Background: European national disparities in the integration of data linkage (ie, being able to match patient data between databases) into routine public health activities were recently highlighted. In France, the claims database covers almost the whole population from birth to death, offering a great research potential for data linkage. As the use of a common unique identifier to directly link personal data is often limited, linkage with a set of indirect key identifiers has been developed, which is associated with the linkage quality challenge to minimize errors in linked data.
Objective: The aim of this systematic review is to analyze the type and quality of research publications on indirect data linkage on health product use and care trajectories in France.
Methods: A comprehensive search for all papers published in PubMed/Medline and Embase databases up to December 31, 2022, involving linked French database focusing on health products use or care trajectories was realized. Only studies based on the use of indirect identifiers were included (ie, without a unique personal identifier available to easily link the databases). A descriptive analysis of data linkage with quality indicators and adherence to the Bohensky framework for evaluating data linkage studies was also realized.
Results: In total, 16 papers were selected. Data linkage was performed at the national level in 7 (43.8%) cases or at the local level in 9 (56.2%) studies. The number of patients included in the different databases and resulting from data linkage varied greatly, respectively, from 713 to 75,000 patients and from 210 to 31,000 linked patients. The diseases studied were mainly chronic diseases and infections. The objectives of the data linkage were multiple: to estimate the risk of adverse drug reactions (ADRs; n=6, 37.5%), to reconstruct the patient’s care trajectory (n=5, 31.3%), to describe therapeutic uses (n=2, 12.5%), to evaluate the benefits of treatments (n=2, 12.5%), and to evaluate treatment adherence (n=1, 6.3%). Registries are the most frequently linked databases with French claims data. No studies have looked at linking with a hospital data warehouse, a clinical trial database, or patient self-reported databases. The linkage approach was deterministic in 7 (43.8%) studies, probabilistic in 4 (25.0%) studies, and not specified in 5 (31.3%) studies. The linkage rate was mainly from 80% to 90% (reported in 11/15, 73.3%, studies). Adherence to the Bohensky framework for evaluating data linkage studies showed that the description of the source databases for the linkage was always performed but that the completion rate and accuracy of the variables to be linked were not systematically described.
Conclusions: This review highlights the growing interest in health data linkage in France. Nevertheless, regulatory, technical, and human constraints remain major obstacles to their deployment. The volume, variety, and validity of the data represent a real challenge, and advanced expertise and skills in statistical analysis and artificial intelligence are required to treat these big data.
Data linkage is a technique for linking data from different sources that relate to the same person . It increases the information available about each patient (clinical and administrative data, disease-related mortality, health care use, etc) and therefore expands research opportunities, particularly for research requiring large sample sizes, detailed data on hard-to-reach populations, or little loss to follow-up, to generate evidence with a high level of external validity [ , ]. The proliferation of the use of data linkage is reflected in the establishment of data linkage research centers worldwide [ ] (Australia [ - ], North America [ , ], the Netherlands [ ], the United Kingdom [ ]). In Europe, a recent review highlighted national disparities in the integration of data linkage into routine public health activities [ ].
In France, since the 2016 law on the modernization of the health system, the use of existing health data and their pooling have been promoted. As a result, French medicoadministrative databases (MADs) were linked together in a single French national health data system (the Système National des Données de Santé [SNDS]) in 2017 based on a reliable identification of individuals by the numéro d’inscription au repertoire (NIR), de-identified using 2 successive hash scrambling operations [, ]. The French national health data system groups together (1) data on claims for reimbursement of outpatient care (the Système National d’Information Inter-Régimes de l’Assurance Maladie [SNIIRAM]), (2) medicoadministrative data on hospitalization in public and private hospitals (the Programme de Médicalisation des Systèmes d’Information [PMSI]), (3) the national death registry (the Center d’Epidémiologie sur les causes médicales de Décès [CépiDc]), and (4) medicosocial data (the Caisse Nationale de Solidarité pour l’Autonomie [CNSA]) [ - ]. The SNDS covers almost 99% of the French population, making it 1 of the largest databases of continuous homogeneous claims in the world [ ]. Between 2007 and 2016, more than 400 scientific publications based on national health insurance data were identified in Medline, mainly on the real-life evaluation of drugs [ ]. The consumption of reimbursed health products (ie, drugs and medical devices) for outpatients is recorded in the SNDS but not treatments used in hospitalized patients (except for expensive reimbursed drugs). Moreover, limited clinical data (no diagnosis, except in the case of hospitalization or chronic disease), no biological test results, scarce sociodemographic data, and no information about occupation [ , ] highlighted the potential for further linkage with other databases.
Clinical cohorts, disease-specific or population-based registries, and hospital data warehouses represent other databases of interest for data linkage in which reimbursed or nonreimbursed health products used in real life can be recorded, often in a heterogeneous way, depending on the initial purpose of the database. All these databases represent important tools for providing information about the safety and benefits of approved health products, with data on real-life use, rare outcomes, and long-term effects that were undetectable in randomized controlled trials . Registries are particularly interesting for rare diseases, with more than 600 registries in Europe, for example, allowing development of clinical research in the field of rare diseases and so patient care improvement [ ]. As these databases potentially represent different points in patient management, linking them can also make possible the reconstruction of care trajectories to better understand therapeutic management throughout the patient’s life.
