pediatrics
August 2015, VOLUME136 /ISSUE 2

# Cost-Effectiveness Analysis of a National Newborn Screening Program for Biotinidase Deficiency

1. Laura Vallejo-Torres, PhDa,b,c,d,
2. Iván Castilla, PhDb,c,e,
3. María L Couce, PhDf,g,h,
4. Celia Pérez-Cerdá, PhDg,i,
5. Elena Martín-Hernández, PhDj,
6. Mercé Pineda, PhDg,k,
7. Jaume Campistol, PhDg,l,
8. Arantzazu Arrospide, MScc,m,
9. Stephen Morris, PhDd, and
10. Pedro Serrano-Aguilar, PhDc,n
1. aDepartamento de Economía de las Instituciones, Estadística Económica y Econometría, Universidad de la Laguna, La Laguna, Spain;
2. bCentro de Investigaciones Biomédicas de Canarias (CIBICAN), Islas Canarias, Spain;
3. cRed de Investigación en Servicios de Salud en Enfermedades Crónicas (REDISSEC), Spain;
4. dDepartment of Applied Health Research, University College London, London, England;
5. eDepartamento de Ingeniería Informática y de Sistemas. Universidad de La Laguna. La Laguna, Spain;
6. fUnidad de Diagnóstico y Tratamiento de Enfermedades Metabólicas Congénitas, Hospital Clínico Universitario de Santiago, Santiago de Compostela, A Coruña, Spain;
7. gCentro de Investigación Biomédica en Red de Enfermedades Raras (CIBERER), Spain;
8. hInstituto para el Desarrollo e Integración de la Sanidad (IDIS), Madrid, Spain;
10. jUnidad Pediátrica de Enfermedades Raras. E. Mitocondriales-E. Metabólicas Hereditarias, Departamento de Pediatría, Hospital Universitario 12 de Octubre, Madrid, Spain;
11. kFundación del Hospital Universitario Sant Joan de Déu, Barcelona, Spain;
12. lServicio Neurología, Hospital Universitario Sant Joan de Déu, Barcelona, Spain;
13. mUnidad de Investigación Sanitaria de AP-OSIs Gipuzkoa, Organización Sanitaria Integrada Alto Deba, Arrasate, Guipúzcoa, Spain; and
14. nServicio de Evaluación del Servicio Canario de la Salud (SESCS), El Rosario, Spain

## Abstract

BACKGROUND AND OBJECTIVES: There are conflicting views as to whether testing for biotinidase deficiency (BD) ought to be incorporated into universal newborn screening (NBS) programs. The aim of this study was to evaluate the cost-effectiveness of adding BD to the panel of conditions currently screened under the national NBS program in Spain.

METHODS: We used information from the regional NBS program for BD that has been in place in the Spanish region of Galicia since 1987. These data, along with other sources, were used to develop a cost-effectiveness decision model that compared lifetime costs and health outcomes of a national birth cohort of newborns with and without an early detection program. The analysis took the perspective of the Spanish National Health Service. Effectiveness was measured in terms of quality-adjusted life years (QALYs). We undertook extensive sensitivity analyses around the main model assumptions, including a probabilistic sensitivity analysis.

### Data

#### Probabilities

Data from the Galicia NBS program for BD were used to estimate some of the key parameters of the model (Table 1). From April 1987 to October 2013, 540 963 newborns were screened for BD, and 24 cases were detected: 16 cases with partial BD and 8 cases with profound BD. The recall rate observed was 0.44%. No false-negative cases were identified, but there were 35 false-positive cases in this period, yielding an estimated sensitivity and specificity of 100% and 99.99%, respectively. It is worth noting the possibility that the lack of false-negative cases might result from cases remaining undetected, and therefore the possibility of false-negative cases cannot be completely ruled out. However, these values are very similar to those reported in previous studies. Kwon and Farrell21 estimated sensitivity and specificity rates in the United States between 1994 and 1995 of 100% and 99.98%, respectively.

