From Data to Discovery: A Fellow's Guide to Starting a Research Project

A Practical, Step-by-Step Guide for the Uninitiated

Dr Neeraj Manikath , claude.ai

Abstract

Research productivity during fellowship training has become an increasingly important component of career advancement in internal medicine and its subspecialties. Despite this emphasis, most fellows receive minimal formal training in research methodology, leaving them to navigate the complex landscape of idea generation, institutional review board (IRB) approval, mentorship, and publication largely on their own. This practical guide provides a structured approach to initiating and completing a research project during fellowship, with emphasis on efficiency, strategic planning, and self-preservation. We present actionable strategies for identifying clinically relevant research questions, streamlining the IRB process, establishing productive mentor relationships, and strategically advancing projects from conception to national presentation and publication.

Introduction

The modern fellowship experience includes an often-unstated expectation: you will conduct research. Whether driven by program requirements, career aspirations, or competitive job markets, fellows face mounting pressure to produce scholarly work alongside demanding clinical responsibilities. Yet paradoxically, most trainees receive little practical guidance on how to actually accomplish this feat.

The result is predictable: anxiety, false starts, abandoned projects, and a pervasive sense that everyone else somehow knows what they're doing. This guide aims to demystify the research process by providing concrete, actionable strategies drawn from successful fellow-initiated projects across multiple institutions. Consider this your practical roadmap—not to becoming a principal investigator overnight, but to completing a meaningful project that advances your career without consuming your sanity.

Idea Generation: How to Find a Research Question in Your Daily Clinical Work

The Clinical-to-Research Pipeline

The best research questions don't emerge from literature reviews conducted at midnight; they arise organically from patterns you notice during clinical work. The key is developing what might be called "research eyes"—a habit of pattern recognition and curiosity that transforms routine observations into investigable hypotheses.

Pearl #1: The "Three Times Rule" If you find yourself explaining the same clinical concept to three different consultants, attendings, or patients within a month, you've likely identified a knowledge gap worth investigating. These recurrent educational moments often represent areas where evidence is ambiguous, guidelines are contradictory, or practice patterns vary widely—fertile ground for research.

The Daily Documentation Audit Start a simple note-taking practice: at the end of each clinical day, spend 60 seconds documenting one clinical surprise, one treatment decision you questioned, or one patient outcome you didn't expect. Review these notes monthly. Patterns will emerge—and these patterns are research questions in embryonic form.

Categories of High-Yield Fellow Projects

Not all research questions are created equal for trainees with limited time and resources. Focus on these project archetypes:

1. Quality Improvement Masquerading as Research Document systematic changes in clinical protocols you're already implementing. Did your program change its approach to managing gastrointestinal bleeding prophylaxis? That's a pre-post study waiting to happen.

2. The "Chart Review That Answers a Real Question" Retrospective analyses remain the workhorse of fellow research for good reason—they're feasible. The trick is asking questions that clinicians actually care about. Example: Rather than "Characteristics of patients with decompensated cirrhosis" (descriptive, low-yield), ask "Does timing of paracentesis within 24 hours versus after 24 hours of admission affect length of stay in patients with tense ascites?" (comparative, actionable).

3. Survey Studies That Fill Knowledge Voids If attendings disagree on management approaches during morning report, surveying regional or national practice patterns can yield publishable insights with minimal resources.

4. Novel Uses of Existing Databases Your institution's electronic health record contains years of structured data. Learning to extract and analyze it (or finding someone who can) unlocks enormous research potential.

Oyster Warning #1: The Unfalsifiable Question Avoid research questions so broad they're impossible to answer with your resources. "What is the optimal management of atrial fibrillation?" is a career's work. "Among hospitalized patients with new-onset atrial fibrillation, what proportion receive guideline-concordant anticoagulation within 24 hours of diagnosis?" is a fellow project.

The Elevator Test

Before committing to a question, test it: Can you explain why this matters to a non-specialist attending in the time it takes to ride an elevator (30 seconds)? If you can't articulate the clinical relevance quickly and clearly, the question needs refinement.

The "Minimum Viable" IRB Proposal: Getting Your Project Approved Without Losing Your Mind

Reframing the IRB Process

Most fellows approach the Institutional Review Board with dread, viewing it as a bureaucratic obstacle designed to thwart progress. Reframe this: the IRB is quality control for your research design. A well-constructed IRB application forces you to think through methodology, data security, and potential confounders before you've wasted months collecting unusable data.

The Three-Document Strategy

Every IRB submission requires three core components. Write these in order:

1. The One-Page Protocol Summary This is your project in plain English: What's the question? Who are the subjects? What data will you collect? How will you analyze it? Write this first—before touching the official IRB forms. This document becomes your north star; every subsequent form field should align with this summary.

2. The Data Collection Sheet Create your actual data collection instrument (REDCap database, Excel spreadsheet, or paper form). This forces specificity: you can't claim you'll "collect laboratory values" in your protocol while leaving it vague. The IRB wants to know which laboratory values, from what time period, and how you'll handle missing data.

Pearl #2: The "Thesis Sentence" Hack Your entire IRB application should support a single, specific thesis sentence. Write this at the top of your working document: "This retrospective chart review will compare readmission rates among patients who received structured telephone follow-up versus usual care after discharge for heart failure exacerbation." Every paragraph of your IRB protocol should reference back to elements of this sentence.

Navigating IRB Categories

Understanding IRB review levels saves enormous time:

Exempt: No identifiable data, no patient contact, minimal risk (retrospective chart reviews often qualify)
Expedited: Minimal risk research, often involving existing data or non-invasive procedures
Full Board Review: Greater than minimal risk, usually requires prospective patient enrollment with interventions

Most fellow projects can be designed for exempt or expedited review. The difference in approval time: weeks versus months.

Hack #1: The Retrospective Chart Review Default When possible, design retrospective studies. They typically qualify for expedited or exempt review, don't require patient consent, and allow you to verify that adequate patient numbers exist before committing months to a project.

Common IRB Pitfalls

The Scope Creep Death Spiral: Fellows often add "just one more variable" repeatedly during planning. Each addition complicates your protocol and data collection. Start minimal. You can always plan a follow-up study.

HIPAA Paranoia: Yes, patient privacy matters. No, you don't need to understand every nuance of HIPAA law. Most institutions have templated language for data security. Use it. Don't reinvent this wheel.

The Perfect Protocol Fallacy: Your IRB application doesn't need to read like a Nature manuscript. It needs to clearly describe a safe, ethical study with adequate methodology. Write clearly, not impressively.

The Two-Week Timeline

With focused effort, IRB submission is achievable in 10 working days:

Days 1-2: Literature review and thesis sentence refinement
Days 3-4: Write one-page protocol summary
Days 5-6: Design data collection instrument
Days 7-8: Complete IRB application forms
Day 9: Internal review by mentor/program director
Day 10: Submit

Pearl #3: The Pre-IRB Consultation Most IRB offices offer pre-submission consultations. Use this. A 15-minute conversation can prevent weeks of revisions. Come prepared with your one-page summary and specific questions about review category and data handling.

Finding a Mentor vs. a Boss: What to Look For and How to Manage the Relationship

The Mentor-Boss Spectrum

Not all research supervisors are created equal. Understanding the difference between a mentor and a boss prevents mismatched expectations and failed projects.

Research Boss Characteristics:

Assigns projects rather than collaborating on ideas
Views you primarily as labor for their research agenda
Expects extended work hours without consideration for clinical duties
Takes first or senior authorship regardless of your contribution
Provides minimal teaching or career development

Research Mentor Characteristics:

Helps refine your ideas rather than replacing them
Teaches research skills transferable beyond the current project
Respects clinical time constraints
Advocates for appropriate authorship
Invests in your long-term career trajectory

Most relationships fall somewhere between these extremes. The goal isn't finding a perfect mentor—they don't exist—but rather identifying someone whose strengths align with your needs and whose limitations you can work around.

The Mentor Selection Matrix

Evaluate potential mentors across four domains:

1. Methodological Expertise: Can they guide you through study design, statistics, and interpretation?

2. Content Knowledge: Do they understand the clinical area well enough to spot flawed assumptions?

3. Protected Time: Do they have institutional support for research, or are they squeezing mentorship between clinical obligations?

4. Track Record: Have their previous trainees published? Presented at national meetings? Secured jobs?

Oyster Warning #2: The Absent Superstar The division chief with 200 publications may look perfect on paper but lacks time for meaningful mentorship. Often, the ambitious junior faculty member hungry to build their research portfolio provides more hands-on support. Consider co-mentorship: one senior figure for credibility and career guidance, one junior faculty for day-to-day project management.

The First Meeting: Questions to Ask

Before committing, have a structured conversation addressing:

"What's your vision for trainee involvement in research?"
"How do you typically handle authorship decisions?"
"What's a realistic timeline for taking a project from idea to submission?"
"How often do you expect to meet about research projects?"
"Can you connect me with a fellow or junior faculty you've previously mentored?"

That last question is critical. Speaking with a mentor's previous trainees reveals more than any interview.

Managing the Relationship: Setting Boundaries

The Research Agreement Document After initial discussions but before substantive work begins, create a brief written agreement outlining:

Project scope and timeline
Expected time commitment (hours per week)
Meeting frequency
Authorship plan
Roles and responsibilities

This isn't legally binding, but it prevents the mission creep that derails projects. Most mentors will respect clear boundaries if you establish them early.

Hack #2: The Bi-Weekly Research Meeting Schedule recurring 30-minute meetings (in-person or virtual) every two weeks. This creates accountability without overwhelming either party. Come to each meeting with:

Progress since last meeting
Current roadblock
Specific question or decision needed

This structure keeps projects moving while respecting everyone's time.

Pearl #4: The Proactive Communication Strategy Don't wait for your mentor to ask about progress. Send brief weekly email updates (3-4 sentences) even when progress is minimal: "This week: extracted data for 60 patients. Next week: complete data extraction and begin data cleaning. No obstacles currently." This demonstrates commitment and prevents mentors from assuming you've abandoned the project during busy clinical rotations.

From Poster to Publication: A Timeline for Taking Your Project to a National Meeting

Strategic Timeline Planning

The path from IRB approval to publication spans 12-18 months for most projects. Here's the realistic timeline with strategic decision points:

Months 1-2: Data Collection

For retrospective studies: 4-8 weeks
For prospective studies: 3-6 months (often extending beyond fellowship)
For survey studies: 6-8 weeks

Pearl #5: The Abstract-Before-Analysis Approach Draft your abstract before completing full analysis. This seems backwards but forces clarity about your primary outcome and key comparisons. You'll revise it, but this initial draft prevents drowning in tangential analyses.

Months 3-4: Analysis and Initial Presentation

Data cleaning and analysis: 2-4 weeks
Abstract preparation: 1 week
Internal presentation (division meeting): 1 week

Months 5-6: National Meeting Submission Most major societies (American College of Physicians, specialty society meetings) have abstract deadlines 6-8 months before conferences. Submit early in your fellowship timeline.

Hack #3: The "Two-Meeting" Strategy Submit to both regional and national meetings. Regional meetings (e.g., regional ACP chapters) often have higher acceptance rates and earlier deadlines, providing presentation experience and preliminary feedback before national venues. Many allow previously presented work, making this a no-risk strategy.

Months 7-12: Meeting Presentation and Manuscript Preparation

Poster design: 1-2 weeks
Meeting attendance: factor in time off requests
Manuscript first draft: 4-6 weeks
Mentor review cycles: 2-4 weeks
Journal submission: allow 2 weeks for formatting and final review

Months 12-18: Peer Review and Publication

Journal review process: 6-12 weeks (often longer)
Revisions: 2-4 weeks
Acceptance to online publication: 4-8 weeks

The Poster Versus Oral Presentation

Poster Advantages:

Higher acceptance rates
Less preparation time
More relaxed presentation environment
Better for complex data or methods

Oral Presentation Advantages:

Higher prestige
Better networking opportunity
More concise message refinement
Stronger CV line

Oyster Warning #3: The Unpublished Presentation Presenting at a national meeting doesn't guarantee publication. Studies suggest less than 50% of presented abstracts reach full publication. Plan manuscript writing before the meeting, when enthusiasm and accountability are highest.

Manuscript Preparation Strategy

The Reverse-Order Writing Method Don't write manuscripts from introduction to conclusion. Use this sequence:

Methods: Write while fresh from IRB approval and data collection
Results: Complete immediately after analysis
Tables and Figures: Create before writing results text
Discussion: Draft with your main findings displayed
Introduction: Write last, now knowing exactly where you're heading
Abstract: Final component, summarizing the complete story

Pearl #6: The "Three-Journal" Submission Strategy Before writing, identify three target journals ranked by impact and feasibility:

Reach journal (higher impact, lower probability)
Target journal (realistic acceptance chance)
Safety journal (high acceptance probability)

Tailor your manuscript to the reach journal's format, but if rejected, you can quickly reformulate for the target journal without complete rewriting.

Timeline Hacks for Busy Fellows

The Protected Research Half-Day Negotiate one protected half-day weekly for research. Four hours of focused time weekly accomplishes more than scattered late-night work.

The Co-Fellow Strategy Partnering with another fellow distributes workload and maintains momentum during heavy clinical rotations. Ensure clear role division to avoid authorship conflicts.

The Professional Medical Writer Consideration Some institutions provide biostatistical and medical writing support. Using these resources isn't "cheating"—it's efficient use of expertise. Your value is clinical insight and study design, not necessarily technical writing skill.

How to Say "No" to Low-Yield Projects That Drain Your Time

The Opportunity Cost Framework

Every research project has two costs: time invested and opportunity cost of alternative projects. Fellows typically have capacity for one, maximum two, meaningful research projects during training. Saying yes to the wrong project means saying no to the right one.

Red Flags: Projects to Avoid

The Endless Data Collection Study Any project requiring more than 200 chart reviews or more than 3 months of prospective enrollment deserves scrutiny. Either scale it down or question whether it's realistic for fellowship.

The "We've Been Collecting Data for Years" Project Inherited projects often have fatal flaws: unclear hypotheses, inconsistent data collection, or analytic challenges. Unless you're joining a well-organized team, these are time sinks.

The Case Report Trap Single case reports rarely advance careers and consume disproportionate time for minimal payoff. Case series (n≥3) of truly unusual presentations have more value, but even these have limited utility unless your career goals include clinical observation rather than research methodology.

The Unfunded Multicenter Trial Unless you're joining an established collaborative group with dedicated research infrastructure, proposing multicenter prospective studies during fellowship demonstrates poor judgment, not ambition.

Oyster Warning #4: The Ego Project Beware of projects you're pursuing because they sound impressive rather than because they're feasible and interesting. "Genomic predictors of treatment response" sounds better than "Chart review of antibiotic timing," but the latter is far more likely to succeed—and successful completion beats impressive failure.

The Polite Decline: Scripting Your "No"

Fellows fear declining projects will damage relationships. The opposite is true: strategic declining establishes you as thoughtful rather than indiscriminant.

Script for Mentor-Proposed Projects: "I really appreciate you thinking of me for this project. Given my current clinical schedule and my committed project on [your existing work], I don't think I could give this the attention it deserves. Could we revisit this in [specific timeframe], or is there someone else in the division who might be interested?"

Script for Peer-Initiated Projects: "This sounds like an important question. I'm at capacity with my current project, but I'd be happy to provide feedback on your study design if helpful."

Script for Division Leadership Requests: "I'm committed to contributing to divisional research. Could we discuss how this fits with my educational goals and timeline? I want to ensure I can deliver quality work rather than overcommitting."

The "Yes, But" Alternative

Sometimes declining entirely isn't possible or advisable. The "yes, but" response sets boundaries while demonstrating collegiality:

"Yes, I can help with data collection for 3 hours weekly, but I can't take on data analysis."
"Yes, I'm interested, but only if we can define a narrow scope that's feasible in 6 months."
"Yes, I'll participate if we can bring on another fellow to share responsibilities."

Hack #4: The Co-Author Contribution Offering minor contributions (e.g., "I can review the manuscript draft and offer clinical input") allows you to support colleagues and add a middle-author publication without derailing your primary work.

Protecting Your Time: The Research Prenup

Before agreeing to any project, clarify:

Time Commitment: "How many hours weekly do you envision this requiring?"
Duration: "What's the expected timeline from start to submission?"
Your Role: "What specific tasks would I be responsible for?"
Authorship: "How are you planning to determine author order?"
Mentorship: "How often will we meet to discuss progress and problem-solve?"

If the proposer hasn't thought through these questions, the project isn't ready for your commitment.

Conclusion: Research as Professional Development

Research during fellowship serves multiple purposes beyond CV building. It teaches critical appraisal, cultivates intellectual curiosity, and develops skills in project management and scientific communication—all transferable to clinical practice. The key is approaching research strategically: selecting feasible projects, establishing productive mentorship, managing timelines realistically, and protecting your time against low-yield distractions.

Remember: completion trumps perfection. A finished chart review that's published in a regional journal advances your career more than an abandoned prospective trial that was going to revolutionize medicine. Start small, finish strong, and build from there.

The research landscape needn't be navigated through trial and error. With intentional planning and strategic decision-making, fellows can produce meaningful scholarly work that enhances rather than detracts from clinical training. The goal isn't becoming a research superstar overnight—it's developing as a clinician-scholar capable of asking good questions, generating reliable evidence, and contributing to medical knowledge throughout your career.

Key Takeaways: Pearls for Success

The Three Times Rule: Recurrent clinical teaching moments reveal knowledge gaps worth investigating
The Thesis Sentence Hack: One clear sentence should guide your entire IRB application
The Pre-IRB Consultation: 15 minutes with IRB staff prevents weeks of revisions
The Proactive Communication Strategy: Brief weekly updates maintain mentor engagement
The Abstract-Before-Analysis Approach: Draft your abstract before full analysis to maintain focus
The Three-Journal Submission Strategy: Identify reach, target, and safety journals before writing

Oyster Warnings: Pitfalls to Avoid

The Unfalsifiable Question: Avoid questions too broad to answer with available resources
The Absent Superstar: Famous mentors often lack time for meaningful guidance
The Unpublished Presentation: Meeting presentations don't guarantee publication—plan manuscript writing early
The Ego Project: Feasibility and completion trump impressive-sounding failures

References

Levine RB, Hebert RS, Wright SM. Resident research and scholarly activity in internal medicine residency training programs. J Gen Intern Med. 2005;20(2):155-159.
Rivera JA, Levine RB, Wright SM. Completing a scholarly project during residency training: perspectives of residents who have been successful. J Gen Intern Med. 2005;20(4):366-369.
Kanna B, Deng C, Erickson SN, et al. The research rotation: competency-based structured and novel approach to research training of internal medicine residents. BMC Med Educ. 2006;6:52.
Chusid MJ, Kozel FJ, Wesolowski W, et al. The health sciences manuscript review project: a model for reviewing manuscripts and teaching scholarly writing. Teach Learn Med. 1992;4(4):208-213.
Weaver AN, McCaw TR, Fifolt M, et al. Impact of elective versus required internal medicine residency research experiences on career outcomes. J Grad Med Educ. 2017;9(3):341-345.
Alguire PC, Anderson WA, Albrecht RR, Poland GA. Resident research in internal medicine residency programs: a review of structures and outcomes. Am J Med. 1996;100(3):357-363.
Segal S, Lloyd T, Houts PS, et al. The association between students' research involvement in medical school and their postgraduate medical activities. Acad Med. 1990;65(8):530-533.
Schoenfeld JD, Ioannidis JP. Is everything we eat associated with cancer? A systematic cookbook review. Am J Clin Nutr. 2013;97(1):127-134.
Bordage G, Dawson B. Experimental study design and grant writing in eight steps and 28 questions. Med Educ. 2003;37(4):376-385.
DeMaria AN. The peer review process. J Am Coll Cardiol. 2009;53(13):1166.
Scherer RW, Langenberg P, von Elm E. Full publication of results initially presented in abstracts. Cochrane Database Syst Rev. 2007;(2):MR000005.
Mimouni M, Krauthammer M, Gershenbaum E, et al. Trends in authorship in ophthalmology: a bibliometric analysis from 1975 to 2020. Am J Ophthalmol. 2021;228:117-124.
Emanuel EJ, Wendler D, Grady C. What makes clinical research ethical? JAMA. 2000;283(20):2701-2711.
Green LA, Seifert CM Jr. Translation of research into practice: why we can't "just do it." J Am Board Fam Pract. 2005;18(6):541-545.
Rosenblatt M, Chretien KC. Flipping the script on trainees and faculty to develop research skills. J Grad Med Educ. 2019;11(1):23-25.