A hybrid inventory management model for pediatric tertiary hospitals in low- and middle-income countries: Integration of ABC–XYZ classification, simulation, and Just-in-Time implementation

This study employed a multi-phase, mixed-methods design to evaluate the performance of hospital inventory management and assess the feasibility of transitioning to a JIT model. Conducted at Children’s Hospital 1, a high-volume tertiary pediatric facility in Ho Chi Minh City, Vietnam, the research covered the period from January 2019 to March 2025. The study combined retrospective data analysis, structured inventory classification, root cause exploration, simulation modeling, and prospective pilot validation. This design was chosen to balance empirical depth with contextual adaptability, ensuring that findings were operationally grounded and implementation-relevant.

Quantitative data were retrospectively extracted from the hospital’s logistics and clinical information systems. These included:

• Monthly consumption and procurement records for over 1,200 high-tech medical supply items (2019–2024),
• Stock-out incidents, procurement cycle times, urgent purchase rates, and inventory turnover metrics,
• Clinical workload data (admissions, procedures, bed occupancy), used to contextualize supply demand.

In parallel, qualitative data were collected through 17 semi-structured interviews with stakeholders across procurement, supply chain, and clinical departments. Interviews focused on identifying systemic inefficiencies, procedural barriers, and gaps in coordination. The use of both retrospective records and real-time stakeholder input ensured triangulation and enhanced the validity of identified bottlenecks.

All medical supply items were classified using a dual-layer ABC–XYZ analysis:

• ABC analysis ranked items by annual consumption value, with Class A accounting for 70%–80% of total value, Class B for 15%–25%, and Class C for the remainder.
• XYZ analysis categorized items based on demand variability using coefficient of variation (CV) thresholds, with Class X indicating low variability, Y moderate, and Z high unpredictability.

The resulting ABC–XYZ matrix informed differentiated control strategies, risk segmentation, and selection criteria for simulation inputs and pilot targets. Items with high volatility (e.g., CZ, BZ) were flagged for additional analysis to understand their operational criticality and explore alternative stocking models beyond JIT.

To identify systemic bottlenecks, a root cause analysis was conducted using a triangulated approach. Quantitative trends in key performance indicators were combined with thematic coding of interview transcripts and cross-referenced with findings from internal audit reports (2020–2023). Bottlenecks were categorized according to their origin (process, data, structure, supplier) and mapped to their operational impact. This diagnostic step established the foundation for simulation modeling, ensuring that projected scenarios were grounded in empirically observed operational challenges rather than theoretical assumptions.

A DES model was developed using Arena 16.1 to evaluate the JIT strategy under realistic constraints. Two scenarios were explored: (1) Baseline (2024), reflecting current workflows, and (2) JIT Scenario (2025), incorporating reduced cycle times (30 to 15 days), dynamic safety stocks, and VMI for high-priority items. Key input parameters are summarized in Table 1 to ensure transparency and reproducibility.

Simulation calibration was performed using historical operational data and refined through three iterative review cycles with logistics, procurement, and clinical stakeholders to ensure alignment with institutional processes. Model outputs were subsequently cross-validated against Q4/2024 performance benchmarks. Sensitivity analyses were conducted under alternative scenarios, including partial supplier compliance and conservative procurement cycle reductions (e.g., to 20 days)—to assess the robustness of projected outcomes. These analyses clarified structural constraints and strengthened the external validity of the proposed inventory reforms.

To validate simulation outputs, a pilot implementation of the JIT model was conducted in Q1/2025 within the Department of Cardiovascular Intervention. A total of 42 supply items, predominantly classified as AX and AY based on the ABC–XYZ matrix, were managed under JIT protocols. Core elements included:

• Integration of daily procedural logs with inventory systems,
• Barcode-enabled real-time tracking,
• Automated replenishment thresholds,
• VMI arrangements for selected high-volume items.

