This was an observational cohort drug utilisation study using retrospective data from the electronic health records of new users of radium-223 who were treated in routine clinical practice in the Netherlands, Germany and Denmark. The study period included time before and after the 2018 EMA label change (Fig. 1). The “before” period started on 13 November 2013, the date of radium-223 approval by the EMA, and ended on 30 November 2017, the date when the first Direct Healthcare Professional Communication letter was sent due to the ERA-223 trial observations. To provide the opportunity for patients to have at least 6 months of follow-up, enrolment for the “before” period ended on 31 May 2017. Enrolment in the “after” period started on 30 April 2019, 6 months after the label change, to allow sufficient time for treating physicians to familiarise themselves with the new prescribing limitations. To maximise enrolment, patients in the Netherlands could enrol through 30 April 2020, and patients in Germany and Denmark could enrol through 30 June 2020. To allow all patients the opportunity to have ≥ 6 months of follow-up, the period after the label change extended through 31 October 2020 for the Netherlands and 31 December 2020 for Germany and Denmark.

Study design. CAPRI 1 and 2 data were from 20 hospitals, CAPRI 3 data were from 6 hospitals, GePaRD data were from 4 statuary health insurance (SHI) providers (2 large providers and 2 small providers) and data from Denmark were from 2 large hospitals. For GePaRD, detailed data on the prior use of systemic metastatic castration-resistant prostate cancer (mCRPC) therapies were not available from all SHI providers; thus, analyses of the second and third objectives were conducted using data from only one large SHI provider (i.e., “restricted sample”). aIn the restricted sample, there were 245 patients in the before-label change period and 34 patients in the after-label change period. CAPRI Castration-Resistant Prostate Cancer Registry, F.u. follow-up, GePaRD German Pharmacoepidemiological Research Database

Data collection occurred between 2021 and 2023 (the Netherlands [May 2021–April 2022], Germany [September 2022–March 2023], Denmark [July 2022–May 2023]). For the main analyses, patients were followed from the date of the first dose of radium-223 (index date) until 6 months after the index date, until the end of the study period (6 months after the end of the defined period before or after the label change), until death or until disenrolment from the database, whichever occurred first. The primary objectives were to estimate the following among the population of patients receiving radium-223 during the period after the 2018 EMA label change: (1) the proportion who receive radium-223 in combination with abiraterone acetate; (2) the proportion who receive radium-223 in combination with other systemic therapies for mCRPC; and (3) the proportion who receive radium-223 without ≥ 2 prior lines of systemic therapy for mCRPC. The secondary objectives were to compare these proportions with those corresponding to the period before the label change and to characterise the population of new users of radium-223, irrespective of combination with other systemic therapies for mCRPC, at treatment start for both study periods.

This study was performed in accordance with the Declaration of Helsinki. In the Netherlands, the study protocol was approved by the Medical Research Ethics Committee Oost-Nederland (CMO 2020-6573); written informed consent was obtained by the treating physician from subjects who were alive, with the exception of patients who were lost to follow-up (due to lack of up-to-date contact details), and involved consent for publication. For Germany, the Ethics Committee of the University of Bremen stated that studies based on the data source (German Pharmacoepidemiological Research Database [GePaRD]) are exempt from institutional review board review. Still, all involved health insurance providers (AOK Bremen/Bremerhaven, DAK-Gesundheit, Techniker Krankenkasse [TK] and hkk Krankenkasse), their responsible authorities (the German Federal Office for Social Security [no. 116–8261-459/2019] and the Senator for Health, Women and Consumer Protection in Bremen [no. 500-006-110-81/2017-2-227]) approved the use of GePaRD data for this study. The responsible authorities approved the use of GePaRD data without informed consent, as informed consent was deemed to be unacceptable and to bias results (in alignment with §75 of the Code of Social Law X). In Denmark, the study received mandatory registration at Aarhus University (journal nr. KEA-2017-36/812) and medical chart abstraction was approved by treating hospitals (Aarhus University Hospital and Copenhagen Hospital) and the Scientific Ethics Committee in the Central Denmark Region [16] (journal nr. 1–45-70-41-20); the ethical committee waived the informed consent requirement.

The study population included new users of radium-223 who were enrolled in each participating data source during the periods before and after the label change. Inclusion and exclusion criteria are summarised in Electronic Supplementary Material (ESM) Table S-1. Briefly, eligible patients were men of any age who initiated treatment with radium-223 during one of the two study periods. A patient could qualify as a new user only once and therefore could be included in only one of the two study periods. Any patient with prior use of radium-223 was excluded from the study.

Analyses in the Netherlands were conducted using data from patients with confirmed mCRPC (as recorded in the Castration-Resistant Prostate Cancer Registry [CAPRI]). In Germany and Denmark, castration sensitivity versus castration resistance could not be assessed, so all patients receiving radium-223 were assumed to have mCRPC.

