Development of a cost-effectiveness analysis model for Duchenne muscular dystrophy utilizing the national registry in Japan

Abstract

Cost-effectiveness analysis models for Duchenne muscular dystrophy (DMD)—a rare muscular disease—have been developed based on data from a limited patient population, such as clinical trials. Thus, this study aimed to construct a more robust cost-effectiveness analysis model based on real-world evidence. The model was constructed using the Registry of Muscular Dystrophy (Remudy) database, the national registry of muscular diseases in Japan. Parameters for transition probability and drug cost were estimated based on this registry, and a quality-of-life (QOL) survey was conducted on Remudy-listed patients for utility. A Markov model was adopted using motor functions as outcomes. Age-specific transition probabilities were estimated by fitting a Weibull distribution to Remudy data of 730 patients. For each drug that was dosed according to body weight (BW), drug costs were estimated from the BW information in the Remudy data and direct medical costs were based on Japanese practice guidelines. For utility, QOL values for each state were estimated for 346 patients who consented to the survey. Using a novel approach that leverages the registry’s comprehensive epidemiological data and patient’s access to research, the refined cost-effectiveness analysis model was developed and this may be fundamental to implementing the health technology assessments of DMD.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-32735-y.

Introduction

Duchenne muscular dystrophy (DMD) is a disease wherein an inborn defect in a gene causes muscle weakness, motor dysfunction, and cardiac dysfunction; it has a poor prognosis and is rare, affecting only 1 in 5,000–6,000 newborn males^1–3. Efforts to develop drugs for treating this disease are currently underway and innovative drugs such as nucleic acid drugs and gene therapies that can significantly improve the causes of disease have been approved^4,5. In 2020, viltolarsen, a nucleic acid drug targeting exon 53, was approved in Japan and the United States. Approximately only 8–10% of patients with DMD have genotypes that can be treated with viltolarsen, thus the treatment is not suitable for the vast majority patients with DMD^6–8. For Japan, this drug is the first disease-modifying drug for DMD. Nucleic acid drugs with similar mechanisms targeting other exons and gene therapies are also currently under clinical development.

Nevertheless, these innovative drugs can be expensive and it is important for decision makers to evaluate their cost-effectiveness in order to continually introduce new innovative drugs, particularly given the healthcare system’s limited financial resources. In the cost-effectiveness analysis of drugs, time to progression of disease is required to estimate the transition probability between each disease state and an analytical model is constructed by incorporating the treatment cost and benefits obtained in each state. There is a lack of comprehensive epidemiological information on disease progression in DMD that would contribute to the development of a cost-effectiveness analysis model. Currently, most clinical trials around the world are focused on how long new drugs can reduce the decline in ambulatory function, and these trial data have been used as a basis for modeling disease progression^9–11. However, most of these trials are designed for patients younger than 10 years and in the early stages of the disease. Furthermore, given the short-term follow-up period of the clinical trials, these data, by representing only a fraction of the DMD patient journey, are insufficient to model disease progression during the entire course of DMD. For innovative new drugs just launched on the market, their long-term efficacy remain clear. On the other hand, to evaluate the cost-effectiveness of these drugs in the future, it is important to construct an elaborate cost-effectiveness analysis model of the standard care group that will be used as a comparison group.

Against this backdrop, the present study aimed to construct a cost-effectiveness analysis model for the DMD standard of care group based on a national registry that collects comprehensive clinical information on DMD.

Methods

All methods were performed in accordance with relevant guidelines and regulations. (The clinical research adhered to the Declaration of Helsinki and the development of the cost-effectiveness analysis model followed the CHEERS guideline.)

