Predicting women at risk for developing obstetric fistula: a fistula index? An observational study comparison of two cohorts




To ascertain if a predictor of obstructed labour and obstetric fistula (OF) occurrence could be devised.


Observational study of two cohorts.


Selian Lutheran Mission Hospital, Arusha, Tanzania and Aberdeen Women's Centre, Freetown, Sierra Leone.


All women presenting with OF caused by obstructed labour and all women having a normal vaginal delivery (NVD) at both institutions were eligible for the study.


All women with OF and those delivering normally had their height in centimetres measured and their intertuberous space measured by the number of examiner's knuckles admitted. The dimensions for OF and normal delivery were examined individually and multiplied to gain a ‘fistula index’.

Main outcomes measures

Dimensions and index were compared statistically between OF and NVD women using t tests. Sensitivity, specificity and predictive values from receiver operating characteristic curves were obtained for predicting OF.


There were statistical differences between the groups with OF women being significantly shorter, having a smaller intertuberous space and lower fistula index than those undergoing NVD (each < 0.001). Sensitivity was high for fistula index ≤ 507.5 (94.9%, 95% confidence interval 83.1–98.6%) and for intertuberous space of at least three knuckles (92.3%, 95% CI 79.7–97.3%) alone.


A simple antenatal measurement of intertuberous space could screen those women at higher risk of needing medical intervention to prevent OF.


Obstetric fistula (OF) is still common in the developing world. Some estimates indicate that there could be 50,000–100,000 new cases per year and up to 2 million women still waiting for treatment.[1, 2]

As seen from the developed world, OF is eminently preventable with timely access to operative delivery in the case of obstructed labour. However, in the developing world there are too few hospitals, health professionals and roads and too little communication and education for all women to have appropriate access to medical attention in labour.

If there were a simple and sensitive screening method to determine which women are at risk of developing an OF, this could help in the battle to reduce the rate of injury. Theoretically high-risk women could be screened at the village level by a birth attendant and referred to either a maternity waiting area (if present) or early in labour to a hospital (again if present) to deliver their child.

It has been known for some time that many ‘at-risk’ women will deliver normally and many ‘low-risk’ women will suffer complications. Making any sort of reliable predictor of labour outcome elusive.

It has long been known that women with OF are shorter than the general population, usually < 150 cm.[3-6] They are shorter by around 7 cm from women delivering normally.[7]

Fistula surgeons are also aware that fistula sufferers have very narrow pelvic outlets, with the angle of the pubic arch very acute, not only making delivery difficult, but also making access for a vaginal fistula repair challenging.

Clinical pelvimetry has long fallen by the wayside, replaced by current teaching recommending the trial of a pelvis by labour. If there are some factors reflecting a severely contracted pelvis that might lead to obstructed labour and OF formation, they could potentially be used as a screening tool.

Knowing that women with OF are shorter than women delivering normally and knowing anecdotally that many fistula sufferers have very narrow pubic arches, an attempt to combine these measurements was made by multiplying the two values together in an attempt to potentiate any predictive value, creating a ‘fistula index’ to see if it could screen for those women at risk of OF.


All women admitted to the Selian Lutheran Mission Hospital, Arusha, Tanzania with OF as a result of obstructed labour had their height in centimetres and intertuberous measurement in knuckles recorded as part of their admission examinations (Figure 1). Women with genitourinary fistula due to other causes were excluded. Data along with particulars of their delivery, demographics and fistula type were recorded in the patient notes and transferred to an excel database. During a fistula surgical camp in Sierra Leone (AB), the same measurements were taken for all 22 women with OF operated on by the primary author.

Figure 1.

Receiver operating characteristic curves for knuckles, height and fistula index.

During the same time period, consecutive women delivering normally at both institutions had their basic demographic particulars recorded as well as their height and intertuberous dimensions as part of routine admission examinations. In most developing nations, these examinations are to be filled in on the admission partogram.

