Preferences for prenatal diagnosis of sickle‐cell disorder: A discrete choice experiment comparing potential service users and health‐care providers

Abstract Background Non‐invasive prenatal diagnosis (NIPD) for sickle‐cell disorder (SCD) is moving closer to implementation and studies considering stakeholder preferences are required to underpin strategies for offering NIPD in clinical practice. Objective Determine service user and provider preferences for key attributes of prenatal diagnostic tests for SCD and examine views on NIPD. Method A questionnaire that includes a discrete choice experiment was used to determine the preferences of service users and providers for prenatal tests that varied across three attributes: accuracy, time of test and risk of miscarriage. Results Adults who were carriers of SCD or affected with the condition (N=67) were recruited from haemoglobinopathy clinics at two maternity units. Health professionals, predominately midwives, who offer antenatal care (N=62) were recruited from one maternity unit. No miscarriage risk was a key driver of decision making for both service users and providers. Service providers placed greater emphasis on accuracy than service users. Current uptake of invasive tests was 63%, whilst predicted uptake of NIPD was 93.8%. Many service users (55.4%) and providers (52.5%) think pressure to have prenatal testing will increase when NIPD for SCD becomes available. Conclusions There are clear differences between service users and health professionals’ preferences for prenatal tests for sickle‐cell disorder. The safety of NIPD is welcomed by parents and uptake is likely to be high. To promote informed choice, pretest counselling should be balanced and not exclusively focused on test safety. Counselling strategies that are sensitive to feelings of pressure to test will be essential.


| BACKGROUND
Sickle-cell disorders (SCD) affect the structure of haemoglobin and result in episodes of acute pain, chronic anaemia and progressive organ damage. Recent improvements in life expectancy for people affected with SCD have been attributed to earlier detection and improvements in comprehensive management. 1 SCD is a relatively common condition, with approximately 12 000-15 000 affected individuals in the UK. 2,3 Whilst the condition is most common in people of African origin, SCD may occur in any ethnic group. 3 SCD is caused by mutations in the β-globin gene. Inheritance is autosomal recessive, so both parents must be carriers of a mutation for the baby to be at risk of SCD. In the UK, there is a universal neonatal screening programme for SCD and carrier screening is offered through an antenatal haemoglobinopathy screening programme that aims to offer screening by 10 weeks in pregnancy. 4 Prenatal diagnosis of SCD is the most frequently requested prenatal test for a single gene disorder in the UK, 5 and currently requires invasive testing (chorionic villus sampling or amniocentesis) which carry a small risk of miscarriage 6,7 and can only be carried out after 11 weeks in pregnancy. A safe alternative is on the horizon, with proof-of-concept studies showing that non-invasive prenatal diagnosis (NIPD) based on analysis of cell-free DNA (cfDNA) in maternal plasma may be possible for SCD. 8 NIPD can be performed early in pregnancy (9-10 weeks) and requires only a maternal blood sample, thereby removing the risk of procedure-related miscarriage associated with invasive diagnostic tests. It is anticipated that NIPD for SCD will be diagnostic and there will be no need to confirm NIPD with invasive testing. As such, NIPD could be considered as a replacement for invasive testing.
It is important that research exploring views of NIPD is carried out alongside the technical development of these new tests to ensure that social and ethical issues are addressed and stakeholder needs are met.
Previous research exploring views of NIPD for single gene disorders, including SCD, has been undertaken using qualitative interviews and focus groups with carriers of single gene disorders and health professionals. [9][10][11] To expand this work, we have utilized a quantitative approach that includes a discrete choice experiment (DCE). DCEs allow us to look closely at people's decision-making processes when making choices about prenatal testing. This is done by giving participants a series of hypothetical options for prenatal tests with differing attributes and asking them to choose between them. Analysing the choices, they make will identify the test attributes that are most important for decision making. We will also gain insights into people's willingness to trade one attribute for another. DCEs have been used to reveal preferences for screening and diagnostic tests for Down syndrome, [12][13][14][15][16][17] and we have also used a DCE to look at preferences for the prenatal diagnosis of cystic fibrosis. 18 Here, we used a DCE to explore service user and health professional preferences for three key attributes of prenatal diagnostic tests for SCD: accuracy, time of test and risk of miscarriage. We also examined views on NIPD, including expected uptake. We tested two hypotheses: 1. Service users and health professionals will differ in their preferences for the three attributes of prenatal tests to be examined: safety, accuracy and time of testing; 2. Service users will value the safety afforded by NIPD and hypothetical uptake of the test will be high.