TABLE 1

Probabilities Related to BD Incidence, Screening Test, and Clinical Symptoms

A systematic review of international literature was undertaken to identify available evidence on the natural history of BD in the absence of a screening program and the effectiveness of NBS in terms of health outcomes among early detected patients (see Supplemental Information). We identified several case series that showed that patients with BD detected through NBS and under treatment remain asymptomatic.710,13,2224 Therefore, we assumed that BD cases detected through NBS and receiving treatment do not experience clinical symptoms. The probabilities of exhibiting symptoms in cases not detected by NBS are based on meta-analyzing the evidence from identified case series studies.6,8,22,2530 The probabilities of each outcome are presented in Table 1, including death, seizures, hypotonia, skin problems, cognitive deficits, optic atrophy, and hearing loss (see Supplemental Information for more details). The base case analysis assumes, conservatively, that only profound cases experience clinical complications, whereas in a separate analysis we allow for partial BD cases to also exhibit some of the features that have been documented in the literature among these patients (ie, seizures, skin problems, hearing loss, and cognitive deficits).

Our model allows any possible combination of symptoms, including no symptoms, and assumes that the probabilities of suffering each symptom are independent. Symptoms of untreated BD usually appear between 1 week and 10 years, with a mean age of 3.5 months.3 We assume that acute episodes of seizures, skin problems, and hypotonia, when they occur, appear during the first year of life, whereas cognitive deficit, optic atrophy, and sensorineural hearing loss would become apparent in the second year of life, varying this last parameter between 2 and 15 years of age in the sensitivity analysis.

BD screening requires measurement of biotinidase activity in blood-saturated filter paper samples on enzymatic assay or semiquantitative fluorometric determination. The cost of the screening test was estimated based on information provided by the NBS program in Galicia. We evaluate the cost-effectiveness of adding BD to a current NBS program, and therefore we considered only the extra resources needed specifically for BD testing. These resources include BD assays, equipment, and staff costs and yielded an estimated cost per test of $1.23 (Table 2). TABLE 2 Unit Costs and QALY Data The confirmatory tests consist of quantitative enzymatic determination of biotinidase activity in serum and plasma and the analysis of mutations of the biotinidase gene. Confirmed cases will then undergo testing of organic acids and determination of acylcarnitines in plasma. The unit costs of these tests are presented in Table 2. Current recommendations for BD treatment involve a dose of oral biotin of 10 mg (ranging from 5 to 20 mg) per day for profound BD and a dose of 5 mg per week for partial BD. The following evaluations are generally recommended annually in profound cases and every 2 years in partial cases1: ophthalmologic, auditory, and medical geneticist or metabolic specialist evaluations and urinary organic acid and acylcarnitine analyses. Treatment and follow-up are assumed to be lifelong; unit costs are shown in Table 2. In our base case partial BD cases received treatment, because this is current practice in Galicia. Because there is controversy about whether partial BD cases ought to receive treatment, we also consider a scenario in which partial BD cases are not treated or followed up. Existing but limited evidence on unscreened children detected after the onset of symptoms suggests that skin problems, seizures, and hypotonia rapidly resolve after biotin treatment, whereas some other features, such as developmental delay, optic atrophy, and hearing loss, are generally irreversible once they occur.1,4 Therefore, we divided BD symptoms into acute and chronic symptoms, assuming that children affected by skin problems, seizures, and hypotonia would incur 1 hospitalization and standard BD diagnosis, treatment, and follow-up. Optic atrophy would require 1 additional visual testing for diagnosis, and once confirmed it would be addressed with extra annual ophthalmologic visits and evaluations. Children with hearing loss will undergo 1 additional auditory test, severe cases will need cochlear implants, and the rest would be treated with hearing aids, based on the proportions of profound and moderate hearing loss findings.6 Hearing aids were assumed to be changed every 4 years up to 16 years of age and every 6 years thereafter. Additionally, patients with hearing loss would receive extra annual auditory evaluations with specialist and speech and language therapist visits. Cognitive deficits might involve a series of tests including MRI, electroencephalography, karyotype analysis, and array comparative genomic hybridization testing performed at time of diagnosis. If deficits were confirmed, these patients were assumed to visit the specialist annually. The cost of infant death was estimated based on the mean national cost of a PICU hospitalization with a discharge status of “death.” Whenever available, unit cost data were measured in terms of the mean of Spanish regional tariffs.31 We also used national data published on the MSSSI Statistics Web site.32 Unit costs of drugs were obtained from the database of the General Council of Spanish Pharmacists.33 In our base case we assume patients suffering chronic consequences of BD would not receive any disability benefits. However, permanent consequences of BD might mean patients are eligible to receive such benefits. Because eligibility is uncertain, we also evaluate the impact of assuming that 20% of the patients suffering chronic consequences of BD would receive disability benefits, and if they do, we conservatively assume that they would receive the lowest approved level of benefits (Table 2). Cost-effectiveness results including these costs are presented in a separate analysis. #### Effectiveness Measure QALY calculations are reported in detail in the Supplemental Information. In brief, we used QoL values related to BD complications taken from the literature and identified by a previous review.34 Disutilities for BD complication based on the most conservative estimates are presented in Table 2; these were applied to the QoL population norm for Spain in our base case evaluation. We ran analyses with several different sets of utility weights. Life expectancy of general Spanish population born in 201235 was applied to individuals who did not die of BD, with the exception of those with cognitive disorders, for whom we considered a lower life expectancy, based on data from Bittles et al,36 applying the same approximation as used by Carroll and Downs.37 #### Sensitivity Analysis We undertook a probabilistic sensitivity analysis by using 1000 simulations in a Monte Carlo analysis to compute cost-effectiveness acceptability curves (CEACs). CEACs indicate the probability that an intervention is cost-effective for different values of the willingness to pay for an extra unit of outcome.38 We applied different probability distributions depending on the nature of the parameter.38 Probabilities were characterized by a β distribution. Resource use data inputs were characterized with a γ distribution, whereas uniform distributions were applied to unit cost parameters. In both cases, we used upper and lower limits of 20% around the mean values, with the exception of the cost of the test, which ranged from$0.7 (€0.5) to $3.4 (€2.5), and the age of the start of chronic features, which ranged from 2 to 15 years, with a uniform distribution. We used β distributions to characterize the uncertainty about the utility values. ## Results The results of the analysis are presented in Table 3, showing the lifetime costs and QALYs for a Spanish birth cohort of 450 000 newborns with and without BD screening. Costs are shown separately for screening, diagnosis, BD treatment and follow-up, and costs of treating complications for the base case. Costs of disability benefits are also reported, although these costs are excluded from the base case, which also assumes that partial BD cases do not exhibit symptoms but are treated and followed up under the screening strategy and uses utility weights from Carroll and Downs.39 The incremental total cost of BD screening compared with no screening in this case is estimated to be$557 044 (mean incremental cost per neonate of $1.24). The incremental QALY gain of BD screening is estimated to be 22.573 QALYs (mean QALY gain of 0.00005 per neonate), which leads to an ICER of$24 677 per QALY. When we consider less conservative scenarios, the cost per QALY is lower. This is especially the case when alternative sets of QALY weights are used, when disability benefits are included, and when the assumption that partial untreated cases do not exhibit any symptoms is relaxed. The latter scenario yielded estimated cost savings.

