The most important fact about NHS data for a UK health dissertation is a rule, not a dataset: since September 2021 the Health Research Authority no longer accepts undergraduate applications for NHS ethics review, so standalone undergraduate research that needs NHS approval “cannot take place”. The published datasets below are how undergraduate and many master’s students in nursing, health sciences, public health and allied health professions do a real project anyway, because every one of them is already collected, already anonymised and already public.
This guide names the datasets, says what each holds, how you get it, what questions it answers and where it fails. It is the health-specific companion to our catalogue of UK data sources for a dissertation by subject, whose health section is two paragraphs long; this is that section, expanded.
The rule first: what the HRA position means for your design
The Health Research Authority’s student research page, last updated on 1 September 2021 and still the published position in August 2026, states that “standalone research at undergraduate level that requires ethics review and/or HRA and HCRW Approval (or devolved administration equivalent) cannot take place”, that undergraduate applications “are no longer being accepted for Research Ethics Committee (REC) review”, that master’s applicants must complete the student research toolkit to check eligibility, and that “arrangements for doctoral research remain unchanged”. The consequence for a taught dissertation is that primary research with NHS patients, NHS staff recruited through the NHS, or identifiable NHS records is off the table at undergraduate level and conditional at master’s level. Secondary analysis of published, anonymised data needs university ethics approval only, through the departmental route described in our guide to ethics approval for an undergraduate dissertation. That is why the datasets below matter.
The datasets
| Dataset | Custodian | What it holds | Access | Typical dissertation question |
|---|---|---|---|---|
| Fingertips public health profiles | Office for Health Improvement and Disparities (DHSC) | Indicators by local authority and ICB: smoking, obesity, mental health, inequalities, mortality | Free, downloadable, API | How does childhood obesity prevalence vary with deprivation across English local authorities? |
| GP Patient Survey | NHS England, run by Ipsos | Annual survey of patient experience by practice, PCN and ICB | Free, practice-level results published | Is access satisfaction lower in practices with larger list sizes? |
| NHS Staff Survey | NHS England | Annual survey of NHS staff by trust: burnout, bullying, intention to leave | Free, trust-level results and benchmark reports | Which trust characteristics predict nurse intention to leave? |
| Hospital Episode Statistics | NHS England | Admitted patient care, outpatient and A&E activity | Published aggregates free; record-level data requires a data access request | Have emergency readmissions within 30 days changed by region? |
| Health Survey for England | NHS England, with NatCen | Annual household survey with measured height, weight, blood pressure | Microdata via UK Data Service | What is the association between physical activity and self-rated health by age group? |
| Prescribing data | NHS Business Services Authority | Monthly prescriptions dispensed by practice and drug | Free statistical collections; OpenPrescribing for analysis | How has antibiotic prescribing varied across practices since 2019? |
| CQC inspection data | Care Quality Commission | Ratings and inspection reports for every registered provider | Free, downloadable | Do care home ratings relate to ownership type? |
| ONS health and social care statistics | Office for National Statistics | Mortality, life expectancy, avoidable deaths, health inequalities | Free | How has the life expectancy gap between the most and least deprived areas moved since 2015? |
| Understanding Society health modules | UK Data Service | Household panel with GHQ-12, self-rated health, long-term conditions | Free registration via university | Does financial strain predict later psychological distress? |
Fingertips: the source most health dissertations should start with
Fingertips describes itself as “a large public health data collection” organised “into themed profiles” that show “local area data alongside relevant comparators”; the site’s own citation line credits the Office for Health Improvement and Disparities, Public health profiles, with the year and the URL under Crown copyright, and the data are released under the Open Government Licence. For a dissertation it offers hundreds of indicators at local authority, region and integrated care board level, each with confidence intervals, deprivation deciles and trend markers, downloadable as CSV or through an API. Who it suits: any question of the form “does outcome X vary with characteristic Y across areas”, which covers most undergraduate public health and nursing projects. Where it fails: it is ecological data, so it cannot tell you anything about individuals, and a dissertation that reads an area-level correlation as an individual-level cause will lose marks for the ecological fallacy. Say the word in your limitations.
The two surveys nobody uses enough
The GP Patient Survey publishes results for every practice in England each year, and the NHS Staff Survey publishes results for every trust. Both are free, both go back more than a decade, and both can be joined to other published characteristics: practice list size and deprivation from Fingertips, trust size and CQC rating from the CQC data. A nursing dissertation asking which organisational factors are associated with staff burnout or intention to leave can be built entirely from the Staff Survey benchmark tables without a single participant being recruited, which is exactly what the HRA position pushes taught students toward. The limitation to state is response bias: the surveys report the views of those who answered, and response rates vary by trust and practice.
Prescribing, inspections and mortality
NHS Business Services Authority publishes monthly prescribing statistics by practice, which a student comfortable with a spreadsheet can use to compare, for example, opioid or antibiotic prescribing across practices or over a policy change, and which the OpenPrescribing tool from the University of Oxford makes explorable without downloading anything. CQC publishes ratings and inspection findings for every registered hospital, care home and GP practice, which supports governance and quality questions across the social care sector that no other source covers. ONS health statistics supply mortality, life expectancy and avoidable mortality by area and deprivation, and are the right source for a health inequalities dissertation that wants a trend over time.
