The integration of inherent uncertainty in prenatal testing for chromosome abnormalities into decision-making processes about test results.

Susman, Marleen

Every pregnant woman wants a healthy baby and for the vast majority of women this is what occurs. There are some women, however, who know they are at risk for fetal abnormality or receive an abnormal result after prenatal testing. Women who know they are at risk can choose to prevent the birth of a baby with a genetic abnormality by deciding not to have children or to have pre-implantation genetic diagnosis with IVF. More commonly, however, they undergo prenatal testing once they are pregnant, hoping to get a normal result. Women who receive abnormal results after prenatal testing have a choice to either continue with the pregnancy and prepare for the birth of a baby with a genetic abnormality, or to terminate the pregnancy. Opinions differ on the appropriateness of using termination to prevent the birth of a baby with a genetic condition, but someone who wants to access this option will understandably seek certainty about the clinical and laboratory outcomes that inform her decision. The desire for certainty is just as strong for results giving confirmation that the baby has 'nothing wrong' as it is for assurance that an abnormal result is accurate and the outcome from that result predictable.

This article considers the role of uncertainty in relation to both current and potential future prenatal testing from laboratory and clinical perspectives. Incorporating and managing uncertainty, in the context of prenatal testing for chromosome abnormalities, is also addressed. The idea that laboratory investigations may be a source of uncertainty may come as a surprise to many people, especially as it is often perceived to be providing certainty to clinical diagnosis. There is a range of different prenatal tests available, but the focus of this article is on the detection of chromosome abnormalities, including Down syndrome, as the vast majority of prenatal testing is done for this reason. There are two different types of tests available: one is a screening test indicating the risk of a particular pregnancy having Down syndrome, while the other diagnoses Down syndrome. The screening test is performed on a maternal blood sample and noninvasive ultrasound measurements of the baby, whilst diagnosis requires a sample of fetal tissue, necessitating an invasive procedure under ultrasound guidance to remove the sample.

UNCERTAINTY IN THE LABORATORY--CURRENT TESTING

Current test strategies to detect chromosome abnormalities prenatally include Down syndrome screening tests, ultrasound and diagnostic tests. Pregnant women in Victoria are offered a screening test for Down syndrome as a routine part of their prenatal care. The majority of pregnant women in Victoria choose to have a test at ten to twelve weeks gestation in their first trimester, which combines the results of an ultrasound, where the nuchal translucency (a fluid-filled pouch at the base of the neck) of the fetus is measured, with results coming from the biochemical analysis of the maternal blood. The test is designed to screen for Down syndrome. The results from these tests are presented as an individual risk figure with increased risk considered to be a risk greater than one in three hundred. Screening tests, by definition, do not give a definitive diagnostic result and there is inevitable uncertainty as incorporating both false positive and false negative results is part of the test strategy. A false positive result occurs when a woman is told she has an increased risk of a Down syndrome pregnancy despite the pregnancy being normal, while a false negative result means that a low-risk result is given to a woman whose baby has Down syndrome.

On the basis of these results, women decide whether they want to have a diagnostic test. The diagnostic test, which gives a definitive result, involves an invasive procedure--either chorionic villus sampling (CVS) at twelve weeks or amniocentesis at sixteen weeks, both of which have associated miscarriage risks. The diagnostic test performed on the specimens taken at either procedure is called a karyotype. This test is done in the laboratory and involves microscope analysis of all twenty-three pairs of chromosomes to ensure that both the number and the structures are 'normal'. As it covers all the genetic material in a cell it is called a genome-wide test. The karyotype analysis requires visual recognition, relying on the experience and expertise of the analyst. Due to international standardisation of what is considered 'normal', there is usually consensus between scientists about a result being normal, but this does not exclude the possibility of disagreement over some karyotypes. (1) Guidelines for analysis acknowledge this by recommending that no result is ever released unless a minimum of two scientists have been involved. (2) This is similar to the degree of uncertainty pertaining to the reliance on human expertise when specialist doctors interpret visual images from ultrasound or x-rays to determine if there is abnormality present. In addition, the standard for determining the karyotype of a fetus requires a minimum of fifteen cells to be analysed. (3) The results of these fifteen cells are assumed to represent all the other cells in the fetus. This number of cells for analysis has been derived statistically to give a good probability that they do represent the other cells, but it does not provide certainty.

