In white Caucasian populations about 10% of all pregnancies involve a mother with rhesus (Rh) D negative blood group and an RhD positive fetus, potentially placing the mother at risk of sensitisation and future babies at risk of haemolytic disease of the fetus and newborn. Anti-D prophylaxis (anti-D IgG) can be given to prevent a woman producing antibodies against fetal RhD-positive blood cells and becoming sensitised. Prophylaxis following delivery was introduced in the 1960s, with a blood cord serology test used to identify the baby's RhD status. This dramatically reduced maternal sensitisations and cases of rhesus disease in babies [1]. In the mid-1990s, routine antenatal anti-D prophylaxis (RAADP) was first used. This was reported to further reduce sensitisation rates (from 1.2% for the earlier policy to 0.28%) [2], with RAADP stated to be 98.4-99% effective [3]. In 2002, the National Institute for Health and Clinical Excellence (NICE) published guidelines for the UK, recommending two doses (500iu each) of anti-D IgG at weeks 28 and 34 of gestation as effective and cost-effective [1]. Figures indicate that 90% of hospitals in England and Wales comply with these guidelines, with 90% of the target population reported to receive the first dose of anti-D IgG and up to 87% the second dose [4, 5]. In 2008, updated NICE guidance stated that a single dose of anti-D (1500iu) between weeks 28 and 30 would also be cost-effective [6]. However, with both RAADP policies, the 40% of RhD negative women whose fetus is also RhD negative will receive antenatal prophylaxis unnecessarily [1].

Non-invasive prenatal diagnosis (NIPD) of fetal RHD blood group is based on the presence of cell-free fetal DNA in maternal plasma [7–10]. Fetal RHD genotyping of this material has the potential to enable antenatal prophylaxis targeted at the 60% of pregnancies with an RhD positive fetus, thereby saving anti-D costs. NIPD test accuracy figures in the range 94.8% - 100% have been reported [8, 9, 11–14], although studies exhibit certain shortcomings [15]. By 2007, many European countries had introduced NIPD testing for the small number of sensitised women in order to identify high risk pregnancies (fetus RhD positive) [10, 14, 16]. Approximately 250-300 sensitised women in England and Wales now undergo RhD NIPD tests annually. In such cases the NIPD test offers the clinical advantage of avoiding an invasive procedure such as amniocentesis with its associated risk of fetal loss [17], as well as possible cost savings.

Several authors have recently recommended a wider roll-out of NIPD testing to the remaining, non-sensitised pregnancies [10, 14, 16], with suggestions that this will be cost saving [16, 18]. Such an approach would extend testing to a significantly larger population. Approximately 16% of white women are RhD negative; RhD negativity in other ethnic groups is lower; South Asians (5.5% - 10.9%), West Africans (5%), Chinese (<1%) [19, 20]. Extension of NIPD testing to this larger population (approximately 93,000 pregnancies annually in England and Wales [21]) is not feasible with the low throughput and labour intensive processes used for testing samples from sensitised women [22]. Mass testing will require an accurate automated laboratory procedure, similar to that recently reported which has a test sensitivity of 99.7% [16]. NIPD testing is not able to identify the degree to which a woman is prone to sensitisation because this is governed by several factors in addition to fetal RHD status [23].

To date the costs and implications of large scale NIPD testing to target antenatal prophylaxis have not been determined. The aim of the present study was to estimate the cost savings (if any) in England and Wales for two implementation strategies compared to the current two-dose RAADP programme (and a possible future single-dose policy). The research was undertaken as part of the SAFE Network of Excellence funded by the European Commission [24].

Management of non-sensitised RhD negative pregnancies has become considerably more effective over time with the sequential introduction of post-delivery and antenatal anti-D prophylaxis [6]. Residual RhD sensitisations still occur, mainly due to poor prophylaxis implementation, administration errors and sub-optimal patient compliance [5, 19, 34]. Advocates of mass RhD NIPD testing therefore focus on the potential for generating net cost savings although this is not quantified [16, 18].

The cost analysis presented in this paper shows that the net financial benefit of implementing mass NIPD testing as an add-on using in-house tests (while maintaining current postnatal testing) will be negligible in England and Wales. The level of royalty fee negotiated will be a major influence on this with zero saving predicted above a fee of £2.18 per test. If antenatal NIPD were to also displace postnatal testing, higher net savings could be realised and the royalty fee might rise to nearly £9 per test before these disappear. Use of a commercial kit, such as that currently being marketed in Europe [30] would make NIPD more expensive in both implementation scenarios. Furthermore, if single-dose prophylaxis, now recommended as equally effective by NICE [6], were to become widespread only the second scenario (displacing postnatal testing) would produce a net saving.

