Healthcare resource utilisation and costs associated with AL amyloidosis: a retrospective matched cohort study

We conducted a retrospective population-based, matched cohort study using the National Health Insurance Research Database to estimate healthcare resource utilisation (HRU) and costs in patients with newly diagnosed AL amyloidosis in Taiwan. Cases were matched 10:1 by age, sex, and area of residence to patients without AL amyloidosis (comparators) randomly selected from the database during the same time period. Annual all-cause HRU and costs for 3 years were quantified. AL amyloidosis-attributable costs were obtained by subtracting all-cause HRU costs incurred by comparators from cases. The mean age of all patients was 60.78 years and 59.07% were male. Co-morbidities were more frequent in cases than comparators. By 6 months after diagnosis, 12.1% of cases had died versus 0.9% of comparators. In the first year, cases had 103% more outpatient visits, 177% more emergency room visits, were hospitalised 4-times more frequently, and spent 5.5-times more days in hospital than comparators, and total healthcare costs were > sixfold higher. Costs incurred during the first year after diagnosis accounted for 55% of the 3-year cumulative cost. High HRU costs associated with delayed diagnosis and end-organ damage indicate a need for earlier diagnosis and more effective treatments for AL amyloidosis.

Survival can be a matter of months in patients who present with advanced disease and extensive cardiac involvement, whereas patients diagnosed at an early stage can survive for more than 10 years 10 .The burden of disease for patients with AL amyloidosis is substantial-the heart is involved in 50-70% of patients, usually presenting as heart failure 11 .Approximately one-third of patients with renal involvement will eventually require dialysis, and a proportion of patients require organ transplant in order to tolerate chemotherapy 11 .A systematic review of the literature published in 2017 found that quality of life is significantly affected in patients with AL amyloidosis 11 .
Information about the baseline healthcare-related costs of AL amyloidosis is needed to inform health technology assessments for new treatments.Studies using healthcare claims databases in the United States (US) have reported that the total all-cause healthcare cost in the first year after a diagnosis of AL amyloidosis was US $122,180 12 .The cost of healthcare in the first year in patients with relapsed/refractory AL amyloidosis was US $139,143 13 .Outside of the US there has been little published about the cost of healthcare in patients diagnosed with AL amyloidosis 11 , and the burden of AL amyloidosis on healthcare systems in Asia is not known.
To bridge the gap in knowledge about the real-world economic burden of AL amyloidosis we conducted a population-based study using a national health insurance claims database to examine the healthcare resource utilisation (HRU) and costs associated with AL amyloidosis in Taiwan.AL amyloidosis-attributable HRU and costs were assessed by comparing outcomes in patients with and without AL amyloidosis.

Data source
This study analysed the National Health Insurance Research Database (NHIRD) provided by National Health Insurance (NHI) Administration and maintained by the Data Science Centre of the Ministry of Health and Welfare in Taiwan.The NHI programme is a compulsory national health insurance which covers 99.9% of the Taiwanese population, which was approximately 24 million people in 2020 14 .Once enrolled, members obtain easy access to a comprehensive benefits package that includes outpatient and inpatient services, dental care services, traditional medicine, prescription drugs, and laboratory and imaging examinations.Data captured by the NHI are released into the NHIRD, a large population-based claims database that contains information on all medical services provided by NHI-contracted hospitals, physicians, and pharmacies to residents throughout Taiwan.Demographic information as well as the dates and types of services received are captured in the database.All claims in the database are accompanied by diagnosis codes.International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes were available in the database up to December 31, 2015 and ICD-10-CM coding was introduced from January 1, 2016.All patients enrolled into the NHIRD have a unique ID that enables all healthcare episodes to be captured regardless of which clinic/hospital the patient attended.The survival status of members was verified by linking NHIRD to the Death Registry using scrambled identification numbers.
De-identified patient data were used for the analysis.The study was granted an exemption from ethical review by the Taipei Medical University-Joint Institutional Review Board, and an exemption from the need for patient consent.The study was conducted according to all applicable guidelines and regulations set by the Health and Welfare Data Science Center (HWDC).

