Cancer Epidemiology 78 (2022) 102145 Contents lists available at ScienceDirect Cancer Epidemiology journal homepage: www.elsevier.com/locate/canep Neuroblastoma in Spain: Linking the national clinical database and epidemiological registries – A study by the Joint Action on Rare Cancers☆ Adela Cañete a, b, *, Rafael Peris-Bonet a, Riccardo Capocaccia c, Elena Pardo-Romaguera a, Vanessa Segura d, Ana Muñoz-López a, Ana Fernández-Teijeiro e, f, Jaume Galceran-Padros g, Gemma Gatta h, Spanish Neuroblastoma Linkage Working Group (SpNbLinkWG)1 a Spanish Registry of Childhood Tumours (RETI-SEHOP), University of Valencia, Valencia, Spain b Paediatric Oncohematology Unit, Hospital La Fe; and Department of Paediatrics, University of Valencia, Valencia, Spain c Direzione scientifica, Epidemiologia & Prevenzione, Milano, Italy d Clinical Translational Cancer Research Group – Paediatric Oncology, IIS La Fe, Valencia, Spain e Spanish Society of Paediatric Haematology and Oncology (SEHOP), Spain f Paediatric Oncohematology Unit, Hospital Universitario Virgen Macarena, University of Sevilla, Sevilla, Spain g Spanish Network of Cancer Registries (REDECAN), Spain h Department of Epidemiological Research and Molecular Medicine, Evaluative Epidemiology Unit, Fondazione IRCCS Istituto Nazionale dei Tumori di Milano, Italy A R T I C L E I N F O A B S T R A C T Keywords: Purpose: Linkage between clinical databases and population-based cancer registries may serve to evaluate Eu­ Neuroblastoma ropean Reference Networks’ (ERNs) activity, by monitoring the proportion of patients benefiting from these and Childhood cancer their impact on survival at a population level. To test this, a study targeting neuroblastoma (Nb) was conducted Cancer registry in Spain by the European Joint Action on Rare Cancers. Clinical registry Record linkage Material and methods: Subjects: Nb cases, incident 1999–2017, aged < 15 years. Linkage included: Spanish Completeness Neuroblastoma Clinical Database (NbCDB) (1217 cases); Spanish Registry of Childhood Tumours (RETI) (1514 Incidence cases); and 10 regional population-based registries (RPBCRs) which cover 33% of the childhood population (332 Survival cases). Linkage was semiautomatic. We estimated completeness, incidence, contribution, deficit, and 5-year European Reference Network survival in the databases and specific subsets. Joint Action on Rare Cancers Results: National completeness estimates for RETI and NbCDB were 91% and 72% respectively, using the Spanish RPBCRs on International Incidence of Childhood Cancer (https://iicc.iarc.fr/) as reference. RPBCRs’ specific contribution was 1.6%. Linkage required manual crossover in 54% of the semiautomatic matches. Five-year survival was 74% (0–14 years) and 90% (0–18 months). Conclusions: All three databases were incomplete as regards Spain as a whole and should therefore be combined to achieve full childhood cancer registration. A unique personal patient identifier could facilitate such linkage. Most children have access to Nb clinical trials. Consolidated interconnections between the national registry and clinical registries (including ERNs and paediatric oncology clinical groups) should be established to evaluate outcomes. facing huge challenges (lack of correct diagnosis and treatment, and hence worse survival). Rare cancer survival rates vary across Europe[3, 1. Introduction 4], reflecting these deficiencies in detection, diagnosis and treatment, and disparities among regions/countries [5]. To overcome this, joint Childhood cancer incidence in Europe is 141 cases/1,000,000 chil­ expertise and networked healthcare are required. European Reference dren aged 0–14 years [1]. According to the published threshold for rare Networks (ERNs) were born as virtual networks involving healthcare cancers [2], all childhood cancers are life-threatening rare diseases, and the SpNbLinkWG. ☆ * Correspondence to: POG Department, University of Valencia, Paediatric Oncohematology Unit, Torre G-21, Hospital La Fe, Avda Fernando Abril Martorell 106, Valencia 46026, Spain. E-mail address: adela.canete@uv.es (A. Cañete). 1 Paediatric oncology units and cancer registries that contributed data to the linkage are included https://doi.org/10.1016/j.canep.2022.102145 Received 13 October 2021; Received in revised form 8 February 2022; Accepted 11 March 2022 Available online 25 March 2022 1877-7821/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by- nc-nd/4.0/). A. Cañete et al. Cancer Epidemiology 78 (2022) 102145 stores data on 1400 patients from 38 hospitals, up until 2017. Hospital Nomenclature coverage remained stable across the study period. Second opinions and non-incident cases were excluded from the study. Cases contributed: ERNs European Reference Networks 1217. CDB Clinical database JARC Joint Action on Rare Cancers 2.1.2. Spanish Registry of Childhood Tumours (RETI) NbCDB Neuroblastoma clinical database The RETI [15] coordinates a nationwide network including all the 48 RETI Spanish Registry of Childhood Tumours paediatric oncology units distributed across all regions of Spain. After RPBCRs Regional population-based cancer registries. obtaining informed consent, these units report all new cases and follow SEHOP Spanish Society of Paediatric Haematology and them up on request. The main goal is survival estimation as an indicator Oncology of results. RETI’s estimated average completeness for all of Spain is CC Childhood cancers currently 95%, and approaches to be virtually complete for children ASRw Age-standardised incidence rate aged 0–14 years in five regions (Aragon, Basque Country, Catalonia, ICCC-3 International Classification of Childhood Cancer Madrid, Navarre) [15,16], covering 37% of Spain [13] (Supplementary INES Infant Neuroblastoma European Study. Table 1). These regions constitute the RETI’s “high-completeness area” EUNS European Unresectable Neuroblastoma Study. included in the European Network of Cancer Registries[17] and SIOPEN Neuroblastoma Group of the European Society of population-based international projects [1,4,9,18]. Cases contributed: Paediatric Oncology 1514. 