Neurosurgery Dissertation Topics: Narrow, Researchable Titles for Medical and Health Students

Questions to settle before you choose a neurosurgery topic

Before you pick from these neurosurgery dissertation topics, settle six things:

• Can I answer my question with a literature review, public data or an audit, or does it need NHS ethics approval that my level does not allow?
• Which area do I mean: tumours, trauma, vascular disease, hydrocephalus, spine or functional surgery?
• Which classification or outcome scale will I use, such as the WHO CNS5 grade or the Glasgow Outcome Scale-Extended?
• Which real source can I build on: a NICE guideline, a trial or a registry?
• Is my question still open after RESCUE-ASDH, CRASH-3 or BASICS, or have those trials already answered it?
• Can I finish it within the data access and governance waits my timeline allows?

Neurosurgery is the surgical treatment of disorders of the brain, spine and peripheral nerves. The topics below cover tumours, trauma, haemorrhage, hydrocephalus, spine, functional surgery, services, AI and global neurosurgery. Questions about drug treatment, stroke medicine or basic brain science fit better under neurology dissertation topics or neurobiology research ideas.

Position correct as of October 2026. Several items rest on guidance that can still move: NICE published its vorasidenib appraisal in April 2026, and the trials of middle meningeal artery embolisation are still being pooled. Check NICE and the named registers for current status before you commit. All topics here are free to use. Ethics, referencing and marking rules differ by country, so check your own university’s requirements too.

Each title below names a guideline, trial, dataset or classification, so it reads as a near-ready proposal. If a title is still too wide, the dissertation topic finder can help you test it.

Glioma, brain metastases and neuro-oncology policy

1. To what extent do post-surgical grade 2 glioma pathways in one NHS unit record IDH status and a vorasidenib decision under NICE TA1147?
NICE TA1147 was published on 29 April 2026. It allows vorasidenib for grade 2 astrocytoma or oligodendroglioma with an IDH1 or IDH2 mutation in people aged 12 and over, and England had to fund it within 90 days. A Master’s service evaluation could check how often molecular status and the treatment decision are documented, with the supervisor inside the trust. Undergraduates can run a document analysis of the appraisal instead.

2. How did NICE’s October 2025 draft and March 2026 final draft guidance on vorasidenib treat uncertainty in the cost-effectiveness model, and what did patient evidence change?
The October draft recommended against routine use, while the final draft recommended it with interim Cancer Drugs Fund access. Brain tumour charities attribute part of the shift to revised cost-effectiveness thresholds from April 2026. Test that claim against the committee papers rather than repeating it. No patient data are needed.

3. How consistently do histopathology reports for adult diffuse glioma use the “not otherwise specified” label under WHO CNS5 when molecular testing fails or is unavailable?
The WHO CNS5 classification asks for layered reports with an integrated diagnosis and applies NOS when the required testing cannot be done. A report audit, run with a neuropathology supervisor, could count NOS use against the tests a laboratory actually offers. Related methods appear under pathology dissertation topics.

4. Does intraoperative smear or frozen-section diagnosis agree with the final CNS5 integrated diagnosis in adult diffuse glioma, and which tumour types account for the disagreements?
Report kappa for agreement, then list discordant cases by tumour type instead of pooling them. This needs a cohort with archived reports, so it suits a Master’s or PhD project with local approval.

5. Which definition of extent of resection do studies of 5-ALA fluorescence-guided glioblastoma surgery use, and how much does the definition change the reported benefit?
Studies differ in the residual volume threshold and in the timing of postoperative MRI. A systematic review that tabulates the definitions is realistic at undergraduate level, while a sensitivity analysis suits a Master’s thesis. The main pitfall is pooling studies that measure residual tumour in different ways.

6. Does adding a verb-production task to object naming during awake craniotomy language mapping change which postoperative language deficits are recorded?
Doctoral work at the University of Trento examined verb and noun tasks in awake surgery, so map what it found before asking what remains open. Below doctoral level, a narrative or systematic review is the realistic route. The anaesthetic side of awake surgery sits with anaesthesiology research topics.

7. Which of the 2015 James Lind Alliance neuro-oncology top ten uncertainties now have a registered randomised trial, and which have none?
The James Lind Alliance neuro-oncology priority setting used a modified Delphi and nominal group process. Cochrane’s neuro-oncology group later prioritised three systematic review titles from it. Search ISRCTN and ClinicalTrials.gov one question at a time. Only public registers are needed, which makes this a safe undergraduate design.

8. How do multidisciplinary team records document the choice between resection and biopsy for suspected glioma under NICE NG99, and is patient preference recorded?
NICE NG99 covers diagnosing, monitoring and managing primary brain tumours and brain metastases in people aged 16 or over. An audit against its recommendations is a clean design. However, it must be registered with the trust audit team, not the research office.

9. Is stereotactic radiosurgery to the surgical cavity after resection of brain metastases delivered as the NHS England commissioning policy sets out, now that NG99 points to that policy?
Since January 2021, NG99 recommendation 1.7.6 has been replaced by a pointer to the NHS England policy. Agree definitions of local control and radionecrosis with oncology colleagues before collecting any data.

