Abstract
Background
Updated data are needed to understand the current implications of Low-risk Myelodysplastic syndromes (LR-MDS) on healthcare resource use (HCRU) and associated costs. Objectives of this study were to describe these outcomes over time but also to compare costs according to transfusion dependency at baseline.
Methods
A retrospective observational study was carried out using the ESND (“Echantillon National du SNDS”) database. All incident adults newly treated for MDS were included between 2018 and 2022. Patients were followed from the start of specific treatment to death, the end of the study period, or loss to follow-up (FU). HCRU and related costs were described according to French National Health Insurance perspective in euro 2022. Reimbursed total costs per patient were computed per year of follow-up and according to transfusion status.
Results
Over the study period, 267 patients newly treated for LR-MDS were included. Median total costs related to HCRU/patient for the entire FU period were estimated at €41,698 (IQR: 19,093; 84,111) with a median cost/patient/year of FU ranging from €26,693 (Y1) to €20,418 (Y3). Main cost drivers were hospitalizations and use of specific drugs. TD patients had more hospitalizations (94.4% vs 89.3%) and transfusions (77.5% vs 47.4%), leading to statistically significantly higher costs after treatment initiation.
Conclusion
This nationwide study is the largest describing the economic burden of treated LR-MDS patients in France using administrative claims data. Elevated costs and substantial burden for treated patients and the health care system were observed. Transfusion dependence was associated with incremental health care resource utilization and related costs. Further studies are needed to assess the impact of emerging therapies in this indication.
Keywords: myelodysplastic syndromes, real-world data analyses, claims, epidemiology, medico-economics
Introduction
Myelodysplastic syndromes (MDS) are heterogeneous clonal disorders of hematopoietic stem and progenitor cells characterized by dysplasia and ineffective hematopoiesis. The majority of MDS are diagnosed as “lower-risk” diseases (LR-MDS).1,2 Patients with LR-MDS present lower risk of death or progression to acute myeloid leukemia in the early period after diagnosis. Thus, management of LR-MDS aims to reduce or correct cytopenia, particularly, symptomatic anemia,3 while in “higher-risk” MDS, it aims at delaying disease progression. Chronic anemia is associated with shorter survival and high clinical burden with complications including cardiovascular disorders, asthenia, increased risks of falls and related bone fracture.4,5 Supportive treatment for LR-MDS involves administration of erythropoiesis-stimulating agents (ESAs) as a first line of treatment.6 In clinical trial data, hematologic improvement is observed in 30–40% of patients with a median time-to-first transfusion of 20–22 months.2 Recently, luspatercept demonstrated a greater efficacy for treating anemia as a first line of treatment with 60% of patients with hematologic improvement and a median time-to-first transfusion of 31 months.7,8
Previous research investigated clinical burden of MDS and the consequences of transfusion dependence in patients. However, most studies have focused on clinical outcomes such as overall survival (OS) and acute myeloid leukemia (AML) progression rather than economic/healthcare resource utilization (HCRU) outcomes. Some studies showed that transfusion dependence is associated with higher health care resource utilization and related costs compared to transfusion independence.9–12 In the US, transfusion dependence was associated with an increased total cost of $31,233 per patient each year. Moreover, transfused patients had greater use of inpatient and outpatient care with significantly higher mean costs than patients without transfusion ($88,824 vs $29,519, p<0.001).11,12 Another study found a mean total cost of $17,815/patient-month in patients with transfusion dependence versus $7874/patient-month in patients with transfusion independence.10 Same results were observed in a nationwide cohort study conducted in Sweden between 2008 and 2017. In this study, costs ranged from $8805 ($6482–$11,625) in low-risk patients to $80,106 ($61,460–$95,792) in the very high risk category.13 Despite these previous results, an update with recent and extensive data is needed to understand the current implications of LR-MDS on healthcare resource use and associated costs. Contemporary estimates are particularly relevant in the context of an evolving therapeutic landscape, with the recent introduction of novel agents such as luspatercept that may substantially modify treatment patterns, transfusion burden, healthcare resource utilization, and associated costs. Specifically, costs after specific treatment initiation and evolution over time are lacking. Moreover, estimation of specific MDS costs such as costs related to transfusions after treatment initiation is needed.
