Potentially Avoidable Pediatric Transfers, Tertiary-Care Utilization, and Regional Care Inequities in Greece: A Retrospective Analysis of Transfer Pathways and Costs

1. Introduction

Regionalized pediatric healthcare systems depend on timely transfer of critically ill or specialized children to tertiary referral centres capable of providing advanced diagnostics, subspecialty consultation, intensive care, and definitive management [1,2,3]. These systems are particularly important in geographically fragmented healthcare settings, where disparities in pediatric expertise and resource availability may substantially influence healthcare utilization and transport-related expenditure [4,5,6].

Greece presents unique challenges for pediatric referral-network organization because of its mainland–island geography, mountainous terrain, seasonal population fluctuations, and unequal regional distribution of pediatric subspecialty services [7,8,9]. Many remote island and rural regions lack continuous access to pediatric intensive care, pediatric surgery, or advanced imaging, frequently necessitating transfer to major referral centres in Athens [10,11].

International studies suggest that a proportion of pediatric interfacility transfers may be avoidable, particularly among children requiring minimal intervention or observation after transfer [12,13,14,15,16]. However, transfer appropriateness represents only one dimension of referral-system performance. Transfer pathways may also reflect regional disparities in pediatric expertise, local resource limitations, and structural inequities in access to specialized care. In geographically dispersed healthcare systems, transport modality and referral-network organization may substantially influence both healthcare utilization and system-level expenditure [17,18,19].

Importantly, prehospital emergency medical service (EMS) transports and interfacility transfers represent fundamentally different healthcare processes. Prehospital EMS transports primarily reflect emergency access patterns and acute illness burden, whereas interfacility transfers more directly reflect referral-network function and tertiary-care dependency [20,21]. Failure to distinguish between these pathways may lead to inaccurate estimation of referral burden and healthcare utilization.

Despite increasing international attention, data regarding pediatric transfer pathways, tertiary-care utilization, potentially avoidable transfers, and transport-related expenditure in Greece remain limited. Accordingly, the present study aimed to evaluate pediatric transfers to a tertiary children’s hospital in Athens, focusing on transfer pathways, tertiary-care utilization, geographic referral patterns, potentially avoidable transfers, and associated transport costs within a geographically fragmented healthcare system.

2. Methods

2.1. Study Design and Setting

This retrospective observational study analyzed all pediatric transfers to Aghia Sophia Children’s Hospital, a tertiary academic children’s hospital in Athens, Greece, and a major national referral center for specialized pediatric care. The study included all children aged 0–16 years who were transferred between December 1st, 2022, and August 31st, 2023. No exclusion criteria were applied, regardless of diagnosis, transfer mode, or clinical condition, in order to provide a comprehensive overview of real-world transfer patterns and minimize selection bias, apart from cases with incomplete key classification variables, as described below.

2.2. Data Collection and Variables

Data were extracted from the hospital’s electronic medical records and transfer logs and, when necessary, supplemented by National Emergency Medical Service (EKAB) dispatch records to verify transport details. Demographic data for the Greek pediatric population were obtained from the Hellenic Statistical Authority (ELSTAT) (https://www.statistics.gr/en/home/).

The following variables were collected for each patient:

• Demographic characteristics: age (in years or months for infants) and sex.
• Clinical information: primary diagnosis or reason for transfer as documented at hospital arrival. Diagnoses were subsequently grouped into predefined clinically relevant categories according to the principal presenting condition: respiratory, neurologic, gastrointestinal, trauma, surgical, and other. The “other” category comprised low-frequency presentations that did not fit the major diagnostic groups, including minor dermatologic, ophthalmologic, musculoskeletal, genitourinary, and nonspecific complaints. Because individual subcategories had small sample sizes, they were combined into a single category to preserve analytical stability. Classification was performed by the research team using predefined grouping rules and clinical judgment rather than a formal coding system (e.g., ICD), reflecting real-world documentation practices. The same diagnostic classification scheme was used throughout all descriptive and multivariable analyses.
• Transport characteristics: mode of transport (ground, air, or sea), geographic origin, and estimated transfer cost. Costs included fuel, personnel, and operational expenses obtained from hospital financial records and/or EKAB and therefore represent mode-based service costs rather than patient-level micro-costing. Economic analyses focused primarily on cost concentration and the burden associated with high-cost transport modalities.
• Geographic origin: the broader region from which the child was transferred (Attica/urban, mainland/provincial, or island region). Descriptive analyses used these broad geographic categories, whereas cost analyses additionally incorporated Greek health authority regions to reflect the administrative framework used for transport expenditure recording. For multivariable analyses, geographic origin was collapsed into three categories (Attica, Aegean, and Mainland/Other) to reduce sparse cells and improve model stability.
• Transfer-source origin: the immediate site from which transport was initiated (patient’s home or incident scene, hospital, or health centre), distinguishing prehospital EMS transports from interfacility transfers. Because geographic origin and transfer-source origin represent different constructs (location versus referral pathway), both variables were retained in the analyses.
• Clinical outcomes: ICU admission upon arrival, length of hospital stay, in-hospital mortality, and final disposition (e.g., discharge, ward admission, or transfer to a surgical or subspecialty service).

