Rethinking EDC Go-Live Metrics: Why EDC Design Phase Work Hours Outperform Days in Measuring DM Efficiency (641 Clinical Trials)

Rethinking EDC Go-Live Metrics: Why EDC Design Phase Work Hours Outperform Days in Measuring DM Efficiency (641 Clinical Trials) | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Rethinking EDC Go-Live Metrics: Why EDC Design Phase Work Hours Outperform Days in Measuring DM Efficiency (641 Clinical Trials) Charles Yan, Shuyu Gou, Huaihai Yan This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7831647/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background The global clinical research industry widely uses EDC go-live days (GLD, time from protocol finalization to EDC launch) to measure data management (DM) efficiency, but GLD is biased by non-DM delays and fails to reflect trial complexity. EDC Design Phase Work Hours (DPWH, total DM effort for EDC setup) is a potential alternative but lacks large-scale validation. Objective To compare GLD and DPWH in reflecting DM efficiency using 641 diverse clinical trials, by assessing their associations with project characteristics (therapeutic area [TA], phase, EDC system) and DM workload proxies (CRF count, edit check [EC] count), and to test their correlation. Methods Retrospective analysis of 641 trials (2019–2025). Descriptive statistics, Spearman’s correlation (including GLD-DPWH correlation), Kruskal-Wallis tests, robust regression, and mixed-effects models (intraclass correlation [ICC]) were used. Results Mean GLD was 101.2 ± 60.1 days, mean DPWH 219.1 ± 148.6 hours. GLD-DPWH correlation was moderate (r = 0.47, p < 0.005). GLD had weak associations with project characteristics (e.g., TA: χ²=60.61, p < 0.0001) and workload (CRF: r = 0.37; EC: r = 0.38). DPWH strongly correlated with characteristics (e.g., Phase III vs. I: 335.5 ± 201.4 vs. 182.0 ± 115.4 hours, p < 0.0001) and workload (CRF: r = 0.48; EC: r = 0.52). Robust regression explained 61% of DPWH variation vs. 26% for GLD. DPWH had higher ICC (TA: 0.39; Phase: 0.5) than GLD (TA: 0.33; Phase: 0.37). Conclusion DPWH outperforms GLD as a DM efficiency metric, aligning with trial complexity and workload. Its adoption can improve resource allocation and process optimization. Electronic Data Capture (EDC) EDC Design Phase Work Hours (DPWH) EDC Go-Live Days (GLD) Data Management Efficiency Clinical Trials 1. Introduction 1.1 Background Electronic Data Capture (EDC) systems are indispensable for modern clinical research, enabling compliant, efficient data collection across all trial phases—from early-phase clinical pharmacology studies to late-phase confirmatory trials [1,2,3]. As trials grow more complex (e.g., integrating biomarkers, global multi-center designs), timely EDC launch directly impacts overall trial progress, making "EDC go-live efficiency" a core priority for DM departments. The most widely adopted metric for this efficiency is EDC go-live days (GLD) , defined as the duration from final protocol approval to EDC system activation [4,5,6]. GLD’s popularity stems from its simplicity: it is easy to quantify and compare across projects, aligning with the timeline pressures of clinical development (e.g., accelerating trial initiation to meet regulatory milestones) [7,8,9]. However, GLD has critical flaws that undermine its validity as a true measure of DM efficiency: Susceptibility to non-DM delays : GLD is heavily influenced by external factors unrelated to DM performance, such as protocol amendments, sponsor review bottlenecks, ethics committee feedback timelines, and IT system downtime. For example, a complex oncology trial may have prolonged GLD due to repeated protocol revisions—even if the DM team efficiently completes EDC setup. Failure to reflect trial complexity: A Phase I trial with 30 case report forms (CRFs) and 300 edit checks (ECs) may have the same GLD as a Phase III trial with 60 CRFs and 1200 ECs, despite the latter requiring twice the DM effort. To address these shortcomings, EDC Design Phase Work Hours (DPWH) has emerged as a promising alternative. DPWH is defined as the total labor hours invested in DM-specific EDC setup tasks during the design phase, encompassing tangible activities: CRF design, EC programming, user acceptance testing (UAT), data management plan (DMP) development, CRF completion guide creation, and database quality control (QC) [unpublished]. Unlike GLD, DPWH directly captures DM workload and is inherently tied to trial complexity, avoiding distortions from non-DM factors. However, few studies have validated DPWH using large-scale, diverse trial datasets, nor systematically compared it to GLD across key project characteristics (e.g., therapeutic area [TA], trial phase, EDC system). 1.2 Research Gap and Objective Existing literature on EDC efficiency metrics has three critical limitations for the global clinical research industry[10,11,12]: 1. Overreliance on GLD : Most studies focus on GLD or process optimization (e.g., automating ECs) but neglect workload-based metrics like DPWH. Even multi-stakeholder studies (involving CROs, sponsors, and academic institutions) rely on descriptive survey data rather than quantifiable empirical analysis, failing to link "efficiency assessment" to "actual DM effort". 2. Small sample sizes: Limited samples fail to capture variability across TA, phase, and EDC systems—key drivers of workload. This also prevents tracking long-term trends: while EDC systems have advanced (e.g., automated ECs, standardized CRF templates), survey data shows GLD has remained stable at ~70 days for years, providing no insight into technological progress. 3. Lack of workload linkage : No studies explicitly connect efficiency metrics to core workload proxies (e.g., CRF count, EC count). For GLD, this means the "stable 70-day GLD" cannot distinguish between "DM efficiency gains offset by complex trials" and "genuine stagnation," making it impossible to validate if GLD responds to workflow or system improvements [13]. To fill these gaps, this study uses data from 641 diverse ** innovative drug clinical trials of Hengrui Pharmaceutical **—all managed by the Clinical Data Science Center of Hengrui and completed EDC go-live by June 30, 2025 (spanning 2019–2025)—to pursue two core objectives: 1. Assess the associations of GLD and DPWH with key project characteristics (therapeutic area [TA], trial phase, EDC system/version, initiation year) across this broad, representative range of trials; 2. Compare GLD and DPWH in their ability to correlate with DM workload proxies (unique CRF count, edit check [EC] count), and verify if DPWH outperforms GLD as a universal DM efficiency metric. Notably, the Clinical Data Science Center of Hengrui adopts a distinct operational model: it has no specialized EDC database build (DBB) personnel, with data managers (DMs) directly leading all EDC setup tasks from trial start. This eliminates the usual communication/coordination costs between DMs (who define data needs) and separate DBB staff in conventional models—allowing us to capture EDC build time with high accuracy, which strengthens DPWH’s reliability (critical for a workload-dependent metric) and ensures DPWH reflects real DM effort (not cross-role inefficiencies), boosting the persuasiveness of our GLD-DPWH comparisons. Further, this DM-led EDC setup aligns with ICH GCP E6R3’s "Quality by Design (QbD)" principle: as stakeholders most familiar with data quality requirements, DMs embed integrity and compliance upfront during build, rather than retrofitting later. It also represents a concrete practice of transitioning from traditional Clinical Data Management (CDM) to Clinical Data Science (CDS)—empowering DMs to integrate data management expertise with technical EDC capabilities, moving beyond transactional tasks to drive data-centric trial efficiency【14】. 1.3 Significance This study contributes to the global clinical research industry in three key ways: Methodological : It validates DPWH—a workload-based metric that addresses GLD’s limitations—providing a more accurate measure of DM efficiency. Practical : It offers actionable insights for resource allocation (e.g., establishing DPWH benchmarks for oncology vs. non-oncology trials) and process optimization (e.g., using DPWH to quantify the impact of EDC system upgrades). Industry-wide : It supports a paradigm shift from "speed-focused" (GLD) to "workload-aligned" (DPWH) efficiency assessment, aligning with regulatory expectations for data quality and reproducibility [15]. 2. Materials and Methods 2.1 Data Source Data were extracted from the Clinical Data Total Management System (CDTMS) of a global pharmaceutical enterprise, covering 641 innovative drug clinical trials managed by the Clinical Data Science Center (CDSC) [16,17] . All trials completed EDC go-live between January 1, 2019, and June 30, 2025, and exhibited broad diversity: Therapeutic areas (TA): Oncology, cardiology, neurology, diabetes, analgesia, autoimmune diseases, and clinical pharmacology; Trial phases : Phase I, Phase II, and Phase III; Investigational products : Innovative drugs: mall-molecule drugs and biologic candidates. This diversity ensures the sample is representative of global clinical research, enhancing the generalizability of results. 2.2 DPWH Tracking timesheet : DM Recorded task-specific labor data daily for predefined task categories with 15-minute increments. For each project, DPWH was calculated by summing labor inputs in EDC designed phase from all participating CDMs, reflecting the collective effort of the team [16]; 2.3 Variable Definition All variables were operationalized based on industry-standard SOPs (aligned with ICH-GCP and CDISC guidelines) and EDC system logs to ensure consistency and reproducibility (Table 1). Table 1. Variable category, name definition and their measurement tools Variable Category Variable Name Definition Measurement Tool Efficiency Metrics GLD Time from protocol finalization (signed by sponsor) to EDC system activation for patient enrollment. Calculation: Approval date of EDC go-live - approval date of protocol version (V1.0) DPWH Total DM work hours for EDC setup during the design phase, including: • CRF design and EDC setup • EC development and UAT • CRF Completion Guide development • Database QC • Data Management Plan, etc CDTMS: Timesheet logs EDCs EDC System/ Version Two system types used across trials: CDTMS：Trial EDC system used • Local EDC (L-EDC): Regional system, subdivided into L-EDC V284 (legacy version) and L-EDC V4.0+/V4.1.x (upgraded version with advanced features: standardization, automation and intelligence)[18,19] • Global EDC (G-EDC): Enterprise-wide system for all type of trials Workload Proxies Unique CRF Count Number of unique valid data collection forms, aligned with CDISC CDASH standards. EDC system report EC Count Number of active edit checks at EDC go-live, includes range checks, consistency checks, and protocol-specific logic. EDC system report 2.4 Statistical Analysis Statistical analyses were performed using SAS (SAS Institute Inc., Cary, NC, USA), with a significance threshold of p<0.05. The analytical approach followed a sequential, "layered" logic to validate DPWH: 1. Descriptive Statistics: - Categorical variables (TA, trial phase, EDC system) were summarized as frequencies and percentages; - Continuous variables (GLD, DPWH, CRF count, EC count) were summarized as mean ± standard deviation (SD) and median (interquartile range [IQR]) to account for outliers. 2. Correlation Analysis: - Spearman’s rank correlation coefficients (r) were calculated to assess associations between: • GLD/DPWH and continuous project characteristics (initiation year); • GLD/DPWH and workload proxies (CRF count, EC count); • GLD and DPWH (to test their linear relationship, pre-specified as a key analysis 3. Stratified Comparison: - Kruskal-Wallis H tests were used to compare DPWH across categorical project characteristics (TA, trial phase, EDC system), with Dunn’s post-hoc procedure (Bonferroni adjustment) for pairwise comparisons. 4. Robust Regression: - Two robust regression models were built: • Model 1: DPWH as the dependent variable, with independent variables (dummy- coded where applicable): TA (reference: Clinical Pharmacology), trial phase (reference: Phase I), EDC system (reference: L-EDC V4.0+), initiation year, CRF count, EC count; • Model 2: GLD as the dependent variable, with the same independent variables. - Model fit was evaluated using adjusted R², with significance assessed via standardized coefficients (β) and p-values. 5. Mixed-Effects Models: - Intraclass Correlation Coefficients (ICC) were calculated to assess homogeneity of projects within subgroups (TA, trial phase). Higher ICC indicates greater within-group homogeneity, reflecting better discriminatory power for efficiency differences. 