To the best of our knowledge, very few studies were conducted in France to assess the economic burden of LR-MDS patients. A retrospective observational study conducted in 100 MDS patient with transfusion dependence estimated the 6-month median transfusion costs at €16,188 per patient (including the cost of the day care facility, transportation to and from the hospital, iron chelation, and lab tests).14 To complete these previous findings with up‑to‑date and extensive data, the Edelweiss study was conducted using exhaustive data from the French health national Insurance database (“Systeme National des données de Santé”, SNDS) to provide real-world evidence on the economic burden of LR-MDS after treatment initiation, complementing data from cancer registries and existing cohorts. As previously described, the SNDS is used for billing and care management purposes and represents an exhaustive national resource to perform economic analyses of reimbursed health care resources in France.15 Therefore, the main objective of this study was to describe healthcare resource utilization (HCRU) and associated costs among incident treated LR-MDS patients in France but also to compare these costs according to transfusion dependency at treatment initiation. The secondary objectives were to describe HCRU and related costs over time (before and after treatment initiation) and to describe costs specifically related to hospitalizations, transfusions and MDS-specific drugs.
Materials and Methods
Study Design and Data Sources
A retrospective observational study was carried out using the ESND (“Echantillon National du SNDS”) database. The ESND is the 2/100ths representative sample of the SNDS.15 It contains individual-level data on all healthcare claims (outpatients and inpatients). The ESND provides near-exhaustive coverage of mortality data, automatically records medication exposure for reimbursement purposes, including expensive medications administered at hospital, and includes diagnostic codes for the causes of hospitalization, enabling comprehensive analysis of healthcare utilization and outcomes. It also reports the presence of some long-term diseases (LTD, “Affection de Longue Durée [ALD]”), that grant full reimbursement of related care. This database serves as a relevant resource for assessing healthcare costs and resource utilization patterns.15
Study Population and Study Period
As previously described in a previous published paper,16 all patients meeting the following criteria were included: incident adults with MDS and who received at least one dose of an ESA or lenalidomide. Incident treated LR-MDS patients were those identified with no prior administration of ESA or lenalidomide and no diagnosis of MDS in the 5-year period preceding the index date. Patients were included between January 1, 2018 and December 31, 2022 (inclusion period). The index date was defined as the first administration of ESAs or lenalidomide recorded in the inclusion period. LR-MDS patients were followed up from the index date to death, the end of the study period (December 31, 2022), loss to follow-up (ie, the absence of any healthcare consumption for two continuous years), whichever occurred first. Supplementary File 1 presents more in depth the algorithm of patients selection and codes used.
Then, as previously described,16 the cohort was categorized into subgroups at index date based on the IWG-2018 classification.17 Red blood cell (RBC) transfusions were identified using ICD-10 code (“Z51.30”) and common classification of medical acts (CCAM) codes (“FELF008” or “FELF011”). The IWG criteria also consider the number of units given in each transfusion. However, since this data is not available in the SNDS, only the number of transfusions was considered. These subgroups included the Transfusion Dependent (TD) patients, with at least one transfusion in the 16 weeks prior to the initiation of ESAs or lenalidomide. Patients were further divided into the High Transfusion Burden (HTB) subgroup (patients receiving ≥4 transfusions), and the Low Transfusion Burden (LTB) subgroup (patients receiving 1–3 transfusions). Finally, the Non-Transfusion Dependent (NTD) group included the patients with no transfusion within the 16 weeks preceding initiation of ESAs or lenalidomide.
Economic Endpoint
HCRU and related costs were described according to French National Health Insurance (NHI) perspective in euro 2022 (ie, amount reimbursed by the payer). For all patients included in this analysis, reimbursed total costs per patient were computed per year of follow-up and according to transfusion status (NTD versus TD). Specific costs of MDS were also computed according to transfusion status: hospitalization for MDS, transfusions and drugs for MDS (Erythropoietin, Darbepoietin alfa, Luspatercept, Lenalidomide, Azacitidine, GCSF).
Covariates
Various demographic and clinical variables were described in the entire study population and on each subgroup of interest (TD, NTD). Characteristics were compared according to transfusion status. The variables analyzed were age, sex, follow-up duration, death during follow-up, comorbidities (auto-immune diseases, chronic obstructive pulmonary disease (COPD), cardiovascular disease, chronic kidney disease, and diabetes). The presence of comorbidities was identified using relevant diagnostic codes from medical records using a 5‑year retrospective period. Red blood cell (RBC) transfusions were identified using ICD-10 code (“Z51.30”) and common classification of medical acts (CCAM) codes (“FELF008” or “FELF011”).
Statistical Methods
The statistical analyses were performed with SAS® software (version 9.4 or later), R software (version 3.6.1 or later), and Python (version 3.7). Descriptive statistics were computed. Continuous data were summarized by their mean, standard deviation (SD) or median, and first (Q1) and third (Q3) quartiles when relevant. Categorical data were summarized by percentage. The mean cost per patient was obtained by dividing the overall cost by the total number of patients, adjusted for the months of follow-up, and then annualized. In order to compare mean costs per year of follow-up according to transfusion status (TD, NTD), general linear models were performed. Different distributions (Poisson, overdispersed Poisson, and negative binomial distributions) were tested and the final model was selected based on the Akaike Information Criterion (AIC). Results were reported as estimated mean costs and relative risks with 95% confidence intervals. No adjustment was performed.