Different qualities in chromosome preparations translate into different levels of analysis and higher levels of analysis reveal more abnormalities. This means that the accuracy of the result is dependent on the quality ofthe chromosome preparation. Guidelines also acknowledge that this difference in quality exists and suggest that the quality of the preparation is reported along with the result. (4) Doctors who discuss these results with patients, however, may not understand the significance of the quality statement, or may not convey this to the patient. Either way, pregnant women do not become aware of this uncertainty. In addition, there are different degrees of uncertainty associated with the result depending on which procedure is chosen, due to the differences between using chorionic villi or amniotic fluid as the source of cells for testing.

Chorionic villi are pieces of tissue taken from the placenta, whilst the fetal cells in the amniotic fluid sample come from the fetus. The embryological development from the fusion of a sperm and egg through to a fully formed fetus occurs on a predetermined pathway. The placenta is also part of this development. The cells that contribute to the formation of the placenta, however, originate from a different subset of cells to those that go on to form the fetus. This can result in the placental cells having a different genetic make-up to the fetal cells. In addition, it is possible for there to be a mixture of normal and abnormal cells in either or both the placental and fetal cells. This is called mosaicism. Figure 1 shows the various possibilities of how these cells can be distributed. The result from a CVS specimen will reflect the cells in the placenta; the results from an amniotic fluid specimen will reflect the cells from the fetus. An abnormal outcome for the fetus usually only occurs when there are some abnormal fetal cells. In Figure 1, the situations in which the fetus may be abnormal and the result from the amniocentesis is abnormal are numbers two, four, five, six, seven and nine. Abnormal CVS results, however, would be reported for situations numbers two, three, four, five, six and eight. Thus situations number three, seven and nine are specimen-dependent as to what result will be reported. Conversely, normal CVS results would be reported for situations number one, seven and nine, whilst normal amniocentesis results would be reported for situations number one, three and eight. Again, there are situations (three, seven, eight and nine) for which the results will differ depending on the specimen used.

[FIGURE 1 OMITTED]

1. Complete fetal-placental concordance--fetus and placenta both have only normal cells

2. Complete fetal-placental concordance--fetus and placenta both have only abnormal cells

3. Confined placental mosaicism--fetus has only normal cells, placenta has a mixture of normal and abnormal cells

4. Fetal-placental mosaicism--both the fetus and the placenta have a mixture of normal and abnormal cells

5. Non-mosaic fetus, mosaic placenta--fetus has only abnormal cells, placenta has a mixture of normal and abnormal cells

6. Fetal mosaicism, non-mosaic placenta--fetus has a mixture of normal and abnormal cells, placenta has only abnormal cells

7. Fetal mosaicism, normal placenta--fetus has a mixture of normal and abnormal cells, placenta has only normal cells

8. Complete fetal-placental discordance--fetus has only normal cells, placental has only abnormal cells

9. Complete fetal-placental discordance--fetus has only abnormal cells, placenta has only normal cells

In addition to this difference between the genetic make-up of fetal and placental cells, there are also differences in the quality of the chromosome preparations, between CVS and amniotic fluid samples, with amniotic fluid samples usually giving a higher quality result. As mentioned above, this matters because better quality preparation allows for a more detailed analysis, giving a better prospect of finding an abnormality. The difference in quality of chromosome preparations, due to the choice of procedure, is often not understood by the referring doctor or the pregnant woman.

The quality of the chromosome preparation can also be affected by the quality of the cells in the original sample. To prepare CVS and amniotic fluid specimens for a karyotype involves growing the cells in tissue culture. To achieve successful cell growth involves an experienced scientist making many subjective decisions based on their visual appraisal of the cells. This job is made more difficult if the quantity or quality of the original sample is poor.

There are some aspects of specimen quality that can be controlled and some that cannot. Obstetricians, who are experienced in ultrasound, take the CVS and amniotic fluid samples. They have control over the type of equipment used (specimens are taken using ultrasound guidance), and the skill of the operator is also controllable, but the position and accessibility of the fetus, which can make collecting a good sample difficult, is patient-dependent.