All data were de-identified before analysis to ensure patient confidentiality. Data collection was performed by trained research personnel. Transfers were classified a priori into two mutually exclusive categories according to the recorded place of origin. Prehospital EMS transports originated from the patient’s home or incident scene, whereas interfacility transfers originated from another hospital or health centre. Patients were additionally stratified by predefined developmental age groups (neonates, infants, preschool children, school-age children, and adolescents) and by geographic region (urban, mainland, or island) to facilitate subgroup analyses of referral patterns and associated resource utilization.

2.3. Definition of Potentially Avoidable Transfers

For the purposes of this study, potentially avoidable pediatric transfers were operationally defined using a predefined composite endpoint based on post-transfer emergency department (ED) resource utilization rather than ED discharge alone. A transfer was classified as potentially avoidable if all of the following criteria were met: (1) no tertiary hospital-level care/admission following ED evaluation; (2) no intensive care unit (ICU) admission; (3) no advanced diagnostic or therapeutic intervention requiring tertiary-level resources, defined as the absence of invasive procedures or surgery, advanced imaging (CT or MRI), or subspecialty procedural intervention; and (4) no documented escalation of care during ED or short-stay observation beyond routine clinical assessment, basic investigations, monitoring, and supportive treatment. Transfers resulting in ED discharge after advanced diagnostics, specialist evaluation, or extended monitoring were not automatically classified as avoidable, acknowledging that tertiary expertise may still be required even when hospitalization is ultimately unnecessary. Classification was conducted retrospectively by the study investigators through chart review using these predefined criteria. Because transfer origin and clinical course were part of the medical record, adjudicators were not blinded to transfer type. Formal inter-rater reliability was not assessed. This definition was intentionally conservative to reduce misclassification and to align with prior literature emphasising post-transfer resource use as a pragmatic proxy for transfer necessity. Accordingly, this construct was treated as a resource-based and exploratory indicator of transfer necessity rather than a definitive measure of clinical appropriateness.

2.4. Statistical Analysis

Data was analysed using IBM SPSS Statistics Version 25 and R Version 4.0. Continuous variables are presented as means with standard deviations or medians with interquartile ranges, as appropriate. Categorical variables are reported as frequencies and percentages. Group comparisons were performed using χ² or Fisher’s exact tests for categorical variables and one-way ANOVA or non-parametric equivalents for continuous variables, as appropriate.

Post-transfer outcome was analysed using a revised binary endpoint: tertiary hospital-level care/admission versus ED/short-stay discharge. ED/short-stay discharge included direct discharge from the emergency department and short-stay observation (≤8 hours) without escalation to inpatient, ICU, surgical, or subspecialty-level tertiary care. Tertiary hospital-level care/admission included ward admission, ICU-coded destination, surgical or subspecialty routing, and other non-ED tertiary-care dispositions. This binary endpoint was selected to avoid ambiguity between conventional ward admission and other clinically meaningful tertiary-care dispositions.

Transfer pathway was analysed as the main explanatory variable. Prehospital EMS transports were defined as transports from home or the incident scene, whereas interfacility transfers were defined as transfers from another hospital or health centre. The association between transfer pathway and tertiary hospital-level care/admission was assessed using logistic regression, with prehospital EMS transport as the reference category. Results are reported as odds ratios with 95% confidence intervals and p-values.

