Clinical trial recruitment software integration is not a single connection. It describes separate handoffs: a system that reads structured EHR data against protocol criteria, a system that passes screened candidates into a clinical trial management system, and a system that receives referrals from external campaigns or portals. Each path creates different obligations for site IT, coordinators, and sponsors [1][2][3]. Bond Health documents an Epic workflow using FHIR R4 APIs to screen charts and return ranked candidates, while OneStudyTeam describes an open API that routes patients from recruitment vendors into site pre-screening queues [1][6]. A 2026 scoping review found that only 9% of EHR-based recruitment support systems were active, notification-driven systems; most remained passive or retrospective [4]. Teams should therefore define which handoff they need before comparing vendor interfaces, because integration capability does not by itself establish workflow adoption [2][4].

Read-side integration: EHR screening via FHIR

Bond Health describes an Epic connection built on Epic's standard FHIR R4 APIs. The vendor states that the software screens charts against a study's criteria and passes ranked candidates to coordinators, without connecting to Epic's Life Sciences platform or Discovery research toolset [1]. The same source notes that a pilot without EHR integration can begin from a site-exported patient list, with EHR screening added after IT approval [1]. This sequence matters for evaluation: a team can test candidate quality and coordinator workflow before committing to an interface project.

The technical decision points are approval burden and data scope. Epic recommends that health systems request a licensing estimate before enabling a third-party app, because the connection can generate costs depending on the Epic contract [1]. Bond states that it charges no integration fee and that its integration work is covered by a volume-based fee per screened patient [1]. The source lists FHIR R4, US Core STU 6.1.0, SMART App Launch Release 2.0.0, Bulk Data Access, and USCDI v3 as the criteria standards [1]. What remains unknown from Bond's page is the false-positive rate of chart screening, the time coordinators spend reviewing ranked candidates, or whether identified patients convert to enrollment at any documented rate [1].

Write-side integration: CTMS and referral handoffs

A second integration class moves candidates from recruitment software into a system of record. Bond states that with CRIO, pre-screened patients go into CRIO through its API and their status returns to Bond, with no double entry [5]. For other CTMS systems, Bond works through a site's own API access or file exports [5]. This distinction is a practical evaluation criterion: a certified two-way integration for one CTMS does not guarantee equivalent behavior for another system at the same site.

RealTime eClinical Solutions describes a different integration surface. Its CTMS can integrate with SubjectWell's study-specific advertising campaigns, and its API supports OData-enabled RESTful requests, signed JWT bearer tokens, and Postman-based setup [7]. OneStudyTeam describes its own API as public and free for recruitment vendors, with documentation online and multi-directional data flow that includes planned screening dates, inclusion/exclusion failure reasons, reasons not screened, and IxRS subject IDs [6]. Vendors also publish different claims: RealTime cites no measured enrollment effect, while ProofPilot states that its scheduler booking boosted conversion rates by over 350% [8]. That figure is a vendor claim without independent verification in the public documents, so it should not be treated as a benchmark.

What to ask before enabling a FHIR connection

Request a licensing estimate from the EHR vendor before enabling a third-party app, as Epic recommends, so the health system understands cost implications before data requests are approved [1]. Confirm which FHIR resources the recruitment system requests and whether they include structured notes, because the public documents do not establish that all systems read unstructured clinical text. Ask whether the vendor can start from a site-exported patient list, and whether that pilot data can later be compared against EHR-screened candidates, as Bond describes for its Epic path [1]. Finally, define review and notification ownership: a passive screening list places the notification burden on coordinators, while only a minority of published systems actively notify targeted users [4].

Workflow integration, not just interface availability

The scoping review by Vaterkowski and colleagues examined EHR-based clinical trial recruitment support systems published between 2013 and 2024. Only 25 of 44 included articles described the interaction model in enough detail to classify the system as active or passive. Four systems, or 9%, were active, meaning they sent notifications to targeted users [4]. Patient-trial matching relied on structured data queries in 55% of studies, while eligibility criteria transformation was computer-assisted in 48% [4]. These findings indicate that interface availability and workflow notification are not the same feature.

The 2026 NPJ Digital Medicine evaluation by Strantz and colleagues offers a sharper conclusion. The authors wrote that process measures should capture whether CTRSS-identified patients were reviewed, confirmed as eligible, approached, consented, and enrolled, and that attribution of enrollment to individual system suggestions would allow more direct evaluation [2]. Their multicenter study used an interrupted time series design because a control group-based design was deemed unfeasible due to anticipated control-group attrition, and a cluster-randomized design would have been resource-intensive and challenging to implement [2]. This is a meaningful evidence limit: even a published evaluation can describe workflow pathways without proving that a system increased enrollment.

What the evidence cannot establish

Vendor pages answer data-flow questions but not outcome questions. Bond Health's Epic page states the FHIR resources requested and the approval path, but it does not publish a conversion-rate benchmark from its own deployments [1]. ProofPilot documents features such as online scheduling, configurable prescreeners, ADA compliance options, and IRT status sync, but the conversion-rate claim appears only as a vendor statement [8]. RealTime documents integration partners and API mechanics without reporting recruitment outcomes [7]. These are different documentation classes: interface specifications are checkable, while outcome claims require study-level evidence.

Academic sources also carry limits. The RIC decision guide from Grout and colleagues emerged from 17 consultations over three years, and its authors state that the guide is not a comprehensive inventory of site-specific technical maturity, that blank cells do not mean a strategy will not work, and that the strategies are generalized [3]. The Vaterkowski scoping review found that 61% of evaluations were retrospective and only 23% produced a complete system evaluation from data transformation to patient-trial matching [4]. A site team cannot use these documents to predict local yield; it can use them to anticipate that notification design, nurse review burden, and criterion computability will determine whether an integration is used at all [2][4].

A concrete evaluation example

Consider an academic medical center running a hypertension study with a narrow inclusion window: systolic blood pressure between 140 and 160 mmHg at two separate visits, no active pregnancy, and baseline eGFR above 45. The site could test Bond's documented Epic path by approving FHIR R4 access for Observation resources that carry blood pressure readings and Condition resources that carry pregnancy and kidney diagnoses [1]. Bond states that a pilot can begin from a site-exported patient list before Epic approves the connection, which would let the coordinator team measure how many ranked candidates survive manual review without any interface project [1].

For the CTMS handoff, the same site could compare two configurations: Bond's CRIO integration, which states a two-way API path for pre-screened patients and statuses, against a file-export flow to a non-CRIO CTMS that Bond says works through a site's own API access [5]. The coordinator team would record whether any patient appears in both systems, whether statuses return from the CTMS, and how long each handoff takes. Those measurements are process metrics, not enrollment proof. The NPJ Digital Medicine authors specifically recommend that such pathway data include confirmation of eligibility, approach, and consent to make later attribution possible [2]. This example is a suggested evaluation method; no testing is reported in the public documents.