TABLE 3

Base Case Estimates of Lifetime Costs and QALY Gains for BD Screening and No BD Screening for a Spanish Birth Cohort, Together With Incremental Analysis (Base Case and Cost per QALY of Additional Analyses)

Figure 2 shows the CEACs computed based on the probabilistic analysis. At a threshold value of €30 000 per QALY (∼$40 000), the probability that the screening strategy is the most cost-effective option reaches >70% in the base case. FIGURE 2 Cost-effectiveness acceptability curves. Note: Threshold value of €30 000 = US$40 000.

## Discussion

Neonatal screening for BD has been instituted in several regions, including the following European countries: Austria, Belgium, Denmark, Germany, Hungary, Italy, the Netherlands, Sweden, and Switzerland.40 BD screening is not currently done at the national population level in Spain. Our study made use of the ongoing experience of the NBS program in 1 Spanish region, which, alongside other sources, has allowed us to provide an estimate of the cost-effectiveness of adding BD to the national screening panel in Spain.

Previous evidence on the cost-effectiveness of BD screening is sparse. Carroll and Downs37 undertook a cost-effectiveness analysis for NBS of a series of conditions in the United States, including BD. They found a program containing BD screening to be both more effective and less costly than a program not containing BD screening; however, they assumed the use of tandem mass spectrometry, which is not an appropriate technique for BD screening.14 In Belgium, Schoos et al41 also included BD among the diseases considered in their analysis, and they concluded that BD screening was cost-effective in the long term; however, health outcomes were not measured in this study.

Our study provides additional evidence of the cost-effectiveness of BD screening in the Spanish context, but our findings might be relevant in other settings. In the United Kingdom, where current policy is to not screen for BD, an external review conducted in 2012 emphasized the need of additional research on the prevalence of the condition and on the cost-effectiveness of newborn BD screening.15 Although region-specific prevalence might only be estimated with pilot studies, this current evaluation might provide some broad indication of the likelihood of the cost-effectiveness of BD screening in similar settings.