Results
Baseline Characteristics of the Study Population
After applying the selection criteria, 267 adults newly treated for an incident LR-MDS, affiliated to general regimen, were included. Among them, 26.6% (N=71) were TD corresponding mainly to LTB patients (72.3% of cases). The median follow-up duration for the overall cohort was 1.9 [IQR: 0.8; 2.9] years with 41.9% (N=112) of deaths over the study period (Figure 1).
Figure 1.
Flowchart of the study selection process (incident LR-MDS).
Abbreviations: MDS, Myelodysplastic syndromes; TD, transfusion‑dependent patients; NTD, non-transfusion‑dependent patients.
Patients included in this analysis had a median age at index date of 81.0 (IQR:74.0;87.0) years and 50.2% (N=134) of them were male (Table 1). They were also comorbid with a high proportion of cardiovascular diseases (46.4%, N=124), chronic kidney disease (22.5%, N=60), history of cancer (25.8%, N=69), and diabetes (24.7%, N=66). Characteristics of patients differed significantly according to transfusion dependence status (chi‑square test). TD patients tended to be more frequently men (56.3% versus 48.0%, p=0.2264) and affected by cardiovascular comorbidities (57.7% versus 42.3%, p=0.0258). Percentage of death during the follow-up was significantly higher in TD patients (54.9% versus 37.2%, p=0.0097).
Table 1.
Baseline and Follow-up Characteristics of Patients with General Regimen Affiliation (Overall and According to Transfusion Status)
| All (N=267) | TD (N= 71) | NTD (N= 196) | P-value (TD vs NTD) | |
|---|---|---|---|---|
| Age at inclusion in years, median [IQR] | 81.0 [74.0;87.0] | 80.0 [74.0;86.0] | 82.0[75.0;87.0] | 0.4761 |
| Male, No. (%) | 134 (50.2) | 40 (56.3) | 94 (48.0) | 0.2264 |
| Comorbidities, No. (%) | ||||
| Cardiovascular disease | 124 (46.4) | 41 (57.7) | 83 (42.3) | 0.0258 |
| Chronic kidney disease | 60 (22.5) | 18 (25.3) | 42 (21.4) | 0.4974 |
| Diabetes | 66 (24.7) | 17 (23.9) | 49 (25.0) | 0.8597 |
| History of cancer | 69 (25.8) | 26 (36.6) | 43 (21.9) | 0.0155 |
| COPD | 35 (13.1) | 12 (16.9) | 23 (11.7) | 0.2691 |
| Auto-immune diseases* | 34 (12.7) | N<11 | N<11 | - |
| Tobacco dependency** | N<11 | N<11 | N<11 | - |
| Death during FU, No. (%) | 112 (41.9) | 39 (54.9) | 73 (37.2) | 0.0097 |
| FU duration in years, mean (±SD) | 2.0 (1.3) | 1.7 (±1.3) | 2.1 (1.3) | 0.0307 |
| Presence of HCRU during FU (at least once), No. (%) | ||||
| GP’s consultation | 253 (94.8) | 66 (92.9) | 187 (95.4) | |
| Specialist’s consultation | 189 (70.8) | 47 (66.2) | 142 (72.4) | - |
| Hospitalizations | 242 (90.6) | 67 (94.4) | 175 (89.3) | |
| Hospitalizations for MDS | 154 (57.7) | 46 (64.8) | 108 (55.1) | |
| Drugs for MDS | 267 (100) | 71 (100) | 196 (100) | |
| Transfusions | 148 (55.4) | 55 (77.5) | 93 (47.4) | |
| Number of HCRU during FU, median (IQR)*** | ||||
| GP’s consultation | 13.0 (7.0;27.0) | 11.0 (5.0; 21.0) | 13.0 (7.0; 28.0) | |
| Specialist’s consultation | 5.0 (2.0; 10.0) | 4.0 (1.0; 8.0) | 6.0 (2.0; 11.0) | |
| Total hospitalizations | 6.5 (3.0; 21.0) | 11.0 (4.0; 29.0) | 5.0 (3.0; 16.0) | - |
| Hospitalizations for MDS | 1.0 (1.0; 4.0) | 2.0 (1.0; 4.0) | 1.0 (1.0; 3.5) | |
| Drugs dispensation for MDS | 31.0 (12.0;68.0) | 23.0 (8.0; 52.0) | 33.0 (13.0; 74.0) | |
| Transfusions | 4.0 (2.0; 11.0) | 5.0 (2.0; 14.0) | 3.0 (2.0; 7.0) |
Notes: Values reported in bold correspond to the overall study population (All, N=267)* Auto-immune diseases include: multiple sclerosis, Guillain–Barré syndrome, Crohn’s disease, hemorrhagic rectocolitis, ankylosing spondylitis, seropositive rheumatoid arthritis, other rheumatoid arthritides, juvenile polyarthritis, systemic lupus erythematosus, psoriasis, autoimmune thyroiditis, type 1 diabetes, celiac disease, systemic scleroderma, Horton’s disease, Sicca syndrome, primary biliary cirrhosis, Addison’s disease. ** Tobacco dependency was tracked with specific diagnosis codes and specific therapy. *** Among patients with at least one HCRU during FU.
