Abstract
Background: Cytokine-adsorption therapy has garnered attention as a potential treatment for conditions such as sepsis, although supporting evidence remains limited. Consequently, its utilization is expected to vary significantly across regions. To date, no ecological studies have investigated this regional heterogeneity.
Objective: This study aimed to examine temporal trends in the use of continuous renal replacement therapy (CRRT) with cytokine-adsorbing hemofilters and polymyxin-B immobilized fiber-direct hemoperfusion (PMX-DHP), as well as the spatial distribution of both across Japan's 47 prefectures.
Methods: This ecological study analyzed National Database (NDB) open data. A longitudinal analysis from 2016 to 2022 assessed temporal trends in the use of adsorption membranes. A cross-sectional analysis of the 2022 data utilized Moran's I statistic to evaluate the spatial autocorrelation of adsorption therapy. To examine the relationship between the two types of adsorption therapy, we calculated the Pearson correlation coefficient and conducted a multivariate analysis.
Results: The longitudinal analysis revealed no significant change in the proportion of cytokine-adsorbing hemofilter use, while PMX-DHP use showed a decreasing trend over the seven-year period. Cross-sectional analysis indicated spatial autocorrelation for both PMX-DHP (Moran's I: 0.34, P < 0.001) and cytokine-adsorption filter use (Moran's I: 0.24, P < 0.001). Univariate analysis (R = -0.29, P = 0.0453) and multivariate analysis (estimated coefficient: 1.27, 95% CI: 0.06-2.49, P = 0.045) demonstrated that higher usage rates of cytokine-adsorbing blood filters were associated with higher PMX-DHP usage rates.
Conclusions: This study identified a declining trend in PMX-DHP use and an association between PMX-DHP and cytokine-adsorbing hemofilter utilization. These findings suggest that physicians' preferences and perceptions regarding cytokine-adsorption therapy may influence its use. Further research with individual patient data is warranted to confirm these findings.
Keywords: an69st, continuous renal replacement therapy, ecological study, hemoadsorption, ndb open data, pmx-dhp, sepxiris○r
Introduction
In recent years, various forms of cytokine-adsorption therapy have been implemented, with an expanding range of clinical indications [1-11]. Among these therapies, polymyxin-B immobilized fiber-direct hemoperfusion (PMX-DHP), CytoSorb®, and continuous renal replacement therapy (CRRT) with cytokine-adsorption hemofilter are the most commonly used. The efficacy of cytokine-adsorption hemofilters for CRRT has been demonstrated in observational studies [12-15]. However, randomized controlled trials (RCTs) have been small in scale and have shown reductions only in inflammatory substances, not in mortality [4,12-14,16]. Recent RCTs involving PMX-DHP have yielded negative results [8-10]. In 2016, the Surviving Sepsis Campaign Guidelines (SSCG) did not recommend PMX-DHP [17]. However, following the results of the 2018 EUPHRATES study [8], the SSCG revised its recommendation in 2021 to "suggest against using polymyxin-B hemoperfusion" (weak recommendation; low quality of evidence) [18]. Despite the limited evidence supporting cytokine-adsorption therapy, new adsorption treatments are being introduced into clinical practice. Supady et al. indicated in their review that seven cytokine-adsorption therapies and 28 RCTs were underway as of 2022 [19]. They noted, "Considering the current lack of evidence for a benefit of hemoadsorption for the treatment of severe inflammation, sepsis, liver failure, and rhabdomyolysis, its routine use in everyday clinical practice cannot be justified."
A notable contradiction exists between the limited evidence supporting adsorption therapy and the reality of its expanding indications, often determined by the discretion of the treating physician. Treatment methods influenced by individual perspectives may contribute to regional variability; however, no ecological study on adsorption therapy has been conducted to date. Ecological research, which uses regions as the unit of analysis, can clarify spatial distributions and identify region-specific factors [20,21]. While causal inference is challenging due to the cross-sectional nature of such research, it can reveal prevalent interpretation trends in specific regions and highlight the impact of information literacy on clinical decision-making [22-25].
Therefore, we conducted an ecological study using the NDB (National Database of Health Insurance Claims and Specific Health Checkups of Japan) open data [26] to clarify the trends in cytokine-adsorption therapies and the factors influencing them.
