Impact of CytoSorb Hemoadsorption Therapy on Cost-Effectiveness and Length of Stay in Critical Care Patients: A Preliminary Study from a Swiss High-Volume Center

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This article examines the socioeconomic value of CytoSorb hemoadsorption therapy in intensive care units, a topic of interest beyond its therapeutic efficacy. My major and minor concerns are as follows:

Major concerns:

1. The consideration of the comparability of the control group is a commendable approach. However, given that the primary aim of this study is to explore the cost-effectiveness of CytoSorb in intensive care units, it's certainly beneficial to control for covariates such as the cause of sepsis, SAPS II, and NEMS. Considering the unique nature of the outcomes, why did the study not include matching patients based on economic or demographic factors (such as gender, age, and baseline health status)? These variables are also likely to influence the outcomes.
2. The inclusion criteria for the sample in the article are not clearly defined. Were patients included based on meeting the indications for CytoSorb therapy, or were all patients who received CytoSorb treatment at your institution included? These two concepts do not completely overlap.
3. How was matching achieved in the study? This aspect is not mentioned in the article. Given that the study appears to be designed retrospectively, was matching implemented using statistical methods? Please provide further clarification on this matter.
4. Patients who meet the indications for CytoSorb therapy often have multiple organ dysfunction. In this study, how was CytoSorb used in these patients? Was it used alone, or in combination with other blood purification methods such as dialysis or ECMO? The use of these additional medical interventions could also impact the cost-effectiveness ratio of CytoSorb. The authors need to clarify this point or include the necessary matched comparisons.
5. The impact of CytoSorb on patient outcomes is currently a subject of debate. Does the article provide any specific references or evidence to support the timing for discontinuation of therapy in patients?

Minor concerns

1. Many abbreviations are not defined in full at their first occurrence, such as DRG. Please ensure that the full terms are provided upon first mention for clarity and consistency.
2. Important baseline information about the patients, such as gender, race, and the presence of other underlying conditions, is missing. The authors should consider reporting this information.

Author Response

This article examines the socioeconomic value of CytoSorb hemoadsorption therapy in intensive care units, a topic of interest beyond its therapeutic efficacy.

 

My major and minor concerns are as follows:

 

1. The consideration of the comparability of the control group is a commendable approach. However, given that the primary aim of this study is to explore the cost-effectiveness of CytoSorb in intensive care units, it's certainly beneficial to control for covariates such as the cause of sepsis, SAPS II, and NEMS. Considering the unique nature of the outcomes, why did the study not include matching patients based on economic or demographic factors (such as gender, age, and baseline health status)? These variables are also likely to influence the outcomes.

 

 

We appreciate the reviewer’s thoughtful comment and agree that demographic and baseline health characteristics may influence clinical and economic outcomes in general. However, there are several reasons why we did not match patients based on variables such as gender, age, or comorbidities in this analysis:

The primary aim of this study was not to evaluate treatment efficacy or outcome differences in terms of survival or complication rates, but rather to examine the cost-effectiveness and operational implications of CytoSorb hemoadsorption therapy in a real-world ICU setting. As such, the analysis was designed from a health economic and hospital management perspective, where variables such as ICU length of stay, resource utilization, and net case results are central. In this context, gender and ethnicity are not typically associated with differential resource needs when disease severity and treatment pathways are comparable.

To ensure clinical comparability while focusing on economically relevant predictors of ICU resource use, we deliberately matched on sepsis origin, SAPS II (reflecting disease severity), and NEMS (a validated measure of nursing effort and workload). These parameters were deemed most impactful for our specific cost-centered endpoints. While age was slightly variable between groups, it did not differ significantly and was not associated with cost differences in our dataset.

Our study focused exclusively on adult patients (≥18 years) with a primary admission diagnosis of septic shock, as defined by the Sepsis-3 consensus criteria, which no longer differentiate between "severe" or "mild" septic shock. According to Singer et al. (2016) (REFERENCE [1], septic shock is now defined as a distinct clinical entity, characterized by circulatory and cellular/metabolic abnormalities with significantly increased mortality. As such, all included patients met identical clinical criteria for septic shock at the time of ICU admission.

Moreover, due to the retrospective design and reliance on routinely collected clinical and DRG-based administrative data, the granularity required to reliably adjust for the full spectrum of comorbidities and socioeconomic status was limited. However, all included patients were admitted primarily for sepsis or septic shock, as defined by the main admission and coding diagnosis, thus limiting potential confounding by unrelated secondary conditions.

Importantly, from a DRG-based hospital accounting perspective, the cost of CytoSorb is fixed per treatment and unaffected by patient gender, age, or comorbidities. Likewise, SwissDRG reimbursement mechanisms do not differentiate by these demographic parameters but rather by diagnosis-related grouping and severity coding. Therefore, for this particular cost-effectiveness question, these variables were not expected to systematically bias the results.

Nevertheless, we acknowledge this limitation and have now clarified our rationale in the manuscript to enhance transparency.

To address this issue, we included the following statement to the methods section:

“Matching was performed based on sepsis origin, SAPS II score, and NEMS score, reflecting clinical severity and nursing workload as key drivers of ICU resource use. Variables such as age, gender, and comorbidities were not included in the matching algorithm, as these were not expected to significantly impact cost-related endpoints within the SwissDRG framework and were not the focus of this retrospective economic evaluation. Furthermore, all patients were admitted with sepsis or septic shock as the primary diagnosis, ensuring a high degree of clinical comparability across cohorts.”

