CL measurement by itself does not have very high sensitivity or
specificity. But as shown by the ROC curve the combination of an AFI >5 cm and a TVCL value of 2.1 cm improved the sensitivity to 79% and specificity to 93%. However for practical purposes of management and comparability to other studies we chose a 2-cm cutoff for analysis. Only 3 of 27 (11%) women with a TVCL >2 cm and AFI >5 cm delivered within 7 days. This indicates a protective effect of both parameters regardless of the GA. Although our numbers are small, counseling can be further refined using GA cutoff of 30 weeks. This knowledge helps one counsel the patient of an improved likelihood of remaining undelivered >7 days. To our knowledge, our study is the first to demonstrate that together TVCL and AFI are predictors of delivery at 7 days in a US population of mixed ethnicity and insurance status who have PPROM. Our findings are in agreement with 3 previous studies conducted outside of the United States that examined the relationship between CL measurements and latency periods in PPROM.13,16,17
TABLE 6
Predicting delivery latency within 7 days following PPROM by each factor £30 wks Variable
Sensitivity
TVCL 2 vs >2 cm (n ¼ 54)
Specificity
PPV
NPV
7/16 ¼ 44%
28/38 ¼ 74%
7/17 ¼ 41%
28/37 ¼ 76%
13/16 ¼ 81%
17/38 ¼ 45%
13/34 ¼ 38%
17/20 ¼ 85%
TVCL 2 cm in women with AFI 5 cm (n ¼ 34)
5/13 ¼ 38%
14/21 ¼ 67%
5/12 ¼ 42%
14/22 ¼ 64%
TVCL 2 cm in women with AFI >5 cm (n ¼ 20)
2/3 ¼ 67%
14/17 ¼ 82%
2/5 ¼ 40%
14/15 ¼ 93%
AFI 5 vs >5 cm (n ¼ 54)
AFI, amniotic fluid index; NPV, negative predictive value; PPROM, preterm premature rupture of membranes; PPV, positive predictive value; TVCL, transvaginal cervical length. Mehra. Cervical length and AFI: can it predict delivery latency PPROM? Am J Obstet Gynecol 2015.
1.e6 American Journal of Obstetrics & Gynecology MONTH 2015 FLA 5.2.0 DTD YMOB10223_proof 27 January 2015 3:25 pm ce
CAOG Papers
ajog.org
TABLE 7
Predicting delivery latency within 7 days following PPROM by each factor >30 wks Variable
Sensitivity
Specificity
PPV
NPV
TVCL 2 vs >2 cm (n ¼ 52)
19/35 ¼ 54%
11/17 ¼ 65%
19/25 ¼ 76%
11/27 ¼ 41%
AFI 5 vs >5 cm (n ¼ 51)
23/34 ¼ 68%
12/17 ¼ 71%
23/28 ¼ 82%
12/23 ¼ 52%
TVCL 2 cm in women with AFI 5 cm (n ¼ 28)
10/23 ¼ 43%
1/5 ¼ 20%
10/14 ¼ 71%
1/14 ¼ 7%
TVCL 2 cm in women with AFI >5 cm (n ¼ 23)
9/11 ¼ 82%
10/12 ¼ 83%
9/11 ¼ 82%
10/12 ¼ 83%
Prevalence of delivery 7 d from test. AFI, amniotic fluid index; NPV, negative predictive value; PPROM, preterm premature rupture of membranes; PPV, positive predictive value; TVCL, transvaginal cervical length. Mehra. Cervical length and AFI: can it predict delivery latency PPROM? Am J Obstet Gynecol 2015.
population delivered within 48 hours of admission. They also reported that TVCL 2.0 cm was associated with significantly shorter latency. Possible reasons for their shorter latency included a much higher clinical chorioamnionitis rate of 67%. They also did not permit tocolysis and patients were actively delivered for an indication of anhydramnios, in contrast to our management. Carlan et al12 in Florida examined 45 women without finding a significant difference in the latency period using a 3-cm cutoff to characterize short CL. Their lack of finding a difference may have been due to a higher and less discriminatory cutoff than a TVCL 2 cm. Our findings appear comparable to previous studies that defined delivery latency from admission or PPROM. We defined latency from TVCL, which was performed within 72 hours after admission. We chose to use this time period to control for the variable amount of time reported from PPROM to admission and time to the ascertainment of TVCL by a qualified sonographer. Of our cases, 64% had a TVCL assessed within 24 hours of PPROM and 97% within 3 days of PPROM. We found no significant difference using the measure of latency from PPROM or from the admission TVCL measurement. This relationship was not specified in previous papers. The strengths of our study include a prospective, well-controlled study design, utilizing multiple statistical approaches. The GA and diagnoses of PPROM were carefully determined.
Biologic mechanisms such as genetic variation and BMI3,25-27 have been proposed as explanations for disparities in PPROM and preterm birth. Other than Tsoi et al,17 none of the other referenced studies provide information on ethnicities or BMI of the study population. Our population consisted of a mixed ethnicity, 55% Caucasian and 42% African American with a median BMI of 29, which differs from the northern and southern European cohort of nonobese, white population and makes our findings more generalizable to a US population. To our knowledge, this is the most comprehensive prospective study in the United States to examine the clinical use of CL in predicting latency in
FIGURE 2
TVCL predicting delivery latency at 7 days in women with AFI >5
AFI, amniotic fluid index; ROC, receiver operating characteristic; TVCL, transvaginal cervical length. Mehra. Cervical length and AFI: can it predict delivery latency PPROM? Am J Obstet Gynecol 2015.
