Uptake, outcomes, and costs of implementing non-invasive prenatal testing for Down’s syndrome into NHS maternity care: prospective cohort study in eight diverse maternity units

Objective To investigate the benefits and costs of implementing non-invasive prenatal testing (NIPT) for Down’s syndrome into the NHS maternity care pathway. Design Prospective cohort study. Setting Eight maternity units across the United Kingdom between 1 November 2013 and 28 February 2015. Participants All pregnant women with a current Down’s syndrome risk on screening of at least 1/1000. Main outcome measures Outcomes were uptake of NIPT, number of cases of Down’s syndrome detected, invasive tests performed, and miscarriages avoided. Pregnancy outcomes and costs associated with implementation of NIPT, compared with current screening, were determined using study data on NIPT uptake and invasive testing in combination with national datasets. Results NIPT was prospectively offered to 3175 pregnant women. In 934 women with a Down’s syndrome risk greater than 1/150, 695 (74.4%) chose NIPT, 166 (17.8%) chose invasive testing, and 73 (7.8%) declined further testing. Of 2241 women with risks between 1/151 and 1/1000, 1799 (80.3%) chose NIPT. Of 71 pregnancies with a confirmed diagnosis of Down’s syndrome, 13/42 (31%) with the diagnosis after NIPT and 2/29 (7%) after direct invasive testing continued, resulting in 12 live births. In an annual screening population of 698 500, offering NIPT as a contingent test to women with a Down’s syndrome screening risk of at least 1/150 would increase detection by 195 (95% uncertainty interval −34 to 480) cases with 3368 (2279 to 4027) fewer invasive tests and 17 (7 to 30) fewer procedure related miscarriages, for a non-significant difference in total costs (£−46 000, £−1 802 000 to £2 661 000). The marginal cost of NIPT testing strategies versus current screening is very sensitive to NIPT costs; at a screening threshold of 1/150, NIPT would be cheaper than current screening if it cost less than £256. Lowering the risk threshold increases the number of Down’s syndrome cases detected and overall costs, while maintaining the reduction in invasive tests and procedure related miscarriages. Conclusions Implementation of NIPT as a contingent test within a public sector Down’s syndrome screening programme can improve quality of care, choices for women, and overall performance within the current budget. As some women use NIPT for information only, the Down’s syndrome live birth rate may not change significantly. Future research should consider NIPT uptake and informed decision making outside of a research setting.