As mentioned above, for a karyotype to be prepared, the cells from CVS and amniotic fluid samples require tissue culturing. During the culturing process, it is possible that an error can occur in cell division, which can lead to a situation in which some of the fifteen cells analysed are abnormal. This situation looks the same as the mosaicism described above but in this case the abnormal cells arise during the culturing process, rather than reflecting the fetal or the placental karyotype. It is also possible for a few maternal cells to get mixed in with either the fetal or the placental cells when the specimen is taken. Sometimes these maternal cells may grow alongside the other cells mimicking a mosaic result. There are standard ways for the laboratory to try to determine whether the observed mixture of abnormal and normal cells is due to the culturing process or to the presence of maternal cells, but there is no certainty--only a reduction of the uncertainty.

UNCERTAINTY IN THE LABORATORY--FUTURE TESTING

As both CVS and amniocentesis have associated miscarriage risks due to the invasiveness of the procedures, being able to provide non-invasive prenatal diagnosis (NIPD) has been the Holy Grail of prenatal diagnosis. With the recent discovery of short lengths of DNA found in maternal blood, this may soon become a reality. (6) NIPD could provide a definitive diagnosis to all pregnant women, instead of the uncertainty of the risk figures currently provided by screening tests. It would also negate the need for tissue culturing and all its associated uncertainties. It would not, however, address the problems of uncertainty related to using placental cells to determine fetal genetic make-up, as the short lengths of DNA actually originate from placental cells.

Increasing the uncertainty with NIPD is the technical reality that a karyotype is not possible on these short lengths of fetal DNA. The tests that are being developed use a range of technologies, but common to all of them (at present) is the need for the test to be targeted. This means that the test will only detect the abnormality it is designed to detect, such as Down syndrome--it will not provide the whole genome coverage of a karyotype. Using population-based data from Victoria, Australia, we conducted a study to determine what changes would have occurred to the number and types of chromosome abnormalities if NIPD had replaced Down syndrome screening programmes for the years 2006 and 2007. Our results indicate there would have been a small increase in the number of Down syndrome pregnancies detected, but a much larger decrease in the number of non-Down syndrome abnormalities detected. (7)

NIPD may therefore avoid the uncertainty in relation to many of the aspects involved in preparing a karyotype, as well as eliminating the uncertainty of screening tests, but it will increase uncertainty in regard to the presence of chromosome abnormalities other than those included in the test. An NIPD test for Down syndrome will not give the same degree of certainty about the health of the baby as karyotyping does, as it will only give a result about Down syndrome. There could still be something wrong with the remaining 22 pairs of chromosomes.

Another test that is already being offered in some circumstances is chromosomal microarrays (CMA). This test is similar to karyotyping in that it provides a test that gives genome-wide coverage and covers many different conditions, but offers even more detail than a karyotype. There are two types: targeted CMA, where only conditions that have a well-established underlying genetic error are included, and genome wide CMA, where any changes from normal are identified. At this point in time CMA still requires a CVS or amniotic fluid specimen to be taken as procedure-associated miscarriage remains a concern. More information is usually considered a 'good thing', and the ability to detect an even greater range of abnormalities could decrease the uncertainty about delivering a 'normal baby'. However, this also runs the risk of delivering an increased number of uncertain results where the clinical outcome cannot be accurately determined.

The uncertainty associated with the scientific aspects of prenatal testing for chromosome abnormalities is often 'black boxed'. Laboratory directors accept that making decisions about laboratory uncertainty is part and parcel of their role. Using their expertise, they come to a conclusion about a test result and only report what they consider to be clinically relevant. This may differ between directors, but again there are professional guidelines and quality assurance programs to ensure a reasonable degree of standardisation. Providing a degree of certainty about test results serves the doctors' interest in wanting to minimise anxiety for their patients. Women also want a degree of certainty about their test results and test accuracy is often mentioned in response to being asked about the advantages and disadvantages of having a prenatal test for chromosome abnormalities. (8) The problem with this approach is that not only do pregnant women not realise the decisions that are being made in the laboratory, but many health professionals are also unaware of the degree of uncertainty behind the results and reports they receive. Most of the time this does not create a problem, as the outcome and test result concur or, whatever the pregnancy outcome, neither the woman nor her doctor question the prenatal result. It is when it does not concur and questions are raised about the process that people are surprised to find out that the result had a degree of uncertainty about it. When there has been an unfortunate outcome and people are looking for someone to blame, the law may be involved and it employs its own definition of certainty, which is 'beyond reasonable doubt'. This definition can be at odds with how both laboratory directors and doctors, who can only ever make decisions based on the best available evidence, manage certainty in their practice. (9)

Compounding the black box around uncertainty of current and future clinical laboratory tests is another layer of uncertainty that arises in the clinical setting.

