Correspondence: Professor S. Meuris, Research Laboratory on Reproduction, Faculté de Médecine (CPi 626), Université Libre de Bruxelles, Route de Lennik 808, B-1070 Brussels, Belgium.
Objective To assess external pelvimetry and maternal height, as predictors of cephalopelvic disproportion.
Design Prospective cohort study.
Setting Four hospitals in Zaire.
Population Six hundred and five nulliparous women.
Methods Maternal height and external pelvimetry were assessed during the third trimester antenatal visit. Cut off values for considering women at risk for cephalopelvic disproportion were height < 150 cm and external pelvic distances < 10th centile for the population. Logistic regression analysis, combining height and pelvic measurements, was performed to predict women at risk for cephalopelvic disproportion.
Main outcome measure Cephalopelvic disproportion was considered when there was caesarean section for failure to progress, vacuum or forceps delivery or intrapartum stillbirth.
Results Cephalopelvic disproportion was present in 42 women. In univariate analysis, height, intertrochanteric diameter and the transverse diagonal of Michaelis sacral rhomboid area were found to be associated with cephalopelvic disproportion. Logistic regression analysis showed that maternal height < 150 cm and/or transverse diagonal < 9.5 cm were the variables most associated with cephalopelvic disproportion. The adjusted odds ratios were 2.2 (95% CI 0.9 to 5.4) and 6.5 (95% CI 3.2 to 13.2), respectively. The positive predictive value and likelihood ratio were 24% and 4.0 (95% CI 2.8 to 5.8), respectively. The addition of transverse diagonal to maternal height increased the sensitivity in predicting cephalopelvic disproportion from 21% to 52%.
Conclusion In addition to height, transverse diagonal measurement is able to predict one out of two cases of cephalopelvic disproportion in nulliparous women. After validation in a separate cohort, this simple predictive method may be used in peripheral centres for timely referral of pregnant women at risk for cephalopelvic disproportion.
Maternal mortality remains a major concern in developing countries. Recent estimates suggest that about 600,000 women die each year of pregnancy related complications1. Among these deaths, 20% to 30% are attributable to the complications of cephalopelvic disproportion1–4. Those who survive are at significant risk of morbidity, including postpartum haemorrhage, perineal tears, genital prolapse and obstetric fistulae3–5. Risks for the fetus include asphyxia, septicaemia, neurological damage and perinatal death6.
In the majority of cases of cephalopelvic disproportion, caesarean section represents the best treatment for both the mother and the fetus. Detection of women at risk for cephalopelvic disproportion is therefore one of the main goals of antenatal clinics. This is particularly crucial in settings where caesarean section is not feasible, so that women at high risk for cephalopelvic disproportion may be referred to a district hospital equipped with an operating theatre prior to the onset of labour. At present, most antenatal care programs rely only on maternal height to identify women at risk for cephalopelvic disproportion7–10. Maternal height has been shown, however, to have limited predictive value8. External pelvimetry was the first technique used to predict cephalopelvic disproportion1,13. Few studies on pelvimetry are available and the prediction of cephalopelvic disproportion by this method was not adequately assessed14–20. The objective of our cohort study was to evaluate external pelvimetry, in addition to maternal height, to predict cephalopelvic disproportion in nulliparous women.
Four obstetricians from the Maternity Hospital of the Kinshasa University were trained to measure external pelvic distances. They were posted for a six-month period between 1986 and 1989 in one of four hospitals of the former Republic of Zaire (Kinshasa, Boende, Mbuji-Mayi and Mambasa). These hospitals served mixed urban and rural Bantu populations. Routinely, pregnant women have a single prenatal visit during the third trimester. All nulliparous women (n= 646) presenting during the period that the visiting obstetrician was there were included in the study. More than 95% of included women were delivered in the same hospital. Women who were limping or who had an obvious asymmetrical pelvis were excluded from the study. Nulliparous women who were delivered of twins (n= 8), a fetus weighing < 2000 g (n= 30) or a non-vertex presentation (n= 3) were also excluded. The analysis included 605 nulliparous women who were delivered of a single fetus weighing ≥ 2000 g in vertex presentation.