Materials and methods
Study design
This ecological study utilizes two open datasets to investigate cytokine-adsorption therapy across Japan's 47 prefectures. The first dataset is the NDB open data [26], a comprehensive nationwide administrative claims database maintained by the Ministry of Health, Labor and Welfare (MHLW), which includes health insurance claims and medical examination records. The second dataset is e-Stat, managed by the Statistics Bureau of the Ministry of Internal Affairs and Communications, which aggregates over 700 datasets from various government ministries and agencies, enabling detailed data searches and outputs [27]. As these datasets do not contain personal information, anonymization was not required.
The study consists of two parts: (i) a longitudinal analysis of NDB open data from the second period (April 1, 2016, to March 31, 2017) to the ninth period (April 1, 2022, to March 31, 2023) to assess trends in PMX-DHP use and the proportion of cytokine-adsorbing hemofilter use; and (ii) a cross-sectional analysis using the ninth NDB open data update along with other open data on a prefecture-by-prefecture basis. In the longitudinal analysis, we excluded the first part of the update due to differences in data availability. This study adhered to the principles of the Declaration of Helsinki. All data used in the analysis were anonymized and did not contain individual patient information. The study protocol was approved by the ethics committee of Tokyo Metropolitan Tama General Medical Center (approval no. 6-122). The study's findings have been reported in accordance with the STROBE (strengthening the reporting of observational studies in epidemiology).
Data collection
The study included all 47 prefectures in Japan. We utilized reports on specific medical materials (annual consumption of normal hemofilters and cytokine-adsorbing hemofilters for CRRT and PMX-DHP) and medical procedures (annual number of PMX-DHP procedures performed) by prefecture from the NDB Open Data. While the usage of PMX-DHP was analyzed as is, the amount of CRRT was expected to influence the amount of adsorbent membrane utilized. Therefore, it was standardized using the following formula: cytokine-adsorbing membrane (AN69ST) use/general membrane use. Additionally, we referenced e-Stat to gather relevant data: population, the proportion of males and proportion of individuals aged 65 and over, the number of physicians, the proportion of intensivists, nephrologists, emergency physicians, areas, and population density. The data on specific medical materials, procedures, and number of physicians were converted to per 100,000 persons.
Statistics
We utilized the second and ninth NDB open data to display trends in the use of cytokine-adsorbing hemofilter use (AN69ST use/total hemofilter use) and PMX-DHP through boxplots. We subsequently performed a cross-sectional analysis using the ninth NDB open dataset. We calculated Moran's I statistics for cytokine-adsorption therapy to assess regional transmissibility. Moran's I statistic is a method for assessing the presence of spatial autocorrelation: a high correlation coefficient indicates similar distributions among neighboring regions, a coefficient near zero signifies a random distribution, and a negative coefficient reveals a "marbled" pattern [28-30]. By smoothing the count of adsorption therapies through a hierarchical Bayesian model, we addressed potential disparities arising from differences in prefectural sizes. Subsequently, we divided the ninth NDB open data for the 47 prefectures into two groups based on the median values of cytokine-adsorbing hemofilter use, categorizing them as high and low use. Scatter plots were created to visualize the relationship between each measurement and the proportion of cytokine-adsorbing hemofilter use as a continuous variable. Pearson's correlation coefficients were calculated to assess these relationships.
We performed a multivariable analysis to examine the relationship between the utilization rates of cytokine-adsorption hemofilters and PMX-DHP. Confounding factors were selected using a directed acyclic graph (DAG), focusing on prefectural characteristics.
In the treatment and specific medical material data from the NDB open data, extremely small numbers of cells (fewer than 20) are reported as missing values. However, the total number of missing values can be obtained. Therefore, we used the mean imputation method, dividing the total number of missing values by the number of missing cells for each variable, and included this in the analysis.
Data were expressed as means and standard deviations where appropriate. The two-tailed significance level was set at p=0.05, and all statistical analyses were performed using R-4.3.1 (R Foundation for Statistical Computing, Vienna, Austria).
