Non-invasive fetal RHD genotyping tests: a systematic review of the quality of reporting of diagnostic accuracy in published studies.

Articles reporting the diagnostic accuracy of non-invasive prenatal diagnostic (NIPD) tests for RHD genotyping using fetal material extracted from maternal blood have been published steadily for over a decade. Health care providers in Europe have started to use this technology for management of the small number of sensitised pregnancies (ca. 220-600 per annum in the Netherlands, Germany, France and the UK). Scientists and clinicians are also advocating widespread implementation for the far larger number of non-sensitised RhD-negative pregnancies (ca. 34,000-125,000 per annum in the same countries). Large-scale, prospective trials are only now underway. Estimates of the technical performance of these tests are currently based on results from small-scale studies, together with formal meta-analysis. The issue of early assessment of test performance is one faced by many new genetic tests. As part of a wider study we have investigated the quality of reporting of diagnostic accuracy in publications and produced guidelines for future studies. A systematic search of the literature identified 27 papers which met predefined inclusion criteria. All 27 papers were, first, assessed against an international quality (STARD) checklist for reporting of diagnostic accuracy and, second, against our own in-house NIPD proforma to assess the implications of the quality of reporting specifically for the RhD NIPD test. Authors were found to generally present an optimistic view of NIPD, bearing in mind weaknesses identified in reporting and conduct of their studies and the analysis of results, as evidenced by the low STARD scores. The NIPD proforma identified that specific biases were potentially introduced through selective population sampling and/or failure to report the make-up of the population tested, omission of inconclusive results, inconsistencies in the handling of repeat results on a sample, and lack of adequate controls. These factors would inevitably affect the validity of diagnostic accuracy as reported in individual publications, as well as any subsequent meta-analyses. Together, published reports to date may provide a biased picture of the actual potential of NIPD testing for fetal RHD genotyping. Generalisation of the available evidence on diagnostic accuracy, especially to large-scale implementation of NIPD testing of non-sensitised women, will also require that decision makers consider further aspects such as test reliability and cost of routine testing in clinical practice. It is recommended that all studies of diagnostic accuracy of NIPD tests adhere to the STARD quality checklist in order to improve reporting, thereby, minimising bias and increasing the comparability of studies. Researchers should also consider specific shortcomings for NIPD and avoid selective participant sampling; report population characteristics; report handling of replicate sampling as well as their failure rates; and include controls for genotypes tested in the study. Furthermore, meta-analyses should consider the quality, as well as the sample size, of NIPD studies in their analysis. Larger trials, required to produce results that are valid and meaningful for clinical practice, must also adhere to these reporting standards.