Why the first 90 days matter
After placement results, clinical workload, rotations, on-call duties and academic expectations start almost at the same time. Thesis planning is often postponed until the clinical routine becomes more predictable. The cost of this delay is usually paid later: ethics approval is late, data access is unclear, sample size is insufficient and the methods section is written under pressure.
The first goal is not to find a perfect topic. It is to build a feasible research line. Ask three questions early: can I actually access these patients or records? Are the measurements standardized and reliable? Does the intended statistical analysis match the research question? If one of these answers is unclear, the project may be attractive but not feasible.
Choose a topic that fits clinical reality
A good residency thesis aligns a meaningful clinical question with accessible data. Retrospective chart reviews, prospective observational studies, surveys, case series and laboratory-marker analyses have different timelines and ethical requirements. Before fixing the title, evaluate patient volume, follow-up duration, documentation quality, measurement consistency and supervisor expertise.
A broad title such as “clinical characteristics of X disease” may look safe, but it often becomes unmanageable during analysis. A stronger title specifies the population, primary outcome and main comparison. This structure makes the protocol, data dictionary, statistical analysis plan and final manuscript easier to build.
Data planning comes before statistics
Statistical analysis is not a technical step added after data collection. Variable coding, missing data rules, continuous and categorical summaries, group definitions and primary analyses should be planned before the spreadsheet is filled. Without that plan, the analysis can become a search for any significant result rather than a test of a defensible question.
Residents should prepare a compact protocol before data extraction. This protocol should include the primary aim, secondary aims, inclusion and exclusion criteria, variable dictionary, expected sample size and analysis plan. It supports the ethics submission, guides supervisor meetings and later becomes the backbone of the methods section.
Plan the thesis as a future manuscript
The thesis and the manuscript should not be treated as unrelated outputs. If the thesis is planned with publication in mind, tables, figures, statistical reporting and discussion structure become clearer from the start. This reduces the effort needed to convert the thesis into an article after the defense.
Journal strategy should be realistic. Study design, sample size, novelty and clinical message determine the target journal range. Not every residency project should start with an unrealistic high-impact target. A short communication, case series, retrospective study or review may be more appropriate depending on the evidence.
Five avoidable mistakes
- Thinking about ethics approval after data extraction has started.
- Collecting many variables without defining a primary outcome.
- Using variables that are not recorded consistently in the clinical archive.
- Separating the statistical analysis plan from the methods section.
- Assuming that converting a thesis into a manuscript is only a language-editing task.
Boss Academy supports residents and clinical researchers with research question development, ethics-file preparation, data templates, statistical analysis and manuscript conversion. The objective is a defensible, transparent and publishable workflow.
FAQ
When should a residency thesis topic be chosen?
Ideally, the research line should be clarified during the first few months, after checking supervisor fit and data access.
Is a retrospective study always easier?
No. Retrospective studies can be practical, but missing data, inconsistent documentation and bias must be planned for carefully.
What is needed to publish a thesis?
A clear question, clean dataset, appropriate analysis, journal-specific formatting and a discussion that does not overstate the findings.