Because sparse data and unstable estimates affected several covariate strata in fully adjusted multivariable models, particularly when diagnosis, transport mode, geographic region, and transfer origin were cross-classified, the revised analysis focused on transparent pathway-level associations rather than an over-parameterized predictive model. Accordingly, Table 6 presents the primary association between transfer type and tertiary hospital-level care/admission.

Potentially avoidable transfers were defined using the predefined conservative composite endpoint described above and were reported overall and by transfer pathway. ED or short-stay discharge alone was not considered sufficient to classify a transfer as potentially avoidable.

Transfer costs were analysed descriptively because costs were highly right-skewed and driven predominantly by transport modality. Total, mean, median, and proportional costs were calculated by transport mode and health authority region. Sensitivity analyses assessed the robustness of cost concentration under alternative unit-cost assumptions.

Statistical significance was set at p < 0.05. Missing data were assessed for all variables. Given the low proportion of missing values, a complete-case analysis was performed for analyses that required those variables.

2.5. Ethical Considerations

This study was reviewed and approved by the Institutional Review Board/Ethics Committee of Aghia Sophia Children’s Hospital (Approval Code:12385/02.06.2023). Given the study’s retrospective, observational nature, informed consent was waived. All procedures adhered to the ethical standards of the Declaration of Helsinki and national guidelines for biomedical research. To protect patient confidentiality, all data were anonymized before analysis, with unique identifiers replaced by coded IDs. Strict data protection protocols were followed throughout data handling and storage. Findings were reported objectively, and all efforts were made to ensure data accuracy, transparency, and integrity.

3. Results

3.1. Demographic Characteristics

A total of 423 pediatric patients were transferred to the tertiary hospital during the 9-month study period: 94/423 (22.2%) between December and February, 135/423 (31.9%) between March and May, and 194/423 (45.9%) between June and August. The mean age was 7.9 ± 5.3 years (range: 0–16 years), with infants (<1 year) comprising 11.3% of transfers, toddlers (1 to <4 years) 25.2%, preschoolers (4 to <6 years) 11.6%, school-age children (6 to <12 years) 23.1%, and adolescents (12–16 years) 28.8%. Males constituted 52.1% of the cohort. No significant sex difference was observed across age categories (χ² = 2.13, df = 4, p = 0.71). Adolescents represented the largest transfer subgroup (28.8%), followed by toddlers aged 1 to <4 years (25.2%). Boys were slightly overrepresented among trauma transfers in the 6–16-year age range (p = 0.048).

Of the 423 pediatric transfers analysed, 216/423 (51.1%) were prehospital EMS transports from home or the incident scene and 207/423 (48.9%) were interfacility transfers. The interfacility cohort comprised 183 transfers from another hospital and 24 from a health centre.

Using the predefined binary post-transfer endpoint, 199/423 children (47.0%) were classified as ED/short-stay discharges, while 224/423 (53.0%) required tertiary hospital-level care/admission. This endpoint was used to describe post-transfer resource utilization and should be distinguished from the stricter potentially avoidable-transfer definition. ED/short-stay discharge alone was not considered sufficient to classify a transfer as potentially avoidable.

Among patients with a documented hospital stay, length of stay did not differ significantly by season. Median (IQR) length of stay was 3.0 days (2.0–5.25) in winter, 3.0 days (2.0–5.0) in spring, and 2.5 days (1.0–6.0) in summer (Kruskal–Wallis p = 0.754). Regarding the place of transfer, 216/423 (51.1%) were directly from home, 183/423 (43.3%) from a secondary pediatric hospital, and 24/423 (5.7%) from a peripheral primary health center. Most transported children were from Athens [248/423 (58.6%)], followed by central Greece [66/423 (15.8%)], southern Greece [52/423 (12.3%)], islands [49/423 (11.6%)], and northern Greece [8/423 (1.7%)].

Figure 1. Flow diagram of pediatric transfers included in the study. The final analytic cohort included 423 pediatric transfers, classified as prehospital EMS transports (n=216) or interfacility transfers (n=207). Outcomes were reclassified into a binary endpoint: ED/short-stay discharge (n=199) versus tertiary hospital-level care/admission (n=224). ED/short-stay discharge included direct ED discharge and short-stay observation without escalation to tertiary inpatient or subspecialty-level care. Tertiary hospital-level care/admission included ward admission, ICU-coded destination, surgical/subspecialty routing, and other non-ED tertiary-care dispositions. Potentially avoidable transfers (n=54) were identified using predefined conservative composite criteria based on post-transfer resource utilization.