Results
We displayed the temporal trend of the annual proportion of cytokine-adsorbing hemofilter use (AN69ST use/total hemofilter use) and PMX-DHP from 2016 to 2022 in Figure 1. The proportion of cytokine-adsorbing hemofilter use neither increased nor decreased, while PMX-DHP use decreased.
Figure 1. Boxplots of the temporal trend of cytokine-adsorption therapies.
The orange box indicates the amount of PMX-DHP use, and the green box indicates the proportion of cytokine-adsorbing filter use (AN69ST use/general filter use).
SSCG: Surviving Sepsis Campaign International Guidelines for Management of Sepsis and Septic Shock; J-SSCG: The Japanese Clinical Practice Guidelines for Management of Sepsis and Septic Shock; PMX-DHP: polymyxin-B immobilized fiber-direct hemoperfusion; AN69ST: acrylonitrile-co-methallyl sulfonate surface-treated; CRRT: continuous renal replacement therapy
Using the ninth NDB open data, we performed cross-sectional analyses, dividing the data for the 47 prefectures into two groups based on the median value of the proportion of cytokine-adsorbing hemofilter use (13.1%). There was a trend toward a larger population, larger areas, fewer physicians, and less CRRT in the high-use group; however, none of these differences were statistically significant (Table 1).
Table 1. Baseline of 47 prefectures in 2022.
* indicates multiplication by 1,000
All data are expressed as means (standard deviation)
CRRT: continuous renal replacement therapy; PMX-DHP: polymyxin-B immobilized fiber-direct hemoperfusion; AN69ST: acrylonitrile-co-methallyl sulfonate surface-treated
| Baseline Data | Low Adsorption Membrane Uses (N=23) | High Adsorption Membrane Uses (N=24) | P-value |
| Prefecture characteristics | |||
| Population (*1,000) | 2081 (2048) | 3262 (3304) | 0.15 |
| Mean age (yrs) | 49 (1.8) | 49 (1.9) | 0.855 |
| Area (km2) | 5916 (3039) | 10079 (15995) | 0.226 |
| Density population (/km2) | 509 (933) | 799 (1455) | 0.421 |
| Aged over 65 (%) | 31 (2.9) | 31 (3.4) | 0.954 |
| Men (%) | 48 (0.8) | 48 (1.1) | 0.528 |
| Number of physicians | |||
| Number of physician (/100,000) | 271 (32.3) | 252 (47) | 0.109 |
| Nephrologist (%) | 1.54 (0.3) | 1.57 (0.6) | 0.855 |
| Emergency physician (%) | 1.18 (0.4) | 1.10 (0.3) | 0.418 |
| Intensivist (%) | 0.28 (0.2) | 0.23 (0.1) | 0.284 |
| Treatment | |||
| PMX-DHP (/100,000) | 3.37 (1.8) | 3.60 (2.2) | 0.7 |
| CRRT (/100,000 person) | 86 (36) | 70 (25) | 0.067 |
| AN69ST (/100,000 person) | 6.31 (4.3) | 19.39 (9.0) | <0.001 |
| Proportion of cytokine-adsorbing hemofilter (AN69ST use/total hemofilter use) | 0.07 (0.1) | 0.22 (0.1) | <0.001 |
Next, we conducted univariable and multivariable analyses with the proportion of cytokine-adsorbing hemofiltration use as the dependent variable. Explanatory variables included PMX-DHP, CRRT, population, and number of physicians, selected based on a DAG. We presented scatterplots showing the relationship between each measurement and the proportion of cytokine-adsorbing hemofilter use as a continuous variable (Figures 2, 3). A higher proportion of cytokine-adsorbing hemofilter use was related to a larger population (R=-0.2, P=0.185), fewer physicians (R=-0.34, P=0.0208), and more PMX-DHP use (R=-0.29, P=0.0453). Correlation coefficients for CRRT, population density, and the proportion of those aged over 65 years were all less than 0.1, and no clear trend could be observed. Statistically significant differences were observed only for PMX-DHP use and the number of physicians.
Figure 2. Scatterplot of PMX-DHP use and the proportion of cytokine-adsorbing hemofilter used for CRRT (AN69ST use/total hemofilter use).