Maternal height and external pelvic measurements were performed by visiting obstetricians during the antenatal visit. Height was measured using a height gauge with the woman standing erect in bare feet. The antero-posterior diameter (also named external conjugate18 or Baudelocque diameter), the distance between the anterior superior iliac spines (intercrestal), between the anterior inferior iliac spines (interspinous), between the femoral trochanters (intertrochanteric) and the intertuberous diameter were measured using a Breisky pelvimeter (Fig. 1). The vertical and transverse diagonals of the Michaelis sacral rhomboid area were measured using a measuring tape (Fig. 1). Height and pelvic measurements were recorded to the nearest 0.5 cm interval. Results of height and pelvic measurements were recorded in a file separate from the antenatal record. Pelvic measurements were not used for decision making and the visiting obstetricians who did the pelvimetry were not involved in the management of delivery.
Information on delivery was recorded by the local medical staff in the obstetric register and collected everyday by the visiting obstetrician. Cephalopelvic disproportion was considered under the following conditions: caesarean section for failure to progress, vacuum or forceps delivery or vaginal delivery complicated by intrapartum stillbirth. All uncomplicated vertex births were considered as the reference group. Another group comprising complicated deliveries (‘other complicated deliveries’ group) included women who had caesarean section for fetal distress, placenta praevia or insufficient uterine contractions. These conditions were not considered as related to cephalopelvic disproportion, because a trial of labour could not be completed.
The mean (SD) were compared using analysis of variance (ANOVA), followed by a Scheffe test. As usually accepted, a maternal height < 150 cm was used as cut off to identify women at risk for cephalopelvic disproportion7–10. Cut off values for pelvic distances were defined as the values closest to the 10th centile of our population. Sensitivity, specificity, positive predictive values and the positive likelihood ratio [sensitivity divided by (1-specificity)] with their 95% confidence intervals (CI) were calculated using these thresholds. Logistic regression models (forced entry method) were constructed, with cephalopelvic disproportion as a dependent variable, to assess the independent predictive properties of various combinations between maternal height and pelvic measurements.
The study protocol was approved by the authorities of the hospitals involved in the study and by academic authorities of the Kinshasa University.
Among included women, the proportion of deliveries complicated by cephalopelvic disproportion, as defined above, was 7.0% (42/605). Among these deliveries, there were 26 caesarean section (including two stillbirths), nine vacuum or forceps (including one stillbirth) and seven vaginal deliveries complicated by intrapartum stillbirth. The proportion of women with cephalopelvic disproportion did not differ between the four maternity hospitals. The proportion of other complicated deliveries was 2.5% (15/605). There were no maternal deaths in the study group.
Mean maternal age did not differ between the cephalopelvic disproportion and normal delivery (n= 548) groups (Table 1). Maternal height and most of the pelvic measurements were smaller in the cephalopelvic disproportion group than in the normal delivery group. The intertuberous diameter was similar between the two groups. Maternal measurements were similar between the ‘other complicated deliveries’ group and the normal delivery group (Table 1).
Table 1. Comparison of maternal characteristics between groups. Values are given as means (SD).
Normal delivery (n= 548)
Cephalopelvic disproportion (n= 42)
Other complicated deliveries(n= 15)
*Computed by ANOVA.
†Significantly different from the normal delivery group at the 0.05 level (Scheffe test).
External pelvic dimensions (cm)
Michaelis sacral rhomboid area (cm)
The proportion of women with height < 150 cm was 7.3% (Table 2). The proportion of women with pelvic measurements below the chosen cut off levels ranged from 8.1% to 12.5%. All anthropometric measurements displayed high specificity ranging from 89.1% to 93.8%. In univariate analysis, maternal height, intertrochanteric diameter and the transverse diagonal of the Michaelis sacral rhomboid area had the highest sensitivity, positive predictive value and positive likelihood ratio (Table 2).
Table 2. Prediction of cephalopelvic disproportion by maternal height and external pelvimetry: univariate analysis. Values are given as %(n/ntotal), unless otherwise indicated.