Figure 1. Flow diagram of pediatric transfers included in the study. The final analytic cohort included 423 pediatric transfers, classified as prehospital EMS transports (n=216) or interfacility transfers (n=207). Outcomes were reclassified into a binary endpoint: ED/short-stay discharge (n=199) versus tertiary hospital-level care/admission (n=224). ED/short-stay discharge included direct ED discharge and short-stay observation without escalation to tertiary inpatient or subspecialty-level care. Tertiary hospital-level care/admission included ward admission, ICU-coded destination, surgical/subspecialty routing, and other non-ED tertiary-care dispositions. Potentially avoidable transfers (n=54) were identified using predefined conservative composite criteria based on post-transfer resource utilization.

3.2. Transfer Origin and Mode of Transport

In total, ground ambulances were the predominant mode of transport (87.9%), followed by air ambulance (9.7%) and boat (2.4%). Transport mode was significantly associated with geographic region (χ² = 89.3, df = 8, p < 0.001), with air and sea transfers being more frequent in remote or island areas.

By contrast, interfacility transfers were predominantly conducted by ground ambulance (77.8%), followed by air ambulance (19.3%) and boat ambulance (2.9%), whereas the majority of prehospital EMS transports used ground ambulance (97.7%).

3.3. Regional and Seasonal Transfer Rates

Transfer rates varied significantly by geographical region, with the highest rates observed in southern Greece and the lowest in northern regions (Table 1). Table 1 summarizes transfers according to broad geographical regions for epidemiological rate estimation, whereas Table 4 summarizes transfers according to Greek health authority divisions for economic analysis. These classification systems are not identical; therefore, transfer counts are not directly comparable across the two tables. For example, Athens/Piraeus is presented as a single geographical category in Table 1, while corresponding cases are distributed across Attika-Athens and Piraeus-Aegean islands in Table 4.

Infants had the highest transfer rate from the islands (0.72/1,000), nearly double that of infants in Athens (0.39/1,000), underscoring the limited neonatal and infant care capacity in remote regions. Conversely, children aged 1 to <4 years had a higher transfer rate in Athens (0.72/1,000), likely reflecting overuse of emergency services for common pediatric conditions in an urban context. Preschool-aged children (4 to <6 years) from the islands had a markedly lower transfer rate (0.14/1,000), suggesting either under-recognition of medical needs or barriers to access.

Transfers peaked in summer (45.9%), compared to spring (31.9%) and winter (22.2%). This seasonal trend was statistically significant (χ² = 27.4, df = 2, p < 0.001), likely reflecting increased pediatric trauma, tourism, and temperature-related illnesses. Interpretation of regional transfer rates during summer should consider that ELSTAT population estimates reflect permanent residential populations and do not capture temporary increases associated with tourism and seasonal migration. Consequently, transfer rates for island and southern regions during periods of peak population influx may be modestly overestimated relative to the true at-risk pediatric population.

3.4. Diagnostic Categories and Tertiary Hospital-Level Care

Respiratory conditions were the most common diagnostic category, followed by neurologic, trauma, gastrointestinal, surgical, and other conditions, according to the predefined diagnostic classification scheme, which was applied consistently across descriptive and inferential analyses (Table 2).

Using the binary endpoint, 224/423 children (53.0%) required tertiary hospital-level care/admission, while 199/423 (47.0%) were managed as ED/short-stay discharges. Tertiary-care utilization differed significantly according to diagnostic category (χ² = 92.4, df = 5, p < 0.001), with the highest rates observed among surgical and neurologic conditions. Respiratory conditions also frequently required tertiary hospital-level care, whereas gastrointestinal, trauma, and miscellaneous low-acuity presentations were more often managed without escalation to inpatient or subspecialty-level tertiary care.

When outcomes were examined according to transfer pathway, tertiary hospital-level care/admission occurred in 143/207 interfacility transfers (69.1%) compared with 81/216 prehospital EMS transports (37.5%), supporting the higher-acuity and referral-driven nature of interfacility transfers.

Age significantly affected diagnostic distribution (p < 0.001). Infants more commonly presented with respiratory disease, whereas adolescents had higher frequencies of trauma and surgical conditions. Sex was not associated with diagnostic pattern (p = 0.74). The same predefined diagnostic categories were used consistently across all descriptive and regression analyses.