MONTH 2015 American Journal of Obstetrics & Gynecology FLA 5.2.0 DTD YMOB10223_proof 27 January 2015 3:25 pm ce
1.e7
web 4C=FPO
Tsoi et al17 examined 101 singleton pregnancies between 24-36 weeks, excluding women who delivered for medical indications. This northern European population in the United Kingdom and Germany was comprised of 77% Caucasian, 12% Afro-Caribbean, and 11% Asian with a median BMI of 25 kg/m2. Delivery within a week occurred in 57% of those pregnancies. In our population, 48% delivered spontaneously and for clinical indications within the first week. The exclusion of women who delivered for clinical indications and inclusion of a later GA by Tsoi et al17 might explain the difference in rates of delivery within the week. In addition our population was demographically different, representing an urban US population. Similar to our findings, Tsoi et al17 reported a positive relationship of GA at PPROM and uterine contractions with an increased delivery risk. Rizzo et al16 from Italy examined 92 singleton pregnancies with PPROM, and demonstrated that a TVCL 7 days. Based on our findings one may be better able to counsel a patient on her risk of delivery, whether spontaneous or indicated, from the time of the admission TVCL. Potential limitations in our study include patients who were lost to follow-up because they left after their membranes resealed or against medical advice. Our study population included patients with PPROM who underwent spontaneous labor and indicated delivery. By not excluding the patients with indicated delivery we may have limited the generalization of our findings to women with spontaneous labor only. However, a clinician cannot often predict on admission who will ultimately be delivered spontaneously. Although GA at PPROM is an important factor for latency we could not substratify the results beyond the GA cutoff of 30 weeks at PPROM due to inadequate numbers. Adjunctive antibiotics are known to prolong delivery at 7 days in women with PPROM.28,29 As all patients in our study received adjunctive antibiotic therapy we were unable to demonstrate the effects of TVCL and AFI on prolonging latency in the absence of antibiotics. This study supports the predictive ability of TVCL for latency and adds predictive value of longer latency with the combined use of TVCL >2 cm and AFI >5 cm in women with PPROM. We recommend the importance of combining clinical factors into the risk assessment of an individual with PPROM. Based on our study we recommend obtaining an initial TVCL and AFI in women as soon as practical after admission. Maternal symptoms of uterine contractions and history of PPROM and preterm birth should be carefully assessed in the setting of PPROM to help identify those women at high risk of shorter latency. There still remains a need to evaluate multiple gestations, previable PPROM patients, and combining ultrasound findings
with biochemical markers to improve prediction of latency. We recommend further research to verify our results and generate a clinically useful prediction model for latency using multiple parameters in women with PPROM at various GA substrata. REFERENCES 1. Mercer BM. Preterm premature rupture of the membranes: current approaches to evaluation and management. Obstet Gynecol Clin North Am 2005;32:411-28. 2. Lee T, Silver H. Etiology and epidemiology of preterm premature rupture of the membranes. Clin Perinatol 2001;28:721-34. 3. Premature rupture of membranes. Practice bulletin no. 139. Obstet Gynecol 2013;122: 918-30. 4. Martin JA, Hamilton BE, Ventura SJ, et al. Births: final data for 2009. Natl Vital Stat Rep 2011;60:1-70. 5. Cunningham F, Leveno K, Bloom S, Spong CY, Dashe J. Williams obstetrics, 24th ed. New York, NY: McGraw-Hill Professional; 2014. 6. Vijgen SM, Van der Ham DP, Bijlenga D, et al. Economic analysis comparing induction of labor and expectant management in women with preterm prelabor rupture of membranes between 34 and 37 weeks (PPROMEXIL trial). Acta Obstet Gynecol Scand 2014;93:374-81. 7. Magee L, Sawchuck D, Synnes A, et al. Magnesium sulphate for fetal neuroprotection: SOGC clinical practice guideline. J Obstet Gynaecol Can 2011;33:516-29. 8. Conde-Agudelo A, Romero R, Hassan SS, et al. Transvaginal sonographic cervical length for the prediction of spontaneous preterm birth in twin pregnancies: a systematic review and metaanalysis. Am J Obstet Gynecol 2010;203: 128.e1-12. 9. Iams JD, Goldenberg RL, Meis PJ, et al. The length of the cervix and the risk of spontaneous premature delivery: National Institute of Child Health and Human Development Maternal Fetal Medicine Unit Network. N Engl J Med 1996;334: 567-72. 10. Hassan SS, Romero R, Berry SM, et al. Patients with an ultrasonographic cervical length < or ¼15 mm have nearly a 50% risk of early spontaneous preterm delivery. Am J Obstet Gynecol 2000;182:1458-67. 11. Crane JM, Hutchens D. Transvaginal sonographic measurement of cervical length to predict preterm birth in asymptomatic women at increased risk: a systematic review. Ultrasound Obstet Gynecol 2008;31:579-87. 12. Carlan SJ, Richmond LB, O’Brien WF. Randomized trial of endovaginal ultrasound in preterm premature rupture of membranes. Obstet Gynecol 1997;89:458-61. 13. Gire C, Faggianelli P, Nicaise C, et al. Ultrasonographic evaluation of cervical length in
pregnancies complicated by preterm premature rupture of membranes. Ultrasound Obstet Gynecol 2002;19:565-9. 14. To MS, Skentou C, Cicero S, et al. Cervical assessment at the routine 23-weeks’ scan: problems with transabdominal sonography. Ultrasound Obstet Gynecol 2000;15:292-6. 15. Fischer RL, Austin JD. Cervical length measurement by translabial sonography in women with preterm premature rupture of membranes: can it be used to predict the latency period or peripartum maternal infection? J Matern Fetal Neonatal Med 2008;21: 105-9. 16. Rizzo G, Capponi A, Angelini E, et al. The value of transvaginal ultrasonographic examination of the uterine cervix in predicting preterm delivery in patients with preterm premature rupture of membranes. Ultrasound Obstet Gynecol 1998;11:23-9. 17. Tsoi E, Fuchs I, Henrich W, et al. Sonographic measurement of cervical length in preterm prelabor amniorrhexis. Ultrasound Obstet Gynecol 2004;24:550-3. 18. Mercer BM, Rabello YA, Thurnau GR, et al. The NICHD-MFMU antibiotic treatment of preterm PROM study: impact of initial amniotic fluid volume on pregnancy outcome. Am J Obstet Gynecol 2006;194:438-45. 19. Goya M, Bernabeu A, Garcia N, et al. Premature rupture of membranes before 34 weeks managed expectantly: maternal and perinatal outcomes in singletons. J Matern Fetal Neonatal Med 2013;26:290-3. 20. Melamed N, Hadar E, Ben-Haroush A, et al. Factors affecting the duration of the latency period in preterm premature rupture of membranes. J Matern Fetal Neonatal Med 2009;22: 1051-6. 21. Piazze J, Anceschi M, Cerekja A, et al. Validity of amniotic fluid index in preterm rupture of membranes. J Perinat Med 2007;35:394-8. 22. Perinatal Quality Foundation. Clear guidelines for cervical length. Available at: https:// clear.perinatalquality.org. Accessed March 20, 2013. 23. Doyle LW, Crowther CA, Middleton P, et al. Magnesium sulphate for women at risk of preterm birth for neuroprotection of the fetus. Cochrane Database Syst Rev 2009;1: CD004661. 24. Goldenberg RL, Nelson KG, Davis RO, et al. Delay in delivery: influence of gestational age and the duration of delay on perinatal outcome. Obstet Gynecol 1984;64:480-4. 25. Mercer BM, Goldenberg RL, Meis PJ, et al. The Preterm Prediction Study: prediction of preterm premature rupture of membranes through clinical findings and ancillary testing; the National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network. Am J Obstet Gynecol 2000;183: 738-45. 26. Creasy RK, Resnik R, Iams JD, et al. Creasy and Resnik’s maternal-fetal medicine: principles and practice, 7th ed. Philadelphia, PA: Saunders; 2013.