3. Results 3.1 Project Basic Characteristics The 641 trials exhibited diverse characteristics, representative of global clinical research (Table 2 ). Phase I trials were most common (57.9%), followed by Phase II (24.8%) and Phase III (17.3%). TA distribution was balanced: Clinical Pharmacology (32.3%), Oncology (33.1%), and Non-Oncology (34.6%). L-EDC was the dominant system (83.5%), with 70.7% using the upgraded V4.0 + version. Mean GLD was 101.2 ± 60.1 days (median: 89 days), and mean DPWH was 219.1 ± 148.6 hours (median: 176.7 hours). Workload proxies also varied by TA: Oncology trials had higher mean CRF count (50.9 ± 7.7) and EC count (837.0 ± 251.7) than Clinical Pharmacology trials (CRF: 30.7 ± 6.1; EC:356.3 ± 126.0), consistent with their higher complexity. Table 2 Basic Characteristics of 641 Clinical Trials, along with their DPWH and GLD Characteristic Category Frequency (n) Percentage (%) DPWH (Mean ± SD; Median) GLD (Mean ± SD; Median) Trial Phase Phase I 371 57.9% 182.0 ± 115.4; 152.8 89.6 ± 52.4; 79 Phase II 159 24.8% 224.5 ± 131.4; 190.8 98.0 ± 56.0; 88 Phase III 111 17.3% 335.5 ± 201.4; 264.6 144.6 ± 70.2; 128 Therapeutic Area (TA) Clinical Pharmacology 207 32.3% 151.9 ± 81.6; 135.7 83.8 ± 54.0; 72 Oncology 212 33.1% 289.3 ± 190.0; 228.9 121.9 ± 69.1; 109 Non-Oncology 222 34.6% 214.7 ± 120.0; 187.5 97.7 ± 49.8; 87 EDC System L-EDC 535 83.5% 178.6 ± 89.1; 158.3 93.6 ± 52.2; 84 G-EDC 106 16.5% 423.3 ± 209.2; 382.9 139.5 ± 80.1; 117 Initiation Year 2019 75 11.7% 285.3 ± 192.3; 217.7 127.1 ± 76.1; 113 2020 83 12.9% 295.0 ± 193.2; 263.7 116.2 ± 81.7; 95 2021 117 18.3% 300.0 ± 177.0; 266.5 109.6 ± 64.7; 96 2022 93 14.5% 202.3 ± 95.5; 185.8 105.3 ± 56.7; 96 2023 97 15.1% 157.9 ± 60.8; 150.0 87.6 ± 40.8; 87 2024 151 23.6% 148.4 ± 57.5; 141.7 84.9 ± 38.5; 80 2025 25 3.9% 114.4 ± 28; 114.8 70.8 ± 23.9; 70 Efficiency Metrics Total 641 100% 219.1 ± 148.6; 176.7 101.2 ± 60.1; 89 3.2 Association Between GLD/DPWH and Project Characteristics 3.2.1 Overall Correlations Spearman’s analysis revealed key differences in how GLD and DPWH correlated with project characteristics (Table 3 ). Notably, GLD and DPWH had a moderate positive correlation (r = 0.47, p < 0.005), indicating some overlap but not redundancy. Initiation year : GLD had a weak negative correlation (r=-0.23, p < 0.0001), meaning EDC workflow optimizations (e.g., system upgrades) did not translate to meaningful GLD reductions—likely offset by rising trial complexity. In contrast, DPWH had a strong negative correlation (r=-0.43, p < 0.0001), reflecting a clear downward trend in workload over time (e.g., 2019: 285.3 ± 192.3 hours vs. 2024: 148.4 ± 57.5 hours), consistent with efficiency gains from advanced EDC tools. Table 3 Spearman's rank correlation coefficients Between GLD/DPWH and Continuous Project Characteristics Variables GLD ( r , p -value) DPWH ( r , p -value) GLD vs. DPWH (r, p-value) Initiation Year (-0.23, < 0.0001) (-0.43, < 0.0001) Unique CRF Count (0.37, < 0.0001) (0.48, < 0.0001) EC Count (0.38, < 0.0001) (0.52, < 0.0001) DPWH (0.47, < 0.005) 3.2.2 DPWH Differences Across Categorical Characteristics Kruskal-Wallis tests confirmed DPWH varied systematically by TA, trial phase, and EDC system—capturing trial complexity (Table 4 ): TA : Oncology trials had the highest DPWH (289.3 ± 190.0 hours), followed by Non-Oncology (214.7 ± 120.0 hours) and Clinical Pharmacology (151.9 ± 81.6 hours). All pairwise differences were significant (p Phase II (224.5 ± 131.4 hours) > Phase I (182.0 ± 115.4 hours). Phase III trials required 85% more DPWH than Phase I (p < 0.001), consistent with larger sample sizes, more endpoints, and stricter regulatory requirements. EDC system : G-EDC trials had significantly higher DPWH than L-EDC trials (423.3 ± 209.2 vs. 178.6 ± 89.1 hours, p < 0.001), due to additional setup demands of G-EDC. Table 4 Kruskal-Wallis tests for DPWH with TA, trial phase, and EDC system Categorical Characteristic Comparison Group DPWH (Mean ± SD) p-value TA Oncology vs. Clinical Pharmacology 289.3 ± 190.0 vs. 151.9 ± 81.6 < 0.001 Non-Oncology vs. Clinical Pharmacology 214.7 ± 120.0 vs. 151.9 ± 81.6 < 0.001 Oncology vs. Non-Oncology 289.3 ± 190.0 vs. 214.7 ± 120.0 < 0.001 Trial Phase Phase III vs. Phase I 335.5 ± 201.4 vs. 182.0 ± 115.4 < 0.001 Phase III vs. Phase II 335.5 ± 201.4 vs. 224.5 ± 131.4 < 0.001 Phase II vs. Phase I 224.5 ± 131.4 vs. 182.0 ± 115.4 < 0.001 EDC System G-EDC vs. L-EDC 423.3 ± 209.2 vs. 178.6 ± 89.1 < 0.001 3.3 Association Between GLD/DPWH and Workload Proxies GLD had weak correlations with both workload proxies, while DPWH had strong correlations (Table 3 ): Unique CRF count: GLD (r = 0.37, p < 0.0001) vs. DPWH (r = 0.48, p < 0.0001). Each additional CRF was associated with ~ 3 more DPWH hours. EC count: GLD (r = 0.38, p < 0.0001) vs. DPWH (r = 0.52, p < 0.0001). For example, two Oncology trials with identical CRF counts (50 forms) had GLD of 70 and 120 days—differences driven by non-DM factors (e.g., protocol amendments)—but DPWH of 290 and 305 hours (minimal variation, reflecting similar workload). This confirms DPWH directly reflects core DM effort, while GLD does not. 3.4 Robust Regression Results Robust regression models highlighted DPWH’s superior ability to be explained by meaningful workload factors (Table 5 ): DPWH model (Adjusted R²=0.61, p < 0.001) : The most impactful predictors were EDC system (β = 187.7, p < 0.001; G-EDC increased DPWH vs. L-EDC V4.0+), initiation year (β=-22.1, p < 0.001; DPWH decreased annually), unique CRF count (β = 3.21, p = 0.0003; more CRFs increased DPWH), and trial phase (β=-85.55, p = 0.0073; Phase III increased DPWH vs. Phase I). Together, these variables explained 61% of DPWH variation. GLD model (Adjusted R²=0.26, p < 0.001) : Only unique CRF count (β = 1.19, p = 0.0005), trial phase (β=-33.89, p = 0.0008), and initiation year (β=-10.96, p < 0.0001) were significant. These variables explained only 26% of GLD variation, indicating GLD is heavily influenced by unmeasured external factors. Table 5 Robust regression models for DPWH and GLD on workload factors Model Independent Variable β (Standardized Coefficient) p-value Adjusted R² DPWH Model Unique CRF Count 3.21 0.0003 0.61 Trial Phase (Phase I vs. III) -85.55 0.0073 EDC System (G-EDC vs. L-EDC V4.0+) 187.7 < 0.001 Initiation Year -22.1 < 0.001 GLD Model Unique CRF Count 1.19 0.0005 0.26 Trial Phase (Phase I vs. III) -33.89 0.0008 EDC System (G-EDC vs. L-EDC V4.0+) 8.93 0.45 Initiation Year -10.96 < 0.0001 3.5 Mixed-Effects Model (ICC Analysis) ICC values confirmed DPWH has better discriminatory power for efficiency differences within and across subgroups (Table 6 ): TA subgroups : DPWH had a higher ICC (0.39) than GLD (0.33), meaning DPWH is more consistent within the same TA and more distinct across TAs. Trial phase subgroups : DPWH also had a higher ICC (0.50) than GLD (0.37), reflecting stronger homogeneity within phases and clearer differences between phases. Table 6 Mixed-Effects Model (ICC Analysis) Metric Subgroup Category ICC DPWH TA 0.39 Trial Phase 0.5 GLD TA 0.33 Trial Phase 0.37 This indicates DPWH is less influenced by external factors and better captures subgroup-specific workload patterns. 4. Discussion The core objective of this study was to validate whether DPWH —a metric directly capturing data management team effort—outperforms GLD in assessing efficiency of DM or EDC. Results from 641 diverse clinical trials (spanning 2019–2025, multiple therapeutic areas, trial phases, and EDC systems) provide robust empirical evidence to support this claim, as detailed below. 4.1 Inherent Limitations of GLD: Why It Fails to Measure True DM Efficiency GLD’s popularity in the industry stems from its simplicity (easy to quantify and compare across projects), but our data expose three critical flaws that disconnect it from "true DM efficiency"—defined as the alignment between effort invested, trial complexity, and output quality. 4.1.1 GLD Is Uncoupled from Trial Complexity Trial complexity (driven by TA, phase, and EDC system type) is the primary determinant of DM workload, yet GLD shows no meaningful ability to distinguish between high- and low-complexity trials. For example: By trial phase : Phase III trials (the most complex, with larger sample sizes and stricter regulatory requirements) had a mean DPWH of 335.5 ± 201.4 hours—85% higher than Phase I trials (182.0 ± 115.4 hours, p < 0.0001). In contrast, GLD for Phase III (144.6 ± 70.2 days) was only moderately higher than Phase I (89.6 ± 52.4 days), and this difference was driven by non-DM delays (e.g., protocol amendments) rather than DM effort. For instance, Phase III oncology trials (mean DPWH: 401.7 ± 230.0 hours, indicating extreme DM workload) had a GLD of 89.5 days—nearly identical to Phase I clinical pharmacology trials (mean GLD: 85.2 days, p = 0.41), which required only 151.7 ± 81.5 hours of DM effort. By EDC system : G-EDC trials demand far more DM effort than Local EDC (L-EDC) trials (mean DPWH: 423.3 ± 209.2 vs. 178.6 ± 89.1 hours, p < 0.001). However, GLD for G-EDC (139.5 ± 80.1 days) was only 50% higher than L-EDC (93.6 ± 52.2 days)—a gap that vastly underestimates the 137% difference in DM workload. These findings align with prior industry observations (e.g., Getz et al. [ 20 ] reported minimal GLD differences between Phase I/II/III trials) and confirm that GLD cannot differentiate between trials requiring vastly different levels of DM effort. 4.1.2 GLD Is Dominated by Non-DM Delays A key marker of a valid efficiency metric is its ability to be explained by factors within the DM team’s control. Our robust regression analysis underscores GLD’s vulnerability to external, non-DM influences: Only 26% of GLD variation was explained by DM-relevant factors (CRF count, trial phase, initiation year; Table 5 ). The remaining 74% of variation likely stems from non-DM issues (e.g., protocol amendments, sponsor review bottlenecks, ethics committee feedback delays)—consistent with an industry survey finding that 50% of "long GLD" cases are caused by non-DM factors [ 19 , 20 ]. For example, two oncology trials with identical CRF counts (50 forms, indicating matched DM workload) had GLD values of 70 and 120 days—a 71% difference. Post-hoc review revealed the longer GLD was due to a 3-week delay in sponsor contract finalization, not DM inefficiency. GLD’s inability to filter out such external noise means it often penalizes or rewards DM teams for factors beyond their control. This disconnect arises because GLD is dominated by non-DM delays common in clinical research: protocol amendments (e.g., adding endpoints), sponsor review bottlenecks, or ethics committee feedback—factors that have no bearing on DM team performance [ 21 , 22 , 23 ]. This is further supported by an industry survey of “long GLD” trials, which found that 50% of delays stemmed from non-DM issues [ 24 , 25 ]. 4.1.3 GLD Misleads Resource Allocation Efficient resource planning requires metrics that link "timeline" to "workload," but GLD’s weak correlation with DM workload proxies (CRF count: r = 0.37; EC count: r = 0.38; Table 2 ) makes this impossible. For instance: Two trials with identical GLD (70 days) had DPWH values of 150 hours (30 unique CRFs) and 250 hours (60 unique CRFs)—a 67% difference in DM effort. If an organization relied on GLD alone, it would assign the same number of FTEs to both trials, leading to under-resourcing of the high-workload trial (60 CRFs) and potential delays in data cleaning or query resolution. This risk is amplified for complex trials: Phase III oncology trials require 2.4x more DPWH than Phase I trials, but their GLD is only 1.6x longer. Using GLD to set FTEs would systematically understaff high-complexity trials, undermining overall trial progress. 4.2 DPWH’s Advantages: Aligning Metric with DM Reality DPWH addresses GLD’s flaws by directly measuring "DM effort invested in EDC setup"—a factor entirely within the DM team’s control. Our data demonstrate four key advantages that make DPWH a superior efficiency metric. 4.2.1 DPWH Captures Trial Complexity Systematically Unlike GLD, DPWH varies in direct proportion to trial complexity—ensuring it reflects the actual demands placed on DM teams. This is evident across all categorical project characteristics: By TA : Oncology trials (the most data-intensive) had the highest DPWH (289.3 ± 190.0 hours), followed by Non-Oncology (214.7 ± 120.0 hours) and Clinical Pharmacology (151.9 ± 81.6 hours). All pairwise differences were statistically significant (p < 0.001), mirroring the higher CRF/EC demands of oncology trials (mean CRF count: 50.9 ± 7.7 vs. 30.7 ± 6.1 for Clinical Pharmacology; mean EC count: 837.0 ± 251.7 vs. 420.5 ± 189.3). By trial phase : DPWH increased monotonically with phase: Phase III (335.5 ± 201.4 hours) > Phase II (224.5 ± 131.4 hours) > Phase I (182.0 ± 115.4 hours). This aligns with regulatory expectations for late-phase trials (e.g., more endpoints, stricter data validation) and ensures DM effort is proportional to trial risk. By EDC system **: G-EDC trials required 2.4x more DPWH than L-EDC trials (423.3 ± 209.2 vs. 178.6 ± 89.1 hours, p < 0.001), reflecting the additional setup tasks required for the systems. This systematic variation means DPWH can be used to establish **evidence-based benchmarks** (e.g., "Phase III oncology trials using G-EDC require 250–270 DPWH hours"), a capability GLD lacks. 4.2.2 DPWH Strongly Correlates with Core DM Workload A valid DM efficiency metric must link to tangible DM tasks—specifically, CRF design and EC programming, which constitute ~ 70% of pre-go-live DM effort [unpublished]. Our correlation analysis confirms DPWH’s superiority in this regard: DPWH had a strong positive correlation with unique CRF count (r = 0.48, p < 0.0001) and EC count (r = 0.52, p < 0.0001; Table 2 ). For every additional CRF, DPWH increased by ~ 3 hours—directly reflecting the time required to design, test, and validate new data collection forms. In contrast, GLD’s correlations with these proxies were weak (CRF: r = 0.37; EC: r = 0.38). This means GLD cannot distinguish between a trial with 30 CRFs (low workload) and 60 CRFs (high workload) if their non-DM delays are similar—rendering it useless for assessing how DM teams handle core tasks. 4.2.3 DPWH Tracks Long-Term DM Efficiency Gains An effective metric should capture improvements in DM processes (e.g., EDC system upgrades, standardized CRF templates). Our data show DPWH is uniquely capable of this: DPWH had a strong negative correlation with initiation year (r=-0.43, p < 0.0001), with mean DPWH dropping from 285.3 ± 192.3 hours (2019) to 148.4 ± 57.5 hours (2024)—a 48% reduction over 5 years. This trend directly reflects efficiency gains from adopting L-EDC V4.0+ (an upgraded system with automation features), which reduced DPWH by ~ 15% for Phase I trials alone【18】. GLD, by contrast, had a weak negative correlation with initiation year (r=-0.23, p < 0.0001), with mean GLD decreasing only 34% (127.1 ± 76.1 days in 2019 to 84.9 ± 38.5 days in 2024). The smaller reduction occurred because DM efficiency gains were offset by rising trial complexity (e.g., new TAs or more advanced trial design methodology)—a tradeoff GLD cannot disentangle, but DPWH can. 4.2.4 DPWH Is More Reliable for Subgroup-Specific Efficiency Assessment The intraclass correlation coefficient (ICC) measures a metric’s ability to reflect homogeneous efficiency within subgroups (e.g., same TA or phase) and distinct differences between subgroups. Our mixed-effects model results confirm DPWH’s higher reliability (Table 6 ): For TA subgroups : DPWH had an ICC of 0.39, compared to 0.33 for GLD. This means DPWH is more consistent in measuring efficiency for trials within the same TA (e.g., all oncology trials) and better at distinguishing efficiency between TAs (e.g., oncology vs. clinical pharmacology). For trial phase subgroups : DPWH’s ICC (0.50) was 35% higher than GLD’s (0.37). A higher ICC indicates DPWH can more accurately identify which Phase III trials have efficient DM teams (lower DPWH relative to peers) versus which are inefficient—critical for targeted process improvement. 