Skip to content
Back to search results

Patient Flow Simulation

Introduced by
Operations research and management science
Year
1990
Type
Discrete event simulation technique
Family
Simulation, Process modeling
Font size

What does it do?

Discrete Event Simulation (DES) is a computational technique that models the movement of patients through healthcare facilities by simulating individual patient journeys and interactions with resources (staff, beds, equipment).

DES allows realistic representation of complex, stochastic healthcare processes and supports 'what-if' analysis without disrupting live operations.

The intuition behind the method

Healthcare systems are too complex for analytical formulas alone. Patients follow different pathways depending on diagnosis and outcomes; staff have multiple roles and variable efficiency; equipment breaks down and requires maintenance. Simulation mimics these real-world complexities: it generates virtual patients, routes them through the system, and tracks outcomes (wait times, resource utilization, patient satisfaction). By running thousands of simulated scenarios, managers can explore the impact of changes before implementation.

History

Discrete event simulation emerged from operations research in the 1950s and 1960s, initially applied to manufacturing and logistics. Healthcare applications began in the 1990s with work by Pidd and colleagues. Early models focused on emergency departments and operating rooms. Modern DES practice includes agent-based extensions (modeling autonomous decision-making) and integration with optimization, allowing not just simulation but automatic search for best policies.

When to use it

Use DES when the process is complex with multiple decision points, when resource constraints create bottlenecks, or when you need to evaluate the combined effect of multiple changes. DES is most valuable for medium- to large-scale systems (emergency departments, surgical centers, full hospital networks). Avoid DES if the system is simple and analytical models suffice, if you lack sufficient data to build a credible model, or if management expects immediate decisions without the time needed for model building and validation.

Limitations and considerations

  • Requires substantial data collection and model building effort, often 3–6 months for detailed models
  • Model credibility depends on data quality and stakeholder validation; garbage in, garbage out
  • Simulationists must make many assumptions (routing rules, break times, priorities); sensitivity analysis is essential
  • Uncertainty in model parameters (e.g., process time distributions) propagates into forecast uncertainty
  • Simulation output is probabilistic; a single scenario does not guarantee a outcome, only a range

Source conditions do not replace independent scientific review or validation by an analysis engine.

Strengths

  • Models real-world complexity: patient heterogeneity, resource constraints, stochasticity, priority rules
  • Allows detailed 'what-if' analysis without disrupting operations: test multiple scenarios rapidly
  • Identifies bottlenecks visually and quantitatively, guiding improvement efforts
  • Captures system interactions that analytical models miss: congestion effects, indirect impacts of changes
  • Scalable from single units to entire health systems

Applications

  • Testing impact of staffing level changes in emergency departments on wait times and patient satisfaction
  • Evaluating surgical suite redesign, including block scheduling, turnover time reduction, and case sequencing
  • Assessing impact of adding new diagnostic equipment (CT scanner, laboratory analyzers) on throughput and revenue
  • Analyzing patient pathways during surge conditions (pandemic, major accident) to plan surge capacity and contingency staffing
  • Optimizing discharge processes to reduce length of stay while maintaining quality and safety

How it works

How it works

1

Collect detailed process data: patient arrival patterns and diagnoses, routing rules (which patients go to emergency vs. elective pathways), service times at each station, staff schedules and availability, equipment capacity and downtime. Observe actual workflows to understand informal rules and workarounds.

2

Validate data completeness, remove anomalies, fit distributions to process times. Classify patients into logical groups (e.g., emergency walk-in, scheduled surgical, ICU transfer). Document all assumptions and resource constraints.

3

Map the current process flow: from entry point (emergency reception, scheduling) through treatment areas and eventual discharge. Identify bottlenecks, resource conflicts, and wait points where improvement is needed.

4

Build a simulation model in discrete event simulation software (Arena, AnyLogic, SimPy, or custom code) that reproduces the observed process. Include patient generation, routing logic, resource allocation, and data collection points.

5

Run baseline simulations to validate the model matches current performance. Evaluate proposed changes (additional staff, new equipment, process redesign) by running alternative scenarios. Report key metrics (average wait times, resource utilization, patient throughput) and variance across replications.

Common pitfalls

  • Building a model that is too detailed, attempting to capture every variation; this becomes unmaintainable and slow
  • Using insufficient replications; if you run the simulation 10 times, results are noisy. Run at least 30–100 replications per scenario.
  • Forgetting to validate the model against current state; a model that does not match reality is unreliable for prediction
  • Presenting simulation results as certainties rather than probabilities; emphasize ranges and confidence intervals
  • Changing multiple variables at once; good practice is to vary one change at a time to isolate its effect

Frequently asked questions

How long does it take to build a patient flow simulation?

A focused model of a single department (emergency, operating room) typically takes 2–4 months. A hospital-wide model can take 6–12 months. The effort is split between data collection (40%), model building (30%), validation (20%), and scenario analysis (10%).

How many replications should I run?

Run enough replications to stabilize output variance. For simple models, 30 may suffice. For complex models with high variability, 100+ is recommended. Use sequential stopping rules: run until the confidence interval width is acceptable.

How do I know if my simulation is valid?

Compare simulation outputs to observed data (current state). Average wait times, resource utilization, and throughput should match within 10–20%. Run sensitivity analysis: if the model is sensitive to unrealistic changes, it may not capture real dynamics.

Can simulation predict the future accurately?

Simulation shows probability ranges, not certainties. Use it to compare relative impact of scenarios: 'Option A is 15% better than Option B.' Do not use it for absolute predictions without external validation.

What software should I use for patient flow simulation?

Arena and AnyLogic are industry standards with extensive healthcare templates. SimPy is free and flexible for Python programmers. Choose based on required features, IT support, and analyst expertise. Consider simulation service vendors if internal expertise is lacking.

Related methods

Sources

  1. Pidd, M. (1992). Computer Simulation in Management Science (3rd ed.). John Wiley & Sons.

    ISBN: 9780471939314
    Copied.
  2. Sokolowski, J. A., & Banks, C. M. (2009). Modeling and Simulation Fundamentals: Theoretical Underpinnings and Practical Domains. John Wiley & Sons.

    Open publication reference
    Copied.
  3. Gunal, M. M., & Pidd, M. (2010). Discrete event simulation for performance modelling in health care: A review of the literature. Journal of Simulation, 4(1), 42–51.

    Open DOI reference
    Copied.