Abbreviations: COPD: Chronic obstructive pulmonary disease; FU: Follow-up; GP: General practitioner; HCRU: Healthcare Resource Utilization; IQR: Interquartile Range MDS: Myelodysplastic syndrome; NTD: Non-transfusion-dependent, SD: Standard deviation; TD: Transfusion dependent.
Total Costs of LR-MDS Patients and Main Cost Drivers
HCRU consumption was high in patients with treated LR-MDS. Over the study period, 90.6% (N=242) of incident LR-MDS were hospitalized at least once with a median number of 6.5 hospitalizations per patient IQR [3.0–21.0]. This percentage was lower in the NTD population compared to the TD population (respectively, 89.3% and 94.4%). Approximately 95% (N=253) of patients had at least one generalist practitioner (GP) consultation and 71% (N=189) at least one specialist consultation. The percentage of NTD patients with at least one specialist consultation was slightly superior in comparison with TD patients (72.4% versus 66.2%). At least one transfusion was required for 55.4% of patients (N=148) over the study period with a median number of transfusion per patient of 4.0 (IQR:2.0;11.0). As expected, 77.5% (N=55) of TD patients had at least one transfusion after treatment initiation with a median number of 5.0 (2.0; 14.0) transfusions per patient. This percentage was significantly lower (47.4%; N=93) in the NTD group with a median number of 3.0 (2.0; 7.0) transfusions per patient.
The median total costs related to HCRU per patient were estimated at €41,698 (IQR: €19,093; €84,111; Figure 2). Main cost drivers were hospitalizations (€15,465 (IQR: €5413; €39,332)) and outpatient drugs (€9,177.8 (IQR: €4455; €21,587)). Median costs per patient related to transfusions were estimated at €662 (IQR: €0; €6,162). Median total costs per patient were increased in TD patients (+28%; €49,658 versus €38,585; Figure 2). These differences were due to higher costs related to hospitalizations (€24,122 (IQR: €7246; €44,100) versus €11,951 (IQR: €4723; €31,312). No differences were observed for costs related to outpatient drugs. Median costs per patient related to transfusions were also higher in TD patients (€3170 (IQR: €48; €12,367) versus €0.0 (IQR: €0.0; €3573)). Also, among patients with at least one transfusion, median costs per patient related to transfusions were €6624 (IQR: €2000; €15,338) in TD patients (N=55) versus €4020 (IQR: €1,337; €12,600) in NTD patients (N=93).
Figure 2.
Median total costs per patient over the follow-up period (total reimbursed costs among total patient). Overall (N) = 267, NTD (N)= 196, TD(N)=71.
Total Costs After Treatment Initiation and Their Evolution Over Time
When regarding reimbursed costs per patient after treatment initiation, median costs per patient per year of follow-up were stable with a slight decrease over time. Median costs were estimated at €26,693 (IQR: 13,706; 46,685) per patient the first year after treatment initiation (Y1), at €24,358 (IQR: 8830; 43,651) the second year (Y2), at €20,418 (IQR: 7256; 46,805) the third year (Y3) and at €21,152 (IQR: 5880; 59,427) the fourth year (Y4). This could be explained by the slight decrease of costs related to hospitalizations, outpatient drugs, and drugs for MDS within Y4. Specifically, the percentage of patients with at least one hospitalization was estimated at 80% in Y1 and at 64% in Y4 with a median number of hospitalizations per patient stable over time. Median costs per patient related to hospitalizations were estimated at €8848 (1262; 22,557) in Y1 and at €3390 (0; 17,820) in Y4. Median costs per patient related to specific drugs for MDS were estimated at €5759 (2215; 10,867) in Y1 and at €1543 (0; 8460) in Y4. Median costs related to total outpatient drugs were estimated at €6581 (3365; 13,691) in Y1 and at €5886 (1281; 14,624) in Y4. In the year before treatment initiation, median reimbursed costs were 3.1 times lower than in the year after treatment initiation estimated at €8568 (IQR: 3272; 19,851). This difference is mainly due to incremental reimbursed costs related to drugs and hospitalizations.
