Organizational Research Support and Initial IRB Submission Quality in Postgraduate Surgical Training: A Five-Year Cohort Study

Abstract

Background

Scholarly activity is a core component of postgraduate surgical training. Yet early research processes particularly Institutional Review Board (IRB) submissions can impose substantial procedural and administrative burdens. These barriers may hinder timely engagement in research, especially among trainees.

Objective

To evaluate whether the organizational research support delivered by trained research assistants is associated with improved initial IRB submission quality and reduced revision burden in a surgical training environment.

Materials and Methods

A single-center retrospective cohort study was conducted using 356 IRB submissions from the Department of Surgery at a tertiary academic center in Bangkok, Thailand (2019–2023). Submissions were classified as research assistant (RA)-supported or non-supported. The primary outcomes were (1) receiving revision comments and (2) the number of unique revision comments during the initial IRB review cycle. Comments were categorized into administrative, ethical, methodological, and scientific domains. Multivariable logistic and Poisson regression were performed, adjusting for investigator level, IRB review type, and year of submission.

Results

RA support was used in 324 submissions (91.0%) and 32 (9.0%) proceeded without RA support. RA support was associated with a lower likelihood of receiving revision comments (76.9% vs 93.8%; adjusted OR 0.18, 95% CI 0.04–0.82) and fewer comments per submission (median 2 vs 3; adjusted IRR 0.51, 95% CI 0.41–0.64). Reductions were primarily observed in administrative domains (39.8% vs 76.7%, p<0.001), with no significant differences in methodological, ethical, or scientific comments. The time to final approval did not differ between the groups.

Conclusions

Organizational research support was associated with improved procedural readiness of IRB submissions, with fewer administrative deficiencies and no change in scientific review outcomes. Effects on scholarly engagement were not assessed and require prospective evaluation.

Introduction

Participation in scholarly activity is a fundamental expectation of postgraduate surgical training.^1–3 Accreditation frameworks, such as the Accreditation Council for Graduate Medical Education (ACGME) and the Royal College of Physicians and Surgeons of Canada (CanMEDS Physician Competency Framework), emphasize the development of research competencies, such as critical appraisal, study design, and translation of evidence into clinical practice.^2,3 Research engagement has also been associated with improved academic productivity and the development of lifelong learning skills among trainees.^4–6

However, in high-volume clinical training environments, competing service demands and limited protected time often constrain trainee participation in research activities.^7,8 These constraints may be particularly pronounced in surgical training, where long operative hours, on-call duties, and the priority placed on technical skill acquisition leave little contiguous time for research tasks, and where scholarly output is often expected within a fixed residency timeline. Consequently, research may be perceived as an additional obligation rather than an integrated component of professional development, particularly during the early stages of the research process. Commonly proposed remedies include protected research time, structured research curricula, faculty mentorship, and research methodology workshops.^4–7 These approaches primarily target the intellectual and motivational dimensions of research; the procedural and administrative dimensions have received comparatively less attention.

One such stage obtaining ethical approval can represent a substantial procedural barrier for novice investigators.^9–11 Institutional Review Board (IRB) submissions require completion of multiple regulatory documents, adherence to formatting standards, and familiarity with specialized terminology and electronic submission platforms. Prior studies in clinical research systems have highlighted the complexity and administrative burden associated with regulatory approval processes, which may delay study initiation and reduce research efficiency.^10–12 These challenges are relevant in training environments, where trainees are concurrently developing both clinical and academic competencies.

From a learning perspective, early research processes involve both intrinsic cognitive demands (eg study design and methodological reasoning) and extraneous workload related to administrative and procedural requirements. According to cognitive load theory, extraneous load does not directly contribute to learning, and an excess of it may impair the acquisition of core competencies.^13–15 In this context, reducing avoidable administrative burden may help preserve cognitive resources for higher-order scholarly tasks.

Structured organizational support may offer a practical approach to addressing these challenges. Conceptually, this form of support can be viewed as scaffolding, whereby targeted assistance enables learners to complete complex tasks beyond their independent capability.^16–18 In medical education, scaffolding strategies have been shown to enhance learning efficiency and facilitate skill acquisition in complex environments. However, within research training, the application of scaffolding at the level of administrative and procedural workflows remains underexplored.

