Selection of IEMs
To guide applicants preparing the material of registration and application for amino acid, carnitine, and succinylacetone detection reagent (MS/MS), the Center for Medical Device Evaluation of National Medical Products Administration of China (NMPA) enacted "Guiding Principles for Amino Acid, Carnitine and Succinylacetone Detection Reagent Registration" in 2019 [59]. The guideline recommends 12 types of IEMs that are relatively common in China and suitable for screening by MS/MS (Table 1). Compared with the numbers of IEMs detected in other countries, it is a conservative recommendation based on China’s actual situation, suiting a referable and applicable pilot program for Shenzhen.
Table 1 IEMs in the nationally recommended program
No.
|
IEMs
|
1
|
Phenylketonuria (PKU)
|
2
|
Methylmalonic acidemia (MMA)
|
3
|
Primary carnitine deficiency (PCD)
|
4
|
Medium-chain acyl-CoA dehydrogenase deficiency (MCAD)
|
5
|
Very long-chain acyl-CoA dehydrogenase deficiency (VLCAD)
|
6
|
Isovaleric acidemia (IVA)
|
7
|
Glutaric acidemia type I (GAI)
|
8
|
Maple syrup urine disease (MSUD)
|
9
|
Citrullinemia type II (CIT-II)
|
10
|
Citrullinemia type I (CIT-Ⅰ)
|
11
|
Propionic acidemia (PA)
|
12
|
Homocystinuria (HCY)
|
Perspective, Discount and Comparators
This study measured the inputs and outputs of MS/MS newborn screening from a social perspective. And then, we estimated both cost and effectiveness to calculate the ICER of expanded screening, with the cost and effectiveness discounted at an annual rate of 3%. For the moment, PKU, CH (congenital hypothyroidism), CAH (congenital adrenal hyperplasia) and G6PD deficiency (Glucose-6-phosphate Dehydrogenase deficiency) are compulsorily detected by IF in Shenzhen. Only PKU is incorporated into the expanded screening. So, we compared the program detecting 12 IEMs by MS/MS with merely detecting PKU by IF.
In order to determine an adequate and reasonable number of IEMs, we compared the results of economic evaluations and budget impact analyses of MS/MS screening programs with different disease combinations. First, we conducted a cost-effectiveness analysis on the program detecting every disease to obtain an ICER ranking of various diseases. Since PKU is already covered in the original NBS program, we needed to concentrate only on the other 11 IEMs. We could then finally determine 11 screening strategies according to the ranking. The first strategy includes PKU and the disease with the smallest ICER, and the second includes PKU and the top two diseases. By parity of reasoning, the 11th strategy is the nationally recommended program covering all 12 diseases.
Model Structure
We developed a decision-tree and Markov model to conduct the cost-effectiveness analysis of expanded newborn screening using the following assumptions:
- A child can have only one kind of disease;
- The progress of IEMs is divided into several independent Markov states (health states) according to the main sequelae. And in each cycle, a child can be in only one of the Markov states.
- The states’ future distribution depends only on current events, and not on those that occurred before.
A half-cycle correction was made to the model. Based on the field investigation, the numbers of people participating in NBS from 2016 to 2018 in Shenzhen were 209,443, 228,206 and 206,670, respectively. Also, there were 194,393 puerperae throughout 2019, so we set the cohort at 200,000 newborns with a cycle-length of one year. The Markov model was terminated at the 82nd cycle since the average life expectancy of Shenzhen residents was 82 years in 2018. States transition of IEMs is shown in Fig. 1, and the Markov model structure in Fig. 2.
Sensitivity and Specificity
The screening methods in Shenzhen include immunofluorescence for the current program while MS/MS only for expanded screening. As is showed in Table 2, the former sensitivity and specificity are 100% and 78.6%, and their counterparts are both 100%.
