Optimal newborn screening algorithms - efficacy and equity

Project Summary/Abstract
Newborn screening (NBS) is a successful public health initiative; newborns in every state are
screened for multiple genetic disorders. There are over 30 genetic disorders recommended for
NBS, many states screen for fewer due to resource restrictions. Cystic fibrosis (CF), an
autosomal recessive disorder with hundreds of known CF-causing mutations, is one of the more
prevalent disorders in NBS, and is included in every state’s NBS initiative. The gold standard
Sweat Chloride (SC) diagnostic test for CF is expensive and impractical as a screening test.
Therefore, all states use multi-tiered screening algorithms (i.e., processes) for CF NBS,
consisting of relatively inexpensive screening tests that are used on dried blood spots routinely
obtained from the newborns.
Most CF screening algorithms start with the low-cost, low-efficacy immunoreactive trypsinogen
(IRT) test, followed by a genetic test, where the latter searches for a subset of known CF-causing
mutations, and end with an SC test for confirmation. However, CF NBS algorithms, namely the
combination of specific tests and the decision rules used (e.g., IRT thresholds, number of testing
tiers, mutations to search for, when to send a newborn for an SC test), vary widely among the
states, leading to quite different rates of false screen negatives and false screen positives, and
testing costs. The primary goal of this proposal is to develop a holistic framework for
designing optimal NBS algorithms that are accurate and equitable, using CF as a model disorder.
This framework will utilize prescriptive analytics, including optimization, risk prediction, and
data analytics, comprised of novel, data-driven models and methodologies, and will consider
novel testing methods, such as pooled genetic testing.
To accomplish our research goals, and validate and demonstrate our results, we will: (1) develop
models to optimize current screening algorithms; (2) develop novel screening algorithms for CF
utilizing pooled and multiplex genetic testing combining disorders; and (3) validate the modeling
framework, and demonstrate its effectiveness by designing an optimal CF NBS algorithm based
on data from the state of New York and validating the results via predictive analytics (via
sophisticated simulations and sensitivity analyses), and descriptive analytics. The models and
ideas developed in this research can be extended to other disorders. This research is in
collaboration between The University of Alabama and New York State Department of Health,
Wadsworth Center, and has the potential to change screening practices nationwide.

Grant Number: 5R21HD113958-02
NIH Institute/Center: NIH
Principal Investigator: Douglas Bish