Journal of Obstetrics and Gynaecology, May 2006; 26(4): 311 – 316

A way to lend objectivity to Bishop score

M. R. M. ELGHORORI, I. HASSAN, W. DARTEY, & E. ABDEL-AZIZ

Department of Obstetrics and Gynaecology, Norfolk and Norwich University Hospital, Norwich, Norfolk, UK

Summary
The aim of this study was to explore the possibility of utilising pre-induction cervical length assessment by trans-vaginal
ultrasound to improve the predictive value of the Bishop score. The idea of this paper has evolved following our prospective
study, which was designed to compare the pre-induction objective assessment of the cervix by ultrasound with the subjective
J Obstet Gynaecol Downloaded from informahealthcare.com by University of Toronto on 03/06/15

one by Bishop Score. The Bishop scores of the 104 women included in the study were modified by replacing the digital
assessment of the cervical length by ultrasound cervical length measurements. There was a significant statistical difference
(p 5 0.0001) between the median of the original and the modified Bishop scores. The original Bishop score showed
insignificant association (p 4 0.05) with the induction-delivery interval (IDI) and the mode of delivery while the modified
score showed a significant association (r ¼ 0.31, p 5 0.05) with mode of delivery and a highly significant one (r ¼ 0.55,
p 5 0.0001) with the IDI. The receiver operating characteristic curve showed that the optimised cut-off value for prediction
of vaginal delivery was 45 for the original Bishop Score and 43 for the modified one. At those optimised cut-off values, the
original Bishop Score predicted vaginal delivery with a sensitivity of 23% (95% CI; 14.6%, 33.2%) and specificity of 88.2%
(95% CI; 63.5%, 98.5% while the modified Bishop score predicted vaginal delivery with a sensitivity of 62% [95%; CI 51 –
72.3] and specificity of 82% [95%; CI 56.6 – 96]). In conclusion, the modified Bishop score is better than the original one in
predicting the IDI and the success of induction of labour. The sensitivity of the Bishop score in predicting the rate of vaginal
delivery has been improved significantly following the modification.
For personal use only.

measurements of the cervical length (Andersen 1991;

Introduction Andersen et al. 1990). Several reports have suggested that
Induction of labour is carried out in approximately 20% of ultrasound offers more reliable assessment for prediction of
all pregnancies (RCOG 2001). There is good evidence to successful labour induction (Pandis et al. 2001; Rane et al.
support the view that induction of labour in prolonged 2003). The main objective of this study was to explore the
pregnancy is associated with a significant reduction in possibility of improving the predictive value of the Bishop
perinatal mortality (Cole et al. 1975; Sande et al. 1983; Sue- score by utilising trans-vaginal ultrasound.
A-Quan 1999). Therefore, the most common indication for
induction of labour is post-term pregnancy. It has been
reported that 20 – 24% of induced labour in women with an Methods
unfavourable cervix result in delivery by caesarean section
Between 2001 and 2003, a prospective study was under-
(Tam et al. 1999; Arulkumaran et al. 1985; Crowley 2001)
taken at Queen Elizabeth Hospital, King’s Lynn to
Methods which identify those women are necessary when
compare the Bishop score with the pre-induction ultra-
prompt delivery is needed in high-risk pregnancies. Pre-
sound cervical length assessment. The main outcome of
induction cervical assessment is, to some extent, the key to
that study was a marked difference between sensitivities of
the prediction of induction-delivery interval (IDI) and the
the ultrasound cervical length and the Bishop score in
success of induction (Fuentes and Williams 1995; Alder-
predicting the rate of vaginal delivery. In addition, an
man 1975; Lange et al. 1982). Since the introduction of the
obvious discrepancy between the digital and the ultrasound
Bishop score (Bishop 1964) to clinical practice, it has been
cervical length assessment was noticed. Therefore, this
the most popular and traditional method of assessing the
study was designed to investigate the likelihood of
favourability of the cervix in women undergoing induction
improvement of the predictive values of the Bishop score
of labour. The five components of the Bishop score depend
by utilising the pre-induction cervical length assessment by
solely on the digital assessment of the cervix and the level of
ultrasound.
the presenting part. Unfortunately the digital assessment is
The study included the 104 pregnant women who
highly subjective with a wide range of intra- and inter-
were recruited in the original research. Those women
observer variability; therefore, the predictive value of the
were booked to have induction of labour at term. Inclu-
Bishop score is poor (Holcomb and Smeltzer 1991; Phelps
sion criteria were: gestational age between 37 and 42
et al. 1995; Goldberg et al. 1997). Over the last decade five
weeks, singleton pregnancy, cephalic presentation and
major scoring systems have been made available but none of
intact membranes. The exclusion criteria were: sponta-
them proved to be much more reliable than the Bishop score
neous rupture of membranes, multiple pregnancies, vagi-
(Hughey et al. 1976). In contrast, trans-vaginal ultrasono-
nal bleeding, previous caesarean section and abnormal
graphy provides more precise, objective and reproducible
presentation.