10. Which patient-reported outcome measures do randomised trials of adult glioma surgery report, and how many share a common core set?
A Frontiers research collection on patient-reported outcomes in neuro-oncology shows continuing interest in the question. A systematic review with a COMET-style mapping of measures is feasible at undergraduate or Master’s level.

11. How often does one unit record the neurorehabilitation needs of people with a brain tumour at discharge, as NG99 addresses?
NG99 lists neurorehabilitation needs as a separate section. A small audit of discharge summaries is possible, and the rehabilitation angle connects to rehabilitation nursing dissertation topics.

12. Which prognostic factors recorded in US SEER research data for glioblastoma can be matched in English cancer registration data, and what is lost in the match?
CBTRUS draws on SEER and CDC registry data, and its 2018 to 2022 report appeared in Neuro-Oncology in October 2025. Both sources need an access application, so request only the variables you can justify and plan weeks for approval. This is a PhD or supervised Master’s design.

Meningioma, pituitary, skull base and posterior fossa surgery

13. How well do Simpson grade and WHO CNS grade together predict recurrence after meningioma surgery in a single unit with at least five years of MRI follow-up?
NG99 includes a separate evidence review on meningioma, so ground the question there. A time-to-event analysis fits, and the statistical test selector can confirm the method. Grading changed between WHO editions, so state which edition applies to each case.

14. What surveillance intervals do UK units use for small meningiomas, and how do they compare with the NG99 evidence review on meningioma follow-up?
A survey of unit protocols is enough for a Master’s project. Ask for documents, not patient data, to keep governance simple.

15. How could pituitary surgery outcomes be compared across units using the pituitary metrics in GIRFT’s neurosurgery data, and what can a student measure without patient-level records?
GIRFT neurosurgery data in the Model Health System cover more than 70 metrics, including pituitary surgery, subarachnoid haemorrhage, brain tumours and cranioplasty. The system needs an NHS login, so students outside the NHS should plan around published reports or ask their supervisor. Do not treat unit-level metrics as patient-level outcomes.

16. Which hearing class, facial nerve grade and tumour control outcomes do vestibular schwannoma studies report in common across microsurgery and radiosurgery, and are they comparable?
Frontiers in Surgery runs a research collection on optimising acoustic neuroma treatment protocols, which shows continuing interest. A PRISMA review is feasible, but follow-up length differs between studies and will limit pooling.

17. Which dural closure methods are associated with fewer cerebrospinal fluid leaks after posterior fossa surgery in retrospective cohorts such as a 2020 Kerala series, and what would a multicentre protocol need?
The Kerala cohort compared different closure methods. A protocol would need a shared definition of a leak, a fixed follow-up window and a plan for confounding by surgeon and tumour site. This is a good example of turning a single-centre finding into a testable question.

18. Can histopathological type (infarction or haemorrhage) and presenting imaging predict outcome in pituitary apoplexy, as a University of Cape Town doctoral thesis asked?
Pituitary apoplexy is uncommon and potentially fatal, so no single unit will have many cases. At undergraduate level, review what the thesis found and what later series add. Pooling several units is the realistic route for primary data, and it relates to endocrinology dissertation ideas.

19. How could the timing of cranioplasty after decompressive craniectomy be estimated from HES operation codes, and what are the limits of that coding?
The Neurosurgical National Audit Programme defines neurosurgical activity through OPCS-4 codes, and GIRFT reports cranioplasty metrics. Hospital Episode Statistics need an application to NHS England’s Data Access Request Service, so this is a PhD or a supervised Master’s with an existing data agreement.

Traumatic brain injury and neurocritical care

20. How consistent are pooled analyses of tranexamic acid in traumatic brain injury with CRASH-3’s finding that benefit appeared in mild-to-moderate injury but not severe injury?
The CRASH-3 trial randomised 12,737 patients. Among those treated within three hours, head-injury-related death was 18.5% with tranexamic acid and 19.8% with placebo. The subgroup estimates were 0.78 for mild-to-moderate injury and 0.99 for severe injury, with a heterogeneity p value of 0.030. Subgroup claims deserve cautious appraisal, which makes this a strong critical review for undergraduates.

21. How do emergency departments apply NG232’s recommendations on CT head scanning for adults taking anticoagulants, and how many return with delayed intracranial bleeding?
NICE NG232, published on 18 May 2023, replaced CG176. The earlier quality statement on CT for anticoagulated patients was removed because the recommendations changed. A service evaluation with a supervisor inside the trust is the route. Define “delayed” before you collect any data.

22. What information do referring hospitals record when requesting transfer to a neuroscience unit after head injury, and how does it compare with NG232’s section on transfer?
NG232 includes recommendations on transfer from hospital to a neuroscience unit. A document audit of referral forms is small enough for a Master’s project.

23. How consistently are NG232’s criteria for investigating cervical spine injury applied to intoxicated adults with head injury?
NG232 covers cervical spine investigation in people with head injury. Intoxication complicates assessment, so define it from the record rather than assuming it. Expect small numbers, which suits a pooled audit across sites.

24. How often do follow-up letters after moderate or severe head injury mention the early diagnosis of hypopituitarism that NG232 covers?
The guideline lists early diagnosis of hypopituitarism among its topics. Counting mentions in letters is simple, but you must decide in advance what counts as a mention.