Cut off value at risk
Women at risk
Positive predictive value
Positive likelihood ratio (95% CI)
External pelvic dimensions (cm)
Michaelis sacral rhomboid area
These predictors were included in four logistic regression models (Table 3). According to the predictive properties derived from these models, women at highest risk for cephalopelvic disproportion were those with a height < 150 cm and/or a transverse diagonal of Michaelis area < 9.5 cm (model 2 in Table 3). The adjusted odds ratios of maternal height and transverse diagonal were 2.2 (95% CI 0.9 to 5.4, P= 0.089) and 6.5 (95% CI 3.2 to 13.2, P < 0.001), respectively. The positive predictive value and likelihood ratio of this model were 24% (95% CI 16 to 34) and 4.0 (95% CI 2.8 to 5.8), respectively. The sensitivity of the two combined tests to identify women who subsequently had a delivery complicated by cephalopelvic disproportion was 52% (95% CI 37 to 68). To achieve a similar sensitivity (55%) using maternal height alone, a cut off level of 159 cm should be used. This would result in a referral rate of 31% (190/605) and a positive predictive value of 13%.
Table 3. Logistic regression analysis: prediction of cephalopelvic disproportion by models combining maternal height (HT), intertrochantericdiameter (IT) and transverse diagonal of the Michaelis area (TD). Values are given as % (n/ntotal), unless otherwise indicated.
Logistic regression models
Women at risk
positive predictive value
Positive likelihood likelihoodratio (95% CT)
Model 1: (HT < 150 cm) or (IT < 27.5 cm)
Model 2: (HT < 150 cm) or (TD < 9.5 cm)
Model 3: (IT < 27.5 cm) or (TD < 9.5 cm)
Model 4: (HT < 150 cm>or (IT < 27.5 cm) or (TD < 9.5 cm)
Mean (SD) birthweight was higher in the cephalopelvic disproportion group [3129 g (425)] than in normal delivery group [2976 g (395); P= 0.015]. Among women identified as being at risk by the model, those having cephalopelvic disproportion were delivered of larger babies [true positives; n= 22; mean (SD) birthweight 3019 g (369)] than women who had an uncomplicated vaginal delivery [false positives; n= 71; mean (SD) birthweight 2840 g (396); P= 0.06]. Birthweight was also higher in women not identified by the model but having cephalopelvic disproportion [false negatives; n= 20; mean (SD) birthweight 3251 g (457)] than among women with normal delivery [true negatives; n= 477; mean (SD) birthweight 2996 g (391); P= 0.004].
In order to evaluate the fetal component of cephalopelvic disproportion on the predictive properties of the proposed model, birthweight was added to the model. The adjusted odds ratios for maternal height, the transverse diagonal of Michaelis area and birthweight ≥ 3500 g (90th centile) were 2.1 (95% CI 0.8 to 5.2; P= 0.112), 7.2 (95% CI 3.5 to 14.9; P < 0.001) and 2.7 (95% CI 1.2 to 6.3, P= 0.019), respectively. The sensitivity, positive predictive value and positive likelihood ratio were 67% (95% CI 50 to 80), 18% (95% CI 13 to 25) and 2.9 (95% CI 2.2 to 3.7), respectively.
Cephalopelvic disproportion is a major risk factor for maternal and perinatal morbidity and mortality. Accurate prediction of women at risk for cephalopelvic cephalopelvic disproportion is not predicted by health centres which are not equipped to perform a caesarean section, long referral distances and poor local transport may lead to obstructed labour and uterine rupture2–4. Conversely, in a resource limited setting, prediction of cephalopelvic disproportion in women at risk must be sufficiently specific to avoid unnecessary referral.
In the present study the proportion of cephalopelvic disproportion in nulliparous women was in the range of 4% to 15%, reported in developed and developing countries6. Variations in the proportion of cephalopelvic disproportion are due to genetic factors, nutritional factors and differences in the definition of cephalopelvic disproportion6,14,18–20. The complications included in our definition of cephalopelvic disproportion are likely to be caused by an absolute or relative mechanical disparity between the fetal size and the birth canal22–24.
