Academic Referencing
Because the examiner can tell when you're making your sources up. (And so can Google Scholar.)
Referencing is not just a bureaucratic chore — it is proof that you have read widely, thought critically, and built your work on solid ground. Whether you are writing a QIP, submitting a journal article, completing a university essay, or supporting a trainee's portfolio, this guide gives you everything you need.
A hand-picked mix of official guidance and real-world academic writing resources. Because the best pearls are not always hiding in the official documents.
Core Referencing Guides
- Academic Skills Library — University of Leeds (Excellent advice on all aspects of writing and publishing)Official
- Cite.me — Free Online Vancouver & Harvard Reference GeneratorTool
- Zotero — Free Reference Management Software (Highly recommended — free forever)Software
- Mendeley — Reference Manager & PDF Organiser (Free for basic use)Software
Harvard Referencing
Vancouver Referencing (Medical & Scientific)
GP Training & Academic Writing
Plagiarism, Academic Integrity & AI
⚡ Quick Summary — If You Only Read One Thing
- There are two main referencing systems used in medicine: Harvard (author-date) and Vancouver (numbered). Both are widely accepted. Most medical journals use Vancouver.
- Before you start writing, ask your publisher, university, or educational body which system they prefer. Always check their "house style guide". Switching mid-project is painful.
- In GP training, referencing matters most for QIPs, audits, case reports, university essays, and journal submissions. Your ARCP assessors notice a strong, well-cited literature review.
- Reference management software (Zotero is free; Mendeley is free for basic use) saves enormous time. Set it up before you start writing — not after 200 references have accumulated.
- Plagiarism detection tools are widely used. AI-generated text is increasingly detectable. Always write in your own words, cite your sources, and — if you use AI tools — follow your institution's disclosure policy.
- When in doubt: cite it. The worst that can happen is an over-referenced essay. The alternative is academic misconduct.
More than just ticking a box — here is what it actually shows about you as a doctor.
🔍 Credibility
A referenced argument shows you have read the evidence, not just your own opinion. Assessors — and editors — take referenced work far more seriously.
🛡️ Academic Integrity
Referencing acknowledges other people's ideas and work. Without it, using their ideas is — legally and ethically — plagiarism, even if unintentional.
🔗 Traceability
Readers can follow your sources, verify your claims, and build on your work. In medicine, this traceability matters for patient safety, not just for marks.
Everything in academic medicine flows from understanding these two families of referencing.
The golden rule — before you write a single word:
Ask your publisher, university, or educational organisation which system they use. Ask for their "house style guide". Being clear about this from the outset saves hours of painful reformatting later. (Yes, this is worth repeating — it really is that important.)
Harvard Style
Also called: Author-date system
- In-text:
(Smith, 2023) - Reference list at end, alphabetical by author
- Popular in: social sciences, health sciences, UK universities
- Reader sees author name instantly in text
Vancouver Style
Also called: Numbered / author-number system
- In-text:
(1)or superscript¹ - Reference list at end, numbered by order of first appearance
- Popular in: medicine, biomedicine, BMJ, Lancet, BJGP
- Keeps text clean; reader looks up number if needed
The author-date system. Clean, instantly readable, and widely used in UK universities.
In Harvard, every citation in your text includes the author's surname and the year of publication in brackets. The full details live in a reference list at the end of your document, sorted alphabetically by the first author's surname.
In-text citation — two formats:
or
(2022) argued that...
Use the first format when you want to cite at the end of a sentence. Use the second when the author is part of your sentence.
Two or more authors:
Three or more: (Jones et al., 2022)
Note: et al. is Latin for "and others" — always italicised.
Below are worked examples for the source types you are most likely to encounter in GP training writing. The format is: Author(s). (Year). Title. Source details.
📄 Journal Article
📚 Book (single author)
📚 Chapter in an edited book
🌐 Website / Webpage
📋 NICE / Clinical Guideline
- If you cite the same author multiple times in the same year, use letters: (Smith, 2022a), (Smith, 2022b)
- If there is no known author, use the organisation name or the title of the document
- For online sources, always include the date you accessed the page — websites change and can disappear
- A DOI (Digital Object Identifier) is more reliable than a URL — use it when available
- Reference list = sources you actually cited. Bibliography = all sources you read, cited or not. Know the difference before you submit.