4.3 Direct Comparison of GLD and DPWH: A Summary of Key Differences To further contextualize DPWH’s superiority, Table 7 synthesizes the core performance metrics of GLD and DPWH using data from this study. Every criterion—from alignment with workload to reliability—confirms DPWH is a more valid measure of DM efficiency. Table 7 Comparison of GLD and DPWH on DM efficiency metrics Evaluation Criterion GLD DPWH Measurement Content Time from protocol finalization to EDC launch (speed-focused) Total DM work hours for EDC setup (workload-aligned) Susceptibility to Non-DM Delays High (e.g., protocol amendments, ethics reviews) Low (only captures DM-specific tasks) Association with Trial Complexity Weak (no meaningful differences by TA/phase/EDC system) Strong (varies systematically by TA/phase/EDC system) Correlation with Workload Proxies Weak (CRF: r = 0.37; EC: r = 0.38) Strong (CRF: r = 0.48; EC: r = 0.52) Ability to Reflect Efficiency Trends Poor (weak correlation with initiation year: r=-0.23) Excellent (strong correlation with initiation year: r=-0.43) Value for Resource Allocation Low (cannot distinguish workload differences) High (enables evidence-based benchmarking) Reliability (ICC for TA/Phase) Lower (TA: 0.33; Phase: 0.37) Higher (TA: 0.39; Phase: 0.50) Large-Scale Validation Limited (mostly survey-based) Validated (641 diverse trials) 4.4 Study Limitations and Future Directions While this study provides strong support for DPWH, three limitations should be noted to guide future research: Single-enterprise data : The 641 trials were from a single global pharmaceutical enterprise. Multi-center studies spanning contract research organizations (CROs) and academic institutions are needed to confirm DPWH’s generalizability across diverse operational models. Lack of post-go-live quality metrics : This study focused on pre-go-live efficiency, but future work should link DPWH to post-launch outcomes (e.g., data query rate, inspection findings). Preliminary industry data suggest trials with higher DPWH (reflecting thorough EC design) have fewer manual queries—a relationship that merits formal validation [ 26 , 27 , 28 ]. No integration of labor costs : DPWH measures hours but not cost (e.g., senior vs. junior DM staff). Adding cost data to create a "cost-per-DPWH" metric would enhance budget optimization, particularly for trials with constrained resources. 4.5 Practical Implications for the Clinical Research Industry The shift from GLD to DPWH has tangible benefits for all stakeholders: Pharmaceutical enterprises : DPWH enables more accurate resource allocation (e.g., assigning 2 FTEs to a Phase III oncology trial vs. 1 FTE to a Phase I trial) and better tracking of DM team performance. CROs : DPWH provides a transparent, workload-based metric for client reporting—reducing disputes over "delayed go-live" caused by non-DM factors. Regulatory Authority : DPWH’s alignment with data quality (via EC count correlation) supports regulatory expectations for reproducible, high-quality clinical data [ 14 ]. 5. Conclusion EDC go-live days is an unreliable metric for measuring DM efficiency across the global clinical research industry. It is disconnected from trial complexity, biased by non-DM delays, and unable to guide effective resource allocation. In contrast, EDC Design Phase Work Hours directly captures DM workload, correlates strongly with project characteristics (TA, phase, EDC system) and workload proxies (CRF count, EC count), and is explained by meaningful efficiency factors (adjusted R²=0.61 vs. 0.26 for GLD). DPWH also has higher reliability (ICC) and better tracks efficiency gains over time. Adopting DPWH as a universal DM efficiency metric can drive three key improvements: Accurate resource allocation Evidence-based DPWH benchmarks for different trial types ensure teams are appropriately staffed, avoiding under-resourcing of complex trials. Measurable efficiency gains DPWH quantifies the impact of EDC system upgrades, standardized workflows, and other process improvements. Quality-aligned assessment DPWH’s link to data quality supports regulatory compliance and reproducibility, aligning with global regulatory expectations. By shifting from "speed-focused GLD" to "workload-aligned DPWH," the clinical research industry can enhance efficiency, improve data quality, and optimize resource stewardship—ultimately accelerating the development of new therapies for patients worldwide. Declarations Funding: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Author Contribution sg collected and analysis the datahy and cy discussed the paper draft and data presentcy wrote the main manuscript text Acknowledgement The authors thank Dr. Yihua Zhao for assistance with data statistical analysis. Data Availability Dataset is available upone request References Ruth C, Huey S, Krisher J, et al. An Electronic Data Capture Framework (ConnEDCt) for Global and Public Health Research: Design and Implementation. J Med Internet Res 2020;22(8):e18580. https://doi.org/10.2196/18580 Pestronk M, Johnson D, Muthanna M, et al. Electronic Data Capture—Selecting an EDC System. J Soc Clin Data Manag 2021;1(1). https://doi.org/10.47912/jscdm.29 Vielhauer J, Mahajan UM, Adorjan K, et al. Electronic data capture in resource-limited settings using the lightweight clinical data acquisition and recording system. Sci Rep 2024;14:19056. https://doi.org/10.1038/s41598-024-69550-w Yan C. Data Management in Clinical Research. Beijing, Science Publisher; 2011. SCDM. Metrics in Clinical Data Management. J Soc Clin Data Manag 2023;1(1):1–9. https://doi.org/10.47912/jscdm.331 Zozus MN, Sanns W, Eisenstein E. Beyond EDC. J Soc Clin Data Manag 2021;1(1). https://doi.org/10.47912/jscdm.33 Datar M. Important Metrics in Clinical Data Management. Cloudbyz Resources; 2023. https://www.cloudbyz.com/resources/edc/10-important-metrics-in-clinical-data-management/ Pomerantseva V. Data Management Efficiencies Through Risk-Based Approaches and Innovations. Appl Clin Trials 2024. https://www.appliedclinicaltrialsonline.com/view/data-management-efficiencies-through-risk-based-approaches-and-innovations Helms RW, Fitzmartin R, Fillow PJ, et al. Metrics and Best Practices in Clinical Data Management: Conclusions of a Dia Roundtable Workshop. Ther Innov Regul Sci 2001;35:681–694. https://doi.org/10.1177/009286150103500306 Wilkinson M, Young R, Harper B, et al. Baseline assessment of the evolving 2019 eClinical Landscape. Ther Innov Regul Sci 2018;53(1):869–876. Harper B, Smith Z, Snowdon J, et al. Characterizing Pain Points in Clinical Data Management and Assessing the Impact of Mid-Study Updates. Ther Innov Regul Sci 2021;55:1006–1012. https://doi.org/10.1007/s43441-021-00301-z Harper B, Smith Z, Snowdon J, et al. Characterizing Clinical Data Management Challenges and Their Impact. Appl Clin Trials 2021;30(1). https://www.appliedclinicaltrialsonline.com/view/characterizing-clinical-data-management-challenges-and-their-impact Eade D, Pestronk M, Russo R, et al. Electronic Data Capture—Study Implementation and Start-up. J Soc Clin Data Manag 2021;4. https://doi.org/10.47912/jscdm.30 FDA. Guidance Document: Electronic Systems, Electronic Records, and Electronic Signatures in Clinical Investigations: Questions and Answers. 2024. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/electronic-systems-electronic-records-and-electronic-signatures-clinical-investigations-questions Yan C et al：DM-led EDC Building: A Quality-by-Design Framework Integrating SAI to Transform Clinical Data Management. 2025 submitted Yan C, Yan H, Gou S, Zhao Y，Zhou E, Wei X，Zhao S and Zhuang Y：Clinical Data Total Management System for Realizing Instant Inspection Readiness in Clinical Data Management 2025 China Food & Drug Administration Magazine (submitted) Yan C, Chen X, Bie L, Peng R, Gou S and Yan H：EDC Design Phase: Where Clinical Data Manager’s Time Goes. A Retrospective Analysis of Worktime Distribution and Efficiency Gains (2019–2021 vs. 2022–2024). 2025 (submitted) Yan C, Yan H, Zhou E, Wei X, Gou S, Zhao S: EDC Building Efficiency: Stratified SAI Contributions—Empirical Analysis of 491 Projects (2020–2025) via HRTAU EDC 2025 (submitted) Yan C, Wei X, Xin X, Yan H, Gou S, Zhao S, Zhuang Y: From Protocol to EDC: Unpacking SAI (Standardization/ Automation/ Intelligence) Technical Mechanisms for Efficient EDC Builds—2022–2024 Worktime Validation. 2025 (In preparation) Getz K, Smith Z, Kravet M. Protocol Design and Performance Benchmarks by Phase and by Oncology and Rare Disease Subgroups. Ther Innov Regul Sci 2023;57(1):49–56. https://doi.org/10.1007/s43441-022-00438-5 Rubio DM. Common metrics to assess the efficiency of clinical research. Eval Health Prof 2013;36(4):432–446. Walden A, Garza M, Rasmussen L. Best Practices for Research Data Management. In: Richesson RL, Andrews JE, Fultz Hollis K, eds. Clinical Research Informatics. Health Informatics. Springer Cham; 2023. https://doi.org/10.1007/978-3-031-27173-1_14 Bajpai N, Chatterjee A, Dang S, et al. Metrics for leveraging more in clinical data management: proof of concept in the context of vaccine trials in an Indian pharmaceutical company. Asian J Pharm Clin Res 2015;8(3):350–357. https://core.ac.uk/download/pdf/477851493.pdf Singh K, Abdul Salam M, Devarajan R, et al. Solving clinical trial delays: innovative solutions. Clin Investig (Lond) 2015;5(9):745–753. https://www.openaccessjournals.com/articles/solving-clinical-trial-delays-innovative-solutions.pdf Industry Survey Reveals Clinical Data Management Delays Slowing Trial Completion. Pharm Exec 2017. https://www.pharmexec.com/view/industry-survey-reveals-clinical-data-management-delays-slowing-trial-completion Walden A, Garza M, Rasmussen L. Best Practices for Research Data Management. In: Richesson RL, Andrews JE, Fultz Hollis K, eds. Clinical Research Informatics. Health Informatics. Springer Cham; 2023. https://doi.org/10.1007/978-3-031-27173-1_14 Rubio DM. Common metrics to assess the efficiency of clinical research. Eval Health Prof 2013;36(4):432–446. Bajpai N, Chatterjee A, Dang S, et al. Metrics for leveraging more in clinical data management: proof of concept in the context of vaccine trials in an Indian pharmaceutical company. Asian J Pharm Clin Res 2015;8(3):350–357. https://core.ac.uk/download/pdf/477851493.pdf Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7831647","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":528050260,"identity":"94973c50-f39a-4869-ba0b-de2911ac6c44","order_by":0,"name":"Charles Yan","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA9UlEQVRIiWNgGAWjYDACCSB+AGEyPobSDYS1JECYzMYka2GTJspd/LObn31IqKhjMDh+9lh1QcUdu+0zkhsYflRsw23JnWPGMxLOHGYwOJOXdnvGmWfJc24kNjD2nLmNU4uBRIIxQ2LbAQaDAzlmt3nbDidLSCQ2MDO24dOS/pkh8R/QYeffmBXz/iNKSw7QlgZmBoMbOWbMvA2H7QhqkbiRU8yQcOwwj+SNN8bSPMcOJ0jwPGw4iM8v/DPSNzN8qKmT4zufY/iZp+awvQR7+sMHPypwa4EBHoUDEEZiA5A4QFA9CMg3QGh7olSPglEwCkbBiAIAofVXoUkJUDQAAAAASUVORK5CYII=","orcid":"","institution":"Jiangsu Hengrui Medicine (China)","correspondingAuthor":true,"prefix":"","firstName":"Charles","middleName":"","lastName":"Yan","suffix":""},{"id":528050261,"identity":"796da7c5-5af3-42d6-87a4-4e060e195d0a","order_by":1,"name":"Shuyu Gou","email":"","orcid":"","institution":"Jiangsu Hengrui Medicine (China)","correspondingAuthor":false,"prefix":"","firstName":"Shuyu","middleName":"","lastName":"Gou","suffix":""},{"id":528050262,"identity":"0e2712db-3097-453e-8a69-9e8b1d75e7f9","order_by":2,"name":"Huaihai Yan","email":"","orcid":"","institution":"Jiangsu Hengrui Medicine (China)","correspondingAuthor":false,"prefix":"","firstName":"Huaihai","middleName":"","lastName":"Yan","suffix":""}],"badges":[],"createdAt":"2025-10-11 04:38:21","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7831647/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7831647/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":93461041,"identity":"73437dae-1d5f-4887-9182-3aa74bc8494b","added_by":"auto","created_at":"2025-10-14 06:20:40","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":55269,"visible":true,"origin":"","legend":"","description":"","filename":"RethinkingEDCGoLiveMetricsWhyEDCDesignPhaseWorkHours1.0.docx","url":"https://assets-eu.researchsquare.com/files/rs-7831647/v1/87c1362b7c1305192987a3fb.docx"},{"id":93462190,"identity":"cf56037a-fd90-4c8b-ad37-8d61832bd44f","added_by":"auto","created_at":"2025-10-14 06:28:40","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5095,"visible":true,"origin":"","legend":"","description":"","filename":"b2bb2966f04c475aa059661bbe49a30d.json","url":"https://assets-eu.researchsquare.com/files/rs-7831647/v1/cc0d94784cb16c1e00268ac2.json"},{"id":93461043,"identity":"64a04693-edd6-4456-adef-01122960f380","added_by":"auto","created_at":"2025-10-14 06:20:40","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":109855,"visible":true,"origin":"","legend":"","description":"","filename":"b2bb2966f04c475aa059661bbe49a30d1enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7831647/v1/495797a30018fae0f044341d.xml"},{"id":93461042,"identity":"bc03dc95-3542-4f3f-811c-0cce49aae4f6","added_by":"auto","created_at":"2025-10-14 06:20:40","extension":"xml","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":108327,"visible":true,"origin":"","legend":"","description":"","filename":"b2bb2966f04c475aa059661bbe49a30d1structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7831647/v1/cc8064a7f1716d2eb6d436b8.xml"},{"id":93461045,"identity":"46fff86e-8b0a-48c4-b45e-7b8e3c7e67d1","added_by":"auto","created_at":"2025-10-14 06:20:41","extension":"html","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":118645,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7831647/v1/6cfc4b40faaf1e17ff2c7965.html"},{"id":101079854,"identity":"54131e55-72dc-400e-8a05-b0e75e70acc5","added_by":"auto","created_at":"2026-01-25 12:54:38","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1656268,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7831647/v1/c2804a49-372f-49dc-b83b-2b5621b1e106.