Study design and population
This is a retrospective population-based, matched cohort study to evaluate HRU and costs from the payer's perspective.The study cohort included patients with newly diagnosed AL amyloidosis identified from the NHIRD using ICD-10 codes E85.4 (organ-limited amyloidosis), E85.8 (other amyloidosis), and E85.9 (amyloidosis, unspecified) between January 1, 2016 and December 31, 2018.The index date was defined as the first eligible occurrence of one of these codes.Eligible patients were those who fulfilled both of the following criteria: (1) had at least one inpatient claim(s) or at least two outpatient claims with primary or secondary diagnosis of AL amyloidosis, and with any pair of claims coming at least 30 days apart; and (2) underwent any biopsy procedure within 12 months prior to, or up to 6 months after the index date.Patients with any claim of amyloidosis with ICD-9 code 277.3 in 2015 in the NHIRD were excluded.Patients were also excluded if their initial diagnosis switched to another amyloidosis code (ICD 10: E85.0, E85.1, E85.2, E83.3) or if they had a record of non-AL amyloidosis (ICD-9: 277.3, ICD-10: E85.1) in the Catastrophic Illness database.Registration in the Catastrophic Illness database is considered confirmed because it requires meeting strict diagnostic evaluation criteria including biopsy confirmation and evaluation by two experts.
A comparator cohort of patients without AL amyloidosis was identified by randomly selecting individuals who had no claim for a diagnosis of amyloidosis from 2008 to 2019 in the database.Patients with newly diagnosed amyloidosis were matched with comparators in a ratio of 1:10 based on age, sex, and residential area, categorised into six areas defined by the NHI; Taipei area, Northern area, Central area, Southern area, Eastern area, and Kaohsiung and Ping-Tung area.Comparators were assigned to a reference date, which was the index date of the corresponding matched case.
The baseline period was defined as 365 days prior to the index/reference date.Relevant comorbidities (i.e., those potentially associated with or caused by amyloidosis) [15][16][17][18][19] were defined using at least two outpatient or one inpatient claim(s) using ICD-9 and ICD-10 codes (Supplementary Table 1) during the baseline period.Co-morbidities were categorised into six major categories (cardiac-related conditions, liver-related diseases, renal-related diseases, pulmonary diseases, neuropathy, and malignancy).

Outcomes
All-cause HRU and costs were captured for both cohorts on a per-patient basis from the index/reference date up until three years after the index date, death, or end of data availability (December 31st, 2019), whichever occurred first.Outpatient, inpatient, and emergency department services were evaluated.Total costs were further broken down by medication costs and non-medication costs (including fees for physician consultations, diagnostic tests, laboratory examinations, injection procedure costs, surgery and inpatient stays).By subtracting the all-cause HRU and costs incurred by matched comparators from the all-cause HRU and costs incurred by patients with AL amyloidosis (i.e., incremental HRU and costs), an estimate of the HRU and costs attributable to AL amyloidosis was obtained.
Costs were not inflated to a base year because health insurance claims within the NHI are mainly based on fixed fee schedules, and thus were reported as documented when the claims were filed from 2016 to 2019.Costs are presented in New Taiwan dollars (NT$).In mid-2022, 100 NT$ converts to around 3.2 Euros and 3.5 USD.