The numbered system. Widely used in biomedical publishing — the language of medical journals.
In Vancouver, every source is given a number in the order it first appears in your text. If you cite the same source again later, it keeps its original number. Numbers appear in the text as (1), [1], or as superscripts¹. The full reference list at the end is arranged numerically — not alphabetically.
In-text citation:
Neighbour has described this as the inner consultation [2].
Recent evidence suggests this approach is effective ³.
All three formats are acceptable — check your journal's preference.
Multiple citations at once:
Non-consecutive: (1, 3, 7)
Mixed: (1–3, 7, 9)
A number range (1–3) means references 1, 2 and 3 all cited together.
Vancouver uses minimal punctuation and no italics. Journal names are typically abbreviated (e.g. N Engl J Med for New England Journal of Medicine). For more than six authors, list the first six then write "et al."
📄 Journal Article
📚 Book
📚 Chapter in edited book
🌐 Website
📄 Journal article — more than 6 authors
The Vancouver system was created in 1978 when a group of medical journal editors met in Vancouver, Canada, and agreed to a unified set of requirements for submitted manuscripts. This group became the International Committee of Medical Journal Editors (ICMJE). Their guidelines — now called the ICMJE Recommendations — standardised how medical research is submitted, reported, and cited. The numbered reference format they adopted became known as the Vancouver system. Today, hundreds of biomedical journals — including the BMJ, The Lancet, the New England Journal of Medicine, and the BJGP — follow Vancouver style. The system's clean, numbered approach keeps text readable and is ideal for technical papers with many citations.
The same information referenced in both systems, so you can see exactly how they differ.
| Feature | Harvard | Vancouver |
|---|---|---|
| In-text format | (Smith, 2022) or Smith (2022) | (1) or [1] or superscript¹ |
| Reference list order | Alphabetical by first author's surname | Numerical (order of appearance in text) |
| Same source cited twice | Same citation format both times | Same number used both times |
| Journal name style | Full name, italicised | Standard abbreviation (e.g. BMJ) |
| Italics/bold used? | Yes — journal/book titles italicised | Minimal — plain style preferred |
| Typical word count impact | Slightly longer (author name in text) | Shorter in-text (just a number) |
| Reader experience | Sees author name without looking it up | Cleaner text; must check list for author |
| Medical journals | Less common | BMJ, Lancet, NEJM, BJGP — standard |
| UK universities | Very common | Less common outside medical schools |
| Best for | Essays, dissertations, university work | Journal articles, research papers, QIPs |
🔁 The Same Sentence — Two Ways
Harvard Style
...
Reference list (alphabetical):
Royal College of General Practitioners (2022). GP Curriculum. London: RCGP.
Vancouver Style
...
Reference list (numbered):
1. Royal College of General Practitioners. GP Curriculum. London: RCGP; 2022.
🌳 Which System Should I Use?
Has your publisher / university specified a style?
If YES → use that style. No further questions.
Is it a medical journal submission?
If YES → Vancouver is almost always correct.
Is it a UK university essay or QIP?
If YES → Harvard is usually expected; check with your deanery or supervisor.
No guidance given at all?
Pick one. Use it consistently throughout. Vancouver saves word count.
The sources GP trainees cite most often — with worked examples in both systems, side by side.
How to use this cheat sheet
Each card below shows one source type. The H tab shows the Harvard format. The V tab shows Vancouver. The red text marks what changes between the two. Copy the pattern — replace the coloured parts with your own details.
Journal Article
Most common in medical writing💡 Key difference: Harvard puts year in brackets after author. Vancouver uses a number and puts year after journal name. Vancouver uses minimal punctuation and abbreviated journal titles.
NICE Guideline
Essential for any GP QIP or essay💡 Always include the guideline code (NG136, CG127 etc.) — this identifies the exact document. Always include the access date for any online source.
Book
e.g. Greenhalgh, Neighbour, Kumar & Clark💡 Vancouver: title is NOT italicised and is in sentence case (only first word and proper nouns capitalised). Harvard: title IS italicised and uses Title Case.