Online supplementary material: cost consequences analysis
Literature review Several studies have examined the costs and benefits of NIPT for Down syndrome screening [1][2][3][4][5][6][7][8][9][10][11][12][13]. Several of these studies focused on the USA [3,4,7,10], Australia [1,9] or Canada [8]. Two studies focused on the UK [5,12], one on the Netherlands [2] and one on Belgium [6]. Palomaki et al. found that compared with current screening, NIPT contingent testing would detect slightly fewer cases, and that fewer invasive tests would be performed, fewer procedure related miscarriages would occur, and costs would be lower [10]. The analysis however did not include costs of NIPT, which may be substantial. Garfield & Armstrong found that NIPT contingent screening in high risk women would be more effective in terms of procedure-related miscarriages and Down syndrome cases detected and cost saving compared to the current screening programme when the costs of NIPT were no higher than $1200 [4]. Cuckle et al. compared different NIPT pathways with the current pathway and showed that the total costs would increase significantly if NIPT was used as a first line test [3] and therefore NIPT as a contingent test was recommended. In another study from the USA, the cost-effectiveness of NIPT contingent testing compared with NIPT as a diagnostic test in high risk women was assessed [7]. NIPT as contingent testing was cost-effective compared to NIPT as a diagnostic test, but the authors did not investigate the cost-effectiveness of NIPT compared to the current screening programme, making it difficult to assess if NIPT contingent testing was costeffective. O'Leary et al. demonstrated that NIPT contingent testing in high risk women would increase the costs of screening in Australia, and that as long as the uptake of NIPT was higher than the uptake of invasive testing in the current pathway, more cases would be detected with NIPT [9]. Song et al. found that offering NIPT to women at high risk of a Down syndrome pregnancy would increase the detection of Down syndrome cases and decrease the total health care costs [11]. The authors included lifetime costs of caring for a child with Down syndrome, which had a large influence on the results. NIPT would lead to increased costs if these costs were not included. Wald and Bestwick investigated the detection rate and genetic test costs of a protocol combining screening and NIPT in the UK, but did not include any costs for invasive testing [12]. They concluded that their protocol would be cheaper than universal DNA testing.
Ayres et al. studied the consequences of replacing the current screening protocol with several different NIPT pathways (first line testing or contingent testing with or without first line NIPT for certain high risk women) [1]. They concluded that NIPT contingent screening was cost saving, but less effective; first line testing was more effective and more costly. The most cost-effective option was NIPT contingent screening with NIPT as first line test for women >40 years. In a Dutch study, the current screening programme was also compared with first line as well as contingent NIPT testing [2]. Both NIPT strategies increased the detection of Down syndrome cases at higher total costs. The authors reported that the costs of the programme would be unacceptably high if NIPT were implemented as a first line test, and hence the NIPT contingent pathway led to more favourable results. A similar comparison was performed by Morris et al., Neyt et al. and Okun et al., all concluding that first line testing was unlikely to be cost-effective and showing more favourable results for NIPT contingent testing [5,6,8]. In the study by Morris et al. NIPT contingent testing could decrease the procedure related miscarriages without increasing the total costs of the screening pathway [5]. The authors looked at different screening risk thresholds for offering NIPT as a contingent test. This threshold value had a large influence on the results, although the detection of Down syndrome cases was lower than in the current programme for each threshold because of assumptions made surrounding the uptake of testing. First line testing was calculated to detect more cases, but at significantly increased total costs of the screening programme. In the study by Walker et al. first line NIPT testing was cost saving and more effective compared to the integrated test from a societal perspective (including lifetime costs of a child with Down syndrome) [13]. However, NIPT was more costly when a narrower perspective was taken, not including lifetime costs of a child with DS.
Influence of uptake and cost of NIPT Based on the information from the studies described above, it is highly unlikely that first line NIPT testing would be cost-effective, unless the costs of NIPT were to fall dramatically. On the other hand, NIPT as contingent testing may have the potential to increase the detection of Down syndrome cases and/or decrease the number of procedure related miscarriages, without increasing costs. In most studies, assumptions regarding the uptake of NIPT were made, because of a lack of empirical data, e.g., assuming 100% uptake or assuming that uptake would be the same as for invasive testing. These assumptions are important because both the number of cases detected and total costs will depend on NIPT uptake, and therefore there is considerable uncertainty about the cost-effectiveness of NIPT due to uncertainty about the uptake of the test. Another area of uncertainty is the cost of NIPT, especially in the NHS since it is not routinely provided at present. In the private sector in the UK, prices for NIPT vary from £400-£900 [5]. In summary, there is little economic evidence for NIPT that is directly relevant to the UK NHS (screening practices, costs and uptake are likely to vary between countries), and what evidence there is has been required to make simplifying assumptions about the uptake of testing.
Morris et al. showed that if NIPT could be offered to more women (for example using a screening risk cut-off of 1/2000), more cases would be detected than with a screening risk cut-off of 1/150, but at increased costs [5]. It is possible that the costs of NIPT contingent screening relative to the current DS screening pathway will vary by screening cut-off, since this affects costs and outcomes. It is therefore useful to investigate the effects of NIPT testing at different screening risk cut-offs to find the best strategy without increasing costs.