Correspondence: M. R. M Elghorori, 45 Elvington, Springwood, Kings Lynn, Norfolk PE30 4TB, UK. E-mail: rabeih@elghorori.fsnet.co.uk
ISSN 0144-3615 print/ISSN 1364-6893 online Ó 2006 Taylor & Francis
DOI: 10.1080/01443610600594922
 312 M. R. M. Elghorori et al.
Vaginal ultrasound examination of the cervix was
explained to the patient and verbal consent was obtained.
All ultrasound examinations were carried out before
induction by one qualified radiographer who was blinded
to the digital cervical assessment and the patient’s history.
An ALOKA ProSound SSD-4000 scanner with a 5.0-MHz
transducer was used for cervical measurements. The
examination was performed with an empty bladder to
avoid elongation of the cervix. The patient was placed in
the lithotomy position and a trans-vaginal probe was placed
in the vagina approximately 3 cm away from the cervix to
avoid any distortion of cervical position or shape. A midline
sagittal view of the cervix, with the echogenic endocervical
mucosa along the length of the canal, was obtained.
Callipers were used to measure the distance between the
internal and external os (Andersen et al. 1990; Sonek et al.
1990).
J Obstet Gynaecol Downloaded from informahealthcare.com by University of Toronto on 03/06/15
The pre-induction Bishop score of each patient was
obtained from the case notes, retrospectively. The Bishop
score was assessed either by a midwife or an on call
obstetrician who were blinded to the results of ultrasound
examination. Unlike ultrasound cervical assessment, the
digital assessment was performed by different operators to
replicate routine practice.
The original Bishop score of each patient was modified
by replacing the digital assessment of the cervical length by
ultrasound cervical length measurements. The mode of
delivery and the IDI were considered to be the targets to
For personal use only.
compare the original and modified Bishop scores.
Induction of labour was performed according to NICE
(National Institute of Clinical Excellence 2001) guidelines
(Inherited Clinical Guidelines D). Women with a Bishop
score of
8 had artificial rupture of membranes. An
oxytocin infusion was started following amniotomy or in
cases of poor progress for labour. Women with an
unfavourable cervix (Bishop score of  4) received 3 mg
Prostin E2 tablet (Dinoprostone, Pharmacia & Upjohn,
Milton Keynes, UK) vaginally. Further vaginal examina-
tions at 6-h intervals were organised for those women to
assess the Bishop score. Depending on the score, another
3 mg of Dinoprostone was administered. The maximum
dose of Dinoprostone used was 6 mg over 24 h.
Statistical analysis was carried out using the MedCalc
software package (Schoonjans 2004).
Regression analysis was used to investigate how well the
Bishop score, the modified score and cervical length by
ultrasound predicted the mode of delivery and IDI. A
Kaplan – Meier survival curve was used to predict any
association between IDI and both the Bishop score and the
ultrasound cervical length. Receiver operating character-
istic (ROC) curves of the mode of delivery for the two
methods were also studied as well as those of the original
and the modified Bishop scores. Differences in the IDI and
the mode of delivery for parity were investigated using
Mann-Whitney U or Median tests where appropriate.
Results
A total of 104 women were recruited into this study; 54
(51.9%) were nulliparous and 50 (48.1%) were multi-
parous. The mean gestational age at induction was 41
weeks (range 37 – 42 weeks). The main indication for
