Investigators reported a notable decline in emergency department use among Madera County residents in the year after Madera Community Hospital (MCH) closed. Combined ED encounters—defined as ED-only plus inpatient-from-ED—fell from 398.47 to 333.61 encounters per 1,000 residents between 2022 and 2023, a relative decrease of 16.3%. By comparison, Fresno County residents showed essentially no change over the same interval (399.61 to 398.49 encounters per 1,000; −0.3%). These county-level rates were calculated using annual population estimates and encounter counts assigned by patient county of residence, regardless of where care was delivered.
The observed drop concentrated in encounters classified as ED-only. The authors reported a 20.4% decline in ED-only rates for Madera residents in the first full year after closure. Short, direct declines in ambulatory emergency visits stood beside more complex shifts in inpatient-from-ED admissions, which increased numerically—authors recorded a 13.4% increase—but adjusted comparative models did not consistently isolate a Madera-specific rise in that inpatient-from-ED component.
Data indicated divergent trajectories for the two encounter types. ED-only encounters fell sharply in Madera County; inpatient-from-ED encounters rose in raw counts. The study team performed adjusted comparative analyses that failed to demonstrate a consistent Madera-specific increase in inpatient-from-ED utilization after accounting for trends in the comparison county and other modeled factors. The source did not report specific covariates or full model specifications in this summary; those methodological details are in the underlying dataset and analysis code referenced by the authors.
Exploratory interrupted time-series regression estimated Madera-specific closure-period rate ratios of 0.871 for combined ED utilization and 0.848 for ED-only utilization. These point estimates indicate lower postclosure rates relative to expected counterfactual trends, according to the authors’ modeling approach. The report emphasized that these results were exploratory and that overdispersion in counts limited formal statistical inference.
The investigators stratified ZIP-code encounter counts by pre-closure dependence on Madera Community Hospital. Combined ED encounter counts declined 18.4% in ZIP codes categorized as high-MCH-dependence, compared with a 5.8% decline in low-dependence ZIP codes. That geographic heterogeneity supports the interpretation that communities previously relying most on MCH experienced larger reductions in ED use after the facility closed.
The source did not provide granular patient-level demographics in the text of the preprint summary. It also did not report whether the ZIP-code stratification adjusted for population shifts, changes in insurance coverage, or contemporaneous health system changes beyond the comparator county framing.
The study used publicly available California hospital Patient Origin/Market Share data spanning 2015–2024. Investigators divided the timeline into pre-pandemic years (2015–2019), pandemic-affected pre-closure years (2020–2022), and two complete postclosure years (2023–2024). Encounters were attributed to the county of patient residence rather than the treating facility. Outcomes were expressed as encounters per 1,000 residents using annual population estimates from the California Department of Finance.
The primary outcomes were combined ED encounters and the two components—ED-only and inpatient-from-ED—analyzed separately. The authors performed an exploratory comparative interrupted time-series analysis comparing Madera County to Fresno County, and they reported both raw percentage changes and modeled rate ratios for the closure period.
Source data and code links were cited by the authors and are publicly available from California Department of Health Care Access and Information and the California Department of Finance. The preprint stated that no restricted patient-level data were used.
The authors explicitly identified several constraints on interpretation. Only two counties were included in the comparative analysis; that limited the scope of statistical inference. The analysis included only two complete closure years (2023–2024) at the time of reporting, which reduces the ability to assess longer-term secular trends. The authors also reported substantial overdispersion in encounter counts, which complicated formal hypothesis testing and the stability of regression estimates.
Crucially, the observational design does not establish causation. The study did not determine whether the reduced ED utilization reflected substitution to other care settings (for example, urgent care or primary care), a true reduction in demand, or unmet medical need among residents. The source did not report patient-level clinical outcomes, morbidity, mortality, or measures of access such as travel time to alternate EDs. Those remain unanswered questions requiring additional data or complementary study designs.
Ethics and data transparency were addressed: the authors declared no competing interests, cited public data sources, and reported IRB review and exemption determinations as appropriate in the preprint record. The preprint format limits peer-review confirmation; readers should interpret the findings with that context in mind.