The triage protocol is a lean diagnostic tool used to improve performance and ensure the proper utilization of available resources. Not knowing the patient's condition upon arrival to allocate priority and offer initial treatment, as well as to allocate appropriate resources, has led to detrimental health conditions for the patient and cost implications for the hospital. Triaging of patients to identify and prioritize them based on the urgency of care is very relevant where demand for care exceeds resource availability [5], [6]. Triaging of patients during admission helps to classify patients based on their criticality by assigning colour codes, red, orange, yellow, and green, to determine urgency [16]. The application of triage as an admission procedure for classifying of patients based on criticality enables a reduction of patient waiting time as well as informs on the resource distribution decisions as it enables one to accurately direct patients to the right resource in the shortest time possible.

There is evident advanced and successful adoption of triage tools in improving the quality of care and patient outcomes in developed countries. These high-income countries have successfully adapted the five-level triage tool; in contrast, the same tools are not suitable for developing low-income countries due to their different epidemiology, high service demand, and limited resource arrangements [18]. Moreover, triage scale tools designed for developed, high-income countries, such as the five-level triage tools, exhibit variations in outcome as well as the degree of validation and reliability. This, therefore, makes it exceptionally difficult to predict an ideal triage tool for a particular context when implemented in low-income countries [19].

Reliable, simple, and efficient triage tools are emphasized by clinicians for developing counties’ hospital settings [20], [21]. Some triage tools developed for low-resource settings in developing countries include the Emergency Triage Assessment and Triage system. This three-level triage tool has seen extensive successful applications across the African region [22]. Other recommended tools for low and middle-income countries include Emergency Triage Assessment and Treatment (ETAT), South African Triage Scale (SATS), and Interagency Integrated Triage Tool (IITT) [23], [24]. These tools are used at the point of entry (admission) of patients into the hospital and utilize signs and symptoms rather than physiologic parameters when defining urgency. They have demonstrated acceptable reliability and validity in a variety of Accident and Emergency facilities.

Several studies have researched the implementation of triage tools in obstetric and emergency care, highlighting both their benefits and limitations. A consistent theme across the literature is that triage protocols enhance accuracy in prioritization and patient safety. Still, their impact on waiting time, LOS, and resource utilization remains underexplored.

For instance, [25] demonstrated that triage improved the accuracy of patient categorization upon arrival, while Best and Sesay [26] and Easterbrook [27] emphasized that standardized tools, such as SOTS and AOTAS, enhanced patient safety, satisfaction, and healthcare provider performance. Similarly, Hesham et al. [28] found that obstetric triage protocols improved both nurses’ knowledge and the timeliness of initial assessments. Collectively, these findings suggest that triage tools facilitate more informed decision-making and clinical judgment among staff, ultimately enhancing the quality of maternity care.

Most studies employing triage have focused on emergency departments in high-resource, urban hospitals, with few addressing rural, low-resource, or non-emergency settings such as maternity units, outpatient clinics, or surgical wards [29]. This limited contextual diversity constrains the applicability of existing findings to healthcare systems where staffing shortages, infrastructural limitations, and patient surges are common.

However, there is less agreement on the effect of triage tools on waiting times and LOS [4]. For example, adapting the Cape Triage Score system in primary healthcare settings found no reduction in waiting time or LOS, partly due to differences in survey tools and a lack of consideration for resource utilization. In contrast, Quaile [29] reported an improvement in timeliness of care from 19 minutes to 10 minutes, although their study did not extend this to overall waiting time or service flow. Likewise, both Hesham [30] and Best et al. [26] did not assess how triage protocols influenced waiting time, LOS, or resource distribution, highlighting a common gap across studies.

Research in broader emergency care contexts further illustrates these limitations. Mitchell et al. [18], [31] examined the Interagency Integrated Triage Tool in low-resource emergency departments, confirming its validity and acceptability, yet again neglecting the dimensions of waiting time, LOS, and resource utilization. A notable lack of evidence remains from African contexts, including Kenya, where hospital workflows and resource constraints differ significantly. There's minimal exploration of how triage protocols affect specialized departments, such as maternity, inpatient wards, and diagnostics [14]. This gap limits the generalizability of existing findings to resource-limited health systems, where operational realities often diverge from those in high-income settings.

Overall, while the literature supports the use of triage tools to improve patient safety, clinical accuracy, and healthcare staff performance, a clear gap remains in understanding their operational impact, specifically on waiting times, patient flow, and resource allocation. To the best of the researcher’s knowledge, no study has combined triage tools with Arena simulation modeling in MDs to address these issues. This study, therefore, seeks to bridge that gap by evaluating patient processes through simulation to optimize waiting time and resource utilization.

Data collection involved directly tracking patients from the moment they arrived in the MD until their discharge, as shown in Fig. 2. At each stage of care, information was gathered on the number of patients present, the staff available, the time patients spent waiting, and the actual service time. In addition, the hourly arrival of patients was recorded, and this arrival pattern was later used to define the patient arrival schedule within the simulation model.