In many academic surgical departments, research assistants (RAs) are integrated into research workflows to support IRB submission preparation. Their roles typically include pre-submission checks for completeness, verification of required documentation, and guidance in navigating institutional electronic systems. This form of support is distinct from research mentorship and methodological supervision, which address study design, analysis, and scientific interpretation. RA support, as defined in this study, is confined to procedural and administrative preparation and does not involve the scientific content of the protocol. In its absence, investigators prepare and submit protocols independently and receive feedback only through the formal IRB review process. RA support is not without potential drawbacks: uptake is usually voluntary, so investigators who use it may differ from those who do not; access may be unequal across investigators; and reducing procedural friction may shift rather than remove the burden. While such system-level support is being increasingly implemented, empirical evidence evaluating its impact on early-stage research processes in postgraduate training remains limited.^16–18

This study was conducted in a tertiary academic surgical department in Thailand, where residency and fellowship training follow a national curriculum under the Royal College of Surgeons of Thailand, and where protocols are reviewed by a single institutional IRB with staff-level administrative screening preceding committee review. Although the accreditation frameworks cited above are North American, the expectation of scholarly activity during training and the administrative burden of ethical approval are shared across settings. We analysed a five-year cohort of IRB submissions from this department to examine whether RA support is associated with improved initial submission readiness and reduced revision burden. We specifically hypothesized that RA-supported submissions would demonstrate fewer administrative and form-related revisions, while methodology- and content-related feedback would remain comparable between groups.

Methods

Study Design and Setting

We conducted a retrospective cohort study of all human research protocols submitted to the IRB of the Faculty of Medicine Ramathibodi Hospital, Mahidol University, Bangkok, Thailand, between January 1, 2019, and December 31, 2023. The institution is a tertiary academic medical center with accredited surgical residency and fellowship programmes and an established departmental research support unit. This study focused on early research workflow processes within the postgraduate surgical training environment.

This study was approved by the IRB of the Faculty of Medicine Ramathibodi Hospital (COA No. MURA2024/195). Informed consent was waived due to the use of de-identified administrative data. Reporting followed the STROBE guidelines.

Study Population

The eligibility criteria included all IRB submissions for which the principal investigator (PI) was affiliated with the Department of Surgery. To ensure a consistent postgraduate training context, protocols led by non-physician investigators were excluded. PIs were categorized as faculty members, fellows, or residents.

Exposure: Organizational Research Support (RA Involvement)

The Department of Surgery maintains a Research Support Unit comprising trained non-physician RAs with expertise in IRB regulations, ethical documentation, and electronic submission procedures. RA support was available to all investigators on a voluntary basis. RA involvement was limited to the procedural and administrative aspects of IRB preparation and did not include the modification of the study design, statistical analysis plans, or scientific content. The RA workflow comprised the following:

1. Pre-submission screening for document completeness and internal consistency

2. Verification of required forms and formatting compliance

3. Technical guidance for navigating the electronic IRB platform

The unit employed eight RAs throughout the study period, with no staff turnover. All RAs had completed institutional human research ethics training and the Collaborative Institutional Training Initiative (CITI) Program before assisting with submissions. Pre-submission review followed the institutional IRB checklist for each review category; the checklist was unchanged during 2019–2023.

Submissions were classified as RA-supported if a pre-submission review by the RA unit was documented.

Outcomes

Primary Outcome: Procedural Completeness During the Initial Review Cycle

Procedural completeness was assessed using two indicators:

1. Revision comments: A binary variable indicating whether any revisions (major or minor) were requested following the first review.

2. Revision burden: The total number of unique comments issued during the initial review cycle. A unique comment was defined as one distinct actionable issue. Multiple sentences addressing the same issue were counted once; the same issue raised by more than one reviewer was counted once; separate issues raised within one paragraph were counted separately. Submissions that received no comments were assigned a count of zero.

These were used as reproducible proxies for document completeness and early workflow quality.