1.e8 American Journal of Obstetrics & Gynecology MONTH 2015 FLA 5.2.0 DTD YMOB10223_proof 27 January 2015 3:25 pm ce
CAOG Papers
ajog.org 27. Guinn DA, Goldenberg RL, Hauth JC, et al. Risk factors for the development of preterm premature rupture of the membranes after arrest of preterm labor. Am J Obstet Gynecol 1995;173:1310-5.
28. Kenyon SL, Taylor DJ, Tarnow-Mordi W, et al. Broad-spectrum antibiotics for preterm, prelabor rupture of fetal membranes: the ORACLE I randomized trial; ORACLE Collaborative Group. Lancet 2001;357:979-88.
29. Amon E, Lewis SV, Sibai BM, et al. Ampicillin prophylaxis in preterm premature rupture of the membranes: a prospective randomized study. Am J Obstet Gynecol 1988;159:539-43.
MONTH 2015 American Journal of Obstetrics & Gynecology FLA 5.2.0 DTD YMOB10223_proof 27 January 2015 3:25 pm ce
1.e9
ajog.org
Transvaginal cervical length and amniotic fluid index: can it predict delivery latency following preterm premature rupture of membranes? Suwan Mehra, MD; Erol Amon, MD; Sarah Hopkins, PhD; Jeffrey Gavard, PhD; Jaye Shyken, MD OBJECTIVE: We sought to determine whether transvaginal cervical
length (TVCL), amniotic fluid index (AFI), or a combination of both can predict delivery latency within 7 days in women presenting with preterm premature rupture of membranes (PPROM). STUDY DESIGN: This was a prospective observational study of
TVCL measurements in 106 singleton pregnancies with PPROM between 23-33 weeks. Delivery latency was defined as the period (in days) from the initial TVCL after PPROM to delivery of the infant, with our primary outcome being delivery within 7 days of TVCL. The independent predictability of significant characteristics for delivery within 7 days was determined using multiple logistic regression. Sensitivity, specificity, and predictive values were used to examine whether the presence of a short TVCL, AFI, or a combination of both affected the risk of delivery within 7 days. RESULTS: Delivery within 7 days occurred in 51/106 (48%) of pregnancies. Median duration (interquartile range) from PPROM to delivery and TVCL to delivery was 8 days (4.0e16.0) and 8 days (3.0e15.0), respectively. Using multiple regression TVCL as a continuous variable (odds ratio, 0.65; 95% confidence interval, 0.44e0.97; P < .05), AFI 5 cm (odds ratio, 4.69; 95% confidence
interval, 1.58e13.93; P < .01) were determined to be independent predictors of delivery within 7 days. In all, 42 women (40%) had a TVCL 2 cm, while 62 (59%) had AFI 5 cm. A total of 26 women (25%) had a combination of both TVCL 2 cm and AFI 5 cm, while 28 women (27%) had neither characteristic. The predictive value of delivery within 7 days for a TVCL 2 cm was 62%, and for an AFI 5 cm was 58%. Having a combination of low TVCL and low AFI did not increase the predictive value of delivery within 7 days (58%). In contrast, only 3 of 27 women (11%) with neither characteristic delivered within 7 days. The predictive value of delivery >7 days for TVCL >2 cm alone was 61%. This predictive value changed when analyzed in conjunction with an AFI 5 cm and >5 cm at 42% and 89%, respectively. CONCLUSION: A shorter TVCL and an AFI 5 cm independently
predict delivery within 7 days in women presenting with PPROM. The combination of an AFI >5 cm and TVCL >2 cm greatly improved the potential to remain undelivered at 7 days following cervical length assessment. These findings may be helpful for counseling and optimizing maternal and neonatal care in women with PPROM. Key words: amniotic fluid index, cervical length, labor latency, preterm birth, preterm premature rupture of membranes
Cite this article as: Mehra S, Amon E, Hopkins S, et al. Transvaginal cervical length and amniotic fluid index: can it predict delivery latency following preterm premature rupture of membranes? Am J Obstet Gynecol 2015;212:x.e1-x.