2.1.3. Regional population-based cancer registries [19] Sixteen RPBCRs were active during the study period, covering 41% providers across Europe, aimed at tackling complex or rare diseases and of children: ten participated in this study (Albacete, Asturias, Canary conditions that require highly specialised treatment and concentration Islands, Castile-Leon, Ciudad Real, Girona, Majorca, Murcia, Tarragona of knowledge and resources [6]. PAEDCAN is the ERN for paediatric and Valencian Region), covering 33% of Spanish children [13] (Sup­ cancer [7]. plementary Table 1). Five registries could not participate. All regional The ERNs’ capacity to work with patients and the final impact on the registries are members of the European Network [17] and the Spanish real population must be monitored by current cancer information sys­ Network of Cancer Registries [19]. They have participated in interna­ tems. Clinical databases (CDBs) and population-based cancer registries tional projects which require and evaluate population-base complete­ (PBCRs) can be linked to monitor the proportion of rare cancer patients ness [1,4,9,20]. Cases contributed: 332. benefiting from the ERNs, and thereby contribute to evaluate ERNs at a All participating registries and databases had regulatory data- population level. exchange agreements with RETI. See variables in Annex-1. To this end, we conducted a linkage study in Spain, within the Eu­ ropean Joint Action on Rare Cancers (JARC) [8], targeting neuroblas­ 2.2. Incidence toma (Nb), the most frequent paediatric extracranial solid tumour (incidence in Spain: 14.9 cases/1,000,000 children aged 0–14 years Incidence rates were calculated as number of cases per million child- [9]). Nb has a heterogeneous clinical presentation and outcome. Trials years, and age-standardised rates adjusted to the world standard popu­ have long shown that infants (children younger than 366 days) have lation (ASRw) [21]. 95% confidence intervals (95%CI) were estimated better outcomes. London et al. [10] showed that age is a continuous by Poisson approximation. The population at risk in the participating variable for risk stratification, and the International Neuroblastoma Risk RPBCRs was calculated according to their geographical areas and Group [11] confirmed this, setting the cut-off point at 18 months (547 participation periods (Supplementary Table 2). In the RETI, NbCDB, days). RETI&NbCDB and (RETI&NbCDB)&RPBCRs sets (see linkage below), We linked the Spanish Neuroblastoma Clinical Database (NbCDB), the population at risk was the entire Spanish population aged 0–14 years the Spanish Registry of Childhood Tumours (RETI) (both national) and in 1999–2017 [13]. regional population-based cancer registries (RPBCRs), and report our results in terms of linkage, feasibility, completeness and survival. 2.3. Completeness 2. Material and methods Completeness of registration by NbCDB, RETI, RETI&NbCDB and (RETI&NbCDB)&RPBCRs was estimated by the ratio of observed-to- We studied children aged 0–14 years diagnosed with Nb (Group IVa, expected cases [22]. 95%CI was estimated using Vandenbroucke’s International Classification of Childhood Cancer [12]) nationwide method [23]. Expected cases were estimated from the 12 Spanish across the period 1999–2017. The average Spanish national population RPBCRs included in the International Incidence of Childhood Cancer in this age group during the study period totalled 6,585,544 (Spanish (IICC-3) [9], excluding RETI (hereinafter RPBCRs-IICC-3) and the cor­ National Statistics Institute [13]). 3063 records from three independent responding population for the whole of Spain in 1999–2017 [13]. The sources were included in the study. IICC-3 was chosen as the most recent world account of childhood cancer incidence by country and registry. The RPBCRs-IICC-3 covered the 2.1. Sources period 1990–2013 and 35% of Spanish children (Supplementary Table 1). 2.1.1. Spanish Neuroblastoma Clinical Database (NbCDB) Since the incidence rate for ages 0–18 months was not available in The NbCDB [14] includes all cases in clinical trials/studies. The the RPBCRs-IICC-3, and the population for this age group was not in the Spanish Neuroblastoma Group was created in 1987 by the Spanish So­ official population figures[13], the incidence rate for this age group was ciety of Paediatric Oncology (SEHOP) to help clinicians with diagnosis, estimated using data from the RETI’s high-completeness area in treatment, and trial enrolment. A procedure is available which facili­ 2000–2016. Cases were those aged 0–18 months, and the population at tates biological, histopathological studies and trial assignment for cen­ risk was approximated by adding half of all children aged 1 year to the tres that participate voluntarily. All patients (or guardians) are required 0-year population [13]. Estimated rate 0–18 months = 66.4 cases/1, to give informed consent. Cases are confirmed centrally and followed up 000,000 (95%CI: 58.9–73.8). according to clinical guidelines. Valencia’s La Fe Hospital database Since RPBCRs are population-based registries previously evaluated 2 A. Cañete et al. Cancer Epidemiology 78 (2022) 102145 Fig. 1. Simplified diagram of the linkage. in international studies, their completeness was assumed to be virtually Co% A = ———————————————————————— complete by current standards. ——————————————————————————————— —— × 100. 