Increased intervals in enzyme replacement therapy for stable type 1 Gaucher disease: A non-inferiority sequential trial emulation.

Abbreviations

chemokine ligand 18

French Evaluation of GD Treatment Committee

enzyme replacement therapy

French Gaucher Disease Registry

Gaucher disease

glucosylsphingosine

every 2 weeks

every 3–4 weeks

Introduction

Gaucher disease (GD) is an autosomal recessive lysosomal disorder caused by mutations in the GBA1 gene leading to a deficiency of glucocerebrosidase [ 1 ]. GD is a rare condition, with an incidence varying between 0.4 and 5.8 per 100,000 inhabitants [ 2 ], and is characterized by various manifestations such as hepatomegaly, splenomegaly, bone involvement (including bone infarcts, avascular osteonecrosis, osteoporosis, or pathological fractures), anemia, and/or thrombocytopenia. Traditionally, the disease has been classified into three primary forms; however, some authors argue it represents a continuum rather than three distinct entities [ 3 ]. Type 1 GD is the most frequent phenotype and is classically described as “non‐neuronopathic” GD, although increasing evidence suggests that patients with type 1 GD may also develop neurological disorders such as Parkinson's disease or Lewy body dementia [ 4 ].

The prognosis of GD patients has substantially improved since the development of enzyme‐replacement therapy (ERT) since the early 1990s, including alglucerase [ 5 ], followed by imiglucerase [ 6 , 7 , 8 , 9 ], velaglucerase alpha [ 10 ], and taliglucerase, which was authorized in France solely for temporary use in 2010–2011. These treatments are efficient to reduce organomegaly and cytopenia, as well as improving bone‐related pain and bone mineral density. Subsequently, oral substrate reduction therapies including miglustat and eliglustat have been available for type 1 GD since 2015 [ 11 , 12 , 13 , 14 ].

In France, imiglucerase and velaglucerase alpha are the most used ERT. These two products have equivalent efficacy and are administered at the same dose of 45–60 U/kg every 2 weeks. Despite an improvement in disease‐related quality of life, intravenous ERT remains burdensome as it requires biweekly infusions, administered either at home or in a hospital setting. In addition, not all patients are eligible to receive oral eliglustat.

Once the disease is stabilized, there is limited data on the efficacy of lower dose regimens or the possibility of increasing the interval between infusions. Clinical practices differ significantly between regions, with some countries, notably Israel, tending to use lower doses than in France, while maintaining a biweekly infusion interval [ 15 ]. We argue that reducing the dose without decreasing infusion frequency does not lessen treatment burden. A single randomized trial published in 2009, compared the efficacy and safety of imiglucerase infusions every 2 weeks (Q2W) versus every 4 weeks (Q4W) in 95 patients with stable type 1 GD [ 16 ]. The trial was negative but it was designed as a superiority rather than a non‐inferiority trial. The primary outcome was assessed relatively early, at 1 year, and patients were required to have been stable for only 6 months prior to enrollment. Consequently, data in patients who have been stable for a longer duration are lacking.

The objective of this study was to evaluate the efficacy and non‐inferiority of dose spacing to every 3–4 weeks (Q3–4W) compared to the standard frequency of every 2 weeks in patients with stable type 1 GD for at least 2 years, using real‐world data from a national registry and a target trial emulation framework.

Patients and methods

The French Gaucher disease registry

The French Gaucher Disease registry (FGDR) is an academic registry of GD patients, created in 2009 by the Referral Center for Lysosomal Diseases and the French Evaluation of GD Treatment Committee (EGDTC). The FGDR aims to collect data for academic studies to improve GD patient clinical care and knowledge.

The registry includes consenting patients with GD managed in France since 1980. All patients were diagnosed based on demonstration of deficient glucocerebrosidase activity in blood. To ensure exhaustive case identification, patients are enrolled in the registry through three different sources: (1) accredited diagnosis laboratories (included in the EGDTC) reporting patients with confirmed enzyme deficiency; (2) data from the French national health insurance informing by Rare Disease Committee members to validate GD diagnosis to ensure health coverage; (3) as part of GD treatment requests during multidisciplinary consultation meeting with the EGDTC members. Treating physicians obtained patient consent for inclusion in the FGDR. Patients who did not consent were not included in the registry ( N = 3).

