Evaluating a universal newborn hearing screening program: tackling several methodological challenges using real-world data

Abstract

Background: The application of real-world data in economic evaluations is a topic of increasing interest to both health economists and decision-makers. In this study the value and challenges of using real-world data was explored in economic evaluations using a case study - Australia’s Universal newborn hearing screening (UNHS) programme. The cost-effectiveness of UNHS is not clear due to methodological limitations in the existing literature. A majority of the limitations stem from the lack of randomised controlled trials related to hearing screening. Consequently, many model parameters are based on a combination of secondary sources or simply assumed by the modeller. This increases the uncertainty in the results. Additionally, many evaluations report outcomes in terms of cost per ‘detected true-positive cases’ only instead of incremental cost per Quality-adjusted life years (QALYs) gained. Outcomes such as cost per ‘detected true-positive cases’ are difficult for decision-makers to interpret, compared to incremental cost per QALY gained. This study estimated the cost-effectiveness of UNHS compared to no hearing screening in the Australian healthcare setting using two real-world longitudinal datasets.

Method: An economic model was parametrised with 100,000 hypothetical newborns. The model consisted of a decision tree reflecting the accuracy of testing for hearing loss, followed by a Markov model reflecting the age of diagnosis of hearing loss, severity levels, hearing interventions received, and the acquisition, progression or remission of hearing loss over time (time horizon = 26 years, the age up to which the Australian government funds surgical and non-surgical hearing interventions). The major model parameters were estimated from two longitudinal datasets: (a) the Statewide Comparison of Outcomes of Hearing Loss (SCOUT) survey to estimate utilities and disutilities using the Health Utilities Index Mark 3 (HUI3) and costs using ‘resource use’ data, and (b) the Longitudinal Outcomes of Children with Hearing Impairment (LOCHI) survey to estimate transition probabilities and language outcomes. The final outcome in terms of the incremental cost (measured in terms of Australian Dollars, AUD$) per QALYs gained. Univariate and probability sensitivity analyses were undertaken to take into account the uncertainty surrounding the model parameters.

Finding: Initial findings suggested that UNHS was cost-effective at AUD$60,000 per QALY gained.

Conclusion: This study addressed several methodological limitations and determined the value-of-money of UNHS programme using real-world, longitudinal data. Initial findings suggested that the UNHS is cost-effective. This research also provides insights into the value and challenges of using real-world, longitudinal data in economic evaluations.