Website / Webpage
e.g. NHS.uk, BNF online, RCGP website💡 Vancouver uses [Internet] after the title for online sources and formats the date as Year Month Day (e.g. 2026 Apr 12). Harvard uses plain English (12 April 2026).
British National Formulary (BNF)
Commonly cited in prescribing-related QIPs💡 The BNF is updated continuously. Always include both the year and your access date. The "author" is the Joint Formulary Committee — use this exact name for the organisation.
Grey Literature & Reports
e.g. NHS reports, deanery documents, GMC guidance💡 Grey literature (reports, guidance, policy documents) follows the same pattern as websites. Always include the publishing organisation, year, and access date.
From finding a source to submitting your work — every step at a glance.
Where you are actually going to need this — and what standard is expected.
📋 Quality Improvement Projects (QIPs)
RCGP expects trainees to describe accepted best practice based on a literature search, citing multiple sources. A single NICE guideline citation is acceptable but weak. Citing several peer-reviewed papers alongside NICE guidance is strong.
🎓 University / Masters Essays
UK universities almost universally require Harvard for health science and medical education essays. Many have their own Harvard variation — check their style guide before starting. Consistency within the document is non-negotiable.
📰 Journal Articles & Case Reports
The BJGP, BMJ, and most UK medical journals use Vancouver. Always download the journal's "instructions for authors" before writing. Most journals have very specific rules about abbreviations, author limits, and URL formatting.
📁 ePortfolio & Log Entries
The 14Fish (FourteenFish) ePortfolio does not require formal references for most learning log entries. However, for significant entries — especially SEAs, QIPs, and reflective entries — citing a guideline or evidence base adds credibility and shows that your reflection is evidence-informed.
What Assessors and Supervisors Actually Look For
- Evidence that you searched for evidence — not just the first guideline you found
- Multiple sources cited, not just one (NICE is great, but one source alone is not a literature review)
- Relevant, recent sources — guidelines that are more than 5 years old should be questioned
- Correct format, used consistently throughout the document
- Sources that are actually available and real (check your references actually exist before submitting)
These insights reflect recurring patterns from UK GP trainees, deanery guidance discussions, and GP educator communities. All have been checked against RCGP and official guidance — only those that align with or extend official advice are included here.
The "one NICE guideline" trap — and how to escape it
One of the most common patterns in GP trainee QIP feedback is this: the trainee searches Google, finds a NICE guideline, cites it, and thinks the job is done. Supervisors consistently mark these entries as "acceptable" rather than "good." The trainees who score higher do a real literature search — they open PubMed, search for two or three relevant terms, and add one or two peer-reviewed papers alongside the NICE guideline. This takes around 20 minutes extra. It makes a measurable difference to the feedback.
Set up Zotero before your first QIP. Not after.
The most painful version of this problem is spending three hours adding references manually to a finished QIP. The second most painful version is trying to switch from Harvard to Vancouver at submission because you chose the wrong system. Zotero takes 15 minutes to set up and avoids both problems. Install the browser extension too — it captures a reference from PubMed with one click. Future you will be genuinely grateful.
Check the date on your guidelines
A common mistake is citing a NICE guideline that has been updated or superseded. NICE updates guidance regularly. Before you cite a guideline, check the publication date on the NICE website and look for a "surveillance review" notice, which tells you if it has been reviewed recently. Citing NICE CG30 (2005) when a 2023 update exists will flag as an issue in your QIP assessment. Always check the date.
Consistency beats perfection
A QIP with Harvard referencing that is perfectly consistent — even if the specific style is slightly off — reads far better than one where half the references are Harvard and half are Vancouver, and one is just a URL pasted into the text. Assessors understand that minor style variations exist. What they cannot overlook is chaos. Pick a system. Use it throughout. If you use Zotero, consistency is almost automatic.
The "I'll add the references later" lie
Almost every trainee who has ever said this has ended up desperately trying to remember where they read something, trawling through browser history at midnight, or simply omitting the reference entirely. The habit that saves the most time is capturing the reference the moment you read the paper. Open Zotero, click the browser extension button — done. It takes three seconds. It saves an hour of searching later.