P
reterm premature rupture of membranes (PPROM) is a breach of the chorioamniotic membrane prior to the onset of labor at 1 digital examination following PPROM. Following study enrollment women were excluded for a variety of reasons (Figure 1). All women were hospitalized and placed on modified bed rest. TVCL was performed within 72 hours of admission using the CLEAR guidelines.22 Measurements of the TVCL were taken after visualizing the endocervical canal in its entirety for 3-5 minutes, with an empty maternal bladder. Calipers were placed where the anterior and posterior walls of the cervix were sonographically opposed and the shortest technically best measurements were used. The presence of funneling was noted. AFI was recorded at the time of the TVCL measurement. Prophylactic antibiotics
1.e2 American Journal of Obstetrics & Gynecology MONTH 2015 FLA 5.2.0 DTD YMOB10223_proof 27 January 2015 3:25 pm ce
CAOG Papers
ajog.org used included ampicillin 2 g intravenously every 6 hours and azithromycin 500 mg intravenously daily for 2 days, followed by oral amoxicillin 250 mg every 8 hours and azithromycin 500 mg daily for 5 days. Two doses of 12 mg betamethasone were given intramuscularly, 24 hours apart. Tocolysis was generally administered during transport. Additional digital examinations were prohibited without visual evidence of cervical change. Expectant management was followed until 34þ weeks’ gestation. Our primary endpoint was a latency period within 7 days from performance of the TVCL at admission. To provide a more meaningful risk assessment for clinical purposes, TVCL and AFI were analyzed as dichotomous variables as 2.0 and >2.0 cm for TVCL, and 5 and >5 cm for AFI. Sensitivity, specificity, and predictive values were used to examine whether the presence of TVCL, AFI, or a combination of both characteristics affected the risk of delivery within 7 days. Comparative analyses were undertaken to determine whether other variables affected latency. Demographic, medical, obstetrical, sonographic, and delivery variables were recorded such as GA at PPROM, history of PPROM or preterm delivery, tobacco and drug use, history of cervical procedures, visual cervical dilation at admission, presence of vaginal bleeding, digital examination performed prior to admission, and presence or absence of funneling at the TVCL assessment. We based our preliminary sample size estimates on previously published PPROM data from Tsoi et al.17 In that study of women with a CL 2.0 cm, 76% delivered within 7 days compared to only 29% of women with a TVCL >2.0 cm. To detect a similar magnitude of difference with a power of 0.8, alpha 7 days using c2, Fisher exact test, and independent Students t test for continuous variables that were normally
TABLE 1
Demographic/medical/obstetric data by delivery latency at 1 week Demographic
£7 d (n [ 51)
>7 d (n [ 55)
Maternal age, y
25.0
(22.0e32.0)
25.0
(20.0e30.0)
Nulliparous
25
49.0
24
43.6
Caucasian
26
51.0
33
60.0
African American
23
45.1
21
38.2
2
3.9
1
1.8
P value 1.00 .58
Race
Other
.58
Body mass index, kg/m2
29.3
(24.5e33.8)
27.8
(23.3e35.2)
.21
Smoking
21
41.2
19
34.5
.48
Illicit drug use
10
19.6
3
5.5
< .05
Private
11
21.6
13
24.1
.10
Medicaid
33
64.7
25
46.3
Self-pay
7
13.7
16
29.6
10
19.6
5
9.1
.12
Pregnancy with PPROM
12
23.5
2
3.6
< .01
Preterm birth
17
33.3
6
10.9
< .01
Cervical incompetence
3
5.9
2
3.6
.67
LEEP or cone biopsy
5
9.8
4
7.3
.74
Insurance
Medical history Asthma Obstetrical history
Data are expressed as median (interquartile range) for continuous variables, and as number and percentage for categorical variables. Insurance was unknown for 1 woman. LEEP, loop electrosurgical excision procedure; PPROM, preterm premature rupture of membranes. Mehra. Cervical length and AFI: can it predict delivery latency PPROM? Am J Obstet Gynecol 2015.
distributed. Kolmogorov-Smirnov test was used to assess normality of data and Mann-Whitney U used for the nonparametric data. Cox regression was used to compare the relationships of latency from PPROM to delivery and from CL to delivery. The independent predictability of statistically significant univariate characteristics on latency from initial TVCL was examined by multiple logistic regression. Sensitivity, specificity, and predictive value were calculated for TVCL (2.0 cm and >2.0 cm) and AFI (5.0 cm and >5.0 cm) in relation to the latency period of 7 days. Positive predictive value (PPV) was defined as the probability for delivery within 7 days from the TVCL. Negative predictive value (NPV) was defined as the probability of remaining pregnant >7
days. A P value < .05 was used to denote statistical significance. All analyses were performed using software (SPSS, Version 21.0; IBM Corp, Armonk, NY).
R ESULTS In all, 129 singleton women with suspected PPROM consented for the study. Figure 1 illustrates the reasons for postconsent exclusions: 106 were included for final analysis. Subjects were between 18-41 years of age, with approximately 90% of cases occurring 7 days of latency from the TVCL measurement are summarized in Tables 1-3. CL and AFI were each significantly associated with delivery within 7 days. The median TVCL was significantly shorter and an AFI (5 cm) was significantly more frequent in women who delivered 7 days. A latent period 7 days was also significantly associated with a later GA at TVCL, prior preterm birth, prior PPROM, illicit drug use, and uterine contractions.