Data source

We sourced data from the Registry of Muscular Dystrophy (Remudy)¹² database, a national registry of neuromuscular diseases in Japan that consists of all items recommended by the Translational Research in Europe-Assessment and Treatment of Neuromuscular Diseases (TREAT-NMD) Network of Excellence¹³. TREAT-NMD is the globally recognized standard framework for neuromuscular disease registries, providing the connections with research and trial networks worldwide. It ensures data comparability and interoperability across countries and studies. This includes clinical data such as age at registration, birth date, area of residence, features of the muscle biopsy, complicating diseases, prednisolone use, present functional mobility, age at loss of ambulation, cardiac function, respiratory function, spinal surgery, serum CK level, and so on.

In this study, we used the data from DMD patients enrolled in Remudy from July 1, 2009 to June 22, 2022. Considering that prednisolone is currently the standard of care in DMD, patients were excluded based on the following criteria: (1) under registration (to exclude incomplete data that has not been fully registered in the database), (2) no data on prednisolone use, (To exclude subjects not taking prednisolone when extracting the standard treatment group, as prednisolone is the current standard treatment stipulated in the guidelines.) (3) comorbidities (To only use the data of subjects with a confirmed diagnosis of DMD), (4) no data on ambulation (To exclude subjects with missing data on walking function, which is the outcome variable in model construction.), and (5) no use of prednisolone (Same rationale of criteria (2)).

Outline of model

DMD causes muscle degeneration and corresponding functional impairment. Therefore, in modeling its disease progression, motor function, respiratory function, and even disease-specific measures can be used as outcomes. On the other hand, in constructing a cost-effectiveness analysis model, outcomes representing disease stage must correspond to cost and utility. In Remudy, the primary source of information, motor function is collected as an outcome representing the stage of DMD. Furthermore, the Japanese guidelines¹⁴ for the treatment of DMD, which provide important information for validating the model, also have a policy for treatment according to the stage of the disease, with motor function as the outcome. Therefore, using motor function as the outcome, we constructed a Markov model consisting of five states: early ambulatory, late ambulatory, early non-ambulatory, late non-ambulatory, and dead. The early ambulatory state was defined as being “ambulatory and able to stand up from the floor,” late ambulatory as “ambulatory and unable to stand up from the floor,” early non-ambulatory as “non-ambulatory and able to sit up,” and late non-ambulatory as “non-ambulatory and unable to sit up.” All patients were in the early ambulatory state at the beginning of the model, and because DMD is an inborn and irreversible disease, the completely cured state was not incorporated into the model (Fig. 1). Since Remudy excluded data related to the ability to stand up from the floor and death, we made the following assumptions based on clinical expert opinions of physicians and epidemiological studies on overall survival of Japanese DMD patients^15,16: (1) patients become unable to stand up from the floor approximately 2 years before becoming unable to walk and (2) no deaths due to DMD occurred before the age of 18 years; 50% of the patients die before the age of 35 years, and the cumulative survival rate was expressed as a straight line, assuming a linear decrease between the ages of 18 and 50, regardless of the patient’s state.

Fig. 1.

Markov model in Duchenne muscular dystrophy. All transfers between states are defined by a yearly transition probability. The mortality rate from each state at a given age was defined as the same, the rationale for this is shown in Appendix 3.

Transition probability

In the cost-effectiveness analysis model, transition probabilities from state to state are required to model the time to disease progression. First, age-specific cumulative probabilities of the late ambulatory, early non-ambulatory, and late non-ambulatory states were estimated by fitting a Weibull distribution to the data. The cumulative probability of early ambulatory was then obtained by shifting that of late ambulatory parallel to two years earlier. The cumulative probability of early ambulatory, late ambulatory and early non-ambulatory are shown in Eqs. (1, 2), and (3) respectively.

1

Scale parameter λ₁ and shape parameter γ₁ are independent of age t.

2

Scale parameter λ₁ and shape parameter γ₁ are independent of age t.

3

Scale parameter λ₂ and shape parameter γ₂ are independent of age t.

The mortality distribution, defined based on assumptions about deaths in DMD, was combined with the Remudy data and rescaled to a distribution where the sum of patients who could walk, could not walk, and died was 1.0.