UNCERTAINTY IN THE CLINIC--CURRENT ISSUES

Once an abnormal result has been issued from the laboratory there is still the important task of trying to predict what the clinical outcome will be for that particular baby. A karyotype result gives information about chromosomes; it does not necessarily tell us what that means for the outcome. Any change to the chromosomes is a differentiation from normal and can be classed as an abnormality, but this refers only to the chromosomes not being normal. It does not imply that the clinical outcome will necessarily be abnormal. Changes to the chromosomes can occur whereby they are rearranged in a different order but there is no gain or loss of material. This is called a balanced rearrangement. In most of these situations, the clinical outcome will be normal. There can also be a situation in which there is additional material but the additional material does not include any critical genes. Once again the clinical outcome may be normal. There have also been reports of material being deleted where the clinical outcome is normal. And there can also be abnormal chromosome conditions where the large range of clinical outcomes can include a normal presentation. Conversely, there are times when the chromosomes appear normal but the clinical outcome is abnormal. This may be due to the quality of the karyotype being insufficient to detect the presence of an abnormality or where the clinical outcome is not related to the karyotype but is due to a mutation in a gene. To detect this kind of abnormality requires a different test.

When babies, children or adults have karyotype testing it is in response to a problem they present as patients. The chromosome result can then be interpreted in conjunction with a clinical examination. A major difference with prenatal testing is that the karyotype is used to predict the outcome. The clinical examination of the fetus is very limited by comparison; it includes only what is visible by ultrasound and results from tests to monitor the pregnancy. In particular, there is no way of predicting the degree of intellectual disability when a chromosome abnormality is detected.

The degree of uncertainty about predicting outcomes after an abnormal chromosome result varies considerably depending on the abnormality. For a diagnosis of Down syndrome, the phenotype ranges from fetal death to independent living as an adult. There are common phenotypic features which make this condition recognisable at birth, but as the website for Down Syndrome Victoria says, 'What happens after birth will be more important in shaping the outlook for a person with Down syndrome, than the occurrence of the extra chromosome at conception'. (10) Even more problematic are results in which the outcome can only be based on a handful of published case reports which may have contradictory findings, or in which educated guesses are made in the total absence of case reports. An educated guess based on years of clinical and laboratory experience with discussion between the laboratory director and the doctor provides the best possible prediction, but it does not provide certainty and it is hardly beyond reasonable doubt.

UNCERTAINTY IN THE CLINIC--FUTURE ISSUES

The introduction of NIPD and CMA testing will also impact on uncertainty in the clinic. Given the differences already discussed between karyotyping and the targeted testing of NIPD, it will be essential to have adequate education of health professionals to ensure that they understand what these differences are. In particular, it will be important for doctors to realise that the NIPD result is only relevant for the targeted conditions and gives no information about other chromosome abnormalities. NIPD may be seen as a replacement test for Down syndrome screening tests, in which case the degree of certainty regarding whether the fetus has Down syndrome is increased. It might also, however, be seen as a replacement test for current diagnosis with a karyotype, in which case women may be given a false impression about the degree of certainty a normal result implies. Alternatively it could be introduced in addition to the current tests, adding to the current layers of uncertainty.

On the other hand, CMA may be considered the replacement test for a karyotype, in which case the certainty of a normal result is increased as MCA will detect additional abnormalities. (11) Depending on the type of MCA test used, it may also lead to a greater number of results of unknown clinical significance. Whether new technologies increase the certainty of results depends to some extent on how they are implemented.