The study used existing data sources (secondary data collection) in the form of electronic medical records and medical record abstraction in the Netherlands (CAPRI 1–3), claims data from a population-based database in Germany (GePaRD) and a combination of routinely collected data from national registries and data from electronic medical record abstraction in Denmark (Danish National Patient Registries [DNPR] and other data sources). The study was conducted following a common statistical analysis plan, and data from each country were managed and analysed locally by investigators at CAPRI in the Netherlands, the Leibniz Institute for Prevention Research and Epidemiology (BIPS) in Germany, and Aarhus University in Denmark, with support from epidemiologists and biostatisticians from RTI Health Solutions (RTI-HS). The size of the study was driven by the uptake of radium-223 in the study population, which included all new users of radium-223 who were captured in the participating data sources and who met the inclusion and exclusion criteria.

In the Netherlands, CAPRI 1 and 2 (comprising data from 20 hospitals) were completed between 2010 and 2017 and included patients diagnosed with CRPC between 1 January 2010 and 31 December 2015. CAPRI 3 (comprising data from 6 hospitals, 5 of which also contributed patients to CAPRI 1 and 2) was started in 2021 and included patients diagnosed with mCRPC after 1 January 2016. Thus, data from the before-label change period were retrieved from CAPRI 1 and 2, and data from the after–label change period were retrieved from CAPRI 3. In Germany, GePaRD included data from four statutory health insurance (SHI) providers (2 large providers and 2 small providers). Detailed data on the prior use of systemic therapies for mCRPC were available from only one large SHI provider. Thus, analyses of the second and third primary objectives (as described above) were conducted using data from only one SHI provider in Germany (i.e. restricted sample). Danish sources included data from two treating hospitals. All patients in the after-label change period were included, while only a random sample of patients in the before-label change period was selected.

For both study periods, the main exposure of interest was new use of radium-223, which defined the study population. It is important to note that each radium-223 dose should be administered intravenously every 4 weeks, for a total of up to six injections, with treatment completed within approximately 6 months. Exposure definitions used for each country are defined in ESM Table S-2.

For the primary outcomes, use of radium-223 in combination with abiraterone acetate was defined as a patient having ≥ 1 administration of abiraterone acetate on the same date, within 30 days after or within 5 days before an administration of radium-223. Therefore, an at-risk period of 36 days was defined by each administration of radium-223. Combined use of radium-223 with other systemic therapies for mCRPC, other than LHRH analogues, was defined as a patient having ≥ 1 administration of systemic therapy for mCRPC, other than abiraterone acetate, on the same date or within 30 days after an administration of radium-223. Lastly, use of radium-223 without having received ≥ 2 prior lines of systemic therapy for mCRPC was defined as new users of radium-223 who received this drug as the first or second line of therapy for mCRPC. Outcome definitions are further detailed in ESM Table S-3.

All sensitivity analyses were restricted to the after-label change period (i.e. the main period of interest). In German data sources, to account for uncertainty surrounding the actual date of abiraterone acetate administration, use of radium-223 in combination with abiraterone acetate was assessed with 5 additional days of lag time after the last dispensing of abiraterone acetate. To account for the possibility that a 30-day dispensing of abiraterone acetate would be used after radium-223 was started, a 30-day lag time was added after dispensing of abiraterone acetate. For data from Germany (restricted sample) and Denmark, where castration-resistant status could not be ascertained, a “worst-case” sensitivity analysis was performed that considered all docetaxel use as a line of treatment for castration-sensitive prostate cancer (CSPC); in this sensitivity analysis, docetaxel use therefore did not count towards the prior lines of systemic therapy for CRPC. Likewise, a sensitivity analysis was conducted using data from Germany and Denmark to assess the proportion of patients who had ≥ 1 cycle of radium-223 in combination with systemic corticosteroids; this analysis could not be conducted with data from the Netherlands since corticosteroid prescription end dates were not available. Lastly, for all countries, to account for cycles dispensed or administered beyond the sixth month after the index date (i.e. the follow-up period of the main analysis), a sensitivity analysis was conducted that extended follow-up until 1 month after the date of the last administration of radium-223; the last administration was defined as an administration of radium-223, up to the sixth dose, that was not followed by a subsequent administration within 2 months.

Analyses of data from the Netherlands were performed using IBM SPSS Statistics 25 (IBM Corp., Armonk, NY, USA), and analyses of data from Germany and Denmark were performed using SAS statistical software (SAS Institute Inc., Cary, NC, USA). In Denmark, when the number of individuals in a cell ranges from 1 to 4 or when back-calculation of cells within or across tables could be performed but would lead to a result of 1 to 4, the exact numbers could not be reported due to local privacy protection policies. Therefore, in such situations, results were reported as ranges. Missing values were described as separate categories for each applicable variable, and data were summarised using descriptive statistics. All analyses adhered to the Guidelines for Good Pharmacoepidemiology Practices of the International Society for Pharmacoepidemiology [17] and the Guide on Methodological Standards in Pharmacoepidemiology from the European Network of Centres for Pharmacoepidemiology and Pharmacovigilance [18].