Analyses

To evaluate the cost-effectiveness of the proposed algorithm, the costs and outcomes (diagnostic rate) of this demonstration study were compared with that of a hypothetical scenario which represents the current algorithm of invasive prenatal diagnosis in Hong Kong. The laboratory workflow of the current algorithm for invasive prenatal diagnosis is illustrated in Fig. 2.

Laboratory workflow of the current algorithm for invasive prenatal diagnosis in the public healthcare system in Hong Kong. *QF-PCR is not commonly offered free of charge for patients with primary indication of DS screening positive / family history of chromosomal or genetic disorders. However, for patients who are willing to pay for self-financed aCGH, the laboratory will first perform QF-PCR for common aneuploidies detection. If QF-PCR results abnormal, aCGH will not be proceeded. ** Samples with inconclusive QFPCR results and subsequent normal karyotyping results will proceed to aCGH if patient is willing to pay for self-financed aCGH. aCGH: array comparative genomic hybridization; CNV: copy number variation; CVS: chorionic villous sampling; DS: Down syndrome; FISH: fluorescence in-situ hybridization; MLPA: multiplex-ligation dependent probe amplification; NT: nuchal translucency; QF-PCR: quantitative fluorescent polymerase chain reaction

In the current algorithm, all patients requiring invasive prenatal testing will be offered amniotic fluid (AF)/chorionic villus (CV) karyotyping. Those with abnormal fetal ultrasound findings and/or increased NT will be offered QF-PCR simultaneously. Self-financed CMA is available to women who are willing to pay $4900. For patients with other primary indications of test such as DS screening positive only, or family history of chromosomal or genetic disorders and are willing to pay for self-financed CMA, the laboratory will also perform QF-PCR for them prior to CMA. The rest of the workflow was similar to the proposed algorithm as described above. Costs and outcome data were estimated by experts and clinicians based on the results from the demonstration study (if the same cohort was to undergo the current algorithm instead of the proposed algorithm).

In the primary analysis, costs and outcomes from the proposed algorithm were compared with that of the current algorithm, under an ideal situation that assumed 100% of the patients are willing to pay 100% out-of-pocket for the aCGH test. In the secondary analysis, unpublished data on willingness-to-pay, which was extracted from the data set collected from the questionnaire used in our previous study [17], on the perceptions of pregnant women and healthcare providers on invasive prenatal testing were incorporated. Only 41.8% of 717 (n = 300) women from that study were willing to undergo aCGH with 100% out-of-pocket payment. Therefore, in the secondary analysis, only 41.8% of the patients in this study would be costed for aCGH in the analysis.

Cost data was replicated 1000 times using non-parametric bootstrapping to mitigate the effects of data skewness and to enable quantification of the uncertainty surrounding the estimates of costs and effects by estimating the 95% confidence intervals (CIs). The difference between the two algorithms could be judged to be significant at p ≤ 0.05 where the bias-corrected CIs of change scores excluded zero. An incremental cost-effectiveness ratio (ICER) was calculated for each cost-outcome combination that showed higher costs and better outcomes, or lower costs and worse outcomes. This was calculated as the bootstrapped mean cost difference divided by the mean effect (diagnostic rate) difference between the two algorithms. The ICER represents the additional cost for every additional unit of effectiveness (an additional 1% of diagnostic rate) made by the proposed algorithm. Data analyses were conducted using STATA (version 15).