25. How should the RESCUE-ASDH economic finding that craniotomy cost less and produced higher QALYs than craniectomy be weighed against the poorer pre-operative condition of the craniectomy group?
RESCUE-ASDH randomised 450 adults: 228 to craniotomy and 222 to craniectomy. Disability and quality of life at 12 months were similar, yet the UK economic analysis estimated lower costs and higher QALYs for craniotomy, even though craniectomy patients were in worse condition before surgery. Additional surgery within two weeks occurred in 14.6% of the craniotomy group and 6.9% of the craniectomy group, while wound complications occurred in 3.9% and 12.2%. A critical appraisal needs no patient data, and the health economics side connects to health economics dissertation topics.

26. How did the 2020 Brain Trauma Foundation update of its decompressive craniectomy recommendations integrate RESCUEicp and 12-month DECRA outcomes, and which earlier recommendation was restated?
The update produced three new level IIA recommendations and restated a fourth. Compare them with the Brain Trauma Foundation guidelines fourth edition, which was finished before RESCUEicp reported. This is a document analysis suitable for any level.

27. Do UK neurocritical care protocols treat intracranial pressure above the 22 mmHg threshold in the Brain Trauma Foundation’s fourth edition, or use other thresholds?
The fourth edition recommends treating intracranial pressure above 22 mmHg (level III) and targeting cerebral perfusion pressure of 60 to 70 mmHg (level IIB). A protocol survey is a Master’s design. Ask units for documents with permission, and keep any patient data out of scope.

28. Which collection methods for the Glasgow Outcome Scale-Extended do UK and European head injury studies use, and how comparable are their scores?
RESCUE-ASDH used the scale as its primary outcome. A methods review comparing postal, telephone and structured interview collection would help any student planning follow-up. It is also a safe design below doctoral level.

29. What does embedding RESCUE-ASDH within the CENTER-TBI observational cohort show about the strengths and limits of nesting a randomised trial inside a registry-style study?
CENTER-TBI is a European brain injury consortium, and RESCUE-ASDH was an embedded study within it. This is a methods critique for epidemiology or trial design projects, and it needs only published protocols.

30. Which neurosurgical procedures for head injury can be identified in the historic TARN dataset, which covers April 2007 to June 2023 for about 400,000 people, and how many cases can be matched by NHS number?
The NHS England catalogue describes a static collection with about 3.5 million injury codes, available for matching through pseudonymised data or NHS number. Access runs through a formal application, so treat this as a PhD design or a supervised project with an existing agreement.

31. How did the end of the Trauma Audit and Research Network in June 2023 and the March 2024 migration of historic data to the National Major Trauma Registry affect continuity for time-trend studies of head injury?
NHS England’s National Major Trauma Registry notes that new fields were added to its minimum dataset. A trend study must therefore check which variables exist on both sides of the changeover before it compares years.

32. How do UK trials recruit adults with severe head injury under the Mental Capacity Act 2005 consultee rules, and what recruitment delays do they report?
Sections 30 to 34 of the Mental Capacity Act 2005 cover research on adults who lack capacity, and the Act’s own notes give early measurements after serious head injury as an example of emergency research. The Act applies in England and Wales, whereas Scotland and Northern Ireland have their own laws. Consent questions of this kind also appear in medical law dissertation ideas.

33. Does weekend admission change the time to craniotomy for traumatic acute subdural haematoma in national hospital data?
First check whether the data release carries the time of admission and operation, because day-level records cannot answer an hours-level question. If they cannot, restate the outcome as a number of days.

34. What does brain tissue oxygen monitoring add to intracranial pressure monitoring in severe paediatric head injury, as the Cape Town multimodality work examined?
A University of Cape Town doctoral thesis studied the contribution of brain oxygen, transcranial Doppler and autoregulation monitoring in children. A review of later studies would show what has since changed. The paediatric critical care angle fits paediatric intensive care nursing topics.

Chronic subdural haematoma, ruptured aneurysms and intracranial haemorrhage

35. Why did EMBOLISE and STEM report fewer treatment failures with middle meningeal artery embolisation while MAGIC-MT (p = 0.10) and EMPROTECT (p = 0.13) did not reach significance, and which design differences could explain the split?
All patients in EMBOLISE received embolisation as an adjunct to surgery. In MAGIC-MT and STEM, roughly a fifth and two fifths of patients received it as primary treatment. Embolic agents and outcome definitions also differed. Position correct as of October 2026, because further trials are registered.

36. Does adding EMPROTECT to the published meta-analyses of three or four trials (1,432 and 1,468 patients) change the pooled estimate of treatment failure?
A 2025 meta-analysis of three trials was registered on PROSPERO as CRD42024618816, and a Canadian review pooled four trials without EMPROTECT. Register your own protocol, state the model, and report heterogeneity.

37. How do primary outcomes, embolic agents and inclusion criteria differ across the 21 registered embolisation trials counted in a 2025 review?
The review found that these design features differed significantly between trials. A registry audit of ISRCTN and ClinicalTrials.gov entries is a good undergraduate design because it needs only public records.