The methodology proposed by Rossetti [32] was adopted to develop the Arena simulation. First, data on waiting and service times were collected, along with resources at each process stage. The data were then converted into model parameters used to build the DES model. The Arena simulation package was used for this study. The model was then verified through animation visualization and validated to ensure it accurately mimicked real-world systems. Finally, the improvement scenario, the Triage protocol, was implemented.

A simplified triage protocol was modeled in Arena using a percentage-based criticality approach. Each patient was assigned a criticality level using discrete probability distributions that reflected the expected case mix. The assigned value was stored as an attribute and used in Decide modules to route patients into priority queues, where higher acuity categories were always served first. Prioritization was achieved through queue ranking, and the model tracked waiting times from arrival to discharge to assess the impact of percentage-based triage on system performance.

Verification was conducted through animation visualization of the simulation model's flow and debugging, resolving errors such as entities not leaving the system and resources not being released. Increasing the patient’s arrival pattern implied that waiting times increased. A comparison was made between the simulation model outputs and the actual (empirical) process flow data to enhance the model's validity and ensure it accurately represents the current situation, as shown in Fig. 4.

waiting time (mins) Antenatal Cash point L/D Gynecology process stagesconsultation observed results simulation results Table 1 illustrates that the deviations for antenatal, cashpoint, and labour/delivery consultations remained under 10%, demonstrating that the simulation provided a strong representation of actual system performance. This aligns with [33], who notes that models with deviations below 10% can be considered reliable reflections of reality. In contrast, the units with deviations above this threshold were influenced by factors outside the model’s scope, such as staff temporarily leaving to handle emergencies, taking breaks, or arriving late.

The model identified critical pressure points within the maternity pathway, where patients experienced extended waiting times and resources were under considerable strain, as shown in Table 2. The most affected areas were the antenatal and gynecology units of the MCH clinic, as well as the triage room and both the standard and cesarean delivery rooms in the Labour and Delivery (L/D) clinic. Compared to other stages in the patient flow, these units consistently recorded the longest delays, underscoring their significant role as operational bottlenecks. These findings highlight the need for targeted interventions.

The uneven distribution of workload and the mismatch between patient arrival rates and available staff capacity further justified the need for targeted interventions, ranging from optimized triage protocols and staffing adjustments to workflow redesign and resource reallocation, as well as streamlining service delivery, since the imbalance between patient inflow and available resource capacity created systemic inefficiencies that prolonged waiting times and reduced overall service performance

In this study, a Design of Experiments (DOE) framework was employed to systematically evaluate the performance of different triage configurations within the MD simulation. Four key control scenarios were modelled: a baseline case without a triage protocol, the introduction of a triage protocol, a triage protocol combined with additional staffing resources, and a stress-test scenario simulating peak patient arrivals. These experimental setups enabled assessment of how triage implementation and resource adjustments influence system behavior under varying operational conditions.

Patients attending the antenatal and gynecology consultation units experienced considerable waiting times of 4.26 minutes and 3.02 minutes, respectively, in addition to an average of 5.05 minutes at the cash point. The antenatal and gynecology units demonstrated the highest levels of resource utilization, at 71% and 69%, respectively, as seen in Table 3, indicating that staff in these areas were operating near or beyond their optimal capacity. Such intensive use often results in service congestion and extended patient delays. In contrast, process stages with lower utilization rates indicate untapped capacity and uneven workload distribution, highlighting the need to redistribute resources and adjust staffing patterns to achieve balanced utilization, as well as enhance patient flow and overall service performance.

In the L/D clinic, patients faced even longer delays. The triage room recorded an average waiting time of 7.91 minutes before service, followed by an additional 2.68 minutes at the consultation stage. Similarly, patients in the standard delivery unit waited 9.50 minutes, while those in the cesarean delivery unit waited 4.11 minutes before receiving care.

L/D clinic

Waiting time (minutes) LD Triaging Normal Cs delivery Normal D. Cs consultant Delivery monitoring monitiring Process stages

Base (no triage) Triage protocol Triage (Added resource) Increased arrival (peak demand)

The initial implementation of the triage protocol did not substantially reduce waiting times under the current resource levels. However, as shown in Figs. 5 and 6, simulations indicate that increasing resources in the most congested areas yields significant improvements. Waiting times in the antenatal and gynecology units dropped from 4.26 and 3.02 minutes to 0 minutes, respectively, while consultation delays in the L/D clinic reduced from 2.68 minutes to 0 minutes, signifying a 100% reduction.

The cash point also improved, with waiting times reduced by more than half, from 5.05 minutes to 1.7 minutes. These improvements were accompanied by a better balance in resource utilization across units, where previously underutilized resources, such as the cashier, showed increased utilization rates, as seen in At optimal resource levels, the triage protocol proved instrumental in reducing waiting times across all stages of the patient process. This, in turn, led to a shorter overall LOS, decreasing from 1,118.1 minutes to 1,055.7 minutes in the L/D clinic, equivalent to a 5.6% reduction, and from 39.29 minutes to 34.52 minutes in the MCH clinic, representing a 12.1% reduction. The relatively small decrease in LOS at the MD aligns with findings by Swart et al. [4], who reported that the introduction of triage protocols alone did not significantly reduce patient LOS.