The FGDR records clinical, biological, and radiological findings at GD diagnosis and during follow‐up, including intercurrent events such as anemia, bone complications, or malignancies. Data collection was retrospective before 2010 and prospective from 2011 onward.

Design

This retrospective study was conducted within the FDGR. We applied the target trial emulation framework using a sequential trial design, a method increasingly recognized for its ability to address selection and immortal‐time bias related to time‐varying treatments [ 17 ]. Briefly, after definition of eligibility criteria, patients were screened for eligibility every trimester throughout the study period. When eligibility criteria were met, patients assigned to the intervention arm were matched to 10 control patients at the corresponding trial date, with follow‐up starting on that date for both arms. Patients in the control arm remained eligible at subsequent timepoints, thereby preventing immortal‐time bias. In contrast, once assigned to the intervention arm, patients were no longer eligible for inclusion in future trials. Full details of the target trial design and its emulation are provided in Table 1 .

Eligibility

Patients with type 1 GD enrolled in the FGDR were eligible if they had received intravenous ERT with imiglucerase or velaglucerase alpha administered Q2W for a minimum of 2 years between January 1990 and March 2025. Other eligibility criteria were evidence of stable disease for at least two consecutive years, defined as the absence of anemia, thrombocytopenia, and bone events (bone infarcts, pathological fractures, or osteonecrosis), as well as the absence of treatment interruption or dose spacing. Given their distinct prognoses and complications, patients with types 2 or 3 GD were not included in the study.

Intervention and follow‐up

The intervention was defined as extending the dosing interval of ERT from Q2W to Q3–4W after a minimum of 2 years without clinical events. Patients without available data on ERT were excluded. As described above, for each trial, follow‐up began on the date of trial and observations were censored at the end of the study period, death, loss to follow‐up, or after a maximum of 10 years, whichever occurred first.

Data collected

The following variables were collected: sociodemographic data, age (at first symptoms and at GD diagnosis), vital status (as of the date of extraction), genotype, clinical symptoms at diagnosis including hepatomegaly, splenomegaly, and bone events, follow‐up in an expert center or not. Data during clinical follow‐up, including splenectomy and bone events, were included, as were biological parameters such as hemoglobin levels (g/dL), platelet counts (G/L), and biomarkers such as ferritin (mg/L), chemokine ligand 18 (CCL18, ng/mL), chitotriosidase (nmol/h/mL), and glucosylsphingosine (lyso‐Gb1) (nmol/L).

Outcomes

The primary endpoint was the occurrence of any of the following events: anemia (hemoglobin <10 g/dL), thrombocytopenia (platelet count <100 Giga/L), or clinically relevant bone events (pathological fracture, clinically significant bone infarct, or avascular osteonecrosis). Bone infiltrates detected on imaging or asymptomatic bone infarcts identified solely by MRI were not considered part of the primary endpoint. Because liver and spleen measurements were not standardized across centers, radiologists, and imaging modalities, these variables were not included in the primary outcome.

Secondary endpoints included longitudinal changes in biological parameters, specifically hemoglobin and platelet levels, and evolution of biomarkers such as ferritin, chitotriosidase, lyso‐Gb1, and CCL18 among patients who spaced their dose to Q3–4W.

An additional secondary analysis assessed economic cost savings and the number of infusions avoided.

Statistical analysis

Descriptive data were presented as numbers (percentages) for categorical data and as medians (Q1Q3) for quantitative variables.

From January 1990 to March 2025, eligibility was assessed every 3 months for all patients. Each patient who spaced their treatment to Q3–4W was matched at each trial to 10 contemporaneously eligible patients who continued Q2W at that same date, based on age at diagnosis, age at inclusion, sex, low dose ERT (<30 U/kg), follow‐up in a referral center for GD, history of bone events, anemia, splenomegaly, hepatomegaly, and splenectomy. After matching, all absolute standardized mean differences were below 0.1 (Fig. S1 ). Patients once included in the Q3–4W group were no longer eligible, whereas patients in the Q2W arm were still eligible in future trials. In the matched cohort, we estimated a risk difference in event‐free survival derived from Kaplan–Meier estimators [ 18 ] at each year during follow‐up. A non‐inferiority margin of 10% was prespecified. We also estimated the hazard ratio (HR) for the primary outcome using a Cox model. Confidence intervals (CI) were obtained using the bootstrap method with 1000 iterations. In primary analysis we did not censor patients if they modified the rhythm of their ERT after inclusion (pseudo‐ITT). In a secondary analysis we censored patients if they stopped or modified the rhythm of their ERT (per‐protocol).