38. How far do volumetric CT endpoints reduce the observer variability that the EMBOLISE imaging analysis raised for clinical failure endpoints in chronic subdural haematoma?
The 2026 Radiology analysis examined haematoma volume, thickness and midline shift in 352 patients, and noted that clinical endpoints may vary between raters. A reliability study would report intraclass correlation for repeated volume measurements. It suits a Master’s or PhD with access to anonymised scans.

39. Where do NICE interventional procedures guidance and NHS England commissioning stand on embolisation for chronic subdural haematoma compared with the 2025 SVIN consensus statement?
The consensus statement concluded that embolisation plus standard treatment may become standard of care in selected patients. No NICE position was verified while this resource was prepared, so check the current status before you write.

40. Does rising antithrombotic prescribing help explain admissions for chronic subdural haematoma in English hospital data, and how should age be handled as a confounder?
A 2026 review notes that incidence is rising with population ageing and antithrombotic use. An interrupted time series or Poisson model could separate the two, using an application to the data service for hospital records.

41. What does the quoted incidence of 20.6 per 100,000 person-years for chronic subdural haematoma rest on, and how does it compare with recent population-based estimates?
The figure is repeated in recent embolisation reviews. Tracing it to its original source and checking the population, years and case definition teaches source appraisal, and it needs no patient data.

42. How does time from admission to securing a ruptured aneurysm differ between weekday and weekend admissions in a UK unit, set against the 48-hour standard and the 2013 NCEPOD findings?
NCEPOD’s 2013 report analysed 427 cases and found that 30% of weekend admissions received care within 24 hours, compared with 70% on weekdays. A Cambridge audit of 2015 to 2016 reported medians of 27.9 hours for weekday and 41.35 hours for weekend admissions. See NCEPOD for the original report, and use NICE NG228 for current practice. Define the ictus time and treatment time exactly.

43. Which national data sources could rebuild NCEPOD’s time-to-treatment analysis today, and what do HES operation codes lose compared with case-note review?
This is a methods project that compares routine coding with manual review for a sample of cases. It needs approval, but the question is specific enough to be answerable in one academic year.

44. How do US inpatient data showing 75.2% early treatment of ruptured aneurysms in 2005 to 2008 compare with recent English proportions treated within 48 hours?
The US analysis covered 32,048 patients and found early treatment more likely with endovascular treatment and weekday admission. Coding and definitions differ between countries, so state them before comparing.

45. What follow-up imaging do units arrange for non-culprit aneurysms after aneurysmal subarachnoid haemorrhage, given that NG228 included a dedicated evidence review on them?
NG228 was published on 23 November 2022. A survey of unit practice is a Master’s-level design, while a narrative review of the evidence review’s findings suits undergraduates.

46. Which monitoring methods for raised intracranial pressure and delayed ischaemia do UK units use after aneurysmal subarachnoid haemorrhage, and how do they map to NG228’s evidence review on monitoring?
Ask units for protocol documents and map them to the guideline’s headings. Keep the sample small and the comparison structured.

47. Which flow-diverting and woven endosaccular device categories described in NG228’s evidence review on preventing rebleeding rest on observational evidence only?
The review describes tubular mesh flow diverters and woven endosaccular devices alongside clipping and coiling. Tabulating study designs by device is a good undergraduate literature project.

48. How do neurovascular teams document the choice between coiling and clipping after subarachnoid haemorrhage, and how often is the reason recorded?
A record audit would need a fixed list of acceptable reasons, such as aneurysm location or neck shape, agreed with a supervisor before data collection.

Hydrocephalus, shunts and paediatric neurosurgery

49. What do the BASICS infection rates (6.0% with standard shunts, 5.9% with silver and 2.2% with antibiotic-impregnated shunts) imply for the number needed to treat when baseline infection risk differs by age?
The BASICS trial is a good place to practise absolute risk reduction. On these figures the difference is 3.8 percentage points, or roughly 26 shunts per infection avoided. A worked calculation with confidence intervals suits undergraduates.

50. How complete is UK Shunt Registry coverage of first shunt insertions when compared with HES counts of shunt operations?
The UK Shunt Registry is held within the Society of British Neurological Surgeons audit programme. A completeness study needs both datasets and careful operation code selection.

51. How should an interrupted time series separate the effect of BASICS reporting from other changes in shunt surgery when antibiotic shunt use rose from 36.9% in children and 20.5% in adults in 2004 to 99.2% and 96.8% in 2021?
A 2026 registry analysis reported the largest rise between 2018 and 2019, the year BASICS was reported. A segmented regression with alternative break points is the method, and you should list co-occurring changes such as theatre practice.

52. How sensitive is the estimated £1,004,572 annual NHS saving from antibiotic shunts to the assumed cost of a revision operation?
The same registry analysis produced the estimate. A one-way and probabilistic sensitivity analysis is a clear Master’s exercise, particularly alongside economics dissertation ideas on health costing.

53. How well does the endoscopic third ventriculostomy success score predict shunt-free survival in a recent UK paediatric cohort, and does it work as well for adults?
The score is built from age, cause of hydrocephalus and previous shunt. A validation study reports discrimination and calibration, and it needs patient-level data from a neurosurgical unit.

54. Does the published evidence support delayed ventriculoperitoneal shunting in post-haemorrhagic hydrocephalus of prematurity, as a 1999 Cape Town master’s study proposed?
That study was prompted by an impression that early surgery brought more shunt-related problems. A systematic review would test the impression against later evidence.