Table 2 Sensitivity and specificity of screening methods
Parameter
|
Mean
|
Reference
|
Fluorometry
|
Sensitivity
|
100.00%
|
[60]
|
Specificity
|
78.60%
|
MS/MS
|
Sensitivity
|
100.00%
|
[61] [21]
|
Specificity
|
100.00%
|
Event probabilities
In Shenzhen, only the incidence of PKU can be collected. Therefore, this study referred to the MS/MS screening results in Zhejiang, covering 1,861,262 neonates and Nanjing, covering 850,486 neonates, as shown in Table 3. Because of the large numbers of people covered and the high maturity of widely used MS/MS, the incidence data is referable for Shenzhen.
Table 3 The incidence of IEMs
No.
|
IEMs
|
Incidence
|
Reference
|
1
|
PKU
|
1:14028
|
0.00007128
|
Filed investigation
|
2
|
MSUD
|
1:206667
|
0.00000484
|
[33]
|
4
|
CIT I
|
1:265700
|
0.00000376
|
[33]
|
5
|
HCY
|
1:212622
|
0.0000047
|
[11]
|
6
|
MMA
|
1:46500
|
0.00002151
|
[34]
|
7
|
IVA
|
1:265900
|
0.00000376
|
[34]
|
8
|
GA I
|
1:310200
|
0.00000322
|
[34]
|
9
|
PA
|
1:310200
|
0.00000322
|
[34]
|
10
|
PCD
|
1:23862
|
0.00004191
|
[35]]
|
11
|
MCAD
|
1:372252
|
0.00000269
|
[35]
|
12
|
VLCAD
|
1:620421
|
0.000001611
|
[35]
|
Data for the age-specific mortality and incidence of sequelae of different IEMs between the screening and non-screening groups remain scarce in China, leaving us no alternative but to refer to studies from other countries. In order to simulate the natural progress of diseases more accurately, the rate of death due to other causes was also added to the model as a parameter. This part of event probabilities is detailed in Additional file 1.
Costs
The cost of MS/MS screening includes direct costs and indirect costs from a social perspective. Direct costs include direct medical cost and direct non-medical cost. The former consists of expenditures for screening, confirmation, treatment (for the disease and its sequela), and follow-up, while the latter consists in family transportation costs and the program cost. Indirect costs are due to the lost productivity of families during the confirmation, treatment, and follow-up stages of NBS. Since detection in newborns occurs within six days in the hospital, families do not need to go to any particular hospital, so transportation and lost productivity costs during the screening stage are not counted.
Cost inputs used in the model are shown in Table 4. The data were derived mainly from the field investigation and estimation based on the Guidelines for the Treatment of Rare Diseases (2019) [62], drug prices at online pharmacies, and assumptions of this study. Also, we referred to published literature, government policies, and statistical yearbooks if parameters could not be directly obtained or estimated.
Table 4 Model cost parameters presented in 2018 RMB
Parameters
|
Mean
|
Reference
|
Screening by IF (per newborn)
|
23
|
Field investigation
|
Screening by MS/MS (per newborn)
|
296
|
Field investigation
|
Confirmation by IF (per newborn)
|
319
|
Field investigation
|
Confirmation by MS/MS (per newborn)
|
296
|
Field investigation
|
Treatment for the diseases (per person-year)a
|
Age groupsb
|
0~5 years
|
6~82 years
|
Guidelines for the Treatment of Rare Diseases (2019) [62]
|
MSUD
|
13440.90
|
6083.00
|
HCY
|
13772.13
|
5952.51
|
CIT I、CIT II
|
48202.46
|
20833.77
|
IVA
|
17567.81
|
13045.01
|
GA I
|
17567.81
|
13045.01
|
MMA
|
19023.76
|
24692.53
|
PA
|
17567.81
|
19839.39
|
MCAD、PCD、VLCAD
|
15853.26
|
15853.26
|
PKU
|
15000.00
|
15000.00
|
Field investigation
|
Treatment for sequelae (per person-year)c
|
DD
|
3,064
|
[63]
|
ND
|
53,400
|
[64]
|
MR
|
12,000
|
[65]
|
RD
|
70,213
|
[66]
|
Follow-up
(per person-year)
|
IF (4 times)
|
1,276
|
Field investigation
|
MS/MS(4 times)
|
1,184
|
Field investigation
|
Transportation (per year)
|
435
|
Field investigation
|
Lost productivity (per year)
|
3,060
|
[67]
|
Program cost
|
Sample transportation
(per year)
|
42,000
|
Field investigation
|
Printing (per year)
|
340,000
|
Cold-chain logistics
(per year)
|
470,000
|
Software development and maintenance, equipment updating, etc.