induction was post-term in 86 (82.7%) women. The other
indications were pre-eclampsia in five (4.8%); oligohy-
dramnios in four (3.8%); small for gestational age fetus in
two (1.9%); maternal cholestasis in two (1.9%); gestational
diabetes mellitus in two (1.9%); reduced fetal movements
in one (1%); epilepsy in one (1%) and maternal request
due to social reasons in one (1%). Demographic char-
acteristics are demonstrated in Table I.
A total of 84 women (80.8 %) had induction by Prostin
E2 vaginal tablets and 20 (19.2%) by artificial rupture of
membranes and oxytocin augmentation; 87 women
(83.7%) delivered vaginally and 17 (16.3%) by caesarean
section. The median cervical length by trans-vaginal
ultrasound was 3.4 cm (interquartile range, IQR 2.4, 4).
The median of the original Bishop Score was 5 (IQR 4, 5)
and the median of the modified Bishop score was 4 (IQR 3,
4). There was a significant statistical difference Z ¼ 4.93
(p 5 0.0001) between the median of the original and the
modified Bishop scores. A highly significant association
(p 5 0.0001) was detected between the ultrasound cervical
length and both the IDI and the mode of delivery (r ¼ 0.87,
0.44, respectively). Although the original Bishop score
failed to show any significant association (p 40.05) with
the IDI and the mode of delivery, the modified score
showed a significant association (r ¼ 0.31, p 5 0.05) with
mode of delivery and highly significant one (r ¼ 0.55,
p 5 0.0001) with the IDI.
Regression analysis of Bishop Score, modified Bishop
score and cervical length with IDI showed that the
ultrasound cervical length was the best predictor of the
IDI followed by the modified Bishop score compared with
original Bishop score, which was a poor predictor of the
IDI. The adjusted R2 value for the model containing the
original Bishop score was 0.001 (p 40.05) compared with
the R2 value of 0.24 (p 5 0.05) for the model containing
modified Bishop Score and R2 value of 0.86 (p 5 0.001) for
the model containing ultrasound cervical length.
In addition, survival analysis of the IDI and ultrasound
cervical length (Figure 1) has shown the likelihood of
shorter IDI occurred with a shorter cervical length. There
was a highly significant (p 5 0.0001) association between
the IDI and the ultrasound cervical length. The median
IDI with cervical length  2.5 cm was 6.4 h (95%; CI 5.4,
6.7 h), while it was 11 h (95%; CI 10.3, 11.7 h) when
cervical length was between 2.6 – 3.5 cm and for cervical
length 43.5 cm, the median IDI was 22.2 h (95%; CI
20.6, 23.9 h). Similarly, a significant association
(p 5 0.001) was detected between the modified Bishop
score and the IDI. Kaplan – Meier survival curve of the IDI
and the modified Bishop score (Figure 2) showed that the
median IDI for those with a Bishop scores 0 – 2 was 21 h
(95%; CI 14.4, 27.6 h), for Bishop scores 3 – 4 the median
IDI was 15.9 h (95%; CI 10.6, 21.2 h), for Bishop score
5 – 6 the median IDI was 7 h (95%; CI 6.7, 7.4 h) and for
Bishop scores 46 the median IDI was 4.4 h (95%; CI 3.5,
5.3 h). In contrast, survival analysis of the original Bishop
score and the IDI showed an insignificant association
(p 40.05).
Table I. Demographic details of the studied women (n ¼ 104)
Age (years): mean and (range) 26.5 (14 – 37)
Nulliparous: n (%) 54 (52)
Multiparous: n (%) 50 (48)
Gestational age at induction (weeks): 41 þ 1 day (37 – 42)
mean and (range)
 Bishop score 313
J Obstet Gynaecol Downloaded from informahealthcare.com by University of Toronto on 03/06/15
For personal use only.

