The limitations of the predictive tests currently available to assess threatened preterm labour.
In terms of a diagnostic quandary, threatened preterm labour continues to vex us. We have all assessed an anxious woman presenting with uterine activity and the price of making the wrong diagnosis at early gestations is high. In 2012, 5456 infants in Australia and New Zealand were born before 34 weeks, representing only 1.7 per cent of the total births, but a huge cost in terms of morbidity, long-term health consequences and inevitably neonatal care dollars.1 However, as few as three per cent of the women who present for attention have been found to progress to delivery within the seven ensuing days in some studies.2
In the absence of clear cervical dilatation or another clinical factor, clinical prediction of which women will deliver is poor. However, sending our patient home runs the risk of false reassurance and a potentially avoidable adverse outcome. Cautiously admitting for observation will be complicated by available neonatal facilities and the need to transfer to urban tertiary centres, with considerable costs to the healthcare system and disruption to family life. In addition, steroids and other antenatal interventions such as magnesium sulphate and antibiotics are critical if delivery is imminent, but expensive and potentially harmful if given without clear indication.
There are two commercially available and competing tests for bio-markers of the pre-labour cascade in Australia and New Zealand: fetal fibronectin (fFN) and Actim Partus. Controversy exists over which is the better test and, in practice, both tests have drawbacks.
The fFN was developed first and extensively validated in clinical trials involving more than 40 000 women in excess of 200 peer-reviewed studies. This is in contrast to the Actim Partus, which has been trialled in fewer than 3000 women in fewer than 30 studies. The fFN therefore paved the way to clinical acceptance, and continues to be the more widely used technology, aided by FDA approval in the US, which Actim Partus has not gained.
The two tests are not however equivalent, although the detection of significant levels of either between 22 weeks and 35 weeks is thought to indicate chorio-decidual separation, regardless of the underlying precipitating mechanism. Fetal fibronectin is an adhesive glycoprotein produced by many cell types, including the fetal amnion. The Actim Partus is an immuno-chromatographic bedside test for phosphorylated insulin-like growth factor binding protein (IGFBP-1) produced by the decidua. The swab specimen collection techniques are also slightly different – the fFN can be collected from the posterior vaginal fornix, but the Actim Partus is potentially more technically challenging and uncomfortable, requiring the removal of fluid from the external cervical os.
Both tests are obviously subject to false positive results, as a background low level of these markers will always be present upon swabbing cervico-vaginal secretions and an increase is not specific for labour. Both tests will usually give false positive results in the presence of amniotic fluid and significant blood. The fFN test can also turn false positive in the presence of semen and lubricating jelly or recent cervical trauma (although a negative result is still valid as a true negative), and false negative results have been obtained owing to Candida vaginal infection. Actim Partus is marketed as able to be used regardless of recent intercourse since IGFBP-1 is not present in semen, but interestingly the product information states there is no data currently to support these assumptions and recent vaginal examination can cause a false negative by removing fluid from the cervical os.
The fFN is currently available in three different modalities: the original TLi-iq qualitative test; the recently released quantitate 10Q Rapid fFN Analyser; and the quantitative QuikCheck fFN, which is a bedside dipstick test retailed for AU $124 per test. The TLi-iq has the slowest turnaround time of 25 minutes and has therefore been phased out by the company in Australia, but is still used in New Zealand. The replacement quantitative 10Q, available in both countries, is designed to be kept in labour wards as opposed to in a laboratory and only takes ten minutes to process. It costs around AU $104 per test processed, but requires an analyser ($4000) and personnel to run daily calibrations. Quantitative results have also been studied as allowing stratification of risk, since the positive predictive value rises with the concentration detected, and fFN has been validated in combination with cervical length screening for asymptomatic women at high risk of pre-term labour.2
Table 1. The only published trials directly comparing clinical performance of fetal fibronectin and IGFBP-1 testing in the same high-risk patients
|Turnell et al 2005 3||99||99||22||16||80||80||85||78|
|Ting et al 2007 4||89||92||39||46||61||72||78||80|
|Eroglu et al 20075||>97||97||35||41||83||83||81||84|
|Audibert et al 20106||98||91||36||20||83||17||84||93|
|Cooper et al 20127||88||86||54||22||33||39||95||74|
|Khambayet al 20128||79||70||67||0|
Table 2. The meaning of the statistical terms used to describe the accuracy of a predictive test for pre-term labour.
|Negative predictive value||The percentage of women with a negative test who will not deliver within 7–10 days|
|Positive predictive value||The percentage of women with a positive test who will deliver within 7–10 days|
|Sensitivity||The percentage of women who will deliver within 7–10 days who are correctly identified ‘positive’|
|Specificity||The percentage of women who will not deliver within 7-10 days who are correctly identified ‘negative’|
The Actim Partus test’s major market advantages were the lower cost (AU $40 per test) and avoidance of the need for an analyser, particularly in a small unit that may only see a few cases a year. The second advantage is less clear-cut now that an equivalent analyser-free fFN test is available and the fFN company will now place an analyser in a hospital free of charge on an equipment loan agreement.
The cost advantage does, however, need to be weighed up against the clinical utility of the test, which is hard to do given the paucity of published data for the Actim Partus test. In symptomatic women, the fFN has a generally better negative predictive value of 88–99.5 per cent overall, compared to 86–95 per cent for Actim Partus. There have only been six studies published directly comparing the two tests in the same high-risk women, of which one trial was performed in asymptomatic patients (see Table 1). The trend of these studies suggests that, although generally comparable, the Actim Partus performs slightly less well over all parameters, particularly in asymptomatic women with a positive predictive value of zero per cent. This equates to a small saving from using a cheaper test potentially negated by a much larger unnecessary cost of admitting and transferring women to tertiary units who will not deliver.
Small-scale studies have described a cost saving when using the fFN test, in the order of an average of AU$3000 per patient tested in an Australian study.9 Larger scale cost-benefit reviews have not taken into account societal costs of admitting women to hospital in terms of lost productivity or childcare costs.10, 11 In reality, the on-the-ground advantages, and possibly visible marketing strategies, have led to widespread uptake of these tests in Australia, New Zealand, Canada, the US and Europe.
Both tests are currently undergoing Medical Services Advisory Committee review in Australia; the Actim Partus for the first time, since the clinical implementation of the Actim Partus has to date relied on the groundwork prepared by 20 years of fFN research and the assumption of equivalent validity. Based on the data so far, the Actim Partus test should not be used in asymptomatic high-risk women and may result in more unnecessary intervention owing to a lower positive predictive value. The current value of both tests lies in the high negative predictive values, allowing up to 90 per cent of symptomatic women to be reassured and discharged. A positive result requires further assessment in a tertiary setting, preferably with ultrasound assessment of cervical length, which has been shown to perform better than either test, to guide ongoing management.