Statistical analysis
Mean, standard deviation (SD), median, and interquartile range (IQR) were used to describe continuous variables, and frequencies and percentages were used to describe categorical variables.T-tests and Chi-square tests were applied to test the differences between the matched patients with AL amyloidosis and comparators for continuous and categorical variables, respectively.
Raw mean annual all-cause HRU and costs for 3 years were reported for patients with AL amyloidosis and comparators.Weighted Kaplan-Meier sample average (KMSA) estimates of HRU and costs were used to account for loss of follow-up due to death or disenrollment during the 3-year follow-up.This was achieved by charting the survival curves of patients with AL amyloidosis and comparators in the 3-year follow-up period using the Kaplan-Meier method.HRU and costs data within 3 years after the index date were then partitioned into 12 observation quarters for both cohorts.To account for the potential over-representation of zeros and right-skewed distribution of HRU and cost data, the mean quarterly HRU and costs were assessed using a two-part model: a logistic regression to predict the probability of service use and a generalised linear model to predict the frequency of HRU and costs among those with positive frequency of service.Covariates entered into the models to predict HRU and costs included age, sex, residential area, and the presence of any co-morbidity under the six relevant disease categories (cardiac-related conditions, liver-related diseases, renal-related diseases, pulmonary diseases, neuropathy, and malignancy).
The regression-predicted mean quarterly HRU, and costs of those patients still alive at the beginning of the quarter were weighted by the KMSA estimator, i.e., the probability of survival in the given quarter was conditional on having survived the previous quarter.This accounted for patient attrition over the study.The weighted estimates were then summed over to obtain an estimate of the mean censored adjusted cumulated HRU and costs up to the 3rd year [20][21][22][23] .This means that the weighted estimates were summed up to get an overall average HRU estimate, taking into account censoring (i.e., incomplete data) in the study population.By using weighted estimates, the results are representative of the larger population and the censored data are properly accounted for.This approach helps to provide a more accurate estimate by using all available information of HRU in the study population.
Since the mortality rate was relatively stable during the second and third year, the mean estimates of quarterly HRU and costs were collapsed into a semi-annual analysis for the second year and an annual analysis for the third year (Supplementary Tables 2 and 3).
Finally, a non-parametric bootstrap re-sampling method 7,8 was performed to calculate the KMSA cost estimates via recycled predictions using 1,000 bootstrap samples of patients with AL amyloidosis and comparators of equal size with replacement.Two-sided 95% confidence intervals (CIs) were constructed.
All statistical analyses were performed using SAS Version 9.4 (SAS Institute, Cary, NC, USA).

Results
In the 2016-2018 data from the NHIRD, 645 patients with newly diagnosed AL amyloidosis were identified with 6450 matched comparators (Table 1).The mean age of patients with AL amyloidosis and comparators was 60.78 years and 59.07% were male.Significantly more patients with AL amyloidosis had relevant co-morbidities; 27.29% of patients had cardiac-related conditions in the 365-day baseline period before their AL amyloidosis diagnosis compared to 12.11% of comparators, 23.41% had a malignancy compared to 4.43% of comparators and 21.71% had renal-related diseases compared to 3.80% of comparators.Eleven patients underwent autologous stem cell transplantation, all within the first 18 months after diagnosis.

Mortality
The Kaplan-Meier survival curve showed lower survival amongst patients with AL amyloidosis compared with comparators, and the majority of deaths occurred within the first 6 months after diagnosis (Fig. 1).Three months after the index date, 8.4% of patients with newly diagnosed amyloidosis died compared with 0.05% of comparators.By 6 months after the index date, 12.1% of patients had died, compared with 0.9% of comparators.By 36 months after the index date, 18.8% of patients had died compared with 3.7% of comparators.

Healthcare utilisation
Patients with newly diagnosed AL amyloidosis had levels of HRU that exceeded those of comparators.In the first year after the index date, patients with AL amyloidosis had a mean of 37.88 outpatient visits, 0.97 emergency room visits, and 1.15 hospitalisations, and spent 12.02 days in hospital (Fig. 2).www.nature.com/scientificreports/emergency room visits, were hospitalised 4-times more frequently, and spent 5.5-times more days in hospital.
The difference between patients with AL amyloidosis and comparators decreased in year 2 and year 3, but HRU continued to remain substantially higher in patients than comparators.HRU remained stable over the three follow-up years in comparators, whereas outpatient and emergency room visits, hospitalisations and days spent in hospital were all higher in the first year in patients with newly diagnosed AL amyloidosis and decreased thereafter.The biggest change from year 1 to year 2 was observed in the number of hospitalisations and days spent in hospital, which decreased by 56% and 38%, respectively in year 2 compared to year 1.Outpatient visits declined between year 2 and year 3, but hospitalisations and emergency room visits in patients increased in year 3 compared to year 2 (Fig. 2).
We explored HRU in patients and comparators for a period of 1 year prior to the index date.HRU in patients subsequently diagnosed with AL amyloidosis well exceeded that of comparators across all domains in the preceding year (Table S4).