 

2. The inclusion criteria for the sample in the article are not clearly defined. Were patients included based on meeting the indications for CytoSorb therapy, or were all patients who received CytoSorb treatment at your institution included? These two concepts do not completely overlap.

 

We thank the reviewer for this important clarification request. We acknowledge that the phrasing in the manuscript may have led to ambiguity and are happy to elaborate.

This was a retrospective, data-driven analysis based on coded clinical and economic records from the SwissDRG accounting system. The initial patient screening included all adult (>18 years) patients with a DRG-coded primary diagnosis of sepsis or septic shock treated across the two acute-care hospitals within our hospital group (Spital Thurgau AG) between 2020 and 2023.

To build the final analytical cohort, we applied the following stepwise inclusion criteria:

1. Selection of all inpatients coded with sepsis or septic shock as a primary diagnosis (n = 6,342).
2. Restriction to patients treated at Kantonsspital Münsterlingen (n = 2,834).
3. Further restriction to those patients who had an ICU stay during their hospitalization.
4. Identification of patients with confirmed ICU diagnosis of septic shock.
5. Division of this final group into:
   • Patients treated with CytoSorb hemoadsorption therapy during their ICU stay, based on internal therapy documentation and device tracking (n = 142), and
   • A control group of ICU patients with identical DRG-coded septic shock diagnoses who were not treated with CytoSorb (n=104).

Thus, all patients who received CytoSorb in the ICU for septic shock during the study period were included in the intervention group. Inclusion was not further restricted by prospective adherence to specific indication thresholds, although the institutional CytoSorb therapy algorithm (as now clarified in the methods section) was routinely followed. We have now clarified this point in the manuscript and refer to the cohort construction flowchart (Figure 1) to visually reinforce the selection process.

To clarify, the following parts were included in the Methods Section (Patient Population and Data Collection)

„Screening was performed retrospectively using SwissDRG-based administrative and clinical coding data. The initial screening population was comprised of all hospitalized patients within the Spital Thurgau hospital network coded with sepsis or septic shock as a primary diagnosis during the study period. This population was subsequently restricted to patients treated at Kantonsspital Münsterlingen and further narrowed to those admitted to the ICU with a confirmed diagnosis of septic shock, as illustrated in the patient selection flow diagram (Figure 1).“

„All ICU patients with septic shock who received CytoSorb hemoadsorption during the study period were included in the intervention group. Inclusion was based on actual treatment exposure rather than post hoc confirmation of formal indication criteria. While institutional guidelines for CytoSorb use were routinely applied in clinical practice, no patients were excluded retrospectively based on adherence to specific threshold parameters.“

 

3. How was matching achieved in the study? This aspect is not mentioned in the article. Given that the study appears to be designed retrospectively, was matching implemented using statistical methods? Please provide further clarification on this matter.

 

We thank the reviewer for this important comment. As clarified in the revised manuscript, the matched control group was derived from the same time period and institutional dataset and included patients with a primary diagnosis of septic shock who were treated in the same ICU but did not receive CytoSorb therapy. Matching was performed based on three clinically and economically relevant parameters: the origin of sepsis (abdominal, pulmonary, urogenital, or other), SAPS II score (disease severity), and NEMS score (nursing workload), all of which are key drivers of resource utilization in the SwissDRG system. Other demographic parameters such as age or gender were not included in the matching algorithm, as these are not considered primary cost determinants in DRG-based reimbursement and were therefore not the focus of this economic evaluation.

 

A clarifying sentence was added in the “Patient Population and Data Collection” section, reading:

 

"A matched control group was created from the same period, comprised of patients with identical primary sepsis diagnoses but who were not treated with CytoSorb. Matching was performed based on sepsis origin (abdominal, pulmonary, urogenital, or other), Simplified Acute Physiology Score II (SAPS II score) [15], and the Nine Equivalents of Nursing Manpower Use Score (NEMS) [16], reflecting clinical severity and nursing workload as key drivers of ICU resource use. "

 

4. Patients who meet the indications for CytoSorb therapy often have multiple organ dysfunction. In this study, how was CytoSorb used in these patients? Was it used alone, or in combination with other blood purification methods such as dialysis or ECMO? The use of these additional medical interventions could also impact the cost-effectiveness ratio of CytoSorb. The authors need to clarify this point or include the necessary matched comparisons.

 

We thank the reviewer for this important comment.

CytoSorb hemoadsorption was exclusively applied in combination with continuous renal replacement therapy (CRRT) in all cases in this cohort. In our institution, CytoSorb is routinely integrated into the CRRT circuit using established connection protocols, regardless of whether the patient presented with acute kidney injury. This reflects both institutional familiarity with CRRT platforms and the infrastructure of our ICU, which operates as a regional nephrology reference center.

CytoSorb was not used as a stand-alone hemoperfusion therapy nor in combination with ECMO or cardiopulmonary bypass during the study period. Accordingly, cost comparisons between groups were not confounded by the presence of these additional extracorporeal therapies.

 

Accordingly, we have added the following sentences to the Manuscript (Methods Section):

“In all cases, CytoSorb was applied via integration into continuous renal replacement therapy (CRRT) circuits. Stand-alone hemoperfusion systems, ECMO, or cardiopulmonary bypass circuits were not used for CytoSorb application during the study period.“

 

 

5. The impact of CytoSorb on patient outcomes is currently a subject of debate. Does the article provide any specific references or evidence to support the timing for discontinuation of therapy in patients?