The probability of transitioning from state A to state B during age t by age t + 1 is the proportion of patients who transferred to state B within 1 year, among the patients who were in state A in age t. Further details of the calculation of transition probability between each state are shown in Appendix 2.

Cost

The cost per year was defined as the direct medical cost of treatment. The treatment of DMD varies widely and patients receive different treatments even for the same disease stage, depending on the hospital they visit and the region in which they live. The complexity of treatment also makes it difficult to obtain detailed information on medical care. In addition, Japan provides medical care under the universal health insurance system, and in the case of rare diseases such as DMD, we are currently not allowed to use receipt data owned by the government in terms of protecting patient privacy. Therefore, in the present study, we generalized the treatment of DMD in each stage (Appendix 1) based on the Japanese practice guidelines¹⁴ and the best-practice DMD treatment provided by two pediatric neurologists at the national center hospital, including the physician responsible for developing the Japanese practice guidelines¹⁴.

Moreover, because the drug was expected to improve or maintain motor function (prednisolone) is dosed per kilogram of body weight (BW), one-year prednisolone cost was calculated based on drug regimens in Japan, using the mean BW of each state obtained from the Remudy data.

The BW at each early and late ambulatory state was the weighted BW mean by the distribution of all ambulatory patients. Other direct medical costs were estimated based on the clinical experience of pediatric neurologists and the medical fee points in Japan.

Utility

DMD patients enrolled in Remudy as of October 1, 2022 were asked to respond to a questionnaire mailed to their homes. The questionnaire consisted mainly of questions regarding age, motor function, prednisolone status, and quality of life (QOL). The EQ-5D-5 L (16 > years old: EQ-5D-Y) was used to evaluate QOL. Conversions to QOL values were based on the tariff for cost-effectiveness analysis in the Japanese guidelines^17–19. The mean QOL values were summarized by motor function status and age.

Cost-effectiveness analysis of a hypothetical nucleic acid drug intervention

To showcase the model, a cost-effectiveness analysis from the healthcare system perspective was conducted using the estimated model parameters, comparing the group that received a standard therapy plus a new drug as a hypothetical nucleic acid drug to the group that received the standard therapy alone. The standard therapy was the current best-practice treatment for DMD in Japan, which we generalized based on the expert opinion described in the cost section. The new drug’s intervention effects were hypothesized to prolong all state transitions by 4 years (loss of ambulation was extended to 8 years). Furthermore, it was assumed that a nucleic acid drug did not migrate into the myocardium, a scenario in which overall survival was not prolonged was also analyzed. For the cost of the new drug, the only approved nucleic acid drug for exon skipping in Japan, viltolarsen, was used as a reference.

The age at the start of the model was set to 4 years old, and the discount rate was 2%¹⁷. Incremental cost-effectiveness ratio (ICER) was defined as the measure of the incremental medical cost relative to the incremental utility of switching from a standard treatment to a new treatment. ICER was calculated by fitting a gamma distribution to costs and a beta distribution to utility and running half-cycle-corrected Monte Carlo simulations. The average value from 10,000 Monte Carlo simulations was used as the ICER. One-way sensitivity analysis was performed for cost, utility, and discount rate. Costs were varied based on the quartile range of BW, utilities were varied based on the quartile range of QOL values, and discount rates were varied in the range of 0–4%, following the Japanese guidelines¹⁷.

Software

Data pre-processing and model parameter estimation were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA) and Microsoft Excel^®. Cost-effectiveness and sensitivity analyses were conducted using Tree Age Pro Healthcare 2022 (Tree Age Software, LLC, Massachusetts, USA).

Results

Transition probability

We analyzed 730 patients after exclusion based on prespecified criteria (Fig. 2). The mean and median (25–75%) of age were 15.0 and 14 (10–19) years, respectively. Of these, 277 patients were ambulatory (mean age 9.7 years), 271 were non-ambulatory but able to take a sitting position (mean age 16.1 years), and 182 were non-ambulatory and unable to take a sitting position (mean age 21.4 years).

