Causal effect of conventional anti-dementia drugs on economic burden: an orthogonal double/debiased machine learning approach

Abstract

Background

The Inflation Reduction Act (IRA) did not introduce a cap on out-of-pocket (OOP) for newly approved Alzheimer’s Disease (AD) drugs, such as lecanemab which is covered under Medicare Part B. Therefore, expanding the use of conventional anti-dementia drugs is critical to addressing the growing economic burden of dementia. In this study, we aimed to evaluate the causal relationship between specific conventional anti-dementia drug use and various healthcare costs with the Double/Debiased Machine Learning (DML) approach.

Methods

Leveraging data from the Medicare Current Beneficiary Survey (MCBS) spanning 2015 to 2019, we utilized a nationally representative survey linked to Medicare data in this study. The presence of Alzheimer’s Disease and Related Dementias (ADRD) and anti-dementia drug use was determined through Medicare claims data. The health care costs were measured as total medical costs and categorized into Medicare costs, OOP costs, inpatient costs, and outpatient costs. Conventional anti-dementia drugs include Cholinesterase inhibitors (ChEIs) and N-methyl-D-aspartate receptor (NMDAR) antagonists. The DML techniques were employed to investigate causal relationships.

Results

A total of 12,764,487 weighted older adults with ADRD were included, with 34.60% of them using anti-dementia drugs. Using anti-dementia drugs could significantly reduce Medicare costs and inpatient costs by $4,804.26 and $2,842.48 on average (P < 0.001), while did not significantly influence total costs, OOP costs, and outpatient costs. ChEIs use could help decrease Medicare costs and inpatient costs significantly (P < 0.05), whereas the NMDAR antagonist (memantine) showed no statistically significant effect across all cost types. Both donepezil and rivastigmine could help significantly decrease Medicare costs and inpatient costs (P < 0.001). Additionally, anti-dementia drug use could significantly reduce Medicare costs and inpatient costs among non-Hispanic Whites, and significantly lower inpatient costs among non-Hispanic Blacks (P < 0.05).

Conclusion

This study revealed the causal relationship between anti-dementia drug use and Medicare costs by employing DML. ChEIs were found to be contributors to the decreased Medicare costs and inpatient costs, which could mainly be attributed to donepezil. The use of donepezil should be expanded, considering the significant benefits. Furthermore, a lower OOP cap for ADRD beneficiaries should be established under the IRA.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12877-025-06298-6.

Introduction

The Inflation Reduction Act (IRA) introduced a cap on out-of-pocket (OOP) drug costs under Medicare Part D starting in 2024, as well as requiring Part D plans and drug manufacturers to pay a greater share of costs for Part D enrollees with high drug costs, offering considerable financial relief to Medicare beneficiaries [1]. However, there is no similar cap on OOP spending for newly approved Alzheimer’s Disease (AD) drugs, such as lecanemab which is covered under Medicare Part B [2]. Additionally, manufacturers of biologic drugs, such as lecanemab, will be exempt from price negotiation with the Centers for Medicare & Medicaid Services (CMS) for 13 years following the drug’s approval date [2], putting a greater economic burden on dementia patients.

Due to the increasing elderly population in the US, the annual number of new Alzheimer’s Disease and Related Dementias (ADRD) cases is expected to double by 2050 [3]. This corresponds with the escalating economic and clinical burden observed in this disease state. After being diagnosed, a patient may live up to 20 years with the condition, with family members more often than not being the primary caregivers [3]. In 2023, over 11 million unpaid caregivers dedicated 18.4 billion hours to caring for individuals with ADRD, the value of unpaid dementia caregiving in 2023 was estimated at $346.6 billion [4]. On average, Medicare payments for beneficiaries aged 65 + with dementia are nearly three times higher than for those without dementia [4]. In 2024, the total cost of health care, long-term care, and hospice services for individuals with dementia is projected to reach $360 billion [4]. Therefore, expanding the use of conventional anti-dementia drugs is critical to addressing the growing economic burden of dementia.