For TD patients, mean costs per patient were always significantly superior compared with NTD patients except in the year before treatment initiation (Table 2). The mean cost of care for TD patients was 32% higher (€45,427 vs €34,382 for NTD) in the Y1. This cost discrepancy escalated to +83% and +57% in Y2 and Y3, respectively.
Table 2.
Comparisons of Mean Costs per Patient Between Transfusion Dependent Patients and Non Transfusion Dependent Patients per year of Follow-up (GLM Models)
| Year | TD | NTD | RR [CI] (NTD vs TD) | P-value |
|---|---|---|---|---|
| Y-1 | ||||
| N | 71 | 196 | ||
| Mean cost per patient (€) | 17,099.4 | 13,707.4 | 0.80 [0.61;1.05] | NS (0.1142) |
| Y1 | ||||
| N | 71 | 196 | ||
| Mean cost per patient (€) | 45,426.9 | 34,382.5 | 0.76 [0.58;0.99] | 0.0467 |
| Y2 | ||||
| N | 41 | 141 | ||
| Mean cost per patient (€) | 57,668.0 | 31,541.6 | 0.53 [0.37;0.76] | 0.0006 |
| Y3 | ||||
| N | 25 | 96 | ||
| Mean cost per patient (€) | 44,023.6 | 27,968.0 | 0.59 [0.42;0.84] | 0.0035 |
| Y4 | ||||
| N | 12 | 49 | ||
| Mean cost per patient (€) | 41,786.6 | 33,700.5 | 0.83 [0.49;1.38] | NS (0.4681) |
Comparisons of Specific Costs Before and After Treatment According to Transfusion Dependency
Figure 3 details mean cost of incident treated LR-MDS per patient per year of follow-up after treatment initiation (specific costs) according to transfusion dependency.
Figure 3.
Comparisons of mean costs per item 1 year before and each year after treatment initiation between TD and NTD patients (GLM models) (a) costs related to hospitalizations (b) costs related to transfusions (c) costs related to specific drugs. Patients included (N): TD: N(Y−1)=71, N(Y1)=71,N(Y2)=41, N(Y3)=25, N(Y4)=12; NTD: N(Y−1)=196, N(Y1)=196, N(Y2)=141, N(Y3)=96, N(Y4)=49.
Higher costs were observed for hospitalizations (Figure 3a) in the TD group both in the year before treatment initiation (at the limit of significance), during Y1 (p-value=0.04), Y2 (p-value=0.0001) and Y3 (p-value=0.002). One year before inclusion, TD patients had a mean hospitalization cost (€9633) slightly higher than NTD patients (€6746). The mean cost of hospitalization was 54% higher for TD patients (€25,240 vs €16,387) within Y1, 157% higher (€38,765 vs €15,064) within Y2 and 105% higher within Y3 (€24,161 vs €11,765).
Higher costs were also observed for transfusions before and after treatment initiation between the two groups (Figure 3b). One year before inclusion, TD patients had a mean transfusion cost (€2409) higher than NTD patients (€183). Within Y1, the mean cost of transfusions was drastically higher for TD patients (€5498 vs €1878, p=0.0001). In the TD group, mean costs per patient related to transfusions seemed stable within 2 years after treatment initiation estimated at €5498 during Y1 and at €5369 during Y2. Within Y3, mean costs increased to €7151. In the NTD group, mean costs related to transfusions increased over time. Within Y1, Y2, Y3 and Y4, mean costs were estimated at €1878, €1877, €2783 and €7663 respectively. As expected, costs related to transfusions were significantly superior in the TD group compared to the NTD group.
Regarding costs related to specific MDS drugs, no differences were observed between the two groups (Figure 3c). In the TD group, mean costs were estimated at €7793, €5500, €4127 within Y1, Y2, Y3 respectively. In the NTD group, mean costs were estimated at €8148, €6197, €5906 within Y1, Y2, Y3 respectively.