Tertiary-care utilization differed significantly according to diagnostic category (χ² = 92.4, df = 5, p < 0.001). Neurologic (88.9%) and surgical (90.0%) conditions demonstrated the highest rates of tertiary hospital-level care/admission, followed by respiratory conditions (70.3%). In contrast, trauma (44.6%), gastrointestinal (43.8%), and miscellaneous low-acuity presentations (10.3%) were more frequently managed as ED/short-stay discharges (Table 2).

3.5. Potentially Avoidable Transfers

Applying the predefined conservative composite criteria, 54/423 transfers (12.8%) were classified as potentially avoidable. These cases represented a subset of children who did not require tertiary hospital-level care, ICU admission, advanced imaging, invasive procedures, surgery, subspecialty procedural intervention, or escalation beyond routine assessment and supportive care. Among the 199 ED/short-stay discharge cases, only 54 met the stricter potentially avoidable-transfer definition.

Potentially avoidable transfers were most commonly associated with minor trauma, uncomplicated gastrointestinal illness, mild respiratory symptoms, and nonspecific complaints. They originated predominantly from urban settings, particularly Athens (78.3%), and were transported almost exclusively by ground ambulance (95%). Adolescents aged 12–16 years represented the largest subgroup (40%), consistent with the lower-acuity profile of these presentations.

Of the 54 potentially avoidable transfers, 30 (55.6%) occurred among interfacility transfers and 24 (44.4%) among prehospital EMS transports, indicating that potentially avoidable transfers arose through both referral pathways, with a slight predominance among interfacility referrals. Because classification was based on retrospective chart review using predefined resource-utilization criteria, these estimates should be interpreted as pragmatic indicators of transfer appropriateness rather than definitive measures of clinical necessity.

3.6. Distribution and Determinants of Transfer Costs

The mean cost of pediatric inter-hospital transfers during the study period was €745.56 (SD: €2,063.87), with a median of €15.00, indicating a right-skewed distribution driven by a small number of high-cost events. This large discrepancy between the mean and median reflects the presence of a small number of extremely costly transfers, primarily air and sea transports, that disproportionately increase the average cost, while the vast majority of transfers are low-cost ground ambulance journeys. In practical terms, this means that overall expenditure is driven by a limited number of high-cost events rather than typical transfers, highlighting that a small number of high-cost transfers account for a disproportionate share of expenditure.

Cost distribution by transport mode and region, as well as sensitivity analyses, are presented in Table 3, Table 4 and Table 5. Total transport expenditure for the study cohort was €315,370. Cost allocation was overwhelmingly driven by transport modality: air ambulance transfers alone accounted for €280,015 (88.79% of total costs), while boat transfers contributed €26,015 (8.25%). In contrast, ground ambulance transfers, despite representing the vast majority of cases (372/423, 87.9%), accounted for only €9,340 (2.96%) of total expenditure.

This concentration of costs indicates that overall system expenditure is determined primarily by a small subset of high-cost transfers rather than by the volume of routine transports. Sensitivity analyses (Table 5) confirmed the robustness of this pattern, demonstrating that even substantial variation in unit costs (±20%) does not materially alter the proportional dominance of air and sea transport. Scenario-specific adjustments affecting only high-cost modalities further amplified this concentration, whereas modifications to ground transport costs had a limited impact on total expenditure despite increasing their proportional share. These findings support interpretation of the economic analysis primarily as a descriptive assessment of cost concentration rather than as evidence of finely resolved patient-level cost determinants.

3.7. Structural Drivers of Transfer Costs

Transfer costs demonstrated a highly right-skewed distribution and were driven predominantly by transport modality. Air and sea transfers accounted for a disproportionately large share of total expenditure despite representing a minority of transfers. Supplementary regression analyses suggested that younger age and transfer characteristics associated with geographically remote cases were linked to increased transport costs. In adjusted models, transfers originating from the Mainland/Other region incurred approximately 2.4-fold higher mean costs than transfers from Attica, whereas estimates for transfers originating from health centres were not reliably estimable because of sparse observations (Tables S1–S3). Although transfers from the Piraeus-Aegean region accounted for most absolute expenditure because of their reliance on air and sea transport, the regression-based geometric mean cost ratios describe average per-transfer costs. Accordingly, the higher Mainland/Other versus Attica cost ratio reflects increased mean per-transfer costs relative to urban transfers and should not be interpreted as inconsistent with the observed concentration of total expenditure in island regions.