2.4. Linkage Cases in final set without repetitions. All cases were included in linkage. Matching was semi-automatic 2.5.2. Specific contribution using Dalink v4.2 software [24]. Since there was no unique personal Cases in A, but not B. identifier common to all files, personal details and shared variables were SCo% A = ———————————————————————— used for linkage. Weights were assigned to each item as follows: 1st —————————————————————————————— surname (25%); 2nd surname (25%); name (24%); and birth and inci­ × 100. dence dates (13% each). A similarity probability was obtained for each Cases in final set without repetitions. matching. A human decision was needed for probabilities < 60% (due to name variability, abbreviations, spelling, etc.). Complementary infor­ 2.5.3. Underreporting mation was consulted, when necessary. Nevertheless, a matching deci­ URT% A = 100 – Co% file A. sion was still impossible for some record pairs, which coincided in some details while differing in others, e.g., some records had the same first 2.5.4. Overlapping between two files and last names, but different birth or incidence dates. Where comple­ Cases common to A and B, without repetitions. mentary information did not permit a decision, records were classified Ov% A (in B) = ————————————————————— as ‘doubtful matchings’ and the patients excluded from all analyses. ——————————————————————————————— The linkage procedure was as follows (see also Fig. 1): —— × 100. 1st Step. Linkage between RETI and NbCDB (all Spain, 0–14 years, Cases in A. 1999–2017). The result was the RETI&NbCDB file, with cases present only in RETI 2.5.5. Overlapping among all files and not in NbCDB (RETI-only), only in NbCDB and not in RETI (NbCDB- Cases common to all files, without repetitions. only), or in both (common-RETI&NbCDB). RETI&NbCDB was the final OvT% = —————————————————————————— set. ——————————————————————————————— 2nd Step. Linkage between RETI&NbCDB and RPBCRs. – × 100. The result was the (RETI&NbCDB)&RPBCRs file, with RETI&NbCDB- Cases in final set without repetitions. only cases, RPBCR-only cases and common-(RETI&NbCDB)&RPBCRs cases. (RETI&NbCDB)&RPBCRs was the final set. 2.6. Survival 2.5. Contribution and deficit Observed 5-year survival was estimated using the Kaplan–Meier method[26]. To allow for complete 5-year follow-up, estimates were Contribution and deficit were evaluated using five indicators [25]. made for the 1999–2011, 1999–2005, and 2006–2011 periods. Cases For simplicity, only two files (A and B) are considered here: were followed up until the 5th anniversary of the incidence date, using the three sources and the National Death Index where necessary. 2.5.1. Total contribution Without a common identifier, searches in the Death Index had to be Cases in A. performed using personal identification details. Patients lost to 3 A. Cañete et al. Cancer Epidemiology 78 (2022) 102145 follow-up were those who were alive at the last documented date of 3. Results follow-up before the 5th anniversary. No estimates were made in < 15 cases. Survival differences were tested using the Log-Rank[27] for ages The three sources supplied 3063 records in total. Linkage required 0–14 years and 0–18 months, period 1999–2011. All calculations were manual crossover in 54% of matches. Fifty-nine doubtful matching pa­ performed using the IBM SPSS Statistics 24.0 programme. tients were excluded (RETI, 28; NbCDB, 28; RPBCRs, 3), leaving 3004 patients (RETI, 1486; NbCDB, 1189; RPBCRs, 329) for analysis purposes. Table 1 Linkage of Nb in Spain study. Number of cases in each file entering the linkage (RETI, NbCDB, RPBCRs) and linked sets (RETI&NbCDB, (RETI&NbCDB)&RPBCs), 1999–2017, and the reference set of registries (RPBCRs-IICC-3), 1990–2013, by age group and incidence rate. Files/sets N Incidence rates per 1,000,000 person-years 0–18 months 0–4 years 0–14 years 0–18 months 0–4 years 0–14 years 0–14 years ASRw (95% CI) RETI 803 1304 1486 63.5 31.0 11.9 13.4 (12.7–14.1) NbCDB 654 1047 1189 51.7 24.9 9.5 10.7 (10.1–11.3) RPBCRs 186 295 329 N/A 30.8 11.6 13.1 (11.7–14.5) RETI&NbCDB 915 1478 1683 72.4 35.1 13.5 15.1 (14.4–15.9) (RETI&NbCDB) &RPBCRs 931 1502 1711 73.7 35.7 13.7 15.4 (14.7–16.1) RPBCRs-IICC-3 N/A 500 561 N/A 35.0 12.3 14.9 (13.8–16.0) NOTES -Nb: Neuroblastoma -RETI: File of the Spanish Registry of Childhood Tumours -NbCDB: File of the Spanish Neuroblastoma Clinical Database -RPBCRs: File of the regional population-based cancer registries participating in the study (Supplementary Table 2). The difference between the total number of cases in this row and the total in Supplementary Table 2 is due to the exclusion of three patients with doubtful matching, as stated in the Results section -RETI&NbCDB: Set resulting from linkage of the RETI and NbCDB files -(RETI&NbCDB)&RPBCRs: Set resulting from linkage of the RETI&NbCDB and RPBCR files -RPBCRs-IICC-3: Set of Spanish RPBCRs in the IICC-3[9], with the RETI excluded, used as a reference. Period: 1990–2013 (see Supplementary Table 1) -ASRw: Age standardised rate based on the world standard population[21] -N/A: Not available -Populations at risk: -RETI, NbCDB, RETI&NbCDB and (RET&NbCDB)&RPBCRs in all Spain (1999–2017): 0–14 years of age, 125,125,334 person-years; 0–18 months, 12,640,259 per­ son-years -RPBCRs, in the geographical area covered by the registries and their period of participation (see Supplementary Table 2) (0–14 years of age): 28,407,194 person-years; 0–18 months, not available -RPBCRs-IICC-3 (1990–2013): 0–14 years, 45,529,309; 0–18 months, not available Table 2 Linkage of Nb in Spain study. Number of cases and completeness of the files entering the linkage (RETI, NbCDB, RPBCRs) and linked sets (RETI&NbCDB, RETI&NbCDB)&RPBCs), 1999–2017, and the reference set of registries (RPBCRs-IICC-3), 1990–2013, by age group. Sources/sets N 0–14 years Completeness for Spain % (CI 95%) 0–18 monthsa 0–4 yearsb 0–14 yearsb RETI 1486 95.7 (89.2–102.4) 88.5 (83.8–93.4) 90.6 (86.1–95.3) NbCDB 1189 77.9 (72.1–84.0) 71.1 (66.8–75.4) 72.5 (68.4–76.7) RPBCRs 329 c RETI&NbCDB 1683 109.0 (102.1–116.2) 100.3 (95.3–105.5) 102.6 (97.8–107.6) (RETI&NbCDB) &RPBCRs 1711 110.9 (103.9–118.2) 101.9 (96.9–107.2) 104.3 (99.5–109.4) RPBCRs-IICC-3 561 c NOTES -Nb: Neuroblastoma -RETI: File of the Spanish Registry of Childhood Tumours -NbCDB: File of the Spanish Neuroblastoma Clinical Database -RPBCRs: File of the regional population-based cancer registries participating in the study (Supplementary Table 2). The difference between the total number of cases in this row and the total in Supplementary Table 2 is due to the exclusion of three patients with doubtful matching, as stated in the Results section -RETI&NbCDB: Set resulting from linkage of the RETI and NbCDB files -(RETI&NbCDB)&RPBCRs: Set resulting from linkage of the RETI&NbCDB and RPBCRs files -RPBCRs-IICC-3: Set of Spanish RPBCRs in the IICC-3[9], with the RETI excluded, used as a reference. Period: 1990–2013 (see Supplementary Table 1) -Completeness: a: The expected number of Nb cases to calculate completeness for children 0–18 months of age is based on the rate estimated from the population-based area of RETI for 0–18 months. b: The expected number of Nb cases to calculate completeness for children aged 0–4 and 0–14 years is based on the rates from the RPBCR-IICC-3 set for the same groups of age. c: Completeness of RPBCRs assumed to be virtually complete in their geographical area of coverage. -Population at risk -RETI, NbCDB, RETI&NbCDB and (RET&NbCDB)&RPBCRs in all Spain (1999–2017): 0–14 years of age, 125,125,334 person-years; 0–18 months, 12,640,259 per­ son-years -RPBCRs, in the geographical area covered by the registries and their period of participation (see Supplementary Table 2) (0–14 years of age): 28,407,194 person-years; 0–18 months, not available -RPBCRs-IICC-3 (1990–2013): 0–14 years, 45,529,309; 0–18 months, not available 4 A. Cañete et al. Cancer Epidemiology 78 (2022) 102145 The RETI incidence rates in Spain were similar to those of the significant differences by sex or age between this and the RETI, NbCDB, RPBCRs set, while the NbCDB rates were slightly lower (Table 1). Taking RETI&NbCDB or (RETI&NbCDB)&RPBCRs sets. The same applied to the the RPBCRs-IICC-3 as reference, both RETI and NbCDB failed to achieve RPBCRs-only, RETI-only, and NbCDB-only subsets. Table 4 shows sim­ total completeness (Table 2). ilarities between the RETI, NbCDB, RETI&NbCDB and RPBCRs files. The linked RETI&NbCDB file had 1683 individual patients (992 Few cases were lost to follow-up. Overall survival was around 90% in common, 494 RETI-only and 197 NbCDB-only; Supplementary Table 3). children aged 0–18 months and about 74% in the 0–14 year age group. Accordingly, the RETI’s and NbCDB’s underreporting rates were 11.7% For children older than 18 months, survival in the set with a higher and 29.4% respectively (Table 3). In the second linkage step, 28 addi­ number of cases varied from 58% for children up to 4 years old tional cases were identified in the RPBCRs (Supplementary Table 3). (n = 389), to 53% for ages 5–9 years (n = 114) and 61% for the older Since participating RPBCRs covered one-third of the Spanish population, group (n = 61). RPBCRs-only cases (n = 23) registered the highest about 100 cases are estimated to be missing from the merged survival. NbCDB-only cases showed the lowest survival at ages 0–18 RETI&NbCDB during the period 1999–2017. Hence, final completeness months (n = 72) and the highest at ages 19 months to 4 years (n = 49) estimates of 84% and 67% were obtained for the RETI and NbCDB (Table 5). Survival was similar, without statistically significant differ­ respectively, figures slightly lower than those obtained taking the ences in the sets with more than 800 cases. The most recently diagnosed RPBCRs-IICC-3 as reference (91% and 73% respectively; Table 2). patients survived slightly longer than did patients in the past (Supple­ The linked files showed completeness of around 100% for age groups mentary Table 4). 0–4 and 0–14 years, for which the RPBCRs-IICC-3 set was the reference. For ages 0–18 months, a result > 100% was obtained (Table 2). 4. Discussion Due to the low number of cases from the RPBCRs, there were no The JARC [8] is a project supported by the European Union and Member States aimed at advancing the quality of care for and research Table 3 on rare cancers. Task 4.4, WP4 aspires to evaluate the impact of ERNs, Linkage of Nb in Spain study. Indicators of contribution and deficit of each file by linking clinical databases and population-based cancer registries to entering the linkages by age group, 1999–2017. monitor the proportion of childhood cancers benefiting from the ERNs, Linkages Co% SCo% URT% Ov% OvT% and the coverage and completeness of epidemiological and clinical st 1 Linkage Result: RETI&NbCDB sources in the target population. 