Use cases of registry‐based randomized controlled trials—A review of the registries' contributions and constraints Authors: Nadine Kubesch, Sneha Gaitonde, Uarda Petriti, Elisabeth Bakker, Swati Basu, Laura Ellen Birks, Elodie Aubrun, Sieta T de Vries, Rahel Schneider Published: 2024 ABSTRACT AbstractRegistry‐based randomized controlled trials (RRCTs) can combine the advantages of registries with those of randomization. This review aimed to expand the current knowledge on RRCT utilization and implementation by providing a comprehensive overview of RRCT use cases. A targeted literature search was conducted through July 2023 to identify articles on RRCTs. Information regarding the RRCT characteristics, their utilization, and the registries' contributions and the constraints faced was extracted. Descriptive statistics were used. We identified 102 RRCTs in 110 publications. RRCTs were mostly performed for the assessment of medical devices or surgical/clinical procedures (n = 45), followed by drugs (n = 30). More than half of the RRCTs were conducted in the Nordic countries (n = 58) and the most used registry types were health service registries/administrative health data (n = 63), followed by disease registries (n = 46). Approximately half of the RRCTs (n = 53) utilized additional data sources aside from registry data. The contribution of a registry to the RRCT was mostly for data collection and study follow‐up (n = 90–92), followed by patient recruitment (n = 56–61), and randomization (n = 28–38), with varying levels of transparency in reporting. We collated author‐reported constraints related to the used registries into four overarching themes, that is, data availability and completeness, data quality, representativeness, and registry infrastructure and accessibility. This review shows that RRCTs are already used in different domains and geographic regions. Guidelines on structured and transparent reporting of RRCT methods and the optimal use are, however, needed to inform decision‐making by health authorities and to reach their full potential. INTRODUCTION Randomized controlled trials (RCTs) are considered the gold standard for obtaining high‐quality clinical evidence to demonstrate therapeutic efficacy and improve clinical care. However, RCTs often face challenges including scientific and operational complexity, waning site and patient participation, burdensome regulatory and institutional requirements, and high costs. 1 Moreover, RCTs enroll a highly selected patient sample, which limits the generalizability of study results to the real‐world patient population. 1 , 2 Due to the challenges of RCTs, researchers have explored alternative methodological approaches to obtain high‐quality clinical evidence that is both affordable and generalizable. 3 , 4 , 5 One option is the registry‐based randomized controlled trial (RRCT). RRCTs fall under the umbrella of “pragmatic trials” and have the potential to combine the advantages of registries (low expense, external validity) with the advantages of randomization (internal validity). There are variable definitions of what constitutes an RRCT, depending on the methods used to recruit patients and the type of registry employed to collect data. 3 , 6 , 7 , 8 Registry data can be utilized in various ways in RRCTs, for example, for identification and recruitment of patients, collecting patients' baseline characteristics including medical history and data related to the intervention, identification of endpoints for outcomes ascertainment post‐intervention, and guidance of trial design. 9 Recently, a review study identified six author‐reported RRCT‐specific strengths, that is, the ability to compare individuals in the registry who were or were not included in the trial, complete and high‐quality trial information, existing data for feasibility estimations, large samples, enhanced efficiency, and enhanced generalizability. 10 However, despite these strengths, the European Medicines Agency (EMA) emphasizes in their guideline on registry‐based studies that questions remain regarding the validity and relevance of RRCTs in scientific research. 9 Echoing the concerns of the EMA, the main author‐reported limitations of RRCTs in the above‐mentioned review study were incomplete or insufficient registry data, insufficient uniformity of measurement processes, insufficient generalizability, and inaccessible data due to privacy/consent issues. 