Performance during the pilot (Q1/2025) was benchmarked against Q1/2024, focusing on key indicators: stock-out incidents, urgent procurement rates, inventory days, holding costs, and procedure disruptions. In addition to quantitative metrics, structured interviews and rapid feedback sessions were conducted with clinical and logistics staff to assess operational feasibility, workload implications, and user acceptability.This dual focus on performance and perception provided a holistic assessment of both system-level impact and frontline experience, enabling a more grounded evaluation of scalability potential.

Figure 1 presents a structured overview of the study’s methodology, including the sequence of research phases, data sources, inventory classification logic, root cause diagnostics, simulation modeling setup, and pilot validation design. The visual framework illustrates the interdependencies across methods and clarifies how evidence from each phase informed the subsequent stage. Such iterative structuring reinforces methodological rigor and enhances the translational relevance of findings.

Figure 1.

Methods framework of the study

Children’s Hospital 1, a major tertiary pediatric referral center in southern Vietnam, experienced considerable fluctuations in clinical service volumes and supply chain demands between 2019 and 2024. These shifts were shaped by both pandemic-related disruptions and post-pandemic recovery surges, creating substantial strain on operational resources.

3.1.1Hospital activity

As shown in Table 2 and Figure 2, outpatient visits and inpatient admissions declined sharply during 2020–2021, reaching their lowest levels in 2021, before rebounding rapidly from 2022 onward. By 2024, service volumes had largely returned to, or exceeded, pre-pandemic levels.

Table 2.Annual hospital activity at Children’s Hospital 1 (2019–2024)

Year	Outpatient Visits	Inpatient Admissions	Bed Occupancy Rate (%)	Avg Length of Stay (days)	Total Surgeries	Cardiac Surgery	Interventional Cardiology
2019	1,836,500	95,000	116.56	6.20	18,606	450	639
2020	1,194,100	70,412	90.15	6.54	15,591	408	555
2021	556,056	37,951	50.13	6.75	10,573	358	456
2022	1,313,591	79,036	87.85	6.09	27,120	300	598
2023	1,347,168	89,705	102.50	6.26	31,154	476	810
2024	1,373,660	91,084	103.87	6.26	29,246	588	840

Figure 2.

Trends in clinical service volume and surgical complexity at Children’s Hospital 1 (2019–2024)

Despite this recovery, operational pressure remained high. Bed occupancy rates exceeded 100% in 2023–2024, indicating sustained capacity strain, while average length of stay remained relatively stable at approximately 6.26 days. Surgical activity followed a similar trajectory, with total procedures more than doubling between 2021 and 2023. High-complexity services, particularly cardiac surgery and interventional cardiology, increased steadily throughout the post-pandemic period.

These findings reflect not only service recovery but also a qualitative shift toward higher clinical intensity. This evolving demand profile provides critical context for assessing inventory performance and underscores the need for more responsive and risk-adaptive supply chain management.

The post-pandemic surge in high-complexity care placed increasing pressure on inventory coordination, particularly for specialty items, highlighting the limits of traditional bulk-based stocking policies.

3.1.2Integrated performance and operational stress metrics

To improve control over high-value medical inventories and support evidence-based decision-making, the hospital applied a dual-layer classification system combining the ABC and XYZ analytical methods. This framework was used to categorize all high-tech medical supplies managed between 2019 and 2024.

The ABC analysis ranked items by their annual consumption value. Class A items represented the top 10%–15% of items that consumed approximately 70%–80% of the total annual inventory budget. Class B and C items accounted for lower proportions of value but formed the majority in terms of item count. The XYZ analysis grouped items according to demand stability. Class X items showed regular, predictable usage; Class Y items had moderate fluctuations; and Class Z items displayed highly erratic or sporadic demand patterns. When combined, the ABC and XYZ classifications formed a risk–value matrix that enabled differentiated inventory policies for each item group. This segmentation facilitated more precise alignment between stocking strategies and consumption risk. Table 4 provides a cross-tabulation of inventory items by value (ABC) and demand variability (XYZ), revealing concentration patterns that inform risk-based stocking strategies and targeted JIT application. These bottlenecks span procurement processes, forecasting, inventory visibility, data integrity, and supplier performance, each contributing to cumulative inefficiency across the supply chain.