Cholinesterase inhibitors (ChEIs) and N-methyl-D-aspartate receptor (NMDAR) antagonists are conventional drugs approved for the treatment of dementia by the US Food and Drug Administration (FDA) previously [5]. These medications are indicated to treat the associated cognitive symptoms and slow disease progression. The ChEIs exert their action by inhibiting the enzyme cholinesterase, which in turn halts the breakdown of acetylcholine leading to increased levels of the neurotransmitter which is involved in neuron signaling related to memory and learning [6]. Drugs from this class used in the treatment of ADRD include donepezil, rivastigmine, and galantamine [7]. The NMDAR antagonists work by blocking the excitatory neurotransmitter glutamate from binding to the N-methyl-D-aspartate receptor [8]. This receptor plays an integral role in the development of new memories but overstimulation by glutamate can lead to excitotoxicity and cell death [8]. The only drug in this class approved for the treatment of ADRD is memantine [7].

Additionally, a previous case-control study involving 687 patients from 1999 to 2002 demonstrated that donepezil therapy in routine clinical practice significantly reduced overall healthcare costs for patients with mild to moderate ADRD [9]. Patients treated with donepezil incurred $2,500 less in annual medical costs than controls, primarily due to shorter hospital stays and lower skilled nursing facility expenses, despite having more physician office visits. These findings reinforce the economic value of donepezil and support the broader conclusion that certain ChEI treatments can yield both clinical and economic benefits over time. Another longitudinal study involving 280 in Utah from 2002 to 2013 found that use of anti-dementia drugs was associated with 32% lower informal costs [10].

However, existing studies are outdated and have focused on aggregate cost comparisons without examining cost components by Medicare coverage type. The five-year incremental costs for Medicare Part A services among dementia patients were reported to be $17,717, whereas the additional costs for services under Part B over the same period were not statistically significant [11]. There is a noticeable literature gap in the causal relationship between the use of specific conventional anti-dementia drugs and various healthcare expenditures. Addressing this gap is extremely important as the projected number of patients suffering from ADRD is expected to grow substantially. While traditional techniques have been widely applied to address confounding in observational settings, they often rely on strong assumptions and limited flexibility in handling high-dimensional data. The Double/Debiased Machine Learning (DML) method offers a robust alternative by allowing for flexible control of confounders and yielding valid treatment effect estimates under weaker conditions, which is especially valuable in health economics research [12]. In this study, we aim to fill this literature gap by evaluating the causal relationship between conventional anti-dementia drug use and healthcare costs with the Double/Debiased Machine Learning (DML) approach.

Methods

Study design and data source

This study used a pooled cross-sectional design utilizing data from the Medicare Current Beneficiary Survey (MCBS) [13], developed by the Centers for Medicare & Medicaid Services (CMS). MCBS employs a sophisticated survey methodology involving a stratified multistage sampling approach and computer-assisted personal interviews. It is designed to be nationally representative and provides extensive data by surveying participants. Furthermore, it links the survey data to their corresponding Medicare claims data. The Medicare claims include Part A inpatient, Part B outpatient, and Part D prescription information on the diagnosis, health care utilization, and health care costs. The linked data enable this study to examine anti-dementia drug utilization and costs related to patients with ADRD.

Patient selection

This study included Medicare beneficiaries who had a diagnosis of ADRD, were 65 years and over, and were included in the MCBS spanning from 2015 to 2019. All participants were restricted to individuals with Medicare Parts A, B, and D.

Variable definitions

The presence of ADRD and anti-dementia drug use was determined through Medicare claims data. The ADRD patients were identified using the International Classification of Diseases, Ninth and Tenth Revisions (ICD-9-CM, ICD-10-M) diagnoses codes for dementia as defined by the Chronic Conditions Warehouse (CCW) [14]. (Supplementary Table 1) The ADRD in our study included Alzheimer’s disease, vascular dementia, frontotemporal dementia, unspecified dementias, and other neurodegenerative and cognitive disorders commonly associated with ADRD.

The health care costs were measured as total medical costs, and categorized into Medicare costs, out-of-pocket (OOP) costs, inpatient costs, and outpatient costs from Medicare Part A (inpatient), B (outpatient/physician), and D (prescription drugs) claims. Conventional anti-dementia drugs in this study consisted of two classes, including ChEIs and NMDAR antagonist. Specifically, ChEIs included the medications rivastigmine, donepezil, and galantamine, while memantine was classified as the NMDAR antagonist. Anti-dementia drug users were defined as individuals with at least one prescription drug fill for ChEIs or NMDAR antagonist during the observation year, using Medicare Part D claims. Those without any such fills were classified as non-users. Healthcare costs and Drug use were measured over the same calendar year. All costs were converted to 2024 US dollars using the Consumer Price Index (CPI).