When delivering normally in Tanzania, women had their intertuberous knuckle dimension measured by AB. Women delivering normally in Sierra Leone had the knuckle dimensions measured by AL, converted to centimetres and then to the equivalent knuckles as per AB to ensure some consistency in knuckle measurement. The number of knuckles was multiplied by the height of the woman in centimetres and recorded as a ‘fistula index’.

As these were routine measurements taken, no consent was asked for and the medical director in each institution granted permission for the study.

Categorical data were summarised using frequencies and percentages. The normality of continuous data was assessed. Normally distributed data were summarised showing means and standard deviations (SD) and analysed using Student's t tests. Data found to be not normally distributed were summarised using medians and range (minimum and maximum) and analysed using Mann–Whitney U tests.

Receiver operating characteristic curves were produced to obtain sensitivity, specificity, and positive and negative predictive values (PPV, NPV) for predicting OF for different threshold values, separately for measurements of knuckles, height and ‘fistula index’.

All analyses used the conventional two-sided 5% significance level. The 95% confidence intervals (CI) for percentages were calculated using the Wilson method in CIA version 2.0.[8] All other summaries and analyses were produced using SPSS version 15 (SPSS Inc., Chicago, IL, USA).


During the study period, 39 women with OF were recruited, 22 (56%) from Freetown, Sierra Leone and 17 (44%) from Selian Hospital, Arusha, Tanzania. Fifty-eight women having a normal vaginal delivery were recruited, 34 (59%) from Freetown and 24 (41%) from Arusha.

There were no statistical differences between the two groups for age, gravity and parity. The length of labour was significantly longer in the OF group, see Table 1.

Table 1. Characteristics of women
Obstetric fistula (= 39)Normal delivery (n = 58) P-value
  1. a

    P-value from Student's t test.

  2. b

    P-value from Mann–Whitney U test.

  3. c

    Data on only 41 women in the normal delivery group.

Age at delivery (years)Mean (SD)24.2 (7.1)26.3 (5.7)0.115a
Length of labour (hours)c Median74.310.0<0.001b

There were highly statistically significant differences between women with OF and normal delivery with respect to knuckles, height and the multiplication between the two, the ‘fistula index’ (Table 2). The women suffering OF had significantly lower measurements in all three domains. Receiver operating characteristic curves with 95% CI are presented in Figure 2. Both the knuckle test and the fistula index were sensitive and specific for predicting a woman with OF with optimal values of sensitivity 92.3% (95% CI 79.7–97.3%) and specificity 91.4% (95% CI 81.4–96.3%) for a measurement of up to three knuckles and sensitivity 94.9% (95% CI 83.1–98.6%) and specificity 91.4% (95% CI 81.4–96.3%) for a fistula index ≤ 507.5. In addition the predictive values were high for both knuckles (PPV 87.8%, 95% CI 74.5–94.7%; NPV 94.6%, 95% CI 85.4–98.2%) and fistula index (PPV 88.1%, 95% CI 75.0–94.8%; NPV 96.4%, 95% CI 87.7–99.0%).

Table 2. Predictors of obstetric fistula
Obstetric fistula (= 39)Normal delivery (= 58) P-value
  1. a

    P-value from Student's t test.

KnucklesMean (SD)2.82 (0.42)3.75 (0.33)<0.001a
Height (cm)Mean (SD)150.5 (5.9)159.3 (5.4)<0.001a
Fistula indexMean (SD)423.32 (65.53)597.74 (59.98)<0.001a
Figure 2.

Measuring the intertuberous space using knuckles.

Height alone was a weak predictor, with a height of < 153 cm having optimal values for sensitivity of 59.0% (95% CI 43.4–72.9%), specificity of 86.2% (95% CI 75.1–92.8%), PPV of 74.2% (95% CI 56.8–86.3%) and NPV of 75.8% (95% CI 64.2–84.55).