Although moderate improvements were observed, the system continued to exhibit several persistent inefficiencies. An integrated review of performance indicators, internal audits, and stakeholder interviews revealed structural misalignments between inventory governance and clinical realities.

Key findings include:

• Procurement delays due to rigid multi-tier approval processes.
• Forecasting inaccuracies stemming from lack of integration with clinical workload data.
• Redundant stock accumulation caused by decentralized tracking and limited oversight.
• Expired stock and reordering errors driven by manual entry and inconsistent coding.
• Low supplier accountability leading to frequent delivery lapses.

Table 5 summarizes the key system-level bottlenecks identified through triangulated analysis of performance metrics, internal audits, and qualitative feedback from logistics and clinical stakeholders.

These systemic constraints limited inventory responsiveness and contributed to recurring operational inefficiencies, particularly during demand surges.

A DES was conducted to assess the impact of transitioning to a JIT model. The simulation assumed real-time consumption tracking, recalibrated safety stocks, and VMI protocols for AX items, with procurement cycle time reduced to 15 working days.

Performance under the simulated JIT scenario demonstrated measurable efficiency gains (see Table 6 and Figure 3):

• Days of inventory decreased by 34.5%
• Stock-out rates declined by 45.8%
• Urgent procurement reduced by 57.0%
• Holding costs fell by 30.6%
• Service level improved modestly to 91.2%

Figure 3.

Inventory performance outcomes: Baseline operations (2024) versus simulated JIT scenario (2025)

Sensitivity analyses showed that even under conservative assumptions, performance improvements were preserved—though to a lesser extent—highlighting institutional readiness as a prerequisite for scalability.

To validate simulation findings, a pilot JIT model was deployed in Q1/2025 in the Cardiovascular Intervention Department. The pilot included 42 high-priority items, mostly AX and AY class, and featured:

• Real-time demand capture from procedural logs
• Barcode-enabled tracking
• Automated reorder thresholds
• Vendor-managed replenishment

Operational results mirrored the simulation (see Table 7):

• Stock-out incidents: ↓72.7%
• Urgent procurements: ↓64.5%
• Inventory days: ↓35.3%
• Holding costs: ↓28.4%
• Zero procedure delays were reported during the pilot

Qualitative feedback emphasized reduced administrative burden, enhanced coordination, and increased staff confidence in inventory reliability.These findings confirm the feasibility of hybrid ABC–XYZ + JIT strategies under real-world constraints, provided adequate institutional support and supplier compliance.

The longitudinal evidence from 2019 to 2024 demonstrates a widening structural gap between the evolving nature of clinical demand and the functional capacity of the hospital’s inventory system. At Children’s Hospital 1, demand volatility manifested not only in patient volumes but also in the acuity and supply intensity of care. While outpatient visits and inpatient admissions declined sharply during the COVID-19 period (2020–2021), the post-pandemic recovery phase was marked by a rapid rebound and subsequent overload, with bed occupancy rates exceeding 100% from 2023 onward and a substantial increase in high-complexity interventions.

In particular, interventional cardiology and surgical procedures more than doubled compared to 2021 levels, creating non-linear pressure on high-cost, short-shelf-life items.These dynamics exposed the limitations of an inventory system designed around historical averages, fixed reorder cycles, and bulk procurement logic. Although selected aggregate indicators—such as average length of stay or service level—appeared relatively stable, disaggregated analysis revealed persistent stress signals. Stock-out rates remained above 10% throughout the study period, urgent procurement peaked at 34.7% in 2021, and expired inventory approached 4% of total stock value, indicating a system that remained fundamentally reactive despite incremental improvements. Furthermore, internal audit logs indicated persistent mismatches between item consumption and ordering frequency in departments with fluctuating patient acuity, such as emergency and surgical units.

