Our Future Health Cohort Tops 1.9 Million Participants, Mapping Disease Patterns Across the UK

A baseline phenomic analysis of approximately 1.9 million adult participants in the UK's Our Future Health (搜索)搜索该词条 (OFH) program has been published in Nature Medicine, describing sociodemographic composition, lifestyle and anthropometric measures, disease prevalence and medication-use patterns across England, Scotland and Wales. The study positions OFH as the largest consented cohort of its kind, while documenting systematic participation differences that shape how its data can be interpreted.

The analysis drew on OFH data releases 12 and 13, issued on 17 September and 11 December 2025, respectively. Data included baseline questionnaire responses (up to 288 questions across five sections), clinic measurements, geolocation data and linked electronic health records (EHRs) covering inpatient and outpatient visits, cancer registry and death registration records for participants registered in England. Recruitment began in late 2022, with appointments starting in England in late 2022 and in Wales and Scotland in early 2024; recruitment in Northern Ireland began only in late 2025 and is not represented in the analyzed releases.

Cohort Composition and Representativeness

The cohort's age–sex structure broadly mirrored the UK population but included a higher proportion of female participants (57% versus 43%) and fewer younger adults. Deprivation was unevenly represented: participants in the most deprived Index of Multiple Deprivation (IMD) quintile accounted for 13% of the cohort versus 20% nationally, while the least deprived quintile accounted for 27% versus 20%. Intermediate quintiles aligned more closely with population estimates.

Despite these gaps, the cohort's scale produces substantial absolute numbers in underrepresented groups. More than 40,000 participants aged 18–39 years identified as Asian British or of any other Asian or Asian British background — more than in any existing UK-based cohort, according to the authors. At registration, 97.7% of participants reported residing in England, reflecting the phased, geography-focused recruitment strategy that prioritized invitation within roughly a 5-mile radius of newly opened appointment centers, concentrated in urban areas.

Self-reported health differed modestly from national benchmarks: 43% of OFH participants reported a long-standing health condition or illness versus 40% in the 2021 Health Survey for England (HSE), while 53% reported no such condition versus 60% in HSE.

Lifestyle and Anthropometric Measures

Anthropometric and lifestyle characteristics were broadly similar to 2021 HSE estimates, with notable exceptions. Mean BMI was comparable for male participants (27.7 kg/m² in OFH versus 27.6 kg/m² in HSE) and female participants (27.0 kg/m² versus 27.4 kg/m²). Waist circumference was lower in OFH for both sexes, though the authors note that HSE waist measurements were available only for surveyed respondents attending a nurse appointment (N = 1,663), a subsample with slightly higher overweight and obese BMI categories.

Current smoking was less commonly reported in OFH among both male participants (7.4% versus 13.2% in HSE) and female participants (6.6% versus 10.6%). E-cigarette use was similar across cohorts, with more than 80% of participants reporting never having used vaping products. Frequent alcohol consumption (three to four times weekly) was higher in OFH, particularly among male participants (22.9% versus 17% in HSE). Physical activity was higher in OFH, with approximately three quarters of participants reporting walking for at least 10 minutes on 16 or more occasions per month, compared with around half of HSE respondents.

Self-rated health diverged: 79% of OFH participants reported their health as "good" or "fair" combined, but only 14% reported "excellent" health versus 36% of HSE respondents. The authors note that differences in questionnaire wording and response options may contribute to these discrepancies. Mean systolic and diastolic blood pressure on the second measurement were 126.6 and 80.1 mm Hg, respectively, with higher values in male participants.

Disease Prevalence and Cross-Cohort Concordance

Prevalence estimates for 109 common self-reported conditions in OFH correlated strongly with UK Biobank (搜索)搜索该词条 (UKB) estimates (Pearson r = 0.784, P = 6.35 × 10⁻²⁴); restricting to participants aged 40–69 years yielded a similar correlation (r = 0.793, P = 1.01 × 10⁻²⁴). Comparison with Global Burden of Disease estimates for the UK across 38 chronic, long-duration conditions also showed strong correlation (r = 0.80, P = 1.65 × 10⁻⁹), though the authors caution that OFH captures lifetime diagnoses while GBD estimates represent active disease at a specific time point.

Divergences were observed. OFH participants reported substantially lower prevalence of age-associated cardiometabolic conditions, including hypertension (搜索)搜索该疾病, myocardial infarction and stroke, while depression (搜索)搜索该疾病, anxiety and related mental health conditions were reported more frequently than in UKB. Category-level analyses confirmed positive correlations for most disease classes except reproductive system conditions and pregnancy- or childbirth-related complications, which the authors link to differences in age structure.

Across 13 well-studied common diseases defined by ICD-10 codes in linked EHRs, OFH showed lower prevalence but substantially larger absolute case counts for most conditions compared with All of Us (搜索)搜索该词条, FinnGen (搜索)搜索该词条 and UKB. Type 2 diabetes (搜索)搜索该疾病 prevalence was higher in All of Us than in OFH and the other European cohorts examined, consistent with established differences in population prevalence between the USA and the UK.