3.8. Cost Concentration Analysis

A Pareto-style cost-concentration analysis demonstrated marked concentration of transport-related expenditure among a small subset of transfers (Figure 2). Air and sea transfers represented only 51/423 transfers (12.1%) but accounted for 97.0% of total transport expenditure. Furthermore, the highest-cost 10% of transfers accounted for 91.6% of total transport costs, confirming that overall expenditure was driven predominantly by a limited number of high-cost air and sea transports rather than routine ground ambulance transfers.

3.9. Association Between Transfer Type and Tertiary Hospital-Level Care

The primary outcome for revised pathway-level analysis was tertiary hospital-level care/admission versus ED/short-stay discharge. This endpoint was selected to avoid ambiguity between conventional ward admission, short-stay observation, and other non-ED tertiary-care dispositions.

Interfacility transfers were substantially more likely than prehospital EMS transports to require tertiary hospital-level care/admission. Tertiary-care use occurred in 143/207 interfacility transfers (69.1%) compared with 81/216 prehospital EMS transports (37.5%), corresponding to an odds ratio of 3.72 (95% CI 2.49–5.57; p<0.001) (Table 6).

3.10. Interfacility Transfers

Within the interfacility cohort, 183/207 transfers originated from another hospital and 24/207 from a health centre. Using the revised binary outcome, 143/207 interfacility transfers (69.1%) required tertiary hospital-level care/admission, compared with 81/216 prehospital EMS transports (37.5%). Conversely, ED/short-stay discharge occurred in 64/207 interfacility transfers (30.9%) and 135/216 prehospital EMS transports (62.5%).

Potentially avoidable transfers were also reported by transfer type. Of the 54/423 transfers meeting the predefined composite avoidability criteria, 30/207 (14.5%) occurred among interfacility transfers and 24/216 (11.1%) among prehospital EMS transports.

Transport mode differed markedly by transfer type. Interfacility transfers were conducted by ground ambulance in 161/207 cases (77.8%), air ambulance in 40/207 cases (19.3%), and by boat ambulance in 6/207 cases (2.9%). In contrast, prehospital EMS transport was conducted almost exclusively by ground ambulance.

Costs remained highly concentrated in the interfacility cohort, which accounted for €295,850/€315,370 (93.8%) of total transport expenditure. This supports the interpretation that interfacility transfers, although representing approximately half of transfers, accounted for most tertiary-care use and nearly all transport-related expenditure.

4. Discussion

To our knowledge, this is the first Greek study to simultaneously examine transfer appropriateness, tertiary-care utilization, and transport-related costs within a geographically fragmented pediatric healthcare system. Three key findings emerge. First, most transfers did not meet the predefined criteria for potential avoidability, with only a small proportion (approximately 13%) classified as potentially avoidable under a conservative definition based on post-transfer resource utilization. Second, interfacility transfers represent the primary system burden, as they are strongly associated with tertiary hospital-level care/admission and account for a disproportionate share of total costs, consistent with previous studies [12,14]. Third, overall expenditure is highly concentrated in a small number of air and sea transfers, indicating that transport modality—rather than transfer volume—is the dominant driver of costs.

A central observation is that children are transferred from both nearby and distant regions, including cases that could potentially be managed at lower levels of care. Emergency department discharge alone is not a reliable indicator of unnecessary transfer, as only a subset of cases met strict criteria for avoidability. This supports the use of post-transfer resource utilization as a more clinically meaningful proxy for the necessity of transfer [12]. In line with international literature, approximately 10–30% of pediatric transfers—particularly for conditions such as abdominal pain and minor trauma—may be avoidable [12,15,28]. Because avoidability was classified retrospectively using resource-utilization criteria, it should be interpreted as a pragmatic system-level indicator rather than a definitive measure of clinical necessity.

Potentially avoidable transfers were predominantly concentrated in urban settings, particularly Athens, and were generally associated with lower-acuity conditions. This pattern may reflect differences in healthcare access, system navigation, and caregiver decision-making. Previous studies have similarly shown that non-urgent ambulance use in children is associated with limited access to primary care, parental anxiety, and socioeconomic factors [17,18,19,20].