Limitations of QALYs
In the UK the recommended outcomes for economic evaluations are quality-adjusted life years (QALYs), and the recommended cost-effectiveness measure is based on the incremental costs per QALY gained [14]. One advantage of using the incremental costs per QALY gained is that there is a published threshold that can be used to judge whether or not an intervention is good value for money (£20,000 to £30,000 per QALY gained [14]). However, measuring QALYs arising from prenatal testing is challenging. Only one study has reported the QALY gains in the context of NIPT testing for Down syndrome [7]. The authors assumed that women would have a lower quality of life during their remaining years of life after giving birth to a baby with Down syndrome, during 2 years after miscarriage or termination and during 1 year after a false positive result. It is difficult to determine the quality of life of a child with Down syndrome and of other unaffected children as well as the quality of life of the parents. There is also uncertainty about what happens when a pregnant mother decides to terminate a Down syndrome affected pregnancy. She might get pregnant again and have another (healthy) baby or not. Next to this, some women might decide not to terminate the pregnancy in case of a Down syndrome affected fetus, but use the information obtained from screening to prepare for the birth of an affected baby. Having this information and being able to prepare for an affected baby, or the reassurance that the baby is unaffected, could also increase the quality of life of the parents. Due to these difficulties, the economic consequences of prenatal testing are not usually reported using QALYs.
As in previous studies, we report instead the number of Down syndrome cases detected, the number of procedure-related miscarriages and costs. Our measure of costs is the combined cost of Down syndrome screening, NIPT and invasive diagnostic testing; it does not include the costs of pregnancy outcomes nor of costs incurred beyond birth, such as lifetime costs of caring for a baby with Down syndrome, according to the NSC's preferred costing methodology. .

Cost consequences analysis of NIPT for Down syndrome as contingent testing compared with the current DS screening programme in the UK National Health Service
Methods Model structure A decision tree was developed to assess the costs and outcomes of 4 different screening pathways: (1) the current pathway; (2) NIPT as contingent screening for women with a ≥1/150 risk; (3) NIPT as contingent screening for women with a ≥1/500 risk; and (4) NIPT as contingent screening for women with a ≥1/1000 risk. Figure 2 in the main paper depicts the current screening pathway and the NIPT contingent screening pathway. Briefly, in the current screening pathway, women are offered invasive testing when their risk based on the combined or quadruple test is ≥1/150. There is a small risk of a procedure related miscarriage, so some high risk women decide not to undergo any further testing. If the result of the invasive test is positive, women can decide to terminate the pregnancy.

RAPID study pathway
In the NIPT pathway, women are offered a NIPT test after a high risk result (depending on the threshold) based on the combined or quadruple test. The pathways for the different thresholds are all similar except for the threshold risk at which NIPT is offered as contingent screening (≥1/150, ≥1/500 or ≥1/1000). If the NIPT test result is positive, an invasive test is offered to confirm the diagnosis. Some women with a risk ≥1/150 after the combined or quadruple test might decide to have an invasive test directly and not have NIPT first. We also assessed the costs and outcomes of modifying the NIPT contingent screening pathway as described to preclude the option of having an invasive test directly after screening, so that only women with a positive NIPT result were offered invasive testing.
Focus on Down syndrome screening While NIPT may have a role in detecting other trisomies, the over-arching aim of the RAPID evaluation study is to evaluate NIPT for Down syndrome. For the economic analysis we therefore focused on the Down's syndrome screening pathway and did not include other trisomies, even though women were offered testing for trisomies 13 or 18 in some clinics. For the present analysis, we categorised women based on Down syndrome risk only and we excluded any testing performed solely based on trisomy 13 or 18 risks. Six women with a Down syndrome risk between 1/151 and 1/1000 underwent invasive testing before NIPT because of a high risk of trisomy 13 or T18. These invasive tests were excluded from the present analysis, as we assumed that only women with a Down syndrome risk ≥1/150 could undergo invasive testing directly after screening. Three of these women also accepted NIPT, the other three women declined NIPT. We therefore assumed that for the DS screening pathway, the first three woman would undergo NIPT and no subsequent invasive test and the other three woman would undergo no further testing after screening.