Discussion
This nationwide study is the largest real-world study describing the economic burden of LR-MDS patients in France. Elevated costs and substantial burden for patients and the health care system were observed. Specifically, the median total costs related to HCRU per patient for the entire follow-up period were estimated at €41,698 (IQR: 19,093; 84,111) with a median cost per patient per year of follow-up ranging from €20,418 to €26,693, illustrating a slight decrease over time. Main cost drivers were hospitalizations and use of specific drugs. To the best of our knowledge, no recent European studies have assessed comparable costs. As expected, transfusions were associated with a significant incremental health care resource utilization and costs (around €11,000 per patient over the follow‑up period). Specifically, TD patients incurred more hospitalizations and transfusions, leading to significant elevated costs after first‑line treatment initiation. In this group, mean costs per patient related to transfusions seemed stable within 2 years after treatment but increased within 3 years after treatment initiation suggesting a significant impact of transfusion dependence over time. In a recent systematic review, transfusion dependence was depicted as a negative prognosis factor on clinical outcomes but also regarding health care utilization and costs. In this review, five studies were identified to assess costs related to transfusion in MDS patients of which 4 reported higher costs in TD patients compared to NTD patients.18 Specifically, transfusion dependence was associated with an increased total cost of $31,233 per patient each year in the US. Same results were observed in a nationwide cohort study conducted in Sweden between 2008 and 2017. In this study, costs ranged from $8805 ($6482–$11,625) in low-risk patients to $80,106 ($61,460–$95,792) in the very high-risk category.13 In our study, TD patients had a mean hospitalization and transfusions cost higher than NTD patients. Same trends were observed in a smaller retrospective study conducted in France in 12 French centers in 100 MDS patients. During the 6-month study period, 22% of the patients required inpatient hospitalization and 36% of them RBC transfusion for symptomatic anemia. In this study, the 6-month median transfusion costs were €16,188 per patient.14 Finally, our study highlights that costs related to transfusions in the NTD group increased over time from Y1 to Y4. Several factors may contribute to this observation, including disease progression over time, potential changes in transfusion requirements and in patient characteristics, or the possibility that the initiation of specific treatment earlier in the disease course delayed the onset of transfusion dependence in some patients. However, our analysis was not designed to investigate the underlying causes of these trends, and these findings should therefore be interpreted cautiously. The decreasing number of patients over follow-up, particularly beyond 3 years after treatment initiation, limits the statistical power of longitudinal analyses and warrants cautious interpretation of cost trends observed in later follow-up years. Moreover, the distribution of follow-up durations indicates that only 25% of patients are followed for more than 3.32 years. This should be considered in the light of survival estimates previously published, with overall survival declining from 70% to 48% in the NTD group and from 53% to 31% in the TD group between 2 years and 46 months. It should also be considered in the context of the rarity of LR-MDS and the use of a representative sample of the SNDS, which inherently restrict the number of patients contributing data at longer follow-up durations.
Indeed, our study was primarily designed to provide the first robust nationwide description of healthcare resource utilization and costs among treated LR-MDS patients in France using real-world medico-administrative data, thereby addressing an important evidence gap. In the context of evolving treatment options, these findings provide a valuable benchmark for future economic evaluations. However, the present data source and study design are less robust to assess the long-term impact of treatment, disease progression, and cost trajectories, highlighting the need for complementary studies using alternative data sources and designs. This is mainly related to the medico-administrative nature of our data offering multiple strengths but also limitations that must be acknowledged in the interpretation of our results.
First, the SNDS serves as a relevant resource for assessing healthcare costs and resource utilization patterns in France. Indeed, all reimbursed health care consumptions are prospectively and independently collected for each patient over several years at national level. Thus, our results offer a robust representation of LR-MDS economic burden after the initiation of specific treatment in France due to the exhaustive nature of the database. However, our results are only generalizable to treated LR-MDS patients because of the treatment-based cohort entry. This design choice may select patients with more symptomatic anaemia or higher healthcare engagement, thereby inflating estimates of healthcare utilization and costs relative to the broader LR-MDS population. However, our objective was specifically to characterize healthcare resource utilization and costs among treated LR-MDS patients. Furthermore, the lack of detailed clinical and laboratory data in the SNDS did not allow a robust identification and characterization of untreated LR-MDS patients. Specifically, the SNDS does not allow direct information on disease severity because no specific codes exists to identify LR-MDS. Therefore, algorithms relying primarily on treatment patterns were used to identify patients with LR-MDS, representing the most appropriate approach available with this data source. The algorithm to identify LR-MDS, outcomes, and variables of interest were developed with a scientific committee in order to get as close as possible to the clinical situation. Secondly, the number of red blood cell units transfused is not recorded in the SNDS. Thus, we assumed to assess transfusion dependence using a specific algorithm at baseline and to describe costs over time. In this context, misclassification of some patients cannot be excluded at baseline but also within follow-up with a potential impact on costs evolution in each subgroup of interest. Indeed, we were not able to study with accuracy transfusion dependence as a dynamic state rather than a fixed baseline characteristic. Moreover, we were unable to stratify our analyses by transfusion burden (low-transfusion burden versus high-transfusion burden) due to the number of patients in our cohort. Also, we were only able to capture costs corresponding to expenses covered by the healthcare system and costs corresponding to medications, devices, or services that are not reimbursed could not be included in our analysis. This may lead to an underestimation of costs related to this condition. Finally, the present analysis did not allow to assess costs according to the newer and high-impact therapies which emerged in the LR-MDS treatment landscape. Indeed, only 9 patients were exposed to newer agents such as luspatercept and 0 to emerging therapies such as imetelstat. Future studies will be needed to provide more granular treatment-specific cost as these new therapies are associated with substantially different acquisition costs, administration patterns, and downstream healthcare utilization compared with traditional ESAs.
