Frequent, Persistent, and Yearly Inpatient Utilization Across a Multi-Hospital Government Health System in Jeddah, Saudi Arabia: A Retrospective Three-Definition Analysis (2022–2024)
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
A small proportion of hospitalized patients generates a disproportionate share of inpatient admissions, bed-day utilization, and associated health expenditure globally. 1,2 In Saudi Arabia, where Vision 2030 mandates measurable reductions in preventable hospitalizations 27 and hospitals consume approximately 79% of public health expenditure, 23 population-level evidence on inpatient frequent utilization is absent from the published literature. A key methodological limitation of existing studies is reliance on a single threshold that cannot distinguish acute high-frequency episodes from sustained multi-year hospital dependence.
Methods
A retrospective cross-sectional study analyzed electronic health records from three public hospitals in Jeddah — East Jeddah Hospital (EJH), King Abdul-Aziz Hospital (KAAH), and Thagher Hospital (TH) — for January 2022 to December 2024. Records from two clinical information systems (Oasis at KAAH and TH; Careware at EJH) were harmonized using an eight-stage data quality protocol applied to 258,391 raw encounters, yielding a final cohort of 82,160 unique patients and 100,685 valid inpatient visits. Three complementary definitions were applied: Frequent Utilizer (FU: ≥3 admissions within any rolling 365-day window4,6), Persistent Utilizer (PU: ≥3 admissions with ≥24 months between first and last), and Yearly Utilizer (YU: ≥1 admission in each of 2022, 2023, and 2024). Analyses were conducted in JASP 0.95.4 (JASP Team, 2025). 36
Results
FU prevalence was 2.96% (n=2,434), PU 0.60% (n=494), and YU 0.62% (n=507). Overlap analysis identified 177 compound utilizers (0.22%) satisfying all three criteria simultaneously, with a median of 7 admissions and 33.44 bed days — more than thirteen times the standard patient median. Strikingly, compound utilizers had the youngest median age of any utilizer group (24 years), while Saudi nationality concentration rose progressively from 75.0% in standard patients to 87.6% in compound utilizers, and female predominance was highest in the persistence-defined groups (PU-only 62.9%, YU-only 63.6%). All three ANOVA models confirmed significant utilizer status × hospital interactions (all p<.001). Logistic regression confirmed age, Saudi nationality, and hospital as independent predictors across all definitions. A gender discrepancy — significant for males in FU Model 1 (OR=1.090, p=.039) but not Model 2 (p=.181) — was attributable to age confounding.
Conclusions
Approximately one in thirty-four inpatients meets the FU criterion in this Jeddah system, with significant between-hospital variation. The three-definition framework reveals clinically distinct utilization phenotypes invisible to any single threshold, including compound utilizers with extraordinary burden and unexpectedly young age, and persistent users entirely missed by annual-window definitions. Saudi nationality is the strongest and most consistent predictor across all definitions. Integrated clinical pathways connecting primary care and community services to hospital care, with shared accountability for quality across levels, are the recommended system response aligned with Vision 2030.