MANAGING UNCERTAINTY IN PRENATAL TESTING

When a woman has a positive pregnancy test and goes to her GP to have it confirmed, the first questions she has to answer are whether she will have any prenatal testing for Down syndrome, and which tests she will choose. As discussed above, the range of tests that she can currently choose from include screening tests, requiring a blood sample from the mother and an ultrasound, or diagnostic tests that involve an invasive procedure and the possibility of miscarriage. Women who want diagnostic testing choose between CVS and amniocentesis. A major advantage of CVS is that it can be done at an early gestation (twelve weeks) while amniocentesis has a lower risk of miscarriage (one in a hundred for CVS and one in two hundred for amniocentesis). Making an informed choice about these tests necessitates that in addition to understanding the timing and safety of the two invasive tests there is complete disclosure about the accuracy of the result and the accuracy of the predicted outcome from that result. Sometimes women only find out how accurate the test is when they receive an unexpected equivocal abnormal result. As has already been discussed, a test result may be 99.9 per cent accurate in terms of the chromosomal abnormality described, but there may be a different degree of accuracy in relation to whether the analysis was performed on fetal cells, whether the fifteen cells analysed actually represent the entire fetus, and whether the clinical outcome of the chromosome abnormality can be accurately predicted from the karyotype. In choosing between these tests, women need to weigh up the value for them of the information the test provides against the miscarriage risk due to the procedure and the timing of the test, and against the potential decision about termination if they receive an abnormal result. Women have these tests to exclude abnormality. For some this translates into wanting to create certainty around the birth of a normal baby, which can translate into a quest for the perfect baby. (12) What this actually means to any one person is subjective and defining 'normal' has always been and remains fraught with difficulty. There are no guarantees of a normal baby no matter how many tests a woman has. When women choose these tests they need to be told not only what the test will tell them, but also the degree of certainty about that result and the predictability of the clinical outcome.

Much of the uncertainty related to current testing strategies is hidden. Laboratory directors are acutely aware of the uncertainty in their domain but often this is not discussed outside the laboratory. Doctors are very aware of the uncertainty involved when they make a clinical diagnosis, but may choose not to relay this information to their patients. It has been noted that the relationship between doctor and patient is relevant to both their perceptions about the certainty of a diagnosis, and also that the effect of disclosure in relation to uncertainty increases a patient's anxiety. (13) With both of these facts in mind, it is no wonder that doctors choose not to relay to patients the degree of uncertainty around their diagnosis. Compounding this non-disclosure, patients are aware at one level that nothing in life is certain, and yet at another level, whether rational or not, expect doctors and laboratories to provide them with certainty. So, in some ways, laboratory directors, doctors and patients are all colluding in their reluctance to accept the reality of uncertainty in relation to prenatal testing for chromosome abnormalities and outcomes of pregnancy.

It is interesting to ponder why there is such a push for the introduction of new technology in laboratory and clinical diagnosis. There is no doubt that the desire to provide increasing certainty about results is one underlying reason. Proof of principle publications about new technologies consistently mention how consumers will welcome new tests such as NIPD, as it removes the risk associated with an invasive procedure, (14) and likewise for MCA, which has improved diagnostic strength compared to karyotyping. (15) As has already been pointed out, it isn't clear that the assumption that these technologies will provide more certainty is justified. In fact, although some uncertainties disappear, the new technologies bring new problems of uncertainty. It has been noted that excessive diagnostic testing sometimes results from the failure of doctors to make decisions where there is uncertainty. (16) This may equally apply to the hunt for better diagnostic tools. That is not to suggest that we shouldn't embrace the benefits of new technology, but rather that it needs to be introduced in an environment that is accepting of uncertainty and of the inability of medical diagnosis to ever being one hundred per cent accurate.