Based on the newly released National Institute on Aging (NIA) Health Disparities Research Framework [15], a total of 56 covariates were used in our study, including and biological factors (e.g., age, sex, race), environmental factors (e.g., residence, cost-related medication nonadherence), sociocultural factors (e.g., education level, income), and behavioral factors (e.g., activities of daily living, instrumental activities of daily living). (Supplementary Table 2)

Statistical analysis

Traditional regression models are often limited by assumptions of linearity and may struggle to account for high-dimensional confounding, leading to biased estimates in observational studies. To overcome these challenges and strengthen causal inference, we employed Double/Debiased Machine Learning (DML) [16] to obtain more accurate and unbiased estimates of the causal effects of anti-dementia drug use on healthcare costs.

DML flexibly models both treatment assignment and outcome using machine learning algorithms, while applying sample splitting and Neyman orthogonality to minimize bias [16]. This orthogonalization effectively isolates the causal effect of interest from the influence of high-dimensional covariates, reducing bias from both regularization and overfitting. Subsampling techniques, such as cross-fitting, further enhance estimation accuracy. Given the complex, nonlinear relationships and large number of covariates in real-world healthcare data, DML is well-suited to produce more robust and unbiased estimates of the causal effects of anti-dementia drug use on healthcare costs.

In our study, we utilized LASSO (Least Absolute Shrinkage and Selection Operator) for predicting both the outcome and the treatment variable. We opted for LASSO regression for several reasons: Firstly, LASSO regression is effective at preventing overfitting and dealing with multicollinearity considering many similar covariates in this study. Additionally, despite being a machine learning technique that optimizes parameters based on the training data, LASSO regression offers a high degree of interpretability. This interpretability is particularly valuable for public health practice [17], as it allows us to examine and understand the coefficients derived from the model.

To estimate the effect of anti-dementia drug use on cost change, we first used machine learning models to estimate each individual’s probability of receiving the drug (treatment model) and their expected change in costs regardless of treatment (outcome model), based on patients’ characteristics. Finally, DML estimated the average treatment effect by comparing the adjusted cost changes between drug users and non-users, using a cross-fitting procedure to improve estimation accuracy and reduce bias.

In this study, individuals with missing values for costs and anti-dementia drug use information were excluded. For covariates with missing values, the missing values were treated as a distinct category, and participants with missing data for these variables were retained in the analysis. In addition, we conducted sensitivity analyses by using multiple imputation for missing variables for each anti-dementia drug. Specifically, we generated five imputed datasets, using logistic regression (“logreg”) for binary variables, multinomial regression (“polyreg”) for multi-categorical variables.

The differences in patient characteristics among patients using or not using anti-dementia drugs were compared using Chi-square tests. Survey sampling weights were applied in this study to generate national estimates. P-value less than 0.05 was considered statistically significant. R package “DoubleML” was employed to conduct DML, and R package “mice” was used to perform multiple imputation in this study.

Results

After applying the inclusion and exclusion criteria, a total of 12,764,487 weighted older beneficiaries with ADRD were included in this study, with 34.60% of them using anti-dementia drugs. (Table 1) Most respondents were aged 85 and over (43.27%), female (62.03%), non-Hispanic Whites (83.66%), widowed (43.14%), high school graduate (41.13%), with income between $10,000 and $24,999 (35.83%), residing in metropolitan areas (78.38%), and living in the South (40.11%). (Table 1) Compared to those not using anti-dementia drugs, anti-dementia drug users were more likely to be young and married (P < 0.001). (Table 1)

Table 1.