 

We thank the reviewer for this important observation. In our study, the discontinuation of CytoSorb therapy followed internal clinical practice protocols, which are aligned with the best practice recommendations outlined by Mitzner et al. (2023). Treatment duration was typically guided by the achievement of hemodynamic stabilization, with adsorbers replaced every 12–24 hours and a maximum of three adsorbers used per treatment course unless clinical deterioration warranted extension.

 

We have now clarified this in the Methods section and added a citation to the Mitzner consensus paper to support the rationale behind our approach.

 

This paragraph has been added to the "Patient Population and Data Collection" section:

 

„Discontinuation of CytoSorb therapy was guided by clinical stabilization and typically occurred after three adsorber cycles, unless persistent hyperinflammation or hemodynamic instability warranted continuation, following our institutional protocols which are consistent with recently published international best-practice recommendations [Mitzner et al., 2023].“

 

 

Minor concerns

1. Many abbreviations are not defined in full at their first occurrence, such as DRG. Please ensure that the full terms are provided upon first mention for clarity and consistency.

 

We thank the reviewer for this helpful comment. We have carefully reviewed the entire manuscript and ensured that all abbreviations, including “DRG” (now defined as Diagnosis-Related Group at first mention), are introduced in full upon their first appearance.

 

2. Important baseline information about the patients, such as gender, race, and the presence of other underlying conditions, is missing. The authors should consider reporting this information.

 

In the context of this retrospective health-economic analysis, variables such as gender, ethnicity, and non-primary comorbidities were not included in the matching algorithm or subgroup analysis. This decision was deliberate and grounded in both methodological rationale and the structure of the SwissDRG system. From a health economics perspective, these demographic factors do not independently influence treatment costs or DRG-based reimbursement. The CytoSorb adsorber has a fixed unit price, and its cost does not vary by patient gender, ethnicity, or age. Similarly, ICU care elements such as medication, nursing effort, and organ support are protocol-driven and based on clinical severity rather than demographic attributes. In other words, a female and a male patient with the same SAPS II score and clinical course will generate comparable ICU costs and revenues within the SwissDRG framework. Moreover, all patients included in this study were admitted with sepsis or septic shock as their primary diagnosis, which anchors DRG coding and reimbursement. While some patients may have had comorbidities, these were not leading contributors to the cost allocation under the DRG model and were coded as secondary diagnoses. This ensures that the analysis focuses on cost drivers directly linked to the index condition (sepsis) and avoids confounding by non-contributory variables.

It is also important to emphasize that the objective of this study was not to assess treatment efficacy or patient-centered clinical outcomes, but rather to evaluate whether CytoSorb therapy in a real-world ICU context leads to cost savings or financial disadvantage for the treating institution. Therefore, including demographic variables that are not cost-relevant under DRG logic would not have enhanced the accuracy or relevance of the findings and could have introduced unnecessary noise in the economic analysis.

 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Thank you for the assignment to review the article entitled " Impact of CytoSorb hemoadsorption therapy on cost-effectiveness and length of stay in critical care patients - a Swiss high-volume centre experience". Tobias Hübner and Oliver Schöffski conducted a single-centre retrospective analysis using DRG-based systems with the aim to evaluate the clinical and economic impact of extracorporeal hemoadsorption (CytoSorb treatment) in ICU patients with septic shock. In this retrospective analysis, the authors found that patients who received hemoadsorption treatment had shorter periods of ICU and hospital length of stay, mechanical ventilation and lower NEMS points. While the total hospital costs registered in the study groups were similar, the authors found that the net financial result was better in the patients treated with CytoSorb compared to those receiving standard care. The authors concluded that the results of this single-centre retrospective analysis provide real-world evidence of the cost-effectiveness of CytoSorb hemoadsorption in severe septic shock underlined by the reduced ICU length of stay and improved financial and nursing workload outcomes.

The authors are to be congratulated for their pursuits aimed to evaluate the cost-effectiveness of the hemoadsorption treatment as a component of the complex care of septic/severe septic shock patients. The topic of this analysis is of interest, but there are several questions regarding the methodology of the study and the interpretation of the results.

Major issues:

• Conception and study design: The critical shortcoming of this retrospective analysis is the objective definition of comparable patient groups, which provides the valid assessment of the hemoadsorption treatment associated benefits in terms of cost-effectiveness in patients treated with septic shock. As the clinical and financial outcomes are significantly influenced by the interaction between patient characteristics and severity of the septic shock, it is obvious that the SAPS II composite score alone is not sufficient to cover the complexity of this patient group.
• Methods: The authors should present the exact date of the institutional guidelines introduction related to the structured indication of the CytoSorb treatment within the investigation period. At present, the exclusion criteria for hemoadsorption treatment (i.e. control group) are not clearly defined in relation to the mentioned institutional guidelines. In addition, it should be indicated how the matched control group was created.
• Results: Figures 2-6: Please indicate the functions of the error bars.
• Discussion and conclusion: In light of the limitations of the study methodology and the questionable validity of the results presented in this study, it is challenging to evaluate the conclusions drawn by the authors of this manuscript.

Comments on the Quality of English Language

The authors' written English requires a native English language editing to achieve more comprehensible text for readers.