Using significant variables from univariate analyses a stepwise multiple logistic regression model was performed (Table 4). We examined screening parameters for TVCL and AFI as dichotomous variables for the prediction of the latent period of 1 week from TVCL (Table 5). In all, 42 women (40%) had a TVCL 2 cm while 62 (59%) had an AFI 5 cm. Of 105 women, 26 (25%) had a combination of both TVCL 2 cm and AFI 5 cm, while 27 of 105 women (26%) had neither characteristic. Having a combination of low TVCL and low AFI did not increase the PPV of delivery within 7 days (58% for low TVCL/low
TABLE 2
Clinical and sonographic characteristics by delivery latency at 1 week £7 d (n [ 51)
Characteristic
>7 d (n [ 55)
P value
Maternal transports
36
70.6
43
79.6
Gestational age at PPROM, wk
31.4
(29.4e33.0)
28.7
(26.9e31.0)
< .001
Gestational age at first cervical length, wk
31.6
(29.4e33.1)
28.9
(27.0e31.3)
< .001
1.0
(1.0e2.0)
1.0
(1.0e2.0)
1.00
Duration from PPROM to first cervical length, d
.28
Digital cervical examinations following PPROM 0
31
60.8
38
69.1
1
20
39.2
17
30.9
Cervical dilation at admission, cm
1.0
.37
(0.0e2.0)
1.0
(0e1.0)
.21
Vaginal bleeding prior to PPROM
10
19.6
7
12.7
.34
Uterine contractions any time before PPROM
18
35.3
10
18.2
< .05
Uterine contractions immediately before PPROM
14
27.5
5
9.1
< .05
Uterine contractions after PPROM
27
52.9
14
25.9
< .01
Positive group-B streptococcus culture
12
24.0
15
27.3
.70
Positive gonococcus or chlamydia DNA
2
3.9
3
5.6
1.00
Cervical length, cm
2.1
(0.9e3.1)
3.0
(1.9e3.7)
< .01
Amniotic fluid index, cm
3.5
(1.3e5.3)
5.2
(2.7e8.4)
< .05
Funneling
8
16.0
13
23.6
.33
Cephalic
37
72.5
37
67.3
.41
Breech
13
25.5
14
25.5
Transverse
0
0.0
03
5.5
Funic
1
2.0
1
1.8
Fetal presentation at first ultrasound
Data are expressed as median (interquartile range) for continuous variables, and as number and percentage for categorical variables. Cervical dilation at admission was unknown for 2 women. Uterine contractions after PPROM, group-B streptococcus, and funneling were unknown for 1 woman. PPROM, preterm premature rupture of membranes. Mehra. Cervical length and AFI: can it predict delivery latency PPROM? Am J Obstet Gynecol 2015.
1.e4 American Journal of Obstetrics & Gynecology MONTH 2015 FLA 5.2.0 DTD YMOB10223_proof 27 January 2015 3:25 pm ce
CAOG Papers
ajog.org
TABLE 3
Delivery characteristics by delivery latency at 1 week £7 d (n [ 51)
Characteristic Gestational age at delivery, wk
>7 d (n [ 55)
31.9
Birth weight, g
1685
(29.4e33.9) (1415e2116)
31.5 1723
P value (29.3e33.5) (1348e2183)
.69 .28
Duration from PPROM to delivery, d
4.0
(3.0e6.0)
16.0
(11.0e22.3)
< .001
Duration from first cervical length to delivery, d
3.0
(2.0e4.0)
15.0
(10.0e20.0)
< .001
Cord prolapse after PPROM
2
3.9
2
3.7
39
76.5
41
75.9
Indicated delivery
9
17.6
6
11.1
Induced for 34 wk
3
5.9
7
13.0
Vaginal
33
64.7
32
59.3
Cesarean
18
35.3
22
40.7
8
15.7
12
22.7
.39
Histological chorioamnionitis
31
70.5
37
80.4
.27
Funisitis
17
38.6
25
54.3
.14
1
2.0
3
5.0
.62
1.00
Delivery Spontaneous delivery
.34
Mode of delivery
Clinical chorioamnionitis
Endometritis
.57
Data are expressed as median (interquartile range) for continuous variables, and as number and percentage for categorical variables. Delivery outcomes were unknown for 1 woman. Placental findings were unknown for 16 women. PPROM, preterm premature rupture of membranes. Mehra. Cervical length and AFI: can it predict delivery latency PPROM? Am J Obstet Gynecol 2015.
AFI compared to either characteristic alone, 62% for low TVCL, and 58% for low AFI) (Table 5). Only 3 of 27 women (11%) who had neither
characteristic delivered within 7 days. The NPV overall for TVCL >2 cm was 61%. The NPV changed when TVCL >2 cm was analyzed together with an AFI
TABLE 4
Final multiple logistic regression models predicting delivery latency for 104 women with PPROM £7 d from initial cervical length Characteristic
OR
95% CI
P value
Prior pregnancy with PPROM
10.62
1.84e61.45
< .01
Gestational age at cervical length, wk
1.35
1.12e1.63
< .01
Uterine contractions after PPROM
5.55
1.91e16.11
< .01
Cervical length, cm
0.65
0.44e0.97
< .05
Amniotic fluid index (5 cm)
4.69
1.58e13.93
< .01
Final multiple logistic regression model predicting delivery latency for each time period was generated from characteristics that were statistically significant in univariate group comparisons. Amniotic fluid index was entered as dichotomous categorical variable (5 cm and >5 cm). Final model was based on 104 women, since uterine contractions after PPROM and amniotic fluid index were unknown for 1 woman each. Only gestational age at cervical length (and not at PPROM) was included in model. CI, confidence interval; OR, odds ratio; PPROM, preterm premature rupture of membranes. Mehra. Cervical length and AFI: can it predict delivery latency PPROM? Am J Obstet Gynecol 2015.