10 Other review studies have shown that RRCTs are mostly conducted in the disease area of cardiology, 3 , 7 , 11 that the included RRCTs were mostly comparative effectiveness studies, 3 , 7 and, indeed, generally have a large sample size. 3 , 7 , 11 In the current review study, we aimed to expand the knowledge about RRCTs by providing a comprehensive overview of RRCT characteristics and an in‐depth exploration of their utilization, and the registries' contributions and the constraints faced in facilitating RRCT implementation. MATERIALS AND METHODS Study design and data sources A review was conducted using a comprehensive search strategy in PubMed to obtain articles on RRCTs (e.g., RRCT study publications, study protocol publications, statistical analysis plan publications) published in the English language through 12 July 2023 (see Appendix S1 for search terms). The title and abstract of each identified article were screened, and all articles with relevant content were screened in full by one author (NK). In addition, a manual search was performed to follow up on allusions to registries or RRCTs mentioned within the identified literature and to find additional relevant articles from the publications' list of references. Once selected, further details (e.g., on methodology) of the RRCTs were extracted from clinicaltrials.gov, if relevant. Outcomes The outcomes of this study were (1) characteristics of the RRCT and used registries, that is, study domain, geographical scope, total study size (if only the protocol was available, planned study size was reported), registry type, and whether or not the RRCT used a mixed approach to data collection (i.e., collection of part or entirety of data outside the registry), (2) contributions of the registry to the RRCT based on the categories mentioned in the EMA guideline on registry‐based studies, 9 that is, sample size calculation, patient recruitment (e.g., to identify patients meeting inclusion and exclusion criteria), randomization allocation, endpoint identification (i.e., the process of selecting and defining the endpoints that will be used to evaluate the impact of the intervention being studied), and data collection and study follow‐up, and (3) author‐reported constraints related to the use of registries in the RRCT. Analyses Descriptive analyses were conducted, and results were reported as frequencies and percentages. For the characteristics of the RRCTs and used registries, an inductive qualitative approach was used to identify the study domains and registry types. Two authors (NK and UP) independently classified each of the reviewed RRCTs into categories and identified and classified the contributions of each of the registries mentioned in the publications. In case of discrepancies between the two authors, cases were discussed among coauthors until consensus was reached. Regarding contributions of the registry to the RRCT, when trials could not be clearly allocated to one specific category (e.g., due to ambiguous reporting), we reported ranges to account for uncertainty. The lowest value of each range represents the number of RRCTs that were certainly assigned to the associated category, while the highest value includes the number of RRCTs that were considered likely but not certainly associated with the respective category. Using an inductive qualitative approach, one of five authors (NK, UP, STdV, RS, or SG) identified author‐reported registry‐related constraints and collated them into overarching themes, which were subsequently verified by another author among the group. Microsoft 365 Excel version 2312 (build 17126.20190) (Microsoft Corporation, Redmond, USA) was used for the analyses and for the graphical presentation of the results. RESULTS Of 819 hits from PubMed and the manual search, 110 publications reporting on 102 RRCTs were included in the review. The dataset is accessible in Appendix S2. Information for 11 of the trials was retrieved from the clinicaltrials.gov website due to either missing 12 , 13 , 14 or incomplete 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 information regarding trial characteristics in the publications. RRCT characteristics The characteristics of the 102 RRCTs are described in Table 1. Four main study domains were identified, that is, medical device or surgical/clinical procedures (n = 45), drugs (n = 30), health service/public health (n = 23), and occupational health (n = 4). Most RRCTs were conducted in the Nordic countries (Denmark, Norway, Sweden, Finland, and Iceland) (n = 58), followed by the United States (n = 18) and Canada (n = 9). No RRCT was conducted solely in South or Central America. However, one of the six multinational RRCTs involved a total of 16 countries, including Argentina, Brazil, and Panama. 15 Of the RRCTs that mentioned the (planned) study size, most RRCTs (n = 52 of 89) aimed to include over 1500 participants. TABLE 1Characteristics of the registry‐based randomized controlled trials (RRCT).CharacteristicRRCT, n (%) N 102Study domainMedical device or surgical/clinical procedures45 (44.1)Drugs a 30 (29.4)Health service/public health23 (22.5)Occupational health4 (3.9)Geographical scopeNordic countries b , c 58 (56.9)United States of America18 (17.6)Canada9 (8.8)Multinational6 (5.9)The Netherlands2 (2.0)United Kingdom2 (2.0)Australia2 (2.0)Germany2 (2.0)Italy1 (1.0)Republic of Korea1 (1.0)Japan1 (1.0)Total (planned) study size≤50018 (17.6)>500–150019 (18.6)>1500–500031 (30.4)>500021 (20.6)Unknown13 (12.7)Registry type d Health service registries/administrative health data63 (61.8)Disease registries46 (45.1)Surgical/clinical procedure registries21 (20.6)Other16 (15.7)Mixed approach to data collection e 53 (52.0) a Including medical oxygen therapy, specific nutrition regimes in care settings, or