(per year)
|
370,000
|
Utilities (per person-year)
|
2.52
|
[68], supposing 200,000 people are screened
|
Administrative management (per person-year)
|
1.81
|
Depreciation of buildings, equipment, etc.
|
Sunk costs
|
aThe calculation process of the treatment cost of the diseases is detailed in Additional file 2.
bSchoen et al. [57] divide the cost of treatment into two parts at 5 years of age since, in some cases, additional care is needed for children with IEMs detected after symptoms manifest during their first five years. We also found discrepancies between different ages in the treatment and medication criteria for IEMs when referring to Guidelines for the Treatment of Rare Diseases (2019). Therefore, we did the same work to estimate the treatment cost by dividing patients into two groups at age 5.
cDD: Development Delay; ND: Neurological Damage; MR: Mental Retardation; RD: Renal Damage.
Effectiveness
Quality-adjusted life-years (QALYs) were estimated through the Markov model, multiplying the length of time in different health states by the utility value for states. We also calculated the ICER between current screening and expanded screening programs. An ICER threshold set at 568.704 RMB, three times per capita GDP in Shenzhen, was used in this study. The utility parameters of health states are listed in Table 5, estimated mainly from data in published articles. We presumed the utility of “alive state” to be 1, which means healthy.
Table 5 Model Utility Parameters
Parametera
|
Mean (range)
|
Reference
|
NS
|
0.900(0.850-0.950)
|
[48]
|
DD
|
0.843(0.792-0.881)
|
[49]
|
ND
|
0.840(0.700-0.850)
|
[69]
|
MR
|
0.790(0.590-0.840)
|
[69]
|
RD
|
0.670(0.580-0.740)
|
[70]
|
Alive
|
1
|
Research assumption
|
aDD: Development Delay; ND: Neurological Damage; MR: Mental Retardation; RD: Renal Damage.
Sensitivity analysis
We carried out one-way sensitivity analysis and constructed tornado diagrams to assess the uncertainty in the model and the robustness of the results. One-way sensitivity analysis in this study evaluated the influence of the discount rate in the range 0-10% (base value is 3%), with 1% as an interval of 10 categories. Tornado diagrams include factors like the incidence of IEMs, costs (e.g., the cost of screening, confirmation, transportation, etc.), and utility of health states. The incidence of IEMs was assumed to vary by 50% from their mean value, and costs were 10%. The value of utility being tested varied based on the upper/lower boundaries illustrated in published articles.
Budget impact analysis
Implementing MS/MS screening, the expansion of diseases screened, and the increase in costs will inevitably place a burden on health expenditure, making it necessary to conduct a budget impact analysis (BIA) of MS/MS screening from the standpoint of Shenzhen’s government.
Health expenditure entailed in the frame of this study included the cost of screening, treatment, and follow-up, as well as the program cost. We assumed that these program costs remain unchanged between the expanded screening and the status quo. What should be noted is that children must receive continuous medical treatment following a positive detection. Because of the increasing numbers of patients year by year, the cost of treatment is necessarily cumulative. The treatment cost of MS/MS screening is mainly that of the IEMs without sequelae, since early detection can sharply reduce the occurrence of sequelae, as indicated in the Markov model. By comparison, the treatment cost of current screening is mainly the treatment cost for the IEMs with sequelae. Due to the high mortality rates of some IEMs, we didn’t calculate the treatment cost of patients with diseases which can cause death within their first two years.
BIA was carried out under the assumption of 200,000 newborns screened annually for ten years (2019-2028), and the cost discounted to 2019 RMB at an annual rate of 3%. Nowadays, in Shenzhen, with relevant policies and regulations released [71] and a national medical insurance system established, the cost of screening and treatment is paid for by families and the government conjointly. The government undertakes 80% of the screening cost and 60% of the treatment cost. How the rate of the medical insurance reimbursement is calculated is shown in Additional file 3.