Figure 1. Kaplan – Meier survival curve between the duration of labour and cervical length.

Figure 2. Kaplan – Meier survival curve between the duration of labour and modified Bishop score.
 314 M. R. M. Elghorori et al.
J Obstet Gynaecol Downloaded from informahealthcare.com by University of Toronto on 03/06/15
Figure 3. ROC curves of the Bishop score, modified Bishop score and ultrasound cervical length.
For personal use only.
Receiver operating characteristics curves (ROC) were
constructed (Figure 3) to work out the best cut off values of
Bishop score, modified Bishop score and cervical length
for prediction of successful induction. The curve for the
Bishop score showed an optimised cut-off value of 45,
corresponding to sensitivity of 23% (95%; CI 14.6 – 33.2)
and specificity of 88% (95%; CI 63.6 – 98.5). The curve for
modified Bishop score showed an optimised cut-off value
of 43, corresponding to sensitivity of 62% (95%; CI 51 –
72.3) and specificity of 82% (95%; CI 56.6 – 96). The
curve for the cervical length showed an optimised cut-off
value of  3.4 cm, corresponding to sensitivity of 62%
(95%; CI 51.0 – 72.3) and specificity of 100% (95%; CI
80.5 – 100). Area under the ROC curve for Bishop
Score ¼ 0.5 (95%; CI: 0.36 – 0.63), while for cervical
length by ultrasound ¼ 0.84 (95%; CI 0.77 – 0.92). The
difference between areas was 0.34 (95%; CI: 0.15 – 0.54)
with significance level of (p 5 0.001). Area under the ROC
curve for the modified Bishop score ¼ 0.74 (95%; CI
0.64 – 0.82). The difference between areas under the curves
of Bishop and the modified Bishop score was 0.24 (95%;
CI: 0.12 – 0.36) with significance level of (p 5 0.001).
However, there was insignificant difference (p 40.05)
between areas under the curves of the modified Bishop
score and the ultrasound cervical length was 0.11 (95%;
CI: 0.03 – 0.25).
Discussion
Although modern technology has provided obstetricians
with a tool to visualise the cervix, pre-induction cervical
assessment is still undertaken digitally using the traditional
Bishop score. The five components of the Bishop score rely
on digital assessment which is subjective in nature with a
lack of consistency from one examiner to the next
(Holcomb and Smeltzer 1991; Phelps et al. 1995; Goldberg
et al. 1997; Harrison et al. 1977). However, the use of
numerical scoring of this information has slightly reduced
the source of error (Hughey et al. 1976). Several studies
claim that the Bishop score has a poor predictive value
for the outcome of the induction (Hughey et al. 1976; Dhall
et al. 1987; Friedman et al. 1966). Trials to modify the
original score have failed to improve predictability (Fuentes
and Williams, 1995; Lange et al. 1982; Hughey et al. 1976;
Dhall et al. 1987; Friedman et al. 1966; Krammer et al.
1995; Paterson-Brown et al. 1991). On the other hand,
some recent studies compared Bishop Score and the
ultrasound cervical length and have concluded that ultra-
sound predicts the outcome of induction better than the
Bishop score (Pandis et al. 2001; Rane et al. 2003).
The aim of this study was to explore the possibility of
utilising trans-vaginal ultrasound assessment of the cervix
before induction of labour to improve the predictive value
of the Bishop score. Simply, the idea is to use the
ultrasound cervical measurements to replace the digital
assessment of the cervical length in the original Bishop
score of each patient and then to compare the original and
the modified Bishop scores.
This study has demonstrated that the pre-induction
sonographically measured cervical length is significantly
associated with the IDI and the rate of vaginal delivery;
however, the original Bishop score lacks such an associa-
tion. Similarly, significant association between the Bishop
score and both the IDI and the rate of vaginal delivery has
been achieved following the modification introduced on
the score. Furthermore, the sensitivity of the Bishop
score in predicting the rate of vaginal delivery has
been improved significantly following the modification
and was equal to the sensitivity of the ultrasound cervical
measurement.
In this context, one can argue that ultrasound alone
should be used to assess the cervix before induction rather