For the secondary objective, we analyzed changes in biological parameters and biomarkers among patients in the Q3–4W arm compared to the period prior to dose spacing, using linear mixed‐effects models with a random intercept for each patient, restricted to those with at least 2 measures. The date of dose spacing (T0) was used as the reference time point.

For the economic analysis, we estimated the number of avoided infusions and dose per infusion in patients in the Q3–4W arm. We then estimated the cost savings based on the total costs if each patient stayed in the Q2W arm.

In a sensitivity analysis, we extended the minimum required period of clinical stability on biweekly ERT without clinical events from 2 to 4 years.

Ethics

The FGDR was approved by the French Data‐Protection Commission and certified by the French Institute for Public Health Surveillance and the National Institute of Health and Medical Research (INSERM). All patients gave written informed consent for use of their data. This study was performed in accordance with the ethical standards of the Helsinki Declaration.

Results

Study population

Out of 534 patients included in the registry, 8 patients had unavailable treatment information, and 280 met the inclusion criteria, of whom 63 switched to Q3–4W during follow‐up: 57 every 3 weeks and 6 every 4 weeks. The characteristics of included patients according to ERT frequency are presented in Table 2 . Patients’ characteristics were overall similar between the two groups. There was no significant difference between sex, age at first symptom, clinical characteristics, or splenectomy rate, which was 8.8% in the Q2W and 11% in the Q3–4W arm. The median age at GD diagnosis was 21 years, with a trend towards an earlier diagnosis in subjects who spaced to Q3–4W ( p = 0.055).

Primary endpoint

After matching, over an average follow‐up of 6.3 years, there were 57 events: 46 in the Q2W arm and 11 events among patients in the Q3–4W arm. In total, we observed 51 bone events, 5 cases of anemia, and 1 cases of thrombocytopenia. In the pseudo‐ITT analysis, the HR was 0.99 (95% CI 0.54–1.51). The upper bound of the CI for risk difference calculated each year was below the non‐inferiority margin at all times (Fig. 1 ). The HR in the per‐protocol analysis was 0.98 (95% CI 0.53–1.52), and upper bound of CI at each year was below non‐inferiority margin (Fig. S2 ). Results were consistent when extending the required clinical stability period to 4 years (Fig. S3 ).

Biological evaluation

For the 63 patients who spaced to Q3–4W, the evolution of biological parameters and biomarkers before and after dose spacing is shown in Fig. 2 . We observed a small increase in hemoglobin levels after spacing compared with before, with a mean difference of 0.39 g/dL (95% CI, 0.25–0.54). Platelet counts also moderately increased, with a mean difference of 21.5 G/L (95% CI, 14.4–28.7). In contrast, ferritin decreased by a mean of 92.6 µg/L (95% CI −133 to −52), although levels during the Q2W period remained elevated at 406 µg/L (±373). A monoclonal component was detected in only two patients, one in each group. No patients developed multiple myeloma. For other biomarkers, although data were available for only a limited number of patients before and after spacing, no significant change in levels of lyso‐Gb1 ( β = +8 nmol/L; 95% CI −16 to 33; n = 11 patients), chitotriosidase ( β = −440 nmol/h/mL; 95% CI −1327 to 448; n = 45), and of CCL18 ( β = −31 ng/mL; 95% CI −14 to 75; n = 14) were observed (Fig. 2 ).

Cost‐benefit evaluation

Among the 63 patients in the Q3–4W arm, over a follow‐up duration of 6.3 years, a total of 55.2 infusion sessions per patient were saved (Table 3 ). The corresponding cost savings were estimated at €445,105 per patient. Overall, more than 3400 infusions were avoided across the cohort, representing approximately €28 million in savings over the follow‐up period.

Discussion

This study evaluated whether extending the interval of ERT from every 2 weeks to every 3–4 weeks in patients with type 1 GD who had been stable for at least 2 years was non‐inferior regarding risk of bone events, anemia, and thrombocytopenia. The composite primary endpoint demonstrated non‐inferiority, as did secondary exploratory analyses assessing biomarker stability. Moreover, ERT dose spacing resulted in an estimated cost saving of nearly €500,000 per patient over an average follow‐up of 6 years, while also substantially reducing treatment burden.