Replication of Clinical Associations

Adjusted odds ratios for associations between 18 selected health conditions and established clinical correlates were positively correlated between OFH and UKB (Pearson r = 0.783, P = 1.21 × 10⁻⁴), with similar effect sizes when OFH participants were restricted to ages 40–69 years (r = 0.767, P = 2.05 × 10⁻⁴). Well-established associations — cardiovascular disease with hypertension (搜索)搜索该疾病, respiratory disease with smoking, metabolic conditions with adiposity, psychiatric conditions with family history, and cancer outcomes with smoking or family history — were consistently replicated. Stratified analyses by self-reported ethnicity showed several associations varied in strength across groups while remaining directionally consistent with UKB.

Medication Use and Self-Report Concordance

Medication use showed clear age-related patterning, with supplements and pain-relief medications commonly reported across all age groups and increasing use of heart and circulatory, bone health and cancer therapies at older ages. Multivariate analyses linking current medication use to corresponding self-reported diagnoses yielded large disease-specific odds ratios, including approximately 550 for cancer, 360 for autoimmune disorders and 220 for neurological disorders. Correlation analyses revealed coherent co-use patterns, including clustering of mental health and pain-relief medications and of supplements with bone-health medications.

Agreement between self-reported diagnoses and linked inpatient EHR-derived diagnoses varied by condition, with Cohen's κ ranging from 0.29 for high cholesterol to 0.66 for cancer. Concordance was generally higher for clinically salient, well-defined conditions such as cancer, and lower for conditions susceptible to differences in ascertainment and healthcare setting, including high cholesterol and depression (搜索)搜索该疾病.

Cancer Prevalence Against National Estimates

Age- and sex-specific cancer prevalence was assessed using self-reported diagnoses, inpatient EHR records and linked cancer registry data, and compared against National Disease Registration Service (NDRS) estimates. For all cancers combined, prevalence across the three OFH sources was broadly comparable to national figures at younger ages, with self-reported prevalence exceeding EHR- and registry-derived estimates at older ages. Among participants aged 85–89 years, the self-report rate was 22,728 per 100,000 versus 16,514 per 100,000 from inpatient EHR and 20,861 per 100,000 from cancer registry data, against a national estimate of 19,622 per 100,000.

Breast and prostate cancer (搜索)搜索该疾病 showed similar patterns, with self-reported rates exceeding national estimates from midlife onwards, potentially indicating recall bias and sample ascertainment differences whereby self-reported diagnoses may capture borderline or preinvasive lesions. Colon and rectal cancer prevalence tracked national estimates closely across data sources. In contrast, lung and bronchial cancer prevalence was substantially lower in OFH than national estimates across all three ascertainment methods, in both sexes and at all ages with sufficient counts — a convergence the authors interpret as probably reflecting true differences in disease prevalence consistent with the cohort's risk factor profile.

Rare and Less Common Conditions

In an exploratory analysis of 187 less common conditions, including rare diseases meeting the European rarity threshold of fewer than 1 in 2,000 people, OFH consistently provided substantially greater case counts than UKB: 668 individuals with myasthenia gravis (搜索)搜索该疾病 (versus 266 in UKB), 464 with primary biliary cholangitis (搜索)搜索该疾病 (versus 312), 172 with cystic fibrosis (搜索)搜索该疾病 (versus 25) and 83 with neuromyelitis optica (搜索)搜索该疾病 (versus 20). Prevalence estimates for the full set were broadly correlated between cohorts (r = 0.79, P = 2.24 × 10⁻⁴¹).

Several conditions showed disproportionate representation beyond what cohort size alone would predict, including idiopathic intracranial hypertension (搜索)搜索该疾病 (1,739 versus 132 cases in UKB) and complex regional pain syndrome type 2 (搜索)搜索该疾病 (922 versus 19), probably reflecting OFH's younger age profile and broader recruitment criteria. Conversely, pleural mesothelioma (搜索)搜索该疾病, strongly associated with older age, was underrepresented relative to UKB (83 versus 277). Marked sex imbalances were observed for conditions including systemic lupus erythematosus (搜索)搜索该疾病 (N = 1,825; 90.4% female) and interstitial cystitis (搜索)搜索该疾病 (N = 1,409; 91.0% female).

Limitations and Interpretation

The authors state that OFH has estimated a response rate of 4.5% based on the proportion of invited individuals who consented, and that a final response rate cannot yet be produced as recruitment continues. They caution that OFH should not currently be used to derive generalizable disease prevalence and incidence rates across all conditions — a limitation particularly relevant for rare conditions, given that approximately 70% of rare diseases begin in childhood and nearly one third of affected patients die before age 5, meaning many are inherently underrepresented in an adult volunteer cohort.

The study relies primarily on self-reported participant data, which may incompletely capture chronic and ambulatory conditions relative to primary care EHR data; primary care records are not yet included, and current EHR-derived definitions should not be treated as a definitive gold standard. The authors also note that the higher burden of self-reported mental health conditions may partially reflect changes in self-reporting and diagnosis over time, as well as the differential impacts of UK austerity measures and the COVID-19 pandemic. All recruitment occurred after the acute phase of the pandemic, and contemporaneous pressures such as prolonged NHS waiting times may have selectively attracted participants with unmet healthcare needs.

Researchers using OFH for causal inference must consider selective participation and collider bias, though the authors note that nonrepresentativeness does not necessarily lead to substantial bias in exposure–disease associations. The cohort is expected to include many pairs of genotyped first-degree relatives, enabling future family-based and within-household analyses. OFH has announced it is developing sampling weights benchmarked to the 2021–2022 UK census, and the authors provide an open-source Python library, Phenofhy, to support reproducible analyses of OFH phenotype data.