Data linkage involves being able to match patient data, sometimes anonymized, between databases. The use of a common unique identifier to directly link personal data is often limited because it does not exist, is not regulatory available, or contains errors . The national individual identifier “NIR” in French MADs is highly protected by privacy rules, leading to complex and lengthy formalities, which discourage potential applicants from using it [ , , , ]. Linkage with a set of indirect key identifiers has therefore been developed with deterministic and probabilistic approaches [ , , ]. Deterministic linkage is applied when there are several identifiers that match perfectly between data sets. In this case, the match of a given identifier is evaluated as a discrete “all-or-nothing” outcome [ ]. Probabilistic linkage uses statistical theory to associate each pattern of matching variable agreement with the likelihood that record pairs exhibiting the pattern are a match [ ]. These data linkages are associated with a methodological challenge on linkage quality to minimize errors in linked data. Data linkage has to balance the risk of missed matches (failing to link data from the same individual) with false matches (mistakenly linking data from different individuals) [ ]. As quality assessment of data linkage is essential to limit biased results and false interpretation, several recommendations highlight elements or information about the linkage pathway (data provision, method of data linkage, and data analyses) to be shared and checked [ , ]. In this context, the objective of this systematic review is to analyze the type and quality of research publications on indirect data linkage on health product use and care trajectories in France.
A comprehensive search was performed by 2 authors (FR and SC) according to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines () [ ] for all papers published up to December 31, 2022, in 2 databases (PubMed/Medline and Embase databases). The search was conducted using a combination of various keywords from 3 categories. The first category included keywords related to different databases potentially used: SNDS, SNIIRAM, PMSI, CépiDc, electronic health record (EHR; Medical Subject Headings [MeSH] term), clinical data warehouse, claims database, registry (MeSH term), cohorts, and analyses (MeSH term). The second category of keywords was related to linkage: data linkage (MeSH term) and medical record linkage (MeSH term). The last category focused on the localization of the study and included 1 keyword: France (MeSH term). Keywords were organized using the following approaches: (1) keywords within 1 category were lined using the OR operator (eg, SNDS OR PMSI), and (2) keywords across different categories were connected using the AND operator (eg, SNDS AND data linkage).
To be eligible for inclusion, studies had to be published in English or French, be human studies, be peer-reviewed papers, focus on health products (drugs or medical devices) or care trajectories, and involve the use of linked French data defined as the linking of 2 or more French data collections at the patient level ().
Only studies based on the use of indirect identifiers were included (ie, without a unique personal identifier available to easily link the databases—the NIR for French databases). Studies on data linkage between the French MADs making up the SNDS were not identified (eg, between SNIIRAM and the PMSI). A manual review of the reference lists of all selected papers was performed to identify any other relevant studies. A full-text search was performed when it was not possible to determine from the abstract whether linked data were used.
|Paper type||Primary research studies||Editorials, letters, commentaries, book chapters, studies where the full text of the publication was not available|
|Intervention||Indirect data linkage||Direct data linkage (ie, unique personal identifier available to easily link the databases—the NIRa for French databases); studies on data linkage between the French MADsb making up the SNDSc not identified (eg, between SNIIRAMd and the PMSIe)|
|Outcome||Studies on health product (drugs or medical devices) use or care trajectories of patients||N/Af|
|Time frame||Up to December 31, 2022||N/A|
aNIR: numéro d’inscription au repertoire.
bMAD: medicoadministrative database.
cSNDS: Système National des Données de Santé.
dSNIIRAM: Système National d’Information Inter-Régimes de l’Assurance Maladie.
ePMSI: Programme de Médicalisation des Systèmes d’Information.
fN/A: not applicable.
Data Extraction and Analysis
The full text of all the eligible papers was screened, and the following criteria were collected: date of publication, research area (type of health products, care trajectories), type of database linked (MAD, registry [set of disease events occurring in a well-characterized population over a given period of time] , cohort [group of individuals sharing a certain number of features and who are followed longitudinally, at the individual level, according to a pre-established protocol] [ ], other databases), number of patients in the different databases, and linkage method (deterministic, probabilistic, or alternative approaches) [ ]. Data linkage quality was assessed using analysis quality indicators and adherence to the Bohensky framework for evaluating data linkage studies [ ] on selected studies published since 2011 (ie, the date of publication of this guide). The Bohensky tool is a checklist for reviewing the specific sources of bias in interpreting the findings of clinical research studies based on data linkage, including the data sets being used, the linkage variables and process, and an assessment of the quality of the linked data sets [ ].
Literature Search Findings
The systematic literature search identified 1155 papers. Screening against the inclusion and exclusion criteria retrieved 39 (3.4%) papers using health database linkage, of which 16 (41.0%) focused on health care product use and care trajectories for the final analysis (). Most papers were recent, with over 87% (n=14) of these papers published since 2017 ( [ - ]). In 1 (6.3%) case included in this review, the publication corresponded to a study protocol in which the data linkage was planned with varying degrees of accuracy [ ]. Data linkage was performed at the national level (ie, for all patients with a disease or who received therapy in France) in 7 (43.8%) cases [ - ] or at the local level in 9 (56.2%) studies [ - ]. The number of patients included in the different databases (medicoadministrative, registries, cohorts, etc) varied greatly, ranging from 713 patients [ ] to 75,000 patients [ , ]. The number of patients resulting from data linkage was also variable, ranging from 210 patients [ ] to 31,000 linked patients [ ].