Characteristics of samples

			Anti-dementia Drug Users		P value
		All	No	Yes
		N = 12,764,487	N = 8,347,410	N = 4,417,076
		%	%	%
Age					< 0.001
65–74		22.68	24.19	19.82
75–84		34.05	31.29	39.27
≥ 85		43.27	44.52	40.91
Sex					0.858
Male		37.97	37.83	38.25
Female		62.03	62.17	61.75
Race					0.387
Non-Hispanic Whites		83.66	83.29	84.37
Non-Hispanic Blacks		9.67	10.16	8.76
Hispanics		2.77	2.39	3.50
Others		3.02	3.11	2.83
Missing		0.88	1.05	0.54
Married Status					< 0.001
Married		39.84	37.20	44.83
Widowed		43.14	43.68	42.12
Divorced/Separated		11.32	12.58	8.95
Single		5.14	5.99	3.52
Missing		0.56	0.55	0.58
Education Level					0.154
< High School		10.49	10.89	9.73
High School		41.13	39.61	44.02
>High School		37.13	37.65	36.16
Missing		11.24	11.85	10.08
Income					0.872
< $10,000		13.15	13.27	12.93
$10,000 - $24,999		35.83	36.32	34.90
$25,000 - $49,999		22.64	22.23	23.42
≥ $50,000		28.38	28.18	28.74
Residence					0.881
Metropolitan		78.38	78.29	78.55
Non-metropolitan		21.62	21.71	21.45
Census					0.206
Northeast		18.77	19.08	18.19
Midwest		23.45	22.38	25.45
South		40.11	39.76	40.77
West		17.67	18.77	15.59

Additionally, users of donepezil, galantamine, rivastigmine, and memantine accounted for 24.57%, 1.08%, 4.32%, and 16.55%, respectively. (Supplementary Table 3) From 2015 to 2019, individuals who did not use anti-dementia medications incurred higher average total healthcare expenditures ($73,020.33) compared to those who did use these medications ($64,988.81). (Supplementary Table 4) This cost differential was consistent across all major expenditure categories, including Medicare costs ($32,331.57 vs. $26,285.89), OOP costs ($19,814.57 vs. $17,929.47), inpatient costs ($11,054.48 vs. $7,924.02), and outpatient costs ($4,111.24 vs. $3,977.59). (Supplementary Table 4)

Results from DML showed that using anti-dementia drugs could significantly reduce Medicare costs and inpatient costs by $4,804.26 and $2,842.48 on average (P < 0.001), while not significantly influencing total costs (costs changing: -$1,612.07; 95% CI: $4,668.72, $1,444.59) and OOP costs (costs changing: -$36.63; 95% CI: -$1,455.29, $1,382.04), and outpatients costs (costs changing: $61.59; 95% CI: -$597.83, $721.02) (Table 2). ChEIs use could help decrease Medicare costs and inpatient costs significantly (P < 0.05), whereas the NMDAR antagonist (memantine) showed no statistically significant effect across all cost types. (Table 2) Further exploring the effects of ChEIs use on costs by specific drugs, results showed that both donepezil and rivastigmine could help significantly decrease Medicare costs and decrease inpatient costs (P < 0.001). (Fig. 1) Additionally, galantamine could help significantly decrease outpatient costs by $1,812.24 (P = 0.003). (Fig. 1)

Table 2.

Effects of Anti-dementia drug use on costs by drug class

	Anti-dementia Drugs				ChEIs				NMDAR Antagonist (Memantine)
	Costs Changing ($)	95% CI		P value	Costs Changing ($)	95% CI		P value	Costs Changing ($)	95% CI		P value
Total Costs	−1,612.07	−4,668.72	1,444.59	0.301	−2,385.62	−5,578.66	807.41	0.143	1,644.27	−2,722.50	6,011.04	0.461
Medicare Costs	−4,804.26	−6,904.09	−2,704.44	< 0.001	−5,267.62	−7,434.26	−3,100.98	< 0.001	−3,216.34	−6,447.73	15.04	0.051
OOP Costs	−36.63	−1,455.29	1,382.04	0.960	−458.08	−1,917.07	1,000.92	0.538	648.58	−1,250.17	2,547.33	0.503
Inpatient Costs	−2,842.48	−4,020.36	−1,664.60	< 0.001	−3,044.49	−4,230.54	−1,858.43	< 0.001	−1,993.62	−4,044.24	57.00	0.057
Outpatient Costs	61.59	−597.83	721.02	0.855	−116.40	−696.44	463.64	0.694	253.75	−779.82	1,287.33	0.630

ChEIs Cholinesterase inhibitors, NMDAR N-methyl-D-aspartate receptors, OOP out-of-pocket

Fig. 1.