5 cm (42%) and AFI >5 cm (89%) (Table 5). Tables 6 and 7 describe the analysis of these testing parameters stratified by a GA in women 30 and >30 weeks, respectively. Compared to the overall cohort the NPV of these tests were enhanced in women 30 weeks at PPROM. In women >30 weeks at PPROM the PPV of delivery within a week was enhanced. These findings further highlight the importance of GA as an independent risk factor for latency. Figure 2 shows a receiver operating characteristic (ROC) curve comparing the initial TVCL in women with AFI >5 cm for predicting latency >7 days with an area under the curve of 0.908. According to this ROC curve, a TVCL of 2.17 cm in the setting of an AFI >5 cm had a 79% sensitivity and 93% specificity in predicting latency >7 days. For practical clinical purposes and for comparability with other studies we
MONTH 2015 American Journal of Obstetrics & Gynecology FLA 5.2.0 DTD YMOB10223_proof 27 January 2015 3:25 pm ce
1.e5
CAOG Papers
ajog.org
TABLE 5
Predicting delivery latency within 7 days following PPROM by each factor Variable
Sensitivity
Specificity
PPV
NPV
TVCL 2 vs >2 cm (n ¼ 106)
26/51 ¼ 51%
39/55 ¼ 71%
26/42 ¼ 62%
39/64 ¼ 61%
AFI 5 vs >5 cm (n ¼ 105)
36/50 ¼ 72%
29/55 ¼ 53%
36/62 ¼ 58%
29/43 ¼ 67%
TVCL 2 cm in women with AFI 5 cm (n ¼ 62)
15/36 ¼ 42%
15/26 ¼ 58%
15/26 ¼ 58%
15/36 ¼ 42%
TVCL 2 cm in women with AFI >5 cm (n ¼ 43)
11/14 ¼ 76%
24/29 ¼ 83%
11/16 ¼ 69%
24/27 ¼ 89%
Prevalence of delivery 7 d from test. AFI, amniotic fluid index; NPV, negative predictive value; PPV, positive predictive value; TVCL, transvaginal cervical length. Mehra. Cervical length and AFI: can it predict delivery latency PPROM? Am J Obstet Gynecol 2015.
decided to use a TVCL of 2 cm. In the setting of an AFI >5 cm, a TVCL of 2 cm had a 79% sensitivity and 83% specificity.
C OMMENT The ability to predict the timing of delivery is helpful to both the patient and physician. To better counsel our patients, we sought to determine whether CL and amniotic fluid volume would independently predict latency in a period of 7 days. Interventions such as the administration of steroids, magnesium for neuroprotection, or the safe transfer to a tertiary center can be optimized with this information.23 This information may be particularly important in counseling women who would reject in-hospital management19 or those who leave against medical advice. Expectant management in women with PPROM improves neonatal survival by approximately 2% for each additional day of in utero maturation, with the optimal benefit between 2-27 weeks.24 The relationship of CL on latency following PPROM was demonstrated
in our cohort. We found a positive association between shorter CL and higher delivery rates within 7 days, even after adjustment for other confounding factors. Without knowledge of the results of the TVCL or AFI approximately 48% of women delivered within 7 days. With the knowledge of a TVCL 2 cm or an AFI 5 cm this delivery probability changed to 62% and 58%, respectively. For a better understanding of delivery latency within 7 days common factors have been identified. This may have important clinical implications as the optimal benefits of corticosteroid therapy are seen in women who deliver within this time frame. Using multiple logistic regression we demonstrated that these factors are the presence of uterine contractions after PPROM, history of PPROM, history of preterm birth, higher GA at TVCL (within 3 days of admission), lower CL, and AFI 5 cm. The knowledge of these additional risk factors for a shorter latency may be helpful in counseling and managing women with PPROM. The CL measurement by itself does not have very high sensitivity or
specificity. But as shown by the ROC curve the combination of an AFI >5 cm and a TVCL value of 2.1 cm improved the sensitivity to 79% and specificity to 93%. However for practical purposes of management and comparability to other studies we chose a 2-cm cutoff for analysis. Only 3 of 27 (11%) women with a TVCL >2 cm and AFI >5 cm delivered within 7 days. This indicates a protective effect of both parameters regardless of the GA. Although our numbers are small, counseling can be further refined using GA cutoff of 30 weeks. This knowledge helps one counsel the patient of an improved likelihood of remaining undelivered >7 days. To our knowledge, our study is the first to demonstrate that together TVCL and AFI are predictors of delivery at 7 days in a US population of mixed ethnicity and insurance status who have PPROM. Our findings are in agreement with 3 previous studies conducted outside of the United States that examined the relationship between CL measurements and latency periods in PPROM.13,16,17
TABLE 6
Predicting delivery latency within 7 days following PPROM by each factor £30 wks Variable
Sensitivity
TVCL 2 vs >2 cm (n ¼ 54)
Specificity
PPV
NPV
7/16 ¼ 44%
28/38 ¼ 74%
7/17 ¼ 41%
28/37 ¼ 76%
13/16 ¼ 81%
17/38 ¼ 45%
13/34 ¼ 38%
17/20 ¼ 85%
TVCL 2 cm in women with AFI 5 cm (n ¼ 34)
5/13 ¼ 38%
14/21 ¼ 67%
5/12 ¼ 42%
14/22 ¼ 64%
TVCL 2 cm in women with AFI >5 cm (n ¼ 20)
2/3 ¼ 67%
14/17 ¼ 82%
2/5 ¼ 40%
14/15 ¼ 93%
AFI 5 vs >5 cm (n ¼ 54)
AFI, amniotic fluid index; NPV, negative predictive value; PPROM, preterm premature rupture of membranes; PPV, positive predictive value; TVCL, transvaginal cervical length. Mehra. Cervical length and AFI: can it predict delivery latency PPROM? Am J Obstet Gynecol 2015.
