Source-by-Source Outline vs. Thematic Outline: Choosing the Right Structure for an Evidence Brief

You have a folder of papers and a deadline. The first structural decision you make — whether to outline paper by paper or theme by theme — shapes how you search, what you extract, and whether your brief answers a question or merely summarizes a reading list. This article compares both approaches on the same task, shows where each fails, and gives you a decision table plus a combined workflow you can run for a focused clinical evidence brief.

Before choosing a structure, name your starting state. Most evidence briefs begin in one of three places:

• A question without a corpus. You know the population and outcome you care about, but you have not yet assembled the literature. Here the outline is a search plan, and structure should follow eligibility criteria.
• A corpus without a question. You have saved papers but no sharp focus. A source-by-source pass can help you see what the literature covers before you commit to themes.
• A question and a corpus. You have both, and the outline is now an extraction and synthesis device.

Cochrane guidance supports planning searches from eligibility criteria and documenting the process, which favors deciding your structure early rather than after reading. NIH rigor guidance similarly treats assessment of prior knowledge, strengths, weaknesses, and knowledge gaps as core to rigorous practice. Both point to the same first artifact: a written scope statement naming your population, exposure or intervention category, comparator context, and outcome — even when you are not conducting a systematic review.

Use a single worked example throughout: a brief on a defined adult population and one clinical outcome, drawing on a modest set of primary studies and reviews. The task is to produce a short evidence brief for a team discussion.

Source-by-source outline. Each H2 is a study or report. Under each, you record design, population, intervention or exposure, outcome definition, result, and stated limitations. The synthesis appears at the end, if at all.

Thematic outline. Each H2 is a question or concept — for example, consistency of the outcome definition across studies, or differences between trial and observational evidence. Studies appear as evidence within themes, and the same study may support more than one theme.

The two structures answer different reader questions. A source-by-source outline answers "what does each paper say?" A thematic outline answers "what do we know about this question, and how strong is it?"

A source-by-source structure is the right first pass when your corpus is small, heterogeneous, or unfamiliar. It forces you to record each study's design and population before you generalize, which surfaces exactly the differences that later become themes.

It also maps cleanly onto staged screening. Cochrane recommends screening in stages and linking multiple reports of one study; a per-study outline gives each study a home and makes duplicate reports visible rather than letting them inflate a theme. When you later need to report search and selection transparently, PRISMA 2020's flow-diagram templates assume you can account for records and studies individually — an artifact a source-by-source pass produces naturally.

Use it when you are building the extraction table that a thematic draft will later draw on, or when your audience needs a study-by-study appraisal.

Its weakness appears as the corpus grows. A stack of per-study sections tells the reader what each paper reported but not how the papers relate. Contradictions sit in separate sections with no reconciliation. A reader cannot tell whether ten studies agree because the evidence is consistent or because the outcome was defined differently in each.

It also hides indirectness. If your population of interest is narrower than the populations studied, a per-study format buries that mismatch inside individual entries instead of raising it as a cross-cutting limitation. And it scales poorly: a brief with twenty study sections is a reading list with headings, not a synthesis.

The failure mode is structural, not a matter of effort. More careful per-study summaries do not fix it.

A thematic structure puts the reader's question at the top of each section. It lets you place converging and conflicting findings side by side, distinguish evidence level from applicability, and state limitations once at the level where they operate — for example, that most included studies used a surrogate or short-term outcome.

It also matches how reporting guidance frames synthesis. PRISMA 2020 addresses transparent reporting of search, selection, and synthesis methods, and synthesis is inherently thematic. When your brief must support a decision discussion, themes let you say what is consistent, what is uncertain, and what would change the picture.

Use it when your question is settled and your corpus is large enough that patterns matter more than individual papers.

Thematic outlines fail in two directions. First, themes chosen before the corpus is mapped tend to reflect the reviewer's assumptions; studies that do not fit a theme get dropped or forced. Second, themes can drift into overgeneralization — a heading like "benefit in adults" can absorb studies with different populations, comparators, and outcome definitions unless each theme carries its own evidence table.

A thematic outline also makes provenance harder to audit. If a reader asks which studies support a claim, you need a traceable link from theme to study. Without that, the outline is persuasive but not checkable.

The practical answer for most evidence briefs is sequential, not either-or.

1. Write the scope statement. Name the population, the exposure or intervention category, the comparator context, and the outcome. This is your eligibility anchor.
2. Plan and document the search. Cochrane recommends combining free-text terms with subject headings, using multiple sources, deduplicating records, and rerunning searches before publication. Record what you did as you go.
3. Run a source-by-source extraction pass. One entry per study: design, population, exposure, outcome definition, result, limitations. Link multiple reports of the same study rather than counting them twice.
4. Derive themes from the extraction table. Let patterns in design, population, and outcome definition generate your H2s rather than imposing them in advance.
5. Rebuild the outline thematically, keeping the study map. Each theme cites the studies behind it and states its own limitations.
6. Check the result against your scope statement. Every theme should bear on the named population and outcome; anything that does not belongs in a clearly labeled adjacent section or out of the brief.

Where WisPaper fits is at the search and organization steps. You can find academic papers through WisPaper search, save and organize the corpus in My Library, and then ask questions grounded in those saved papers using Library QA — useful when you want to check how a theme is represented across your own collection rather than across the open web. For longer synthesis tasks, Scholar Agent supports literature review work. If your brief will carry a reference list, TrueCite checks BibTeX against real academic databases to flag hallucinated references before the brief circulates.

These four are often collapsed into a single "quality" note. Keep them separate in either outline type:

• Evidence level is about design and the strength of inference the study permits.
• Applicability is whether the studied context resembles the one your readers care about.
• Indirectness is the specific mismatch — population, outcome definition, or comparator — between the evidence and your question.
• Limitations are the study's own constraints, stated by its authors or visible in its methods.

In a source-by-source outline, record all four per study. In a thematic outline, aggregate them per theme: state the predominant evidence level, flag where applicability narrows, name the indirectness explicitly, and list limitations that recur across studies. This is where the thematic structure earns its keep — a limitation shared by eight studies is a property of the evidence base, not of any one paper.

A brief is ready when a reader can move from any claim to the studies behind it and back again. Practical checks:

• Every theme names the studies that support it, and every included study appears in at least one theme or in a labeled excluded list.
• The scope statement's population and outcome are visible in the structure, not just the introduction.
• Conflicting findings sit together with an explicit statement of what differs.
• Evidence level, applicability, indirectness, and limitations are addressed at the theme level.
• The search and selection process is documented well enough to describe in a methods note, consistent with Cochrane's documentation guidance and PRISMA 2020's reporting expectations.
• References have been checked against real records before circulation.

If any check fails, the fix is usually structural: return to the extraction table, not to the prose.