Clinical research coordinators reviewing patient records during eligibility screening
    Clinical Operations

    The Hidden Cost of Manual Screening in Clinical Trials

    Manual patient screening drains coordinator hours, inflates screen failure costs, and delays enrollment. See what the data shows and how automated pre-screening may help.

    Published by Kitsa Editorial Team·December 11, 2024·~18 min read

    Every clinical research coordinator knows the workflow: pull up a patient record, work through a protocol's inclusion and exclusion criteria one by one, document the finding, move to the next candidate. For a single Phase III oncology trial, that process can take close to an hour per evaluation [1]. For a coordinator juggling several concurrent studies, that hour rarely exists without trade-offs. The result is a workflow that looks operational on the surface while quietly consuming staff hours and generating costs that trial budgets may not fully account for.

    Manual eligibility screening is a well-documented contributor to recruitment difficulty and one of the more resource-intensive tasks in clinical trial conduct. This article examines what the published evidence shows about the scale of that burden, where its costs accumulate, and what automated and AI-supported approaches are demonstrating in real trial environments.

    A note on terminology: this article discusses pre-screening (informal, staff-initiated review to identify potentially eligible patients before formal consent) and formal eligibility screening (the protocol-defined process after informed consent is obtained) as distinct steps, because their resource implications differ. Most of the coordinator workload discussed here falls in the pre-screening and initial eligibility review phase.

    Why manual screening quietly drains clinical trial operations

    [1]
    31.62%
    Observed coordinator time spent on patient eligibility screening in one pediatric emergency department study [2]
    [2]
    0.84 hrs
    Mean Phase III oncology eligibility evaluation time per candidate [1]
    [3]
    7.55 hrs
    Mean screening effort per patient ultimately enrolled in Phase III oncology trials [1]
    [4]
    $129-$336
    Screening cost per enrolled patient across trial phases in Penberthy et al. [1]
    [5]
    $90,000+
    Estimated annual institutional screening burden in one cancer research setting [1]
    [6]
    51%
    Screen failure rate in a single-institution oncology audit of 7,481 screened patients [3]

    Why Manual Screening Consumes More Than Anyone Budgets For

    The scale of the problem becomes visible when you measure time rather than dollars. A time-and-motion study published in Pediatric Emergency Care by Dexheimer et al. (2019), conducted across 15 observation sessions with eight clinical research coordinators at a pediatric emergency department, found that patient eligibility screening consumed 31.62% of observed coordinator time, the single largest individual task category recorded [2]. Patient contact accounted for 18.67%, performing procedures 17.6%, and physician contact 1%, with other activities making up the remainder. The study was limited to one emergency department setting and 30 total hours of observation; whether its proportions generalize to other site types and therapeutic areas requires further study.

    For complex oncology protocols, the time burden is documented to be substantially higher. Penberthy et al. (2012), publishing in the Journal of Oncology Practice, used prospectively tracked screening effort at a cancer research institution and found that mean Phase III eligibility evaluation time was approximately 0.84 hours per evaluation, with a mean of 7.55 hours of screening effort per patient ultimately enrolled [1]. Because a large proportion of screened candidates do not qualify, most of that labor is expended on patients who never enroll. Those figures reflect a specific institutional setting and phase distribution and should not be treated as universal benchmarks, but they illustrate the compounding effect of high screen failure rates on staff time.

    What Screen Failures Actually Cost

    Screen failures are not merely a statistical inconvenience. Each patient who undergoes eligibility assessment but does not enroll represents a real financial event for the site. Penberthy et al. found that the cost of eligibility screening ranged from $129 to $336 per enrolled patient across trial phases, with an estimated annual burden of more than $90,000 to a single institution, and that these costs were typically not reimbursed through sponsor contracts [1].

    Screen failure rates in published trial data are substantial. A retrospective audit of 15 randomized studies at Tata Memorial Centre's medical oncology solid tumor unit found that 51% of 7,481 screened patients failed to enroll, with the most frequent documented reason being failure to meet inclusion criteria (54.9% of failures) [3]. This was a single-institution audit from one specialized oncology unit and should not be generalized to all trial settings. Among phase II/III genitourinary cancer trials, Wong et al. (2018), reviewing 50 published studies conducted between 1999 and 2016, found a mean screen failure rate of 26% in prostate cancer trials (range 12% to 45%), with ineligibility the most frequently cited reason [4].