The health products covered by the studies included drugs (anticoagulants or antiplatelet agents [, - ], 5-alpha-reductase inhibitors [ , ], immunosuppressive treatments [ , ], antihypertensive drugs [ , ], and statins [ ]) and medical devices (transcatheter aortic valve implantation [ ]). The diseases studied were mainly chronic diseases (chronic kidney disease [ - , ]; cancer [ , , ]; cardiovascular diseases, eg, arterial hypertension [ ], aortic stenosis [ ], stroke [ ]; diabetes [ ]) and infections [ ].
Although data linkage increases the information available on each patient, the objectives of the data linkage were multiples: to estimate the risk of adverse drug reactions (ADRs) (n=6, 37.5%) [, - , ], to reconstruct the patient’s care trajectory (n=5, 31.3%) [ - , , ], to describe therapeutic uses (n=2, 12.5%) [ , ], to evaluate the benefits of treatments (n=2, 12.5%) [ , ], and to evaluate treatment adherence (n=1, 6.3%; see ) [ ].
Type of Database Linkage
All (n=15, 93.7%) but 1 (6.3%) study  reported indirect linkage using French claims databases (SNIIRAM, the PMSI, or the SNDS) at the national or the regional level. Registries are the most frequently linked databases with French claims data, with 6 (37.5%) studies based on the Renal Epidemiology and Information Network (REIN) registry [ - ]. The REIN registry records information about all patients with end-stage renal disease who start renal replacement therapy (dialysis or pre-emptive kidney transplantation) in France [ ]. It includes data on patient and center identification, primary kidney disease, initial clinical characteristics, comorbidities, and management of end-stage renal disease, but it does not contain data on health care consumption. In addition, 2 (12.5%) studies involved registries of cardiological procedures [ , ], and 1 (6.3%) study involved cancer registries [ ].
Regarding cohort databases, Mechtouff et al  linked the AVC69 cohort (patients with suspected acute stroke admitted to an emergency department or stroke unit in the Rhône area) and a MAD to assess the use of and adherence to secondary prevention drugs 3 and 6 years after a transient ischemic attack or ischemic stroke. Ad hoc emergency hospital clinical databases from 5 regions in France (Angers, Brest, Grenoble, Nantes, and Rennes) were used to study the risk of major bleeding with anticoagulant and antiplatelet agents [ - ].
Finally, 3 (18.8%) studies focused on outpatient data. The results of biological explorations (Gleason score on prostate biopsy sample) from pathological anatomy laboratories in Brittany were linked with a MAD (SNIIRAM) in 2 (12.5%) studies in order to assess the risk of prostate cancer in patients treated with 5-alpha-reductase inhibitors for symptomatic benign prostatic hyperplasia [, ]. Perlbarg et al [ ] reported on the feasibility of matching a general practice database from ambulatory care with the French health insurance database (SNIIRAM), which represents a novel approach to analyze treatment, comorbidities, medical practices, and care pathways. No studies looked at linking with a hospital data warehouse, a clinical trial database, or patient self-reported databases.
In this review, the deterministic approach [- , , , ] was used in 7 (43.8%) studies and the probabilistic linkage in 4 (25.0%) studies [ , , , ]. In addition, 5 (31.3%) studies used linkage with indirect key identifiers but did not specify whether it was a deterministic or a probabilistic approach [ - ]. They all performed data linkage at the local level. The linkage rate, when specified in the study (11/15, 73.3%, studies), was mainly from 80% to 90% [ , , , , ]. Only 3/15 (20%) studies achieved a rate of over 90% [ , , ].
The number of key identifiers varied from 3 [, ] to 11 [ ]. The application or software used for the linkage was poorly described. For some studies, manual linkage was probably performed but not clearly described.
Quality of the Linkage
As the publication of Raffray et al  was a study protocol in which the results of data linkage were not specified, the quality of the data linkage was not assessed.
Analysis of the characteristics of matched versus unmatched records to assess the completeness of the linkage or to identify a potential selection bias (ie, a specific population might be missed by the algorithm) was used in 26.7% (4/15) of the studies [, , , ]. Didier et al [ ] used comparisons of survival curves between merged populations, registries, and the SNDS to assess the quality of the linkage results. Another method to determine whether a pair is a true match was to check that the comorbidity recorded in database 1 was also recorded in database 2 for all linked patients. This method was used by Raffray et al [ ] to compare diabetes in matched patients. The sensitivity of linkage (ie, the proportion of truly matched records detected), specificity (ie, the proportion of truly unmatched records detected), and the positive predictive value and negative predictive value were reported in only 1 (6.7%) of 15 studies at 99.9%, 97.9%, 99.9%, and 96.9%, respectively [ ].