In the context of prenatal testing, in which women's autonomous decision making is considered paramount, the doctors' failure is not in failing to make decisions but rather in failing to assist women in making decisions that involve elements of uncertainty. As Djulbegovic notes, 'Evidence is expressed on a continuum scale of credibility, whereas decision making is about choice and is a categorical exercise--we decide or do not.' (17) It is not helpful for uncertainty to be laid on the table without some assistance from an expert as to how to make an appropriate decision. The catchcry of the genetic counselling profession has for many years been the need for it to be 'non-directive'. (18) Non-directive counselling has been defined as giving value-neutral information. This has been challenged both from a theoretical and from a practical standpoint. (19) A survey of counsellors and their clients indicated that counselling should be tailored to individuals' needs and that successful decision making contributes to decreasing anxiety. (20) Both of these conclusions are relevant to the successful integration of uncertainty into prenatal decisions. As there will always be individual difference in what uncertainty means, and in individual responses to it, (21) individualising how uncertainty is integrated into the doctor-patient relationship is essential. It is also reasonable to extrapolate if successful decision making contributes to decreasing anxiety and acceptance of a degree of uncertainty improves the decision-making process. That is, contrary to current perceptions, acceptance of uncertainty may lead to a decrease in anxiety. Backing up this claim, Wellbery suggests that 'grappling with uncertainty can paradoxically allow for a clarification of values'. (22) Integrating an individual's values into prenatal decisions is considered to be part and parcel of a genetic counsellor's role. (23) If accepting uncertainty encourages decision making that respects and includes a woman's value system, it will inevitably lead to better decisions. So how do doctors or genetic counsellors assist patients in accepting uncertainty in prenatal testing, regardless of new technology or better diagnostic tests, to assist them in making successful decisions, ones that reduce rather than increase their anxiety?

Practical suggestions for doctors to improve how they cope with uncertainty include trust, honesty, awareness and kindness as essential components of the doctor-patient relationship. (24) I suggest that this may not be sufficient, as it is reasonable to assume that these traits exist in most doctor-patient relationships already, coexisting with doctors' reluctance to relay the complexity of uncertainty that exists around prenatal testing. It has also been suggested that 'our leaders and the public understand the inherent limitations of medical knowledge and the role of research in reducing uncertainty'. (25) This is all well and good, but reducing uncertainty does not solve the practical problem of how to deal with it, even if it is reduced. A recent opinion piece in The Lancet sheds more insight into the integration of uncertainty in medicine by suggesting:

   Art can demonstrate the interpretive process that leads to
   resolution of uncertainty: first, by defining its precise nature;
   second, by identifying which information is lacking; third, by
   recognising that interpretation is a process that occurs over time;
   and finally, by knowing the contributions of various aspects of the
   context. (26)

At a practical level, this implies that doctors need to ensure that they are familiar with all the aspects of uncertainty, including the laboratory aspects, and that they understand that interpreting results is an art in which opinions may differ and change over time. This self-reflection may help them be more comfortable with uncertainty, which will in turn be reflected in their relationships with their patients. An editorial in the same journal, reflecting on this article, suggests that a doctor's role (and this equally applies to a genetic counsellor's role) should contain both objective and subjective approaches, necessitating that training includes both the science and the art of medicine to ensure that they can minimise uncertainty while being comfortable with the uncertainty that remains. (27)

If all of these suggestions are taken together--a trusting and honest relationship between doctor and patient, public awareness of the uncertainty in medical diagnoses and doctors' self reflection and ability to be comfortable with a degree of uncertainty--then it is more likely that uncertainty will be part of the decision making, ultimately leading to better decisions that relieve rather than exacerbate anxiety for both doctor and patient.

CONCLUSION

Uncertainty is a part of life and an inevitable part of medical diagnoses. The particular elements that contribute to uncertainty are numerous, both from a laboratory and from a clinical perspective, and for both current and future prenatal testing for chromosome abnormalities. Although new technology will improve the ability to diagnose abnormalities in some instances and provide diagnosis for some abnormalities in a risk-free environment, it will not provide certainty. It is not possible for any diagnostic test to ever provide complete certainty, so although reducing uncertainty is a positive goal for research, acceptance of a degree of uncertainty will always be an essential part of medical practice. In prenatal testing there will always be the issue of having to try to predict the clinical outcome on the basis of the test result, whether it be a karyotype, NIPD or MCA. Denial of its existence is not beneficial for doctors or for patients. In fact, it has been demonstrated that acceptance and integration of uncertainty into decision making is beneficial to patients, resulting in decisions that reduce anxiety. Acceptance of uncertainty should not impede the development or acceptance of new technology either, but should lead to appropriate understanding and education about the uncertainties that new tests may bring, alongside the acknowledgement of the uncertainties that may disappear.

