Maximizing Benefits and Minimizing Harms: Diagnostic Uncertainty Arising From Newborn Screening

Screening newborns for rare diseases for which effective treatments are available is a crucial public health practice. Indeed, the expansion of population-wide newborn screening (NBS) programs was named by the Centers for Disease Control and Prevention as 1 of the 10 great public health achievements in the first 10 years of the 21st century.¹  Pivotal randomized controlled trials of NBS for cystic fibrosis (CF) revealed potential benefits,^2,3  with evidence from the Wisconsin trial revealing that early ascertainment through screening could prevent severe malnutrition and improve long-term growth.²  NBS for CF is now widely implemented in many jurisdictions and is included in the influential US Recommended Uniform Screening Panel.⁴ 

As Gray et al⁵  provocatively noted, however, “All screening programmes do harm; some do good as well, and, of these, some do more good than harm at reasonable cost. The first task of any public health service is to identify beneficial programmes by appraising the evidence.”⁵ (p480) The current study by Gonska et al⁶  is an important example of ongoing research required to measure benefits and understand harms of NBS, thus supporting program improvement.

Although the benefit of NBS for CF is unequivocal,⁷  there are potential harms to children and their families who receive false-positive or inconclusive results after a positive screen result. Although most false-positive findings are resolved in early infancy, with evidence mainly revealing transient psychosocial impacts that are mitigated with timely follow-up and effective communication,^8,9  inconclusive results tend to be longer lasting. This is indeed the case for children identified through NBS as having a diagnosis of "Cystic Fibrosis Screen Result Positive, with Inconclusive Diagnosis" (CFSPID), for whom clinical guidelines recommend ongoing monitoring because the diagnosis is difficult to rule out.^10,11  Such uncertainty can present challenges for families.^12–15  Health care providers also report challenges in navigating inconclusive findings from NBS, including a need to balance potential overmedicalization of children with a good prognosis with a desire to err on the side of caution to prevent adverse outcomes.¹⁶  Screening policy must account for these complexities in decisions to start or even stop screening for a disease and when strategizing to minimize such circumstances.

The study by Gonska et al⁶  fills a critically important evidence gap by reporting on a Canada-wide cohort of children with CFSPID managed to school age. The findings with respect to nutritional and pulmonary status were reassuring, but a relatively high proportion (n = 24; 21%) of children diagnosed with CFSPID through NBS met criteria for a diagnosis of CF during follow-up. Gonska et al⁶  identified that among those children later diagnosed with CF based on an elevated sweat chloride level ≥60 mmol/L (n = 12) an initial newborn diagnostic sweat chloride level of ≥40 mmol/L predicted this diagnosis. The authors further tentatively suggested that a high initial immunoreactive trypsinogen (IRT) level may be predictive of a future CF diagnosis based on reinterpretation of gene variants.

These findings require validation and refinement but hold promise for the future implementation of clinical prediction tools that can be used to make decisions related to discharging those children who are least likely to go on to meet CF diagnostic criteria. The development of such tools is difficult when relying on results across different screening programs that have used inconsistent screening and diagnostic testing algorithms. For example, as acknowledged by Gonska et al,⁶  the methods used to screen and ascertain cases in their study cohort were not homogenous, and this could have led to an increased risk of bias. In addition, for NBS IRT to be used in such tools, careful consideration will need to be given to ensuring that results are comparable between laboratories, assays, and over time, for example, by calibrating against certified reference materials or using normalized measures of IRT.

These issues of uncertainty are increasingly important to consider as the scope of NBS programs and their use of genomic technologies expands. To achieve benefits for children with targeted diseases while minimizing the impact of uncertain results, longitudinal studies of the health of children with a screening-ascertained attenuated form of disease or uncertain diagnosis are needed to inform the development of clinical prediction tools and guide interventions and discharge decisions. For example, CFTR sequencing is now increasingly being used in the NBS process. Although it is reducing uncertainty for some infants, it is identifying other asymptomatic newborns with genotypes of uncertain significance for whom treatment, follow-up, or discharge decisions may prove difficult.^17,18  Other diseases recently added to the Recommended Uniform Screening Panel with as much or even more clinical heterogeneity as CF, such as mucopolysaccharidosis type 1 and X-linked adrenoleukodystrophy, present similar challenges.^19–24  Looking even further ahead, next-generation sequencing approaches in NBS are being investigated and will present similar questions for a much broader range of rare diseases.²⁵  The ability to generate the type of evidence Gonska et al⁶  present for other existing or candidate NBS target diseases will be key to maximizing benefits while minimizing harms.