For IMGs: UK referencing is probably different from what you were taught
In many countries, academic writing uses a different citation system — or referencing is not taught explicitly at all. In the UK, most university essays use Harvard, and most medical journals use Vancouver. Neither system is harder than the other once you understand the logic. But mixing the two up — or using a third system from another country — will generate unhelpful feedback. If you are unsure which to use, ask your educational supervisor before you start writing.
🎯 What Actually Gets You Higher Marks in QIP Referencing
Based on real assessor feedback patterns reported by UK GP trainee supervisors:
❌ What "acceptable" looks like:
- One NICE guideline cited, no other sources
- References formatted inconsistently throughout
- Outdated guidance cited without acknowledgement
- No in-text citations — just a reference list at the end
- Website URLs without access dates or author information
✅ What "above expectation" looks like:
- Three to five sources: NICE guideline + peer-reviewed papers + maybe a systematic review
- All references formatted consistently, using one system throughout
- In-text citations at each specific claim — not just at the end
- Sources are current (within 5 years, unless foundational)
- Evidence directly relevant to the QIP topic — not generic background reading
📊 The Evidence Hierarchy — Which Sources Carry the Most Weight?
When building your literature review, include sources from as high up this pyramid as you can find. The higher the level, the stronger the evidence.
Higher = stronger evidence. For a GP QIP, aiming for NICE guidelines + one or two higher-level studies is the gold standard.
Most trainees Googled "NICE hypertension" and called it a literature review. Here is what a real one looks like — and it does not take as long as you think.
🗺️ A Simple 6-Step Literature Search
Define your question clearly
Use the PICO framework: Patient or Problem, Intervention, Comparison, Outcome. For example: "In adults with hypertension (P), does lifestyle modification (I) compared to medication alone (C) reduce cardiovascular risk (O)?" This sounds formal, but it forces clarity. A vague question produces a vague search.
Choose your databases
For GP training writing, you need at least two sources:
PubMed is free worldwide. The Cochrane Library is free to anyone in England. Google Scholar is useful for finding papers but check the quality of what you find.
Search with good terms — and try more than one
Medical databases use MeSH terms (Medical Subject Headings) — these are standardised labels for medical concepts. PubMed translates many plain-English searches into MeSH automatically. But try a few different search terms:
"blood pressure management AND general practice"
"hypertension AND non-pharmacological treatment"
Use AND to narrow results. Use OR to broaden them (e.g. "GP OR general practice OR primary care").
Filter and select — do not read everything
For a GP QIP, you do not need a comprehensive systematic review. You need a few high-quality, relevant papers. Use PubMed's filters to limit to: last 5 years, English language, and optionally systematic reviews or RCTs. Read the abstract first. Most papers can be accepted or rejected from the abstract alone. You are aiming for three to five papers — not fifty.
Capture each reference immediately
The moment you find a useful paper, add it to Zotero (one click with the browser extension). If you are using a database that does not have the extension, copy the DOI and paste it into Zotero's "Add by identifier" box. Never rely on memory. References that are not captured when found are references that get lost.
Write, cite, and check
As you write, insert citations directly from Zotero into your Word document. When finished, generate the reference list automatically. Then do one final check: are all in-text citations in the list? Are all list entries cited in the text? Are any references more than five years old — and if so, is there a reason? Verify that every web URL still works.
🔵 PubMed in 60 Seconds — For Busy GP Trainees
To search PubMed:
- Go to pubmed.ncbi.nlm.nih.gov (free, no login needed)
- Type your topic in plain English — e.g. "atrial fibrillation primary care"
- Click Filters → Publication date → Last 5 years
- Click Article types → Systematic Review or RCT for stronger evidence
- Read abstracts first — full text available free on many articles
Useful PubMed tricks:
- Use quotation marks for exact phrases: "general practice" finds those exact words together
- Use the Clinical Queries feature for evidence-based practice filters
- Click Similar articles on a paper you like — finds related work quickly
- Save searches using My NCBI (free account) — useful for large projects
🔗 pubmed.ncbi.nlm.nih.gov — free worldwide access to over 40 million biomedical citations.