    The Budget Lines That Are Routinely Missed

    Trial budgets may not fully account for all costs of screening visits beyond the direct items: lab tests ordered per protocol, investigator time, imaging required at baseline. The indirect cost of eligibility review labor is frequently missing from budget models. Penberthy et al. noted explicitly that these costs were typically not compensated through contracts supporting clinical trials and described them as a significant and largely nonreimbursed financial burden to research institutions [1]. Sites conducting multiple concurrent oncology studies face a compounding effect: coordinator capacity absorbed by screening one study is unavailable for enrollment or data management activities in another.

    Current Evidence and Research Landscape

    The evidence base on manual screening inefficiency is now substantial enough to shift the conversation from whether the burden exists to how urgently it should be addressed operationally.

    Cancer trial accrual is a well-recognized challenge. Peterson et al. (2022), analyzing 1,197 NCI-affiliated phase II/III interventional cancer trials initiated between 2008 and 2018, found that 19.3% (231 trials) failed because of low accrual, and that accrual failure rate rose from 11.8% in the lowest decile of eligibility criteria complexity to 29.4% in the highest decile [5]. The same study documented that median eligibility criteria word count increased by 95%, from 214 unique content words in 2008 to 417 in 2018, and that this growth was independently associated with accrual failure [5]. A systematic review by Unger et al. (2019), published in the Journal of the National Cancer Institute, pooled enrollment data from 13 studies and 8,883 patients represented across trial decision-pathway studies and estimated a pooled enrollment rate of 8.1%, with markedly lower rates in community settings (7.0%) than in academic ones (15.9%) [6]. The review also identified structural, clinical, physician, and patient-level barriers contributing to this gap, noting that even patients who are potentially eligible and willing to enroll face significant practical obstacles [6].

    Manual pre-screening may not reach all potentially eligible patients across a large health system. A coordinator reviewing a tumor board roster or an admission census works within the bounds of available hours and familiarity with actively recruiting studies. Patients who meet criteria but are not flagged through these informal channels may be missed without documentation of the missed opportunity.

    The Protocol Complexity Problem

    Eligibility criteria have grown more intricate over time, driven by precision oncology and biomarker-stratified trial designs. The ASPE's examination of clinical trial costs and barriers noted that complex eligibility requirements create competition for limited patient pools, particularly in targeted oncology and rare disease indications [7]. The Peterson et al. analysis found that eligibility criteria word count increased by 95%, nearly doubling across NCI-affiliated trials between 2008 and 2018, and that criteria complexity was independently associated with accrual failure [5]. When a single protocol carries 30 or more inclusion and exclusion criteria spanning laboratory values, prior therapy history, molecular markers, and organ function thresholds, manual chart review is not merely slow; it is prone to omission. In the Tata Memorial Centre audit, the most common documented reason for screen failure was failure to meet inclusion criteria [3]. The audit did not establish whether structured EHR-based pre-filtering would have changed those outcomes.

    How manual screening turns into hidden trial cost

    1
    Protocol complexity
    More eligibility criteria, biomarkers, lab thresholds, prior therapy rules, and organ function requirements
    2
    Manual chart review
    Coordinator time spent one candidate at a time across fragmented patient records
    3
    Screen failure
    Many reviewed patients do not qualify or do not enroll after assessment
    4
    Unreimbursed site burden
    Screening labor and pre-screening effort often fall outside sponsor reimbursement
    5
    Enrollment delay
    Coordinator capacity shifts away from outreach, retention, data queries, and enrollment workflows
    6
    Trial-level inefficiency
    Low accrual, missed candidates, site mismatch, and delayed enrollment milestones

    Manual screening cost is not only the cost of one review. It is the cumulative opportunity cost of coordinator time spent on patients who never enroll.

    Operational Implications for Sites and Sponsors

    The operational consequences of manual pre-screening extend beyond the assessment visit itself. Two downstream effects are worth examining with appropriate caution.