Effects of cholinesterase inhibitors use on costs OOP out-of-pocket

Racial/ethnic (Supplementary Table 5) and sex (Supplementary Table 6) disparities in the effects of anti-dementia drugs on total costs were also evaluated. Among non-Hispanic Whites, anti-dementia drug use could significantly reduce Medicare costs (costs changing: -$5,491.28; P < 0.001) and inpatient costs (costs changing: -$2,788.01; P < 0.001). Among non-Hispanic Blacks, drug use could significantly lower inpatient costs (costs changing: -$5,703.83; P = 0.006). For the other racial/ethnic group (including Hispanics and others), no significant effects were observed across any cost category. Among both males and females, anti-dementia drug use could significantly reduce Medicare costs and inpatient costs (P < 0.05).

Sensitivity analysis was conducted to examine the robustness of results by using multiple imputation for missing variables, and the results further confirmed the cost-reducing effects of anti-dementia drug use, particularly in Medicare and inpatient costs. (Supplementary Tables 7 & 8) Notably, sensitivity analysis also revealed statistically significant reductions in total costs among anti-dementia drug users (costs changing: -$3,978.64; P = 0.020), ChEIs users (costs changing: -$4,847.38; P = 0.007), and donepezil users (costs changing: -$6,165.65; P = 0.001), which was not observed in the primary analysis.

Discussion

This study revealed the causal relationship between conventional anti-dementia drug use and Medicare costs by employing DML. Results showed that anti-dementia drugs use could significantly reduce annual Medicare costs and inpatient costs by $4,804.26 and $2,842.48 on average. ChEIs were found to be contributors, specifically, donepezil played a crucial role in reducing Medicare costs and inpatient costs. The study didn’t observe significantly reduced total costs, OOP costs, and outpatient costs resulting from conventional anti-dementia drug use.

ChEIs were the first drugs granted approval in the US for the treatment of AD. These drugs operate by blocking the action of the acetylcholinesterase enzyme in the brain, leading to a boost in the levels of acetylcholine at the synaptic cleft, which in turn enhances cholinergic neurotransmission. According to a systematic review, the likelihood that ChEIs are more cost-effective than best supportive care (BSC) exceeds 99%.¹⁸ This finding is consistent with this study, demonstrating that ChEIs have the potential to decrease Medicare costs and inpatient costs. A retrospective cohort study found that ChEIs were associated with shorter duration of hospital admissions and decreased risk of mortality [18]. Therefore, ChEIs could potentially lead to cost savings for patients in terms of reduced inpatient expenditures, despite the additional expense incurred for the medication itself. Although ChEIs have similar clinical mechanisms, the variation in their effects on healthcare costs was observed. This may be explained by differences in approved indications (e.g., donepezil is approved for all stages of AD), real-world prescribing patterns, and adherence associated with different formulations (e.g., oral vs. transdermal rivastigmine). Additionally, sample size variation, particularly the smaller number of galantamine users, may have limited the statistical power to detect significant cost effects for some agents.

The benefits of ChEIs contributing to decreased costs could be mainly attributed to donepezil. Cost-effectiveness analysis of donepezil among AD patients suggested that donepezil was cost-effective stemming from the improvement in cognitive function, which led to a reduction in care expenses and delayed progression to more expensive disease stages and care settings [19]. A systematic review also demonstrated that donepezil is the most cost-effective among ChEIs drugs [20]. In addition, a retrospective case-control study using Medicare managed care claims data found that donepezil use significantly reduced total healthcare costs, particularly inpatient and skilled nursing facility expenses [9]. Therefore, our results have been consistent with studies using different methodological approaches, including modeling-based cost-effectiveness analysis and real-world retrospective observational designs.

Additionally, findings from other countries, such as Germany [21] and the UK [22]also corroborate these results. Hartz et al. (2012) [21] used discrete event simulation modeling to evaluate donepezil in the German healthcare context and found it to be a cost-effective intervention for AD, primarily through delaying institutionalization and reducing care costs. Similarly, Getsios et al. (2010) [22] applied a discrete-event simulation model in the UK and demonstrated that donepezil was cost-effective in treating mild to moderate AD, with improved quality-adjusted life years and reduced long-term care needs. These international findings reinforce the economic value of donepezil and support the broader applicability of our results across healthcare systems with varying structures and financing mechanisms. The benefits of donepezil could be elucidated by its clinical efficacy. Presently, while all commonly prescribed ChEIs are approved for managing AD in its mild-to-moderate stages, donepezil stands out as the sole selective ChEIs that has approval for all AD stages, encompassing mild, moderate, and severe conditions [23]. Donepezil offers several distinct advantages in comparison to other ChEIs, including extended half-life, improved tolerability, minimal food effects, limited drug interactions, and efficient oral absorption [23].