vaccine administration. b Denmark, Norway, Sweden, Finland, and/or Iceland. c Seven RRCTs in this group were multinational (i.e., included several Nordic countries). d Each RRCT can contribute to multiple categories (e.g., if several different registries were used). e Defined as collection of part or entirety of data outside the registry. Characteristics of the registry‐based randomized controlled trials (RRCT). Including medical oxygen therapy, specific nutrition regimes in care settings, or vaccine administration. Denmark, Norway, Sweden, Finland, and/or Iceland. Seven RRCTs in this group were multinational (i.e., included several Nordic countries). Each RRCT can contribute to multiple categories (e.g., if several different registries were used). Defined as collection of part or entirety of data outside the registry. Four categories of registry types were created among the included RRCTs based on the registries' nature and aim for data collection: health service registries/administrative health data, disease registries, surgical/clinical procedure registries, and “other.” The category of health service registries/administrative health data comprises databases that collect information on healthcare services provided, focusing on their utilization, resource allocation, and health outcomes. Disease registries aim to capture comprehensive information on specific diseases or medical conditions. The category of surgical/clinical procedure registries encompasses databases that collect information on surgical procedures, medical devices, or clinical interventions. The “other” category mostly comprises registries that originated from the merging of different registry types. The two most common categories of registry types were health service registries/administrative health data (n = 63) and disease registries (n = 46) (Table 1). Surgical/clinical procedure registries (n = 21) and registries falling into the “other” category (n = 16) were less commonly utilized (Table 1). The “other” category included 11 RRCTs using the SWEDEHEART registry, which combines a disease registry, surgical/clinical procedure registries, and immunization registries. 5 , 12 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 , 36 About half of the RRCTs (n = 53) utilized a mixed approach to data collection. Among the reported additional data collection methods for supplementation of registry data were, for example, the administration of questionnaires and collection of patient‐reported and/or provider‐reported information, 17 , 37 , 38 , 39 , 40 , 41 , 42 , 43 , 44 telephone and in‐person interview follow‐ups, 20 , 45 , 46 , 47 and study visits, clinical exams, and other health check‐ups. 38 , 45 , 48 Registry contributions to the RRCTs In most of the RRCTs, the registry was used for data collection and study follow‐up (n = 90–92) (Figure 1), which could include the collection of data at baseline and/or during follow‐up. More than half of the RRCTs (n = 56–61) used the registry for patient identification and/or recruitment. About a third of the RRCTs (n = 28–38) used the registry for participants' randomization allocation (mainly based on integrated tools, such as web‐based randomization modules). Due to a lack of clarity in the reporting, the randomization allocation had the largest width of the range of all contribution categories. The use of the registry for sample size calculation was reported for 7 to 13 RRCTs. None of the RRCTs used the registry data for endpoint identification. Contribution of registries to the registry‐based randomized controlled trials (RRCT). Furthermore, 15 RRCTs 21 , 37 , 49 , 50 , 51 , 52 , 53 , 54 , 55 , 56 , 57 , 58 , 59 , 60 , 61 , 62 used registry data for other purposes, including the administration of the intervention itself, such as sending reminders to participants, 37 , 61 , 62 site identification, 59 intervention feedback, 21 or investigation of the generalizability of the study results. 37 , 49 , 55 , 56 One RRCT used a population‐based register to create an external comparator group. 60 In one use case, the contribution of the registry involved was not specified in the publication and could not be attributed to any of the predefined categories. 63 Constraints regarding the use of registries for RRCT conduct We collated author‐reported constraints into four overarching themes, that is, data availability and completeness, data quality and reliability, representativeness, and registry infrastructure and accessibility. Specific examples are described in Table 2. Data availability and completeness issues of registry data included both the absence of data (i.e., variables not captured in the registry) and the missingness of data (i.e., missing observations among captured data). In addition, timeliness constraints regarding registry data were observed, for example, based on a time lag of data reporting, which impacted outcome assessment. This issue is closely linked with the challenge of an unclear temporal relationship between an intervention and the specific health outcomes reported to the registry as information regarding interventions or outcomes may not populate at the time of the event. 