Conclusions
LR-MDS treatment and management lead to significant economic burden for patients and health care system. Transfusion dependence is associated with incremental health care resource utilization and related costs in patients treated for LR-MDS in France. These results provide a robust nationwide estimate of the economic burden among treated LR-MDS patients and establish an important benchmark for future evaluations of emerging therapies. Future studies using complementary data sources are needed to further characterize the long-term clinical and economic impact of these therapies.
Acknowledgments
The authors gratefully acknowledge the contributions of Sacha Hiridjee, Mallory Cals-Maurette, and Arnaud Panes for data management, statistical analyses, and study documents redaction, respectively. All are employees of Heva (Lyon, France), the CRO which conducted the study on behalf of BMS. Writing and editorial assistance was provided by Cécile Conte, who is an employee of Heva (Lyon, France), the CRO which conducted the study on behalf of BMS.
Funding Statement
This study was funded by Bristol Myers Squibb.
Data Sharing Statement
The data supporting the study findings are part of the National health data system (SNDS, Système national des Données de Santé) and are available from the HDH (Health Data Hub https://www.health-data-hub.fr/). Restrictions apply to the availability of these data containing potentially identifying and sensitive patient information. Special permission to access these data for this study was granted by the Ethics and scientific committee for health research, studies, and evaluations (CESREES).
Ethics Approval
Approval of protocol and data access for this study was obtained from the Ethics and scientific committee for health research, studies, and evaluations (CESREES, Comité Ethique et Scientifique pour les Recherches, les Etudes et les Evaluations dans le domaine de la Santé) according to French regulation. Transparency and patient data protection are granted in this study with details available on the HDH website (Health Data Hub https://www.health-data-hub.fr/).
Author Contributions
All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising, or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.
Disclosure
Professor Maud D’Aveni reports support for the manuscript from BMS: writing, during the conduct of the study; Honoraria from Pfizer, Sanofi, Servier; meeting/travel support from Sanofi, Novartis and Servier, outside the submitted work. Dr Jérôme Fernandes reports support for the manuscript from BMS, during the conduct of the study; grants or contracts from Janssen, Abbvie, Astra Zeneca, Astellas Pharma, outside the submitted work. Mrs Mélanie Chartier reports stock or stock options from Bristol Myers Squibb and is an employee of Bristol Myers Squibb, outside the submitted work. Dr Nicolas Pagès reports support for the manuscript from Bristol Myers Squibb, during the conduct of the study and works for Heva, the company contracted by BMS to conduct the study. Mrs Aurélie Schmidt reports support for the manuscript from BMS, during the conduct of the study, and works for Heva, the company contracted by BMS to conduct the study. Mrs Léa Webert is an employee of Bristol Myers Squibb. Dr François-Emery Cotté reports support for the manuscript from Bristol Myers Squibb, during the conduct of the study, and is an employee of Bristol Myers Squibb. Dr Thibault Comont declared consulting fees from BMS for this work.
