Review Reports - Potentially Avoidable Pediatric Transfers, Tertiary Hospital-Level Care, and Regional Differences in Referral Burden in Greece: A Retrospective Analysis of Transfer Pathways and Costs

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

In this retrospective observational study, Dr. Filippatos et al. report on pediatric transfers to a tertiary children’s hospital in Athens, Greece, over a 10-month period. The results show that: i) interfacility transfers accounted for most tertiary hospital-level care/admission events and the large majority of transport costs; ii) a small proportion of air and sea transfers accounted for nearly all transport expenditure; and iii) only a measurable minority of transfers—particularly in urban settings—met the predefined criteria for potential avoidability.

This is a very detailed report of pediatric transfers occurring at a tertiary university hospital in Athens. In some parts, the article is difficult to read because it contains a large amount of information. However, I believe that this information could be useful to healthcare professionals involved in the care of children requiring emergency transfers.

My main concern relates to the calculation of transport costs. The authors state that the “cost included fuel, personnel, and operational expenses…” (Line 154). However, the median cost of each pediatric transfer was €15.00. How is this figure possible, considering the components included in the cost calculation? Could the authors please provide further details on how the costs were calculated and clarify whether the reported median cost represents the actual cost per transfer or, for example, an estimated or reimbursed amount?

A crucial aspect of pediatric transport is the timing of the mission, including activation time, stabilization time, transport time, and total mission time. Do the authors have information on these time intervals? If available, reporting these data would provide valuable additional information, particularly for assessing the efficiency and timeliness of emergency pediatric transfers.

Author Response

Reviewer 1

Response: We thank the reviewer for the positive assessment and constructive comments. To improve readability, we streamlined repetitive sections, clarified the economic methodology, and revised Figure 1 and the presentation of post-transfer outcomes.

Comment 1:

Response:
We thank the reviewer for identifying this lack of clarity. The reported median cost of €15 does not represent a patient-level calculation of the actual resources consumed during each ambulance mission, nor does it represent a reimbursed amount. For ground ambulance transfers, the cost assigned to each journey was an estimated administrative amount derived from EKAB records and calculated according to the kilometric distance between the recorded transfer origin and the receiving hospital. It therefore represents a standardized distance-based operational estimate rather than detailed micro-costing of fuel consumption, personnel time, vehicle depreciation, dispatch activity, or other mission-specific expenses.

The relatively low median can be explained by two characteristics of the cohort. First, ground ambulance journeys accounted for the vast majority of transfers (372/423, 87.9%) and had a median assigned cost of €15. Second, most transfers originated within the Athens metropolitan area, where the distances to the receiving tertiary hospital were generally shorter than those associated with transfers from provincial or island regions. Consequently, these short-distance urban ground transfers determined the median of the overall cohort. In contrast, the relatively small number of high-cost air and sea transfers produced the marked right-skew of the distribution and increased the overall mean cost to €745.56.

We agree with the reviewer’s comment that the original statement that costs “included fuel, personnel, and operational expenses” could incorrectly imply that these components were measured individually for each transfer. We have therefore revised the Methods, Results, Table 3 footnote, and Limitations to clarify how we calculated and interpreted the reported cost estimates.

Comment 2:

Response:
We thank the reviewer for highlighting the importance of mission-time intervals in evaluating pediatric transport systems. The present study was designed to examine transfer pathways, tertiary hospital-level care/admission, potentially avoidable transfers, regional patterns, and transport-related costs rather than operational mission timing. Therefore, activation, arrival, stabilization, departure, transport, and total mission times were not predefined study variables and were not systematically collected. Consequently, these data are not available for reliable retrospective analysis.

We have nevertheless acknowledged the reviewer’s important concern in the Discussion and identified standardized assessment of mission-time intervals as a complementary objective for future prospective research. The following sentence has been added: Future prospective studies using standardized mission-time definitions could complement these findings by evaluating the operational efficiency and timeliness of pediatric transport pathways.

Reviewer 2 Report

Comments and Suggestions for Authors

Comments to manuscript [ Potentially Avoidable Pediatric Transfers, Tertiary-Care Utilization, and Regional Differences in Referral Burden in Greece: A Retrospective Analysis of Transfer Pathways and Cost

1- I would like to thank the authors for their efforts

2- General comments

This manuscript addresses an important aspect of pediatric healthcare delivery in Greece, namely the clinical utilization, referral patterns, and economic burden of pediatric transfers within the healthcare system. The study provides useful descriptive information regarding transfer pathways, tertiary-care utilization, potentially avoidable transfers, and transport-related costs. The topic is relevant to health services research and healthcare policy. The manuscript is generally well written, the methodology is transparent, and the authors appropriately acknowledge many limitations. However, several methodological concerns limit the manuscript's strength, particularly regarding the assessment of transfer appropriateness, regional comparisons, and the interpretation of explanatory factors not directly measured in the study. The absence of standardized severity measures, subjective retrospective adjudication, lack of independent review of avoidable transfers, and the single-center design limit the interpretation of transfer appropriateness. Several explanatory factors discussed in the manuscript were not directly measured and should be presented with greater caution. These limitations do not invalidate the findings but should be emphasized when interpreting implications for healthcare policy and referral-system redesign.