📚 Key Databases at a Glance — For UK GP Trainees
| Database | What It Contains | Cost | Best For |
|---|---|---|---|
| PubMed / MEDLINE | 40+ million biomedical citations; journal articles, reviews, trials | Free | Main search for any clinical topic |
| NICE / NICE CKS | UK clinical guidelines and primary care summaries | Free | UK-specific best practice; always include this |
| Cochrane Library | Systematic reviews and meta-analyses on healthcare interventions | Free in England | Highest-quality evidence for treatment questions |
| Google Scholar | Broad academic search across many disciplines | Free | Quick overview; finding grey literature and theses |
| BJGP / BJGP Open | UK general practice research; open access options available | Open access | GP-specific research, highly relevant to UK practice |
| Embase | Broader European biomedical database; useful for pharmacology | Paid (via NHS/Uni) | Systematic reviews; drug-related questions |
For most GP training tasks, PubMed + NICE + Cochrane covers everything you need — and all are free.
Referencing is not just about formatting. It is a habit of thinking — one that makes you a better, safer doctor.
🩺 In Daily GP Practice
You do not write formal references every day in the consultation room. But the thinking behind referencing — "where did this idea come from, and how solid is the evidence?" — is something the best GPs do constantly. They do it when a patient asks about a new treatment they read about online. They do it when a colleague recommends a different approach to a guideline. They do it at every clinical decision point.
Evidence-based thinking is referencing at speed, without the footnotes.
📁 In the 14Fish ePortfolio
You do not need formal references in every log entry. But adding a citation — even a simple one — to a significant event analysis, a learning entry, or a reflection on a difficult case shows that you looked it up, not just reflected on it in isolation. Even one well-placed (NICE, 2023) in a reflective entry signals that your reflection is evidence-informed, not just anecdotal. Supervisors notice this.
Evidence in your reflection = thinking, not just feeling.
What Your Referencing Style Tells a Supervisor About You
Multiple sources cited
"This trainee searched widely and thought critically about the evidence."
Consistent formatting
"This trainee is organised, careful, and attentive to detail."
Current sources only
"This trainee keeps up to date and does not rely on outdated guidance."
Your references are a window into how you think. Make sure they show what you want them to show.
The single most useful habit you can develop right now
Every time you look something up in clinical practice — a guideline, a drug dose, a management option — note it. If it influences a clinical decision, it is worth referencing in your reflection. If it changes the plan, write it down with the source. Over time, this habit transforms your portfolio from a list of things you did into a record of how you think.
Stop formatting references by hand. Seriously. These tools do it for you.
| Software | Cost | Best For | Works With | Notable Feature |
|---|---|---|---|---|
| Zotero | Free | Most people — especially beginners | Word, Google Docs, LibreOffice | Open source; one-click capture from web; best free option |
| Mendeley | Free (basic) | PDF-heavy workflows; collaborative use | Word, LibreOffice | Built-in PDF viewer; academic social network features |
| EndNote | Paid (~£100+) | Large research projects; PhD theses | Word (primarily) | Largest library of citation styles; most powerful for bulk work |
| RefWorks | University licence | Students whose university provides it | Word, Google Docs | Web-based; no install needed |
Our recommendation: Start with Zotero
Zotero is completely free, works on Mac and Windows, integrates with Word and Google Docs, and has one-click capture from PubMed, NICE, and most websites. It handles Harvard and Vancouver equally well. You can keep using it forever without paying anything. Install it before you start your next project.
🔄 Cite While You Write
Reference managers plug into Word or Google Docs. As you write, you insert citations from your library with one click. The reference list builds itself automatically.
🎨 Style Switching
Need to change from Harvard to Vancouver? One click. All your citations and reference list reformat automatically. This is the feature that makes grown researchers weep with gratitude.
⚠️ Always Double-Check
Reference managers are excellent but not perfect. Always review the final reference list before submitting. Minor formatting errors — especially punctuation and capitalisation — are common.
The mistakes that cause the most pain — and how to avoid them.
Using the wrong system for the context
Submitting a Vancouver-formatted QIP when your deanery expects Harvard, or vice versa. Always ask first. Always check the house style guide before you start.
Formatting inconsistently
Some references with italics, some without. Journal names abbreviated in some entries, written in full in others. Assessors notice this immediately. Use a reference manager and it is almost impossible to be inconsistent.
Citing only one source for a major claim
Especially for QIPs: citing just one NICE guideline and nothing else suggests a limited literature search. Multiple converging sources = stronger evidence = better assessment.