Model inputs
We ran our analyses using two datasets; the first was based predominantly on data from the RAPID evaluation study, supplemented with national data where necessary; the second was based predominantly on national data, using data from the RAPID evaluation study mainly to quantify NIPT uptake behaviours. Data from the RAPID evaluation study were used to assess the uptake of NIPT contingent screening and the uptake of subsequent NIPT and invasive testing. Data from the RAPID study were also available for the different outcomes of the tests and pregnancy outcomes. The uptake of screening and invasive testing in the current pathway was assessed using a historical dataset collected in the same clinics as the RAPID data, but before introduction of NIPT, during the period 2011 to 2012. The key parameters of the model, and the sources of these data, are listed in Table 2 (main paper). The uptake of screening in the eight clinics involved in the RAPID study was slightly higher before NIPT was offered than during the RAPID study (76.0 vs 78.7%). This might be caused by the fact that some women undergo NIPT in the private sector instead of undergoing screening in the national programme. In the present analysis we assumed that this decrease in uptake of screening is unrelated to the RAPID study, so we used the same uptake value as in the current programme (78.7%). Our analysis considers different Down syndrome screening cut-offs and we require NIPT uptake figures for these cut-offs. The uptake of NIPT was 70.5% in women with a risk between 1/150 and 1/1000. This percentage was 72.5 for women with a high risk result. More than 80% of the women with a positive NIPT result underwent invasive testing to confirm the diagnosis, while 54% of the women with a high risk result in the current screening programme underwent invasive testing.
The number of women with a serum screen risk of ≥1/150 was higher than the national figure. In the analysis using the RAPID data, screening results were age-standardised to make them more representative of the national screening programme. For some input parameters, no data was available from the RAPID study (for example fetal loss after invasive testing or the incidence of Down syndrome in the general population) and national sources were used instead. No national data is available yet on any outcomes related to NIPT (e.g., uptake of NIPT, number of positive results), so for the analysis based on national sources, we used RAPID data for these input parameters. In the RAPID study, five of the positive NIPT results were found to be false-positives (9.1%), and the positive predictive value of NIPT has been found to be less than 100% in several other studies. For the national dataset, we therefore assumed a positive predictive value of 90.9% and varied this value in the sensitivity analysis.
For the additional scenarios in which we restricted the possibility of women undergoing invasive testing without having NIPT first we assumed that the uptake of NIPT among women with a high risk result (≥1/150) would be 91%. In the RAPID study several very high risk women underwent invasive testing directly, and therefore the proportion of positive NIPT results was expected to be higher when these women were to undergo NIPT first. We estimated an average of 7.9% positive NIPT results for this scenario.

Costs
To assess the costs of NIPT we have taken into account the costs of all different tests (combined/quadruple Down syndrome screening tests, NIPT and invasive testing), including sampling, laboratory testing, and feeding back the results. The unit costs are listed in Table 2 (main paper). Costs of screening were taken from the UK National Screening Committee decision planning tool and inflated to 2012-13 UK£ [15]. Laboratory charges for NIPT (£250) in the RAPID study were used for the costs of NIPT in the base case and these costs were varied over a wide range in the sensitivity analysis. Costs for counselling and feedback by a midwife as well as phlebotomy costs were included [16,17]. Total charges for invasive testing were collected from the eight participating clinics and the weighted mean unit cost of CVS/QF-PCR and amniocentesis/full karyotype was used; this included pre-test counselling, consultant obstetrician time to perform the procedure, cytogenetic laboratory costs, and post-test feedback and counselling. This value was varied in the sensitivity analysis. All costs are expressed in 2012-13 UK£. The time horizon in this study was the duration of pregnancy, therefore no discounting was necessary.
There were a small number of failed and inconclusive tests in the RAPID evaluation study (n=31 out of 1971 tests performed, excluding the period 26/09/2014 to 10/10/2014 when the study was suspended due to failure of and subsequent unavailability of library preparation kits) Of the women that were offered a repeat test, 79% accepted to have a repeat test. For the purpose of the analysis, we assumed that 1.2% (79% of 31/1971) of NIPT tests would result in a failed/inconclusive test and all these women would undergo a repeat test. To capture this we inflated the costs of NIPT by the proportion of patients requiring extra tests. Because of the small numbers, we did not take into account a difference in costs between rerunning the sample in the lab or redrawing blood and running the test using the new sample.