1.e6 American Journal of Obstetrics & Gynecology MONTH 2015 FLA 5.2.0 DTD YMOB10223_proof 27 January 2015 3:25 pm ce
CAOG Papers
ajog.org
TABLE 7
Predicting delivery latency within 7 days following PPROM by each factor >30 wks Variable
Sensitivity
Specificity
PPV
NPV
TVCL 2 vs >2 cm (n ¼ 52)
19/35 ¼ 54%
11/17 ¼ 65%
19/25 ¼ 76%
11/27 ¼ 41%
AFI 5 vs >5 cm (n ¼ 51)
23/34 ¼ 68%
12/17 ¼ 71%
23/28 ¼ 82%
12/23 ¼ 52%
TVCL 2 cm in women with AFI 5 cm (n ¼ 28)
10/23 ¼ 43%
1/5 ¼ 20%
10/14 ¼ 71%
1/14 ¼ 7%
TVCL 2 cm in women with AFI >5 cm (n ¼ 23)
9/11 ¼ 82%
10/12 ¼ 83%
9/11 ¼ 82%
10/12 ¼ 83%
Prevalence of delivery 7 d from test. AFI, amniotic fluid index; NPV, negative predictive value; PPROM, preterm premature rupture of membranes; PPV, positive predictive value; TVCL, transvaginal cervical length. Mehra. Cervical length and AFI: can it predict delivery latency PPROM? Am J Obstet Gynecol 2015.
population delivered within 48 hours of admission. They also reported that TVCL 2.0 cm was associated with significantly shorter latency. Possible reasons for their shorter latency included a much higher clinical chorioamnionitis rate of 67%. They also did not permit tocolysis and patients were actively delivered for an indication of anhydramnios, in contrast to our management. Carlan et al12 in Florida examined 45 women without finding a significant difference in the latency period using a 3-cm cutoff to characterize short CL. Their lack of finding a difference may have been due to a higher and less discriminatory cutoff than a TVCL 2 cm. Our findings appear comparable to previous studies that defined delivery latency from admission or PPROM. We defined latency from TVCL, which was performed within 72 hours after admission. We chose to use this time period to control for the variable amount of time reported from PPROM to admission and time to the ascertainment of TVCL by a qualified sonographer. Of our cases, 64% had a TVCL assessed within 24 hours of PPROM and 97% within 3 days of PPROM. We found no significant difference using the measure of latency from PPROM or from the admission TVCL measurement. This relationship was not specified in previous papers. The strengths of our study include a prospective, well-controlled study design, utilizing multiple statistical approaches. The GA and diagnoses of PPROM were carefully determined.
Biologic mechanisms such as genetic variation and BMI3,25-27 have been proposed as explanations for disparities in PPROM and preterm birth. Other than Tsoi et al,17 none of the other referenced studies provide information on ethnicities or BMI of the study population. Our population consisted of a mixed ethnicity, 55% Caucasian and 42% African American with a median BMI of 29, which differs from the northern and southern European cohort of nonobese, white population and makes our findings more generalizable to a US population. To our knowledge, this is the most comprehensive prospective study in the United States to examine the clinical use of CL in predicting latency in
FIGURE 2
TVCL predicting delivery latency at 7 days in women with AFI >5
AFI, amniotic fluid index; ROC, receiver operating characteristic; TVCL, transvaginal cervical length. Mehra. Cervical length and AFI: can it predict delivery latency PPROM? Am J Obstet Gynecol 2015.
MONTH 2015 American Journal of Obstetrics & Gynecology FLA 5.2.0 DTD YMOB10223_proof 27 January 2015 3:25 pm ce
1.e7
web 4C=FPO
Tsoi et al17 examined 101 singleton pregnancies between 24-36 weeks, excluding women who delivered for medical indications. This northern European population in the United Kingdom and Germany was comprised of 77% Caucasian, 12% Afro-Caribbean, and 11% Asian with a median BMI of 25 kg/m2. Delivery within a week occurred in 57% of those pregnancies. In our population, 48% delivered spontaneously and for clinical indications within the first week. The exclusion of women who delivered for clinical indications and inclusion of a later GA by Tsoi et al17 might explain the difference in rates of delivery within the week. In addition our population was demographically different, representing an urban US population. Similar to our findings, Tsoi et al17 reported a positive relationship of GA at PPROM and uterine contractions with an increased delivery risk. Rizzo et al16 from Italy examined 92 singleton pregnancies with PPROM, and demonstrated that a TVCL 7 days. Based on our findings one may be better able to counsel a patient on her risk of delivery, whether spontaneous or indicated, from the time of the admission TVCL. Potential limitations in our study include patients who were lost to follow-up because they left after their membranes resealed or against medical advice. Our study population included patients with PPROM who underwent spontaneous labor and indicated delivery. By not excluding the patients with indicated delivery we may have limited the generalization of our findings to women with spontaneous labor only. However, a clinician cannot often predict on admission who will ultimately be delivered spontaneously. Although GA at PPROM is an important factor for latency we could not substratify the results beyond the GA cutoff of 30 weeks at PPROM due to inadequate numbers. Adjunctive antibiotics are known to prolong delivery at 7 days in women with PPROM.28,29 As all patients in our study received adjunctive antibiotic therapy we were unable to demonstrate the effects of TVCL and AFI on prolonging latency in the absence of antibiotics. This study supports the predictive ability of TVCL for latency and adds predictive value of longer latency with the combined use of TVCL >2 cm and AFI >5 cm in women with PPROM. We recommend the importance of combining clinical factors into the risk assessment of an individual with PPROM. Based on our study we recommend obtaining an initial TVCL and AFI in women as soon as practical after admission. Maternal symptoms of uterine contractions and history of PPROM and preterm birth should be carefully assessed in the setting of PPROM to help identify those women at high risk of shorter latency. There still remains a need to evaluate multiple gestations, previable PPROM patients, and combining ultrasound findings