Finally, adherence to the Bohensky framework for evaluating data linkage studies  is presented in . This analysis showed that the description of the source databases for the linkage was always performed but that the completion rate and accuracy of the variables to be linked were not systematically described.
|Framework item and description||Studies|
|Perlbarg et al ||Béchade et al ||Hogan et al ||Sitruk et al ||Mechtouff et al ||Scailteux et al ||Lesaine et al ||Ferrerira et al ||Bouget et al [- ]||Didier et al ||Raffray et al [, ]|
|1. Completeness of source databases|
|A description of the data sources to be used in the study should be included.||Given||Given||Given||Given||Given||Given||Given||Given||Given||Given||Given|
|The number of eligible records obtained from each data set and the reasons for differences should be reported.||Given||Given||Given||Not performed||Given||Given||Given||Not performed||Given||Given||Given|
|2. Accuracy of data sources|
|Variables selected by researchers for linkage and analysis should be reported.||Given||Given||Given||Given||Given||Given||Given||Given||Given||Given||Given|
|The completion rate and accuracy of variables to be linked should be presented.||Given||Not performed||Not performed||Not performed||Not performed||Not performed||Given||Not performed||Given||Given||Given|
|Coding practices and the use of standardized definitions should be stated, if used.||Given||Given||Given||Given||Given||Given||Given||Given||Given||Given||Given|
|3.Linkage methodology and technology|
|A measure of the validity of the linked data sets (false-positive and false-negative rates, if available) should be given.||Not performed||Not performed||Not performed||Not performed||Not performed||Not performed||Given||Not performed||Not performed||Given||Given|
|An analysis of potential sources of bias among nonlinked cases should be reported.||Given||Not performed||Not performed||Not performed||Not performed||Given||Given||Not performed||Given||Given||Given|
|The denominator used to derive linkage rates and the justification for this should be reported.||Given||Not performed||Not performed||Not performed||Not performed||Given||Given||Not performed||Not performed||Given||Given|
|A description of the data linkage methods (ie, deterministic or probabilistic) with a justification for these should be reported.||Given||Not performed||Given||Given||Not performed||Given||Given||Given||Not performed||Given||Given|
Health data linkage is a powerful research resource that is being increasingly developing worldwide but with national disparities [, ]. This review aimed to share research publications on indirect health data linkage focusing on health product use and care trajectories in a country with a centralized health data system. As the French claims database is 1 of the largest in the world, covering almost the whole population (99%) from birth to death, irrespective of provider, socioeconomic status, or retirement [ ], linkage with other data offers great research potential to study the safety and effectiveness of drugs in routine care [ ]. Nevertheless, only 16 studies describing their indirect record linkage methodology were included in our review. They were mainly published since 2017, which corresponds to the start of facilitated access to health data, which was also illustrated with the increase in accepted health data access projects after the reform [ ]. This result suggests that the recent increase in data linkage in France is encouraging and needs to be confirmed. This review also highlights that indirect linkage mainly actually involves claims databases and disease-specific or population-based registries. However, many others database, including the use of health products, exist, emphasizing the big potential and challenges associated with data linkage. Finally, the quality of data linkage is poorly described: few studies have assessed or specified potential errors associated with data processing before, during, and after linkage. Moreover, 31% of the studies included in this review did not define the method of indirect data linkage (ie, deterministic or probabilistic approaches). Wider appropriation and dissemination of recommendations for the proper use of database linkage seem important to achieve.
Comparing our results on the use of health data linkage to other countries is difficult due to the use of different linkage models [, ]. For example, Young et al [ ] reported over 1200 publications in Australia based on linked data since 2009, without specifying the linkage method. In 2020, Haneef et al [ ] reported in their European survey that France uses advanced data linkage for routine public health activities at national and subnational levels. As direct linkage between French health databases involves the use of a unique identification number (NIR), and because its sharing is highly protected, we focused in this review on indirect linkage. However, several large health cohorts have been directly linked to the SNDS [ ]. For example, CONSTANCES is a large prospective population-based cohort (200,000 persons included), in which an annual direct linkage is performed with 3 French social and health databases (SNIIRAM; Caisse nationale d'assurance vieillesse [CNAV], the national salaried employees retirement fund; and the National Death Registry, CépiDc) [ - ]. Another example of a database directly linked to the SNDS, and not included in its formal framework, is Resid’EPHAD, which allows nursing homes to transmit information about the residents and their health care consumption [ , ].
Claims databases and disease-specific or population-based registries appeared to be the most used database for data linkage on health product use and care trajectories in this review. Nevertheless, other databases are likely to be linked in the future. Health insurance claims databases and EHRs have been identified as the preferred data sources for studying the safety and effectiveness of drugs in routine care . Surprisingly, Lin and Schneeweiss [ ] reported in 2016 in their review only 9 papers on linking electronic medical records to claims data to study drug safety and effectiveness. All the included studies were found to be based in the United States, although this was not part of the inclusion criteria [ ]. The development of hospital data warehouses, defined as a large computerized database that processes all data generated during hospital stays from the hospital information system (eg, medical observations and diagnoses, biology, imaging, prescribing, and drug administration), is on the rise [ , - ]. These databases can quickly become powerful tools because real-time data collection is automated and reflects the clinical practice in hospitals for all inpatients [ ]. In particular, they offer the possibility of supplementing data on health products used in hospitalized patients, which are data rarely captured in other databases [ ]. The pharmaceutical record (the Dossier Pharmaceutique), which is a centralized national electronic database shared by all French community pharmacists on all dispensing, could provide added value to study over-the-counter drug consumption [ , ]. Another possible source of data to be linked to health databases is data generated by new tools, from various fields, such as mobile phones, social networks, eHealth, and connected medical devices. The quantified self-tracking movement, defined as the regular collection of any measurable data about oneself, such as biological, physical, behavioral, or environmental information, offers an additional opportunity to enrich knowledge about health product use and care trajectories [ ]. Patient-reported outcomes are increasingly used in routine medical care, improving patient-clinician communication, clinician knowledge of symptoms, symptom management, patient satisfaction, quality of life, and overall survival [ ]. In this context, Tran et al [ ] proposed the “COOP’ e-cohort,” which aims to build a large community of patients willing to participate in research by contributing to the creation of a large database, passively enriched, at the individual level, through linkage with routinely collected care or medicoadministrative data. All of these real-world data offer the possibility of describing new insights into the use of health products in daily clinical practice [ ].