| Ethical approval
Ethical approval was obtained from a National Research Ethics Service Committee (10/H0714/3).

| Recruitment
Two groups of participants were recruited: (i) service users: adults who were either carriers of SCD or affected with the condition, aged 18 or over and attending haemoglobinopathy antenatal clinics at either St Thomas's Hospital or University College London Hospital, and (ii) service providers: health professionals who deliver antenatal care and see women with pregnancies at risk of SCD at St Thomas' Hospital.
Convenience sampling was used to participants to the study. For the service users, group potential participants were invited to anonymously complete the questionnaire whilst waiting for their clinical appointment. For the service provider group, potential participants were approached in person in their workplace and invited to anonymously complete a hard copy of the questionnaire.

| Questionnaire design
The questionnaire had three components: (i) DCE choice sets, (ii) structured questions about prenatal testing and NIPD and (iii) demographic questions. Questionnaire design has been described previously, as the questionnaire used here was a modified version of the questionnaire used in our previous study looking at prenatal testing for cystic fibrosis. 18 Briefly, design of the choice sets followed DCE guidelines [19][20][21] and the attributes of safety, accuracy and time of test results were derived from focus groups with carriers of single gene disorders (SCD, cystic fibrosis and thalassaemia). 10 The attributes and levels used in the DCE choice sets are presented in Table 1

| Analysis
DCE analysis followed published guidance. [19][20][21] The DCE preference data were analysed using a conditional logit regression model as previously described. 18 As the sign (+ or −) of the coefficients from the regression analysis shows the direction of the preference,

| Participants
Questionnaires were completed by 78 service users and 62 health professionals. Questionnaires were excluded if the internal con- A small proportion had a child with SCD (16.4%). Health professionals were predominately midwives (88.1%) and female (91.8%).
Demographic information for service users and providers is summarized in Tables 2 and 3.

| Regression results
Service users and providers prefer a test with greater accuracy, early testing and no risk of miscarriage (

| Marginal rates of substitution
The MRS demonstrates the strong preference held by both service users and health professionals for a test with no risk of miscarriage.
Notably, service users were prepared to wait longer and accept lower accuracy compared to service providers for a test that had no risk of miscarriage (Table S1).

| Ranking of attributes
Participants ranked the importance of five attributes of prenatal tests: safety, early testing, accuracy, financial cost and comprehensive information (Tables S2 and S3). Safety was ranked highest by service users (43.2%) followed by full information (21.6%). Accuracy was ranked highest by health professionals (45.6%) followed by safety (42.1%).
Cost was ranked lowest by both service users (86.5%) and health professionals (57.9%).

| Views on prenatal testing and the introduction of NIPD
When service users were asked whether they would have invasive testing for SCD, 63.1% said they have had or are likely to have invasive testing, with the most common reasons for testing being to "help make a decision about whether or not to continue the pregnancy" (47.4%) and "to plan and prepare for the possible birth of a baby with SCD" (36.8%) (

