JMIR Publications

ABSTRACT

Background:

Population health management and precision public health aim to reduce healthcare costs and improve outcomes. To support these strategies, integrated dashboards and analytics tools are essential for tracking utilization and informing interventions.

Objective:

To develop a dashboard to characterize inpatient utilization across diverse clinical conditions, supporting care coordination and system-level planning.

Methods:

We used the Clinical Classifications Software Refined (CCSR), which aggregates thousands of International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM) codes into ~530 clinically meaningful categories. As our institution uses the ICD-10 Australian Modification (ICD-10-AM), we developed a semi-automated algorithm to map ICD-10-AM codes to CCSR categories via ICD-10-CM as an intermediary. This mapping was applied to principal diagnoses from inpatient encounters in the Singapore Health Services (SingHealth) Diabetes Registry (SDR) from 2019 to 2023. An interactive dashboard was built to visualize admission patterns and geospatial utilization.

Results:

The SDR captured 564,209 admissions with 5,892 unique ICD-10-AM codes. In 2023, hospital utilization was visualized using nested treemaps by admission count and inpatient length of stay (LOS). Circulatory system diseases accounted for the highest share of admissions (16.2%) and LOS (17.9%), including coronary atherosclerosis, myocardial infarction, and stroke. Respiratory and endocrine conditions (e.g., pneumonia, diabetes complications) were also major contributors. Short stays were linked to acute conditions, while longer stays were associated with cerebrovascular and injury-related diagnoses. Geospatial mapping showed high utilization within the institutional catchment and cross-cluster admissions across Singapore.

Conclusions:

The CCSR-based dashboard enables structured analysis of inpatient utilization and supports planning of community-based interventions.