Notable racial/ethnic differences in the cost-related effects of anti-dementia drug use were observed as well. Among non-Hispanic White patients, significant reductions were observed in Medicare costs and inpatient costs, suggesting that this group may benefit more directly from pharmacologic treatment in terms of healthcare savings. In contrast, the estimated cost changes for non-Hispanic Black patients and patients of other races/ethnicities were not statistically significant across most cost categories, with the exception of inpatient cost reduction among non-Hispanic Blacks. These findings may reflect underlying disparities in access to follow-up care, quality of care, or differences in disease severity and treatment adherence across racial/ethnic groups. Prior studies have consistently documented racial disparities in dementia care and outcomes, including differential diagnosis rates, access to healthcare services, and medication utilization patterns [24–26]. Structural factors such as healthcare system bias, socioeconomic inequality, and geographic barriers may further contribute to these differences [27]. Our results highlight the importance of tailoring dementia care strategies to address racial and ethnic disparities, not only in clinical outcomes but also in economic impact.

This was the first study exploring the causality between conventional anti-dementia drug use and medical costs by taking advantage of DML which offers a solution to eliminate the bias in causal inference. Our results suggest that expanding the use of donepezil should be considered by policymakers given the significant benefits of donepezil. Additionally, our study found that none of the conventional anti-dementia drugs contributed to a significant reduction in OOP expenses for ADRD patients. To alleviate the economic burden for ADRD patients, further measures might need to be implemented, such as establishing a lower OOP cap for Medicare beneficiaries under the IRA.

Several limitations should be noted. First, ADRD diagnoses were identified using Medicare claims data, which may underestimate the true prevalence of the condition, as many individuals with cognitive impairment remain undiagnosed or are not captured through billing codes. Second, although we examined the causal effects of specific anti-dementia drugs, the severity of ADRD, which plays a critical role in treatment selection and outcomes, could not be directly measured due to the lack of clinical indicators in claims data. This may partially explain the lack of significant cost impact observed for NMDAR antagonists such as memantine, which are more often prescribed in advanced stages. To address this, we included ADLs and IADLs as proxies for functional severity. Third, our analysis did not account for informal caregiving costs, such as unpaid caregiver time and related burden. Although we included both Medicare expenditures and OOP costs, these estimates do not capture the broader societal costs of ADRD, potentially underestimating its full economic burden. Fourth, while DML helps reduce bias from regularization and overfitting in high-dimensional models, its validity relies on key assumptions, such as no unmeasured confounding, sufficient sample size, and proper model specification. Thus, while DML improves robustness, it does not eliminate all sources of bias and should be interpreted within these methodological constraints. Lastly, this study was limited to four conventional anti-dementia drugs due to the limitation of MCBS data, further studies should include newly covered anti-dementia drugs such as lecanemab and donanemab.

Conclusion

This study revealed the causal relationship between anti-dementia drug use and Medicare costs by employing DML. ChEIs were found to be contributors to the decreased Medicare costs and inpatient costs, which could mainly be attributed to donepezil. The use of donepezil should be expanded, considering the significant benefits. Furthermore, a lower OOP cap for ADRD beneficiaries should be established under the IRA.

Authors’ contributions

ZKL and XJ conceived the study. XJ performed the statistical analyses and drafted the manuscript. JF drafted the manuscript. GL, ML, and ZKL reviewed/edited the manuscript. All authors gave full approval of the version to be published.

Funding

This research did not receive any supporting funding.

Data availability

All data generated or analyzed during this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

The University of South Carolina Institutional Review Board approved this study. Because we used deidentified data from the Medicare Current Beneficiary Survey (MCBS), the University of South Carolina Institutional Review Board has waived the need for informed consent for this study. Specifically, the Office of Research Compliance, on behalf of the Institutional Review Board, approved the referenced study. This study was conducted in accordance with the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.