Data on ERT dose spacing in type 1 GD remain limited. In the international GOS registry, Deegan reported in 2018 that only a very small proportion of patients received ERT every 3 or 4 weeks [ 15 ]. This study combined heterogeneous practices, including frequent use of reduced doses (<40 U/kg and sometimes <20 U/kg), that have not been formally evaluated in clinical trials. In contrast, ERT dosing practices observed in France are similar to those reported in the United States, where most patients receive 40–60 U/kg, and differ from those in Israel, where lower doses are commonly used. Although dose spacing and dose reduction pursue a similar objective of lowering overall treatment intensity and cost, the vast majority of patients outside of France are treated every 2 weeks.

In the randomized trial comparing spacing to Q4W with continuing on Q2W dosing, 63% of patients who spaced their doses met the composite efficacy endpoint, compared with 81% in the Q2W group (difference −17.6%; 95% CI, −35.7 to 5.8) [ 16 ]. This trial included patients treated Q2W for only 6 months and clinically stable for 1 year, whereas we imposed at least 2 years of stability. The composite endpoint was stringent and largely driven by hematological fluctuations, whereas bone events were rare. A secondary analysis based on therapeutic goals showed no significant difference between groups (89% vs. 100%; difference −11.5%; 95% CI, −23.1 to 6). Although the authors suggested that dose spacing might be feasible in selected stable patients, the study was underpowered for this question and designed as a superiority trial, precluding conclusions on non‐inferiority of Q4W dosing.

In contrast, we used a more clinically relevant endpoint of absence of anemia, thrombocytopenia, and new clinical bone events. The majority of observed events in our study were bone events, likely due to the extended duration of follow‐up. Of note, most patients spaced dosing to every 3 weeks rather than every 4 weeks. The 10% non‐inferiority margin, although partly empirical, was conservative compared with margins used in prior GD trials [ 12 ]. Our endpoint was limited to major clinical events and did not include subclinical skeletal markers, which may have favored non‐inferiority and should be considered when interpreting the results.

We acknowledge some limitations to our study. First, most patients switched ERT to every 3 weeks. Consequently, we were unable to perform separate subgroup analyses for Q3W and Q4W regimens. Second, liver and spleen volumes were not included in the primary outcome. In stable patients, these measures may be of limited relevance because of measurement variability and the persistence of asymptomatic organomegaly without clinical consequences. Third, as this was a retrospective observational study, residual confounding and indication bias cannot be excludes. Fourth, longitudinal biomarker data were not available for all patients, which limits their interpretation. Finally, the impact of interval spacing on health‐related quality of life (HRQoL) was not assessed because standardized HRQoL questionnaires were not collected in the registry. Nevertheless, the reduction in infusion frequency likely translates into improved HRQoL by decreasing treatment burden and minimizing disruptions to their daily lives.

Nevertheless, this study has several strengths. We applied a target trial emulation framework using sequential trials to address time‐varying exposure, as eligible patients could switch between treatment strategies at multiple time points. This approach explicitly defines the target trial and ensures proper alignment of eligibility criteria, treatment assignment, and start of follow‐up. In this setting, it is more robust than time‐varying Cox models or marginal structural models for addressing the specific methodological challenges of our study design. We additionally used propensity score matching to account for multiple potential confounders, including center, splenectomy, organomegaly, bone events, and dose of ERT. Finally, the substantial sample size and the extended follow‐up time of 6.3 years in average enabled the assessment of long‐term outcomes, particularly bone events, which are difficult to capture in randomized trials.

Conclusion

Using a target trial emulation framework, we showed that in patients with type 1 GD who were clinically stable on biweekly ERT for at least 2 years, extending the dosing interval to every 3–4 weeks was non‐inferior to continued biweekly administration with respect to major clinical outcomes. Dose spacing substantially reduced treatment burden and was associated with considerable cost savings over long‐term follow‐up. These findings support the feasibility and safety of individualized ERT dose‐spacing strategies in selected patients with GD1.

Author contributions

Study conception and design: Maxime Beydon, Yann Nguyen Acquisition of data: All authors. Statistical analysis: Maxime Beydon, Yann Nguyen. Analysis and interpretation of data: All authors. Drafting the manuscript: Maxime Beydon, Yann Nguyen. Revising the manuscript: All authors.