This apparent contradiction between stable macro-level indicators and deteriorating item-level performance echoes concerns raised by Ahmad et al.,^[22] who cautioned that aggregated hospital metrics often mask operational fragility at the SKU or departmental level. In the present study, this masking effect translated into delayed replenishment of critical supplies during periods of clinical surge, while surplus inventory accumulated elsewhere, particularly among low-value but erratically consumed items. For example, during peak months of 2023, departments experienced simultaneous shortages of cardiac catheters and oversupply of low-use consumables like specific feeding kits, highlighting a lack of dynamic coordination.

Unlike earlier qualitative or cross-sectional studies that identified procurement rigidity as a general bottleneck,^[23] the six-year dataset demonstrates how repeated episodes of unmodeled demand variability accumulate into systemic inefficiency.Attempts to expedite purchasing during high-demand periods were often slowed by rigid quarterly budgeting processes and lack of real-time inventory visibility, further compounding system stress.

In a pediatric tertiary context, these misalignments carry heightened risk. Short-term stock-outs of specialized devices or consumables can delay time-sensitive interventions, amplifying clinical consequences beyond financial inefficiency.

Collectively, the findings challenge the continued reliance on conventional inventory models that assume demand regularity and cost optimization through volume purchasing. Instead, they point to the necessity of inventory governance frameworks that explicitly recognize volatility as a core operational condition rather than an exception.^[8,21,23] This includes embedding risk tolerance thresholds, proactive buffer mechanisms for high-acuity items, and digital synchronization between clinical service forecasts and supply chain decisions.

The implementation of the ABC–XYZ matrix in this study marked a significant transition from undifferentiated inventory control to a structured, risk-based governance model. By jointly stratifying inventory items based on annual consumption value (ABC) and demand variability (XYZ), the matrix unveiled nuanced consumption-risk patterns that would have remained obscured under one-dimensional approaches. This dual-layer stratification provided a granular understanding of item-level behavior, facilitating more precise risk prioritization and resource allocation. Empirical results revealed that Class A items—comprising approximately 80% of total inventory value but only a small fraction of SKUs—were predominantly classified as X or Y in variability. These items demonstrated stable or moderately fluctuating usage, suggesting strong suitability for tighter control, automated replenishment, and JIT-style protocols. Yet, in prior practice, they had often been managed under uniform policies alongside lower-priority items.^[5,20] Conversely, CZ and BZ items, although low in financial impact, were disproportionately associated with demand unpredictability, stock expiry, and redundant intra-hospital distribution.^[8,19,22] Analysis of expired stock reports from 2021–2023 confirmed that over 60% of expired items belonged to CZ and BZ categories, despite their low overall budget share.

Crucially, this reinforces the need to treat low-value/high-volatility items not as negligible but as high-risk categories for waste and inefficiency, particularly in budget-constrained LMIC settings. The value of the ABC–XYZ framework in this study lies not only in its analytical clarity but in its translation into operational policy. Specific procurement thresholds, review frequencies, and stock management strategies were tailored to each matrix segment. For example, AX and AY items were assigned continuous monitoring protocols with real-time alerts and automated restocking, while CZ items were excluded from bulk procurement cycles and managed via controlled, just-in-need requisition. This represents a significant advancement from prior literature, where ABC–XYZ matrices are often limited to descriptive analysis rather than integrated into procedural governance.^[8,11,15] The hospital developed matrix-specific SOPs (Standard Operating Procedures), linking inventory review cadence and approval layers directly to classification category, thereby institutionalizing the framework into daily operations. However, a key limitation of the ABC–XYZ model is its omission of clinical criticality, perishability, and emergency-use frequency—factors crucial in high-acuity domains such as pediatric surgery and intensive care.

For instance, low-cost items like resuscitation drugs or antitoxins, though low on ABC, may carry life-or-death urgency. In future applications, the integration of multidimensional frameworks (e.g., ABC–VEN, criticality scoring) could enhance decision fidelity by aligning financial and operational risk dimensions with clinical importance.^[8,10] Preliminary mapping exercises initiated in Q4/2024 have begun incorporating VEN tags and emergency-use flags to supplement ABC–XYZ stratification, indicating institutional readiness for multi-factorial classification. Ultimately, risk-informed stratification using ABC–XYZ created the structural foundation necessary for subsequent optimization strategies. Without clear differentiation between predictable and erratic demand, any attempt to implement lean or JIT models would risk either over-buffering or dangerously thin inventory levels.^[5,12,19] By aligning classification with policy, the hospital was able to operationalize a responsive, tiered inventory governance approach suitable for the volatile demands of a tertiary pediatric institution.