No access date for websites
Websites change and disappear. Without an access date, a web reference cannot be verified. Always include "Accessed: [day month year]" for any online source.
Trusting AI-generated references
AI tools frequently invent plausible-sounding but completely fabricated references. Every reference from any AI tool must be verified in PubMed or the original source before use.
Leaving referencing until the end
Adding references retrospectively is a nightmare. Set up your reference manager before you start writing. Capture each source as you read it. You will thank yourself later.
Insider Pearl — From Trainee Experience
The most common piece of feedback from supervisors on QIP entries is: "Good project, but the literature review only cites one guideline." Trainees often search NICE, find one relevant guideline, cite it, and think the job is done. A true literature review searches PubMed (or MEDLINE) using relevant terms, considers systematic reviews and primary studies, and cites at least three to five sources — not just the top Google result.
The essentials — distilled for fast recall.
🔤 Harvard in 5 seconds
- 📝 In-text: (Author, Year)
- 📋 Reference list: Alphabetical by surname
- ✏️ Journal names: Written in full, italicised
- 🔑 Key phrase: "Who said it and when"
- 🏫 Used in: UK universities, essays, dissertations
🔢 Vancouver in 5 seconds
- 📝 In-text: (1), [1], or superscript¹
- 📋 Reference list: Numbered in order of appearance
- ✏️ Journal names: Abbreviated, minimal punctuation
- 🔑 Key phrase: "Just the number"
- 🏥 Used in: BMJ, Lancet, BJGP, medical journals
🧩 Remember it this way
Or simply: Harvard = (Human, year) | Vancouver = (1)
The rules are evolving fast — but the underlying principles have not changed at all.
⛔ What NOT to do with AI
- Do not copy AI-generated text and submit it as your own without disclosure
- Do not use AI to generate reference lists — the references may not exist
- Do not rely on AI to interpret or summarise guidelines — it can be confidently wrong
- Do not use AI for reflective writing — reflection must come from you
- Do not assume AI-generated text is undetectable — detection tools are improving
✅ Responsible use of AI in writing
- Use AI to help structure ideas and outlines — then write in your own words
- Use AI for grammar and language polishing — where permitted
- Always verify any clinical information AI provides against NICE or BNF
- Follow your institution's AI disclosure policy — requirements vary
- If AI contributed to your work, cite it appropriately (policies vary; check with your institution)
Plagiarism means presenting someone else's work or ideas as your own — without acknowledgement. In medicine, this is not just an academic offence. It undermines the credibility of evidence, can mislead clinical practice, and is a professional conduct issue that can affect your registration.
Types of plagiarism to avoid:
- Copying text without quotation marks or citation
- Paraphrasing without citation
- Using another trainee's work (self-plagiarism counts too)
- Fabricating or falsifying references
- Submitting AI-generated text as your own (where prohibited)
Safe practices:
- Always cite the source when using someone else's idea — even if paraphrased
- Keep notes of where every idea came from as you research
- Use a plagiarism checker (Turnitin, Scribbr) before submitting
- When in doubt — cite it
Referencing errors are rarely catastrophic — but some carry serious consequences. Know which is which.
Minor formatting errors
Wrong punctuation, slightly wrong date format, journal name not abbreviated correctly. These are common and rarely penalised heavily — as long as the overall system is consistent. Fix before submitting, but do not panic.
Mixed or absent referencing
Using both systems in the same document, or submitting no references for a QIP. This leads to lower marks and specific written feedback from supervisors. In a university essay, it may affect your grade significantly.
Plagiarism and fabricated references
Submitting others' work as your own, or citing papers that do not exist (including AI-generated ones). These are serious academic and professional misconduct issues. In medicine, they can lead to GMC investigation and affect your registration. Never take this risk.
Common trainee difficulties with referencing
- Confusion between the two systems — Many trainees have never been formally taught referencing. They may mix elements of Harvard and Vancouver in the same document without realising.
- The "one source is enough" mindset — Trainees often cite a single NICE guideline and consider their literature review complete. Teach them what a real evidence search involves.
- Over-reliance on Google — Trainees search Google rather than PubMed, Cochrane, or NICE. Introduce them to proper search strategies and databases.