than modifying the Bishop score. However, some of the
Bishop score components cannot be assessed by ultrasound
such as the consistency of the cervix and the level of the
presenting part. Although the cervical dilatation and
effacement are the most important components of the
Bishop score (Lange et al. 1982; Watson et al. 1996;
Friedman and Sachtleben 1962), the consistency and the
level of the presenting part remain as important compo-
nents to obstetricians (Crump 1987; Cocks 1955; Harrison
et al. 1977).
Previous trials were made to modify the original Bishop
score aiming to improve the accuracy of the score and to
expand its clinical usefulness (Hughey et al. 1976; Dhall
et al. 1987; Friedman et al. 1967). Modification was
introduced either by taking away from, or adding more
components to, the original Bishop score (Hughey et al.
1976; Cocks 1955). Consequently, several scoring systems
J Obstet Gynaecol Downloaded from informahealthcare.com by University of Toronto on 03/06/15
have evolved but none of them has proved to be useful or
gained popularity (Hughey et al. 1976; Williams et al.
1997). We claim in this study a significant improvement of
the sensitivity of the Bishop score just by replacing the
digital cervical length assessment by the ultrasound one in
the original Bishop score.
Although this idea seems to be simple, it will raise some
debate regarding its practical application which includes
skill, training and workload. Trans-vaginal cervical length
assessment by ultrasound at term can be more difficult
compared with the assessment in the mid-trimester due to
For personal use only.
the distortion of the cervical alignment by the engaged
presenting part (Burger et al. 1997). An inaccurate cervical
length assessment may also occur at the extremes of
cervical length. Excessive cervical effacement may lead to
false identification of the cervical canal, which is an
important landmark, due to mucus or debris filling this
space (Burger et al. 1997; O’Brien et al. 1999). Therefore,
appropriate training is required for those undertaking this
measurement. Introduction of this idea to practice will
raise an important question: would this task be the
responsibility of the radiographers who are already over-
loaded with the dating and the anomaly scans or the
obstetricians who may lack the required expertise in
transvaginal ultrasound scanning. Thus, the Bishop score
will continue to be the traditional method of pre-induction
cervical assessment due to its simplicity.
References
Alderman B. 1975. Factors influencing planned delivery. Clinical
Obstetrics and Gynaecology 2:19 – 47.
Andersen HF, Nugent CE, Wanty SD, Hayashi RH. 1990.
Prediction of risk of preterm delivery by ultrasonographic
measurement of cervical length. American Journal of Obstetrics
and Gynecology 163:859 – 867.
Andersen HF. 1991. Transvaginal and transabdominal ultrasono-
graphy of the uterine cervix during pregnancy. Journal of
Clinical Ultrasound 19:77 – 83.
Arulkumaran S, Gibb DM, Tambyraja RL, Heng SH, Ratnam SS.
1985. Failed induction of labour. Australia and New Zealand
Journal of Obstetrics and Gynaecology 25:190 – 193.
Bishop EM. 1964. Pelvic scoring for elective induction. Obstetrics
and Gynaecology 24:266 – 268.
Burger M, Weber-Rössler T, Willmann M. 1997. Measurement of
the pregnant cervix by transvaginal sonography: an interobserver
study and new standards to improve the interobserver variability.
Ultrasound in Obstetrics and Gynecology 9:188 – 193.
Cocks DP. 1955. Significance of initial condition of cervix uteri to
course of labour. British Medical Journal 1:327 – 328.
Bishop score 315
Cole RA, Howie PW, Macnaughton MC. 1975. Elective induction
of labour. A randomised prospective trial. Lancet 1:767 – 770.
Crowley P. 2001. Interventions for preventing or improving the
outcome of delivery at or beyond term (Cochrane Review). In
The Cochrane Library. Issue 3. Oxford: Update Software.
Crump WJ. 1987. Bishop score and labor duration: A new look.
South Medical Journal 80:1294 – 1295.
Dhall K, Mittal SC, Kumar A. 1987. Evaluation of pre-induction
scoring system. Australia and New Zealand Journal of Obstetrics
and Gynecology 27:309 – 311.
Friedman EA, Niswander KR, Bayonet-Rivera NP, Sachtleben