One woman had a higher index within the fistula group (of 608, being 152 cm tall with an intertuberous diameter of four knuckles). This woman developed OF after her fourth vaginal delivery at home after a 2-day labour. She subsequently delivered a live born child vaginally at home again. It was likely that she had malpresentation causing obstruction and development of OF in her fourth pregnancy.

Only one woman in the fistula group had a measurement of four knuckles (above). It is interesting that there were no women in the normal vaginal delivery group with a knuckle measurement less than three knuckles, whereas 14 of the 39 women (26%) with OF had measurements of less then three knuckles (Table 3).

Table 3. Number (%) of knuckles for obstetric fistula versus normal delivery
KnucklesObstetric fistula (n = 39)Normal delivery (= 58)
2.04 (10.3%)0
2.510 (25.6%)0
3.022 (56.4%)5 (8.6%)
3.52 (5.1%)19 (32.8%)
≥4.01 (2.6%)34 (58.6%)


Main findings

All parameters—knuckles, height and ‘fistula index’—were significantly lower in women with an OF secondary to obstructed labour than in those women delivering normally. Height alone was a weak predictor of OF with low sensitivity and specificity and added little to the fistula index. The intertuberous measurement alone was a very similar predictor to the fistula index and hence could be used alone, not as part of a ‘fistula index’ with height.

Strengths and weaknesses

Although there was a small sample in this study, the differences were still highly significant. The differences could have been greater but the maximum number of knuckles between the ischial tuberosities were four for obvious anatomical reasons; several women had room for more than four knuckles if the examiners' anatomy had permitted!

Measurement of knuckles between the ischial tuberosities is not a standard measurement because not everyone's knuckles are of the same dimension. Four of the author's knuckles measured 8 cm and two knuckles 4 cm. If this information was used then either an examiner would know the equivalent measurement with respect to his/her hand or some other standard measurement of intertuberous space could be devised. This is especially important in rural villages were primary healthcare providers and traditional birth attendants might have smaller frames and hands secondary to environmental factors. However, using knuckles is reliable with regards to availability of resources in areas where something like a ruler might get broken or stolen and not be able to be replaced.

This study was also open to bias as the examiners were not blinded to which women had delivered normally and who was a fistula patient.

It could also be argued that as the women with OF had delivered up to 30 years before the examination but the women who delivered normally were often examined on the same day as their delivery, pelvic dimensions could change with time. Severe fibrosis forming after the ischaemic insult of the obstructed labour could contract the pelvis and its dimension, but on the other hand many women with OF had no, or minimal, vaginal fibrosis. A prospective study could shed light on this, but for ethical reasons, an OF could not be the end point, but that of obstructed labour and operative delivery.


There have been several attempts at finding ways to screen high-risk women before delivery with few useful results. Several papers have been published looking at pelvimetry, mainly using technology such as X-ray and magnetic resonance imaging. Interestingly some of these papers have shown that the intertuberous diameter is predictive of a poor labour outcome, but these measurements were taken with magnetic resonance imaging.[9, 10] These technologies are out of the reach of most women in the developing world.

Some publications have detailed simpler, technology-independent ways of screening women. A height < 155 cm led to a 4.9 times increased chance of caesarean section in Burkina Faso; or if the mother was <19 years of age and < 150 cm tall, this was again predictive.[11] Another study from Congo looked at the transverse diagonal on pelvic examination[12] and in Tanzania, the ‘reachability’ of the sacral promontory on pelvic examination.[13] Both of these were mildly predictive for obstructed labour but required an invasive pelvic examination.

The intertuberous diameter, as measured by the number of knuckles, is simple, not dependent on technology, easily taught, less intrusive than an internal pelvic assessment and can be performed at the village level. It could potentially help to identify women at risk of OF.

A severely contracted outlet, with a knuckle measurement of 2 or 2.5, appears highly predictive of poor delivery outcome and a risk of OF development (less than three knuckles gave a 100% prediction of OF (95% CI 78–100%).