Fig. 2.

Selection of Duchenne muscular dystrophy patients from the Remudy database for this study. All data of patients diagnosed with DMD between May 1, 2009 and June 22, 2022, and whose registration forms were sent to the Patient Registration Division of Remudy, National Center of Neurology and Psychiatry. PSL, prednisolone.

The Weibull distributions were fitted to the data of these 730 patients and they were combined with the mortality distribution. The estimated scale parameter λ₁ and shape parameter γ₁ were 0.0338 (95% confidence interval: 0.0247–0.0430) and 1.7646 (95% confidence interval: 1.6283–1.9009), which were in the cumulative probabilities of early ambulatory and late ambulatory. The estimated scale parameter λ₂ and shape parameter γ₂ were 0.0128 (95% confidence interval: 0.0027–0.0229) and 1.6826 (95% confidence interval: 1.3847–1.9805), which were in the cumulative probability of early non-ambulatory.

These distributions are shown in Fig. 3, and the age-specific transition probabilities between each state are shown Table 1.

Fig. 3.

Distribution of patients in the Duchenne muscular dystrophy standard care group.

Table 1.

Age-specific transition probabilities between each state.

Age	Probability between EA and LA	Probability between LA and ENA	Probability between ENA and LNA	Mortality
4	0.033	0.000	0.000	0.000
5	0.078	0.000	0.000	0.000
6	0.113	0.306	0.000	0.000
7	0.144	0.427	0.000	0.000
8	0.172	0.470	0.000	0.000
9	0.197	0.495	0.061	0.000
10	0.221	0.512	0.089	0.000
11	0.243	0.527	0.095	0.000
12	0.264	0.538	0.096	0.000
13	0.284	0.549	0.097	0.000
14	0.302	0.558	0.098	0.000
15	0.320	0.567	0.100	0.000
16	0.337	0.575	0.102	0.000
17	0.354	0.583	0.104	0.000
18	0.369	0.559	0.074	0.031
19	0.353	0.543	0.076	0.032
20	0.967	0.549	0.079	0.033
21	0.000	0.350	0.083	0.034
22	0.000	0.964	0.086	0.036
23	0.000	0.000	0.088	0.037
24	0.000	0.000	0.092	0.038
25	0.000	0.000	0.096	0.040
26	0.000	0.000	0.099	0.042
27	0.000	0.000	0.103	0.043
28	0.000	0.000	0.105	0.045
29	0.000	0.000	0.108	0.048
30	0.000	0.000	0.110	0.050
31	0.000	0.000	0.112	0.053
32	0.000	0.000	0.114	0.056
33	0.000	0.000	0.115	0.059
34	0.000	0.000	0.116	0.063
35	0.000	0.000	0.116	0.067
36	0.000	0.000	0.115	0.071
37	0.000	0.000	0.114	0.077
38	0.000	0.000	0.112	0.083
39	0.000	0.000	0.108	0.091
40	0.000	0.000	0.103	0.100
41	0.000	0.000	0.096	0.111
42	0.000	0.000	0.875	0.125
43	0.000	0.000	0.000	0.143
44	0.000	0.000	0.000	0.167
45	0.000	0.000	0.000	0.200
46	0.000	0.000	0.000	0.250
47	0.000	0.000	0.000	0.333
48	0.000	0.000	0.000	0.500
49	0.000	0.000	0.000	1.000

This table shows the probability of transition to the next state by the next year at each age.

EA: Early Ambulatory, LA: Late Ambulatory, ENA: Early Non-Ambulatory, LNA: Late Non-Ambulatory.

Cost

To ensure comparability with other studies, costs were also presented in dollar equivalents.