MR. 1966. Relationship of prelabor evaluation to inducibility
and the course of labor. Obstetrics and Gynecology 28:459 –
501.
Friedman EA, Niswander KR, Bayonet-Rivera NP, Sachtleben
MR. 1967. Prelabor status evaluation II. Weighted score.
Obstetrics and Gynecology 29:539 – 544.
Friedman EA, Sachtleben MR. 1962. Determinant role of initial
cervical dilatation on the course of labor. American Journal of
Obstetrics and Gynecology 84:930 – 935.
Fuentes A, Williams M. 1995. Cervical assessment. Clinical
Obstetrics and Gynaecology 38: 224 – 231.
Goldberg J, Newman RB, Rust PF. 1997. Interobserver reliability
of digital and endovaginal ultrasonographic cervical length
measurements. American Journal of Obstetrics and Gynecology
177:853 – 858.
Harrison RF, Flynn M, Craft I. 1977. Assessment of factors
constituting an ‘inducibility profile’. Obstetrics and Gynecology
49:270 – 274.
Holcomb WL, Smeltzer JS. 1991. Cervical effacement: variation
in belief among clinicians. Obstetrics and Gynecology 78:43 –
45.
Hughey MJ, McElin TW, Bird CC. 1976. An evaluation of
preinduction scoring systems. Obstetrics and Gynecology
48:635 – 641.
Krammer J, Williams MC, Sawai SK, O’Brien WF. 1995. Pre-
induction cervical ripening: A randomised comparison of two
methods. Obstetrics and Gynecology 85:614 – 618.
Lange AP, Secher NJ, Westergaard JG, Skovgard I. 1982. Prelabor
evaluation of inducibility. Obstetrics and Gynaecology 60:137 –
147.
National Institute for Clinical Excellence (NICE). 2001.
Inherited Clinical Guidelines D. Induction of Labour. London:
NICE.
O’Brien JM, Allen AA, Barton JR. 1999. Intravaginal saline as a
contrast agent for cervical sonography in the obstetric patient.
Ultrasound in Obstetrics and Gynecology 13:137 – 139.
Pandis GK, Papageorghiou AT, Ramanathan VG, Thompson
MO, Nicolaides KH. 2001. Pre-induction sonographic mea-
surement of cervical length in the prediction of successful
induction of labor. Ultrasound in Obstetrics and Gynecology
18:623 – 628.
Paterson-Brown S, Fisk NM, Edmonds DK, Rodeck CH. 1991.
Preinduction cervical assessment by Bishop’s score and transva-
ginal ultrasound. European Journal of Obstetrics and
Gynecology and Reproductive Biology 40:17 – 23.
Phelps JY, Higby K, Smyth MH, Ward JA, Arredondo F, Mayer
AR. 1995. Accuracy and intraobserver variability of simulated
cervical dilatation measurements. American Journal of Obste-
trics and Gynecology 173:942 – 945.
Rane SM, Pandis GK, Guirgis RR, Higgins B, Nicolaides KH.
2003. Pre-induction sonographic measurement of cervical
length in prolonged pregnancy: the effect of parity in the
prediction of induction-to-delivery interval. Ultrasound in
Obstetrics and Gynecology 22:40 – 44.
Royal College of Obstetricians and Gynaecologist (RCOG). 2001.
Clinical Support Unit. Induction of labour. In: RCOG,
editors. Evidence-based Clinical Guideline. No. 9. London:
RCOG.
Sande HA, Tuveng J, Fonstelien T. 1983. A prospective
randomised study of induction of labor. International Journal
of Gynaecology and Obstetrics 21:333 – 336.
 316 M. R. M. Elghorori et al.
Schoonjans F. 2004. MedCalc Software for windows. Statistics for
biomedical research. Version 7.3, Belgium.
Sonek JD, Iams JD, Blumenfeld M, Johnson F, Landon M, Gabbe
S. 1990. Measurement of cervical length in pregnancy:
comparison between vaginal ultrasonography and digital exam-
ination. Obstetrics and Gynecology 76:172 – 175.
Sue-A-Quan AK, Hannah ME, Cohen MM, Foster GA, Liston
RM. 1999. Effect of labour induction on rates of stillbirth and
caesarean section in post-term pregnancies. Canadian Medical
Association Journal 160:1145 – 1149.
J Obstet Gynaecol Downloaded from informahealthcare.com by University of Toronto on 03/06/15
For personal use only.
Tam WH, Tai SM, Rogers MS. 1999. Prediction of cervical
response to prostaglandin E2 using fetal fibronectin. Acta
Obstetricia et Gynecologica Scandinavica 78:861 – 865.
Watson WJ, Stevens D, Welter S, Day D. 1996. Factors predicting
successful labor induction. Obstetrics and Gynecology 88:990 –
992.
Williams MC, Krammer J, O’Brien WF. 1997. The value of the
cervical score in predicting successful outcome of labor
induction. Obstetrics and Gynecology 90:784 – 789.