In this review, we chose the Bohensky framework for evaluating data linkage studies  because of its simplicity of use. Nevertheless, more comprehensive recent guidelines (Guidance for Information about Linking Data sets [GUILD] guidance) [ ] and a checklist (expanded Reporting of Studies Conducted Using Observational Routinely Collected Health Data for Pharmacoepidemiology [RECORD-PE]) [ ] have been developed on the information that needs to be made available about the data linkage process by data providers, data linkers, analysts, and researchers and could have provided further relevant information. The methodology chosen for record linkage has the potential to introduce misclassification into research studies and should be discussed [ ]. Efforts need to be made to improve the health scientific community’s understanding of data linkage methodology and the interpretation of linked data. The recent creation of the Health Data Hub [ ], enabled by the law of July 24, 2019, on the organization and transformation of the French health system [ ] should improve this. Its objective is to enable the implementation of authorized innovative projects using nonnominal data via a state-of-the-art secure technological platform [ ]. Once all the regulatory authorizations relating to data security have been obtained, the Health Data Hub platform would centralize databases from patient registries, cohorts, or electronic medical records and allow the linkage of consolidated databases with SNDS data [ ].
Limitations of this review include our research strategy based on the MeSH terms “medical record linkage” and “data linkage,” which probably excluded some potentially eligible studies that were not referenced with these terms. In addition, Tuppin et al  highlighted a lack of homogeneity in the English terms used to describe French national health insurance databases, which made it more difficult to detect papers. Therefore, the use of indirect data linkage may be underestimated in our review. Moreover, some interesting information on data linkage studies may have escaped a traditional bibliography (ie, reports provided to health authorities or other gray literature) and may thus contribute to the underreporting of indirect data linkage experiences in France. Wider application of RECORD guidelines [ ], which aims to increase the discovery of publications involving the use of routinely collected data, including data linkage, would help overcome this underreporting issue.
Implications for the Future
For French health organizations and regulators, this review suggests that efforts initiated with the creation of the Health Data Hub be continued in order to facilitate the reuse of data, while ensuring respect of data privacy. A European dimension is also planned for sharing with other databases. One of the points raised in this review is also the need to promote and intensify the training in health data science in medical and pharmaceutical universities in collaboration with data scientists.
This review highlighted the growing interest in health data linkage on health product use and care trajectories in France. Nevertheless, regulatory, technical, and human constraints remain major obstacles to their deployment. The volume, variety, and validity of the data represent a real challenge, and advanced expertise and skills in statistical analysis and artificial intelligence are required to treat these big data.
Conflicts of Interest
PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines.PDF File (Adobe PDF File), 1265 KB
Description of included studies (N=16).DOCX File , 29 KB
- Young A, Flack F. Recent trends in the use of linked data in Australia. Aust Health Rev 2018;42(5):584. [CrossRef]
- Harron K, Dibben C, Boyd J, Hjern A, Azimaee M, Barreto ML, et al. Challenges in administrative data linkage for research. Big Data Soc 2017 Dec 05;4(2):2053951717745678 [FREE Full text] [CrossRef] [Medline]
- Holman CDJ, Bass JA, Rosman DL, Smith MB, Semmens JB, Glasson EJ, et al. A decade of data linkage in Western Australia: strategic design, applications and benefits of the WA data linkage system. Aust Health Rev 2008 Nov;32(4):766-777. [CrossRef] [Medline]
- Bohensky MA, Jolley D, Sundararajan V, Evans S, Pilcher DV, Scott I, et al. Data linkage: a powerful research tool with potential problems. BMC Health Serv Res 2010 Dec 22;10:346 [FREE Full text] [CrossRef] [Medline]