Model outputs
The main outcomes of the analysis were: screening and diagnosis-related costs of the different pathways; number of Down syndrome cases detected; and, number of miscarriages avoided. Uncertainty around our input parameters was taken into account using one-way sensitivity analysis and probabilistic sensitivity analysis. In the one-way sensitivity analysis we varied one parameter at a time over a plausible range (mostly equal to 95% confidence limits, but for some parameters such as uptake of screening wider intervals were used) to identify the maximum value at which introducing NIPT to the national screening programme would be cost neutral. In the probabilistic sensitivity analysis, 1000 simulations of the outputs were produced, based on drawing random samples from the probability distributions of all input parameters. Beta distributions were used for most probabilities, Dirichlet distributions for probabilities if there were 3 or more possible outcomes and gamma distributions for all costs. The distribution of the 1000 simulations was used to calculate 95% uncertainty intervals for each of the model outputs.
Type of clinic We expect a difference in NIPT uptake between one stop clinics (where NIPT is done on the same day as screening) and clinics where women need to come back for NIPT at a later date (screening results are phoned out). In a previous analysis (NSC report) we estimated the uptake of NIPT and direct IPD in both types of clinics. We applied these estimates to our current central estimate as shown in the main paper. As might be expected, in the RAPID study, the uptake of NIPT in women with a risk between 1/150 and 1/1000 was higher in the one-stop clinics than in the clinics where an extra visit for NIPT was required (74.4% versus 59.3%). However, in one stop clinics, high risk women (≥1/150) were more likely to undergo invasive testing directly after screening (23.0% versus 13.9%) and less likely to undergo NIPT first (70.0% versus 77.7%) ( Table S1). The reason for this is unclear. Because of this difference, we ran the analysis separately for both types of clinic in a sensitivity analysis. The costs of the different steps in the pathways were assumed to be the same in each type of clinic, because the NIPT test and counselling is charged as a separate contact in both types of clinics. The total costs of the NIPT pathways and the number of cases detected and the number of miscarriages were expected to differ between the two clinics, because of the difference in uptake of NIPT. Because no extra visit for NIPT is required in the one-stop clinic, the costs incurred by the woman are expected to be lower.

19.
The      -means that varying the parameter within the specified range did not change the results from cost saving to cost neutral, * means that varying the parameter within the specified range did not change the results from increased costs to cost neutral -means that varying the parameter within the specified range did not change the results from cost saving to cost neutral, * means that varying the parameter within the specified range did not change the results from increased costs to cost neutral Uptake figures were derived from Table 1 in the main text but including T21 only. Figure S1. Benefits and costs of the DS screening pathway nationally for the current pathway and using NIPT as a contingent test for wom of > 1/150, 1/500 and 1/1000.

RAPID DATA
Benefits and costs of the DS screening pathway nationally for the current pathway and using NIPT as a contingent test for wom 14 Benefits and costs of the DS screening pathway nationally for the current pathway and using NIPT as a contingent test for women with a DSS risk Figure S2. Marginal costs for different base prices of implementing NIPT as a contingent test for DS for women with a DS screening risk indicates costs without the option of direct IPD and the other li direct IPD without prior NIPT.

RAPID DATA
Marginal costs for different base prices of implementing NIPT as a contingent test for DS for women with a DS screening risk > 1 in 150 ( ), 1 in 500 ( ), or 1 in 1000 ( ). The dashed line indicates costs without the option of direct IPD and the other lines costs allowing the option of 15 Marginal costs for different base prices of implementing NIPT as a contingent test for DS ). The dashed line nes costs allowing the option of NATIONAL DATA Figure S3a. Benefits and costs of the DS screening pathway nationally for the current pathway and using NIPT as a contingent test for wom risk of > 1/150, 1/500 and 1/1000. One stop clinics Benefits and costs of the DS screening pathway nationally for the current pathway and using NIPT as a contingent test for wom 16 Benefits and costs of the DS screening pathway nationally for the current pathway and using NIPT as a contingent test for women with a DSS Figure S3b. Benefits and costs of the DS screening pathway n risk of > 1/150, 1/500 and 1/1000. Two stop clinics Benefits and costs of the DS screening pathway nationally for the current pathway and using NIPT as a contingent test for women with a DSS 17 ationally for the current pathway and using NIPT as a contingent test for women with a DSS RAPID DATA Figure S4a. Benefits and costs of the DS screening pathway nationally for the current pathway and using NIPT as a contingent test for wom risk of > 1/150, 1/500 and 1/1000. One stop clinics Benefits and costs of the DS screening pathway nationally for the current pathway and using NIPT as a contingent test for wom 18 Benefits and costs of the DS screening pathway nationally for the current pathway and using NIPT as a contingent test for women with a DSS Figure S4b. Benefits and costs of the DS screening pathway nationally for the current pathway and using NIPT as a contingent test for wom risk of > 1/150, 1/500 and 1/1000. Two stop clinics Benefits and costs of the DS screening pathway nationally for the current pathway and using NIPT as a contingent test for wom