How is the accuracy of NIPT measured?
This is a more detailed response to the question, “How accurate is NIPT?”.
You are welcome to read this section (that is why we have written it!), but It is not necessary that you do. Your doctor will provide you with the necessary information and advice.
When a laboratory does an NIPT, the result is presented as a yes/no answer i.e. “yes, there is sufficient evidence to recommend further assessment of this pregnancy” or “no, there is insufficient evidence to recommend further assessment of this pregnancy”.
The ideal test would only provide “true” results: all affected pregnancies would have a “yes” result, and all unaffected pregnancies would have a “no” result”. In practice, this is not feasible. Incorrect results can occur for a variety of reasons. In particular, the chromosome status of the placenta (the source of the DNA used in NIPT) may differ from that of the fetus.
For this reason, there are four possible outcomes from an NIPT:
It is possible to adjust the reporting of NIPT to reduce the chance of either a false positive or a false negative result, but it is usually not possible to reduce the chance of both at the same time. There needs to be a trade-off between the chance of getting a false negative result (and missing an affected pregnancy) and getting a false positive result (and potentially having an unnecessary amniocentesis).
For each part of an NIPT e.g. the test for trisomy 21, for trisomy 18, for a chromosome deletion etc, the trade-off is different. These differences reflect the performance of the test and the consequences of getting a false result.
The trade-offs established by most providers of NIPT for trisomy 21 in Australia would result in the following outcomes for a population of 100,000 women. The approximate number of women with each type of result would be as follows:
This means that almost all of the women with an affected pregnancy (298 out of 300) would be correctly identified by NIPT, but about a quarter of those with an abnormal NIPT result would, in fact, have an unaffected pregnancy (99 out of 397). It would be possible to adjust the test to reduce the number of false positives, but this would increase the number of false negatives. It is not possible to reduce the number of false negatives any further because of those rare pregnancies in which the placenta and the fetus have different numbers of chromosomes.
There are statistical terms for describing the relationship between these four outcomes. These terms are summarised in the Table, along with a calculation based on the example above.
You will note that sensitivity and specificity refer to the performance of the test overall i.e. from the laboratory’s perspective. These are the principle measures used by the laboratory in validating a new test. These measures are provided for each component of NIPT on our website and on test reports.
The positive and negative predictive values refer to the performance of the test from the perspective of an individual patient (and her doctor). The laboratory cannot provide a precise measure of the positive or negative predictive value for a particular patient. This is because there are a variety of patient-specific factors such as maternal age, paternal age, family history, ethnicity etc, that can alter the predictive value of a test – and this information is not known to the laboratory.
For these reasons, NIPT is regarded as a screening test. It does not provide a definitive diagnosis. NIPT is remarkably accurate, but it is not perfect.
Your doctor will include the result of your NIPT with other clinical, imaging and pathology information to advise you about the appropriate management of your pregnancy.