Systemic Population Segmentation to Enable Systemic Care Integration: A Singapore Regional Population Health System Experience

Background: Population segmentation is an important planning and development activity to enable efficient health system functioning. Together with segment-specific interventions and systematic measurement, these activities form the core value-driven strategy for health and healthcare systems to enable care integration to improve outcomes and value. Yishun Health (YH), a regional population health and hospital system under Singapore’s National Healthcare Group (NHG) caring for more than 330,000 residents in Northern Singapore, developed the Unified Care Model (UCM) and its subsidiary Lifelong Care (LC) and Episodic Care (EC) services models to guide the development of its future-state integrated Service Delivery System. Population segmentation models that generate mutually exclusive and collectively exhaustive populations at the macrosystem level, will provide clearer planning parameters to enable services planning, care redesign and resource allocation to accelerate health services integration and health system transformation. Methods: A River of Life framework was designed by NHG to meet health needs for the rapidly aging population. Under this framework, YH population was segmented to five categories of needs i.e., Living Well, Living with Illness, Living with Frailty, Crisis and Complex Care and Leaving well. Building upon this, residents known to YH were segmented into seven LC segments representing different level of biopsychosocial needs. Next, residents admitted to YH’s hospital system were further segmented into seven EC segments by adding onto the LC segmentation model different levels of crisis care needs surrogated by the Patient Acuity Category (PAC) value. LC and EC segmentation models were evaluated for their ability to stratify patients into distinct risk-groups using LC and EC outcomes. The interdependence between LC and EC population segmentation models were used to enable systemic services planning at the regional population health system level. Results: All 330,000 residents were segmented extrapolating from the results of a Population Health Survey conducted by YH, of which 78,810 residents visited YH in 2022 were further segmented into LC segments. Healthcare utilisation (i.e., number of Emergency Department visits, Specialised Outpatient Clinic visits, admissions and annual bed-days), and annual total healthcare cost were found to be statistically different across LC segments (p<0.05). Of 41,521 admissions in 2022, residents with higher EC needs was associated with longer average length of stay in our hospital system, higher rates of avoidable admissions, 30-day emergency readmissions and higher admission costs (p<0.05). Psychosocial issues were demonstrably the key driver for poorer outcomes. LC segmentation enabled assignment of health outcomes accountability and population health outcomes measurement at regional population health system level, whereas EC segmentation enabled assignment of crisis and complex care accountability and measurement of EC outcomes at the hospital system level. Discussion & Conclusion: Using a novel approach that coupled LC and EC needs, our systemic population segmentation models effectively segmented our population into mutually exclusive and collectively exhaustive groups of homogenous needs and outcomes for both LC and EC. Together with the UCM, systemic LC and EC population segmentation models enabled more systemic care integration that accelerates healthcare transformation towards people-centered, integrated and value-driven regional population health systems.