Author Response

Thank you for the assignment to review the article entitled " Impact of CytoSorb hemoadsorption therapy on cost-effectiveness and length of stay in critical care patients - a Swiss high-volume centre experience". Tobias Hübner and Oliver Schöffski conducted a single-centre retrospective analysis using DRG-based systems with the aim to evaluate the clinical and economic impact of extracorporeal hemoadsorption (CytoSorb treatment) in ICU patients with septic shock. In this retrospective analysis, the authors found that patients who received hemoadsorption treatment had shorter periods of ICU and hospital length of stay, mechanical ventilation and lower NEMS points. While the total hospital costs registered in the study groups were similar, the authors found that the net financial result was better in the patients treated with CytoSorb compared to those receiving standard care. The authors concluded that the results of this single-centre retrospective analysis provide real-world evidence of the cost-effectiveness of CytoSorb hemoadsorption in severe septic shock underlined by the reduced ICU length of stay and improved financial and nursing workload outcomes.

 

The authors are to be congratulated for their pursuits aimed to evaluate the cost-effectiveness of the hemoadsorption treatment as a component of the complex care of septic/severe septic shock patients. The topic of this analysis is of interest, but there are several questions regarding the methodology of the study and the interpretation of the results.

 

1. Conception and study design: The critical shortcoming of this retrospective analysis is the objective definition of comparable patient groups, which provides the valid assessment of the hemoadsorption treatment associated benefits in terms of cost-effectiveness in patients treated with septic shock. As the clinical and financial outcomes are significantly influenced by the interaction between patient characteristics and severity of the septic shock, it is obvious that the SAPS II composite score alone is not sufficient to cover the complexity of this patient group.

 

We thank the reviewer for raising this important point regarding group comparability. We would like to clarify and expand on the rationale behind our matching approach.

Our study focused exclusively on adult patients (≥18 years) with a primary admission diagnosis of septic shock, as defined by the Sepsis-3 consensus criteria, which no longer differentiate between "severe" or "mild" septic shock. According to Singer et al. (2016) (REFERENCE [1], septic shock is now defined as a distinct clinical entity, characterized by circulatory and cellular/metabolic abnormalities with significantly increased mortality. As such, all included patients met identical clinical criteria for septic shock at the time of ICU admission.

 

To ensure cohort comparability, we matched patients based on:

• Sepsis origin (abdominal, pneumogenic, urogenital, or other),
• SAPS II score, as this is the standardized severity score mandated by the Swiss Society for Intensive Care Medicine’s MDSI (Minimal Dataset Intensive Care) for all ICU admissions,
• NEMS score (Nine Equivalents of Nursing Manpower Use Score), which reflects nursing workload and resource utilization, both critical for economic evaluations in ICU settings.

 

While we acknowledge that additional variables such as age, gender, or underlying comorbidities may influence individual outcomes, they were not included in our matching algorithm for the following reasons:

1. The primary aim of this study was to assess the economic impact (not clinical outcomes) of CytoSorb therapy under the SwissDRG reimbursement system. In this context, age and gender have minimal relevance, as treatment costs for comparable severity of illness are not stratified by these demographic variables.
2. All patients were selected based on their primary diagnosis (septic shock), which, per DRG logic, defines the main cost-driving condition during hospitalization. Any comorbidities were secondary and had no dominant influence on case coding or cost allocation.
3. Our approach reflects real-world ICU practice and billing structures, where economic outcomes are tied to diagnosis-related grouping and not adjusted for demographic parameters.
4. SAPS II remains the only validated score systematically collected in Swiss ICUs for benchmarking and reimbursement purposes. While we agree it may not capture every nuance of disease complexity, it provides a robust, standardized comparator and is integrated into clinical and economic workflows.

 

Given this, while additional covariate adjustment could offer further granularity, our matching strategy was tailored to the study’s economic focus and conformed with both Swiss healthcare practice and international sepsis definitions. We have clarified these points in the revised manuscript (see Methods, “Patient Population and Matching Criteria”).

 

“Matching was performed based on sepsis origin (abdominal, pulmonary, urogenital, or other), Simplified Acute Physiology Score II (SAPS II score) [15], and the Nine Equivalents of Nursing Manpower Use Score (NEMS) score [16], reflecting clinical severity and nursing workload as key drivers of ICU resource use.”

 

“Variables such as age, gender, and comorbidities were not included in the matching algorithm, as these were not expected to significantly impact cost-related endpoints within the SwissDRG framework and were not the focus of this retrospective economic evaluation.”

 

“Furthermore, all patients were admitted with sepsis or septic shock as the primary diagnosis, ensuring a high degree of clinical comparability across cohorts.”

 

 

2. Methods: The authors should present the exact date of the institutional guidelines introduction related to the structured indication of the CytoSorb treatment within the investigation period.

 

We thank the reviewer for this important remark. CytoSorb hemoadsorption was formally integrated into institutional clinical practice at our center in 2016, supported by a standardized internal guideline (SOP) outlining indication criteria and treatment protocols. As all patients included in this study were treated between January 2020 and December 2023, the data exclusively reflect a period during which the SOP had been well established and consistently applied for over four years. This clarification has now been explicitly added to the end of the “Study Design and Setting” section of the revised manuscript as follows:

 

“CytoSorb hemoadsorption was formally integrated into institutional clinical practice in 2016, supported by a standardized operating procedure (SOP) outlining indication criteria and treatment protocols. As all patients included in the current analysis were treated between 2020 and 2023, the study exclusively reflects a period in which the SOP had been fully implemented and consistently applied for over four years.”