64 TABLE 2Author‐reported registry‐related constraints in registry‐based randomized controlled trials.Constraints related toExamples with a supporting reference a Data availability and completeness Absence of core variables 80 Missingness of data due to the registry context 55 Need for modification of outcome definition due to insufficient/lack of data availability in registry 59 Time lag in reporting 88 Data quality Potential inaccuracy of data 89 Variation of accuracy or quality of coding between clinicians and/or sites 64 Variation of willingness and ability of healthcare providers to enter data in the registry between sites 64 Data inconsistencies due to use of different registries from different countries 36 Potential misclassification of outcome or exposure due to the lack of internal adjudication or validation 90 Potential misclassification of participant eligibility due to lack of (baseline) parameters needed for recruitment 65 Representativeness Incomplete population capture 28 Registry sites may not be representative 16 Registry infrastructure and accessibility Need for different randomization methods across involved countries due to lack of randomization modules within some registries 45 Technical complexity and extra work due to changes in the data system during the study period 43 Additional investment of time and effort due to establishment of data use agreements with registries 67 Collecting outcome data from routine administrative registers is not encouraged due to current norms of research ethics approvals 43 a The references provided are illustrative examples of articles mentioning the specified constraints and do not represent an exhaustive list. Author‐reported registry‐related constraints in registry‐based randomized controlled trials. Absence of core variables 80 Absence of core variables 80 Missingness of data due to the registry context 55 Missingness of data due to the registry context 55 Need for modification of outcome definition due to insufficient/lack of data availability in registry 59 Need for modification of outcome definition due to insufficient/lack of data availability in registry 59 Time lag in reporting 88 Time lag in reporting 88 Potential inaccuracy of data 89 Potential inaccuracy of data 89 Variation of accuracy or quality of coding between clinicians and/or sites 64 Variation of accuracy or quality of coding between clinicians and/or sites 64 Variation of willingness and ability of healthcare providers to enter data in the registry between sites 64 Variation of willingness and ability of healthcare providers to enter data in the registry between sites 64 Data inconsistencies due to use of different registries from different countries 36 Data inconsistencies due to use of different registries from different countries 36 Potential misclassification of outcome or exposure due to the lack of internal adjudication or validation 90 Potential misclassification of outcome or exposure due to the lack of internal adjudication or validation 90 Potential misclassification of participant eligibility due to lack of (baseline) parameters needed for recruitment 65 Potential misclassification of participant eligibility due to lack of (baseline) parameters needed for recruitment 65 Incomplete population capture 28 Incomplete population capture 28 Registry sites may not be representative 16 Registry sites may not be representative 16 Need for different randomization methods across involved countries due to lack of randomization modules within some registries 45 Need for different randomization methods across involved countries due to lack of randomization modules within some registries 45 Technical complexity and extra work due to changes in the data system during the study period 43 Technical complexity and extra work due to changes in the data system during the study period 43 Additional investment of time and effort due to establishment of data use agreements with registries 67 Additional investment of time and effort due to establishment of data use agreements with registries 67 Collecting outcome data from routine administrative registers is not encouraged due to current norms of research ethics approvals 43 Collecting outcome data from routine administrative registers is not encouraged due to current norms of research ethics approvals 43 The references provided are illustrative examples of articles mentioning the specified constraints and do not represent an exhaustive list. Another overarching theme of constraints was related to the data quality (i.e., the extent to which the entered data are reliable, timely, and coherent, thus reflecting the validity of the information). For example, the ICD‐10 coding may vary between clinicians and/or sites according to experience or knowledge of the illness and the quality framework in place or may not reflect all subtypes of a disease or condition. 