It is worth noting that data from the NBS program in the Galicia region shows higher BD prevalence than that reported in the worldwide study by Wolf.5 However, other countries have shown similar, and sometimes higher, prevalence values, such as Canada,42 Hungary,24,43,44 Austria,23,45 and Brazil.9 We explored the implications of this issue by rerunning the analysis and assuming the same prevalence as that reported by Wolf. The conclusions remained the same: BD testing was found to be more cost-effective than clinical detection.

A critical issue around BD is the controversy as to whether identified partial cases ought to be treated with biotin supplementation. We found that identifying and treating both forms of BD is likely to be cost-effective, even if partial deficiency cases were assumed to not exhibit symptoms if untreated (which was our base case). The cost per QALY gained decreased only slightly when we assumed no treatment of partial BD cases. This finding, alongside the anecdotal evidence that some patients with partial BD do experience clinical symptoms if untreated,22,30 and the lack of toxicity for biotin1 build a stronger case for the treatment of detected partial BD.

This study has several limitations. BD is a rare disease, and therefore the information available on the natural history of the condition, the screening and treatment effects, and the health care resource implications is limited and of low quality. Information on long-term consequences of unscreened children is particularly sparse, although most evidence suggests that features related to cognitive, hearing, and vision problems are permanent. Few published cases have reported improvement to normal ranges,46,47 which in some instances have been contradicted by long-term follow-up information.48 Therefore, we could not estimate how often these complications are reversible, but we conducted an additional analysis imposing the strong assumption that these complications would last for a maximum of 5 years in every affected patient and then return to normal ranges. The ICER was affected by this assumption but was still under the reference cost-effectiveness threshold.

QALY measurement in newborn populations poses serious methodological challenges,49 and so we also acknowledge the large degree of uncertainty with regard to these measures. Therefore, we used alternative sets of QALY weights, which were found to yield the same conclusions. Finally, the perspective of the analysis was that of the MSSSI; however, if a societal perspective were considered, wider costs related to BD should have been taken into account, such as patients’ and carers’ costs including parents taking days off work and productivity losses due to illness. Because BD screening would reduce clinical symptoms in identified newborns, accounting for these factors would probably increase savings related to the screening strategy, strengthening the conclusions of this article.

## Conclusions

The inclusion of BD in the panel of conditions screened under the national NBS program in Spain is likely to be cost-effective compared with clinical detection. At standard threshold values, the probability that BD screening is cost-effective is >70%.

## Acknowledgments

We thank Raquel Zubizarreta, director of the Neonatal Screening Program of Galicia, members of the Metabolic Pathologies Laboratory from Hospital Clínico Universitario de Santiago, and Mercedes Espada Sáenz-Torre, coordinator of the Unit for Neonatal Screening in the Basque Country.

## Footnotes

• Accepted May 26, 2015.
• Address correspondence to Laura Vallejo-Torres, PhD, Centro de Salud de San Isidro–El Chorrillo, 1a Planta, Camino Candelaria, 44 38109 S/C, Tenerife, Spain. E-mail: lvallejo{at}ull.es
• Dr Vallejo-Torres participated in the concept and design of the study and in the collection, analysis, and interpretation of data and drafted and revised the manuscript; Dr Castilla participated in the design of the study and in the collection and interpretation of data and revised the manuscript; Drs Couce, Pérez-Cerdá, Martín-Hernández, Pineda, and Campistol participated in the concept and design of the study and in the collection and interpretation of data and revised the manuscript; Ms Arrospide and Dr Morris participated in the concept and design of the study and in the analysis and interpretation of data and revised the manuscript; Dr Serrano-Aguilar coordinated the research team and critically reviewed the manuscript; and all authors approved the final manuscript as submitted.

• FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.

• FUNDING: Funded under a collaboration agreement between the Carlos III Health Institute, an autonomous organization of the Ministry of Economics and Competitiveness, and the Canary Islands Foundation of Research and Health (FUNCIS). This work was undertaken in the framework of activities run by the Network of Health Technology Assessment Agencies, funded by the Ministry of Health, Social Services and Equality. L.V.T. acknowledges the IMBRAIN project (FP7-REGPOT-2012-CT2012-31637-IMBRAIN), funded under the 7th Framework Programme (Capacities).

• POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.