 

 

At present, the exclusion criteria for hemoadsorption treatment (i.e. control group) are not clearly defined in relation to the mentioned institutional guidelines

 

We appreciate the reviewer’s comment and welcome the opportunity to clarify. The control group consisted exclusively of patients treated at our center (Kantonsspital Münsterlingen) in the ICU during the study period who met the general inclusion criteria (admission with sepsis or septic shock) but did not meet the institutional thresholds for CytoSorb therapy initiation (e.g. did not reach the dynamic score cut-off of ≥6 within the first 6 hours after ICU admission) (Kogelmann K, Hübner T, Schwameis F, Drüner M, Scheller M, Jarczak D. First Evaluation of a New Dynamic Scoring System Intended to Support Prescription of Adjuvant CytoSorb Hemoadsorption Therapy in Patients with Septic Shock. J Clin Med 2021;10:2939. https://doi.org/10.3390/jcm10132939), those who were <18 years of age and patients with a diagnosis of septic shock who declined intensive care treatment or were subject to end-of-life decisions. Furthermore, patients who were treated in other hospitals, not admitted to the ICU, or who did not fulfill the diagnostic criteria for septic shock were excluded during the initial screening process.

To improve clarity, this has now been explicitly added at the end of the Patient Population and Data Collection section of the revised manuscript as follows:

 

“Control patients met the general inclusion criteria but did not fulfill institutional thresholds for CytoSorb therapy (e.g., sub-threshold dynamic score levels according to Kogelmann et al., 2021), were under 18 years of age, or had a diagnosis of septic shock but declined intensive care treatment as part of an end-of-life decision. These patients were managed with standard care accordingly.”

 

In addition, it should be indicated how the matched control group was created.

 

We thank the reviewer for this important comment. As clarified in the revised manuscript, the matched control group was derived from the same time period and institutional dataset and included patients with a primary diagnosis of septic shock who were treated in the same ICU but did not receive CytoSorb therapy. Matching was performed based on three clinically and economically relevant parameters: the origin of sepsis (abdominal, pulmonary, urogenital, or other), SAPS II score (disease severity), and NEMS score (nursing workload), all of which are key drivers of resource utilization in the SwissDRG system. Other demographic parameters such as age or gender were not included in the matching algorithm, as these are not considered primary cost determinants in DRG-based reimbursement and were therefore not the focus of this economic evaluation.

 

A clarifying sentence was added in the “Patient Population and Data Collection” section, reading:

 

"A matched control group was created from the same period, comprised of patients with identical primary sepsis diagnoses but who were not treated with CytoSorb. Matching was performed based on sepsis origin (abdominal, pulmonary, urogenital, or other), Simplified Acute Physiology Score II (SAPS II score) [15], and the Nine Equivalents of Nursing Manpower Use Score (NEMS) [16], reflecting clinical severity and nursing workload as key drivers of ICU resource use. "

 

3. Results: Figures 2-6: Please indicate the functions of the error bars.

 

We thank the reviewer for this comment. The figure legends for Figures 2–6 have been revised to explicitly define the error bars. We now state that data are presented as medians with error bars indicating the interquartile range (Q1–Q3), ensuring clarity and consistency across all figures.

 

4. Discussion and conclusion: In light of the limitations of the study methodology and the questionable validity of the results presented in this study, it is challenging to evaluate the conclusions drawn by the authors of this manuscript.

We appreciate the reviewer’s concern and fully acknowledge the methodological limitations inherent to retrospective, single-center analyses. In response, we have taken care to further nuance and moderate our conclusions. Specifically, we revised the discussion and conclusion sections to emphasize that the findings are hypothesis-generating and reflective of a specific institutional context. Moreover, we now explicitly frame our results as exploratory rather than definitive, and highlight the need for prospective, multicenter studies to validate our observations. We hope these clarifications adequately address the reviewer’s reservations and reinforce the transparency and contextual relevance of our conclusions.

 

Comments on the Quality of English Language

The authors' written English requires a native English language editing to achieve more comprehensible text for readers.

 

We thank the reviewer for this important observation. The entire manuscript has undergone comprehensive language editing by a native English speaker with scientific writing expertise to improve clarity, grammar, and overall readability. We trust the revised version now meets the journal's standards.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Well designed and executed study. I do, however, feel the study conclusions are not entirely objective in terms of the therapeutic benefit of the intervention. Why would we implement an intervention that does not improve outcome? How do we justify its use? Should we rather look at a non-inferiority design or look at more long-term outcomes etc.?

In terms of the results I would urge the authors to depict the graphs with time on the y-axis be expressed in days and not hours since this makes more clinical sense and is more understandable.

Could we see the differences in SOFA score rather than SAPS scores? Did the authors look at any biochemical indices while receiving the extra-corporeal blood purification? Could we see LOS rather be depicted as days alive rather which eliminates the confounder of early mortality in the Cytosorb group.

Finally, did the authors declare any financial or industry funding for the use of Cytosorb?

 

Author Response

Well designed and executed study. I do, however, feel the study conclusions are not entirely objective in terms of the therapeutic benefit of the intervention. Why would we implement an intervention that does not improve outcome? How do we justify its use? Should we rather look at a non-inferiority design or look at more long-term outcomes etc.?