55. Which tap-test and gait thresholds do services use before offering a shunt for idiopathic normal pressure hydrocephalus, and how do they differ from international consensus guidance?
A Masaryk University doctoral topic on selecting patients with normal pressure hydrocephalus for shunting shows the question has been studied for years. Do not mix normal pressure hydrocephalus with other types in your sample.

56. How do GIRFT’s paediatric neurosurgery metrics on shunts, brain tumours and endoscopic third ventriculostomy differ by operational delivery network?
GIRFT’s paediatric neurosurgery data are organised by network and unit and drawn from deep dive visits. Access needs an NHS login, so check what your supervisor can share. Related child health themes appear in child health nursing research topics.

57. Does molecular profiling change risk stratification of medulloblastoma in resource-limited settings compared with histology alone, as a Cape Town paediatric cohort explored?
The cohort also covered juvenile pilocytic astrocytoma. At undergraduate level, review what profiling adds and what it costs. Primary data would require a laboratory partner and approval.

58. How should long-term outcome be defined after paediatric craniopharyngioma treatment when hormone dependence, independence and schooling all matter?
A Cape Town master’s series combined those three elements into one functional outcome. Compare definitions across published series before proposing your own.

59. Can early intracranial pressure monitoring predict the outcome of endoscopic third ventriculostomy, as a University of Cape Town master’s study tested?
Replication would need a prospective cohort with a fixed monitoring protocol and a defined failure time. A review of later work shows what is already known.

Spine surgery: degenerative disease, cauda equina syndrome and metastatic compression

60. Which of the AO Spine RECODE-DCM top ten research priorities, agreed in November 2019, now have a registered trial?
RECODE-DCM gathered 3,404 research ideas from 429 people before agreeing its priorities. Search ISRCTN and ClinicalTrials.gov using the older terms for the condition as well as the agreed name.

61. Which items of the RECODE-DCM minimum dataset can be mapped to British Spine Registry fields?
The British Spine Registry was established in May 2012 by the British Association of Spine Surgeons. A mapping exercise needs only form documentation, but check what documentation the association shares.

62. How did adopting the name degenerative cervical myelopathy in October 2020 change the way literature searches should be built for older terms such as cervical spondylotic myelopathy?
Test a search strategy with and without the older terms and report the records you would have missed. This is a clean methods project at any level.

63. How well do British Spine Registry records match HES-recorded spinal procedures, and what does a Queen Mary matching study suggest about completeness?
That study joins the registry to Hospital Episode Statistics, Office for National Statistics and patient-reported outcome data. A student version can review its published methods and design a smaller comparison.

64. Did the spinal best practice tariff change the completeness of British Spine Registry entries and patient consent beyond the single trust that a Bone and Joint Open study examined?
The study looked at how tariff incentives affected compliance in one trust. A multi-site comparison would need permissions from each site, so a Master’s student may be better placed to design it than to run it.

65. Which patient-reported outcomes does the British Spine Registry collect before and after surgery for lumbar spinal stenosis and degenerative spondylolisthesis, and what response rates limit analysis?
Both conditions appear among the registry’s research areas. Describe the type of data you would request, and plan for missing follow-up forms.

66. How closely can the 52,523 NHS spinal procedures reported by GIRFT for 2017 to 2018 be reproduced from published HES operation tables?
GIRFT’s cranial neurosurgery and spinal surgery reports set out the counts. Reproducing them teaches how coding groups shape totals, and published tables avoid a patient-level application.

67. What proportion of MRI scans for suspected cauda equina syndrome meet the GIRFT standard of four hours from request, and what predicts delay?
The GIRFT national suspected cauda equina pathway asks for emergency MRI as soon as possible and certainly within four hours of request. One London district hospital quality improvement project found 89.19% of scans within 24 hours, so choose your time clock carefully: presentation to scan or request to scan.

68. Which staffing and scanner constraints do district general hospitals cite for lacking 24-hour MRI for suspected cauda equina syndrome, as published audits describe them?
A pathway published in February 2023 expected 24-hour MRI provision by June 2024, or a documented out-of-hours alternative. A qualitative synthesis of published audits is feasible without patient data.

69. Does using bladder scans as a rule-in test in the 2025 GIRFT interactive pathway change MRI request rates in one emergency department?
A 2025 narrative review describes bladder scanning as rule-in rather than rule-out in the updated pathway. Before and after audit periods must be matched for staffing and season.

70. How often is the assessment of spinal stability recorded at first contact for suspected metastatic spinal cord compression, as NG234 expects?
NICE NG234 was published on 6 September 2023 and includes tools for assessing spinal stability and prognosis. A record audit is a good Master’s design with a supervisor in an oncology or spinal service.

71. What changed between CG75 in 2008 and NG234 in 2023 in how metastatic spinal cord compression is recognised, and which changes could be measured in referral times?
NG234 replaced CG75, and a March 2026 review simplified it without changing practice. Compare the two documents recommendation by recommendation, then pick one that is measurable.

72. Which survival scoring systems did NG234’s evidence reviews compare for people with spinal cord compression, and which have external validation outside the UK?
A literature review that tabulates each system and its validation cohorts fits undergraduate level well.