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Rethinking EDC Go-Live Metrics: Why EDC Design Phase Work Hours Outperform Days in Measuring DM Efficiency (641 Clinical Trials)","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003e\u003cstrong\u003e1.1 Background \u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eElectronic Data Capture (EDC) systems are indispensable for modern clinical research, enabling compliant, efficient data collection across all trial phases\u0026mdash;from early-phase clinical pharmacology studies to late-phase confirmatory trials [1,2,3]. As trials grow more complex (e.g., integrating biomarkers, global multi-center designs), timely EDC launch directly impacts overall trial progress, making \u0026quot;EDC go-live efficiency\u0026quot; a core priority for DM departments. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe most widely adopted metric for this efficiency is \u003cstrong\u003eEDC go-live days (GLD)\u003c/strong\u003e, defined as the duration from final protocol approval to EDC system activation [4,5,6]. GLD\u0026rsquo;s popularity stems from its simplicity: it is easy to quantify and compare across projects, aligning with the timeline pressures of clinical development (e.g., accelerating trial initiation to meet regulatory milestones) [7,8,9]. However, GLD has critical flaws that undermine its validity as a true measure of DM efficiency: \u0026nbsp;\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003e\u003cstrong\u003eSusceptibility to non-DM delays\u003c/strong\u003e: GLD is heavily influenced by external factors unrelated to DM performance, such as protocol amendments, sponsor review bottlenecks, ethics committee feedback timelines, and IT system downtime. For example, a complex oncology trial may have prolonged GLD due to repeated protocol revisions\u0026mdash;even if the DM team efficiently completes EDC setup. \u0026nbsp;\u003c/li\u003e\n \u003cli\u003e\u003cstrong\u003eFailure to reflect trial complexity:\u003c/strong\u003e A Phase I trial with 30 case report forms (CRFs) and 300 edit checks (ECs) may have the same GLD as a Phase III trial with 60 CRFs and 1200 ECs, despite the latter requiring twice the DM effort. \u0026nbsp;\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eTo address these shortcomings, \u003cstrong\u003eEDC Design Phase Work Hours (DPWH)\u003c/strong\u003e has emerged as a promising alternative. DPWH is defined as the total labor hours invested in DM-specific EDC setup tasks during the design phase, encompassing tangible activities: CRF design, EC programming, user acceptance testing (UAT), data management plan (DMP) development, CRF completion guide creation, and database quality control (QC) [unpublished]. Unlike GLD, DPWH directly captures DM workload and is inherently tied to trial complexity, avoiding distortions from non-DM factors. However, few studies have validated DPWH using large-scale, diverse trial datasets, nor systematically compared it to GLD across key project characteristics (e.g., therapeutic area [TA], trial phase, EDC system). \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e1.2 Research Gap and Objective \u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eExisting literature on EDC efficiency metrics has three critical limitations for the global clinical research industry[10,11,12]: \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e1. \u003cstrong\u003eOverreliance on GLD\u003c/strong\u003e: Most studies focus on GLD or process optimization (e.g., automating ECs) but neglect workload-based metrics like DPWH. Even multi-stakeholder studies (involving CROs, sponsors, and academic institutions) rely on descriptive survey data rather than quantifiable empirical analysis, failing to link \u0026quot;efficiency assessment\u0026quot; to \u0026quot;actual DM effort\u0026quot;. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e2. \u003cstrong\u003eSmall sample sizes:\u003c/strong\u003e Limited samples fail to capture variability across TA, phase, and EDC systems\u0026mdash;key drivers of workload. This also prevents tracking long-term trends: while EDC systems have advanced (e.g., automated ECs, standardized CRF templates), survey data shows GLD has remained stable at ~70 days for years, providing no insight into technological progress. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e3. \u003cstrong\u003eLack of workload linkage\u003c/strong\u003e: No studies explicitly connect efficiency metrics to core workload proxies (e.g., CRF count, EC count). For GLD, this means the \u0026quot;stable 70-day GLD\u0026quot; cannot distinguish between \u0026quot;DM efficiency gains offset by complex trials\u0026quot; and \u0026quot;genuine stagnation,\u0026quot; making it impossible to validate if GLD responds to workflow or system improvements [13]. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo fill these gaps, this study uses data from 641 diverse **\u003cstrong\u003einnovative drug clinical trials of Hengrui Pharmaceutical\u003c/strong\u003e**\u0026mdash;all managed by the Clinical Data Science Center of Hengrui and completed EDC go-live by June 30, 2025 (spanning 2019\u0026ndash;2025)\u0026mdash;to pursue two core objectives: \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e1. Assess the associations of GLD and DPWH with key project characteristics (therapeutic area [TA], trial phase, EDC system/version, initiation year) across this broad, representative range of trials; \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e2. Compare GLD and DPWH in their ability to correlate with DM workload proxies (unique CRF count, edit check [EC] count), and verify if DPWH outperforms GLD as a universal DM efficiency metric. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eNotably, the Clinical Data Science Center of Hengrui adopts a distinct operational model: it has no specialized EDC database build (DBB) personnel, with data managers (DMs) directly leading all EDC setup tasks from trial start. This eliminates the usual communication/coordination costs between DMs (who define data needs) and separate DBB staff in conventional models\u0026mdash;allowing us to capture EDC build time with high accuracy, which strengthens DPWH\u0026rsquo;s reliability (critical for a workload-dependent metric) and ensures DPWH reflects real DM effort (not cross-role inefficiencies), boosting the persuasiveness of our GLD-DPWH comparisons.\u003c/p\u003e\n\u003cp\u003eFurther, this DM-led EDC setup aligns with ICH GCP E6R3\u0026rsquo;s \u0026quot;Quality by Design (QbD)\u0026quot; principle: as stakeholders most familiar with data quality requirements, DMs embed integrity and compliance upfront during build, rather than retrofitting later. It also represents a concrete practice of transitioning from traditional Clinical Data Management (CDM) to Clinical Data Science (CDS)\u0026mdash;empowering DMs to integrate data management expertise with technical EDC capabilities, moving beyond transactional tasks to drive data-centric trial efficiency【14】.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e1.3 Significance \u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study contributes to the global clinical research industry in three key ways: \u0026nbsp;\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003e\u003cstrong\u003eMethodological\u003c/strong\u003e: It validates DPWH\u0026mdash;a workload-based metric that addresses GLD\u0026rsquo;s limitations\u0026mdash;providing a more accurate measure of DM efficiency. \u0026nbsp;\u003c/li\u003e\n \u003cli\u003e\u003cstrong\u003ePractical\u003c/strong\u003e: It offers actionable insights for resource allocation (e.g., establishing DPWH benchmarks for oncology vs. non-oncology trials) and process optimization (e.g., using DPWH to quantify the impact of EDC system upgrades). \u0026nbsp;\u003c/li\u003e\n \u003cli\u003e\u0026nbsp;\u003cstrong\u003eIndustry-wide\u003c/strong\u003e: It supports a paradigm shift from \u0026quot;speed-focused\u0026quot; (GLD) to \u0026quot;workload-aligned\u0026quot; (DPWH) efficiency assessment, aligning with regulatory expectations for data quality and reproducibility [15]. \u0026nbsp;\u003c/li\u003e\n\u003c/ul\u003e"},{"header":"2. Materials and Methods","content":"\u003cp\u003e\u003cstrong\u003e2.1 Data Source \u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData were extracted from the Clinical Data Total Management System (CDTMS) of a global pharmaceutical enterprise, covering 641 innovative drug clinical trials managed by the Clinical Data Science Center (CDSC) [16,17] . All trials completed EDC go-live between January 1, 2019, and June 30, 2025, and exhibited broad diversity: \u0026nbsp;\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003e\u003cstrong\u003eTherapeutic areas (TA):\u003c/strong\u003e Oncology, cardiology, neurology, diabetes, analgesia, autoimmune diseases, and clinical pharmacology; \u0026nbsp;\u0026nbsp;\u003c/li\u003e\n \u003cli\u003e\u003cstrong\u003eTrial phases\u003c/strong\u003e: Phase I, Phase II, and Phase III; \u0026nbsp;\u0026nbsp;\u003c/li\u003e\n \u003cli\u003e\u0026nbsp;\u003cstrong\u003eInvestigational products\u003c/strong\u003e: Innovative drugs: mall-molecule drugs and biologic candidates. \u0026nbsp;\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eThis diversity ensures the sample is representative of global clinical research, enhancing the generalizability of results. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.2 DPWH Tracking timesheet\u003c/strong\u003e:\u003c/p\u003e\n\u003cp\u003eDM Recorded task-specific labor data daily for predefined task categories with 15-minute increments. For each project, DPWH was calculated by summing labor inputs in EDC designed phase from all participating CDMs, reflecting the collective effort of the team [16];\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.3 Variable Definition \u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll variables were operationalized based on industry-standard SOPs (aligned with ICH-GCP and CDISC guidelines) and EDC system logs to ensure consistency and reproducibility (Table 1).\u003c/p\u003e\n\u003cp\u003eTable 1. \u0026nbsp;Variable category, name definition and their measurement tools\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"614\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eVariable Category\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eVariable Name\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eDefinition\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 198px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMeasurement Tool\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 89px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eEfficiency Metrics\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eGLD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eTime from protocol finalization (signed by sponsor) to EDC system activation for patient enrollment.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 198px;\"\u003e\n \u003cp\u003eCalculation:\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eApproval date of EDC go-live - approval date of protocol version (V1.0)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eDPWH\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eTotal DM work hours for EDC setup during the design phase, including:\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u0026bull; \u0026nbsp;CRF design and EDC setup\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u0026bull; \u0026nbsp;EC development and UAT \u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u0026bull; \u0026nbsp;CRF Completion Guide development\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u0026bull; \u0026nbsp;Database QC\u003c/p\u003e\n \u003cp\u003e\u0026bull; \u0026nbsp;Data Management Plan, etc\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 198px;\"\u003e\n \u003cp\u003eCDTMS: Timesheet logs\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"3\" style=\"width: 89px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eEDCs\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\" style=\"width: 71px;\"\u003e\n \u003cp\u003eEDC System/ Version\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eTwo system types used across trials:\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\" style=\"width: 198px;\"\u003e\n \u003cp\u003eCDTMS：Trial EDC system used\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003e\u0026bull; Local EDC (L-EDC): Regional system, subdivided into L-EDC V284 (legacy version) and L-EDC V4.0+/V4.1.x (upgraded version with advanced features: standardization, automation and intelligence)[18,19]\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003e\u0026bull; Global EDC (G-EDC): Enterprise-wide system for all type of trials\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 89px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eWorkload Proxies\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eUnique CRF Count\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eNumber of unique valid data collection forms, aligned with CDISC CDASH standards.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 198px;\"\u003e\n \u003cp\u003eEDC system report\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003eEC Count\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eNumber of active edit checks at EDC go-live, includes range checks, consistency checks, and protocol-specific logic.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 198px;\"\u003e\n \u003cp\u003eEDC system report\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003e2.4 Statistical Analysis \u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eStatistical analyses were performed using SAS (SAS Institute Inc., Cary, NC, USA), with a significance threshold of p\u0026lt;0.05. The analytical approach followed a sequential, \u0026quot;layered\u0026quot; logic to validate DPWH: \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e1. Descriptive Statistics: \u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp;- Categorical variables (TA, trial phase, EDC system) were summarized as frequencies and percentages; \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp;- Continuous variables (GLD, DPWH, CRF count, EC count) were summarized as mean \u0026plusmn; standard deviation (SD) and median (interquartile range [IQR]) to account for outliers. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2. Correlation Analysis: \u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp;- Spearman\u0026rsquo;s rank correlation coefficients (r) were calculated to assess associations between: \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026bull; GLD/DPWH and continuous project characteristics (initiation year); \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026bull; GLD/DPWH and workload proxies (CRF count, EC count); \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026bull; GLD and DPWH (to test their linear relationship, pre-specified as a key analysis\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3. Stratified Comparison: \u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp;- Kruskal-Wallis H tests were used to compare DPWH across categorical project characteristics (TA, trial phase, EDC system), with Dunn\u0026rsquo;s post-hoc procedure (Bonferroni adjustment) for pairwise comparisons. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4. Robust Regression: \u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp;- Two robust regression models were built: \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026bull; Model 1: DPWH as the dependent variable, with independent variables (dummy-\u003c/p\u003e\n\u003cp\u003ecoded where applicable): TA (reference: Clinical Pharmacology), trial phase (reference: Phase I), EDC system (reference: L-EDC V4.0+), initiation year, CRF count, EC count; \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026bull; Model 2: GLD as the dependent variable, with the same independent variables. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp;- Model fit was evaluated using adjusted R\u0026sup2;, with significance assessed via standardized\u003c/p\u003e\n\u003cp\u003ecoefficients (\u0026beta;) and p-values. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e5. Mixed-Effects Models: \u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp;- Intraclass Correlation Coefficients (ICC) were calculated to assess homogeneity of projects within subgroups (TA, trial phase). Higher ICC indicates greater within-group homogeneity, reflecting better discriminatory power for efficiency differences. \u0026nbsp;\u003c/p\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003e3.1 Project Basic Characteristics\u003c/h2\u003e\u003cp\u003eThe 641 trials exhibited diverse characteristics, representative of global clinical research (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Phase I trials were most common (57.9%), followed by Phase II (24.8%) and Phase III (17.3%). TA distribution was balanced: Clinical Pharmacology (32.3%), Oncology (33.1%), and Non-Oncology (34.6%). L-EDC was the dominant system (83.5%), with 70.7% using the upgraded V4.0\u0026thinsp;+\u0026thinsp;version.\u003c/p\u003e\u003cp\u003eMean GLD was 101.2\u0026thinsp;\u0026plusmn;\u0026thinsp;60.1 days (median: 89 days), and mean DPWH was 219.1\u0026thinsp;\u0026plusmn;\u0026thinsp;148.6 hours (median: 176.7 hours). Workload proxies also varied by TA: Oncology trials had higher mean CRF count (50.9\u0026thinsp;\u0026plusmn;\u0026thinsp;7.7) and EC count (837.0\u0026thinsp;\u0026plusmn;\u0026thinsp;251.7) than Clinical Pharmacology trials (CRF: 30.7\u0026thinsp;\u0026plusmn;\u0026thinsp;6.1; EC:356.3\u0026thinsp;\u0026plusmn;\u0026thinsp;126.0), consistent with their higher complexity.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eBasic Characteristics of 641 Clinical Trials, along with their DPWH and GLD\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"6\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCharacteristic\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eCategory\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eFrequency (n)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003ePercentage (%)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eDPWH (Mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD; Median)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eGLD (Mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD; Median)\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e\u003cp\u003eTrial Phase\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePhase I\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e371\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e57.9%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e182.0\u0026thinsp;\u0026plusmn;\u0026thinsp;115.4;\u003c/p\u003e\u003cp\u003e152.8\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e89.6\u0026thinsp;\u0026plusmn;\u0026thinsp;52.4;\u003c/p\u003e\u003cp\u003e79\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePhase II\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e159\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e24.8%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e224.5\u0026thinsp;\u0026plusmn;\u0026thinsp;131.4;\u003c/p\u003e\u003cp\u003e190.8\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e98.0\u0026thinsp;\u0026plusmn;\u0026thinsp;56.0;\u003c/p\u003e\u003cp\u003e88\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePhase III\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e111\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e17.3%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e335.5\u0026thinsp;\u0026plusmn;\u0026thinsp;201.4;\u003c/p\u003e\u003cp\u003e264.6\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e144.6\u0026thinsp;\u0026plusmn;\u0026thinsp;70.2;\u003c/p\u003e\u003cp\u003e128\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e\u003cp\u003eTherapeutic Area (TA)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eClinical Pharmacology\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e207\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e32.3%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e151.9\u0026thinsp;\u0026plusmn;\u0026thinsp;81.6;\u003c/p\u003e\u003cp\u003e135.7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e83.8\u0026thinsp;\u0026plusmn;\u0026thinsp;54.0;\u003c/p\u003e\u003cp\u003e72\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eOncology\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e212\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e33.1%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e289.3\u0026thinsp;\u0026plusmn;\u0026thinsp;190.0;\u003c/p\u003e\u003cp\u003e228.9\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e121.9\u0026thinsp;\u0026plusmn;\u0026thinsp;69.1;\u003c/p\u003e\u003cp\u003e109\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNon-Oncology\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e222\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e34.6%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e214.7\u0026thinsp;\u0026plusmn;\u0026thinsp;120.0;\u003c/p\u003e\u003cp\u003e187.5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e97.7\u0026thinsp;\u0026plusmn;\u0026thinsp;49.8;\u003c/p\u003e\u003cp\u003e87\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003eEDC System\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eL-EDC\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e535\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e83.5%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e178.6\u0026thinsp;\u0026plusmn;\u0026thinsp;89.1;\u003c/p\u003e\u003cp\u003e158.3\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e93.6\u0026thinsp;\u0026plusmn;\u0026thinsp;52.2;\u003c/p\u003e\u003cp\u003e84\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eG-EDC\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e106\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e16.5%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e423.3\u0026thinsp;\u0026plusmn;\u0026thinsp;209.2;\u003c/p\u003e\u003cp\u003e382.9\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e139.5\u0026thinsp;\u0026plusmn;\u0026thinsp;80.1;\u003c/p\u003e\u003cp\u003e117\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"6\" rowspan=\"7\"\u003e\u003cp\u003eInitiation Year\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2019\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e75\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e11.7%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e285.3\u0026thinsp;\u0026plusmn;\u0026thinsp;192.3;\u003c/p\u003e\u003cp\u003e217.7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e127.1\u0026thinsp;\u0026plusmn;\u0026thinsp;76.1;\u003c/p\u003e\u003cp\u003e113\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2020\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e83\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e12.9%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e295.0\u0026thinsp;\u0026plusmn;\u0026thinsp;193.2;\u003c/p\u003e\u003cp\u003e263.7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e116.2\u0026thinsp;\u0026plusmn;\u0026thinsp;81.7;\u003c/p\u003e\u003cp\u003e95\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2021\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e117\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e18.3%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e300.0\u0026thinsp;\u0026plusmn;\u0026thinsp;177.0;\u003c/p\u003e\u003cp\u003e266.5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e109.6\u0026thinsp;\u0026plusmn;\u0026thinsp;64.7;\u003c/p\u003e\u003cp\u003e96\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2022\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e93\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e14.5%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e202.3\u0026thinsp;\u0026plusmn;\u0026thinsp;95.5;\u003c/p\u003e\u003cp\u003e185.8\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e105.3\u0026thinsp;\u0026plusmn;\u0026thinsp;56.7;\u003c/p\u003e\u003cp\u003e96\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2023\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e97\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e15.1%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e157.9\u0026thinsp;\u0026plusmn;\u0026thinsp;60.8;\u003c/p\u003e\u003cp\u003e150.0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e87.6\u0026thinsp;\u0026plusmn;\u0026thinsp;40.8;\u003c/p\u003e\u003cp\u003e87\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2024\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e151\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e23.6%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e148.4\u0026thinsp;\u0026plusmn;\u0026thinsp;57.5;\u003c/p\u003e\u003cp\u003e141.7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e84.9\u0026thinsp;\u0026plusmn;\u0026thinsp;38.5;\u003c/p\u003e\u003cp\u003e80\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2025\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e25\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3.9%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e114.4\u0026thinsp;\u0026plusmn;\u0026thinsp;28;\u003c/p\u003e\u003cp\u003e114.8\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e70.8\u0026thinsp;\u0026plusmn;\u0026thinsp;23.9;\u003c/p\u003e\u003cp\u003e70\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEfficiency Metrics\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTotal\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e641\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e100%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e219.1\u0026thinsp;\u0026plusmn;\u0026thinsp;148.6;\u003c/p\u003e\u003cp\u003e176.7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e101.2\u0026thinsp;\u0026plusmn;\u0026thinsp;60.1;\u003c/p\u003e\u003cp\u003e89\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003e3.2 Association Between GLD/DPWH and Project Characteristics\u003c/h2\u003e\u003cdiv id=\"Sec13\" class=\"Section3\"\u003e\u003ch2\u003e3.2.1 Overall Correlations\u003c/h2\u003e\u003cp\u003eSpearman\u0026rsquo;s analysis revealed key differences in how GLD and DPWH correlated with project characteristics (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Notably, GLD and DPWH had a moderate positive correlation (r\u0026thinsp;=\u0026thinsp;0.47, p\u0026thinsp;\u0026lt;\u0026thinsp;0.005), indicating some overlap but not redundancy.\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eInitiation year\u003c/b\u003e: GLD had a weak negative correlation (r=-0.23, p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), meaning EDC workflow optimizations (e.g., system upgrades) did not translate to meaningful GLD reductions\u0026mdash;likely offset by rising trial complexity. In contrast, DPWH had a strong negative correlation (r=-0.43, p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), reflecting a clear downward trend in workload over time (e.g., 2019: 285.3\u0026thinsp;\u0026plusmn;\u0026thinsp;192.3 hours vs. 2024: 148.4\u0026thinsp;\u0026plusmn;\u0026thinsp;57.5 hours), consistent with efficiency gains from advanced EDC tools.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eSpearman's rank correlation coefficients Between GLD/DPWH and Continuous Project Characteristics\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eVariables\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eGLD (\u003cem\u003er\u003c/em\u003e,\u0026nbsp;\u003cem\u003ep\u003c/em\u003e-value)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eDPWH (\u003cem\u003er\u003c/em\u003e,\u0026nbsp;\u003cem\u003ep\u003c/em\u003e-value)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eGLD vs. DPWH\u003c/p\u003e\u003cp\u003e(r, p-value)\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eInitiation Year\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e(-0.23, \u0026lt;\u0026thinsp;0.0001)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e(-0.43, \u0026lt;\u0026thinsp;0.0001)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eUnique CRF Count\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e(0.37, \u0026lt;\u0026thinsp;0.0001)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e(0.48, \u0026lt;\u0026thinsp;0.0001)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEC Count\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e(0.38, \u0026lt;\u0026thinsp;0.0001)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e(0.52, \u0026lt;\u0026thinsp;0.0001)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDPWH\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e(0.47, \u0026lt;\u0026thinsp;0.005)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section3\"\u003e\u003ch2\u003e3.2.2 DPWH Differences Across Categorical Characteristics\u003c/h2\u003e\u003cp\u003eKruskal-Wallis tests confirmed DPWH varied systematically by TA, trial phase, and EDC system\u0026mdash;capturing trial complexity (Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e):\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eTA\u003c/b\u003e: Oncology trials had the highest DPWH (289.3\u0026thinsp;\u0026plusmn;\u0026thinsp;190.0 hours), followed by Non-Oncology (214.7\u0026thinsp;\u0026plusmn;\u0026thinsp;120.0 hours) and Clinical Pharmacology (151.9\u0026thinsp;\u0026plusmn;\u0026thinsp;81.6 hours). All pairwise differences were significant (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), reflecting the higher CRF/EC demands of oncology trials (e.g., biomarker data forms, toxicity monitoring ECs).