There was very little difference between the predictive value of the fistula index and the number of knuckles alone as the measurement of height was weakly associated. Using the intertuberous measurement alone simplifies things, as a measurement of risk is not reliant on a tape measure (usually not available) or a multiplication (again, not assessable by illiterate birth attendants in remote areas of the developing world).

It could potentially be a simple screening method identifying which women will need a supervised delivery in an institution with facilities for a caesarean section if needed and those who might not. Further prospective blinded studies are needed.


Clinical measurement of the intertuberous space could be used as a screening tool to identify women at risk of developing OF. Used antenatally, women at risk of obstructed labour could be referred to a health facility or maternal waiting area to have timely access to an assisted or operative delivery if needed.

Disclosure of interest

The authors report that there are no conflicts of interest.

Contribution to authorship

AB designed the study, collected data and wrote the paper. AL collected data and reviewed the paper and SW performed the statistical analysis of the data and reviewed the written paper.

Details of ethics approval

There was no ethics board in either institution and approval to perform the study was given by the medical directors of each institution.


There was no funding for this study.


We wish to thank the staff of Selian Lutheran Mission Hospital, Tanzania and Aberdeen Women's Centre, Sierra Leone, for their cooperation with this study.

Commentary on ‘Pelvimetry and prevention of obstetric fistula: start with the basics’

Browning and colleagues have revisited the prospect of using an easily measured pelvic dimension, the intertuberous measurement, to predict which women will have obstructed labour and develop obstetric fistula. This small, retrospective study demonstrates that both the fistula index (number of knuckles × height) and the number of knuckles alone were significantly lower in fistula patients when compared with those who delivered normally. One need only pass through a typical obstetric fistula ward and see our patients to surmise these findings. However, these findings have been well validated by this study with highly significant differences between groups, especially for the intertuberous diameter. This study benefits from a simple design and practical measurements, which any healthcare provider can perform. The study is limited by small size and by the fact that the examiner was not blinded to the groups. Also, cases and controls were not matched, though there was no overall statistically significant difference in gravity, parity and age between groups.

The design illustrates well only the first step in studying a relatively uncommon condition—estimated prevalence of obstetric fistula 188 per 100 000 women in low-resource settings (Stanton et al. Int J Gynaecol Obstet 2007;99:S4–S9) by looking backwards towards some plausible exposures that may have contributed to the outcome. These retrospective findings encourage us to look forward in time with a large group of women in whom simple pelvic measurements are taken and then determine which patients go on to obstructed labour. As noted by the authors, it would be both impractical and probably unethical to use obstetric fistula as an outcome in a prospective analysis, so obstructed labour would be the best proxy. Quality prospective research is largely neglected in obstetric fistula patients because they are a vulnerable and disenfranchised group. If obstetric fistula were a disease of the wealthy, we would have thousands of studies dedicated to its eradication. Instead, we have a handful of prospective studies, some well-performed cohorts and a host of observational accounts.

In the end, it will still be difficult to draw any causal association between small pelvis, as measured by the intertuberous diameter, and development of obstetric fistula. The challenge will be to determine which variables are sufficiently predictive of obstructed labour to warrant the time and cost of finding these women before labour and referring them to a maternity waiting home with an adjacent health facility that performs cesarean deliveries so as to avoid unchecked labour in these patients at all cost. However, this remains the pervasive challenge for all high-risk conditions in low-resource settings that lead to maternal and neonatal morbidity and mortality. Women with undiagnosed pre-eclampsia, previous cesarean delivery, malpresentation or multiple gestation, undiagnosed or poorly controlled HIV and many other conditions also remain at high risk of death or disability. In fact, most poor obstetric outcomes still occur to those women in whom no specific risk factor exists. Helpful predictors of obstructed labour and other obstetric tragedies may be useful stop-gap measures towards the ultimate goal of having safe, local obstetric care for all women independent of where they live or how much they earn.

Disclosure of interests


  • JP Wilkinson

  • Associate Professor of Obstetrics and Gynecology, University of North Carolina and Freedom from Fistula Foundation, Lilongwe, Malawi