    Coordinator capacity allocation. Time spent on manual chart review is time not directed toward patient communication, retention activities, data query resolution, or study-specific training. When pre-screening consumes a substantial share of coordinator effort, other trial functions absorb the deficit in ways that are difficult to trace directly but are nonetheless real. Whether this relationship holds consistently across different site types and therapeutic areas requires more systematic study than a single time-motion observation can provide.

    Site selection and patient population mismatch. Sites can be selected on the basis of investigator relationships or historical enrollment performance, without adequate structured analysis of whether the actual patient population in their EHR matches the protocol's eligibility profile. When that mismatch surfaces during the pre-screening phase, it appears as low screen-to-enroll conversion. Feasibility assessments built on structured, prospective patient data queries could expose these gaps earlier in the process, though the operational feasibility of implementing such queries consistently across diverse EHR environments remains a practical challenge.

    Regulatory and Documentation Considerations

    The ICH E6(R3) Good Clinical Practice guideline was finalized by the ICH Assembly on January 6, 2025. The EMA adopted it effective July 23, 2025, making it legally effective for EU clinical trials from that date [14]. The FDA published its final guidance for industry on September 8, 2025, making the guideline officially available in the United States [8]. The MHRA published UK-specific annotations, and full UK legal compliance became required from April 28, 2026 [15]. Implementation timelines and the precise scope of regulatory obligation differ by jurisdiction; sponsors should confirm requirements with relevant regional regulators.

    E6(R3) is relevant to automated screening approaches in several ways. The guideline provides updated principles for using electronic health records as source documentation, acknowledging that original data may be captured in various formats including EHR systems and patient-reported outcome platforms [9]. Its risk-proportionate framework encourages fit-for-purpose solutions, including centralized and remote monitoring approaches, and is designed to remain relevant as technology evolves.

    The guideline does not specifically validate AI eligibility screening tools, and it does not establish a requirement that every automated pre-screening workflow be described in the protocol or ICF. Sponsors deploying AI-supported screening tools should document their validation approach, define the process for qualified human review before final enrollment decisions, and confirm that eligibility determination records meet the source data integrity requirements of the guideline. Final eligibility determination remains an investigator responsibility under GCP regardless of what automated systems are used upstream.

    The FDA's final guidance, like the E6(R3) document it accompanies, describes the agency's current thinking and does not establish legally enforceable obligations [8].

    A safer model for automated pre-screening

    1
    Protocol criteria structuring
    Inclusion and exclusion criteria converted into structured eligibility logic
    2
    EHR and FHIR query layer
    Diagnoses, medications, labs, procedures, clinical notes, and relevant structured data queried where available
    3
    Candidate prioritization
    Likely eligible patients surfaced for coordinator review
    4
    Human assessment
    Coordinator, investigator, or delegated qualified personnel validates eligibility context
    5
    Enrollment workflow
    Only human-confirmed candidates proceed into consent, formal screening, and trial-specific enrollment steps

    Automation should reduce the volume of records requiring full manual review, not replace investigator responsibility for final eligibility determination.

    AI and Automation in Eligibility Pre-Screening: What the Evidence Shows

    The research literature on AI-assisted clinical trial screening has grown substantially in recent years, with several studies now reporting accuracy and efficiency data from real-world and retrospective settings. These results are meaningful but require context: most high-performing systems operate within defined EHR environments with relatively clean structured data and well-characterized patient populations. Performance in settings with fragmented EHR infrastructure, rare disease populations, or highly heterogeneous eligibility criteria may differ.

    A retrospective evaluation of MSK-MATCH (Memorial Sloan Kettering Multi-Agent Trial Coordination Hub), published as a preprint by Rosenthal et al. in November 2025, tested the system across 88,518 clinical documents from 731 breast cancer patients across six trials. The system classified 61.9% of eligibility cases in the retrospective dataset without flagging them for human review, and triaged 38.1% for coordinator assessment. For triaged cases, pre-populating the eligibility screen with AI-generated explanations reduced review time from 20 minutes to 43 seconds per case, at an average computational cost of $0.96 per patient-trial pair. Across the full cohort, the workflow achieved 98.6% accuracy, 98.4% sensitivity, and 98.7% specificity for patient-level eligibility classification [10]. As a preprint, these findings have not completed peer review and should be interpreted accordingly.