- Not using reference management software — Many trainees format references manually, which is error-prone and time-consuming. A short tutorial on Zotero can transform their practice.
💬 Discussion Prompts for Tutorials
- "What databases did you search for your QIP literature review — and why?"
- "If you had to justify this management choice with evidence, what would you cite?"
- "How do you know this guideline is the most current version?"
- "What is the difference between a reference list and a bibliography?"
- "How would you reference an algorithm you found on the NHS website?"
🛠️ Practical Tutorial Ideas
- Ask the trainee to bring their QIP draft — review the reference list together
- Give the trainee one NICE guideline and one BMJ article and ask them to reference both in Harvard and Vancouver
- Show a worked example of referencing the same source two different ways
- Introduce Zotero together in a 20-minute tutorial and install it on their laptop
- Spot-check: give a reference and ask "what is wrong with this formatting?"
The questions trainees actually ask.
No — but you should reference every specific factual claim, every guideline threshold or recommendation, and every statement about current best practice. Your own observations, data you collected, and your personal reflections do not need citing. As a rule: "could someone reasonably question where this came from?" → cite it.
Use the organisation name as the author (e.g. NHS England, NICE, RCGP). If no date is available, write "(no date)" in Harvard, or "n.d." in some styles. For the access date, use today. Important: if a webpage has no author, no date, and no clear organisational authority, consider whether it is reliable enough to cite at all.
Yes — NICE CKS (Clinical Knowledge Summaries) pages are a valid reference for general practice writing. Cite them as a website: NICE (year). Title of topic. Available at: [URL]. Accessed: [date]. However, for research-level work, primary NICE guidelines (e.g. NG136) carry more weight than CKS summaries, which are themselves derived from those guidelines.
The BNF is available in print and online. For the online version:
Vancouver: Joint Formulary Committee. British National Formulary [Internet]. London: BMJ Group and Pharmaceutical Press; 2026 [cited 2026 Apr 12]. Available from: https://bnf.nice.org.uk
A DOI (Digital Object Identifier) is a permanent identifier for a digital document — it does not break or change like URLs. For academic journal articles, always use the DOI instead of, or in addition to, a URL. DOIs look like: 10.1136/bmj.n1234. In your reference, write it as: https://doi.org/10.1136/bmj.n1234
🏁 Take-Home Points — The Bits to Remember Tomorrow
- Two systems, one rule: Harvard uses (Author, Year) in-text with an alphabetical list. Vancouver uses (1) in-text with a numbered list. Medical journals almost always use Vancouver. UK university essays usually use Harvard. Ask before you start writing.
- Ask for the house style guide first. Before you write a single word, find out which system your publisher, university, or deanery requires. Switching systems after writing 5,000 words is exactly as unpleasant as it sounds.
- Set up Zotero before you start. Zotero is free, integrates with Word and Google Docs, and switches between Harvard and Vancouver with one click. Install it now, before your next piece of writing.
- Never trust AI-generated references. AI tools routinely invent convincing-looking but completely fictional papers. Always verify every reference in PubMed or the original source before submitting.
- One NICE guideline is not a literature review. For QIPs, search PubMed, add one or two peer-reviewed papers alongside the guideline, and demonstrate genuine engagement with the evidence. This is what separates "acceptable" from "above expectation" in supervisor feedback.
- Consistency beats perfection. A document formatted consistently throughout — even if slightly imperfect — reads far better than one that mixes Harvard and Vancouver at random. Pick one system and stick with it.
- Capture references the moment you find them. The habit of adding a paper to Zotero immediately — not later — saves hours of retroactive hunting. Three seconds now versus forty-five minutes at midnight before submission.
- Check the date on your guidelines. NICE updates guidance regularly. A 2005 guideline cited in 2026 will attract feedback unless you can justify it. Always check the date on the NICE website before citing.
- When in doubt, cite it. Over-referencing is a minor stylistic issue. Under-referencing or plagiarism is a professional conduct issue. The decision is easy.
- Referencing is a habit of mind, not just a formatting rule. The best GPs ask "where did this idea come from, and how solid is the evidence?" at every clinical decision point. Referencing in writing is just that habit made visible on paper.