**Background:** Research in geriatric medicine is expanding but remains fragmented, with small single-centre studies and a lack of standardised assessment tools. This heterogeneity prevents pooling of data and cross-study comparisons. Previous efforts, such as the Gerontonet minimum dataset for European clinical trials, saw limited UK uptake due to concerns about burden, outdated items (e.g., Charlson index, copyright-protected MMSE), and omission of newer tools like the Clinical Frailty Scale (CFS). The COMET initiative demonstrated the value of standardising outcome measurement, but the authors argue that standardising data collection tools themselves is equally important for improving research quality, protocol efficiency, and collaboration.
**Methods:** A modified Delphi process was conducted in six steps. Step 1: An exploratory (non-systematic) literature review identified common domains and measures from UK prospective geriatric medicine studies published in high-impact journals (1 Jan 2014 – 31 Dec 2018). Results were presented at the NIHR Newcastle BRC academic geriatric medicine event (May 2019, 29 delegates). Step 2: A broad questionnaire (18 short-answer questions) was distributed to delegates and additional expert academic geriatricians; 19 responses were obtained. Five additional items were suggested. Step 3: A focused questionnaire (26 items, 108 within-category options) was sent to all previous invitees; 21 responses were obtained. Participants indicated whether each aspect was “core” and selected preferred measurement tools. Step 4: Items considered core by >80% were proposed for agreement; items deemed not relevant by >75% were removed. Where two measures were in contention, participants selected one; where more than two, they ranked them. This questionnaire was distributed more widely (UK Association of Academic Geriatric Medicine, British Geriatrics Society Research and Academic Development Committee, national mailing lists); 58 responses were obtained. Step 5: A virtual consensus meeting (13 attendees, 5 Nov 2020) discussed results and made final decisions. Step 6: A separate survey on long-term conditions (18 responses) asked clinicians to rank 34 pre-specified conditions; the ten with the lowest mean ranking were included in the core dataset.
**Key Results:** Across all rounds, 98 responses were obtained (Initial: 19, Second: 21, Third: 58). The final core dataset comprises six domains: (1) Demographics (age, gender, ethnicity, place of residence, socioeconomic status – with specific guidance for UK census categories and indices of multiple deprivation); (2) Specified morbidities (dementia, stroke, ischaemic heart disease, diabetes mellitus, cancer, congestive cardiac failure, COPD, parkinsonian syndromes, hypertension, depression – recorded as individual binary variables plus a count); (3) Medication count (whole integer, with detailed inclusion/exclusion criteria); (4) Functional ability (Barthel Index for basic ADLs and/or Nottingham Extended Activities of Daily Living for instrumental ADLs; Disability Assessment of Dementia recommended for dementia populations); (5) Frailty assessment (Clinical Frailty Scale 2.0, ordinal 1–9, assessed holistically; for acute admissions, based on function two weeks prior); (6) Cognition (hospital setting: delirium screening with 4AT ± DSM-5, consider IQCODE; community setting: prospective objective assessment using e.g., MMSE, MoCA, Mini-ACE, ACE-III, Stroop test – raw total score and assessment used must be recorded); (7) Patient-reported outcome measures (recommended for all studies, with no single tool mandated; options include EQ-5D, SF-36, PROMIS Physical Function Short Form 10, PROCOG). The extended dataset provides standardised guidance for additional demographic data (disability, religion, biological sex, education, sexual orientation), CIRS-G for multi-morbidity, handgrip strength (Jamar dynamometer, best of each side and overall, record dominance), walking speed (m/s to two decimals, specify course length and protocol), frailty phenotype (score 0–5, specify definition of low physical activity), frailty index (record to two decimals, suggest deficits from validated indices), mood (GDS-15 total score), and nutrition (MNA-SF for screening, full MNA for assessment, categorised as not malnourished/at risk/malnourished).
**Clinical Implications:** The core and extended datasets are intended for all prospectively conducted geriatric medicine research in the UK, including clinical trials, cross-sectional studies, and cohort studies. Standardisation is expected to enable individual patient data meta-analyses and secondary data analyses without requiring new prospective studies, reducing participant burden and accelerating the research timeline. The datasets are designed to be minimum standards, not exhaustive; researchers may add further variables. They are also considered applicable to non-clinical ageing research and other specialties conducting research in older adults. Limitations include: the non-systematic initial literature review; predominantly doctor respondents (despite inviting nurses, AHPs, pharmacists); inability to reach consensus on some items (e.g., specific cognitive tool); the small number of core items may reduce representativeness; the dataset has not been prospectively trialled to demonstrate impact; and it may require updating as new tools emerge. The authors plan dissemination via the British Geriatrics Society, Geriatric Medicine Research Collaborative, and NIHR Ageing Clinical Research Network, and encourage citation of this manuscript when the dataset is used.