    Published in the Journal of Cardiac Failure in 2026, Martyn et al. studied an AI system (Synapsis AI, Dyania Health) deployed within a unified EHR system across multiple Cleveland Clinic hospitals in Ohio and Florida beginning in August 2024. The system assessed patient eligibility for a transthyretin amyloid cardiomyopathy trial using both structured and unstructured EHR data and achieved 96% accuracy across nine prespecified domains of trial criteria [11]. The study noted the importance of auditable AI justifications for supporting investigator review.

    In cardiovascular research, van Dijk et al. (2021), publishing in the Journal of Clinical Epidemiology, validated EHR text-mining for trial eligibility pre-screening across three medical centers with different EHR vendors as part of the LoDoCo2 trial. Automated EHR screening of 92,466 patients showed that the number of patients requiring manual review could be reduced by 73,863 (79.9%), with the remaining 20.1% containing 82.4% of actual trial participants. The overall accuracy of automatically extracted data was 88.0% [12]. The study noted that 17.6% of actual trial participants were missed by the text-mining tool, primarily due to absent or incomplete data fields, underscoring that automated pre-screening systems require validation before deployment and cannot substitute for human eligibility determination.

    An earlier automated pre-screening algorithm tested at Massachusetts General Hospital, published in Clinical Trials by Beauharnais et al. in 2012, halved total daily screening time from four hours to two in an inpatient diabetes trial, while more than doubling the average number of patients pre-screened daily (from 13 to 30) and increasing enrollment rates. Developing the algorithm added a fixed cost of $3,000 to the study [13].

    What Automation Cannot Do

    These results should not be read as evidence that manual screening can be eliminated. The MSK-MATCH system triaged 38.1% of retrospective cases for human assessment. The van Dijk cardiovascular EHR system missed roughly one in six actual trial participants. The Cleveland Clinic system required a defined diagnostic code pre-filter before AI processing began. Each of these systems also operated in a controlled environment with a defined trial population and structured EHR access that many sites cannot replicate.

    The appropriate operational model, supported by both the evidence and the regulatory framework of ICH E6(R3), is human-AI collaboration: automated pre-screening that reduces the volume of records requiring full coordinator review and identifies likely eligible candidates at scale, with qualified human assessment before any final eligibility determination. False negatives in automated pre-screening (missing patients who would qualify) carry patient access and recruitment implications that make human oversight not merely procedurally required but practically necessary.

    How Kitsa Fits Into This Problem

    KScreener, Kitsa's patient pre-screening product, is designed to connect EHR data to a trial's eligibility criteria and flag likely eligible patients before manual chart review begins. The intended use is to reduce the volume of patients requiring full coordinator-level assessment. For sponsors building out broader trial infrastructure, KScreener is designed to operate within the Kitsa platform alongside KScout's site feasibility intelligence and KScribe's regulatory document generation tools.

    KScreener · Patient Pre-Screening

    Manual screening costs accumulate before most patients ever reach formal enrollment. KScreener is designed to connect EHR data to trial eligibility criteria, surface likely eligible patients earlier, and reduce the volume of records requiring full coordinator-level review. Used alongside KScout for site feasibility intelligence, it helps sponsors, CROs, and sites move from one-record-at-a-time screening toward structured, scalable patient matching.

    Key Takeaways

    • Manual eligibility pre-screening was the single largest individual task category for clinical research coordinators observed in a published time-motion study, consuming 31.62% of recorded time in one pediatric emergency department setting. The finding is specific to that setting and should not be assumed universal.
    • Penberthy et al. (2012) found mean Phase III eligibility evaluation time of approximately 0.84 hours per evaluation and 7.55 hours of staff effort per enrolled patient, with these costs largely unreimbursed through sponsor contracts. Screening costs ranged from $129 to $336 per enrolled patient.
    • Screen failure rates vary substantially by indication and trial phase. A single-institution oncology audit found 51% of screened patients failed to enroll; a review of GU cancer trials found mean Phase III prostate cancer screen failure rates of 26% (range 12% to 45%).
    • Peterson et al. (2022) found that 19.3% of 1,197 NCI-affiliated cancer trials failed because of low accrual, with failure rates rising from 11.8% among trials with the simplest eligibility criteria to 29.4% among the most complex.
    • Automated EHR pre-screening systems have demonstrated meaningful reductions in manual review volume in published studies, with one cardiovascular multicenter validation showing 79.9% fewer patients requiring full manual review. Missing rates, data quality requirements, and EHR infrastructure constraints vary across settings.
    • ICH E6(R3) was finalized in January 2025. FDA published final guidance in September 2025; EMA adopted it effective July 2025. Implementation timelines and obligations are jurisdiction-specific. Final eligibility determination remains an investigator responsibility under GCP, regardless of upstream automation.
    • None of the systems discussed in this article eliminates the need for qualified human review; the appropriate model is automated candidate identification followed by coordinator-level assessment before enrollment decisions.

