BackgroundHomeless health care is often characterised by physical health, mental health, and substance use problems, resulting in high use of emergency care, poor outcomes, and extreme social inequities. We assessed health needs as prevention opportunities for hospitalised people experiencing homelessness (PEH) in England. MethodsThis population-based retrospective cross-sectional study used anonymised national Hospital Episodes Statistics Admitted Patient Care data. PEH were identified as having at least one homeless code (“no fixed abode”, “registered with a homeless-exclusive GP practice”, “clinical diagnosis of homelessness”) from April 1, 2017, to March 31, 2018. We analysed admissions for PEH and for housed people. We estimated the prevalence of demographic and admission characteristics and diagnoses by 10th International Classification of Disease (ICD-10) chapter. We developed novel diagnostic phenotypes for physical health (internal disease processes) and psychosocial adversity (mental health, substance use, violence, and social factors). We compared admissions between PEH and housed people using sex-stratified logistic regression adjusted for age and ethnicity. FindingsThere were 15 566 010 admissions (51 643 PEH and 15 514 367 housed people). Compared with housed people, proportionately more PEH were younger (PEH aged 26–45 years, n=24 224 [46·9%], housed people n=3 323 951 [21·4%]), male (PEH n=37 662 [72·9%], housed people n=6 819 157 [44·0%]), and not White British (PEH n=14 605 [28·3%], housed people n=3 447 183 [22·2%]). Emergency admissions were more common among PEH (PEH male n=30 958 [82·2%], housed people male n=5 321 428 [34·3%], adjusted odds ratio [aOR] 8·76, 95% CI 8·53–9·00). The most common primary diagnoses by ICD-10 chapter for PEH were mental and behavioural conditions (PEH male n=7118 admissions [18·9%], housed people male n=155 144 [1·0%], 12·97, 12·61–13·34). Admissions for the psychosocial adversity phenotype were higher in PEH, particularly for women (PEH female n=3922 [28·1%], housed people female n=155 644 [1·79%], 18·18, 17·50–18·88). Physical health phenotype admissions were less common in PEH (PEH male n=7510 [19·9%], housed people male n=1 821 397 [26·7%], 0·91, 0·89–0·94), but specific infections, cancers, respiratory, and cardiovascular diseases were more common among PEH for both men and women. InterpretationThese results support targeting of preventative interventions for PEH before, during and after admission to hospital, highlighting psychosocial needs. Future research should aim to produce reliable estimates of the size of the national homeless population to enable calculation of admission rates for psychosocial and physical health diagnoses. FundingNational Institute for Health and Care Research (NIHR).