73. Which radiological measures define fusion and stability after craniovertebral junction fixation in single-centre series such as the 2023 Thrissur report?
A review of definitions would show whether results can be compared. The craniocervical junction also has its own Frontiers research collection, which suggests continuing interest.

74. Which risk factors for recurrent lumbar disc herniation recur across single-centre studies, and can a small unit’s records support a replication?
A 2023 Kerala study reported risk factors, which gives you a starting list. Decide the recurrence definition and minimum follow-up before extracting data.

75. How should minimally invasive transpedicular screw fixation for thoracolumbar fractures be compared with open fixation in a basic neurosurgery unit, using a 2021 retrospective analysis as the template?
Define fracture type, neurological status and the outcomes you will compare. Without randomisation, report how surgeons chose the technique.

76. Which published series on minimally invasive excision of thoracolumbar intradural extramedullary tumours may include overlapping patients from one department, and how should a review handle them?
Several papers from one Kerala department report open and minimally invasive excision. Overlap is a real risk in systematic reviews, so contact authors or exclude duplicates transparently.

Functional neurosurgery, epilepsy and pain

77. What outcome data would the research recommendations in NICE’s guidance on focused ultrasound thalamotomy require, and which have since been reported?
NICE’s guidance on focused ultrasound thalamotomy for treatment-resistant essential tremor, first published in June 2018, asks for work on patient selection, functional improvement, quality of life and long-term follow-up. Map each requirement to later studies in a systematic review.

78. Does staged bilateral focused ultrasound thalamotomy add benefit or risk compared with unilateral treatment, as NICE’s scope on staged procedures asks?
NICE has published a final scope covering unilateral and staged bilateral treatment for medication-refractory essential tremor. Verify whether guidance has followed the scope before you write, because that decides what evidence you can cite.

79. How do connectivity-based targets for deep brain stimulation in Parkinson’s disease and essential tremor compare with atlas-based targets for predicting outcome?
A University College London doctoral thesis built predictive models of stimulation outcome from MRI connectivity, and a Szeged department lists related work. This is a PhD-level method comparison, and a review is the realistic alternative.

80. What outcome measures do adaptive deep brain stimulation studies for Parkinson’s disease report, and are they comparable with conventional stimulation trials?
The European Association of Neurosurgical Societies journal Brain and Spine has invited papers on adaptive neuromodulation, which shows the area is active. Tabulate outcomes before judging comparability.

81. How long do adults with drug-resistant focal epilepsy wait between failing anti-seizure treatment and referral for surgical evaluation in one region?
Define “failing treatment” from clinic records before starting. Small sample sizes are common, so a pooled audit across clinics is more practical than a single unit.

82. How do pain freedom and facial numbness compare between microvascular decompression and stereotactic radiosurgery for trigeminal neuralgia in published series, and which pain scale do they use?
Studies often use the Barrow Neurological Institute pain scale, though not always. A review can compare scales and follow-up length.

83. How do clinics define a successful trial of spinal cord stimulation compared with the definition in NICE TA159, and does the definition affect who receives an implant?
NICE TA159 was published in October 2008, and a 2014 review found nothing new that affected its recommendations. A comparison of clinic protocols with the appraisal is a Master’s design, and the pain management view connects to pain management nursing topics.

84. Which functional outcomes beyond pain scores do spinal cord stimulation studies for failed back surgery syndrome report?
TA159 names pain relief, functional outcomes and quality of life among the benefits to consider. A review can count how many studies report each.

85. Which neuropsychological measures were used in studies of neurosurgery for mental disorder, building on a 2003 Edinburgh thesis, and how do later deep brain stimulation studies for obsessive-compulsive disorder measure cognition?
The thesis examined outcomes after ablative surgery for severe, intractable illness. Handle the ethics of this population carefully, and prefer review designs.

86. What does a 2011 Oxford doctoral thesis on deep brain stimulation and cardiovascular and respiratory indices imply for monitoring autonomic effects in current stimulation trials?
The thesis evaluated stimulation effects on the motor, cardiovascular and respiratory systems. A review could ask whether current trials report any autonomic outcomes at all.

87. Which analgesic protocols are compared in post-craniotomy pain studies, and do they report opioid consumption in a common unit such as oral morphine equivalents?
Report whether studies convert doses or only list drugs. The answer shows whether pooling is possible.

Technology, AI and robotics in neurosurgery

88. How many neurosurgical outcome-prediction studies published since April 2024 report the 27 items of TRIPOD+AI, and which items are most often missing?
TRIPOD+AI was published in the BMJ in April 2024 and supersedes the 2015 checklist. A reporting audit scores each paper against the checklist. Two scorers and a reliability check make it credible.

89. Do studies of large language models answering neurosurgical board-style questions follow the TRIPOD-LLM reporting guideline?
TRIPOD-LLM is listed alongside TRIPOD+AI on the EQUATOR network. A meta-research review is feasible at undergraduate level, and the question will probably still be open.

90. How often do machine learning models that predict outcomes after neurosurgery use an external validation cohort, and does it change reported accuracy?
A Frontiers research collection on machine learning in predicting neurosurgical outcomes shows continuing activity. Extract the validation type for each model and compare reported discrimination.