Building upon the ABC–XYZ stratification, a DES model was developed to assess the operational feasibility and projected impact of transitioning to a JIT inventory strategy. Unlike idealized simulations, this model incorporated hospital-specific parameters, including historical consumption data, supplier performance variability, lead-time distributions, and procurement cycle governance. This ensured that the outputs were grounded in contextual realism rather than theoretical abstraction.The simulation was conducted using Arena software version 16.1, with sensitivity analyses performed on procurement lead time, demand fluctuation, and service-level targets to validate robustness of results.

The simulated results revealed considerable performance gains. Relative to the 2024 baseline, stock-out rates were projected to decline by 45.8%, average inventory days reduced by 34.5%, and urgent procurement requests decreased by over half. Holding costs dropped by approximately 30%, and service levels showed a modest increase.These improvements suggest not only increased efficiency but also enhanced system responsiveness under constrained resource settings. However, these improvements were not evenly distributed across item classes. AX-class items—those with high financial value and stable demand—performed exceptionally well under JIT conditions,^[12,15] with reduced stock volatility and predictable replenishment cycles.

In contrast, items categorized as BZ or CZ—characterized by low value but high variability—remained vulnerable to disruptions. The simulation emphasized that for these classes, JIT must be complemented by auxiliary strategies, such as buffer stock reserves, consignment models, or multi-source contracts.^[5,8] Notably, BZ items showed a simulated increase in stock-out events when modeled under full JIT without safety stock buffers, reaffirming the importance of differentiated policy application. A critical modeling assumption involved reducing the procurement cycle time from 30 to 15 days-an improvement that hinges on broader system reforms, including enhanced digital integration, streamlined approval hierarchies, and supplier performance accountability. This highlights that the success of JIT is not merely a matter of inventory mathematics but is contingent on structural and governance transformation.^{[13,18,21,23]}

During scenario testing, the model demonstrated that without at least a 30% reduction in cycle time, service level gains plateaued and urgent orders remained above 15%, reinforcing the dependency on institutional reform. To validate the model and manage risk, the pilot implementation deliberately focused on AX and AY items, which inherently carry lower operational uncertainty. This targeted selection aligns with risk management best practices,^[7,8,9] but it also introduces an element of selection bias. The findings should therefore be interpreted as proof-of-concept for phased application, beginning with predictable items and gradually expanding toward more complex and variable categories. Moreover, the simulation did not incorporate labor availability constraints, which may influence real-world responsiveness during scale-up. Future modeling iterations should consider workforce dynamics, warehousing limitations, and supplier behavior in stressed contexts.

The Q1/2025 pilot implementation in interventional cardiovascular supplies provided critical real-world validation of the simulation findings. By focusing on a high-risk, high-value category dominated by AX and AY items, the pilot tested JIT principles under operational conditions rather than theoretical assumptions.The pilot selection process used a multi-criteria prioritization tool that considered historical demand stability, stock-out frequency, and financial value to identify suitable items for initial JIT rollout. The observed improvements were substantial. Stock-out incidents declined by 72.7%, urgent procurement requests by 64.5%, and average days of inventory by over 35%, while holding costs decreased by nearly 30%. Importantly, no procedures were delayed due to supply unavailability during the pilot period. These results underscore not only quantitative performance gains but also a high degree of implementation feasibility within constrained environments.