0–14 years RETI 88.3 29.4 11.7 66.8 58.9 Nb was chosen because it has a dedicated national clinical database NbCDB 70.6 11.7 29.4 83.4 in Spain, international clinical research collaboration in Europe estab­ 0–18 months RETI 87.8 28.5 12.2 67.5 59.2 lished under SIOPEN [28], and national clinical databases in most NbCDB 71.5 12.2 28.5 82.9 2nd Linkage: Result: countries. As regards cancer registration, Spain has yet to complete its (RETI&NbCDB)&RPBCRs national childhood cancer registration. Two systems coexist: several 0–14 years RETI&NbCDB 98.4 80.8 1.6 17.9 17.6 RPBCRs (described above); and the RETI, which is striving for national RPBCRs 19.2 1.6 80.8 91.5 completeness [16]. Additionally, every SEHOP clinical working group 0–18 months RETI&NbCDB 98.3 80.0 1.7 18.6 18.3 maintains its database for clinical and outcome evaluations. The NbCDB RPBCRs 20.0 1.7 80.0 91.4 was developed to co-ordinate sample delivery, trial recruitment, second NOTES consultations, and clinical follow-up. Accordingly, the NbCDB, RETI and -Nb: Neuroblastoma 10 RPBCRs were included in the study. -Co%: Contribution In Spain, children do not have a national identification number, -SCo%: Specific contribution whereas adults are required by law to have one for purely administrative -URT%: Under-reporting of a file in relation to the total final set -Ov%: overlapping between files purposes. Since there is no unique patient identifier in Spain for clinical, -OvT%: overlapping in total final set public health or research issues, linkage had to be semi-automatic, using -RETI: File of the Spanish Registry of Childhood Tumours personal identification details. Manual crossover was required in 54% of -NbCDB: File of the Spanish Neuroblastoma Clinical Database. matches: as this is neither efficient nor sustainable, we recommend -RETI&NbCDB: Complete set resulting from the 1st linkage, RETI and NbCDB implementing a unique (national/European) personal health identifier files capable of tracking patients in the health system and research envi­ -RPBCRs: File of the regional population-based cancer registries participating in ronment, such as the European Unified Patient Identity Management the study (see Supplementary Table 2) (EUPID20) [29]. -(RETI&NbCDB)&RPBCRs: Complete set resulting from the 2nd linkage, Completeness of expected Nb incidence in Spain was insufficient in RETI&NbCDB and RPBCRs files both RETI and NbCDB. However, the merged RETI&NbCDB and Table 4 Linkage of Nb in Spain study. Differences by age, sex, year of diagnosis and survival between the main files and sets resulting from the linkages. RETI NbCDB RETI&NbCDB RPBCRs N 1486 1189 1683 329 Age in years (mean (st err)) 1.80 (0.065) 1.74 (0.070) 1.78 (0.061) 1.71 (0.137) Age 0–18 months % respect 0–14years) 54.0% 55.0% 54.4% 56.5% Sex (% female) 46.2% 44.9% 46.3% 46.8% Year of diagnosis (mean (st err)) 2008.52 (0.138) 2008.34 (0.141) 2008.40 (0.129) 2006.53 (0.250) 5-year survival (st err) Age: 0–18 months 90.2 (1.3) 90.1 (1.4) 89.2 (1.3) 90.4 (2.4) 5-year survival (st err) Age: 0–14 years 73.6 (1.4) 73.5 (1.5) 74.0 (1.3) 73.3 (2.7) NOTES -Nb: Neuroblastoma -RETI: File of the Spanish Registry of Childhood Tumours -NbCDB: File of the Spanish Neuroblastoma Clinical Database -RETI&NbCDB: Set resulting from the linkage of files RETI and NbCDB -RPBCRs: File of the regional population-based cancer registries participating in the study (see Supplementary Table 2) 5 A. Cañete et al. Cancer Epidemiology 78 (2022) 102145 Table 5 Linkage of Nb in Spain study. 5-year survival by age group in the main files, sets, and subsets resulting from the linkages, with percentage of cases lost to follow-up. Period of diagnosis: 1999–2011. Files/sets/subsets Group of age 0–14 years 0–18 months 19 months-4 years 5–9 years 10–14 years No. of 5-year No. of 5-year No. of 5-year No. of 5-year No. of % 5-year cases survival (95% cases survival (95% cases survival (95% cases survival (95% cases Lost survival (95% CI) CI) CI) CI) CI) RETI 529 90 (88–93) 334 55 (50–60) 102 54 (44–64) 26 54 (35–73) 991 0.0 74 (71–76) NbCDB 446 90 (87–93) 282 56 (50–62) 82 46 (36–57) 18 61 (39–84) 828 0.4 74 (71–76) RPBCRs 156 90 (86–95) 95 53 (43–63) 26 46 (27–65)a 277 0.0 73 (68–79) RETI&NbCDB 601 89 (87–92) 383 58 (53–63) 113 52 (43–61) 29 59 (41–76) 1126 0.3 74 (71–77) Comm- 374 92 (89–94) 233 52 (45–58) 71 48 (36–59) 15 53 (28–79) 693 0.0 73 (70–76) RETI&NbCDB (RETI&NbCDB)& 615 89 (87–92) 389 58 (53–63) 114 53 (43–62) 31 61 (44–78) 1149 0.3 74 (72–77) RPBCRs a RETI-only 155 87 (81–92) 101 62 (53–72) 42 64 (50–79) 298 0.0 75 (70–80) NbCDB-only 72 82 (73–91) 49 77 (66–89) 14 50 (24–76)a 135 2.2 77 (70–84) RPBCRs-only 14 b 6 b 1 b 2 b 23 0.0 87 (73–100)c NOTES -Nb: Neuroblastoma -RETI: File of the Spanish Registry of Childhood Tumours -NbCDB: File of the Spanish Neuroblastoma Clinical Database -RPBCRs: File of the regional population-based cancer registries participating in the study (see Supplementary Table 2) -RETI&NbCDB: Set resulting from the linkage of files RETI and NbCDB -Comm-(RETI&NbCDB): Subset of cases common to RETI and NbCDB -(RETI&NbCDB)&RPBCRs: Set resulting from the linkage of files RETI&NbCDB and RPBCRs -RETI-Only: Subset of cases present in RETI but not in NbCDB -NbCDB-Only: Subset of cases present in NbCDB but not in RETI -RPBCRs-Only: Subset of cases present in RPBCRs but not in (RETI&NbCDB) -a: Group of age 5–14 years -b: Survival not calculated because < 15 cases were available -c: 95% CI truncated at 100 (RETI&NbCDB)&RPBCRs files approached full completeness when the followed up. RPBCRs-IICC-3 set was used as reference (0–4 and 0–14 age groups), This study has several limitations: (1) the NbCDB could not be though some overestimation cannot be ruled out (see limitations below). restricted to regions covered by the RPBCRs for linkage purposes Nevertheless, the result for the 0–18 month age group, i.e., completeness because residence data are not recorded in the NbCDB. We therefore > 100%, is illogical. The following explanation might at least partially recommend introducing patient place/province of residence in clinical account for this: since completeness for cases aged 0–18 months had to databases; (2) some duplications were likely present in the be calculated by taking the rate obtained from the RETI’s high- RETI&NbCDB file, due to matching errors; (3) some cases not resident in completeness area as reference, the observed overestimate of Spain may be present in the linked files, thus contributing to an over­ completeness for the merged file could be the effect of an underestimate estimate of completeness; (4) five RPBCRs accounting for 8% of of expected cases in the 0–18 month age group. This underestimate coverage were absent from the analysis. Nevertheless, since all RPBCRs would have been the result of a deficit in registration of 0–18 month in Spain covered 41% of the total childhood population, even if these 5 cases in the RETI and its high-completeness area, which would have registries had participated, total completeness for all of Spain would not produced an erroneously low reference rate and thus a correspondingly have been achieved; (5) the RPBCRs participating in the study covered low number of expected cases. Concurrently, clinical experience shows only 33% of the child population in Spain; (6) the reference rate used for that there might be a small number of cases not registered in the RETI the completeness estimation was based on the 12 RPBCRs included in (low-stage cases diagnosed in neonatology or local hospitals and not IICC-3 [9], which together covered 35% of the child population (Sup­ referred to paediatric oncology units, due to spontaneous regression or plementary Table 1); and (7) the time-span covered by the only surgical treatment). Hence, the observed overestimate of RPBCRs-IICC-3 was 1990–2013, whereas the study period was completeness must be assumed to have highlighted a specific RETI 1999–2017. This temporal discrepancy might have led to a slight limitation for patients aged 0–18 months. Furthermore, some artefacts overestimate of completeness, since childhood cancer incidence gradu­ may have spuriously increased the number of observed cases in all ages, ally increased in southern Europe from 1991 to 2010 [18]. including the undetected presence of cases not resident in Spain, which Despite some incompleteness, the RETI&NbCDB file is currently the cannot be definitively ruled out (see limitations below). most complete available set of cases for studying Nb survival in Spain. The specific contribution of the participating RPBCRs to the final set Further analyses should be conducted using clinical criteria which were (RETI&NbCDB)&RPBCRs was small (1.6%), as they were substantially not the focus of this paper. Our 5-year survival results are consistent overlapped by the RETI and NbCDB. All but 28 of the RPBCRs’ Nb cases with those in Europe [4]. Age-specific survival comparisons are difficult had been included in clinical studies and/or treated at Nb units, sug­ because we applied clinical age stratification [10]. In general, there gesting good access to specialised paediatric oncological care and were no significant differences in 5-year survival. The discrepancies in complementarity of the databases, which could be improved by making survival between the NbCDB-only and common-RETI&NbCDB subsets them interoperable. Importantly for paediatric oncologists, missing are minimal and should be clarified by examining the clinical data, cases were identified in the linkage of the NbCDB and RETI, helping identifying the possible effects of a lack of follow-up, and analysing clinician researchers to retrieve such patents, improve NbCDB individual cases and registration practice from 1999 to 2011, when completeness, and understand how these patients were treated and several trials (INES [30], EUNS [31]) and interim studies were ongoing. 6 A. Cañete et al. Cancer Epidemiology 78 (2022) 102145 From our standpoint, the absence of differences in survival among the Appendix A. Supporting information larger subsets indicates good access to care, and confirms the good work and quality of clinical practices and evaluation by the different infor­ Supplementary data associated with this article can be found in the mation systems (clinical/ epidemiological). online version at doi:10.1016/j.canep.2022.102145. The small differences in incidence and survival reaffirm the complementarity of the two files combined in the RETI&NbCDB set. We References therefore recommend integrating the NbCDB and RETI, together with RPBCRs, thereby approximating population-based levels, and enabling [1] E. Steliarova-Foucher, M. Colombet, L.A.G. Ries, F. Moreno, A. Dolya, F. Bray, et al., International incidence of childhood cancer, 2001-10: a population-based dynamic, updated survival analyses without placing constraints on the registry study, Lancet Oncol. 