41 Data reliability concerns stemming from a less rigorous or missing internal adjudication or validation of variables in RRCTs compared to conventional RCTs create concerns regarding misclassification of eligibility or of the outcome. For example, transplant‐ineligible patients may be misclassified as eligible for transplant due to the lack of administrative health data on certain exclusion criteria (e.g., substance abuse 65 ). To mitigate the consequences related to this constraint, some trials implemented external adjudication/validation practices. 31 , 66 Constraints regarding representativeness were related to insufficient generalizability of the included registry sites and the incomplete population capture of the used registries (i.e., the extent to which all necessary individuals who could have been included have actually been included). An example of the latter, which was mentioned by authors of several RRCT publications, was that conservatively treated patients may have been missed because the used registry data were based on hospital admissions. 28 Several authors reported a difference in baseline characteristics between participants who were enrolled into the RRCT and eligible non‐enrolled participants within the registry. As a result, individuals participating in randomized trials may not fully represent the overall real‐world patient population, even in a registry setting. 16 , 41 , 45 , 48 , 54 , 55 , 66 Registry infrastructure and accessibility concerns related to the overall structure, organization, and availability of the registry system, including its technological capabilities, ease of use, and ability to be accessed by relevant stakeholders for data entry and retrieval. If, for example, the registry is not used routinely due to the registry interface being too burdensome to use, usability of data may be limited. In addition, it was noted that the establishment of data use agreements with registries should be considered when designing an RRCT as it may require additional investment of time and effort. 67 Another challenge includes the need for different randomization allocation methods among participating countries depending on the (un)availability of registry‐embedded randomization modules. 45 DISCUSSION This review provides a comprehensive overview of the characteristics of RRCTs, their registries' contributions and author‐reported constraints. Regarding the characteristics, we found that the RRCTs conducted so far have been more commonly conducted in the study domain of medical devices and surgical or clinical procedures than in other domains such as drugs. It is likely that RRCTs play a particularly important role in evaluating clinical questions where funding is difficult to access, which can be the case for the evaluation of medical devices and clinical procedures. 68 Other explanations could be the differences in the approval processes between the study domains and the recognition of the potential value of RRCTs in the specific field. Differences in requirements, classifications, and timelines of the approval processes of medical devices and drugs have for instance been observed. 69 Further studies are needed to investigate such potential explanations. The majority of the RRCTs identified in our study were conducted in the Nordic countries. This has also been shown in previous review studies. 7 , 11 The accessibility of registries and the possibility of linkage to other data sources in these countries may have contributed to this finding. Of the RRCTs conducted in other countries, only two were conducted solely in Asian countries 16 , 60 and only one multinational RRCT included countries from South and Central America. 15 This may indicate, for instance, a potential lack of registries in these regions, a lack of possibilities to link data sources, less reliable health registers, less recognition for the registries' potential value, or differences in data protection laws, in cultural perceptions of health systems or in openness to health data sharing. Various types of registries exist, including disease‐, procedure‐, or pathology‐specific registries, administrative‐, health‐systems‐based, or combined/linked registries, and product registries. 70 We found that health service registries/administrative health data are the registry type most often used in RRCTs (i.e., in 62% of the identified RRCTs). However, it is noteworthy that about half of the RRCTs in our review either could not or did not rely solely on the data available from registries. Additional data collection in conjunction with a registry may add complexity and cost to the RRCT. This shows the importance of an initial feasibility assessment and the inclusion of a fit‐for‐purpose registry. 9 Nevertheless, additional data collection can address several of the identified registry constraints by minimizing the occurrence of missing or inaccurate data and by enhancing the understanding of data validity and integrity. This is corroborated by a systematic review of RRCTs, indicating that the smallest amount of missing data was found in studies that combined different types of registries. 