Considering the study period, the conversion rate used was the 2022 average (1 USD = 131 yen). The mean BWs in each state, estimated from the BW data of DMD patients enrolled in Remudy, were 27.6 kg, 35.8 kg, 48.9 kg, and 45.1 kg in the early ambulatory, late ambulatory, early non-ambulatory, and late non-ambulatory states, respectively. The regimen and price of viltolarsen in Japan were 80 mg/kg weekly infusion at 91,136 yen (695.7 USD)/250 mg, and those of prednisolone were 1.0 mg/kg daily, with a maximum dose of 15 mg/day at 9.8 yen (7.5 cent)/5 mg. The costs of each drug, calculated from the above, are shown in Table 2. The details of standard medical care and medical fee points for DMD in Japan are shown in Appendix 1.

Table 2.

Cost and utility parameters in the model.

State	Body weight (kg)	Cost			Utility
		Direct medical cost (yen/year)	Prednisolone cost (yen/year)	The new drug cost (yen/year)
Early ambulatory	27.6	797,390 (6,086.9 USD)	10,738 (82.0 USD)	42,038,703 (320,906.1 USD)	0.817
Late ambulatory	35.8	825,198 (6,299.2 USD)	10,738 (82.0 USD)	54,525,272 (416,223.4 USD)	0.744
Early non-ambulatory	48.9	1,194,097 (9,115.2 USD)	10,738 (82.0 USD)	74,473,328 (568,498.7 USD)	0.511
Late non-ambulatory	45.1	8,374,458 (63,927.2 USD)	10,738 (82.0 USD)	68,686,869 (524,327.2 USD)	0.369

Direct medical cost does not include the cost of drugs that are expected to improve or maintain motor function.

To ensure comparability with other studies, costs were also presented in dollar equivalents. Considering the study period, the conversion rate used was the 2022 average (1 USD = 131 yen).

Utility

Among the 1,212 patients who were sent the questionnaire, 384 patients gave their consent to participate. Data from 346 patients who had no deficits in their survey answers or discrepancies between their motor function and EQ-5D responses were used. Responses were obtained from 130 ambulatory (Mean age: 9.3 years), 116 early non-ambulatory (Mean age: 18.3 years), and 100 late non-ambulatory (Mean age: 25.2 years) patients, and the QOL values for each state were estimated, as shown in Table 2.

Cost-effectiveness of a hypothetical nucleic acid drug and sensitivity analysis

The ICER for the new drug delaying all state transitions except death by 4 years was 408 million yen (3.1 million USD)/QALY when used up to the early non-ambulatory state and 434 million yen (3.3 million USD)/QALY when used up to the late non-ambulatory state. The ICER for the new drug delaying all state transitions, including death by 4 years, was 320 million yen (2.4 million USD)/QALY when used up to the early non-ambulatory state and 365 million yen (2.8 million USD)/QALY when used up to the late non-ambulatory state (Table 3).

Table 3.

Results of cost-effectiveness analysis for each scenario.

Scenario A
New drug used until	Mean of total cost (yen)	⊿C (yen)	Mean of total QALY	⊿Q	ICER (yen/QALY)
ENA	1,184,144,624 (9,039,271.9 USD)	1,100,893,685 (8,403,768.6 USD)	15.4	2.7	407,738,402 (3,112,506.9 USD/QALY)
LNA	1,255,484,927 (9,583,854.4 USD)	1,172,771,357 (8,952,453.1 USD)	15.4	2.7	434,359,762 (3,315,723.4 USD/QALY)

Scenario B
New drug used until	Mean of total cost (yen)	⊿C (yen)	Mean of total QALY	⊿Q	ICER (yen/QALY)
ENA	1,232,363,160 (9,407,352.4 USD)	1,150,643,058 (8,783,534.8 USD)	16.3	3.6	319,623,072 (2,439,870.8 USD/QALY)
LNA	1,395,816,017 (10,655,084.1 USD)	1,312,726,294 (10,020,811.4 USD)	16.3	3.6	364,646,193 (2,783,558.7 USD/QALY)

Scenario A: The new drug delays all state transitions except death by 4 years.