 

We thank the reviewer for this thoughtful comment. We fully agree that therapeutic benefit in critical care must ultimately be measured by improved patient outcomes. However, this study was not designed or powered to detect survival differences, and mortality was not a predefined primary endpoint. Instead, our objective was to assess economic and operational endpoints in a real-world ICU population, in particular resource utilization, length of stay, and financial outcomes, which are also of high relevance in modern, value-based healthcare systems.

We believe the findings justify consideration of CytoSorb therapy not as a replacement for standard sepsis management, but rather as a potential adjunct in resource-intensive cases where it may contribute to earlier stabilization and more efficient ICU throughput, as reflected in shorter LOS, lower nursing workload, and favorable net case results and without additional cost.

We adjusted the discussion and conclusion section to reflect this distinction more clearly and to temper claims regarding clinical benefit, emphasizing instead that these are hypothesis-generating results which should lead to prospective studies with long-term or patient-centered outcomes and possibly non-inferiority designs, as the reviewer rightly suggests. Moreover, we now explicitly frame our results as exploratory rather than definitive, and highlight the need for prospective, multicenter studies to validate our observations.

 

The following changes were made to the manuscript:

 

“As such, these results are rather hypothesis-generating than definitive and support the design of prospective multicenter trials focused on the health economic impact of hemoadsorption therapy in septic and other critically ill populations.”

 

“Future research may also consider non-inferiority approaches to substantiate clinical equivalence and assess long-term outcomes, with an additional focus on demonstrating measurable gains in resource utilization and operational efficiency.”

 

In terms of the results I would urge the authors to depict the graphs with time on the y-axis be expressed in days and not hours since this makes more clinical sense and is more understandable.

 

This has been done accordingly.

 

Could we see the differences in SOFA score rather than SAPS scores?

We appreciate the reviewer’s suggestion. In our ICU, the SAPS II score is mandated as part of the national Minimal Data Set (MDSi) for coding, quality benchmarking, and reimbursement purposes. Therefore, SAPS II was consistently and retrospectively available for all patients and chosen as the primary severity marker. While SOFA scores are routinely used for bedside assessment, they were not available in a structured format suitable for retrospective extraction and analysis in this dataset.

 

 

Did the authors look at any biochemical indices while receiving the extra-corporeal blood purification?

 

We acknowledge the relevance of biochemical monitoring in the context of hemoadsorption. However, the focus of this study was economic and operational rather than mechanistic. While markers such as IL-6 and PCT were routinely measured to guide therapy, they were not uniformly available in a structured format suitable for retrospective analysis and were therefore not included. This represents an important area for future prospective studies.

 

Could we see LOS rather be depicted as days alive rather which eliminates the confounder of early mortality in the Cytosorb group.

 

This is an excellent point. In order to account for this confounder, we stratified ICU and hospital length of stay (LOS) by survivor status, which was reported in both tables and figures. While ‘days alive’ is a valuable metric, it was not a predefined outcome in this retrospective dataset. We agree it would be a useful endpoint for future prospective designs.

 

Finally, did the authors declare any financial or industry funding for the use of Cytosorb?

 

We confirm that all relevant disclosures have been included in the manuscript. As stated, this study received no external or industry funding. The CytoSorb devices were purchased through standard hospital procurement channels without support or discounts from the manufacturer. Competing interests are transparently declared in the dedicated section.

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

The authors present a cost-effectiveness and length of stay analysis for patients treated for septic shock with or without hemoadsorption with Cytosorb adsorber in a retrospective case control study. I see several major issues, which need major adjustements:

• although authors claim otherwise in the introduction, Cytosorb therapy is not evidence-based and has a very poor evidence base from RCTs. In that context cost-effectiveness analysis is not what is needed...
• authors performed matching, but do not provide results of how well the groups were matched.
• Furthermore, the cost analysis was performed on a sub-group, which seems even less matched (different SAPS II, worse mortality in Cytosorb group), and no explanation is given why cost data was only available for a subgroup- there is a very high risk of selection bias. We do not know if the cost-analysis subgroups are also properly matched, as the whole group. Results of such an ambigous group are hard to interpret.
• the authors report a significantly higher ICU mortality in the matched cohort - this can not be given as a minor/irrelevant result, if the authors truly believe in the effectiveness of their matching - which is likely not true
• regarding the cost and LOS analysis. I see some methodological issues, given the high (and significantly different between the groups) mortality. High mortality actually reduces the cost and LOS for patients that died (dying soon after admission is "cheap"). How was this accounted for? The Cytosorb group had higher mortality and shorter ICU LOS, which both might produce lower costs and result iin the observed simmilar costs between the groups. I presume that with 50% mortality, LOS should be estimated with Kaplan-Meier (censoring for death), some other advanced techniques or only reporting LOS comparison for survivors.

To conclude, I find the analysis poorly matched and of questionable methodological rigour, which also poses the conclusions under question.

Author Response

The authors present a cost-effectiveness and length of stay analysis for patients treated for

septic shock with or without hemoadsorption with Cytosorb adsorber in a retrospective case

control study.

 

I see several major issues, which need major adjustements:

 

1. Although authors claim otherwise in the introduction, Cytosorb therapy is not evidence-based and has a very poor evidence base from RCTs. In that context cost-effectiveness analysis is not what is needed...

 

We thank the reviewer for this important remark and agree that the randomized evidence base for hemoadsorption in septic shock remains controversial and evolving. We would like to clarify two points.