In contrast, transfers from island and remote regions were less frequent but substantially more resource-intensive and typically involved patients with higher clinical acuity. These findings likely reflect structural limitations in regional healthcare capacity rather than inefficiencies in clinical decision-making. Comparable patterns have been reported in other geographically dispersed healthcare systems, where long-distance transfers are both necessary and resource intensive [14].

Distinguishing between prehospital EMS transports and interfacility transfers is essential for accurate interpretation. These pathways reflect different underlying processes within the healthcare system. Interfacility transfers are closely linked to referral pathways, regional care capacity, and clinical decision-making in peripheral facilities, whereas prehospital EMS transports more often involve lower-acuity cases and appear to be influenced by access barriers and caregiver behaviour. This distinction has been emphasised in prior research as critical for evaluating healthcare utilization and system performance [14,17,18], and similar patterns have been observed internationally.

Transfer decisions are inherently multifactorial and influenced by clinical judgment, local resource availability, transport logistics, and organizational factors that cannot be fully captured retrospectively. In addition, sparse data and unstable estimates in selected covariate strata limited the precision of some regression estimates.

From an economic perspective, transport-related expenditure demonstrated extreme concentration within a small subset of high-cost transfers. Air and sea transport accounted for nearly all expenditure despite representing a minority of transfers, whereas routine ground ambulance transfers contributed minimally to total costs. The Pareto-style cost-concentration analysis further demonstrated that the highest-cost 10% of transfers accounted for more than 90% of total transport expenditure. These findings suggest that transport modality and geographic constraints represent the principal structural drivers of system-level costs within pediatric transfer networks. Although island transfers generated most total expenditure because of a small number of extremely costly air and sea evacuations, regression analyses suggested that transfers originating outside Attica generally incurred substantially higher average per-transfer costs than urban transfers.

These findings should be considered within the broader context of the Greek healthcare system. Persistent structural challenges—including underdeveloped primary care, limited pediatric capacity in peripheral hospitals, and the absence of standardized referral pathways—likely contribute to both unnecessary transfers and overreliance on tertiary care [4,22,23]. Geographic disparities further exacerbate these issues, as healthcare resources and specialist workforce are concentrated in urban centers, particularly in Attica, while island and rural regions face significant shortages in infrastructure and personnel [24,25]. As a result, peripheral facilities are often unable to manage complex pediatric cases, necessitating transfer even when clinically appropriate [26].

International comparisons provide additional context. In the United States, respiratory and neurologic conditions remain leading causes of pediatric admissions, while improvements in outpatient care and coordination have contributed to reductions in hospitalization rates [27,28]. Similarly, studies from North America and Europe report that 10–30% of pediatric interfacility transfers may be avoidable, particularly for low-acuity conditions requiring minimal intervention [12,15,16]. In contrast, healthcare systems with well-developed primary care infrastructure, structured referral pathways, and integrated regional networks—such as those in Northern Europe—tend to demonstrate lower rates of unnecessary transfers and more efficient resource utilization [22]. These systems emphasize gatekeeping, standardized triage, and access to specialist consultation, including telemedicine support.

These findings represent observations from a single tertiary referral center and provide insight into pediatric referral dynamics within this setting. However, the study provides a valuable perspective on how pediatric referral pathways operate at the central node of the healthcare system, where cases from diverse geographic regions converge. This setting enables an internally consistent, real-world assessment of referral dynamics, case complexity, and transport-related resource use, complementing future multicenter and population-based analyses.

From a policy perspective, the findings highlight several potential areas for improvement. Reducing avoidable transfers—particularly in urban settings—may require strengthening primary care access, improving triage processes, and supporting clinical decision-making. Evidence from other settings suggests that pediatric observation units can reduce unnecessary admissions and associated costs [29], while standardized referral criteria may improve consistency in transfer decisions. Telemedicine also represents an important opportunity to support clinicians in peripheral settings by enabling real-time consultation with tertiary centers, thereby reducing uncertainty and potentially avoiding unnecessary transfers [30].

At the same time, targeted strategies are needed to address non-urgent ambulance use, particularly in urban areas. Improving access to primary care and enhancing parental health literacy may help reduce inappropriate utilization of emergency services [17,19,20,31,32]. Given the disproportionate costs of air and sea transport, optimising triage for high-cost transfers should be a priority. Ensuring that these resources are reserved for patients with clear clinical need may improve system efficiency without compromising quality of care [14].