Major concerns :

Methods

1- Single-center design that limits external validity, also this means not represent Greece nationally reflect institutional practice only.

2-Diagnostic grouping was based on clinical judgement rather than a standardized coding system; this introduces potential classification bias

3-The transfer classifications: concerns include retrospective assessment, the reviewers were not blinded, with no independent adjudication and no inter-rater agreement statistics. The estimated 12.8% avoidability rate should therefore be interpreted cautiously.

4- The severity of cases cannot be fully estimated as no validated indicators of illness severity were available. Missing variables include triage acuity, PEWS, physiological scores, and severity indices. These may affect transfer necessity, resource utilization, and outcome interpretation

Results

5-OR = 3.72 should not be interpreted as an independent effect, as the final logistic model includes only transfer type as the primary predictor. Residual confounding remains possible; diagnosis, age, geography, and transport mode may explain part of the observed association.

6-The economic analysis is useful but should be recognized as operational costing, not patient-level costing and not societal costing. Therefore, conclusions regarding efficiency should remain cautious

7- Regional transfer rate comparisons may overinterpret differences because healthcare capacity was not measured, referral thresholds were not measured, and population movements were not measured. The observed differences may have multiple explanations.

8-Small numbers in some regional subgroups, Crete (n=2) and Macedonia-Thrace (n=2) , make estimates unstable.

Discussion

9-The manuscript frequently discusses primary care access, health literacy, regional service availability, and referral culture as explanations. However, none were directly measured.

10-The telemedicine discussion is reasonable but largely speculative. The study provides no direct data demonstrating telemedicine effectiveness, avoided transfers, and cost reduction. Thus, policy recommendations should remain cautious.

Minor concerns

Duplication: Table 1 and parts of Section 3.4 appear duplicated in the manuscript text submitted.

Terminology: The manuscript alternates between tertiary-care utilization, hospital-level care, and tertiary hospital-level care. Consistency would improve readability.

The flow diagram could more clearly distinguish ED discharge, short-stay observation, inpatient admission, and ICU admission

Comments for author File: Comments.pdf

Author Response

Reviewer 2

2- General comments

Response: We thank the reviewer for this balanced assessment. We have addressed the methodological concerns by clarifying the study’s single-center scope, diagnostic and transfer-classification procedures, absence of standardized severity measures, operational-costing perspective, and limitations of regional comparisons and policy interpretation.

Major concerns:

Methods

1-Single-center design that limits external validity, also this means not represent Greece nationally reflect institutional practice only.

Response: We thank the reviewer for this valuable comment and completely agree that the single-center design limits the external validity of the findings. Aghia Sophia Children’s Hospital is the largest tertiary pediatric referral center in Greece and receives transfers from multiple metropolitan, mainland, and island regions. Nevertheless, the study captures transfers arriving at a single institution and therefore reflects its referral catchment, transfer pathways, documentation procedures, and institutional practices. The findings should not be interpreted as nationally representative estimates of pediatric transfer activity in Greece. We have clarified this point in the revised Discussion and Limitations. Multicenter or population-based studies incorporating other tertiary hospitals and regional referral networks are required to determine the generalizability of these findings across the country.

2-Diagnostic grouping was based on clinical judgement rather than a standardized coding system; this introduces potential classification bias

Response: We absolutely agree that the absence of a standardized diagnostic coding system introduces a risk of classification bias. We assigned diagnoses retrospectively to predefined, clinically relevant categories based on the principal presenting condition documented at hospital arrival. However, this process required clinical judgment, did not use ICD coding, and was not independently duplicated. Misclassification was therefore possible, particularly for children with overlapping or multisystem presentations. Although we applied the same predefined grouping rules throughout the analysis, this ensured internal consistency but did not eliminate potential classification bias. We have clarified the classification process in the Methods, explicitly acknowledged this limitation, and moderated the corresponding statement in the Strengths section.

Response: We thank the reviewer for this comment and agree that the estimated 12.8% rate requires cautious interpretation. We performed classification retrospectively using predefined, conservative resource-utilisation criteria. Because transfer origin and the subsequent clinical course were integral components of the medical record, the investigators were not blinded to the transfer pathway. In addition, a second reviewer did not independently adjudicate cases, so inter-rater agreement statistics could not be calculated.

Although the predefined criteria promoted consistency and reduced reliance on an entirely subjective assessment, they do not eliminate the possibility of reviewer-dependent classification or retrospective information bias. We have therefore clarified that 12.8% represents the proportion of transfers meeting our study-specific operational criteria for potential avoidability, rather than an independently adjudicated prevalence of clinically inappropriate transfers. We revised the Methods, Results, and Limitations accordingly.

Response: We thank the reviewer and agree that the absence of validated illness-severity measures limits characterization of baseline clinical severity. Standardized triage acuity, Pediatric Early Warning Scores (PEWS), physiological scores, and validated severity indices were not consistently available in the retrospective records and could not be included in the analyses. Consequently, we could not determine the extent to which differences in baseline severity influenced transfer necessity, subsequent resource utilization, or clinical outcomes.