7 In this review, numerous RRCTs used multiple data sources merged into one registry or merged several registries involving a comprehensive data linkage to overcome data limitations. For RRCTs that were not conducted in Nordic countries, challenges to merging registry data included a lack of standardized data and the requirement of complex data pooling. 3 , 43 In general, a team of diverse stakeholders and experts on both RCTs and real‐world data (including registry data) is vital in the set‐up of RRCTs, which includes aligning on core data elements, to maximize their usability and impact. The contribution of registries to RRCTs consisted most frequently of data collection and study follow‐up, followed by patient recruitment and randomization. Recruitment and follow‐up can be two of the most laborious and expensive parts of conducting RCTs. 71 Therefore, these are areas in which registry data use is highly advantageous for researchers. 70 Registries were not used for endpoint identification in any of the RRCTs found in our review. This may partly be attributed to the limited number of RRCTs observed in the field of rare diseases, 13 , 14 , 33 an area in which uncertainties in the choice of endpoints often arise and to which a registry could contribute. 72 In about a third of the RRCTs, registries with integrated randomization tools were used to allocate interventions to the different groups. The methods used in RRCTs to randomize patients or investigational sites, however, were not always transparent in the publications. In the case of 10 RRCTs, it was likely, though not reported with enough clarity whether the registry played a role in the randomization process. 19 , 21 , 22 , 36 , 64 , 73 , 74 , 75 , 76 , 77 , 78 Uncertainty regarding the use of registries was also observed for sample size calculation, patient recruitment, and data collection and study follow‐up in six, 67 five, 18 , 19 , 45 , 64 , 79 and two 15 , 80 RRCTs, respectively. Such uncertainty is in line with the findings of another review, which called out a lack of information and transparency from RRCTs with respect to which datasets have been accessed and for what purposes. 81 Similarly, a large review study found that adequate reporting of critical features of RRCTs is often inadequate in biomedical journals, thereby hampering replicability, bias assessment, interpretation, and applicability of results. 6 Thus, clearly describing the registry contribution within the body of the publication is crucial for allowing transparency of methodology. Structural use of the Consolidated Standards of Reporting Trials (CONSORT) extension for randomized controlled trials conducted using cohorts and routinely collected data (CONSORT‐ROUTINE) 82 could be a way forward. Moreover, it has been suggested that RRCTs should be registered as is the case for RCTs 3 and that guidance is needed for the critical appraisal and synthesis of evidence from RRCTs. 8 As the concept of RRCTs lacks a clear definition, 3 , 6 , 7 , 8 RRCTs are expected to be treated like conventional RCTs by regulators. This means imposing strict rules and regulations that are either not applicable to RRCTs or may be challenging to follow. 83 The EMA provides general guidance on the conduct of registry‐based observational studies. 9 Specific guidance on the conduct of RRCTs has been developed in Sweden. These national guidelines are freely available from the regional node for clinical studies or the regional office of the national registry center. 84 Moreover, a 5‐year project that started in January 2023 aims to establish the value of registry‐based real‐world data in augmenting RCT data and enable the More Effective and ethical Use of Registry data to suppOrt PAtient‐centered regulatory and health technology assessment (HTA) decision‐making (More‐EUROPA). 85 The project seeks in part to develop guidance and a protocol for RRCT conduct that could be used to answer questions posed by regulatory authorities and HTA bodies. How to evaluate and design RRCTs will also be part of the provided training program. Guidance for RRCT conduct should not only consider the potential of RRCTs, but also their possible constraints and information on how to mitigate them. The author‐reported constraints identified in our study were related to (a) data availability and completeness, (b) data quality and reliability, (c) representativeness, and (d) registry infrastructure and accessibility. These are in line with the concerns about registry‐based studies reported in the EMA guideline as well as with constraints found in previous studies. 3 , 10 , 34 , 86 While findings of RRCTs are deemed more generalizable to the real‐world population than those of RCTs, 10 insufficient population capture or included registry sites that are not representative may still compromise external validity. In addition to reported issues regarding representativeness identified in this review, a survey study found that barriers to inhibit patients from joining a registry for RCTs include data safety concerns, potential side effects, and concerns that registry participation would impact current treatment. However, despite these concerns, roughly four of five patients were still willing to consent to have their data collected in a registry for conducting RCTs. 87 Unless directly addressed within the study design, the identified author‐reported constraints could lead to spurious associations between study interventions and outcomes, hindering the decision‐making process. Especially, timeliness of data is a serious concern, since it is a core part of determining the causality of an outcome. 