\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eTrial phase\u003c/b\u003e: DPWH increased with phase: Phase III (335.5\u0026thinsp;\u0026plusmn;\u0026thinsp;201.4 hours)\u0026thinsp;\u0026gt;\u0026thinsp;Phase II (224.5\u0026thinsp;\u0026plusmn;\u0026thinsp;131.4 hours)\u0026thinsp;\u0026gt;\u0026thinsp;Phase I (182.0\u0026thinsp;\u0026plusmn;\u0026thinsp;115.4 hours). Phase III trials required 85% more DPWH than Phase I (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), consistent with larger sample sizes, more endpoints, and stricter regulatory requirements.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eEDC system\u003c/b\u003e: G-EDC trials had significantly higher DPWH than L-EDC trials (423.3\u0026thinsp;\u0026plusmn;\u0026thinsp;209.2 vs. 178.6\u0026thinsp;\u0026plusmn;\u0026thinsp;89.1 hours, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), due to additional setup demands of G-EDC.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eKruskal-Wallis tests for DPWH with TA, trial phase, and EDC system\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCategorical Characteristic\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eComparison Group\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eDPWH (Mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003ep-value\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e\u003cp\u003eTA\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eOncology vs. Clinical Pharmacology\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e289.3\u0026thinsp;\u0026plusmn;\u0026thinsp;190.0 vs. 151.9\u0026thinsp;\u0026plusmn;\u0026thinsp;81.6\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNon-Oncology vs. Clinical Pharmacology\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e214.7\u0026thinsp;\u0026plusmn;\u0026thinsp;120.0 vs. 151.9\u0026thinsp;\u0026plusmn;\u0026thinsp;81.6\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eOncology vs. Non-Oncology\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e289.3\u0026thinsp;\u0026plusmn;\u0026thinsp;190.0 vs. 214.7\u0026thinsp;\u0026plusmn;\u0026thinsp;120.0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e\u003cp\u003eTrial Phase\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePhase III vs. Phase I\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e335.5\u0026thinsp;\u0026plusmn;\u0026thinsp;201.4 vs. 182.0\u0026thinsp;\u0026plusmn;\u0026thinsp;115.4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePhase III vs. Phase II\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e335.5\u0026thinsp;\u0026plusmn;\u0026thinsp;201.4 vs. 224.5\u0026thinsp;\u0026plusmn;\u0026thinsp;131.4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePhase II vs. Phase I\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e224.5\u0026thinsp;\u0026plusmn;\u0026thinsp;131.4 vs. 182.0\u0026thinsp;\u0026plusmn;\u0026thinsp;115.4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEDC System\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eG-EDC vs. L-EDC\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e423.3\u0026thinsp;\u0026plusmn;\u0026thinsp;209.2 vs. 178.6\u0026thinsp;\u0026plusmn;\u0026thinsp;89.1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\u003ch2\u003e3.3 Association Between GLD/DPWH and Workload Proxies\u003c/h2\u003e\u003cp\u003eGLD had weak correlations with both workload proxies, while DPWH had strong correlations (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e):\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eUnique CRF count: GLD (r\u0026thinsp;=\u0026thinsp;0.37, p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001) vs. DPWH (r\u0026thinsp;=\u0026thinsp;0.48, p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001). Each additional CRF was associated with ~\u0026thinsp;3 more DPWH hours.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eEC count: GLD (r\u0026thinsp;=\u0026thinsp;0.38, p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001) vs. DPWH (r\u0026thinsp;=\u0026thinsp;0.52, p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001).\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003eFor example, two Oncology trials with identical CRF counts (50 forms) had GLD of 70 and 120 days\u0026mdash;differences driven by non-DM factors (e.g., protocol amendments)\u0026mdash;but DPWH of 290 and 305 hours (minimal variation, reflecting similar workload). This confirms DPWH directly reflects core DM effort, while GLD does not.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e\u003ch2\u003e3.4 Robust Regression Results\u003c/h2\u003e\u003cp\u003eRobust regression models highlighted DPWH\u0026rsquo;s superior ability to be explained by meaningful workload factors (Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e):\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eDPWH model (Adjusted R\u0026sup2;=0.61, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001)\u003c/b\u003e: The most impactful predictors were EDC system (β\u0026thinsp;=\u0026thinsp;187.7, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001; G-EDC increased DPWH vs. L-EDC V4.0+), initiation year (β=-22.1, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001; DPWH decreased annually), unique CRF count (β\u0026thinsp;=\u0026thinsp;3.21, p\u0026thinsp;=\u0026thinsp;0.0003; more CRFs increased DPWH), and trial phase (β=-85.55, p\u0026thinsp;=\u0026thinsp;0.0073; Phase III increased DPWH vs. Phase I). Together, these variables explained 61% of DPWH variation.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eGLD model (Adjusted R\u0026sup2;=0.26, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001)\u003c/b\u003e: Only unique CRF count (β\u0026thinsp;=\u0026thinsp;1.19, p\u0026thinsp;=\u0026thinsp;0.0005), trial phase (β=-33.89, p\u0026thinsp;=\u0026thinsp;0.0008), and initiation year (β=-10.96, p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001) were significant. These variables explained only 26% of GLD variation, indicating GLD is heavily influenced by unmeasured external factors.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eRobust regression models for DPWH and GLD on workload factors\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eModel\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eIndependent Variable\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eβ (Standardized Coefficient)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003ep-value\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eAdjusted R\u0026sup2;\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e\u003cp\u003eDPWH Model\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eUnique CRF Count\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e3.21\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.0003\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\" morerows=\"3\" rowspan=\"4\"\u003e\u003cp\u003e0.61\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTrial Phase (Phase I vs. III)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e-85.55\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.0073\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eEDC System (G-EDC vs. L-EDC V4.0+)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e187.7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eInitiation Year\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e-22.1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e\u003cp\u003eGLD Model\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eUnique CRF Count\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1.19\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.0005\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\" morerows=\"3\" rowspan=\"4\"\u003e\u003cp\u003e0.26\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTrial Phase (Phase I vs. III)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e-33.89\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.0008\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eEDC System (G-EDC vs. L-EDC V4.0+)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e8.93\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.45\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eInitiation Year\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e-10.96\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.0001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec17\" class=\"Section2\"\u003e\u003ch2\u003e3.5 Mixed-Effects Model (ICC Analysis)\u003c/h2\u003e\u003cp\u003eICC values confirmed DPWH has better discriminatory power for efficiency differences within and across subgroups (Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e6\u003c/span\u003e):\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eTA subgroups\u003c/b\u003e: DPWH had a higher ICC (0.39) than GLD (0.33), meaning DPWH is more consistent within the same TA and more distinct across TAs.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eTrial phase subgroups\u003c/b\u003e: DPWH also had a higher ICC (0.50) than GLD (0.37), reflecting stronger homogeneity within phases and clearer differences between phases.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab6\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 6\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eMixed-Effects Model (ICC Analysis)\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"3\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMetric\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSubgroup Category\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eICC\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003eDPWH\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTA\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.39\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTrial Phase\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.5\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003eGLD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTA\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.33\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTrial Phase\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.37\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eThis indicates DPWH is less influenced by external factors and better captures subgroup-specific workload patterns.\u003c/p\u003e\u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eThe core objective of this study was to validate whether \u003cb\u003eDPWH\u003c/b\u003e\u0026mdash;a metric directly capturing data management team effort\u0026mdash;outperforms \u003cb\u003eGLD\u003c/b\u003e in assessing efficiency of DM or EDC. Results from 641 diverse clinical trials (spanning 2019\u0026ndash;2025, multiple therapeutic areas, trial phases, and EDC systems) provide robust empirical evidence to support this claim, as detailed below.\u003c/p\u003e\u003cdiv id=\"Sec19\" class=\"Section2\"\u003e\u003ch2\u003e4.1 Inherent Limitations of GLD: Why It Fails to Measure True DM Efficiency\u003c/h2\u003e\u003cp\u003eGLD\u0026rsquo;s popularity in the industry stems from its simplicity (easy to quantify and compare across projects), but our data expose three critical flaws that disconnect it from \"true DM efficiency\"\u0026mdash;defined as the alignment between effort invested, trial complexity, and output quality.\u003c/p\u003e\u003cdiv id=\"Sec20\" class=\"Section3\"\u003e\u003ch2\u003e4.1.1 GLD Is Uncoupled from Trial Complexity\u003c/h2\u003e\u003cp\u003eTrial complexity (driven by TA, phase, and EDC system type) is the primary determinant of DM workload, yet GLD shows no meaningful ability to distinguish between high- and low-complexity trials. For example:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eBy trial phase\u003c/b\u003e: Phase III trials (the most complex, with larger sample sizes and stricter regulatory requirements) had a mean DPWH of 335.5\u0026thinsp;\u0026plusmn;\u0026thinsp;201.4 hours\u0026mdash;85% higher than Phase I trials (182.0\u0026thinsp;\u0026plusmn;\u0026thinsp;115.4 hours, p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001). In contrast, GLD for Phase III (144.6\u0026thinsp;\u0026plusmn;\u0026thinsp;70.2 days) was only moderately higher than Phase I (89.6\u0026thinsp;\u0026plusmn;\u0026thinsp;52.4 days), and this difference was driven by non-DM delays (e.g., protocol amendments) rather than DM effort. For instance, Phase III oncology trials (mean DPWH: 401.7\u0026thinsp;\u0026plusmn;\u0026thinsp;230.0 hours, indicating extreme DM workload) had a GLD of 89.5 days\u0026mdash;nearly identical to Phase I clinical pharmacology trials (mean GLD: 85.2 days, p\u0026thinsp;=\u0026thinsp;0.41), which required only 151.7\u0026thinsp;\u0026plusmn;\u0026thinsp;81.5 hours of DM effort.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eBy EDC system\u003c/b\u003e: G-EDC trials demand far more DM effort than Local EDC (L-EDC) trials (mean DPWH: 423.3\u0026thinsp;\u0026plusmn;\u0026thinsp;209.2 vs. 178.6\u0026thinsp;\u0026plusmn;\u0026thinsp;89.1 hours, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). However, GLD for G-EDC (139.5\u0026thinsp;\u0026plusmn;\u0026thinsp;80.1 days) was only 50% higher than L-EDC (93.6\u0026thinsp;\u0026plusmn;\u0026thinsp;52.2 days)\u0026mdash;a gap that vastly underestimates the 137% difference in DM workload.