| DISCUSSION
The aim of this study was to determine the preferences held by potential service users and providers for prenatal tests for SCD. Test safety was an important driver of decision making for both groups.
Differences were evident in the emphasis placed on test accuracy, with service providers placing greater emphasis on this attribute than service users. It was clear that service users were prepared to wait longer and accept lower accuracy than service providers for a test with no risk of miscarriage. In addition, over half of the service users chose tests based only on test safety and did not consider the other attributes. Differences between health professionals and potential service users have been seen in other DCE studies comparing noninvasive and invasive tests for cystic fibrosis 18 and also for Down syndrome. 15 of partners had carrier screening. 28 It is possible that the number of women opting for NIPD for SCD may increase further because women whose partners decline carrier testing may find NIPD more acceptable than invasive testing when the exact risk of the child inheriting SCD is unknown. 10,11 Overall, it appears that uptake of NIPD for single gene disorders will be high and many couples who would not consider invasive testing due to the risk of miscarriage would have NIPD.
The changes to the population accepting the offer of prenatal testing need to be considered in our approaches to counselling for NIPD for SCD. Most notably, many couples who would not previously have had prenatal testing may take up NIPD, and as a result may find themselves faced with a decision about termination of pregnancy.
This issue is particularly important in the setting of prenatal testing for SCD where in the UK and other countries a prenatal carrier screening programme is in place and there are existing concerns about informed consent processes 29 and carrier screening being presented as routine. 30 Without careful pretest counselling, the addition of NIPD to the care pathway has the potential to undermine informed consent as women newly identified as carriers of SCD may see NIPD as routine next step following a positive carrier screening result. 31 As such, it will be especially important to allow sufficient time in pretest counselling to talk through the impact of having NIPD which includes a discussion around the implications of the possible test results that is guided by the parents values and preferences. 32 It must be made clear to parents that accepting a "simple blood test" could lead to a decision about whether to continue or terminate the pregnancy. Individualized support through post-test counselling to assist decisions about next steps will be essential.
The cost of NIPD for SCD will need to be considered in strategies to implement this test in the NHS. The potential cost of NIPD relative to the current invasive testing pathway has been explored and NIPD for SCD using current approaches was estimated to cost £1210, which was £190 more than invasive testing. Moreover, the anticipated high uptake of NIPD we found in this study (approximately 95% compared to 65% for invasive testing) would result in an incremental cost of NIPD over invasive testing of £48,635 per 100 pregnancies at risk of SCD. As the increased uptake will include parents who would want to have NIPD for information only and would not consider termination of pregnancy, there is a need to address the issue of whether it is appropriate to direct resources to test when pregnancy management would not change. Consideration of this question must include the benefits of the information for early reassurance or for planning and preparation if the baby is found to be affected by SCD. 31 In the study, we report here only a very small proportion of health professionals (6.6%), when asked, thought NIPD should only be offered if it would change pregnancy management.

| Limitations
Several limitations of our study may mean that our findings are not widely generalizable. Recruitment was only conducted at two centres for service users and only one centre for service providers and both centres were located in central London. As this was not a random sample and convenience sampling was used, it is possible that sampling bias will limit the generalizability of our results. In addition, participants were self-selected and there may be responder bias towards people with strong pre-existing views on NIPD. Another limitation of the study was that the numbers of participants recruited were not sufficient to allow subgroup comparisons. In an equivalent study looking at NIPD for cystic fibrosis, we found there were differences in preferences between people affected with the condition and those who were carriers. 18 In addition, only a small number of men were recruited and their viewpoints may differ to those of women.

| CONCLUSIONS
When making decisions about prenatal testing for SCD, potential service users and providers do not place the same emphasis on the test attributes. It is likely that the safety of NIPD will be welcomed by parents and uptake will be high. It is therefore important that pretest counselling is balanced and not predominately focused on the safety of NIPD. Care must also be taken to minimize feelings of pressure to have NIPD. Considerations for implementation need to include current carrier screening and prenatal diagnosis pathways. Offering NIPD as a next step in the current pathway could create pressure to test at a time when news of carrier status is still being processed and decisions need to be made quickly about next steps. Thorough pre-and post-test counselling will be essential and NIPD should be offered by health professionals specifically trained in counselling for prenatal testing for SCD.

ACKNOWLEDGEMENTS
We are grateful to the patients and health professionals who participated in this study. We also thank the staff assisting with the research.