First, the present analysis reflects real-world data collected during a period in which several of the subsequently published negative RCTs were not yet available. At that time, hemoadsorption was implemented under structured institutional criteria in selected patients with therapy-refractory septic shock, based on the available literature, pathophysiological rationale, and accumulating clinical experience.

Second, while high-quality randomized trials are essential to determine definitive efficacy, hemoadsorption is used in clinical practice across several indications beyond septic shock (e.g., liver failure, rhabdomyolysis, drug removal including NOAC-associated bleeding), where its mechanistic plausibility and biochemical effects are well established and largely based on consensus conferences. In our institutional experience, clinical stabilization and hemodynamic improvement in selected high-risk patients were observed, which was not entirely congruent with the uniformly negative interpretation of the current RCT landscape.

Importantly, our study does not aim to demonstrate therapeutic superiority. Rather, it addresses a complementary and pragmatic question: whether the use of hemoadsorption under real-world conditions imposes an additional economic burden within a DRG-based healthcare system. Given that the intervention was already part of institutional practice, evaluating its operational and financial implications is both relevant and necessary.

To avoid overinterpretation, we have further clarified in the revised manuscript that our findings are health-economic and hypothesis-generating, and do not establish or discuss clinical efficacy.

We have therefore implemented the following statement into the discussion section:

“It should be noted that this cohort was treated during a period preceding the publication of controversial results from some randomized trials, and institutional practice was guided by the evidence and clinical experience available at that time.”

 

 

2. authors performed matching, but do not provide results of how well the groups were matched.

 

We thank the reviewer for this important observation. In the revised manuscript, we have now explicitly reported baseline characteristics for both the full matched cohort and the cost-analysis subgroup to transparently demonstrate comparability between groups. Matching was performed based on sepsis origin, SAPS II score, and NEMS score, reflecting clinical severity and nursing workload as key drivers of ICU resource utilization within the SwissDRG framework. These parameters were deemed most impactful for our specific cost-centered endpoints and are now clearly presented in Table 1, including statistical comparisons between groups. As reported, sepsis origin distribution and age were comparable between groups. As expected in a real-world setting, the CytoSorb group exhibited higher SAPS II scores, reflecting indication bias toward more critically ill patients rather than imperfect matching. This has been explicitly acknowledged and discussed. For the cost-analysis subgroup, we have additionally provided baseline characteristics and clarified that cost data availability was limited to the years 2022–2023 due to an accounting system transition. We now explicitly state this in the Methods sections and interpret the economic findings accordingly. We believe that these revisions improve transparency regarding matching performance and allow readers to better contextualize the findings within the constraints of a retrospective observational design.

To address this issue, we included the following statements into the manuscript:

“Matching was performed based on sepsis origin, SAPS II score, and NEMS score, reflecting clinical severity and nursing workload as key drivers of ICU resource use. Variables such as age, gender, and comorbidities were not included in the matching algorithm, as these were not expected to significantly impact cost-related endpoints within the SwissDRG framework and were not the focus of this retrospective economic evaluation. Furthermore, all patients were admitted with sepsis or septic shock as the primary diagnosis, ensuring a high degree of clinical comparability across cohorts.”

3. Furthermore, the cost analysis was performed on a sub-group, which seems even less matched (different SAPS II, worse mortality in Cytosorb group), and no explanation is given why cost data was only available for a subgroup- there is a very high risk of selection bias. We do not know if the cost-analysis subgroups are also properly matched, as the whole group. Results of such an ambigous group are hard to interpret.

 

We thank the reviewer for this important methodological observation and agree that the restriction of cost analyses to a subgroup requires transparent clarification.

Cost data were available only for the years 2022–2023 due to a transition in the hospital’s accounting software system, which limited retrospective extraction of standardized cost variables for earlier years. This was a technical data availability constraint rather than a post hoc selection of cases. We have now explicitly clarified this in the Methods section.

We fully acknowledge that restricting the economic analysis to a subset of the overall cohort introduces a potential risk of selection bias. For this reason, we interpret the cost findings cautiously and explicitly frame them as exploratory rather than definitive. Importantly, the economic observations are directionally consistent with the clinical resource metrics (shorter ICU length of stay and reduced nursing workload) observed in the full cohort, suggesting internal coherence rather than contradiction.

Given the retrospective design and data constraints, we agree that the results should be interpreted within the limitations of observational real-world analyses. We have further tempered the conclusions accordingly.

 

4. the authors report a significantly higher ICU mortality in the matched cohort – this can not be given as a minor/irrelevant result, if the authors truly believe in the effectiveness of their matching - which is likely not true

 

We thank the reviewer for this important comment. We agree that mortality is a clinically relevant endpoint and should not be considered irrelevant. However, mortality was not a matching variable nor a primary endpoint of this study. The purpose of matching was to improve comparability with respect to key drivers of ICU resource utilization within the SwissDRG system (sepsis origin, SAPS II, and NEMS), rather than to create prognostically identical cohorts. In real-world clinical practice, CytoSorb was preferentially used in patients with therapy-refractory septic shock and limited remaining treatment options. Despite matching on severity metrics, residual confounding and indication bias cannot be eliminated in a retrospective design. The higher mortality observed in the CytoSorb group is therefore interpreted as reflecting underlying disease severity and treatment selection rather than treatment effect. We have revised the manuscript to emphasize more clearly that this study was not designed or powered to assess survival benefit and that mortality findings must be interpreted cautiously within the observational framework.