Scenario B: The new drug delays all state transitions, including death, by 4 years.

⊿C: The difference in total cost between the new drug and prednisolone.

⊿Q: The difference in total quality-adjusted life years (QALY) between the new drug and prednisolone.

ENA: early non-ambulatory.

LNA: late non-ambulatory.

To ensure comparability with other studies, costs were also presented in dollar equivalents. Considering the study period, the conversion rate used was the 2022 average (1 USD = 131 yen).

When the cost, utility, and discount rate were altered, the utility in the late non-ambulatory state and the cost in the early non-ambulatory and early ambulatory states in a nucleic acid drug group significantly affected ICER in all scenarios. For the utility of the late non-ambulatory state, ICER was higher for higher QOL values.

Discussion

This study is the first to construct a cost-effectiveness analysis model for DMD treatments based on a large patient registry system. With the limited epidemiological evidence available for DMD, model parameters have previously been estimated solely from physicians’ clinical experience or the results of short-term clinical trials. In the present study, the use of Remudy data, which includes detailed long-term clinical outcomes of DMD patients, enabled more precise modeling of the disease progression. The cost-effectiveness analysis model for DMD first proposed in 2017⁹ estimated disease progression from only one event: loss of ambulation. However, with this method of estimation, the calculation of the transition probability had to be simplified, and the probability remained constant regardless of how the disease progressed in a real and practical setting. Consequently, the model showed that more than 20% of patients could walk after the age of 20. In contrast, our model incorporated a distribution in which 50% of all patients lose ambulation around age 11, and most patients lose ambulatory function at age 20 or older. This is consistent with clinical experience and previous epidemiological studies^1,20. As described above, Remudy is considered an important data source for describing the long prognosis of DMD, a chronic disease. It also has the virtue of following the global standard registry that collects all items recommended by TREAT-NMD. Our model, which used the Remudy registry and a methodology for model-construction, is applicable in many countries, and its results are comparable by applying a muscular disease registry of the same standard. Cost-effectiveness analyses in DMD have also been conducted in other countries with different healthcare systems and contexts, such as the studies by Landfeldt et al. (Sweden) and Atehortúa et al. (Colombia). Landfelt et al. presented the first cost-effectiveness analysis model for medication intervention in DMD. While our study followed the broad framework proposed by Landfelt et al., we estimated and modeled disease progression more precisely by utilizing comprehensive national registry data that included patients in the non-ambulatory phase. Furthermore, Landfelt et al. analyzed prednisolone, currently the standard treatment for DMD, whereas our study used a hypothetical nucleic acid medicine. This provides insights for evaluating the cost-effectiveness of nucleic acid medicines which are likely to be launched as many products in the future.

In addition, Atehortúa et al. analyzed the cost-effectiveness of diagnostic genetic testing strategies for Duchenne and Becker muscular dystrophies in Colombia. While that study addressed the diagnostic phase rather than post-diagnosis treatment, it noted the low sensitivity of testing and diagnosis in Colombia. Given that disease progression and treatment costs vary across countries with different healthcare systems and contexts, generalizing our findings to other nations requires careful consideration of these differences in healthcare settings and clinical practice guidelines.