Comment Domain Classification

Revision comments were categorized into four domains:

• Administrative/forms (including formatting): missing documents, incorrect form completion, and formatting issues (eg “add the version number and date to the case report form”; “attach the co-investigator’s CITI certificate”)

• Ethics: informed consent, privacy, or ethical safeguards (eg “the information sheet should state that withdrawal will not affect clinical care”)

• Methodology: study design, statistical plan, or sample size (eg “clarify whether recruitment stops at 100 participants or continues to the end of the study period”)

• Content-specific scientific recommendations: clinical or surgical scientific feedback (eg “specify whether CT or colonoscopy confirmation is required for inclusion”)

A comment addressing more than one domain was assigned to each applicable domain. All comments were classified by one investigator. A second investigator, blinded to RA status, independently classified all comments using the same criteria. Inter-rater agreement was almost perfect (Cohen’s κ = 0.95, 95% CI 0.92–0.98). Disagreements were resolved by consensus, and the consensus classification was used in all analyses.

Secondary Outcome: Process Efficiency

Process efficiency was defined as time to approval, calculated as the number of days from the date of initial submission to issuance of the Certificate of Approval (COA). This outcome was analysed only for submissions that obtained a COA (n=333); submissions that were withdrawn or not approved (n=23) were excluded from this analysis.

Covariates

The covariates that were recorded are listed below.

• Investigator level (faculty, fellow, or resident)

• IRB review type (exempt, expedited, or full board)

• Year of submission (continuous)

Statistical Analysis

No a priori sample-size calculation was performed. The study included all eligible IRB submissions during the predefined five-year period, so the sample size was fixed by the available data. With 32 non-supported and 324 supported submissions, the study had 80% power (two-sided α=0.05) to detect an absolute reduction in the proportion receiving revision comments of approximately 20% points from a baseline of 94%. Smaller differences, including those observed for the domain-specific and secondary outcomes, may not have been detectable.

Categorical variables were compared using chi-square or Fisher’s exact tests. Continuous variables were assessed for normality and compared using Mann–Whitney U-tests. A multivariable logistic regression model was fitted with RA support, investigator level, IRB review type, and year of submission as covariates. All covariates were retained in the final model regardless of statistical significance. Study design was not entered as a separate covariate because it is encoded within the IRB review type at our institution (full board review for interventional and prospective studies; expedited review for observational studies; exempt review for case reports and minimal-risk studies). Adjusted odds ratios (aORs) with 95% confidence intervals (CIs) were reported. Model fit was assessed using the Hosmer–Lemeshow test.

For the number of unique revision comments, a Poisson regression model was fitted with the same covariates, including submissions that received no comments. Overdispersion was assessed by the Pearson dispersion statistic and by comparison with a negative binomial model; neither indicated overdispersion (dispersion = 1.01). Adjusted incidence rate ratios (IRRs) with 95% CIs were reported.

Given the numerical imbalance between RA-supported and non-supported submissions, a bootstrap sensitivity analysis (1,000 iterations) was conducted to evaluate the stability of the logistic regression estimates. Because some investigators submitted more than one protocol, the logistic regression model was additionally re-estimated with standard errors clustered by investigator, and refitted using only each investigator’s first submission. All analyses were performed using Stata version 14.2 (StataCorp, College Station, TX, USA). A two-sided p-value <0.05 was considered statistically significant.

Results

Baseline Characteristics

Of 358 IRB submissions from the Department of Surgery during the study period, 356 met the inclusion criteria and were led by 196 investigators; 53 investigators submitted more than one protocol (range 2–27) (Figure 1). Of all submissions, 54.8% were led by trainees and 45.2% by faculty. Expedited review was the most common IRB category (59.6%), followed by full board review (27.3%) and exempt/case report submissions (13.2%) (Table 1).

Figure 1.

Flow diagram of institutional review board submissions, eligibility assessment, classification according to research assistant support, and study outcomes.

Abbreviations: IRB, institutional review board; RA, research assistant; IQR, interquartile range.

Table 1.

General Characteristics of IRB Submissions from the Department of Surgery (N = 356)

Characteristics	Value
IRB review type, n (%)
Full board review	97 (27.3)
Expedited review	212 (59.6)
Exempt/Case report	47 (13.2)
Investigator level, n (%)
Faculty	161 (45.2)
Trainee (total)	195 (54.8)
– Fellow	36 (10.1)
– Resident	159 (44.7)
Submission outcomes
IRB approval obtained, n (%)	333 (93.5)
Time to approval, median (IQR), days a	58 (27–97)
Received revision comments, n (%)	279 (78.4)
Number of comments, median (IQR)b	2 (1–3)
Reviewer source of initial comments (n=279), n (%)
IRB staff (administrative screening)	204 (73.1)
IRB committee (formal review)	75 (26.9)

Notes: Protocols led by non-physician investigators were excluded. ^aAmong the 333 submissions that obtained approval. ^bAmong the 279 submissions that received at least one comment.