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003eThese findings align with prior industry observations (e.g., Getz et al. [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e] reported minimal GLD differences between Phase I/II/III trials) and confirm that GLD cannot differentiate between trials requiring vastly different levels of DM effort.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec21\" class=\"Section3\"\u003e\u003ch2\u003e4.1.2 GLD Is Dominated by Non-DM Delays\u003c/h2\u003e\u003cp\u003eA key marker of a valid efficiency metric is its ability to be explained by factors within the DM team\u0026rsquo;s control. Our robust regression analysis underscores GLD\u0026rsquo;s vulnerability to external, non-DM influences:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eOnly 26% of GLD variation was explained by DM-relevant factors (CRF count, trial phase, initiation year; Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). The remaining 74% of variation likely stems from non-DM issues (e.g., protocol amendments, sponsor review bottlenecks, ethics committee feedback delays)\u0026mdash;consistent with an industry survey finding that 50% of \"long GLD\" cases are caused by non-DM factors [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e].\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eFor example, two oncology trials with identical CRF counts (50 forms, indicating matched DM workload) had GLD values of 70 and 120 days\u0026mdash;a 71% difference. Post-hoc review revealed the longer GLD was due to a 3-week delay in sponsor contract finalization, not DM inefficiency. GLD\u0026rsquo;s inability to filter out such external noise means it often penalizes or rewards DM teams for factors beyond their control.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003eThis disconnect arises because GLD is dominated by non-DM delays common in clinical research: protocol amendments (e.g., adding endpoints), sponsor review bottlenecks, or ethics committee feedback\u0026mdash;factors that have no bearing on DM team performance [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. This is further supported by an industry survey of \u0026ldquo;long GLD\u0026rdquo; trials, which found that 50% of delays stemmed from non-DM issues [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e].\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec22\" class=\"Section3\"\u003e\u003ch2\u003e4.1.3 GLD Misleads Resource Allocation\u003c/h2\u003e\u003cp\u003eEfficient resource planning requires metrics that link \"timeline\" to \"workload,\" but GLD\u0026rsquo;s weak correlation with DM workload proxies (CRF count: r\u0026thinsp;=\u0026thinsp;0.37; EC count: r\u0026thinsp;=\u0026thinsp;0.38; Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e) makes this impossible. For instance:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eTwo trials with identical GLD (70 days) had DPWH values of 150 hours (30 unique CRFs) and 250 hours (60 unique CRFs)\u0026mdash;a 67% difference in DM effort. If an organization relied on GLD alone, it would assign the same number of FTEs to both trials, leading to under-resourcing of the high-workload trial (60 CRFs) and potential delays in data cleaning or query resolution.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eThis risk is amplified for complex trials: Phase III oncology trials require 2.4x more DPWH than Phase I trials, but their GLD is only 1.6x longer. Using GLD to set FTEs would systematically understaff high-complexity trials, undermining overall trial progress.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec23\" class=\"Section2\"\u003e\u003ch2\u003e4.2 DPWH\u0026rsquo;s Advantages: Aligning Metric with DM Reality\u003c/h2\u003e\u003cp\u003eDPWH addresses GLD\u0026rsquo;s flaws by directly measuring \"DM effort invested in EDC setup\"\u0026mdash;a factor entirely within the DM team\u0026rsquo;s control. Our data demonstrate four key advantages that make DPWH a superior efficiency metric.\u003c/p\u003e\u003cdiv id=\"Sec24\" class=\"Section3\"\u003e\u003ch2\u003e4.2.1 DPWH Captures Trial Complexity Systematically\u003c/h2\u003e\u003cp\u003eUnlike GLD, DPWH varies in direct proportion to trial complexity\u0026mdash;ensuring it reflects the actual demands placed on DM teams. This is evident across all categorical project characteristics:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eBy TA\u003c/b\u003e: Oncology trials (the most data-intensive) had the highest DPWH (289.3\u0026thinsp;\u0026plusmn;\u0026thinsp;190.0 hours), followed by Non-Oncology (214.7\u0026thinsp;\u0026plusmn;\u0026thinsp;120.0 hours) and Clinical Pharmacology (151.9\u0026thinsp;\u0026plusmn;\u0026thinsp;81.6 hours). All pairwise differences were statistically significant (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), mirroring the higher CRF/EC demands of oncology trials (mean CRF count: 50.9\u0026thinsp;\u0026plusmn;\u0026thinsp;7.7 vs. 30.7\u0026thinsp;\u0026plusmn;\u0026thinsp;6.1 for Clinical Pharmacology; mean EC count: 837.0\u0026thinsp;\u0026plusmn;\u0026thinsp;251.7 vs. 420.5\u0026thinsp;\u0026plusmn;\u0026thinsp;189.3).\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eBy trial phase\u003c/b\u003e: DPWH increased monotonically with phase: Phase III (335.5\u0026thinsp;\u0026plusmn;\u0026thinsp;201.4 hours)\u0026thinsp;\u0026gt;\u0026thinsp;Phase II (224.5\u0026thinsp;\u0026plusmn;\u0026thinsp;131.4 hours)\u0026thinsp;\u0026gt;\u0026thinsp;Phase I (182.0\u0026thinsp;\u0026plusmn;\u0026thinsp;115.4 hours). This aligns with regulatory expectations for late-phase trials (e.g., more endpoints, stricter data validation) and ensures DM effort is proportional to trial risk.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eBy EDC system\u003c/b\u003e**: G-EDC trials required 2.4x more DPWH than L-EDC trials (423.3\u0026thinsp;\u0026plusmn;\u0026thinsp;209.2 vs. 178.6\u0026thinsp;\u0026plusmn;\u0026thinsp;89.1 hours, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), reflecting the additional setup tasks required for the systems.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003eThis systematic variation means DPWH can be used to establish **evidence-based benchmarks** (e.g., \"Phase III oncology trials using G-EDC require 250\u0026ndash;270 DPWH hours\"), a capability GLD lacks.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec25\" class=\"Section3\"\u003e\u003ch2\u003e4.2.2 DPWH Strongly Correlates with Core DM Workload\u003c/h2\u003e\u003cp\u003eA valid DM efficiency metric must link to tangible DM tasks\u0026mdash;specifically, CRF design and EC programming, which constitute\u0026thinsp;~\u0026thinsp;70% of pre-go-live DM effort [unpublished]. Our correlation analysis confirms DPWH\u0026rsquo;s superiority in this regard:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eDPWH had a strong positive correlation with unique CRF count (r\u0026thinsp;=\u0026thinsp;0.48, p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001) and EC count (r\u0026thinsp;=\u0026thinsp;0.52, p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001; Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). For every additional CRF, DPWH increased by ~\u0026thinsp;3 hours\u0026mdash;directly reflecting the time required to design, test, and validate new data collection forms.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eIn contrast, GLD\u0026rsquo;s correlations with these proxies were weak (CRF: r\u0026thinsp;=\u0026thinsp;0.37; EC: r\u0026thinsp;=\u0026thinsp;0.38). This means GLD cannot distinguish between a trial with 30 CRFs (low workload) and 60 CRFs (high workload) if their non-DM delays are similar\u0026mdash;rendering it useless for assessing how DM teams handle core tasks.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec26\" class=\"Section3\"\u003e\u003ch2\u003e4.2.3 DPWH Tracks Long-Term DM Efficiency Gains\u003c/h2\u003e\u003cp\u003eAn effective metric should capture improvements in DM processes (e.g., EDC system upgrades, standardized CRF templates). Our data show DPWH is uniquely capable of this:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eDPWH had a strong negative correlation with initiation year (r=-0.43, p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), with mean DPWH dropping from 285.3\u0026thinsp;\u0026plusmn;\u0026thinsp;192.3 hours (2019) to 148.4\u0026thinsp;\u0026plusmn;\u0026thinsp;57.5 hours (2024)\u0026mdash;a 48% reduction over 5 years. This trend directly reflects efficiency gains from adopting L-EDC V4.0+ (an upgraded system with automation features), which reduced DPWH by ~\u0026thinsp;15% for Phase I trials alone【18】.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eGLD, by contrast, had a weak negative correlation with initiation year (r=-0.23, p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), with mean GLD decreasing only 34% (127.1\u0026thinsp;\u0026plusmn;\u0026thinsp;76.1 days in 2019 to 84.9\u0026thinsp;\u0026plusmn;\u0026thinsp;38.5 days in 2024). The smaller reduction occurred because DM efficiency gains were offset by rising trial complexity (e.g., new TAs or more advanced trial design methodology)\u0026mdash;a tradeoff GLD cannot disentangle, but DPWH can.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec27\" class=\"Section3\"\u003e\u003ch2\u003e4.2.4 DPWH Is More Reliable for Subgroup-Specific Efficiency Assessment\u003c/h2\u003e\u003cp\u003eThe intraclass correlation coefficient (ICC) measures a metric\u0026rsquo;s ability to reflect homogeneous efficiency within subgroups (e.g., same TA or phase) and distinct differences between subgroups. Our mixed-effects model results confirm DPWH\u0026rsquo;s higher reliability (Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e6\u003c/span\u003e):\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eFor TA subgroups\u003c/b\u003e: DPWH had an ICC of 0.39, compared to 0.33 for GLD. This means DPWH is more consistent in measuring efficiency for trials within the same TA (e.g., all oncology trials) and better at distinguishing efficiency between TAs (e.g., oncology vs. clinical pharmacology).\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eFor trial phase subgroups\u003c/b\u003e: DPWH\u0026rsquo;s ICC (0.50) was 35% higher than GLD\u0026rsquo;s (0.37). A higher ICC indicates DPWH can more accurately identify which Phase III trials have efficient DM teams (lower DPWH relative to peers) versus which are inefficient\u0026mdash;critical for targeted process improvement.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec28\" class=\"Section2\"\u003e\u003ch2\u003e4.3 Direct Comparison of GLD and DPWH: A Summary of Key Differences\u003c/h2\u003e\u003cp\u003eTo further contextualize DPWH\u0026rsquo;s superiority, Table\u0026nbsp;\u003cspan refid=\"Tab7\" class=\"InternalRef\"\u003e7\u003c/span\u003e synthesizes the core performance metrics of GLD and DPWH using data from this study. Every criterion\u0026mdash;from alignment with workload to reliability\u0026mdash;confirms DPWH is a more valid measure of DM efficiency.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab7\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 7\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eComparison of GLD and DPWH on DM efficiency metrics\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"3\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEvaluation Criterion\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eGLD\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eDPWH\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMeasurement Content\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTime from protocol finalization to EDC launch (speed-focused)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eTotal DM work hours for EDC setup (workload-aligned)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSusceptibility to Non-DM Delays\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eHigh (e.g., protocol amendments, ethics reviews)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eLow (only captures DM-specific tasks)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAssociation with Trial Complexity\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eWeak (no meaningful differences by TA/phase/EDC system)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eStrong (varies systematically by TA/phase/EDC system)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCorrelation with Workload Proxies\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eWeak (CRF: r\u0026thinsp;=\u0026thinsp;0.37; EC: r\u0026thinsp;=\u0026thinsp;0.38)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eStrong (CRF: r\u0026thinsp;=\u0026thinsp;0.48; EC: r\u0026thinsp;=\u0026thinsp;0.52)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAbility to Reflect Efficiency Trends\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePoor (weak correlation with initiation year: r=-0.23)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eExcellent (strong correlation with initiation year: r=-0.43)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eValue for Resource Allocation\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLow (cannot distinguish workload differences)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eHigh (enables evidence-based benchmarking)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eReliability (ICC for TA/Phase)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLower (TA: 0.33; Phase: 0.37)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eHigher (TA: 0.39; Phase: 0.50)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLarge-Scale Validation\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLimited (mostly survey-based)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eValidated (641 diverse trials)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec29\" class=\"Section2\"\u003e\u003ch2\u003e4.4 Study Limitations and Future Directions\u003c/h2\u003e\u003cp\u003eWhile this study provides strong support for DPWH, three limitations should be noted to guide future research:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eSingle-enterprise data\u003c/b\u003e: The 641 trials were from a single global pharmaceutical enterprise. Multi-center studies spanning contract research organizations (CROs) and academic institutions are needed to confirm DPWH\u0026rsquo;s generalizability across diverse operational models.