- Holman C, Bass A, Rouse I, Hobbs MS. Population-based linkage of health records in Western Australia: development of a health services research linked database. Aust N Z J Public Health 1999 Oct;23(5):453-459 [FREE Full text] [CrossRef] [Medline]
- Boyd JH, Randall SM, Ferrante AM, Bauer JK, McInneny K, Brown AP, et al. Accuracy and completeness of patient pathways--the benefits of national data linkage in Australia. BMC Health Serv Res 2015 Aug 08;15(1):312 [FREE Full text] [CrossRef] [Medline]
- Tew M, Dalziel KM, Petrie DJ, Clarke PM. Growth of linked hospital data use in Australia: a systematic review. Aust Health Rev 2017;41(4):394. [CrossRef]
- Chamberlayne R, Green B, Barer ML, Hertzman C, Lawrence WJ, Sheps SB. Creating a population-based linked health database: a new resource for health services research. Can J Public Health 1998 Jul 1;89(4):270-273. [CrossRef]
- Potosky AL, Riley GF, Lubitz JD, Mentnech RM, Kessler LG. Potential for cancer related health services research using a linked Medicare-tumor registry database. Med Care 1993;31(8):749-756. [CrossRef]
- Kuiper JG, Bakker M, Penning-van Beest FJ, Herings RM. Existing data sources for clinical epidemiology: the PHARMO Database Network. Clin Epidemiol 2020 Apr;12:415-422. [CrossRef]
- Kendrick S, Clarke J. The Scottish Record Linkage System. Health Bull (Edinb) 1993;51:72-79. [CrossRef]
- Haneef R, Delnord M, Vernay M, Bauchet E, Gaidelyte R, Van Oyen H, et al. Innovative use of data sources: a cross-sectional study of data linkage and artificial intelligence practices across European countries. Arch Public Health 2020 Jun 10;78(1):55 [FREE Full text] [CrossRef] [Medline]
- Tuppin P, Rudant J, Constantinou P, Gastaldi-Ménager C, Rachas A, de Roquefeuil L, et al. Value of a national administrative database to guide public decisions: from the système national d'information interrégimes de l'Assurance Maladie (SNIIRAM) to the système national des données de santé (SNDS) in France. Rev Epidemiol Sante Publique 2017 Oct;65(Suppl 4):S149-S167. [CrossRef] [Medline]
- Bezin J, Duong M, Lassalle R, Droz C, Pariente A, Blin P, et al. The national healthcare system claims databases in France, SNIIRAM and EGB: powerful tools for pharmacoepidemiology. Pharmacoepidemiol Drug Saf 2017 Aug 24;26(8):954-962. [CrossRef] [Medline]
- Casarotto E, Noize P, Gouverneur A, Berdaï D, Bégaud B, Levy-Bachelot L, et al. Overview of French databases available for studying anticancer drugs in real-life setting. Fundam Clin Pharmacol 2021 Feb 17;35(1):76-85. [CrossRef] [Medline]
- Scailteux L, Droitcourt C, Balusson F, Nowak E, Kerbrat S, Dupuy A, et al. French administrative health care database (SNDS): the value of its enrichment. Therapie 2019 Apr;74(2):215-223 [FREE Full text] [CrossRef] [Medline]
- Suissa S, Garbe E. Primer: administrative health databases in observational studies of drug effects--advantages and disadvantages. Nat Clin Pract Rheumatol 2007 Dec;3(12):725-732. [CrossRef] [Medline]
- Lapeyre-Mestre M. The challenges of pharmacoepidemiology of orphan drugs in rare diseases. Therapie 2020 Apr;75(2):215-220. [CrossRef] [Medline]
- Doidge JC, Harron K. Demystifying probabilistic linkage: common myths and misconceptions. Int J Popul Data Sci 2018 Jan 10;3(1):410-417 [FREE Full text] [CrossRef] [Medline]
- Looten V, Simon M. Impact analysis of the policy for access of administrative data in France: a before-after study. Stud Health Technol Inform 2020;270:1133-1137. [CrossRef]
- Combes S, Bacry E, Fontbonne C. [Health Data Hub in France, use cases in oncology and radiation oncology]. Cancer Radiother 2020 Oct;24(6-7):762-767. [CrossRef] [Medline]
- Gilbert R, Lafferty R, Hagger-Johnson G, Harron K, Zhang LC, Smith P, et al. GUILD: GUidance for Information about Linking Data sets. J Public Health (Oxf) 2018 Mar 01;40(1):191-198 [FREE Full text] [CrossRef] [Medline]
- Dusetzina S, Tyree S, Meyer AM, Meyer A, Green L, Carpenter WR. Linking data for health services research: a framework and instructional guide. Agency for Healthcare Research and Quality, Rockville (MD). 2014 Sep. URL: https://www.ncbi.nlm.nih.gov/books/NBK253313/ [accessed 2023-10-04]
- Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gøtzsche PC, Ioannidis JPA, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate healthcare interventions: explanation and elaboration. BMJ 2009 Jul 21;339(jul21 1):b2700-b2700 [FREE Full text] [CrossRef] [Medline]