91. Which grading systems for pedicle screw accuracy do robotic spine studies use, and how does the choice change the reported accuracy?
The Gertzbein-Robbins grading is common, but definitions of an acceptable screw differ. A review can recalculate accuracy under one common definition where papers allow it.

92. Which outcomes do comparisons of the exoscope and the operating microscope report in cranial and spinal surgery?
List the outcomes reported, such as operative time and surgeon ergonomics, and note which studies report patient outcomes at all.

93. How is neurosurgical simulation performance validated against operating room performance, and which validity types do studies report?
A review organised by validity type (face, content, construct, predictive) is realistic at Master’s level. Many studies stop at construct validity.

94. How accurate is automated MRI segmentation of vestibular schwannoma volume against manual segmentation, and which overlap metric do studies report?
Many segmentation studies report the Dice coefficient. A student can use a public annotated dataset if one is available, which avoids ethics entirely.

95. Which intraoperative imaging methods (MRI, ultrasound, fluorescence) do glioma studies compare, and which report survival rather than extent of resection alone?
A Frontiers research collection on intraoperative imaging in neuro-oncology suggests the question is current. Tabulate endpoints, because extent of resection is only a surrogate.

Services, audit, safety and access

96. What can the 2024 annual report of the Neurosurgical National Audit Programme tell a student about case-mix adjustment in unit-level 30-day mortality, and what does it not show?
The Neurosurgical National Audit Programme was set up in 2013 by the Society of British Neurological Surgeons and uses Hospital Episode Statistics. The 2024 report was its first annual report. A critical reading needs no data access.

97. How sensitive are NNAP activity counts to the OPCS-4 codes chosen to define neurosurgical procedures?
The programme publishes a coding spreadsheet. Re-run a count with a wider or narrower code list and report the shift.

98. How does the SBNS outlier policy handle risk-adjusted mortality, and which statistical methods suit small neurosurgical units?
A simulation using funnel plots can show how often small units are flagged by chance. This is a good statistics-led Master’s project.

99. How do referral-to-treatment waiting times for neurosurgery compare with neighbouring specialties, and what does the 18-week standard exclude?
Neurosurgery has its own treatment function code (150) in NHS England’s waiting times statistics. Those pathways cover non-emergency consultant-led treatment, so emergency neurosurgery sits outside them. Published tables make this a good public-data project.

100. How do Scotland’s Annual Review of Consultant Outcomes and Clinical Audit of Paediatric Neurosurgical Activity differ from NNAP in outcome definition and coverage?
The Scottish Managed Service Network for neurosurgery describes both audits in its 2023 to 2024 annual report. A devolved comparison is rare and could be examined in a policy dissertation.

101. What educational value do neurosurgical student-selected components provide after the HRA’s 2021 change to student research, building on a 2022 survey of 24 respondents?
In that survey, 62% said they had published or presented work. Compare such outcomes with the alternatives the HRA suggests, including group projects and mock ethics committees.

102. How should a time-to-MRI project on suspected cauda equina syndrome be classified using the HRA decision tool, and what changes if the aim moves from checking a standard to testing a hypothesis?
Measuring performance against a four-hour standard is audit. Testing a new explanation for delay is research. Save the decision tool output with your project records.

103. How do research ethics committees apply flagged-committee rules when a neurosurgical study includes adults who lack capacity?
Intrusive research on adults lacking capacity in England and Wales needs review by a flagged NHS committee. A policy review can compare how universities and trusts explain that requirement.

104. Which of the 15 recommendations in GIRFT’s 2018 cranial neurosurgery report have measurable follow-up metrics in the Model Health System?
The report made 15 recommendations on pathways, length of stay and cancellations, and the system now holds more than 70 metrics. Matching them is a clear document project.

Global neurosurgery

105. How were the more than 5 million unmet essential neurosurgical cases and 23,300 additional neurosurgeons estimated, and how sensitive are those figures to assumptions about incidence and surgical need?
The estimates come from a Global Neurosurgery Initiative summary of essential neurosurgical capacity, published in the Journal of Neurosurgery. Rebuild the logic from its methods and test alternative inputs.

106. Which of the five essential neurosurgical conditions (head injury, stroke-related conditions, tumours, hydrocephalus and epilepsy) has the least published capacity data in sub-Saharan Africa?
A scoping review could map the literature by condition and country. The scale of the gap is not established, so frame it as a question and report what the search shows.

107. How do outcomes of decompressive craniectomy in a South African series of 76 patients treated between 2005 and 2011 compare with those in randomised trials such as RESCUEicp?
The Groote Schuur study offers a low-resource benchmark. Compare case mix first, because a survival difference may reflect who was treated.

108. How have national surgical plans referred to neurosurgery since World Health Assembly resolution 68.15 in 2015?
The resolution on emergency and essential surgical care followed the Lancet Commission on Global Surgery. A document analysis of published plans from several countries is a safe design for any level.

Choosing a neurosurgery topic that your level allows

Level matters more in neurosurgery than in most subjects, because the Health Research Authority limits what students can do with NHS patients. Read its student research rules before you fix a design.