RA support was utilized in 324 submissions (91.0%), while 32 submissions (9.0%) proceeded without RA involvement. Trainees constituted a higher proportion of the non-RA group (84.4%) than the RA-supported group (51.9%) (Table 2).

Table 2.

Comparison of IRB Submission Characteristics and Procedural Completeness According to Research Assistant (RA) Support (n=356)

Characteristics	Without RA (n=32)	With RA (n=324)	p-value
Investigator level, n (%)			0.001
Faculty	5 (15.6)	156 (48.2)
Trainee	27 (84.4)	168 (51.9)
Procedural completeness
Received revision comments, n (%)	30 (93.8)	249 (76.9)	0.027
Number of comments, median (IQR)a	3 (2–4)	2 (1–3)	<0.001
Comment domainsa, n (%)	(n=30)	(n=249)
Administrative/forms	23 (76.7)	99 (39.8)	<0.001
Ethics	2 (6.7)	37 (14.9)	0.222
Methodology	8 (26.7)	79 (31.7)	0.572
Content-specific scientific recommendations	16 (53.3)	135 (54.2)	0.927
Process efficiency
Time to approval, median (IQR), daysb	60 (39–104)	58 (27–97)	0.360

Notes: ^aAmong protocols that received at least one revision comment in each group (without RA n=30; with RA n=249). ^bAmong protocols that obtained approval (without RA n=29; with RA n=304).

Overall, 279 submissions (78.4%) received at least one revision comment during the initial IRB review cycle; among these, the median number of comments was 2 (IQR 1–3), and 204 (73.1%) originated from IRB staff screening and 75 (26.9%) from committee review. IRB approval was obtained for 333 submissions (93.5%), with a median time to approval of 58 days (IQR 27–97) (Table 1).

Procedural Completeness During Initial Review

RA-supported submissions were less likely to receive revision comments than non-supported submissions (76.9% vs 93.8%, p=0.027). The revision burden was also lower in the RA-supported group (median 2 [IQR 1–3] vs 3 [IQR 2–4], p<0.001) (Table 2).

Domain-Specific Revision Comments

Reductions associated with RA support were limited to the administrative and form-related domain (39.8% vs 76.7%, p<0.001). There were no significant differences between RA-supported and non-supported submissions in methodology-related comments (31.7% vs 26.7%, p=0.572), ethics-related comments (14.9% vs 6.7%, p=0.222), or content-specific scientific recommendations (54.2% vs 53.3%, p=0.927) (Table 2).

Multivariable Analysis

After adjusting for investigator level, IRB review type, and year of submission, RA support remained associated with lower odds of receiving revision comments (aOR 0.18; 95% CI 0.04–0.82; p=0.027) (Table 3). Expedited and exempt protocols had lower odds than full board protocols (p<0.001), and the odds of receiving comments increased with year of submission (aOR 1.36 per year; 95% CI 1.11–1.67; p=0.003). Investigator level was not associated with revision likelihood.

Table 3.

Logistic Regression Analysis of Factors Associated with Receiving Revision Comments (n=356)

Variable	Crude OR (95% CI)	p-value	Adjusted OR (95% CI)	p-value
RA support
No	Reference		Reference
Yes	0.22 (0.05–0.95)	0.042	0.18 (0.04–0.82)	0.027
Investigator level
Faculty	Reference		Reference
Fellow	1.98 (0.72–5.45)	0.185	1.06 (0.35–3.22)	0.925
Resident	1.22 (0.72–2.07)	0.458	0.70 (0.38–1.31)	0.266
IRB review type
Full board	Reference		Reference
Expedited	0.03 (0.004–0.22)	0.001	0.02 (0.003–0.18)	<0.001
Exempt/Case report	0.01 (0.001–0.10)	<0.001	0.008 (0.001–0.06)	<0.001
Year of submission (per year)	1.13 (0.95–1.35)	0.176	1.36 (1.11–1.67)	0.003

Notes: Adjusted ORs are from a single multivariable model including all variables shown. Hosmer–Lemeshow p = 0.69.