The cost of the drugs was estimated by tracking changes in BW over time in the Remudy data. As drugs like nucleic acid drugs and gene therapies have high unit costs that change drastically depending on the dose, a probability distribution was fitted to the BW distribution in each state obtained from the Remudy data. This estimate of drug costs may provide more realistic estimates of the potential range of ICERs. Drug prices could be estimated based on epidemiological evidence rather than assumptions, indicating the usefulness of the Remudy data for drug cost estimation. However, direct medical costs of DMD in Japan have not been reported, and estimating them from the Remudy data was difficult due to the lack of items regarding medical costs. Therefore, we made assumptions based on clinicians’ experience. In Europe and the United States, direct medical costs have been estimated based on questionnaires for patients and caregivers regarding the economic burden of DMD^21–23. In Germany, the estimated costs in 2013 ranged from 4,220 to 7,629 €/year for the ambulatory and 11,666–68,968 €/year for the non-ambulatory states. Medical costs were found to increase significantly with disease progression due to the need for ventilators, medical aids, rehabilitation, and so on²¹. By using the exchange rate in 2013 (1 € = 130 yen), the estimated direct medical costs and changes in costs associated with disease progression showed no significant discrepancy with those in Germany, which has a similar level of medical care to Japan, the assumptions regarding medical costs in this study can be interpreted as reasonable. The questionnaire survey conducted on patients enrolled in Remudy was the first large-scale survey of QOL in DMD patients in Japan. As QOL is a parameter that has a significant impact on ICER, conducting an actual survey enhanced the reliability of the ICER. For the cost-effectiveness analysis, we used the EQ-5D-5 L, which is the first choice as a practical indicator in the Japanese guidelines¹⁷. The QOL values for each condition estimated in this study were higher than those estimated in previous studies using the Health Utilities Index Mark 3 (HUI3)^9,22. Previous reports comparing the values of EQ-5D-5 L and HUI3^24–27 have shown that the EQ-5D-5 L value is higher than that of the HUI3 when the EQ-5D-5 L value is in the 0–0.8 range. When the HUI3 values reported in the previous study were mapped to the EQ-5D-5 L using the conversion table published by the Assessment of Quality of Life (AQOL)²⁸, the values were 0.408 for the early non-ambulatory and 0.353 for the late non-ambulatory states, which did not differ much from the results of this study.

The ICER for a hypothetical nucleic acid drug intervention calculated in this study and the true efficacy of nucleic acid drugs for DMD will be clarified through long-term clinical trials and epidemiological studies. In the case of rare diseases with limited treatment options, it is controversial to use the willingness to pay (WTP) method, which is defined in each country, to determine insurance coverage and drug price adjustment. In contrast, the ICERs calculated in this study under several scenarios suggest the kind of epidemiological evidence that should be generated in the future. The ICER varied widely from about 300–400 million yen (2.3–3.1 million USD)/QALY, depending on whether a nucleic acid drug was administered during the ambulatory state and continued until the early or late ambulatory states. The duration of a nucleic acid drug administration depends on the clinician’s judgment at this time, and additional evidence is necessary to establish the clinical significance. It is also important to follow the progression of disease in patients treated with nucleic acid drugs over time, as ICER varies greatly depending on whether these drugs prolong mortality.

Through sensitivity analysis, it was clear that the utility of each state had a relatively significant effect on ICER. Therefore, it is highly possible that ICER may change, even when the same model framework is used, depending on the evaluation index and QOL tariff. In this study, 96% of ambulatory patients aged ≥ 16 years answered the EQ-5D-Y. The EQ-5D-Y is a 3-level index, and a ceiling effect has been reported for the EQ-5D-3 L, which also evaluates at 3 levels^29,30. Therefore, it is possible that the QOL values for ambulatory patients used as model parameters in this study were calculated as higher than the true values. Additionally, highly generalized indices such as the EQ-5D and the HUI3, which were adopted in past cost-effectiveness evaluations of DMD, have also been pointed out as concerns in evaluating the QOL of DMD^31,32. Therefore, disease-specific indices that can appropriately assess the QOL of patients with DMD and other neuromuscular diseases have been developed^33,34, but their comparability with the QOL of other diseases has not been ensured, and there is no validation regarding their use in health economic evaluation that contributes to payer decision-making. When a validated disease-specific QOL index for DMD is developed in the future, replacing the utility values in this model with such an index should be considered.