Additionally, cycle time for procurement approvals within the pilot department was reduced by 40%, largely due to delegation of authority and digitized requisition workflows. Beyond quantitative gains, qualitative feedback from clinicians and logistics staff highlighted improved predictability, reduced administrative burden, and greater confidence in the supply system.^[8,11,17] These behavioral dimensions are critical, as trust in inventory reliability often determines whether clinical teams adhere to formal procurement pathways or resort to informal workarounds. Interviews with clinical leads indicated that improved visibility into item availability through integrated dashboards reduced pre-procedure uncertainty and eliminated informal stock hoarding practices.Such behavioral shifts are essential enablers of system transformation, but often overlooked in quantitative evaluations.

At the same time, the pilot revealed practical nuances. Certain items required micro-buffering due to clustering of procedures, and supplier performance variability necessitated closer monitoring. These observations informed the refinement of a hybrid model, blending JIT with selective redundancy—a pragmatic adjustment that enhances resilience without reverting to blanket overstocking.^[19,24] As a result, the hospital developed a contingency protocol defining buffer thresholds for procedure-linked items based on historical clustering data, reviewed quarterly.

Despite the encouraging results observed in both the simulation and pilot phases, the transition toward full-scale implementation of a hybrid inventory management model remains constrained by persistent system-level barriers. These constraints extend beyond technical limitations and are deeply embedded within institutional structures, governance arrangements, and organizational norms that shape procurement and logistics practices. A primary obstacle lies in procurement rigidity. Fixed quarterly budgeting cycles and multi-tiered approval processes continue to restrict the hospital’s ability to respond dynamically to fluctuations in clinical demand.^[22,24] Such procedural inflexibility prolongs replenishment lead times and weakens alignment between inventory decisions and real-time service requirements. Even when analytical tools indicate imminent shortages or excesses, corrective action is often delayed by administrative procedures. This structural lag undermines the potential efficiency gains projected under JIT scenarios. Equally consequential is the fragmentation of information systems. Persistent data silos between inventory management, procurement units, and clinical departments limit visibility across the supply chain and compromise forecasting accuracy.^[13,17,22] In practice, procurement decisions are frequently based on partial or outdated data, reinforcing reactive purchasing behaviors and increasing reliance on urgent procurement. Supplier performance management represents a further systemic vulnerability. The absence of formalized monitoring mechanisms, performance benchmarking, and enforceable accountability frameworks allows external variability—such as delayed deliveries or incomplete order fulfillment—to propagate unchecked into internal operations. Such instability poses particular risks for JIT-based systems, which rely fundamentally on predictable and reliable supply streams.^[1,18,25] Organizational culture also exerts a powerful influence. Long-standing reliance on buffer stock—while sometimes operationally necessary—reflects not only logistical risk mitigation but also institutional memory shaped by prior supply failures. Over time, this experience fosters risk-averse procurement norms, in which lean approaches such as JIT are perceived as unsafe or impractical.

Although the pilot intervention contributed to rebuilding confidence in tighter inventory cycles, scaling such reforms will require deliberate and sustained change management. Importantly, the findings do not support the wholesale replacement of conventional inventory models with JIT. Rather, they point toward the value of a strategically hybrid approach. JIT principles appear most appropriate for items that combine high clinical criticality with relatively stable demand profiles, while buffers or redundancy remain necessary for items characterized by high variability or supplier uncertainty.^[3,8,14,26] Such selective integration allows hospitals to balance efficiency gains with resilience.

In pediatric tertiary care settings, the consequences of inventory failure are particularly severe. Clinical areas such as interventional cardiology or intensive care demand near-absolute supply reliability, as even short-term stock-outs may delay life-saving interventions. In this study, these risks were mitigated through embedded safeguards, including micro-buffering of time-sensitive items, integration of procedural logs with real-time inventory dashboards, and predefined fallback protocols for supplier non-fulfillment. While these measures enabled efficiency improvements without compromising patient safety, their sustainability depends on disciplined execution, continuous performance monitoring, and clearly defined cross-functional accountability.^[2,16,20] Ultimately, the hybrid inventory governance model advanced in this study offers a pragmatic pathway for reform in LMIC hospital systems—balancing cost-efficiency with clinical responsiveness under real-world institutional constraints.