    FAQ

    What is a screen failure in a clinical trial?
    A screen failure occurs when a patient who has consented to participate in a clinical trial undergoes formal eligibility assessment but does not meet the protocol's inclusion and exclusion criteria and cannot be enrolled. Screen failures generate cost and staff effort without contributing data to the study. Common documented reasons include laboratory values outside protocol-specified ranges, ineligible disease staging, prior therapy history, and administrative or scheduling barriers. The term is also sometimes used informally to describe pre-screening rejections before consent, though these are operationally distinct events.
    How much does eligibility screening cost at the site level?
    Published estimates vary by trial phase, therapeutic area, and how costs are defined. Penberthy et al. found that screening costs ranged from $129 to $336 per enrolled patient across trial phases, with more than $90,000 in estimated annual institutional burden, and that these costs were typically not reimbursed through sponsor contracts [1].
    How common is accrual failure in cancer clinical trials?
    Peterson et al. (2022), analyzing 1,197 NCI-affiliated phase II/III cancer trials, found that 19.3% (231 trials) failed because of low accrual. Accrual failure was significantly associated with eligibility criteria complexity: failure rates ranged from 11.8% in trials with the simplest eligibility criteria to 29.4% in those with the most complex. Unger et al. (2019), in a meta-analysis of 13 studies and 8,883 patients, estimated a pooled cancer trial enrollment rate of 8.1%, with markedly lower rates in community versus academic settings, reflecting both patient-facing and structural barriers.
    What regulatory guidance governs eligibility screening documentation?
    ICH E6(R3), finalized by the ICH Assembly in January 2025, provides the current international Good Clinical Practice standard. EMA adopted it effective July 23, 2025. The FDA published final guidance for industry in September 2025, reflecting the agency's current thinking though not constituting binding requirements. The guideline acknowledges EHR-derived source data and encourages risk-proportionate, fit-for-purpose approaches to trial conduct. It does not specifically mandate or validate AI pre-screening workflows, and final eligibility determination remains an investigator responsibility under all versions of the guideline.
    Can AI systems replace manual eligibility screening?
    Current published evidence does not support replacing qualified human judgment in eligibility determination with fully automated systems. AI and NLP systems have demonstrated meaningful reductions in the volume of records requiring full manual review, and in some settings have flagged eligible patients that would have been missed through informal referral-only processes. However, each of the systems discussed in this article that has reported real-world results has retained human review for a proportion of cases, and missing rates in automated systems carry patient access implications. The appropriate model is human-AI collaboration: automated pre-screening concentrates coordinator attention on likely candidates, with qualified human assessment before enrollment decisions.
    What is FHIR and why does it matter for clinical trial pre-screening?
    Fast Healthcare Interoperability Resources (FHIR) is a data standard for exchanging healthcare information electronically. FHIR-enabled EHR systems allow structured patient data (diagnoses, laboratory results, medications, procedures) to be queried programmatically against eligibility criteria. This makes it technically possible to build automated pre-screening workflows that operate at a population level rather than one patient at a time. Not all EHR systems are fully FHIR-enabled, and data quality and completeness vary significantly across implementations, which affects the performance of any automated pre-screening system that depends on structured EHR data.