18 (2017) 719e31, https://doi.org/10.1016/S1470- development of the clinical database. 2045(17)30186-9. To assess the extent to which the target population accesses appro­ [2] G. Gatta, A.P. Dei Tos, J.M. Van Der Zwan, P.G. Casali, S. Siesling, A. Tavilla, et al., Rare cancers are not so rare: the rare cancer burden in Europe, Eur. J. Cancer 47 priate healthcare for childhood cancer and its subsequent outcome, it is (2011) 2493–2511, https://doi.org/10.1016/j.ejca.2011.08.008. essential that Spain, like other European countries, achieve full national [3] G. Gatta, A. Trama, R. Capocaccia, RARECARENet Working Group. Epidemiology registration of childhood cancer [16]. According to JARC’s Rare Cancer of rare cancers and inequalities in oncologic outcomes, Eur. J. Surg. Oncol. 45 (2019) 3–11, https://doi.org/10.1016/j.ejso.2017.08.018. Agenda 2030 [32], registries must incorporate variables such as the [4] G. Gatta, L. Botta, S. Rossi, T. Aareleid, M. Bielska-Lasota, J. Clavel, et al., Toronto Staging Guidelines [33,34]. Moreover, a stable interconnection Childhood cancer survival in Europe 1999-2007: results of EUROCARE-5-a between the national registry and the clinical databases of the relevant population-based study, Lancet Oncol. 15 (2014) 35–47, https://doi.org/10.1016/ S1470-2045(13)70548-5. healthcare providers (including ERNs) and SEHOP clinical groups [5] S. Mathoulin-Pélissier, K. Pritchard-Jones, Evidence-based data and rare cancers: should be established for all childhood cancers. To this end, a unique the need for a new methodological approach in research and investigation, Eur. J. personal patient identifier common to all databases, systematic Surg. Oncol. 45 (2019) 22–30, https://doi.org/10.1016/j.ejso.2018.02.015. cross-linkage, and appropriate IT tools are needed. European [6] European Reference Networks. Available from: 〈https://ec.europa.eu/health/ern _en〉 [accessed 23 November 2020]. data-protection regulations [35] must be positively implemented to [7] ERN PAEDCAN. European Reference Network for Paediatric Oncology. 〈https://pa ensure that clinical and population-based databases are interoperable, edcan.ern-net.eu/〉 [accessed 23 May 2021]. and so allow for more efficient evaluation of outcomes. [8] Joint Action on Rare Cancers (JARC). 〈https://jointactionrarecancers.eu/〉 [accessed 23 November 2020]. [9] E. Steliarova-Foucher, M. Colombet, L.A.G. Ries, P. Hesseling, F. Moreno, H. Role of the funding source Y. Shin, et al. (Eds.), International Incidence of Childhood Cancer, Volume III (electronic version), International Agency for Research on Cancer, Lyon, France, 2017. 〈http://iicc.iarc.fr/results/〉. accessed 23 November 2020. The funding sources had no role in study conception or design, data [10] W.B. London, R.P. Castleberry, K.K. Matthay, A.T. Look, R.C. Seeger, H. Shimada, collection, data analysis, data interpretation, writing of the paper or in et al., Evidence for an age cutoff greater than 365 days for neuroblastoma risk the decision to submit for publication. group stratification in the Children’s Oncology Group, J. Clin. Oncol. 23 (2005) 6459–6465, https://doi.org/10.1200/JCO.2005.05.571. [11] S.L. Cohn, A.D. Pearson, W.B. London, T. Monclair, P.F. Ambros, G.M. Brodeur, et Funding al., The International Neuroblastoma Risk Group (INRG) Classification System: An INRG Task Force Report, J. Clin. Oncol. 27 (2008) 289–297, https://doi.org/ 10.1200/JCO.2008.16.6785. This work is part of the project “Joint Action on Rare Cancers”/JARC [12] E. Steliarova-Foucher, C. Stiller, B. Lacour, et al., International Classification of funded by the European Union’s Health Programme (2014–2020) Childhood Cancer, Cancer 103 (2005) 1457–1467, https://doi.org/10.1002/ [Grant No. 724161]. The Spanish Registry of Childhood Tumours has cncr.20910. the support of the Ministry of Health of Spain, the University of Valencia [13] Instituto Nacional de Estadística (INE). 〈http://www.ine.es〉 [accessed 27 March 2019]. and the Spanish Society of Paediatric Oncology; as well as the collabo­ [14] V. Castel Sánchez, C. Melero Moreno, P. García-Miguel García-Rosados, A. Navajas ration of the Enriqueta Villavecchia Foundation. Gutiérrez, J.I. Ruiz Jiménez, S. Navarro Fos, et al., Neuroblastoma en niños menores de 1 año, An Esp Pediat. 47 (1997) 584–590. [15] A. Cañete Nieto, E. Pardo Romaguera, A. Muñoz López, S. Valero Poveda, S. Porta CRediT authorship contribution statement Cebolla, M.S. Barreda Reines, R. Peris Bonet, Cáncer infantil en España. Estadísticas 1980-2020. Registro Español de Tumores Infantiles ((RETI-SEHOP)), A.C.N., G.G., R.C., and R.P.B. conceived and designed the study. E.P. Universitat de València,, Valencia, 2021 (Available from), 〈https://www.uv.es/rn ti/informes.html〉. R. and V.S.C., in collaboration with A.F.T., J.G.P., and W.G. members [16] E. Steliarova-Foucher, C. Stiller, M. Colombet, P. Kaatsch, R. Zanetti, R. Peris- acquired the data for this study. A.M.L., E.P.R., and V.S.C. performed the Bonet, Registration of childhood cancer: moving towards pan-European coverage? quality control. A.M.L. did the statistical analyses. A.C.N., A.M.L., E.P. Eur. J. Cancer 51 (2015) 1064–1079, https://doi.org/10.1016/j.ejca.2015.03.009. [17] European Network of Cancer Registries (ENCR). Available from: 〈https://www.en R., G.G., R.C., and R.P.B. interpreted the results. A.C.N., G.G., R.C., and cr.eu/encr-members-contact-list〉 [accesed 3 January 2022]. R.P.B. drafted the manuscript, which was reviewed and approved in its [18] E. Steliarova-Foucher, M.M. Fidler, M. Colombet, B. Lacour, P. Kaatsch, M. Piñeros, final version by all authors. et al., Changing geographical patterns and trends in cancer incidence in children and adolescents in Europe, 1991-2010 (Automated Childhood Cancer Information System): a population-based study, Lancet Oncol. 