64 To ensure that RRCTs result in reliable and high‐quality clinical evidence, the challenges of data integrity, data quality, and endpoint adjudication need to be addressed by incorporating quality assurance processes into the administration of registries. 3 RRCTs may also have repercussions for registry data that were originally meant to reflect routine clinical practice in a real‐world setting. Due to the nature of RCTs, participants do not necessarily receive the current standard of care during RRCT participation. If these data are recorded within the registry and not specifically flagged as being used for an RCT, subsequent observational studies based on them may come to biased conclusions (e.g., in a study describing drug utilization). Registration of such information is needed so that it can be taken into account in observational studies by, for instance, excluding those participants or conducting sensitivity analyses. The most important strength of the current review was the broad inclusion criteria resulting in a comprehensive overview of the current landscape of RRCTs. The following limitations should be considered when interpreting this review's findings. As there was a variety of specific methodologies used and a lack of literature describing them, the classification of study domains and registry types was done using an inductive qualitative approach. Thus, there is a risk of misclassification. However, this risk was minimized by having two reviewers independently perform the classification. In addition, for the registry contributions, the degree of uncertainty of the classification was reported, as reflected by the ranges introduced in the quantitative data. Moreover, we only considered author‐reported registry‐related constraints within included RRCT publications and provided a summary of the constraints in this review. As there may have been unreported constraints in working with registry data, we did not attempt to quantify them. Additionally, because this review did not follow a systematic review approach but rather aimed to explore the breadth of RRCT applications, some publications may not have been identified in the search. Research that was unpublished or published in a language other than English may have been overlooked. Lastly, although our search strategy was broad, differences in terminology used in reporting RRCTs could have led to some publications not being identified in this review. This could have been the case if terms such as “pragmatic trials” or “registry‐based clinical trials” had not been used in earlier publications. The use of different search terms, data sources and inclusion and exclusion criteria could also explain differences in included RRCTs across other review studies. 7 , 11 In conclusion, this review study found that registries are used in a variety of ways in the RCT setting across many study domains, from drug‐related research to public health research. Registry data were most frequently used for data collection, study follow‐up, and patient recruitment. Yet, in about half of all RRCTs, additional data sources aside from registry data were utilized. To improve the suitability of registries for RRCTs, validation of outcomes and standardized processes to increase data quality and data completeness should be considered. Forthcoming discussions should include the establishment of guidelines on transparent reporting of RRCT methods and the optimal use of RRCTs to inform decision‐making by health authorities. FUNDING INFORMATION This work was supported by Novartis Pharma AG, Basel, Switzerland. EB and STdV have received funding from the European Union's Horizon Europe Research and Innovation Actions under grant no. 101095479 (More‐EUROPA). Views and opinions expressed are however those of the authors only and do not necessarily reflect those of the European Union nor the granting authority. Neither the European Union nor the granting authority can be held responsible for them. CONFLICT OF INTEREST STATEMENT EA, LB, SB, RS, and SG are Novartis employees and EA, LB, SB, and RS own shares in the company. NK and UP are employees at Cognizant Technology Solutions, which has a service agreement with Novartis Pharma AG. EB and STdV are employees at the University Medical Centre Groningen and declare no competing interests in this work. Supporting information Appendix S1 Appendix S2 Kubesch N, Gaitonde S, Petriti U, et al. Use cases of registry‐based randomized controlled trials—A review of the registries' contributions and constraints. Clin Transl Sci. 2024;17:e70072. doi: 10.1111/cts.70072 Nadine Kubesch and Sneha Gaitonde shared first authorship. REFERENCES Associated Data Supplementary Materials Appendix S1 Appendix S2