Undergraduate neurosurgery research topics

Standalone undergraduate research that needs NHS ethics review or HRA approval cannot take place. For undergraduate and Master’s courses, the HRA says the supervisor should take the role of chief investigator. Safe routes therefore include systematic or narrative reviews, critical appraisals of trials, policy analysis, public registers and published statistics. The topics numbered 2, 5, 7, 20, 25, 37, 49 and 60 are examples of designs that need no NHS patient data.

Master’s neurosurgery dissertation topics

Master’s students should complete the HRA student research toolkit first, because only some projects are accepted for ethics review. Audits and service evaluations are common alternatives, as long as the trust registers them. If you want a second reader for a medical methods chapter, medical dissertation help can support planning and feedback. Our guide on getting ethics approval for dissertation research explains the general process.

PhD neurosurgery research proposal ideas

Doctoral applications are accepted for NHS ethics review and HRA approval. Doctoral designs can therefore use patient data, imaging cohorts and registry applications, as in topics 12, 19, 30 and 38. Plan data access first, because approval timelines often decide the project rather than the idea.

Ethics, capacity and governance for neurosurgical projects

Start by deciding whether your project is research, audit or service evaluation. The HRA decision tool walks through this, and audits and service evaluations generally fall outside NHS ethics review while still needing local sign-off. The flow below shows the three common routes.

Neurosurgical patients often lack capacity at the time of enrolment, for example after head injury or haemorrhage. In England and Wales, the Mental Capacity Act 2005 allows approved research to include them through a personal or nominated consultee, and intrusive research needs a flagged NHS ethics committee. Scotland and Northern Ireland use different laws.

Children need parental consent and, where appropriate, the child’s assent. Photographs and imaging carry identifiers, so state how you de-identify them. If you are unsure, ask your supervisor and your own university’s ethics office before collecting anything.

Design choices that suit neurosurgical questions

Match your design to the question. Systematic reviews suit questions about conflicting trials, and the guide to writing a literature review covers the structure. Registry and hospital data suit questions about practice variation, while audits suit questions about whether a unit meets a standard.

Prediction models need TRIPOD+AI reporting, and qualitative projects on clinician or family experience need a defined analysis method, such as the one in the guide to thematic analysis. For quantitative work, the guide to choosing a statistical test and the sample size calculator save time.

Four worked examples with aims and objectives

Example 1: why the embolisation trials disagreed (topic 35)

Research aim: To explain why randomised trials of middle meningeal artery embolisation for chronic subdural haematoma reached different conclusions on treatment failure.

• Extract design features from EMBOLISE, STEM, MAGIC-MT and EMPROTECT, including adjunct versus primary use, embolic agent and outcome definition.
• Compare pooled estimates with and without trials that mixed primary and adjunct use.
• Judge risk of bias in each trial with a named tool and report disagreements between two reviewers.

Example 2: weekend delay in aneurysm treatment (topic 42)

Research aim: To measure time from admission to securing a ruptured aneurysm in one unit and compare weekday with weekend admissions against the 48-hour standard.

• Identify consecutive confirmed aneurysmal subarachnoid haemorrhage cases from the unit database.
• Calculate median hours to securing the aneurysm, with interquartile range, by admission day.
• Record documented reasons for treatment after 48 hours and group them.

Example 3: the effect of BASICS on shunt practice (topic 51)

Research aim: To estimate how far the publication of BASICS explains the rise in antibiotic shunt use in the UK.

• Obtain annual counts of antibiotic and standard shunts from registry reports or a data request.
• Fit a segmented regression with a break around 2018 to 2019.
• Test alternative break points and list other changes that might have co-occurred.

Example 4: mapping cervical myelopathy priorities to trials (topic 60)

Research aim: To identify which RECODE-DCM top ten priorities have a registered trial.

• Build a search strategy using the agreed name and older synonyms.
• Classify each registered trial against the priority it addresses, with a second reviewer for disagreements.
• Report which priorities have no registered trial, and describe the limits of register searching.

Neurosurgery dissertation questions answered

Can an undergraduate use NHS patient records for a neurosurgery dissertation?

Not through a standalone research project that needs NHS ethics review. Since 1 September 2021, the HRA does not accept those applications. Instead, choose a review, a trial appraisal, a policy analysis or public data. Some universities also run supervised audits that the trust registers, but an audit is not research and needs local governance.

Is a time-to-scan project on suspected cauda equina syndrome research or audit?

It is audit if you measure practice against an existing standard, such as the four-hour target. It becomes research if you test a new hypothesis, for example whether a patient factor predicts delay. Use the HRA decision tool and keep the output.

What if my participants lack capacity to consent?

In England and Wales, sections 30 to 34 of the Mental Capacity Act 2005 allow approved research with a consultee. Scotland uses the Adults with Incapacity (Scotland) Act 2000, and Northern Ireland uses its own 2016 Act. Never plan this route without ethics approval.

How narrow should a neurosurgery topic be?

Narrow enough to name a population, an intervention or exposure, and an outcome. For example, “chronic subdural haematoma” becomes “middle meningeal artery embolisation” and then “why randomised trials disagreed on treatment failure”. The guide on writing a research question shows each step.

From topic to proposal

Need help narrowing one of these into a question? The research question generator and the guide to aims, objectives and research questions can turn it into something a supervisor can approve. For a structured outline, research proposal support offers feedback and planning. Any material you receive is for reference and learning, and you should use it according to your institution’s policy.