There are some limitations to the model developed in this study. Remudy does not include data from a full study of deaths, so we made assumptions regarding the distribution of deaths based on clinical experience. The distribution was simplified by assuming that there were no deaths before the age of 18 years, although it has been reported that a small number of deaths occur in patients under the age of 18 years^15,16. Furthermore, mortality from each state was assumed to be the same, but respiratory failure has been reported to account for 20–30% of deaths in DMD^15,16, and respiratory function were different among states. Therefore, it cannot be ruled out that mortality rates may differ. It is necessary to collect such information on deaths and reflect it in the model. Additionally, above these limitations related to modeling disease progression, there are two points to consider regarding utility. First, the response rate for this study’s questionnaire was approximately 30%, and there may be systematic differences between respondents and non-respondents. Therefore, non-response bias may have influenced the utility value estimation. Second, new modalities innovative therapies are recently launched for which long-term side effects and adverse events have not been identified. Therefore, the cost-effectiveness analysis in this study does not account for the impact of side effects or adverse events on utility values. As evidence accumulates in the future, it would be desirable to incorporate side effects as a disutility factor in the utility values.

DMD presents challenges in modeling indirect affects due to its complex disease progression and the wide range of subsequent medical interventions and care required. However, given the disease’s characteristic need for long-term caregivers, a broader analysis incorporating indirect impacts on utility and costs—considering patient’s own labor productivity and affects on caregivers—is important for more precise economic evaluation.

Conclusion

In this study, the utilization of a registry enabled the development of a model that accurately describes the disease progression of DMD, resulting in the refinement of a cost-effectiveness analysis model. As such, this model framework is expected to be valuable for accurately evaluating health technologies for DMD, from a medical economics perspective.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

- Conceptualization and Methodology: Takuro Okada, Takami Ishizuka, Mari Oba, Satoko Hori, Harumasa Nakamura- Statistical analysis: Takuro Okada, Mari Oba- Interpretation of data: Takuro Okada, Hirofumi Komaki, Takami Ishizuka, Mari Oba, Satoko Hori, Harumasa Nakamura- Drafting of the manuscript: Takuro Okada, Mari Oba, Satoko Hori- Obtaining funding: Hirofumi Komaki, Harumasa Nakamura- Supervision: Hirofumi Komaki, Harumasa Nakamura.

Funding

This study was supported by an Intramural Research Grant (2–4, 2–9) for Neurological and Psychiatric Disorders of NCNP.

Data availability

The data used in this study were obtained from the Registry of Muscular Dystrophy (Remudy), which is not publicly available. Data access is limited to approved investigators upon request and approval by the registry’s steering committee. If it is necessary to request access to the data used in this study, please contact Takuro Okada (Email: t.okada.keio@gmail.com).

Competing interests

Mr. Okada is currently an employee of Astellas Pharma Inc, but from the start to the end of the study, he was affiliated with Keio University, and thus has no conflicts of interest to disclose. Dr. Komaki reported receiving grants from Taiho, Nippon Shinyaku, Chugai, Sanofi, Pfizer Japan, Daiichi Sankyo, and PTC Therapeutics; consulting fees from PTC Therapeutics, Astellas, Daiichi Sankyo, and Sarepta Therapeutics; and payment for educational events from Chugai, Biogen, Sanofi, Novartis, and Nippon Shinyaku. Dr. Oba reported receiving payments for educational events from Pfizer Japan, Astra Zeneca, SAS Japan, and EP-SOGO. Dr. Nakamura reported receiving grants from Pfizer Japan, Nippon Shinyaku, Daiichi Sankyo, and Astellas; and consulting fee from Nippon Shinyaku. Dr. Ishizuka, and Dr. Hori reported no conflicts of interest.

Consent for publication

The patients have provided written informed consent for the publication of the collected data.

Ethical considerations

All methods were performed in accordance with relevant guidelines and regulations. The study that analyzed the Remudy data was approved by the ethics committee of the National Center of the Neurology and Psychiatry (NCNP). (No. A2022-001) The questionnaire survey on QOL of DMD patients was approved by the ethics committee of both NCNP and Keio University Faculty of Pharmacy. (No. B2022-023, 220808)

Consent to participate

The patients have provided written informed consent for participating studies.