- Palmieri L, Veronesi G, Corrao G, Traversa G, Ferrario MM, Nicoletti G, et al. Cardiovascular diseases monitoring: lessons from population-based registries to address future opportunities and challenges in Europe. Arch Public Health 2018 Jun 28;76(1):31 [FREE Full text] [CrossRef] [Medline]
- Spira A. [Large health cohorts in France]. Bull Acad Natl Med 2013 Feb;197(2):315-328. [Medline]
- Didier R, Gouysse M, Eltchaninoff H, Le Breton H, Commeau P, Cayla G, et al. Successful linkage of French large-scale national registry populations to national reimbursement data: improved data completeness and minimized loss to follow-up. Arch Cardiovasc Dis 2020 Aug;113(8-9):534-541 [FREE Full text] [CrossRef] [Medline]
- Raffray M, Bayat S, Campéon A, Laude L, Vigneau C. The pre-dialysis care trajectory of chronic kidney disease patients and the start of dialysis in emergency: a mixed method study protocol. Int J Environ Res Public Health 2019 Dec 09;16(24):5010 [FREE Full text] [CrossRef] [Medline]
- Raffray M, Bayat S, Lassalle M, Couchoud C. Linking disease registries and nationwide healthcare administrative databases: the French renal epidemiology and information network (REIN) insight. BMC Nephrol 2020 Jan 28;21(1):25 [FREE Full text] [CrossRef] [Medline]
- Raffray M, Vigneau C, Couchoud C, Bayat S. Predialysis care trajectories of patients with ESKD starting dialysis in emergency in France. Kidney Int Rep 2021 Jan;6(1):156-167 [FREE Full text] [CrossRef] [Medline]
- Ferreira J, Couchoud C, Gregson J, Tiple A, Glowacki F, London G, et al. Angiotensin-converting enzyme inhibitors/angiotensin receptor blockers, β-blockers or both in incident end-stage renal disease patients without cardiovascular disease: a propensity-matched longitudinal cohort study. Nephrol Dial Transplant 2019 Jul 01;34(7):1216-1222. [CrossRef] [Medline]
- Sitruk L, Couchoud C, Hourmant M, Tuppin P, Macher M, Legeai C. [Description of immunosuppressive maintenance treatments post kidney transplant through the National System of Health Insurance]. Nephrol Ther 2018 Dec;14(7):523-530. [CrossRef] [Medline]
- Hogan J, Pietrement C, Sellier-Leclerc A, Louillet F, Salomon R, Macher M, et al. Infection-related hospitalizations after kidney transplantation in children: incidence, risk factors, and cost. Pediatr Nephrol 2017 Dec 25;32(12):2331-2341. [CrossRef] [Medline]
- Mechtouff L, Haesebaert J, Viprey M, Tainturier V, Termoz A, Porthault-Chatard S, et al. Secondary prevention three and six years after stroke using the French National Insurance Healthcare System Database. Eur Neurol 2018 May 14;79(5-6):272-280. [CrossRef] [Medline]
- Béchade C, Dejardin O, Bara S, Bouvier V, Guizard A, De Mil R, et al. Incidence and characteristics of chronic renal replacement therapy in patients with cancer: data from kidney and cancer registries in Basse-Normandie. J Nephrol 2018 Feb 4;31(1):111-118. [CrossRef] [Medline]
- Bouget J, Balusson F, Viglino D, Roy P, Lacut K, Pavageau L, et al. Major bleeding risk and mortality associated with antiplatelet drugs in real-world clinical practice. A prospective cohort study. PLoS One 2020 Aug 7;15(8):e0237022 [FREE Full text] [CrossRef] [Medline]
- Bouget J, Balusson F, Maignan M, Pavageau L, Roy P, Lacut K, et al. Major bleeding risk associated with oral anticoagulant in real clinical practice. A multicentre 3-year period population-based prospective cohort study. Br J Clin Pharmacol 2020 Dec;86(12):2519-2529 [FREE Full text] [CrossRef] [Medline]
- Bouget J, Balusson F, Scailteux L, Maignan M, Roy P, L'her E, et al. Major bleeding with antithrombotic agents: a 2012-2015 study using the French nationwide Health Insurance database linked to emergency department records within five areas - rationale and design of SACHA study. Fundam Clin Pharmacol 2019 Aug 11;33(4):443-462 [FREE Full text] [CrossRef] [Medline]
- Scailteux L, Balusson F, Vincendeau S, Rioux-Leclercq N, Nowak E. Rationale and design of the CANARI study: a case-control study investigating the association between prostate cancer and 5-alpha-reductase inhibitors for symptomatic benign prostate hypertrophy by linking SNIIRAM and pathology laboratories in a specific region in France. Fundam Clin Pharmacol 2018 Feb 14;32(1):120-129. [CrossRef] [Medline]
- Perlbarg J, Allonier C, Boisnault P. Feasibility and practical value of statistical matching of a general practice database and a health insurance database applied to diabetes and hypertension. Sante Publique Vandoeuvre--Nancy Fr 2014;26:355-363.