We have clarified that tertiary hospital-level care/admission is a post-transfer utilization outcome and should not be considered a substitute for a validated illness-severity measure. We have also moderated the interpretation that interfacility transfers involved greater clinical complexity, added a separate limitation concerning unmeasured severity, and identified standardized severity assessment as an important component of future prospective multicenter studies.

Results

Response: We thank the reviewer and agree. The OR of 3.72 was derived from a univariable logistic regression model containing transfer type as the sole predictor. It should therefore be reported as an unadjusted, pathway-level association and not interpreted as an independent effect of interfacility transfer. We explored fully adjusted modelling but did not retain it because sparse observations and unstable estimates arose when diagnosis, age, geographic origin, and transport mode were included across several strata. Differences in these measured characteristics, together with unmeasured illness severity and other clinical factors, may explain part of the observed association.

We have revised the Abstract, Methods, Results, Table 6, Discussion, and Limitations to identify the OR explicitly as unadjusted, avoid causal or independent-effect language, and acknowledge residual confounding.

Response: We thank the reviewer and agree. Our analysis uses administrative operational transport-cost estimates rather than patient-level or societal costing. It therefore describes the distribution and concentration of recorded transport expenditure but cannot assess economic efficiency, cost-effectiveness, or the full economic burden of pediatric transfers. We have clarified this distinction in the Methods and Limitations and moderated the corresponding statements in the Discussion, Conclusions, and key messages.

Response: We thank the reviewer and agree that regional comparisons require cautious interpretation. The calculated rates represent transfers to the study hospital relative to the permanent pediatric population reported by ELSTAT; they do not capture transfers to other hospitals, regional healthcare capacity, referral thresholds, or temporary population movements. The observed differences may therefore reflect several interacting factors and should not be interpreted as population incidence or as direct evidence of regional service deficiencies. We have revised the Methods, Results, Table 1, Discussion, and Limitations to describe these measures as study-period referral rates to the study center and to acknowledge the multiple possible explanations.

8-Small numbers in some regional subgroups, Crete (n=2) and Macedonia-Thrace (n=2), make estimates unstable.

Response: We appreciate this important comment. The regional cost estimates for Crete and Macedonia–Thrace are based on only two transfers each and are therefore unstable. These subgroups were retained for descriptive completeness and to preserve the administrative regional classification, but no reliable regional inference should be drawn from their mean or median costs. We have added explicit caution in the Results, Table 4 footnote, and Limitations.

Discussion

9-The manuscript frequently discusses primary care access, health literacy, regional service availability, and referral culture as explanations. However, none were directly measured.

Response: We thank the reviewer and agree. Primary-care access, parental health literacy, regional service availability, referral thresholds, and referral culture were not measured. They should therefore not be presented as explanations derived from our data. We have revised the Discussion to distinguish observed findings from literature-based hypotheses, removed causal or prescriptive language, and explicitly acknowledged these unmeasured factors in the Limitations. They are now presented only as potential determinants requiring prospective evaluation.

Response: We agree. The study did not collect data on telemedicine availability or use and therefore cannot determine whether telemedicine would prevent transfers, reduce costs, or improve clinical outcomes. We have revised the Discussion and Conclusions to present telemedicine only as a hypothesis for prospective evaluation rather than as an evidence-based policy recommendation arising from this study.

Minor concerns

Duplication: Table 1 and parts of Section 3.4 appear duplicated in the manuscript text submitted.

Response: We thank the reviewer for identifying these duplications. We reviewed the revised manuscript and retained only one version of Table 1 and its corresponding title. We also consolidated the repeated findings in Section 3.4, particularly the duplicate reporting of the association between diagnostic category and tertiary hospital-level care/admission. We removed the redundant text without changing the reported results.

Terminology: The manuscript alternates between tertiary-care utilization, hospital-level care, and tertiary hospital-level care. Consistency would improve readability.

Response: We have standardized the terminology throughout the manuscript. The predefined binary outcome is now consistently termed “tertiary hospital-level care/admission,” contrasted with “ED/short-stay discharge.” The shorter term “tertiary hospital-level care” is retained only in the manuscript title and broad descriptive contexts. The alternative expressions “tertiary-care utilization,” “tertiary-care use,” “hospital-level care,” and “tertiary-level care” have been removed or replaced.

The flow diagram could more clearly distinguish ED discharge, short-stay observation, inpatient admission, and ICU admission

Response: We thank the reviewer and agree. We retained the overall structure of Figure 1 because it reflects the predefined binary endpoint used in the primary analysis. However, we revised the outcome boxes and caption to distinguish direct ED discharge (n=117), short-stay observation (n=82), non-ICU tertiary hospital-level care/admission (n=221), and ICU admission (n=3). Because the primary outcome included clinically meaningful non-ED tertiary dispositions in addition to conventional ward admission, the 221 non-ICU cases are labelled “non-ICU tertiary hospital-level care/admission” rather than “inpatient admission” alone. The revised caption explicitly identifies the dispositions included in this category.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

I have no more comments .