    References

    1. [1]Penberthy LT, Dahman BA, Petkov VI, DeShazo JP. "Effort Required in Eligibility Screening for Clinical Trials." Journal of Oncology Practice, 2012;8(6):365-370. doi:10.1200/JOP.2012.000646. PMC3500483. Link
    2. [2]Dexheimer JW, Tang H, Kachelmeyer A, Hounchell M, Kennebeck S, Solti I, Ni Y. "A Time-and-Motion Study of Clinical Trial Eligibility Screening in a Pediatric Emergency Department." Pediatric Emergency Care, 2019;35(12):868-873. doi:10.1097/PEC.0000000000001592. PMC6445787. Link
    3. [3]Patil V, et al. "Audit of Screen Failure in 15 Randomised Studies from a Low and Middle-Income Country." ecancermedicalscience, 2022;16:1476. doi:10.3332/ecancer.2022.1476. PMC9934872. Link
    4. [4]Wong SE, North SA, Sweeney CJ, Stockler MR, Sridhar SS. "Screen Failure Rates in Contemporary Randomized Clinical Phase II/III Therapeutic Trials in Genitourinary Malignancies." Clinical Genitourinary Cancer, 2018;16(1):e233-e242. doi:10.1016/j.clgc.2017.08.019. PMID:28993132. Link
    5. [5]Peterson JS, Plana D, Bitterman DS, Johnson SB, Aerts HJWL, Kann BH. "Growth in Eligibility Criteria Content and Failure to Accrue Among National Cancer Institute (NCI)-Affiliated Clinical Trials." Cancer Medicine, 2022. doi:10.1002/cam4.5276. PMID:36398619. PMC9972031. Link
    6. [6]Unger JM, Vaidya R, Hershman DL, Minasian LM, Fleury ME. "Systematic Review and Meta-Analysis of the Magnitude of Structural, Clinical, and Physician and Patient Barriers to Cancer Clinical Trial Participation." JNCI, 2019;111(3):245-255. doi:10.1093/jnci/djy221. PMC6410951. Link
    7. [7]U.S. Department of Health and Human Services, ASPE. "Examination of Clinical Trial Costs and Barriers for Drug Development." ASPE Report, 2014. Link
    8. [8]U.S. Food and Drug Administration. "E6(R3) Good Clinical Practice: Guidance for Industry." September 2025. Federal Register, September 9, 2025. FR Doc. 2025-17311. Link
    9. [9]International Council for Harmonisation. "ICH E6(R3) Guideline for Good Clinical Practice." Finalized January 6, 2025. Link
    10. [10]Rosenthal JT, Hahesy E, Chalise S, Zhu M, Sabuncu MR, Braunstein LZ, Li A. "AI-Assisted Workflow Enables Rapid, High-Fidelity Breast Cancer Clinical Trial Eligibility Prescreening." arXiv, November 2025. (Preprint; not yet peer-reviewed.) Link
    11. [11]Martyn T, Hanna M, Nissen SE, et al. "Automating Chart Review Using an Artificial Intelligence-Enabled System for Assessing Transthyretin Amyloid Cardiomyopathy Trial Eligibility." Journal of Cardiac Failure, 2026. doi:10.1016/j.cardfail.2026.01.010. Link
    12. [12]van Dijk WB, Fiolet ATL, Schuit E, Sammani A, Groenhof TKJ, van der Graaf R, et al. "Text-Mining in Electronic Healthcare Records Can Be Used as Efficient Tool for Screening and Data Collection in Cardiovascular Trials: A Multicenter Validation Study." Journal of Clinical Epidemiology, 2021;132:97-105. doi:10.1016/j.jclinepi.2020.11.014. PMID:33248277. Link
    13. [13]Beauharnais CC, Larkin ME, Zai AH, Boykin EC, Luttrell J, Wexler DJ. "Efficacy and Cost-Effectiveness of an Automated Screening Algorithm in an Inpatient Clinical Trial." Clinical Trials, 2012;9(2):198-203. doi:10.1177/1740774511434844. PMC5222546. Link
    14. [14]European Medicines Agency (EMA). "ICH E6 Good Clinical Practice: Scientific Guideline." Effective July 23, 2025 for EU clinical trials. Link
    15. [15]Medicines and Healthcare products Regulatory Agency (MHRA). "UK-Specific Annotations to ICH E6(R3)." UK legal compliance required from April 28, 2026. Link

    Related posts