Our analysis also indicates that NIPD implementation is unlikely to produce important clinical benefits. The number of sensitisations will not fall appreciably and in fact might rise if NIPD test sensitivity is below 99.9%. This is especially true for the scenario in which postnatal blood cord serology is eliminated in order to maximise any net cost saving. Introduction of NIPD typing must also meet the needs of testing in a diverse population [35]. In the UK, the Race Relations Amendment Act 2000 lays a statutory duty upon the NHS to consider the implications of any new policy on racial equality [36]. In white Caucasians, the most frequent RhD negative genotype is caused by an RHD deletion, while in Black Africans and South Asians other variants exist which can result in mistyping [37]. There has been a recent call for careful consideration of patients' ethnic background by obstetricians [38]. Antenatal prophylaxis has been reported to be equally effective, and also more cost-effective, in ethnic minority populations in the UK and elsewhere [19, 23, 39, 40]. However, to date NIPD assays have primarily been validated in white Caucasian populations and their reliability in ethnic minority populations still has to be demonstrated [41]. Due to the absence of a 'gold standard' comparator, it is not possible to establish the absolute accuracy of the serology test, although since the test is performed manually it will be prone to some human error. Thus, even in scenario 1, sensitisations may occur because of inaccuracy in the postnatal serology test. Two recent studies have suggested that RhD NIPD may provide more accurate results than postnatal serology, both by minimising human error and by helping to detect D variants and weak D newborns [14, 28].

Based on our findings it is difficult to argue that NIPD targeted antenatal prophylaxis will be sufficiently cost-effective to warrant its large scale introduction. Increased magnitude of savings will be at the expense of additional sensitisations. Even the maximum amount of £9,190 saved per additional sensitisation is much lower than the figure of £15,903 - £17,668 judged acceptable for avoiding a sensitisation by NICE [42].

However, if NIPD testing could be carried out earlier in pregnancy cost-effectiveness might be improved. For example, adherence to prophylaxis guidelines for 'high risk' antenatal events might be improved if women are identified as carrying an RhD positive child early in their pregnancy [43, 44]. Balanced against this is the fact that NIPD test accuracy is known to deteriorate at lower gestational age [45–47]. The accuracy of mass testing in early pregnancy is uncertain. A large scale UK trial of NIPD testing in the first trimester of pregnancy has still to report [Dr. Kirstin Finning, National Blood Service, Bristol, UK, personal communication]. However, a US company which has recently launched a fetal RHD genotyping test for use in first-trimester pregnancies reports a sensitivity of 97.2% [48].

During the development of the original 2002 NICE guidelines, questions were raised about the safety of anti-D IgG because of isolated cases of hepatitis C infection in the early 1990's [49, 50]. Avoidance of unnecessary use of anti-D has therefore been suggested as a 'soft benefit' for NIPD even though anti-D continues to be judged safe for routine antenatal use by NICE [1, 6]. Modern processes make it among the lowest-risk biological products in use [51], especially in combination with avoidance of high risk plasma donors [52–54]. A second linked argument is that NIPD might improve prophylaxis coverage [4]. Some authors have suggested that low uptake might be linked to anxiety about anti-D blood products [4, 5, 55]. At the same time, it is recognised that poor antenatal attendance, knowledge that the father is RhD negative, or certainty that this is a woman's last pregnancy influence uptake [19]. Such factors would be unaffected by introduction of NIPD.

A final advantage put forward is linked to anti-D supply. In England and Wales, where anti-D IgG is produced using plasma purchased from the USA, there have been no supply problems. However, in countries such as the Netherlands where in-country sourcing requires the purposeful hyperimmunisation of volunteers, decreased use of anti-D has been viewed as an ethical benefit in its own right [56]. Scientists are currently working on the development of recombinant anti-D IgG as a replacement for human plasma products [57]. Although there is no completely satisfactory product on the market [58], a phase 2 clinical trial has recently commenced [59]. In future, this has the potential to address these concerns [60, 61]. If, as is likely, the final recombinant product is more expensive than the human product then NIPD would become financially more attractive by generating greater anti-D savings.

The main limitation in the present study is uncertainty over the level of royalty fee for in-house NIPD tests. The speed of development and diffusion of recombinant anti-D IgG may also affect the conclusion drawn, but this is difficult to estimate without information on the eventual cost of this product. Further new technologies which might influence cost-effectiveness include NIPD markers for conditions such as haemoglobinopathies or Trisomy 21 (Down's syndrome) which appear to be nearing final development [62–66]. If a battery of markers, including fetal RHD status, were combined in a single NIPD test then this would reduce costs through economies of scale in common processes such as DNA extraction and amplification. Finally, our economic analysis, which only considered England and Wales, may be less applicable in other countries. For example, in some European countries there is no policy of routine antenatal prophylaxis although this is now being considered [67, 68]. In such cases, the introduction of targeted prophylaxis underpinned by RhD NIPD testing might offer similar clinical benefit to introduction of RAADP at a potentially acceptable cost.

Our analysis does not support routine implementation of RhD NIPD testing in England and Wales at present for non-sensitised pregnancies. Annual savings will be relatively small and balanced against this increased maternal sensitisations may be unacceptably high. First trimester testing is unlikely to alter this picture significantly although other emerging technologies may. There is also a need to demonstrate the reliability of assays in different ethnic minority populations.