Because the dataset includes transfers to a single tertiary hospital, regional transfer rates should be interpreted as indicators of referral burden rather than population incidence.

5. Limitations

This study has several limitations. First, its retrospective observational design limits causal inference and depends on the accuracy and completeness of routinely documented clinical and administrative data. Unmeasured confounding and variability in documentation practices may therefore have influenced classification and outcome assessment.

Second, the analysis was conducted at a single tertiary pediatric referral center. Although the hospital receives referrals from multiple geographic regions across Greece, the findings primarily reflect referral dynamics and healthcare utilization patterns within this tertiary-center setting and should not be interpreted as national population-level estimates.

Third, transport costs were estimated using operational transport-cost data rather than detailed patient-level micro-costing methodologies. Consequently, the economic analyses primarily reflect transport-modality–based expenditure and cost concentration rather than precise individual healthcare costs. Because air and sea transport structurally dominated expenditure, much of the observed cost variation was attributable to transport modality rather than patient-specific clinical factors.

Fourth, diagnostic grouping was based on pragmatic clinical categorization rather than standardized ICD coding, and standardized severity-of-illness scores were unavailable, limiting adjustment for clinical complexity.

Fifth, potentially avoidable transfers were classified retrospectively using predefined resource-utilization criteria. Adjudicators were not blinded to transfer pathway, and inter-rater reliability was not formally assessed. Therefore, this construct should be interpreted as a conservative exploratory proxy for transfer necessity rather than a definitive assessment of clinical appropriateness.

Finally, regression analyses were limited by sparse data within selected covariate strata, reducing the stability of some estimates.

6. Strengths

Despite these limitations, this study has several important strengths. To our knowledge, it represents one of the first systematic analyses in Greece to jointly examine the clinical appropriateness, geographic distribution, and economic burden of pediatric interhospital transfers within a single analytical framework.

A key strength is the comprehensive inclusion of all pediatric transfers over a defined period, without restriction by diagnosis, transfer modality, or clinical severity. This approach minimises selection bias and provides a real-world representation of transfer practices across the full spectrum of pediatric conditions.

The study setting further enhances its relevance. Conducted at one of the largest tertiary pediatric hospitals in Greece, which serves as a major national referral center, the analysis captures transfer patterns from both metropolitan and geographically remote regions. This position at the convergence of regional referral pathways enables a uniquely informative assessment of how the pediatric transfer system operates in practice, including the interplay between referral decisions, case complexity, and resource utilization.

Another important strength is the integration of clinical, geographic, and economic data. By linking patient outcomes with transport modality and cost, the study moves beyond purely descriptive epidemiology and provides a system-level perspective on how structural factors—such as regional care capacity and transport logistics—shape both utilization and expenditure. This integrated approach enables identification of key patterns, including the concentration of costs in a small number of high-resource transfers, which may not be apparent when clinical or economic data are examined in isolation.

The use of consistent diagnostic grouping across descriptive and multivariable analyses enhances internal validity and interpretability, while detailed stratification by age, region, seasonality, and transport mode enables granular subgroup analyses and the identification of high-risk and high-cost patterns.

Finally, by explicitly distinguishing between prehospital EMS transports and interfacility transfers, the study provides a clearer conceptual framework for evaluating pediatric transfer systems. This distinction allows for a more accurate interpretation of healthcare utilization and highlights different targets for intervention, thereby increasing the study’s relevance for health system planning and policy development.

7. Conclusions

Pediatric interhospital transfers in Greece reflect important structural characteristics of the healthcare system in addition to clinical need. Most transfers were associated with clinically meaningful tertiary-level care, although a measurable minority—particularly in urban settings—met predefined criteria for potential avoidability. Interfacility transfers accounted for most tertiary hospital-level care/admission events and transport expenditure, whereas transfers from remote and island regions, although less frequent, generated disproportionately high resource use because of reliance on air and sea transport.

These findings suggest that regional differences in healthcare capacity, referral pathways, and access to primary care substantially influence pediatric transfer patterns and resource utilization. A dual strategy may therefore be required: strengthening regional pediatric services to support clinically appropriate long-distance transfers while improving primary care access, referral coordination, and telemedicine support to reduce potentially avoidable transfers. Future prospective multicenter studies incorporating standardized severity assessment are needed to evaluate the effectiveness of these strategies and inform the development of more efficient pediatric transfer networks.