with biochemical markers to improve prediction of latency. We recommend further research to verify our results and generate a clinically useful prediction model for latency using multiple parameters in women with PPROM at various GA substrata. REFERENCES 1. Mercer BM. Preterm premature rupture of the membranes: current approaches to evaluation and management. Obstet Gynecol Clin North Am 2005;32:411-28. 2. Lee T, Silver H. Etiology and epidemiology of preterm premature rupture of the membranes. Clin Perinatol 2001;28:721-34. 3. Premature rupture of membranes. Practice bulletin no. 139. Obstet Gynecol 2013;122: 918-30. 4. Martin JA, Hamilton BE, Ventura SJ, et al. Births: final data for 2009. Natl Vital Stat Rep 2011;60:1-70. 5. Cunningham F, Leveno K, Bloom S, Spong CY, Dashe J. Williams obstetrics, 24th ed. New York, NY: McGraw-Hill Professional; 2014. 6. Vijgen SM, Van der Ham DP, Bijlenga D, et al. Economic analysis comparing induction of labor and expectant management in women with preterm prelabor rupture of membranes between 34 and 37 weeks (PPROMEXIL trial). Acta Obstet Gynecol Scand 2014;93:374-81. 7. Magee L, Sawchuck D, Synnes A, et al. Magnesium sulphate for fetal neuroprotection: SOGC clinical practice guideline. J Obstet Gynaecol Can 2011;33:516-29. 8. Conde-Agudelo A, Romero R, Hassan SS, et al. Transvaginal sonographic cervical length for the prediction of spontaneous preterm birth in twin pregnancies: a systematic review and metaanalysis. Am J Obstet Gynecol 2010;203: 128.e1-12. 9. Iams JD, Goldenberg RL, Meis PJ, et al. The length of the cervix and the risk of spontaneous premature delivery: National Institute of Child Health and Human Development Maternal Fetal Medicine Unit Network. N Engl J Med 1996;334: 567-72. 10. Hassan SS, Romero R, Berry SM, et al. Patients with an ultrasonographic cervical length < or ¼15 mm have nearly a 50% risk of early spontaneous preterm delivery. Am J Obstet Gynecol 2000;182:1458-67. 11. Crane JM, Hutchens D. Transvaginal sonographic measurement of cervical length to predict preterm birth in asymptomatic women at increased risk: a systematic review. Ultrasound Obstet Gynecol 2008;31:579-87. 12. Carlan SJ, Richmond LB, O’Brien WF. Randomized trial of endovaginal ultrasound in preterm premature rupture of membranes. Obstet Gynecol 1997;89:458-61. 13. Gire C, Faggianelli P, Nicaise C, et al. Ultrasonographic evaluation of cervical length in
pregnancies complicated by preterm premature rupture of membranes. Ultrasound Obstet Gynecol 2002;19:565-9. 14. To MS, Skentou C, Cicero S, et al. Cervical assessment at the routine 23-weeks’ scan: problems with transabdominal sonography. Ultrasound Obstet Gynecol 2000;15:292-6. 15. Fischer RL, Austin JD. Cervical length measurement by translabial sonography in women with preterm premature rupture of membranes: can it be used to predict the latency period or peripartum maternal infection? J Matern Fetal Neonatal Med 2008;21: 105-9. 16. Rizzo G, Capponi A, Angelini E, et al. The value of transvaginal ultrasonographic examination of the uterine cervix in predicting preterm delivery in patients with preterm premature rupture of membranes. Ultrasound Obstet Gynecol 1998;11:23-9. 17. Tsoi E, Fuchs I, Henrich W, et al. Sonographic measurement of cervical length in preterm prelabor amniorrhexis. Ultrasound Obstet Gynecol 2004;24:550-3. 18. Mercer BM, Rabello YA, Thurnau GR, et al. The NICHD-MFMU antibiotic treatment of preterm PROM study: impact of initial amniotic fluid volume on pregnancy outcome. Am J Obstet Gynecol 2006;194:438-45. 19. Goya M, Bernabeu A, Garcia N, et al. Premature rupture of membranes before 34 weeks managed expectantly: maternal and perinatal outcomes in singletons. J Matern Fetal Neonatal Med 2013;26:290-3. 20. Melamed N, Hadar E, Ben-Haroush A, et al. Factors affecting the duration of the latency period in preterm premature rupture of membranes. J Matern Fetal Neonatal Med 2009;22: 1051-6. 21. Piazze J, Anceschi M, Cerekja A, et al. Validity of amniotic fluid index in preterm rupture of membranes. J Perinat Med 2007;35:394-8. 22. Perinatal Quality Foundation. Clear guidelines for cervical length. Available at: https:// clear.perinatalquality.org. Accessed March 20, 2013. 23. Doyle LW, Crowther CA, Middleton P, et al. Magnesium sulphate for women at risk of preterm birth for neuroprotection of the fetus. Cochrane Database Syst Rev 2009;1: CD004661. 24. Goldenberg RL, Nelson KG, Davis RO, et al. Delay in delivery: influence of gestational age and the duration of delay on perinatal outcome. Obstet Gynecol 1984;64:480-4. 25. Mercer BM, Goldenberg RL, Meis PJ, et al. The Preterm Prediction Study: prediction of preterm premature rupture of membranes through clinical findings and ancillary testing; the National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network. Am J Obstet Gynecol 2000;183: 738-45. 26. Creasy RK, Resnik R, Iams JD, et al. Creasy and Resnik’s maternal-fetal medicine: principles and practice, 7th ed. Philadelphia, PA: Saunders; 2013.
1.e8 American Journal of Obstetrics & Gynecology MONTH 2015 FLA 5.2.0 DTD YMOB10223_proof 27 January 2015 3:25 pm ce
CAOG Papers
ajog.org 27. Guinn DA, Goldenberg RL, Hauth JC, et al. Risk factors for the development of preterm premature rupture of the membranes after arrest of preterm labor. Am J Obstet Gynecol 1995;173:1310-5.
28. Kenyon SL, Taylor DJ, Tarnow-Mordi W, et al. Broad-spectrum antibiotics for preterm, prelabor rupture of fetal membranes: the ORACLE I randomized trial; ORACLE Collaborative Group. Lancet 2001;357:979-88.
29. Amon E, Lewis SV, Sibai BM, et al. Ampicillin prophylaxis in preterm premature rupture of the membranes: a prospective randomized study. Am J Obstet Gynecol 1988;159:539-43.
MONTH 2015 American Journal of Obstetrics & Gynecology FLA 5.2.0 DTD YMOB10223_proof 27 January 2015 3:25 pm ce
1.e9
