Background Integrated Care Systems (ICSs) are being introduced into the National Health Service (NHS) in England to replace Sustainability and Transformation Partnerships (STPs). They aim to improve care through place-based collaboration between primary, secondary and community providers. It is important that new organisational configurations adequately reflect existing patterns of patient care to minimise disruption resulting from patients crossing between ICSs. Methods All planned outpatient hospital clinic appointments from 1st April 2017 to 31st March 2018 for patients resident in England to NHS hospitals in England were identified from Hospital Episode Statistics. Markov Multiscale Community Detection (MMCD), an unsupervised network clustering technique, was used to identify natural communities of GP practices, hospitals and geographic regions according to patterns of GP practice registration and outpatient clinic referral. Two primary measures of care coverage were calculated; the proportion of patients registered to a GP practice in a different community than they reside, and the proportion of outpatient clinic appointments to hospitals in a different community to the referring GP practice. Results 109,830,647 outpatient clinic appointments were identified for 20,992,695 patients. A configuration of 42 ICSs was identified from MMCD to match the 42 STPs of the current configuration. In the current STP configuration, 534,946 patients (2.6%) were registered to a GP practice in a different STP than their residence, compared to 334,192 (1.6%) in the optimal MMCD configuration. 16,110,267 hospital clinic appointments (14.7%) occurred in a different STP to the referring GP practice, compared to 11,518,735 (10.5%) in the MMCD configuration. Conclusions Millions of hospital appointments annually occur in hospitals outside of the STP of the referring GP practice. Applying MMCD we derive spatially consistent partitions of NHS Trusts and GPs into ICSs that are more representative of existing patient flows while maintaining the intended population size and number of ICSs. The findings of this study should guide policymakers locally and nationally to determine where ICS boundaries may be redrawn and the extent to which such changes would better reflect the current needs of patients.