Conflict of interest statement

Yann Nguyen, Maxime Beydon, Karima Yousfi, Dalil Hamroun, Samira Zebiche, Thierry Billette de Villemeur, Laure Swiader, Anne‐Sophie Guemann, Vanessa Leguy‐Seguin, Yves‐Marie Pers, Soumeya Bekri, Catherine Caillaud, Samia Pichard, Marie Szymanowski, Leonardo Astudillo, Bénédicte Hivert, Marc G. Berger, and Jérôme Stirnemann declare that they have no conflict of interest. Bénédicte Héron received travel fees from Sanofi. Anaïs Brassier received travel fees from Sanofi and Takeda, and consulting fees or other remuneration including fee as speaker from Sanofi and Takeda. Florence Dalbies received consulting fees or other remuneration including fee as speaker from Sanofi. Bérengère Cador received travel fees from Sanofi and Takeda. Francis Gaches received travel fees from Sanofi. Agathe Masseau received fee as speaker from Sanofi and Takeda. Magali Pettazzoni received travel fees from Sanofi, and consulting fees or other remuneration including fee as speaker from Sanofi and Takeda. Wladimir Mauhin received fees for consultancy speaking, travel grants, meetings from Sanofi, Takeda, and Amicus therapeutics. Yann Nadjar received research grants by Takeda and received consulting fee or other remuneration including fee as speaker from Sanofi; assisting in the design of and/or participating in clinical studies using products manufactured by the Sanofi. Christine Serratrice received consulting fee or other remuneration including fee as speaker from Takeda and Sanofi. Fabrice Camou received fees for consultancy speaking, travel grants, and meetings from Sanofi and Takeda. Nadia Belmatoug received fees for consultancy speaking, travel grants, and meetings from Sanofi and Takeda.

Funding information

The FDGR received a research grant from Sanofi and Takeda.

Disclosure

The present article reflects only the authors’ view and Sanofi and Takeda had no link with the design of the study and had no access to the data.

Ethics statement

The FGDR was approved by the French Data‐Protection Commission and certified by the French Institute for Public Health Surveillance and the National Institute of Health and Medical Research (INSERM). This study was performed in accordance with the ethical standards of the Helsinki Declaration and was approved by the local Institutional Review Board.

Patient consent statement

All patients gave written informed consent for use of their data.

Supporting information

Figure S1 : Standardized mean differences before and after matching.

Figure S2 : Primary outcome in per‐protocol analysis.

Figure S3 : Primary outcome in intention‐to‐treat analysis, extending stability period from 2 to 4 years (Q2W n = 166, Q3‐4W n = 46).

Acknowledgments

The authors are indebted to all patients included to the FGDR. They acknowledge all author members of the Evaluation of Gaucher Disease treatment committee.

Open access publication funding provided by COUPERIN CY26.

Data availability statement

Data are available upon reasonable request after approvement of the scientific committee.

References

1. Brady RO, Kanfer JN, Shapiro D. Metabolism of glucocerebrosides. II. Evidence of an enzymatic deficiency in Gaucher's disease. Biochem Biophys Res Commun. 1965;18:221–225.

2. Nalysnyk L, Rotella P, Simeone JC, Hamed A, Weinreb N. Gaucher disease epidemiology and natural history: a comprehensive review of the literature. Hematol Amst Neth. 2017;22(2):65–73. 10.1080/10245332.2016.1240391.

3. Sidransky E. Gaucher disease: complexity in a “simple” disorder. Mol Genet Metab. 2004;83(1):6–15. 10.1016/j.ymgme.2004.08.015.

4. Rosenbloom B, Balwani M, Bronstein JM, Kolodny E, Sathe S, Gwosdow AR, et al. The incidence of Parkinsonism in patients with type 1 Gaucher disease: data from the ICGG Gaucher registry. Blood Cells Mol Dis. 2011;46(1):95–102. 10.1016/j.bcmd.2010.10.006.

5. Barton NW, Brady RO, Dambrosia JM, Di Bisceglie AM, Doppelt SH, Hill SC, et al. Replacement therapy for inherited enzyme deficiency‐macrophage‐targeted glucocerebrosidase for Gaucher's disease. N Engl J Med. 1991;324(21):1464–1470. 10.1056/NEJM199105233242104.

6. Wenstrup RJ, Kacena KA, Kaplan P, Pastores GM, Prakash‐Cheng A, Zimran A, et al. Effect of enzyme replacement therapy with imiglucerase on BMD in type 1 Gaucher disease. J Bone Miner Res. 22(1):119–126. 10.1359/jbmr.061004.