Costs of healthcare utilisation
The mean total annual all-cause cost of HRU was NT$ 294,490 (USD 9,347) per patient in the first year after the index date versus NT$ 47,570 (USD 1,510) in comparators (Fig. 3A).In the first year, all-cause costs of outpatient visits accounted for 61% of the total cost (NT$ 179,740 [USD 5,705]), followed by hospitalisations (NT$ 108,450 [USD 3,442]) and emergency room visits (NT$ 6,300 [USD 200]).NT$ 121,250 (USD 3,848) was accrued in medication costs and NT$ 173,240 (USD 5498) in non-medication costs (Fig. 3B).
Costs accrued by hospitalisation and for medication were proportionally higher for patients with newly diagnosed AL amyloidosis than for comparators.Hospitalisation and medication costs accounted for 37% and 41%, respectively, of the all-cause total cost for year 1 in patients with AL amyloidosis.
Total mean total annual all-cause costs in patients decreased to NT$ 175,400 (USD 5,567) in year 2 and to NT$ 152,180 (USD 4,830) in year 3 yet remained at least threefold higher than costs accrued by comparators in each follow-up year.www.nature.com/scientificreports/Cumulative mean annual costs of HRU were significantly higher in patients with newly diagnosed AL amyloidosis than comparators over the entire 3-year follow-up period (P < 0.0001 for all comparisons) (Table 2).The mean total cost of HRU attributable to AL amyloidosis was NT$ 449,643 (SD 36,419) (USD 14,271, SD 1,156) after 3 years, made up of NT$ 259,560 (SD 20,379) (USD 8238, SD 646) for outpatient visits, NT$ 148,207 (SD 20,704) (USD 4704, SD 657) for inpatient episodes, and NT$ 13,242 (SD 4,575) (USD 420, SD 145) for emergency room visits (Table 2).Outpatient care contributed about 62%-64% and inpatient care 34%-36% to the 1-year, 2-year, and 3-year cumulated total costs (Fig. 4A).Costs incurred during the first year after diagnosis accounted for 55% of the 3-year cumulative cost.The total cost of all-cause healthcare after 3 years was fourfold higher in patients with AL amyloidosis compared to comparators.
Similarly, cumulative mean annual medication and non-medication costs were higher in patients than comparators over the entire 3-year follow-up period (P < 0.0001 for all comparisons) (Table 3).Medication costs consisted about 45%-48% of the 1-year, 2-year, and 3-year cumulated total costs (Fig. 4B).The mean total cost of medications attributable to AL amyloidosis was NT$ 212,606 (SD 27,497) (USD 6,748, SD 872) after 3 years, and the mean non-medication cost was NT$ 257,232 (SD 25,720) (USD 8,164, SD 816).58% of the AL amyloidosisattributable medication costs and 52% of the attributable non-medication costs over the 3-year follow-up period were incurred in the first year after diagnosis.
Frequency of care and all health services, medication, and non-medication costs were substantially higher in the year prior to the index date in patients subsequently diagnosed with AL amyloidosis versus comparators (Table S5).