In a Poisson regression model adjusted for the same covariates, RA support was associated with fewer unique revision comments (adjusted IRR 0.51; 95% CI 0.41–0.64; p<0.001), corresponding to approximately half as many comments per submission. Expedited (IRR 0.51; 95% CI 0.43–0.61) and exempt (IRR 0.26; 95% CI 0.18–0.38) protocols received fewer comments than full board protocols. Comment count increased slightly with year of submission (IRR 1.06 per year; 95% CI 1.00–1.12; p=0.043), and investigator level was not associated with comment count.

Sensitivity Analyses

Bootstrap resampling (1,000 iterations) yielded estimates consistent with the primary logistic model. With standard errors clustered by investigator, the association between RA support and revision comments was unchanged (aOR 0.18; 95% CI 0.05–0.71; p=0.014). Restricting the analysis to each investigator’s first submission (n=196; 23 without RA support) gave a similar point estimate with a wider confidence interval (aOR 0.14; 95% CI 0.02–1.15; p=0.068).

Process Efficiency

Among submissions that obtained approval, the median time to approval did not differ between RA-supported and non-supported submissions (58 vs 60 days, p=0.36) (Table 2).

Discussion

Principal Findings

In this 5-year institutional cohort of postgraduate surgical research activity, organizational research support in the form of structured research assistant (RA) involvement was associated with improved initial submission readiness. The RA-supported submissions were less likely to receive revision comments and demonstrated a lower revision burden. Importantly, these differences were confined to administrative and form-related domains, while methodological and scientific feedback remained comparable between the groups.

This pattern suggests that organizational support primarily influences procedural quality and regulatory navigation rather than altering the intellectual or scientific evaluation of research proposals. In this context, RA involvement appears to function as a workflow-level quality assurance mechanism, improving submission readiness without attenuating the rigor of scholarly critique. These findings align with prior work, suggesting that research support infrastructure can enhance process efficiency without necessarily modifying academic output quality.^{4–6,19–22}

Organizational Support as an Educational Infrastructure

Participation in research during postgraduate surgical training is widely recognized as a core professional competency; however, early stages of the research process particularly ethical approval may represent a threshold barrier for novice investigators.^9–12 In clinically intensive environments, procedural and regulatory requirements can compete with limited cognitive and temporal resources, potentially delaying or discouraging scholarly engagement.^7,8

The interpretation that follows is theoretical. Cognitive load was not measured in this study, and the mechanism proposed is a hypothesis for future testing rather than a finding. From a cognitive load theory perspective, administrative tasks associated with IRB submission may constitute extraneous cognitive load necessary for regulatory compliance but not directly contributory to methodological learning.^23–25 Excessive extraneous load may impair the acquisition of core competencies by diverting attention from meaningful learning activities. By providing structured pre-submission reviews and technical guidance, RA support could reduce this extraneous burden, preserving cognitive capacity for core scholarly tasks, such as study design, hypothesis refinement, and critical appraisal.

Within this framework, organizational research support could be conceptualized as part of the educational infrastructure of a training program, although the present data address only procedural outcomes. Consistent with socio-cultural learning theory, this form of support may function as an educational and operational scaffold, enabling trainees to engage in complex research processes beyond their independent capability while maintaining exposure to authentic scientific evaluation.^{13–15,21,26–29}

Standardization and Equity in Early Research Engagement

The covariate of investigator level (faculty, fellow, or resident) was not independently associated with revision likelihood after adjusting for RA support, review type, and year. Although causal inference cannot be established, structured administrative workflows may reduce variability in procedural compliance across research experience levels. Whether this translates into reduced disparities in research participation was not examined and is a hypothesis for future study.

This interpretation is consistent with prior literature demonstrating that structured research support and institutional infrastructure can reduce variability in research participation and performance across trainees.^{4–6,30–32} The comparable levels of methodological and scientific critique across groups further suggest that RA support did not dilute the evaluative function of the IRB.

Implementation of RA support also carries risks. Uptake in this cohort was voluntary, and trainees were over-represented among non-supported submissions (84% vs 52% of supported submissions), suggesting that those who might benefit most were less likely to use the service. Reasons for non-uptake were not recorded and may include lack of awareness, time pressure, or a preference for independence. If access depends on investigator initiative, such support could widen rather than narrow disparities. Programmes adopting this model should consider routing all trainee submissions through pre-submission review by default and monitoring uptake by investigator group.

Process Efficiency and System-Level Determinants

Despite fewer administrative and form-related revisions, the time to approval did not differ between groups. This suggests that IRB timelines are influenced predominantly by institutional-level processes, such as committee scheduling, reviewer workload, and administrative throughput rather than submission completeness alone.^9–12,33 Therefore, improvements in procedural quality may reduce friction within the review process without necessarily shortening total approval duration.

More granular process metrics such as the number of revision cycles, reviewer response time, and investigator turnaround time may better capture workflow efficiency and should be considered in future studies.

Implications for Postgraduate Research Training

The findings of this study have several implications for postgraduate medical education. Interventions to enhance trainee scholarship have traditionally focused on protected time, structured curricula, and mentorship models.^4–7 Our findings highlight a complementary systems-based strategy optimizing the procedural environment in which research is conducted. Whether reducing administrative barriers at the ethical submission stage leads to earlier or more sustained scholarly engagement was not assessed in this study and remains to be tested, particularly in high-volume surgical settings.^7,8

These findings should not be interpreted as evidence that administrative support enhances research competence, engagement, or learning; none of these outcomes were measured. Rather, organizational support may serve as a gateway facilitator improving access to the research process while preserving methodological rigor and independent scientific evaluation. Similar system-level approaches have been advocated in academic medicine to support trainee productivity and engagement.^{19–22,26–29} Prospective studies incorporating educational outcomes, such as research self-efficacy, engagement, and scholarly productivity, are needed to evaluate whether any downstream educational benefit exists.

Strengths and Limitations

This study has several strengths, including a 5-year institutional dataset with complete investigator identifiers, duplicate classification of revision comment domains with measured inter-rater agreement, and multivariable analysis with bootstrap and investigator-clustered sensitivity testing. These features strengthen the internal validity of the findings within a real-world training environment.

However, several limitations should be considered. First, RA utilization was voluntary and was adopted in 91% of submissions, leaving only 32 protocols in the comparison group. Selection into RA support was evident: trainees accounted for 84% of non-supported submissions compared with 52% of supported submissions. Investigators who declined RA support may also have differed in motivation, prior research experience, or familiarity with IRB procedures, none of which were captured in the dataset. Although we adjusted for investigator level, review type, and year, residual confounding is likely, and the small comparison group limits the precision of the effect estimates, as reflected in the wide confidence interval for the adjusted odds ratio (0.04–0.82). The findings should therefore be interpreted as associations, not causal effects.

Second, revision comments were used as proxy measures of procedural completeness and submission quality. These measures do not reflect the intrinsic scientific quality of the research and may be influenced by reviewer variability and protocol complexity, neither of which was captured.

Third, the likelihood of receiving comments increased over the study period independent of RA support. RA staffing and the pre-submission checklist were unchanged, so this trend most likely reflects evolving IRB reviewer expectations, which we could not characterize further.

Finally, the findings come from a single Thai academic center with a two-stage IRB process (staff screening followed by committee review), a departmental RA unit of eight staff, and a national residency curriculum. Institutions with different IRB structures, fewer administrative staff, or electronic submission systems that perform completeness checks automatically may observe smaller or different effects. The administrative burden of ethical approval is common across settings, but the size of any benefit from RA support will depend on how much of that burden is avoidable at each site.

Future Directions

Future research should focus on prospective and multi-institutional studies to validate these findings across diverse training environments. Incorporating measures of protocol complexity, prior research experience, and mentorship intensity would strengthen causal interpretation. Importantly, longitudinal educational outcomes such as research self-efficacy, engagement, and scholarly productivity should be evaluated to determine whether improvements in procedural access translate to sustained academic development. In addition, cost-effectiveness analyses may provide valuable insights into the scalability and sustainability of structured research support systems, particularly in resource-constrained settings.

Conclusions

Structured organizational research support was associated with improved procedural completeness of IRB submissions, with fewer administrative and form-related revisions and no difference in methodological or scientific critique. Whether such administrative scaffolding also lowers barriers to scholarly engagement was not measured and requires prospective evaluation with educational outcomes.

Disclosure

The authors report no conflicts of interest in this work.