19 (2018) 1159–1169. Declaration of Competing Interest [19] REDECAN. Red Española de Registros de Cáncer. Available from: 〈https://redecan. org/redecan.org/es/page04b2.html?id=14&title=quienes-somos〉 [accessed 23 November 2020]. The authors declare that they have no known competing financial [20] Cancer Incidence in Five Continents (CI5C). Available from: 〈https://ci5.iarc.fr〉 interests or personal relationships that could have appeared to influence [accesed 3 January 2022]. the work reported in this paper. [21] M. Segi, Cancer mortality for selected sites in 24 countries (1950–57), Department of Public Health, Tohoku University of Medicine,, Sendai, Japan, 1960. [22] J.A. Ross, R.K. Severson, B.H. Pollock, L.L. Robinson, Childhood cancer in the Acknowledgements United States, Cancer 77 (1996) 201–207, https://doi.org/10.1002/(SICI)1097- 0142(19960101)77:1<201::AID-CNCR32>3.0.CO;2-7. The authors gratefully acknowledge the Ministry of Health of Spain, [23] J.P. Vandenbroucke (Ed.), Am. J. Epidemiol. 115 (1982) 303–304. [24] J. Gomez-Bao, J.L. Larriba-Pey, J. Ribes Puig, Record linkage performance for large the Spanish Society of Haematology and Paediatric Oncology, and the data sets. PAVLAD ’09 Proceedings of the ACM First International Workshop on Spanish Network of Cancer Registries for endorsing this study. We also Privacy and Anonymity for Very Large Databases, ACM,, New York, NY, USA, thank K. Pritchard Jones, A. Trama, M. Regaña, and J. Calabuig for their 2009, pp. 9–16, https://doi.org/10.1145/1651449.1651453. [25] M.F. Abad Garcia, Investigación Evaluativa en Documentación. Aplicación a la suggestions; V. Castel and D. Ramal for their work with the Spanish Documentación Médica, Universitat de València, Valencia, 1997. ISBN 84-370- Neuroblastoma Clinical Database; J.M. Borras and J. Ribes for the 3305-5. DALINK program; and R. Fernández-Delgado and M.S. Barreda for their continuous help in the Spanish Registry of Childhood Tumours. 7 A. Cañete et al. Cancer Epidemiology 78 (2022) 102145 [26] E.L. Kaplan, P. Meier, Nonparametric estimation from incomplete observations, without MYCN amplification: results of the SIOPEN study, Eur. J. Cancer 49 (2013) J. Am. Statist. Assoc. 53 (282) (1958) 457–481, https://doi.org/10.1080/ 3671–3679, https://doi.org/10.1016/j.ejca.2013.07.002. 01621459.1958.10501452. [32] PG Casali, A. Trama (Eds), Rare Cancer Agenda 2030. Ten recommendations from [27] R. Peto, M.C. Pike, P. Armitage, N.E. Breslow, D.R. Cox, S.V. Howard, et al., Design the EU Joint Action on Rare Cancers. Lecce, It. Yucanprint, 2020 [ISBN 978–88- and analysis of randomized clinical trials requiring prolonged observation of each 31642–71-2]. Available from: 〈https://www.jointactionrarecancers.eu/attachment patient II Analysis and examples, Br. J. Cancer 35 (1977) 1–39, https://doi.org/ s/article/265/Rare_Cancer_Agenda_2030.pdf〉 [accessed 23 November 2020]. 10.1038/bjc.1977.1. [33] J.F. Aitken, D.R. Youlden, A.S. Moore, P.D. Baade, L.J. Ward, V.J. Thursfield et al. [28] SIOPEN. Available from: 〈https://www.siopen.net/〉 [accessed 23 November Childhood cancer staging for population registries according to the Toronto 2020]. Childhood Cancer Staging Guidelines. Cancer Council Queensland Australia. [29] H. Ebner, D. Hayn, M. Falgenhauer, M. Nitzlnader, G. Schleiermacher, R. Haupt, et Brisbane, Australia, 2017. Available from: 〈https://cancerqld.blob.core.windows. al., Piloting the European Unified Patient Identity Management (EUPID) Concept net/content/docs/childhood-cancer-staging-for-population-registries.pdf〉 to facilitate secondary use of neuroblastoma data from clinical trials and [accessed 23 November 2020]. biobanking, Stud. Health Technol. Inform. 223 (2016) 31–38, https://doi.org/ [34] J.F. Aitken, D.R. Youlden, A.S. Moore, P.D. Baade, L.J. Ward, V.J. Thursfield, P.C. 10.3233/978-1-61499-645-3-31. Valery, A.C. Green, S Gupta, AL Frazier, Childhood cancer staging for population [30] H. Rubie, B. De Bernardi, M. Gerrard, A. Canete, R. Ladenstein, J. Couturier, et al., registries according to the Toronto Childhood Cancer Stage Guidelines. Cancer Excellent outcome with reduced treatment in infants with non-metastatic and Council Queensland and Cancer Australia: Brisbane, Australia; 2017. unresectable neuroblastoma without MYCN amplification: results of the [Estadificación del cáncer infantil para registros de base poblacional. Traducida prospective INES 99.1, J. Clin. Oncol. 29 (2011) 449–455, https://doi.org/ por el Grupo de Trabajo del RETI en la SEHOP, 2020. DOI: 10.7203/72724]. 10.1200/JCO.2010.29.5196. Available from: 〈http://doi.org/10.7203/72724〉 [accessed 23 November 2020]. [31] J.A. Kohler, H. Rubie, V. Castel, K. Beiske, K. Holmes, C. Gambini, F. Casale, [35] Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 C. Munzer, G. Erminio, S. Parodi, S. Navarro, C. Marquez, M. Peuchmaur, April 2016 on the protection of natural persons with regard to the processing of C. Cullinane, P. Brock, D. Valteau-Couanet, A. Garaventa, R. Haupt, Treatment of personal data and on the free movement of such data, and repealing Directive 95/ children over the age of one year with unresectable localised neuroblastoma 46/EC (General Data Protection Regulation). Available from: 〈http://data.europa. eu/eli/reg/2016/679/oj〉 [accessed 23 November 2020]. 8