 

The following addition has been made to the discussion:

 

“Overall, mortality differences between groups likely reflect residual confounding and indication bias inherent to retrospective observational designs and should not be interpreted as evidence of treatment harm or benefit. As such, these results are rather hypothesis-generating than definitive and support the design of prospective multicenter trials focused on the health economic impact of hemoadsorption therapy in septic and other critically ill populations.”

 

5. regarding the cost and LOS analysis. I see some methodological issues, given the high (and significantly different between the groups) mortality. High mortality actually reduces the cost and LOS for patients that died (dying soon after admission is "cheap"). How was this accounted for? The Cytosorb group had higher mortality and shorter ICU LOS, which both might produce lower costs and result in the observed similar costs between the groups. I presume that with 50% mortality, LOS should be estimated with Kaplan-Meier (censoring for death), some other advanced techniques or only reporting LOS comparison for survivors.

 

We thank the reviewer for this important methodological point and agree that mortality can confound both length of stay and cost analyses, as early death may shorten ICU exposure and thereby reduce resource consumption.

To address this, we did not interpret crude LOS comparisons in isolation. Instead, we stratified the analyses by survivor status (survivors vs. non-survivors) and reported ICU and hospital LOS as well as ventilation duration accordingly. This approach was chosen specifically to reduce the bias introduced by differential mortality between groups, and it allows interpretation of resource use in patients who survived long enough to complete a full ICU/hospital course.

We acknowledge that time-to-event approaches (e.g., Kaplan–Meier estimates or competing-risk models treating death as a competing event for discharge) may provide additional insights, and we agree that “days alive and out of ICU/hospital” could be a robust endpoint in future studies. However, such analyses were not prespecified for this retrospective health-economic evaluation and were not feasible within the available dataset structure. We have now explicitly added this as a limitation and have further tempered the interpretation of LOS and cost findings accordingly. It should also be noted that patients who stayed in hospital for a shorter period than the minimum length of stay (e.g. due to death) are subject to financial deductions in the Swiss DRG system and can therefore become not cheaper, but more expensive.

 

The following addition has been made to the limitations section:

 

“Because death acts as a competing event for discharge, we used survivor-stratified analyses to mitigate mortality confounding; competing-risk/time-to-discharge methods (e.g., Kaplan–Meier–based approaches) and ‘days alive’ endpoints were not applied and should be considered in future prospective studies.“

 

To conclude, I find the analysis poorly matched and of questionable methodological rigour,

which also poses the conclusions under question.

 

Author Response File: Author Response.pdf

Round 2

Reviewer 4 Report

Comments and Suggestions for Authors

I would like to thank the authors for repliying extensively to my remarks. Some issues are resolved, but some crucial are still lacking:

- the authors performed matching on the whole cohort, in the methods they state: "Matching was performed based on sepsis origin ...,SAPS II, and ...NEMS, reflecting clinical severity and nursing workload as key drivers of ICU resource use. Variables such as age, gender, and comorbidities were not included in the matching algorithm, as these were not expected to significantly impact cost-related endpoints within the SwissDRG framework ..." and in their reply "These parameters were deemed most impactful for our specific cost-centered endpoints and are now clearly presented in Table 1," -  but the Table1_R1 shows data for overall cohort (and sepsis origin subgroups and (non)survivors), where there is significant difference in NEMS score, for which the patients were supposed to be matched. This significant difference is also commentated in the results "patients in the CytoSorb group demonstrated a significantly lower cumulative nursing workload on the ICU, as reflected by a >20% reduction in total NEMS points (1624.5 vs. 2054.5; p = 0.015).". Therefore, I do not see results of two comparable groups, based on matching according to the criteria, authors claim to use.

- furthermore, and much more importantly, the cost-effectiveness sub-group was not matched, or the results of matching are not presented. In Table 2, we only have costs, and no age, NEMS, SAPS etc. I do understand that data were only available for certain time periods, but for the cost-effectiveness analysis to be meaningful, patients need to be matched by at least some of the criteria other than the sepsis origin. The authors claim "we have additionally provided baseline characteristics and clarified that cost data availability was limited to the years 2022–2023 due to an accounting system transition" but I do not see such analysis in the manuscript.

- regarding the cost-effectiveness analysis: First of all, I'm looking at absolute amounts: costs per case are about 10.000 CHR, revenues are about 100.000 CHR and net balance is reported as about +15.000 CHF - how do these numbers add up? The balance should be about +90.000 CHF.   Second, the difference in earnings per case was significant in the overall cohort. But you also claim that  "this difference remained significant in the subgroup of survivors and in abdominal sepsis patients." - but in Table 2 the p value for "earnings per case" for survivors is 0.09. The most important issue is, that you claim that 
These are the most basic issues ti be resolved.

I maintain that:

- a cost analysis is not needed for a therapy without proven benefits

• the cost analysis was performed on a sub-group, which seems even less well-matched (different SAPS II, worse mortality in Cytosorb group) than the full cohort and no explanation is given why cost data was only available for a subgroup. This introduces a very high risk of selection bias.
• The high (early) mortality likely significantly reduces the cost per patient. Reported mortality was different between the groups. I do not see how mortality was considered as a competing event to higher costs in the comparison.
• Although the study focused on cost analysis, significantly different mortality between the groups is not something one can ignore and report it in a footnote, if the authors truly believe in effective matching of the cohorts.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf