What are the steps to conduct a medical research paper?

How to Conduct a Medical Research Paper

The foundation of any successful medical research paper begins with developing a clear, focused research question using the PICO framework (Population, Intervention, Comparator, Outcome), followed by a rigorous protocol that serves as your roadmap through study design, data collection, analysis, and dissemination. 1

Phase 1: Formulating the Research Question

Apply the FINER Criteria

Your research question must be Feasible, Interesting, Novel, Ethical, and Relevant (FINER) to ensure it generates clinically impactful knowledge. 1, 2 This prevents wasting resources on questions of marginal interest or those that ignore important existing work. 1

Structure Using PICO Framework

• Population: Define your target patient population with specific characteristics 1, 3
• Intervention/Exposure: Specify the intervention or exposure being studied 1, 3
• Comparator: Identify what you're comparing against (control, standard care, placebo) 1, 3
• Outcome: Define outcome measures as precisely as if conducting an original study yourself 1
• Timeframe: Establish the duration reasonable for outcome development 1

Conduct Comprehensive Literature Review

Perform an extensive literature search to understand what is already known, identify gaps in current evidence, and justify why a new study is needed. 1 Search ClinicalTrials.gov to avoid unnecessary duplication and learn from similar ongoing studies. 1

Phase 2: Protocol Development

Create a Detailed Written Protocol

A well-developed protocol is the single most critical determinant of overall study quality and potential bias. 1, 3 The protocol serves as your complete roadmap and should include: 1, 3

• Study rationale and objectives: Classify as descriptive and/or analytical 3
• Study design: Clearly identify whether RCT, prospective cohort, case-control, cross-sectional, or case series 1
• Eligibility criteria: Detailed inclusion/exclusion criteria 3
• Sample size calculation: Pre-planned with power calculations 3
• Data collection methods: Specify timing and assessment methods for all variables 3
• Statistical analysis plan: Include pre-planned strategies to identify and mitigate bias 3

Register Your Protocol

Register the protocol prospectively in ClinicalTrials.gov or appropriate registry before patient enrollment or data collection begins. 1, 3 This prevents selective publication, protocol changes, and unnecessary research duplication. 1

Adhere to Reporting Standards

Follow established guidelines: 3

• PRISMA-P for systematic reviews and meta-analyses 1
• SPIRIT for clinical trial protocols 3
• ARRIVE for animal research 3

Phase 3: Study Design Selection

Match Design to Question

The study design must be appropriate for the specific question being asked. 1 Consider this hierarchy: 1

• Randomized Controlled Trials (RCTs): Provide maximum-quality evidence on benefits and risks of interventions; use when sufficient preliminary data exists 1
• Prospective Cohort Studies: Preferred for higher-level evidence when RCTs are not feasible 1, 3
• Registries: Essential for assessing real-world use, costs, and effectiveness 1
• Observational Studies: Appropriate when generating hypotheses or preparing ground for RCTs 1

Common Pitfall to Avoid

Do not combine results from observational cohorts and randomized trials in the same analysis, as differences in study design, follow-up duration, and exposure definitions lead to misleading findings. 1

Phase 4: Data Collection and Management

Define All Variables Precisely

Provide a complete list of core variables grouped as: 3

• Baseline characteristics
• Exposures
• Outcomes (primary and secondary, pre-specified before analysis) 3

Describe each variable in detail including timing of collection and assessment methods. 3

Ensure Reproducibility

Test and report: 1

• Observer reproducibility: Inter-observer and intra-observer variability
• Test-retest reproducibility: Using intra-class correlation coefficients (ICC) with 95% CI and Bland-Altman analysis as minimum standards 1
• Never use simple correlations or mean value comparisons for reporting reproducibility 1

Enroll Consecutively

Begin enrolling consecutive patients at study onset to avoid selection bias. 3 Document recruitment strategies to understand potential biases in your target sample. 3

Phase 5: Statistical Analysis

Pre-Specify Analysis Plan

The statistical analysis plan must be developed during protocol creation, not after data collection. 3 Include: 1

• For prognostic studies: Univariable and multivariable analysis (logistic or Cox regression) adjusted for traditional risk factors 1
• For prediction models: Assess discrimination (ROC curve, sensitivity, specificity), calibration (calibration slopes, Hosmer-Lemeshow test), and additional predictive value (C-statistic increment, NRI, IDI) 1
• For heterogeneity assessment: Use I² and Cochran Q statistics with forest plots for visual inspection 1

Collaborate with Statisticians

Engage expert statisticians early for sample size estimation and analysis planning. 1 This collaboration is essential for avoiding fatal methodological flaws. 1

Phase 6: Quality Assessment and Bias Mitigation

Assess Study Quality

Use established tools to assess potential for bias: 1

• Risk of bias assessment tools for included studies
• GRADE or similar frameworks for certainty of evidence 1

Address Ethical Considerations

Obtain appropriate institutional review board or ethics committee approval before study initiation. 3 Include trial registration numbers in all publications. 3

Phase 7: Dissemination

Plan Multi-Directional Dissemination

• Identify target audiences (specialists, clinicians, patients, stakeholders) 1
• Use multiple venues while respecting policies on data reuse 1
• Encourage collaborator participation to amplify dissemination 1

Report Completely and Transparently

Follow reporting guidelines specific to your study design to ensure concise, explicit, and complete reporting. 4 Both positive and negative findings should be published to advance the field. 1

Critical Success Factors

The three most common reasons research fails are: 1

1. Poorly defined research question leading to inappropriate study inclusion and substantial heterogeneity 1
2. Vague protocol criteria introducing unnecessary subjectivity during screening and data extraction 1
3. Inadequate subject matter expertise in translating the hypothesis into an executable protocol 1

Avoid these pitfalls by investing substantial time upfront in question formulation and protocol development before collecting any data. 1