- Scailteux L, Rioux-Leclercq N, Vincendeau S, Balusson F, Nowak E, Oger E, Network of Pathologists in Brittany. Use of 5α-reductase inhibitors for benign prostate hypertrophy and risk of high grade prostate cancer: a French population-based study. BJU Int 2019 Feb 11;123(2):293-299 [FREE Full text] [CrossRef] [Medline]
- Lesaine E, Belhamri N, Legrand J, Domecq S, Coste P, Lacroix A, ACIRA investigators. [Linking Interventional Cardiology clinical registry data with French hospital administrative data: development and validation of deterministic record linkage]. Rev Epidemiol Sante Publique 2021 Apr;69(2):78-87. [CrossRef] [Medline]
- Couchoud C, Stengel B, Landais P, Aldigier JC, de Cornelissen F, Dabot C, et al. The renal epidemiology and information network (REIN): a new registry for end-stage renal disease in France. Nephrol Dial Transplant 2006 Feb;21(2):411-418. [CrossRef] [Medline]
- Pacurariu A, Plueschke K, McGettigan P, Morales DR, Slattery J, Vogl D, et al. Electronic healthcare databases in Europe: descriptive analysis of characteristics and potential for use in medicines regulation. BMJ Open 2018 Sep 05;8(9):e023090 [FREE Full text] [CrossRef] [Medline]
- Goldberg M, Carton M, Descatha A, Leclerc A, Roquelaure Y, Santin G, CONSTANCES team. CONSTANCES: a general prospective population-based cohort for occupational and environmental epidemiology: cohort profile. Occup Environ Med 2017 Jan 24;74(1):66-71 [FREE Full text] [CrossRef] [Medline]
- Zins M, Bonenfant S, Carton M, Coeuret-Pellicer M, Guéguen A, Gourmelen J, et al. The CONSTANCES cohort: an open epidemiological laboratory. BMC Public Health 2010 Aug 12;10(1):479 [FREE Full text] [CrossRef] [Medline]
- Zins M, Goldberg M, CONSTANCES team. The French CONSTANCES population-based cohort: design, inclusion and follow-up. Eur J Epidemiol 2015 Dec 31;30(12):1317-1328 [FREE Full text] [CrossRef] [Medline]
- Atramont A, Bonnet-Zamponi D, Bourdel-Marchasson I, Tangre I, Fagot-Campagna A, Tuppin P. Health status and drug use 1 year before and 1 year after skilled nursing home admission during the first quarter of 2013 in France: a study based on the French National Health Insurance Information System. Eur J Clin Pharmacol 2018 Jan;74(1):109-118. [CrossRef] [Medline]
- Lin K, Schneeweiss S. Considerations for the analysis of longitudinal electronic health records linked to claims data to study the effectiveness and safety of drugs. Clin Pharmacol Ther 2016 Aug 12;100(2):147-159. [CrossRef] [Medline]
- Cancé C, Madiot P, Lenne C. Cohort creation and visualization using graph model in the PREDIMED health data warehouse. Stud Health Technol Inform 2020;270:108-112. [CrossRef]
- Cuggia M, Garcelon N, Campillo-Gimenez B. Roogle: an information retrieval engine for clinical data warehouse. Stud Health Technol Inform 2011;169:588.
- Daniel C, Serre P, Orlova N, Bréant S, Paris N, Griffon N. Initializing a hospital-wide data quality program. The AP-HP experience. Comput Methods Programs Biomed 2019 Nov;181:104804. [CrossRef] [Medline]
- Adenot I. [The pharmaceutical record]. Ann Pharm Fr 2007 Sep;65(5):325-330. [CrossRef] [Medline]
- Gillibert A, Griffon N, Schuers M, Hardy K, Elmerini A, Letord C, et al. Impact on medical practice of accessing pharmaceutical records. Int J Med Inform 2019 Jan;121:58-63. [CrossRef] [Medline]
- Swan M. Emerging patient-driven health care models: an examination of health social networks, consumer personalized medicine and quantified self-tracking. Int J Environ Res Public Health 2009 Feb 05;6(2):492-525 [FREE Full text] [CrossRef] [Medline]
- Basch E, Barbera L, Kerrigan CL, Velikova G. Implementation of patient-reported outcomes in routine medical care. Am Soc Clin Oncol Educ Book 2018 May(38):122-134. [CrossRef]
- Tran V, Ravaud P. COllaborative open platform e-cohorts for research acceleration in trials and epidemiology. J Clin Epidemiol 2020 Aug;124:139-148. [CrossRef] [Medline]
- Cave A, Kurz X, Arlett P. Real-world data for regulatory decision making: challenges and possible solutions for Europe. Clin Pharmacol Ther 2019 Jul 10;106(1):36-39 [FREE Full text] [CrossRef] [Medline]
- Pratt NL, Mack CD, Meyer AM, Davis KJ, Hammill BG, Hampp C, et al. Data linkage in pharmacoepidemiology: a call for rigorous evaluation and reporting. Pharmacoepidemiol Drug Saf 2020 Jan 17;29(1):9-17. [CrossRef] [Medline]
- Padmanabhan S, Carty L, Cameron E, Ghosh RE, Williams R, Strongman H. Approach to record linkage of primary care data from Clinical Practice Research Datalink to other health-related patient data: overview and implications. Eur J Epidemiol 2019 Jan 15;34(1):91-99 [FREE Full text] [CrossRef] [Medline]
- Health data hub. Health Data Hub. URL: https://www.health-data-hub.fr/ [accessed 2023-04-10]
- Benchimol EI, Smeeth L, Guttmann A, Harron K, Moher D, Petersen I, RECORD Working Committee. The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) statement. PLoS Med 2015 Oct 6;12(10):e1001885 [FREE Full text] [CrossRef] [Medline]
|ADR: adverse drug reaction|
|CépiDc: Center d’Epidémiologie sur les causes médicales de Décès|
|EHR: electronic health record|
|MAD: medicoadministrative database|
|MeSH: Medical Subject Headings|
|NIR: numéro d’inscription au repertoire|
|PMSI: Programme de Médicalisation des Systèmes d’Information|
|REIN: Renal Epidemiology and Information Network|
|SNDS: Système National des Données de Santé|
|SNIIRAM: Système National d’Information Inter-Régimes de l’Assurance Maladie|
Edited by A Mavragani; submitted 18.07.22; peer-reviewed by L Kanbar, T Karen; comments to author 15.02.23; revised version received 22.03.23; accepted 22.03.23; published 18.05.23Copyright
©Florence Ranchon, Sébastien Chanoine, Sophie Lambert-Lacroix, Jean-Luc Bosson, Alexandre Moreau-Gaudry, Pierrick Bedouch. Originally published in the Journal of Medical Internet Research (https://www.jmir.org), 18.05.2023.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on https://www.jmir.org/, as well as this copyright and license information must be included.