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eLack of post-go-live quality metrics\u003c/b\u003e: This study focused on pre-go-live efficiency, but future work should link DPWH to post-launch outcomes (e.g., data query rate, inspection findings). Preliminary industry data suggest trials with higher DPWH (reflecting thorough EC design) have fewer manual queries\u0026mdash;a relationship that merits formal validation [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e].\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eNo integration of labor costs\u003c/b\u003e: DPWH measures hours but not cost (e.g., senior vs. junior DM staff). Adding cost data to create a \"cost-per-DPWH\" metric would enhance budget optimization, particularly for trials with constrained resources.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec30\" class=\"Section2\"\u003e\u003ch2\u003e4.5 Practical Implications for the Clinical Research Industry\u003c/h2\u003e\u003cp\u003eThe shift from GLD to DPWH has tangible benefits for all stakeholders:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003ePharmaceutical enterprises\u003c/b\u003e: DPWH enables more accurate resource allocation (e.g., assigning 2 FTEs to a Phase III oncology trial vs. 1 FTE to a Phase I trial) and better tracking of DM team performance.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eCROs\u003c/b\u003e: DPWH provides a transparent, workload-based metric for client reporting\u0026mdash;reducing disputes over \"delayed go-live\" caused by non-DM factors.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eRegulatory Authority\u003c/b\u003e: DPWH\u0026rsquo;s alignment with data quality (via EC count correlation) supports regulatory expectations for reproducible, high-quality clinical data [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e].\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003c/div\u003e"},{"header":"5. Conclusion","content":"\u003cp\u003eEDC go-live days is an unreliable metric for measuring DM efficiency across the global clinical research industry. It is disconnected from trial complexity, biased by non-DM delays, and unable to guide effective resource allocation. In contrast, EDC Design Phase Work Hours directly captures DM workload, correlates strongly with project characteristics (TA, phase, EDC system) and workload proxies (CRF count, EC count), and is explained by meaningful efficiency factors (adjusted R\u0026sup2;=0.61 vs. 0.26 for GLD). DPWH also has higher reliability (ICC) and better tracks efficiency gains over time.\u003c/p\u003e\u003cp\u003eAdopting DPWH as a universal DM efficiency metric can drive three key improvements:\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eAccurate resource allocation\u003c/strong\u003e\u003cp\u003eEvidence-based DPWH benchmarks for different trial types ensure teams are appropriately staffed, avoiding under-resourcing of complex trials.\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eMeasurable efficiency gains\u003c/strong\u003e\u003cp\u003eDPWH quantifies the impact of EDC system upgrades, standardized workflows, and other process improvements.\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eQuality-aligned assessment\u003c/strong\u003e\u003cp\u003eDPWH\u0026rsquo;s link to data quality supports regulatory compliance and reproducibility, aligning with global regulatory expectations.\u003c/p\u003e\u003c/p\u003e\u003cp\u003eBy shifting from \"speed-focused GLD\" to \"workload-aligned DPWH,\" the clinical research industry can enhance efficiency, improve data quality, and optimize resource stewardship\u0026mdash;ultimately accelerating the development of new therapies for patients worldwide.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eFunding:\u003c/strong\u003e\u0026nbsp; This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.\u003c/p\u003e\n\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003esg collected and analysis the datahy and cy discussed the paper draft and data presentcy wrote the main manuscript text\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eThe authors thank Dr. Yihua Zhao for assistance with data statistical analysis.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eDataset is available upone request\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eRuth C, Huey S, Krisher J, et al. An Electronic Data Capture Framework (ConnEDCt) for Global and Public Health Research: Design and Implementation. J Med Internet Res 2020;22(8):e18580. https://doi.org/10.2196/18580\u003c/li\u003e\n \u003cli\u003ePestronk M, Johnson D, Muthanna M, et al. Electronic Data Capture\u0026mdash;Selecting an EDC System. J Soc Clin Data Manag 2021;1(1). https://doi.org/10.47912/jscdm.29\u003c/li\u003e\n \u003cli\u003eVielhauer J, Mahajan UM, Adorjan K, et al. Electronic data capture in resource-limited settings using the lightweight clinical data acquisition and recording system. Sci Rep 2024;14:19056. https://doi.org/10.1038/s41598-024-69550-w\u003c/li\u003e\n \u003cli\u003eYan C. Data Management in Clinical Research. Beijing, Science Publisher; 2011.\u003c/li\u003e\n \u003cli\u003eSCDM. Metrics in Clinical Data Management. J Soc Clin Data Manag 2023;1(1):1\u0026ndash;9. https://doi.org/10.47912/jscdm.331\u003c/li\u003e\n \u003cli\u003eZozus MN, Sanns W, Eisenstein E. Beyond EDC. J Soc Clin Data Manag 2021;1(1). https://doi.org/10.47912/jscdm.33\u003c/li\u003e\n \u003cli\u003eDatar M. Important Metrics in Clinical Data Management. Cloudbyz Resources; 2023. https://www.cloudbyz.com/resources/edc/10-important-metrics-in-clinical-data-management/\u003c/li\u003e\n \u003cli\u003ePomerantseva V. Data Management Efficiencies Through Risk-Based Approaches and Innovations. Appl Clin Trials 2024. https://www.appliedclinicaltrialsonline.com/view/data-management-efficiencies-through-risk-based-approaches-and-innovations\u003c/li\u003e\n \u003cli\u003eHelms RW, Fitzmartin R, Fillow PJ, et al. Metrics and Best Practices in Clinical Data Management: Conclusions of a Dia Roundtable Workshop. Ther Innov Regul Sci 2001;35:681\u0026ndash;694. https://doi.org/10.1177/009286150103500306\u003c/li\u003e\n \u003cli\u003eWilkinson M, Young R, Harper B, et al. Baseline assessment of the evolving 2019 eClinical Landscape. Ther Innov Regul Sci 2018;53(1):869\u0026ndash;876.\u003c/li\u003e\n \u003cli\u003eHarper B, Smith Z, Snowdon J, et al. Characterizing Pain Points in Clinical Data Management and Assessing the Impact of Mid-Study Updates. Ther Innov Regul Sci 2021;55:1006\u0026ndash;1012. https://doi.org/10.1007/s43441-021-00301-z\u003c/li\u003e\n \u003cli\u003eHarper B, Smith Z, Snowdon J, et al. Characterizing Clinical Data Management Challenges and Their Impact. Appl Clin Trials 2021;30(1). https://www.appliedclinicaltrialsonline.com/view/characterizing-clinical-data-management-challenges-and-their-impact\u003c/li\u003e\n \u003cli\u003eEade D, Pestronk M, Russo R, et al. Electronic Data Capture\u0026mdash;Study Implementation and Start-up. J Soc Clin Data Manag 2021;4. https://doi.org/10.47912/jscdm.30\u003c/li\u003e\n \u003cli\u003eFDA. Guidance Document: Electronic Systems, Electronic Records, and Electronic Signatures in Clinical Investigations: Questions and Answers. 2024. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/electronic-systems-electronic-records-and-electronic-signatures-clinical-investigations-questions\u003c/li\u003e\n \u003cli\u003eYan C et al：DM-led EDC Building: A Quality-by-Design Framework Integrating SAI to Transform Clinical Data Management. 2025 submitted\u003c/li\u003e\n \u003cli\u003eYan C, Yan H, Gou S, Zhao Y，Zhou E, Wei X，Zhao S and Zhuang Y：Clinical Data Total Management System for Realizing Instant Inspection Readiness in Clinical Data Management 2025 China Food \u0026amp; Drug Administration Magazine (submitted)\u003c/li\u003e\n \u003cli\u003eYan C, Chen X, Bie L, Peng R, Gou S and Yan H：EDC Design Phase: Where Clinical Data Manager\u0026rsquo;s Time Goes. A Retrospective Analysis of Worktime Distribution and Efficiency Gains (2019\u0026ndash;2021 vs. 2022\u0026ndash;2024). 2025 (submitted)\u003c/li\u003e\n \u003cli\u003eYan C, Yan H, Zhou E, Wei X, Gou S, Zhao S: EDC Building Efficiency: Stratified SAI Contributions\u0026mdash;Empirical Analysis of 491 Projects (2020\u0026ndash;2025) via HRTAU EDC 2025 (submitted)\u003c/li\u003e\n \u003cli\u003eYan C, Wei X, Xin X, Yan H, Gou S, Zhao S, Zhuang Y: From Protocol to EDC: Unpacking SAI (Standardization/ Automation/ Intelligence) Technical Mechanisms for Efficient EDC Builds\u0026mdash;2022\u0026ndash;2024 Worktime Validation. 2025 (In preparation)\u003c/li\u003e\n \u003cli\u003eGetz K, Smith Z, Kravet M. Protocol Design and Performance Benchmarks by Phase and by Oncology and Rare Disease Subgroups. Ther Innov Regul Sci 2023;57(1):49\u0026ndash;56. https://doi.org/10.1007/s43441-022-00438-5\u003c/li\u003e\n \u003cli\u003eRubio DM. Common metrics to assess the efficiency of clinical research. Eval Health Prof 2013;36(4):432\u0026ndash;446.\u003c/li\u003e\n \u003cli\u003eWalden A, Garza M, Rasmussen L. Best Practices for Research Data Management. In: Richesson RL, Andrews JE, Fultz Hollis K, eds. Clinical Research Informatics. Health Informatics. Springer Cham; 2023. https://doi.org/10.1007/978-3-031-27173-1_14\u003c/li\u003e\n \u003cli\u003eBajpai N, Chatterjee A, Dang S, et al. Metrics for leveraging more in clinical data management: proof of concept in the context of vaccine trials in an Indian pharmaceutical company. Asian J Pharm Clin Res 2015;8(3):350\u0026ndash;357. https://core.ac.uk/download/pdf/477851493.pdf\u003c/li\u003e\n \u003cli\u003eSingh K, Abdul Salam M, Devarajan R, et al. Solving clinical trial delays: innovative solutions. Clin Investig (Lond) 2015;5(9):745\u0026ndash;753. https://www.openaccessjournals.com/articles/solving-clinical-trial-delays-innovative-solutions.pdf\u003c/li\u003e\n \u003cli\u003eIndustry Survey Reveals Clinical Data Management Delays Slowing Trial Completion. Pharm Exec 2017. https://www.pharmexec.com/view/industry-survey-reveals-clinical-data-management-delays-slowing-trial-completion\u003c/li\u003e\n \u003cli\u003eWalden A, Garza M, Rasmussen L. Best Practices for Research Data Management. In: Richesson RL, Andrews JE, Fultz Hollis K, eds. Clinical Research Informatics. Health Informatics. Springer Cham; 2023. https://doi.org/10.1007/978-3-031-27173-1_14\u003c/li\u003e\n \u003cli\u003eRubio DM. Common metrics to assess the efficiency of clinical research. Eval Health Prof 2013;36(4):432\u0026ndash;446.\u003c/li\u003e\n \u003cli\u003eBajpai N, Chatterjee A, Dang S, et al. Metrics for leveraging more in clinical data management: proof of concept in the context of vaccine trials in an Indian pharmaceutical company. Asian J Pharm Clin Res 2015;8(3):350\u0026ndash;357. https://core.ac.uk/download/pdf/477851493.pdf\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Electronic Data Capture (EDC), EDC Design Phase Work Hours (DPWH), EDC Go-Live Days (GLD), Data Management Efficiency, Clinical Trials","lastPublishedDoi":"10.21203/rs.3.rs-7831647/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7831647/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cb\u003eBackground\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe global clinical research industry widely uses EDC go-live days (GLD, time from protocol finalization to EDC launch) to measure data management (DM) efficiency, but GLD is biased by non-DM delays and fails to reflect trial complexity. EDC Design Phase Work Hours (DPWH, total DM effort for EDC setup) is a potential alternative but lacks large-scale validation.\u003c/p\u003e\u003cp\u003e\u003cb\u003eObjective\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTo compare GLD and DPWH in reflecting DM efficiency using 641 diverse clinical trials, by assessing their associations with project characteristics (therapeutic area [TA], phase, EDC system) and DM workload proxies (CRF count, edit check [EC] count), and to test their correlation.\u003c/p\u003e\u003cp\u003e\u003cb\u003eMethods\u003c/b\u003e\u003c/p\u003e\u003cp\u003eRetrospective analysis of 641 trials (2019\u0026ndash;2025). Descriptive statistics, Spearman\u0026rsquo;s correlation (including GLD-DPWH correlation), Kruskal-Wallis tests, robust regression, and mixed-effects models (intraclass correlation [ICC]) were used.\u003c/p\u003e\u003cp\u003e\u003cb\u003eResults\u003c/b\u003e\u003c/p\u003e\u003cp\u003eMean GLD was 101.2\u0026thinsp;\u0026plusmn;\u0026thinsp;60.1 days, mean DPWH 219.1\u0026thinsp;\u0026plusmn;\u0026thinsp;148.6 hours. GLD-DPWH correlation was moderate (r\u0026thinsp;=\u0026thinsp;0.47, p\u0026thinsp;\u0026lt;\u0026thinsp;0.005). GLD had weak associations with project characteristics (e.g., TA: χ\u0026sup2;=60.61, p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001) and workload (CRF: r\u0026thinsp;=\u0026thinsp;0.37; EC: r\u0026thinsp;=\u0026thinsp;0.38). DPWH strongly correlated with characteristics (e.g., Phase III vs. I: 335.5\u0026thinsp;\u0026plusmn;\u0026thinsp;201.4 vs. 182.0\u0026thinsp;\u0026plusmn;\u0026thinsp;115.4 hours, p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001) and workload (CRF: r\u0026thinsp;=\u0026thinsp;0.48; EC: r\u0026thinsp;=\u0026thinsp;0.52). Robust regression explained 61% of DPWH variation vs. 26% for GLD. DPWH had higher ICC (TA: 0.39; Phase: 0.5) than GLD (TA: 0.33; Phase: 0.37).\u003c/p\u003e\u003cp\u003e\u003cb\u003eConclusion\u003c/b\u003e\u003c/p\u003e\u003cp\u003eDPWH outperforms GLD as a DM efficiency metric, aligning with trial complexity and workload. Its adoption can improve resource allocation and process optimization.\u003c/p\u003e","manuscriptTitle":"Rethinking EDC Go-Live Metrics: Why EDC Design Phase Work Hours Outperform Days in Measuring DM Efficiency (641 Clinical Trials)","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-10-14 06:20:35","doi":"10.21203/rs.3.rs-7831647/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"e154ec0f-c259-48c2-87c6-8694cdf17c2a","owner":[],"postedDate":"October 14th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-01-25T12:54:11+00:00","versionOfRecord":[],"versionCreatedAt":"2025-10-14 06:20:35","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7831647","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7831647","identity":"rs-7831647","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}