ENDNOTES

(1) L G Shaffer, M L Slovak and L J Campbell, ISCN 2009 An International System for Human Cytogenetic Nomenclature (2009), Karger in collaboration with Cytogenetic and Genome Research, 2009.

(2) Requirements for Cytogenetic Testing--2007: National Pathology Accreditation Advisory Council, Australian Government Department of Health and Ageing, 2007.

(3) Requirements for Cytogenetic Testing.

(4) Requirements for Cytogenetic Testing.

(5) R J Gardner and GR Sutherland, Chromosome Abnormalities and Genetic Counseling, Third Edition, Oxford University Press, New York, 2004.

(6) Y M Lo, 'Circulating nucleic acids in plasma and serum: an overview', Annals of the New York Academy of Science, vol.945, September 2001, 1-7.

(7) M R Susman, D J Amor, E Muggli, A M Jaques and J Halliday, 'Using population-based data to predict the impact of introducing noninvasive prenatal diagnosis for Down syndrome', Genetics in Medicine, vol.12, no.5, 2010, 298-303.

(8) M Susman, Exploring what pregnant women, in the first half of their pregnancy, want to know about their baby's chromosomes during pregnancy, 2010.

(9) R Hayward, 'Balancing certainty and uncertainty in clinical medicine', Developmental Medicine & Child Neurology, vol.48, no.1, 2006, 74-7.

(10) Down Syndrome Victoria Website, 2010, http://www.dsav.asn.au/

(11) L Rickman, H Fiegler, C Shaw-Smith et al., 'Prenatal detection of unbalanced chromosomal rearrangements by array CGH', Journal of Medical Genetics, vol.43, no.4, 2006, 353-61; L G Shaffer, J Coppinger, S Alliman et al., 'Comparison of microarray-based detection rates for cytogenetic abnormalities in prenatal and neonatal specimens', Prenatal Diagnosis, vol.28, no.9, 2008, 789-95.

(12) Susman; L Remennick, 'The quest for the perfect baby: why do Israeli women seek prenatal genetic testing?' Sociology of Health & Illness, vol.28, no.1, 2006, 21-53.

(13) R Hayward 'Balancing certainty and uncertainty in clinical medicine', Developmental Medicine & Child Neurology, vol.48, no.1, 74-7.

(14) Cell-free fetal nucleic acids for non-invasive prenatal diagnosis, Report of the UK expert working group, PHG Foundation, http://www.phgfoundation.org/download/ffdan/ffDNA_report.pdf. Accessed 28 October 2009.

(15) Rickman et al.; Shaffer et al.; Coppinger et al.

(16) J P Kassiere, 'Our stubborn quest for diagnostic certainty. A case of excessive testing', New England Journal of Medicine, vol.320, 1989, 1489-91.

(17) B Djulbegovic, 'Lifting the fog of uncertainty from the practice of medicine', BMJ, vol.329 (2004):1419-20.

(18) P S Harper, Practical Genetic Counseling, Wright, Bristol, 2004.

(19) C A Rentmeester, 'Value neutrality in genetic counseling: an unattained ideal', Medicine, Health Care and Philosophy, vol.4, no.1, 2001, 47-51.

(20) A Toth, T Nyari and J Szabo, 'Changing views on the goal of reproductive genetic counselling in Hungary', European Journal of Obstetrics Gynecology and Reproductive Biology, vol. 137, no. 1, 2008, 3-9.

(21) C Wellbery, 'The value of medical uncertainty?' The Lancet, vol. 375, 2010, 1686-7.

(22) Wellbery.

(23) Counsellors ASoG, Guidelines for the Practice of Genetic Counselling--Version 1, 2008.

(24) R Hayward, 'Balancing certainty and uncertainty in clinical medicine', Developmental Medicine and Child Neurology, vol.48, no.1, 2006, 74-7.

(25) I Chalmers, D Sackett, C Silagy, 'The Cochrane collaboration', in A Maynard and I Chalmers (eds), Non-random Reflections on Health Services Research, BMJ, London, 231-49.

(26) Wellbery.

(27) 'Uncertainty in medicine', The Lancet, vol.375, 2010, 1666.

Marleen Susman

Public Health Genetics, Murdoch Childrens Research

Institute, Royal Children's Hospital