A quantitative evidence base for population health: applying utilization-based cluster analysis to segment a patient...

Area of Science:

• Health Services Research
• Data Science in Healthcare
• Population Health Management

Background:

• Traditional population segmentation relies on demographics or morbidities, often overlooking specific care needs across settings and focusing only on high-need patients.
• Understanding diverse care needs is essential for improving overall population health.
• Existing methods may not adequately capture the heterogeneity within patient populations, particularly for lower-need individuals.

Purpose of the Study:

• To explore the application of utilization-based cluster analysis for segmenting a general patient population into homogeneous groups.
• To identify distinct patient segments based on their healthcare utilization patterns.
• To provide a quantitative evidence base for improving population health strategies.

Main Methods:

• A k-means cluster analysis was performed on administrative datasets of 300,000 patients, incorporating socio-demographic variables, morbidities, and care utilization.
• Patient segmentation was based on six key utilization variables: non-elective and elective inpatient admissions, outpatient visits, GP practice visits, GP home visits, and prescriptions.
• Segments were analyzed post-hoc to determine their distinct morbidity and demographic profiles.

Main Results:

• Eight distinct patient segments were identified, each exhibiting significantly different utilization patterns across primary and secondary care settings.
• Each segment presented unique morbidity patterns and demographic characteristics, defining eight distinct care user types.
• Comparison with traditional patient grouping methods highlighted the heterogeneity missed by conventional approaches, especially among lower-need patients.

Conclusions:

• Utilization-based cluster analysis provides a robust method for segmenting patient populations into groups with unique care priorities, supporting evidence-based population health improvements.
• This approach offers advantages over traditional methods by also segmenting lower-need populations, enabling targeted preventive interventions.
• The identification of diverse care user types offers valuable insights into patient needs across the entire care continuum.