Discussion
We used the NHIRD to identify patients in Taiwan with a new diagnosis of AL amyloidosis between 2016 and 2018 and assessed their mortality and HRU over the first three years after diagnosis.Patients with newly diagnosed AL amyloidosis had high levels of co-morbidity and high early mortality, with 12.1% of patients dying within 6 months after diagnosis.HRU and associated costs were substantially higher in patients than in comparators, and were highest in the first year after diagnosis, which likely reflects the burden of morbidity and early mortality amongst those patients who presented with advanced disease.Delays in diagnosis and the late presentation of patients with severe end organ damage are well recognised challenges in the management of AL amyloidosis 2 .High rates of early mortality persist, despite improvements in overall survival that have followed the availability of more effective treatment options 2,24 .Higher costs incurred during the first year after diagnosis were observed in patients with AL amyloidosis in the US 25 ; total all-cause healthcare costs decreased by 35% in the second year after diagnosis, which is similar to the 30% decrease in total all-cause healthcare costs observed in our study between year 1 and year 2. While total all-cause healthcare costs reduced by a further 13% in the third year after diagnosis in our study, total annual healthcare costs were still threefold higher than those in comparators.
We also observed higher HRU and associated costs in patients than in comparators in the year prior to diagnosis.Due to the insidious onset and chronicity of the disease, and because diagnosis may often be delayed, that HRU and cost estimates from the time of diagnosis do not reflect the entire burden of this disease on patients or the healthcare system.
There is no specific ICD code for AL amyloidosis and patients with AL amyloidosis may have multiple comorbidities.Clinical data are not captured in claims databases and it is not possible to attribute causality to  S3.
diseases that co-occur with AL amyloidosis.This means that disease-specific HRU and costs cannot be precisely identified in the NHIRD using claims with principal or secondary diagnosis codes of AL amyloidosis alone.By matching a cohort of patients without AL amyloidosis we were able to control for the occurrence of coincident co-morbidities in the regression model to determine the AL amyloid-attributable cost of healthcare over the first 3 years after diagnosis.We considered that relevant co-morbidities (i.e., those potentially associated with amyloidosis) in the AL amyloid population were unlikely to be present in the control population (without AL amyloidosis).
The AL amyloidosis-attributable cost was NT$ 247,730 (USD 7,862) in year 1, accumulating to NT$ 363,487 (11,537) by the second year, and NT$ 449,643 (USD 14,271) by the third year.
Potential limitations of our study are the absence of a specific diagnosis code in ICD-10 for AL amyloidosis, which is challenge common to all studies that use claims databases to evaluate this disease 12 .Such studies typically use a combination of disease and treatment codes to confirm the target diagnosis 12,13 .We were unable to incorporate treatment codes into our case definition because several therapies used for the treatment of AL amyloidosis, such as novel agents, were not re-imbursed by the NHI at the time of the study, and so were not captured in the NHIRD.Instead, we included a claim for any biopsy in the period 12 months before or 6 months after the index date as confirmatory evidence of an AL amyloidosis diagnosis.The validity of this approach has been reported previously 7 .Our study only considered direct healthcare costs and was not able to estimate the indirect costs associated with AL amyloidosis, or the effects on quality of life which is significantly impacted in AL amyloidosis 11 .Finally, we had no way of differentiating between localised and systemic amyloidosis in the claims database.
The main strength of our study is that the NHI is a mandatory health insurance for the whole population of Taiwan, and the NHIRD provides complete and longitudinal data collection from birth until death and provides a full picture of the treatment journey regardless of where the patient was treated.This eliminates issues related to data loss when patients are lost to follow-up.
Our study is the first to describe the healthcare burden associated with AL amyloidosis in an Asian country.We showed that patients with AL amyloidosis have a high burden of co-morbidities and suffer a high rate of early mortality in the first 6 months after diagnosis, suggesting that like other countries, disease awareness is

Figure 1 .
Figure 1.Kaplan-Meier estimates of survival in patients with newly diagnosed AL amyloidosis and comparators from the index date.

Figure 2 .
Figure 2. Crude means of annual outpatient visits, admissions, emergency room visits, and days spent in hospital, by each year after the index date for patients with newly diagnosed AL amyloidosis and comparators.Data are tabulated in TableS2.

Figure 3 .
Figure 3. Crude mean annual total costs in NT$ per patient broken down by outpatient visits, admissions, and emergency room visits (A); and by medication costs and non-medication costs, by each year after the index date for patients with newly diagnosed AL amyloidosis and comparators (B).Data are tabulated in TableS3.

Figure 4 .
Figure 4. KMSA estimates of cumulative annual total costs of AL amyloidosis per patient broken down by (A) outpatient (OP), inpatient (IP), and emergency room (ER) care; and (B) medication costs and non-medication costs from the index date up to the third year.

Table 1 .
Demographic and clinical characteristics of patients with AL amyloidosis and comparators.SD standard deviation, (Q1, Q3), interquartile range.*Relevantco-morbiditieswere identified using ICD codes (TableS1) with at least two outpatient or one inpatient claim(s) during the baseline period.

Table 2 .
KMSA estimates of cumulative annual total costs per patient broken down by outpatient, inpatient, and emergency department from the index date up to the third year, for patients with newly diagnosed AL amyloidosis and comparators.CI confidence interval; KMSA Kaplan-Meier sample average; SD standard deviation; NT$ New Taiwan dollars.*t test; ** In mid-2022, 100 NT$ coverts to around 3.2 Euros and 3.5 USD.

Table 3 .
KMSA estimates of cumulative annual total costs per patient broken down by medication costs and non-medication costs from the index date up to the third year, for patients with newly diagnosed AL amyloidosis and comparators.CI confidence interval; KMSA Kaplan-Meier sample average; SD standard deviation; NT$ New Taiwan dollars.*t test; ** In mid-2022, 100 NT$ coverts to around 3.2 Euros and 3.5 USD.