Risk Assessment of Emergency Medical Resource Exhaustion Under Major Infectious Disease Outbreaks: Based on Discrete-Event Simulation Models

Abstract

Objective

To quantify the risk of emergency medical resource exhaustion during the early stages of major infectious disease outbreaks and proposed resource allocation optimization strategies.

Methods

This study integrated discrete event simulation and patient flow theory to develop a dynamic medical resource allocation simulation system in Hubei Province from January 23 to February 21, 2020. This period referred to the first 30 days interval following the implementation of the lockdown in Wuhan. The system simulated the utilization of general hospital beds, ICU beds, and ventilators under three distinct resource supply scenarios: baseline (Expected), optimal (Best case), and worst (Worst case). Simulation outputs including cumulative depletion days, waiting time, and deaths attributable to delayed access to critical resources were summarized using descriptive statistical analysis.

Results

In the Best case scenario, the cumulative depletion days of ICU beds and ventilators persisted for 4 days and 6 days respectively, while no shortages occurred for general beds; 384 deaths were attributable to waiting for resources. In the Expected scenario, the cumulative depletion days of ICU beds, ventilators, and general beds were 11 days, 17 days, and 1 day respectively, with 766 deaths attributable to waiting for resources. In the Worst case scenario, the cumulative depletion days of both ICU beds and ventilators were 28 days, and general beds also experienced severe congestion resulting in an average wait time of 4.18 days, a maximum wait time of 11.28 days, and with 15,029 deaths attributable to waiting for resources.

Conclusion

The abrupt surge in cases at the onset of the epidemic exerted considerable pressure on medical resources. The intensity of resource supply is highly correlated with the risk of patient death, and ventilators and ICU beds are the key resources affecting the death risk. This simulation model can provide a scientific tool for emergency resource reserve and allocation in public health emergencies.

Introduction

The outbreak of major infectious disease epidemics often causes a catastrophic impact on the public health system. The resulting strain on medical resources may significantly increase patient mortality as well as socioeconomic losses.¹ The global pandemic of Coronavirus Disease 2019 (COVID-19) has exposed the vulnerability of medical institutions in responding to public health emergencies. Especially when key medical resources such as Intensive Care Units (ICUs) and ventilators are insufficiently reserved, the risks of delayed patient treatment and excess mortality increase significantly.^2,3 Hubei Province was the first province in China to experience an outbreak of the COVID-19 epidemic. In the early stage of the epidemic in Hubei Province in 2020, the sudden surge of cases in a short period caused the medical system to operate beyond its capacity, exposing the core contradiction of dynamic imbalance between resource supply and demand.⁴

Over the past three decades, medical systems worldwide have generally adopted the “zero inventory” management model, continuously reducing safety stocks and pursuing maximum resource utilization through the Just-in-Time supply chain. Although this strategy reduces operating costs under normal conditions, it has led public health institutions to fall into an extreme resource shortage dilemma during the COVID-19 pandemic.⁵ To achieve the timely supply of medical resources during the epidemic, the academic community has constructed a series of medical resource demand forecasting models,⁶ providing decision support tools for the strategic allocation of medical resources involving tens of millions of people. Currently, epidemiological models are mainly used to predict the spread of infectious diseases at the national or regional level. These models predict the number of COVID-19 infections and hospitalizations, which are then converted into the demand for hospital beds and the number of deaths.^7–9 Walker et al calibrated a global model using multi-country data, which includes indicators such as population structure, social structure, and access to medical services to predict the medical resource capacity and demand of various countries during the COVID-19 outbreak from a macro perspective.¹⁰ Rainisch et al used the SEIR partition model to predict the number of COVID-19 patients to estimate the demand for ICU beds and ventilators.¹¹ The World Health Organization (WHO) integrated the SEIR model to develop the COVID-19 Essential Supplies Forecasting Tool (ESFT), which predicts the required quantities of personal protective equipment, diagnostic equipment, consumable medical supplies, biomedical equipment for case management, and essential drugs for supportive treatment and care of COVID-19.¹² Furthermore, several studies have enhanced epidemiological models by integrating lockdown measures and various non-pharmaceutical interventions, thereby offering a robust foundation for informing public health policies.¹³

Existing studies on the prediction of medical resource demand for emerging infectious diseases are mainly based on static supply–demand analysis frameworks.^7–11 However, this approach has two key limitations. Firstly, static models typically assume that resource supply and demand match instantaneously, ignoring the dynamic coupling between the surge characteristics of patient admissions and the delay effects of treatment cycles and downstream transfers.^14,15 Secondly, traditional models often do not explicitly account for the elevated death risk associated with waiting for critical resources. For example, when ICU beds are in short supply, critically ill patients detained in general wards may face an increased risk of death due to delayed transfer to the ICU, and such capacity-dependent mortality has not been adequately quantified in many forecasting frameworks.^14,16 It should be noted that the demand for medical resources during the epidemic is closely related to the flow process of patients within the medical system. The dynamic path of patients (ie., “patient flow”)—from admission assessment, department triage to treatment turnover—directly affects the rhythm of resource consumption and the risk of shortage. However, existing static frameworks seldom incorporate this critical dimension, thereby complicating an accurate representation of the mechanism from resource scarcity to patient flow disruption and ultimately to increased mortality risk.

Discrete Event Simulation (DES) offers an effective approach to addressing the aforementioned limitations. DES is a computer simulation technology usually used to model and analyze dynamic systems driven by a series of discrete events. The system models operations as discrete events occurring at specific time points—such as patient arrival, service initiation, and service completion—and advances the simulation clock sequentially according to the occurrence of these events. It records dynamic changes in system state variables, including queue length and resource utilization, thereby enabling performance evaluation through key metrics such as waiting times and resource efficiency, and supporting informed decision-making.

DES has been widely used in simulating the use of medical resources during the COVID-19 pandemic. It is primarily used to address two categories of problems: the first is to provide decision support for pandemic-related scenarios through scenario simulation, such as strategies for mitigating the impact of the pandemic, balancing non-COVID-19 patient care activities, and optimizing vaccination programs. Currie et al discussed how simulation models can help mitigate the impact of COVID-19, proposed multiple issues where simulation can be used as decision support, and emphasized that one of the issues is related to inpatient bed capacity and intensive care capabilities.¹⁷ Wood and Melman et al (2021) both studied the trade-offs between the advantages and disadvantages of reducing the diagnosis and treatment activities for non-COVID-19 patients during the epidemic.^16,18 Asgary et al applied DES to evaluate different setting schemes related to large-scale vaccination immunization facilities.¹⁹ Another type of research is aimed at estimating the number of required hospital beds. For example, Wood et al sought to predict mortality associated with the COVID-19 pandemic and classified such deaths into two categories: those attributable to healthcare capacity constraints and those not attributable to such constraints.¹⁴ The study analyzed various scenarios concerning the number of patients admitted with confirmed positive COVID-19 test results and the availability of intensive care unit (ICU) beds Mallor et al aimed to predict the number of beds required by COVID-19 patients in the ICU and other areas of the hospital in the coming weeks.¹⁵ In addition to estimating the bed demand brought by COVID-19 patients, Le Lay et al also evaluated different strategies for managing the surge in bed demand.²⁰ However, most of the existing DES studies are based on the parameters of the European and American medical systems, lacking modeling verification for the specific context of China. Moreover, during the early stages of an epidemic outbreak when pathogen characteristics remain unclear and clinical management pathways have not yet been standardized simulation studies often face significant challenges, including high uncertainty in fundamental parameters and substantial difficulties in real-time data calibration.

This study utilizes the early post-lockdown phase of the COVID-19 outbreak in Hubei Province, China (January 23 to February 21, 2020; the first 30 days after the Wuhan lockdown) as the empirical context to develop a dynamic medical resource simulation system. This window was selected to capture the acute surge period and to ensure consistent daily reporting of both patient demand and resource availability for model calibration. The system is based on an enhanced version of the COVID-19 Resource Estimator (CORE) model originally developed in Ontario, Canada21. By integrating discrete event simulation and patient flow theory, this study quantifies the influence mechanism of the demand-supply gaps in general beds, ICU beds, and ventilators on patient queuing and capacity-dependent mortality, and simulates the system’s response characteristics under various resource supply scenarios. Research can provide evidence-based support for the risk assessment of medical resource shortages in uncertain situations.

Materials and Methods

Data Sources

The time period selected for this study spanned from January 23, 2020—the date of the lockdown in Wuhan City—to February 21, 2020. Daily simulations were conducted over this 30-day period. This period was selected to represent the earliest emergency stage after lockdown, during which patient demand increased sharply. Importantly, during this interval, both patient demand indicators and the daily availability of general beds, ICU beds, and ventilators were reported in a relatively consistent manner, enabling coherent calibration of the demand–supply process in the simulation model. Although the epidemic continued to evolve outside this window, later phases were strongly affected by large, policy-driven capacity expansions and changes in reporting/clinical practice, which would require a multi-phase model beyond the scope of this paper. We therefore acknowledge this as a key limitation and discuss its implications. Despite the implementation of stringent public health interventions by the government, limited understanding of the disease and the absence of established clinical treatment protocols led to a substantial influx of patients into hospitals. The scenario observed over these 30 days aligns with the pattern of regional medical treatment resources utilization during the sudden emergence of an unknown acute infectious disease.

The data sources of this study comprise two main categories. First, official data were obtained from the website of Hubei Province, China, during the outbreak of COVID-19, including the number of cases, deaths, and total bed capacity. Second, we incorporated published academic articles and publicly available literature related to the COVID-19 pandemic, which provided parameters such as average length of hospital stay, probabilities of medical resource utilization, and ICU resource allocation. Detailed references are presented in Table 1.

Table 1.

Key Variables for Modelling

Key Variable	Value	Source
Daily cumulative number of COVID-19 cases in Hubei Province	Supplementary Figure 2a	Official website of Hubei Provincial Health Commission21
Total number of beds	-	Hubei Province Health Statistics Yearbook4
Number of general beds in designated hospitals	Total beds * 26% (Supplementary Figure 2b)	Xinhua News Agency22
Number of ICU beds in designated hospitals	Total beds/10 (Supplementary Figure 2c)	Official website of the State Council of the People’s Republic of China23
Number of ventilators in designated hospitals (units)	Number of ICU beds * 68.75% (Supplementary Figure 2d)	Consensus on Diagnosis, Treatment and Management of Severe COVID-1924
Patient hospitalization rate (%)	43	Cumulative number of inpatients/Cumulative number of cases21
Proportion of hospitalized patients requiring ICU (%)	63	Severe case rate/Hospitalization rate = 27%/43%
Probability of ICU patients needing ventilators (%)	78	Barrett et al, 202025
Length of stay in general beds (without ICU admission) (Days)	14	Rees et al, 202026
Length of stay in ICU beds (Days)	8	Rees et al, 202026
Length of stay in general beds (after transfer from ICU) (Days)	6	Rees et al, 202026
Severe case rate (%)	27	Song et al, 202027
Mortality rate of ICU patients (%)	39	Qian et al, 202028
Mortality rate of patients waiting for ICU beds or ventilators for more than 1 day (%)	100	Assumption in this paper

The medical treatment resources examined in this article—general hospital beds, ICU beds, and ventilators—were selected because they represent the most direct, capacity-defining constraints along the inpatient care pathway (admission, escalation to critical care, and life-support). These resources have clear, quantifiable availability and directly determine queuing and resource-exhaustion risk in the discrete-event simulation framework. Other emergency supplies (eg., personal protective equipment and medications) are also critical. However, consistent daily inventory/consumption data at the provincial level were not available for calibration and are therefore not explicitly modeled in this study. These resources are all components of designated medical institutions. These designated institutions encompass hospitals specifically assigned for the treatment of COVID-19 patients during the epidemic, temporary facilities such as Huoshenshan and Leishenshan Hospitals, and mobile cabin hospitals established to address surges in patient volume. The total number of beds includes beds for treating non-COVID-19 patients and beds for treating COVID-19 patients (general beds for mild cases, ICU beds for critical cases, and isolation and medical observation beds for close contacts, etc). Due to the unavailability of comprehensive provincial-level data, this study estimates the provincial situation by using the proportion of ordinary hospital beds in Wuhan as of February 10, 2020, which accounts for 26% (4,250 out of 16,350).²² The hospitalization rate is determined by dividing the cumulative number of inpatients by the total number of confirmed cases. The mortality probability in the intensive care unit (ICU), both for patients receiving mechanical ventilation and those not, is based on data from Qian et al’s study²⁸ (2020) on the case severity and mortality of early COVID-19 patients in China, which reported a severe-case fatality rate of 39%. The data regarding the average length of hospital stay for patients in regular wards and intensive care units (ICUs) were derived from a retrospective review study conducted in China by Rees et al²⁶ The average length of stay for patients in regular wards was defined as 14 days, while that for ICU patients was set at 8 days.²⁹

This paper defines the condition in which available resources on a given day are zero as the resource depletion state. The primary findings include: (1) the number of days on which general beds, ICU beds, and ventilators are fully utilized; (2) the number of deaths occurring both during treatment and while patients were awaiting access to medical resources; (3) the waiting time and the number of patients awaiting access to medical treatment resources.

Research Methods

COVID-19 Resource Estimator Model

The Ontario COVID-19 Resource Estimator (CORE) model (Canada) was jointly developed by the Ontario government, epidemiologists, and data scientists. It is mainly used to predict the development trend of the epidemic, assess the demand for medical resources, and simulate the impact of different prevention and control measures on the medical system. The model integrates real-time epidemic data and population mobility information, and provides a quantitative basis for the government to formulate medical resource allocation strategies and public health intervention measures by dynamically simulating resource gaps under different scenarios. Its core advantage lies in integration of short-term prediction and long-term trend analysis, enabling not only the prediction of case growth in the coming weeks to inform emergency response measures, but also the simulation of epidemic progression over several months to support strategic resource planning.¹⁷

The structure of the CORE model is shown in Supplementary Figure 1. Hospitalized patients are admitted to general wards or directly to the ICU according to the severity of their illness, and some ICU patients need ventilators. In the overall model, the patient can be in one of three distinct states: undergoing treatment, recovered, or deceased. In the event that any resources (ie., general hospital beds, ICU beds, or ventilators) are unavailable, this article assumes that the patient remains in their current status and awaits the release of the required resources. For example, if an eligible patient lacks access to an available ICU bed, they will be placed in a waiting queue until an ICU bed becomes accessible. Upon transfer to a general ward or death, the ventilator and ICU bed are released. Similarly, once a patient recovers and is discharged from the hospital, the general ward bed is vacated.

Patient Flow Theory

Due to persistent pressure on hospital bed capacity, healthcare administrators have consistently prioritized the development of systems aimed at maximizing bed utilization rates. Achieving this objective effectively requires a thorough understanding of the primary patient populations served by the system and the patterns of patient flow within the facility (Patient Flow).³⁰

The core idea of Patient Flow Theory originated from industrial engineering and operations research. It was promoted and applied in the medical system by institutions such as the Institute for Healthcare Improvement from the late 20th century to the early 21st century. The theory aims to identify bottlenecks in the process by visualizing, quantifying, and optimizing the flow path of patients within medical institutions, thereby improving the utilization efficiency of medical resources and service quality. By dynamically monitoring the matching between patient flow and resources, it is possible to accelerate bed turnover, reduce congestion rates, reduce medical waste, and simultaneously improve patient satisfaction and the work efficiency of medical staff.

Discrete Event Simulation

Discrete Event Simulation (DES) is a computational technique used to model the dynamic behavior of real-world systems by representing changes in system state as discrete events occurring at specific points in time. These events, such as patient arrivals, machine failures, or service completions, serve as triggers that drive state transitions within the simulated environment. This method simulates the operational processes of complex systems such as bank queue congestion, production line bottlenecks, and hospital resource scheduling by defining key system components, including entities (eg., patients), events (ie., occurrences that trigger state changes), and event rules (eg., queue prioritization policies). Its core advantage lies in the intuitive handling of randomness such as random arrival times or service durations and concurrent events, enabling hypothesis-driven analysis to quantitatively assess the impact of various strategies on system performance metrics, including waiting time and resource utilization, thereby providing a robust foundation for optimization decisions.^17,31

Based on the Ontario COVID-19 Resource Estimator (CORE) model,²⁵ this study integrates the patient flow theory to construct a dynamic medical resource simulation model, and uses the discrete event simulation method¹² to simulate the entire process of COVID-19 patients in Hubei Province from admission to discharge. Based on the multi-level patient flow network proposed by Melman,¹⁸ patients are divided into two categories: complex process (general ward → ICU → recovery/death) and simple process (general ward → recovery/death). The patient transfer probability and length of stay parameters are calibrated using actual data (Table 1), and a resource competition mechanism is introduced — when ICU beds or ventilators are in short supply, patients enter a dynamic waiting queue. If the waiting period exceeds 24 hours, a termination event will be initiated. Upon release of the resources, a priority-based allocation strategy will be implemented.

Based on the simulation model, the following assumptions were adopted in this study:

1. The severe case rate among COVID-19 patients in Hubei Province was reported to be 0.27 on January 27, 2020.²⁷ Subsequently, the probability of hospitalized patients requiring intensive care unit (ICU) admission was estimated at 0.63, calculated as the ratio of severe cases to total hospitalizations. Furthermore, the probability of ICU patients requiring mechanical ventilation was determined to be 0.78.

2. Only inpatients requiring ICU admission are at risk of mortality during hospitalization, and the risk of death is comparable between those who require mechanical ventilation and those who do not. This assumption was adopted for two primary reasons: first, provincial-level aggregated data on the early-stage clinical stratification (the distinction between general wards and ICU wards, as well as the ventilator requirement status of ICU patients) was unavailable; second, incorporating more granular subcategories was deliberately avoided to prevent model overparameterization and maintain estimation robustness.

3. The patient passed away after waiting for over 24 hours in the ICU for access to a ventilator.

4. The number of ICU beds constitutes 10% of the total bed capacity,²³ and the number of ventilators—encompassing both invasive and non-invasive types, which are not differentiated in this article—amounts to 68.75% of the total number of ICU beds.²⁴

5. Based on the design of the patient triage mechanism, patients admitted to general wards are not involved in the transfer process to ICU treatment pathways. All patients requiring ICU care are triaged at the time of admission. Patients admitted to general wards may experience clinical deterioration necessitating transfer to the intensive care unit (ICU). This assumption simplifies the patient-flow structure for estimating resource exhaustion risk and may lead to an underestimation of ICU capacity requirements.

It is assumed that the number of available general beds, ICU beds, and ventilators in Hubei Province on January 23 represents the initial level of medical resource availability in the region. This study established three distinct resource utilization scenarios. The first scenario (Expected) is that the number of available general beds, ICU beds, and ventilators is the initial value of Hubei Province, and changes according to the actual available value of Hubei Province within the 30 days of simulation. The second scenario (Best case) assumes that the number of available general beds in 30 days is 120% of the available beds, and the number of available ICU beds and ventilators in 30 days is 120% of the existing available number. This scenario represented a feasible short-term expansion through emergency mobilization and operational optimization. The third scenario (Worst case) assumes that the number of available general beds in 30 days is 20% of the available beds, and the number of available ICU beds and ventilators in 30 days is 20% of the existing available number. This scenario represented an extreme disruption in effective capacity (eg., staffing shortages, infection-control constraints, and competition with non-COVID care). These multipliers are intended as planning stress-test bounds and can be adjusted in other settings to reflect local context. By simulating the three aforementioned resource utilization strategies, data such as the number of days until resource depletion and the number of fatalities under different scenarios can be obtained. Since the three scenarios studied in this paper only differ in the available number of medical resources, while the transmission capacity of the virus, the severity of patients’ conditions, and the mortality rate are the same in each scenario.

Simulation outputs including cumulative depletion days, waiting time (and queue length), and deaths attributable to delayed access to critical resources were summarized using descriptive statistical analysis. This paper uses Arena 14.0 simulation software to simulate the process of patients from admission to discharge. Arena software is a relatively mature simulation software, and the entire system is stable during simulation.¹ Arena software is often used to simulate logistics and supply chain management activities. It is designed for dynamic models and can simulate continuous, discrete, and mixed continuous-discrete models.³²

Research Results

Descriptive Statistics

The cumulative number of COVID-19 cases is shown in Supplementary Figure 2a. The total cumulative number of cases in the 30 days from January 23 to February 21, 2020, was 59,804, with the highest number of new cases in a single day on February 12 (14,717 cases). The changes in the number of general beds, ICU beds, and ventilators under different scenarios are shown in Supplementary Figures 2b, c, and d respectively. The number of the three types of resources all reached the peak on February 10. Taking the real situation (Expected scenario) as an example, the number of general beds, ICU beds, and ventilators was 19,310, 8,252, and 5,673 respectively.

Patient Flow Model

The hospital patient flow model constructed during the COVID-19 outbreaks in Arena software is illustrated in Figure 1. Upon arrival at the hospital, COVID-19 patients undergo clinical evaluation by medical professionals. Based on the assessment, some patients may be advised to remain in home isolation until recovery, particularly those with mild symptoms or stable conditions. It is generally assumed that individuals presenting with early-stage infection are not suitable for outpatient management, in order to minimize the risk of community transmission. Others, especially those with severe symptoms or underlying health conditions, will be admitted for inpatient care. Inpatients are categorized into two groups based on the severity of their condition: those with milder symptoms are admitted to general wards, whereas those with more severe conditions are admitted to the intensive care unit (ICU). Patients in the ICU can further be classified into two subgroups: one requiring ICU care alone, and the other requiring both ICU admission and mechanical ventilation. ICU patients face a certain risk of mortality during treatment. Once their condition stabilizes, they are transferred from the ICU to a regular ward for continued care prior to discharge. When critical resources are unavailable, patients must wait in queues until such resources become accessible. During this waiting period, patients who do not receive access to ventilators and intensive care unit (ICU) services within 24 hours face a significantly increased risk of mortality.

Figure 1.

Hospital patient flow model during the COVID-19 outbreak.

Results of Resource Discrete Simulation

This study simulated the utilization of medical resources and patient mortality in Hubei Province over a 30-day period from January 23, 2020, to February 21, 2020. During this period, there were a cumulative total of 59,804 confirmed cases of COVID-19, of which 26,246 required hospitalization. The simulation outcomes regarding medical resource utilization in Figure 2 and patient mortality are presented in Table 2. In the best-case scenario, the cumulative number of days with exhausted ICU bed was 4 days, and that of ventilator was 6, while no shortages occurred for ordinary beds. A total of 384 individuals died while awaiting medical resources, and 2,584 patients succumbed to their conditions despite receiving treatment. In the expected scenario, ICU bed depletion persisted for 11 days, ventilator depletion for 17 days, and regular bed depletion for 1 day. There were 766 deaths attributed to delays in accessing medical resources, and 2,255 patients died after receiving care. In the worst-case scenario, both ICU bed and ventilator exhaustion extended for 28 days. A total of 15,029 individuals died while waiting for medical resources, and an additional 912 patients died during treatment.

Figure 2.

Depletion and occupancy of medical resources. (A) The depletion and occupancy of available general beds. (B) The depletion and occupancy of available ICU beds. (C) The depletion and occupancy of available ventilators.

Table 2.

Depletion of Medical Resources and Patient Deaths

Resource Scenario	The Cumulative Number of Days When Resource Was Unavailable (Days)			The Number of Deaths
	ICU Beds	Ventilators	Ward Beds	While Waiting for Needed Resources	While Receiving Needed Resources
Best	4	6	0	384	2584
Expected	11	17	1	766	2255
Worst	28	28	24	15,029	912

The utilization of medical resources and patient waiting times are presented in Table 3. In the Expected scenario, the average waiting time for patients to access general hospital beds is 0.25 days, with a maximum of 0.84 days; on average, 35.63 patients await general hospital beds daily. For ICU beds, the average waiting time is 0.14 days, reaching a peak of 2.59 days, with an average of 17.21 patients waiting per day. Regarding ventilator access, the average waiting time is 1.31 days, with the longest wait extending to 4.45 days, and an average of 262.60 patients awaiting ventilators each day.

Table 3.

Utilization of Medical Resources and Patient Waiting Conditions Across Three Scenarios

Resource	Indicator	Worst Case	Expected	Best Case
Wards	Average waiting days	4.18	0.25	0
	Longest waiting days	11.28	0.84	0
	Average waiting patients	1991.85	35.63	0
ICU beds	Average waiting days	4.25	0.14	0.10
	Longest waiting days	7.27	2.59	1.97
	Average waiting patients	55.30	17.21	11.75
Ventilators	Average waiting days	6.61	1.31	1.44
	Longest waiting days	9.96	4.45	4.96
	Average waiting patients	578.98	262.60	109.81

In the best-case scenario, the average waiting time for ICU beds was 0.10 days, with a maximum of 1.97 days, and an average of 11.75 patients awaited ICU bed access daily. The average waiting time for ventilator use was 1.44 days, reaching a maximum of 4.96 days. On average, 109.81 patients were waiting for ventilator access every day.

In the worst-case scenario, the average waiting time for patients requiring general beds is 4.18 days, with a maximum waiting time of 11.28 days, and an average daily queue of 1,991.85 individuals. For ICU beds, the average waiting time is 4.25 days, peaking at 7.27 days, with an average of 55.30 patients awaiting access daily. Regarding ventilator availability, patients face an average waiting time of 6.61 days, reaching a maximum of 9.96 days, while the average number of individuals waiting for ventilators daily stands at 578.98.

Discussion

Based on empirical data from the early stage of the COVID-19 epidemic in Hubei Province in 2020 (January 23 to February 21), this study integrates discrete event simulation and patient flow theory through an enhanced CORE model to simulate system response characteristics under three medical resource supply scenarios. The impact of exhaustion in general hospital beds, ICU beds, and ventilators on patient queuing and mortality rates is quantified. The findings reveal patterns of imbalance between supply and demand for medical resources during the initial phase of the outbreak, providing a critical theoretical foundation for resource allocation strategies in public health emergencies.

Medical Treatment Resources are Being Scaled Up in Response to the Surge in Patient Cases

The significant rise in cases has substantially affected the availability of medical resources. The results of descriptive statistics analysis show that the cumulative number of COVID-19 cases in Hubei Province reached 59,804 from January 23 to February 21, 2020, with the highest number of new cases in a single day on February 12. The number of general beds, ICU beds, and ventilators all peaked on February 10. In early February 2020, among the many designated hospitals for treating COVID-19 patients in Wuhan, there was a severe shortage of hospital beds, resulting in an inability to admit confirmed cases. Tens of thousands of individuals with mild or moderate symptoms were consequently required to remain under home isolation and observation. Due to the extreme shortage of hospital beds, Wuhan urgently needed a measure to quickly and largely improve its isolation and admission capacity. During the most severe period of the epidemic in Wuhan, thousands of new COVID-19 patients were added every day. For this reason, on February 5, 2020, Wuhan City opened the first batch of three makeshift shelters by repurposing large-scale convention and exhibition centers and sports facilities. In the subsequent weeks, an additional 13 makeshift shelters were established across Wuhan.³³

This increase in cases and the variability in resource availability mirror the challenges identified in a study by Saiin et al examining the impact of the COVID-19 pandemic on Japan’s emergency medical system for patients with acute coronary syndrome.³⁴ During the Japanese epidemic, since the first case was reported on January 15, 2020, the emergency medical system experienced increased operational strain, while the number of hospitalized patients with acute coronary syndrome decreased significantly. This phenomenon may be attributable to the disproportionate allocation of medical resources toward infectious disease management, which has consequently reduced the volume of routine outpatient visits. Together, these observations suggest that the rapid, wave-like transmission of major infectious diseases can severely pressure healthcare systems across regions. It is noteworthy that the volume of resources in Hubei Province peaked on February 10, closely corresponding to the timely implementation of emergency resource allocation measures by the government at that time.

A Comparative Analysis of the Patient Flow Model and Its Applicability in the Chinese Healthcare Context

The patient flow model clearly shows the complete path of COVID-19 patients from admission assessment to final recovery or death. To a certain extent, this model is also applicable to other countries, but there are differences in applicable conditions. For example, during the epidemic in the United States, although medical resources are abundant, due to the lack of a unified coordination mechanism, each state acts independently, leading to uneven resource allocation.³⁵ In Singapore, characterized by high population density and rigorous urban management, an optimized influenza epidemic simulation model was employed during the early stages of the outbreak to formulate effective prevention and control strategies. These included coordinated intervention measures such as isolation and quarantine of infected individuals and their household contacts, as well as temporary closures of schools and workplaces to mitigate transmission.³⁶ This approach bears similarities to China’s emphasis on community-based prevention and control as well as precision public health measures. During the initial phase of epidemic control, the components of the model pertaining to community transmission containment and patient isolation demonstrated certain applicability in the Singapore context.

There are obvious differences compared with other models. For example, the SIDARTHE model constructed by Italian scholars focuses on distinguishing confirmed and unconfirmed cases and classifying cases according to severity to predict the epidemic trend.³⁷ In contrast, the patient flow model in this paper focuses more on the flow process of patients within medical institutions under medical resource allocation. Another example is the SEIR model, which integrates a fine-grained dynamic mobility network to simulate virus transmission and focuses on the impact of population mobility on the epidemic.³⁸ In contrast, the model in this paper focuses more on in-hospital processes and the relationship between medical resources and patient treatment. In terms of applicable conditions, foreign models are mostly constructed based on their own social systems, medical system characteristics, and population structures. The application of the SEIR model presupposes a highly market-oriented medical system and a relatively unrestricted lifestyle among the population. The SIDARTHE model, in contrast, is grounded in the distribution of medical resources and the implementation of early epidemic prevention and control measures. The model proposed in this paper is built upon China’s robust governmental coordination capacity, comprehensive primary healthcare system, and the highly cooperative nature of its society, which collectively influence the extent and effectiveness of the model’s applicability across different countries.

The Intensity of Resource Supply is Strongly Correlated with the Risk of Mortality. Critical Care Resources, Including Intensive Care Unit (ICU) Beds and Ventilators, Should Be Prioritized for Protection

The discrete event simulation results reveal variations in system responses under different resource supply scenarios. In the Expected scenario of this study, general beds were only depleted for 1 day, while ICU beds and ventilators had longer cumulative depletion days. As the resource supply scenario becomes more severe, the number of people waiting for medical resources and dying due to waiting for medical resources increases. Although mobile cabin hospitals, Huoshenshan and Leishenshan Hospitals were established and medical resources from other provinces were provided, ICU beds and ventilators in Hubei Province were in short supply from January 23 to February 21. This is because with the increase in mobile cabin hospitals, Huoshenshan, and Leishenshan Hospitals, the demand for ICU beds and ventilators increased much faster than the supply.³⁹

By observing the patient death situation in Table 2 and the patient waiting situation in Table 3, it can be found that when the resource supply situation becomes severe (Worst case), the average waiting days and average number of waiting patients increase significantly, leading to far more deaths due to waiting for resources than in the other two scenarios. However, the number of patients who died after receiving treatment is relatively low. This is because the base number of patients who can receive treatment in this scenario is small, so the number of patients who died after receiving treatment is small. Ji et al attributed the high mortality rate of COVID-19 patients in Hubei Province to the medical crowding-out and severe shortage of critical care resources caused by the rapid escalation of the early epidemic in Wuhan,⁴⁰ which overburdened the local health system and thus had an adverse impact on patient prognosis. This finding is highly consistent with the logic revealed by the simulation model, namely that “the intensity of resource supply determines the risk of mortality.” Meanwhile, this finding is consistent with the clinical observations reported by Qian et al, who noted that “competition for treatment resources among critically ill patients is more intense”.²⁸ This suggests that in the allocation of emergency medical resources, ICU beds and ventilators represent critical limiting factors influencing patient mortality and should therefore be prioritized to ensure optimal outcomes. Furthermore, this conclusion provides empirical support—particularly from the early stages of the pandemic in developing countries—for the “fair distribution of scarce medical resources” principle proposed by Emanuel et al³ This indicates the critical importance of prioritizing and scientifically allocation of essential medical resources during public health emergencies, such as epidemics to effectively reduce patient mortality risk.

The simulation results highlight the core role of critical care resources such as ICU beds and ventilators in reducing the risk of death. This is mainly due to the high dependence of critically ill patients on advanced life support equipment, and their shortage will directly lead to delayed treatment and poor prognosis. China has attached great importance to this issue at the policy level and has recently put forward clear construction goals: by the end of 2027, the number of intensive care beds nationwide should reach 18 beds per 100,000 people, and the number of convertible intensive care beds should reach 12 beds per 100,000 people.⁴¹ This means that the number of ICU beds will increase significantly in the next few years, and higher requirements will also be put forward for the configuration of related equipment. For example, the policy suggests that 1 extracorporeal membrane oxygenation (ECMO) device should be equipped for every 10 ICU beds. However, there is a huge gap between the current domestic stock of ECMO devices and the target demand, which further confirms the practicality of resource constraints in the simulation model and the urgency of prioritizing critical care resources.

Strengths and Limitations

This study integrates discrete-event simulation with patient flow theory to quantify resource exhaustion risk under multiple resource-availability scenarios using empirical early-outbreak data. Several limitations should be noted. First, the empirical window focuses on the first 30 days after lockdown and may not represent later phases when capacity expansion and clinical/reporting practices changed substantially. Second, due to the use of aggregated provincial-level data, the model simplifies certain clinical pathways (eg., no explicit ward-to-ICU escalation) and mortality structure; these assumptions may affect absolute mortality estimates but are unlikely to change the qualitative patterns of queuing and depletion that are driven primarily by admissions, length of stay, and effective capacity. Third, epidemic parameters are held constant across scenarios to isolate supply-side effects; thus, results should be interpreted as controlled comparisons rather than full epidemic forecasts. Future work can extend the framework to multi-phase settings with more granular patient-level transition and outcome data.

Conclusion

Findings indicate that the rapid surge in case numbers during the initial phase of the epidemic significantly strains regional medical resource capacity. There is a strong correlation between the adequacy of medical resource supply and patient mortality risk. Among the resources examined, ventilators—critical for treating severe cases—exhibit the most pronounced mortality amplification effect when in short supply. Furthermore, adopting an aggressive expansion strategy at the onset of the outbreak can effectively alleviate resource pressure and reduce mortality risk. Dynamic monitoring of patient waiting times and queue lengths serves as a key indicator for early warning of medical resource overcrowding. The dynamic simulation model developed in this study provides a robust analytical framework for assessing medical resource exhaustion risks during the early stages of major infectious disease outbreaks. The findings offer scientific support for optimizing public health emergency resource stockpiling strategies and enhancing the resilience of healthcare systems in responding to public health emergencies.

Funding Statement

This research was funded by the Humanities and Social Sciences Research Planning Fund Program of the Ministry of Education of China (23YJAZH166) and Health Commission of Hubei Province scientific research project (WJ2023Z009).

Data Sharing Statement

The data that support the findings of this study are available from the corresponding authors on reasonable request.

Ethics Approval and Consent to Participate

Not applicable. This study used publicly available, aggregated data reported by official sources and parameters from published literature, and did not involve individual-level identifiable information or interventions with human participants.

Author Contributions

All authors made significant contributions to the work reported, whether in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors declare that they have no competing interests in this work.