Which dataset for which question: five worked matches
The table is easier to apply when the question comes first. Five questions taught health students actually propose, and the dataset each one needs.
- Is smoking in pregnancy more common in more deprived local authorities? Fingertips: the smoking status at time of delivery indicator by upper-tier local authority, read against the deprivation decile the site already attaches to each area. An ecological design; say so in the first line of the methods.
- Do nurses in trusts rated Inadequate or Requires Improvement report more burnout? NHS Staff Survey benchmark tables by trust for the burnout and intention-to-leave items, merged with CQC ratings by provider on the trust code. Two free downloads and one join.
- Has antibiotic prescribing per patient fallen since the 2019 antimicrobial resistance action plan? NHS Business Services Authority prescribing data by practice, monthly, expressed per registered patient; OpenPrescribing supplies the denominators so you do not build them by hand.
- Is physical activity associated with self-rated health in adults over 65? Health Survey for England microdata through the UK Data Service, at the individual level, which makes it the only design on this list that escapes the ecological fallacy altogether.
- Has the life expectancy gap between the most and least deprived areas widened since 2015? ONS health state life expectancy by deprivation decile for England, published as a repeated cross-section, so the trend is a table rather than a model.
Notice that four of the five never touch a patient record, and the fifth uses an anonymised survey a student registers for in a week. That is the shape of a taught health dissertation after September 2021: the question is the same one a clinical study would ask, and the data are the published version of the answer.
Record-level data: what a taught student can and cannot get
Hospital Episode Statistics at record level, and any identifiable NHS data, require a data access request and a lawful basis that a taught dissertation almost never has. Use the published aggregates instead; NHS England publishes HES summaries by provider, diagnosis and procedure, and they are enough for most descriptive and comparative questions. For individual-level analysis, the Health Survey for England and Understanding Society through the UK Data Service are the routes: both hold measured or self-reported health variables at the individual level, both are anonymised and both are available to students by registration, using the process in our guide to getting UK Data Service data. Note that NHS England, the custodian of HES and the Health Survey for England, absorbed the former NHS Digital in 2023, so older reading lists cite a name that no longer exists.
How to write the methods section for a secondary-data health dissertation
- Design. Name it: secondary analysis of published, anonymised data; cross-sectional or repeated cross-sectional; ecological where it is.
- Data source. Custodian, dataset, release date, date accessed, and the exact indicators or tables used, with their definitions copied from the metadata rather than paraphrased.
- Population and units. Local authorities, practices, trusts or individuals; the year or years; exclusions and the final count.
- Ethics. State that the data are published and anonymised, that no participants were recruited, and which university committee approved the project. Cite the HRA position if a reader might wonder why you did not go to the NHS.
- Quality appraisal. If the dissertation is a literature-based project rather than a data analysis, the appraisal tools are in our guide to the CASP checklist for a nursing dissertation and the search is built with the method in our guide to the nursing dissertation search strategy.
- Analysis. The test, the software, and the sensitivity check; the sample-size logic for area-level regressions is the same as in our guide to sample size for an undergraduate dissertation.
Where health students lose marks on data
On two sentences. The first names the custodian and the release; the second names the limitation the dataset carries. A dissertation that says “data were obtained from the NHS” has written neither. If the indicators are downloaded and the methods chapter is still blank, Tesify can draft the data and methods sections from your own source notes, your own indicator definitions and your own counts, in the order a health marker checks them. Every sentence is 100% written by you; what you get is a chapter that exists before the deadline.
Frequently asked questions
Can an undergraduate do research with NHS patients for a dissertation?
Not as a standalone project. Since September 2021 the Health Research Authority no longer accepts undergraduate applications for NHS ethics review, so primary research needing NHS approval cannot take place at that level. Secondary analysis of published data needs only university ethics approval.
What is the best free NHS dataset for a nursing dissertation?
The NHS Staff Survey for workforce questions and Fingertips for population health questions. Both are free, published every year and available at trust or local authority level without any application.
Do I need ethics approval to use published NHS data?
Not NHS ethics approval. You need your university’s departmental approval, and the methods chapter should state that the data are published and anonymised and that no participants were recruited.
Can I get Hospital Episode Statistics for my dissertation?
Published aggregates, yes, free from NHS England. Record-level HES requires a formal data access request and a lawful basis that a taught dissertation almost never has.
What is the ecological fallacy and why do markers mention it?
It is the error of reading an association between areas as an association between individuals. Fingertips and ONS area data can show that deprived areas have worse outcomes, not that deprived people do; the limitation must be stated.
Is NHS Digital still the right name to cite?
No. NHS Digital merged into NHS England in 2023; cite NHS England as the custodian of Hospital Episode Statistics and the Health Survey for England, and note the former name if your sources use it.
Can master’s students do primary NHS research?
Sometimes. The HRA asks master’s applicants to complete its student research toolkit to check eligibility, and some master’s applications are no longer accepted. Check the toolkit before designing a primary study.
How do I cite Fingertips?
Use the site’s own line: Office for Health Improvement and Disparities, Public health profiles, the year and the URL, then add the profile, the indicator and the date you downloaded it, in whatever referencing style your department requires.