7. Grabowski GA, Barton NW, Pastores G, Dambrosia JM, Banerjee TK, McKee MA, et al. Enzyme therapy in type 1 Gaucher disease: comparative efficacy of mannose‐terminated glucocerebrosidase from natural and recombinant sources. Ann Intern Med. 1995;122(1):33–39.

8. Weinreb NJ, Charrow J, Andersson HC, Kaplan P, Kolodny EH, Mistry P, et al. Effectiveness of enzyme replacement therapy in 1028 patients with type 1 Gaucher disease after 2 to 5 years of treatment: a report from the Gaucher registry. Am J Med. 2002;113(2):112–119.

9. Weinreb NJ, Goldblatt J, Villalobos J, Charrow J, Cole JA, Kerstenetzky M, et al. Long‐term clinical outcomes in type 1 Gaucher disease following 10 years of imiglucerase treatment. J Inherit Metab Dis. 2013;36(3):543–553. 10.1007/s10545-012-9528-4.

10. Gonzalez DE, Turkia HB, Lukina EA, Kisinovsky I, Dridi M‐FB, Elstein D, et al. Enzyme replacement therapy with velaglucerase alfa in Gaucher disease: results from a randomized, double‐blind, multinational, phase 3 study. Am J Hematol. 2013;88(3):166–171. 10.1002/ajh.23381.

11. Lukina E, Watman N, Dragosky M, Pastores GM, Arreguin EA, Rosenbaum H, et al. Eliglustat, an investigational oral therapy for Gaucher disease type 1: phase 2 trial results after 4 years of treatment. Blood Cells Mol Dis. 2014;53(4):274–276. 10.1016/j.bcmd.2014.04.002.

12. Cox TM, Drelichman G, Cravo R, Balwani M, Burrow TA, Martins AM, et al. Eliglustat compared with imiglucerase in patients with Gaucher's disease type 1 stabilised on enzyme replacement therapy: a phase 3, randomised, open‐label, non‐inferiority trial. Lancet Lond Engl. 2015;385(9985):2355–2362. 10.1016/S0140-6736(14)61841-9.

13. Mistry PK, Lukina E, Ben Turkia H, Amato D, Baris H, Dasouki M, et al. Effect of oral eliglustat on splenomegaly in patients with gaucher disease type 1: the ENGAGE randomized clinical trial. JAMA. 2015;313(7):695–706. 10.1001/jama.2015.459.

14. Belmatoug N, Di Rocco M, Fraga C, Giraldo P, Hughes D, Lukina E, et al. Management and monitoring recommendations for the use of eliglustat in adults with type 1 Gaucher disease in Europe. Eur J Intern Med. 2017;37:25–32. 10.1016/j.ejim.2016.07.011.

15. Deegan P, Fernandez‐Sasso D, Giraldo P, Lau H, Panahloo Z, Zimran A. Treatment patterns from 647 patients with Gaucher disease: an analysis from the Gaucher outcome survey. Blood Cells Mol Dis. 2018;68:218–225. 10.1016/j.bcmd.2016.10.014.

16. Kishnani PS, DiRocco M, Kaplan P, Mehta A, Pastores GM, Smith SE, et al. A randomized trial comparing the efficacy and safety of imiglucerase (Cerezyme) infusions every 4 weeks versus every 2 weeks in the maintenance therapy of adult patients with Gaucher disease type 1. Mol Genet Metab. 2009;96(4):164–170. 10.1016/j.ymgme.2008.12.015.

17. Hernán MA, Robins JM. Using big data to Emulate a target trial when a randomized trial is not available. Am J Epidemiol. 2016;183(8):758–764. 10.1093/aje/kwv254.

18. Maringe C, Benitez Majano S, Exarchakou A, Smith M, Rachet B, Belot A, et al. Reflection on modern methods: trial emulation in the presence of immortal‐time bias. Assessing the benefit of major surgery for elderly lung cancer patients using observational data. Int J Epidemiol. 2020;49(5):1719–1729. 10.1093/ije/dyaa057.

Associated Data

Supplementary Materials

Figure S1 : Standardized mean differences before and after matching.

Figure S2 : Primary outcome in per‐protocol analysis.

Figure S3 : Primary outcome in intention‐to‐treat analysis, extending stability period from 2 to 4 years (Q2W n = 166, Q3‐4W n = 46).

Data Availability Statement

Data are available upon reasonable request after approvement of the scientific committee.