Women's height is correlated to pelvic size and is currently used to predict cephalopelvic disproportion7–10. In our study women who had cephalopelvic disproportion were shorter than those having a normal delivery. However, the sensitivity related to a maternal height cut off for risk set at < 150 cm was low. Our results confirm previous studies showing the limitations of maternal height to predict cephalopelvic disproportion7,8.
Pelvic measurements, performed either by external pelvimetry or by X-ray techniques, can provide markers of the risk for cephalopelvic disproportion. External pelvic measurements have been found to be correlated with internal pelvic measurements by X-ray25,26. In our study, the majority of external pelvic measurements were smaller among women having cephalopelvic disproportion than among women with uncomplicated vaginal delivery. The antero-posterior diameter (Baudelocque diameter) is the only external distance associated with the pelvic anterior posterior size6,18,26. Its sensitivity, positive predictive value and likelihood ratio remained lower than those of maternal height. The intercrestal, interspinous and intertrochanteric diameters can be used to determine pelvic shape and transverse capacity6,18,19–26. Among them, the intertrochanteric diameter was the best predictor of cephalopelvic disproportion. The intertuberous diameter was not related to cephalopelvic disproportion. This distance, nevertheless, is usually described as related to outlet dystocia15,16
Abnormal size and shape of the Michaelis sacral rhomboid area, first described in 1851, was reported to be associated with abnormal pelvis12,13. Our study shows that the transverse diagonal of this sacral area was the strongest anthropometric predictor for cephalopelvic disproportion. Moreover, the clinical algorithm obtained by the addition of this risk factor to the maternal height was found to be the best model to identify women at risk for cephalopelvic disproportion. This model identified more than half of the cases of cephalopelvic disproportion among nulliparous women, with a proportion of women presumed to be at risk of 16%. It is important to note that in order to achieve a similar sensitivity using only maternal height, a cut off value of 159 cm should be used. Using this cut off value will lead to referral of 31% of nulliparous women, which is not acceptable in most settings. The proposed model greatly reduced the referral rate to a limit that may be more suitable in a resource limited setting. However, a lower cut off may be chosen (e.g. 5th centile for both measures) to minimise the number of women referred.
Large babies are at higher risk of dystocia6. In our study, mean birthweight was higher in the cephalopelvic disproportion group than in the group having a normal delivery. Babies born to women who had normal height and transverse diagonal and who subsequently had a cephalopelvic disproportion (false negatives) were larger. Conversely, babies born to false positive women were smaller and thus the mothers might have been able to deliver normally despite having a smaller pelvis. After addition of birthweight in the logistic regression model, the association between transverse diagonal and cephalopelvic disproportion remained high. Consequently, as birthweight and fundal height are correlated27, it may be of great interest to assess the prediction of cephalopelvic disproportion by fundal height and/or abdominal circumference measurement, in addition to external pelvimetry and maternal height.
The results of our study should be validated prospectively in a separate cohort before implementing them in routine practice. Acceptability of referral based on the proposed model must also be evaluated. The advantage of the proposed measurements are that only measurements that can be performed easily during a routine antenatal visit by health workers, using a simple and widely available measuring tape, are needed.
In conclusion measurements of maternal height and the transverse diagonal of the Michaelis sacral rhomboid area using a measuring tape may represent a simple method to detect nulliparous women at risk for cephalopelvic disproportion. This method, after being validated in other populations, may be very useful in peripheral antenatal clinics to identify pregnant women at risk for cephalopelvic disproportion and to refer them to district hospitals before the onset of labour.
The authors would like to thank Drs J. E. Mboloko, J. K. Kizonde, and D. Nk. Ndidwa, obstetricians at the maternity hospital of the University of Kinshasa. Local medical staffs at Boende, Mbuji-Mayi and Mambasa are greatly thanked for recruiting the participants. They would also like to thank Drs D. Walker and S. Ako for their help in editing the manuscript. Mr S. Meuris is a research director from the Belgium National Fund for Scientific Research.