References
- 1.Brunner AM, Leitch HA, van de Loosdrecht AA, Bonadies N. Management of patients with lower-risk myelodysplastic syndromes. Blood Cancer J. 2022;12:166. doi: 10.1038/s41408-022-00765-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Fenaux P, Adès L. How we treat lower-risk myelodysplastic syndromes. Blood. 2013;121:4280–10. doi: 10.1182/blood-2013-02-453068 [DOI] [PubMed] [Google Scholar]
- 3.Fenaux P, Santini V, Spiriti MAA, et al. A Phase 3 randomized, placebo-controlled study assessing the efficacy and safety of epoetin-α in anemic patients with low-risk MDS. Leukemia. 2018;32:2648–2658. doi: 10.1038/s41375-018-0118-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Battaglia MR, Cannova J, Madero-Marroquin R, Patel AA. Treatment of anemia in lower-risk myelodysplastic syndrome. Curr Treat Opt Oncol. 2024;25:752–68. doi: 10.1007/s11864-024-01217-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Sekeres MA, Taylor J. Diagnosis and treatment of myelodysplastic syndromes: a review. JAMA. 2022;328:872–880. doi: 10.1001/jama.2022.14578 [DOI] [PubMed] [Google Scholar]
- 6.Fenaux P, Haase D, Santini V, Sanz GF, Platzbecker U, Mey U. Myelodysplastic syndromes: ESMO clinical practice guidelines for diagnosis, treatment and follow-up†☆. Ann Oncol. 2021;32:142–156. doi: 10.1016/j.annonc.2020.11.002 [DOI] [PubMed] [Google Scholar]
- 7.Della Porta MG, Garcia-Manero G, Santini V, et al. Luspatercept versus epoetin alfa in erythropoiesis-stimulating agent-naive, transfusion-dependent, lower-risk myelodysplastic syndromes (COMMANDS): primary analysis of a phase 3, open-label, randomised, controlled trial. Lancet Haematol. 2024;11:e646–58. doi: 10.1016/S2352-3026(24)00203-5 [DOI] [PubMed] [Google Scholar]
- 8.Platzbecker U, Della Porta MG, Santini V, et al. Efficacy and safety of luspatercept versus epoetin alfa in erythropoiesis-stimulating agent-naive, transfusion-dependent, lower-risk myelodysplastic syndromes (COMMANDS): interim analysis of a phase 3, open-label, randomised controlled trial. Lancet. 2023;402:373–385. doi: 10.1016/S0140-6736(23)00874-7 [DOI] [PubMed] [Google Scholar]
- 9.Cogle CR. Incidence and burden of the myelodysplastic syndromes. Curr Hematol Malig Rep. 2015;10:272–281. doi: 10.1007/s11899-015-0269-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.DeZern AE, Binder G, Rizvi S, et al. Patterns of treatment and costs associated with transfusion burden in patients with myelodysplastic syndromes. Leuk Lymphoma. 2017;58:2649–2656. doi: 10.1080/10428194.2017.1312372 [DOI] [PubMed] [Google Scholar]
- 11.Frytak JR, Henk HJ, De Castro CM, Halpern R, Nelson M. Estimation of economic costs associated with transfusion dependence in adults with MDS. Curr Med Res Opin. 2009;25:1941–1951. doi: 10.1185/03007990903076699 [DOI] [PubMed] [Google Scholar]
- 12.Goldberg SL, Chen E, Sasane M, Paley C, Guo A, Laouri M. Economic impact on US Medicare of a new diagnosis of myelodysplastic syndromes and the incremental costs associated with blood transfusion need. Transfusion. 2012;52:2131–2138. doi: 10.1111/j.1537-2995.2012.03626.x [DOI] [PubMed] [Google Scholar]
- 13.Zhao J, Dahlén T, Edgren G. Costs associated with transfusion therapy in patients with myelodysplastic syndromes in Sweden: a nationwide retrospective cohort study. Vox Sang. 2021;116:581–590. doi: 10.1111/vox.13034 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Jouzier C, Cherait A, Cony-Makhoul P, et al. Red blood cell transfusion burden in myelodysplastic syndromes (MDS) with ring Sideroblasts (RS): a retrospective multicenter study by the Groupe Francophone des Myélodysplasies (GFM). Transfusion. 2022;62:961–973. doi: 10.1111/trf.16884 [DOI] [PubMed] [Google Scholar]
- 15.de Germay S, Conte C, Micallef J, et al. Performing pharmacoepidemiological studies using the French health insurance data warehouse (SNDS): how to translate guidelines into practice. Therapie. 2023;78:679–689. doi: 10.1016/j.therap.2023.01.009 [DOI] [PubMed] [Google Scholar]
- 16.D’Aveni M, Fernandes J, Chartier M, et al. Incidence and burden of lower-risk myelodysplastic syndrome: a nationwide population study. Clin Lymphoma Myeloma Leuk. 2025;25(11):e933–e941.e2. doi: 10.1016/j.clml.2025.06.009 [DOI] [PubMed] [Google Scholar]
- 17.Platzbecker U, Fenaux P, Adès L, et al. Proposals for revised IWG 2018 hematological response criteria in patients with MDS included in clinical trials. Blood. 2018;133:1020. doi: 10.1182/blood-2018-06-857102 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Kaka S, Jahangirnia A, Beauregard N, Davis A, Tinmouth A, Chin-Yee N. Red blood cell transfusion in myelodysplastic syndromes: a systematic review. Transfus Med. 2022;32:3–23. doi: 10.1111/tme.12841 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data supporting the study findings are part of the National health data system (SNDS, Système national des Données de Santé) and are available from the HDH (Health Data Hub https://www.health-data-hub.fr/). Restrictions apply to the availability of these data containing potentially identifying and sensitive patient information. Special permission to access these data for this study was granted by the Ethics and scientific committee for health research, studies, and evaluations (CESREES).