PMX-DHP: polymyxin-B immobilized fiber-direct hemoperfusion; CRRT: continuous renal replacement therapy; AN69ST: acrylonitrile-co-methallyl sulfonate surface-treated
Figure 3. Scatterplot of the proportion of cytokine-adsorbing hemofilter used for CRRT (AN69ST use/total hemofilter use) and each variable.
* indicates multiplication by 1,000
CRRT: continuous renal replacement therapy; AN69ST: acrylonitrile-co-methallyl sulfonate surface-treated
We then performed univariable and multivariable analyses. We set PMX-DHP, CRRT, population, and number of physicians as explanatory variables and the proportion of cytokine-adsorbing hemofilter use as the objective variable. The estimated coefficients, 95% confidence intervals (CI), and P-values for the explanatory variables in the multivariable analysis are presented in Table 2. Statistically significant associations were observed for PMX-DHP use (crude coefficient: 1.40, 95% CI: 0.24 to 2.56, P-value: 0.023; adjusted coefficient: 1.27, 95% CI: 0.06 to 2.49, P-value: 0.045) and the number of physicians (crude coefficient: -0.08, 95% CI: -0.14 to -0.02, P-value: 0.016; adjusted coefficient: -0.08, 95% CI: -0.14 to -0.01, P-value: 0.021). Other variables were not statistically significant. We used a DAG to select the explanatory variables for adjustment in the multivariate analysis (Figure 4).
Table 2. Results of the multivariable analysis with the proportion of cytokine-adsorbing hemofilter use (AN69ST use/total hemofilter use) as the objective variable.
We showed all estimated coefficients and 95% confidence intervals, both per 100,000 population.
CI: confidence interval; CRRT: continuous renal replacement therapy; PMX-DHP: polymyxin-B immobilized fiber-direct hemoperfusion
| Variables | Unadjusted | Adjusted | ||
| Estimated coefficients (95% CI) | P-value | Estimated coefficients (95% CI) | P-value | |
| PMX-DHP | 1.40 (0.24 to 2.56) | 0.023 | 1.27 (0.06 to 2.49) | 0.045 |
| CRRT | 0.00 (-0.05 to 0.06) | 0.889 | 0.01 (-0.06 to 0.07) | 0.877 |
| Population | 0.001 (0 to 0.001) | 0.232 | 0.001 (0 to 0.002) | 0.185 |
| Physician | -0.08 (-0.14 to -0.02) | 0.016 | -0.08 (-0.14 to -0.01) | 0.021 |
Figure 4. Directed acyclic graph.
A directed acyclic graph (DAG) is a tool for organizing hypotheses about causal relationships between variables. This graph illustrates that potential confounding can be controlled by adjusting for CRRT, population size, and the number of physicians.
CRRT: continuous renal replacement therapy
Image created by the authors using DAGitty (https://www.dagitty.net/dags.html)
Finally, we included all variables listed in Table 1 in a multivariable model using a stepwise approach for sensitivity analysis (Table 3). The results showed that the number of physicians lost statistical significance, but the estimated coefficient for PMX-DHP use remained statistically significant, indicating the robustness of the main analysis.
Table 3. Results of multivariable analysis by stepwise approach.
We showed all estimated coefficients and 95% confidence intervals, both per 100,000 population. Akaike information criterion of best model: -94.67.
CI: confidence interval; CRRT: continuous renal replacement therapy; PMX-DHP: polymyxin-B immobilized fiber-direct hemoperfusion
| Variables | Estimated Coefficients (95% CI) | P-value |
| PMX-DHP | 1.48 (0.35 to 2.6) | 0.014 |
| Population | 0.001 (0 to 0.002) | 0.087 |
| Physician | -0.08 (-0.14 to -0.02) | 0.011 |
| Aged over 65 | 0.65 (-0.39 to 1.70) | 0.226 |
Discussion
This ecological study aimed to identify trends and factors associated with cytokine-adsorption therapies in Japan. Longitudinal analysis showed no clear trend of increase or decrease in cytokine-adsorbing hemofilter use; however, PMX-DHP use showed a decreasing trend over a seven-year period. The cross-sectional analysis revealed that the use of PMX-DHP and cytokine-adsorbing hemofilter exhibited spatial autocorrelation, indicating a geographic association between areas where these interventions were utilized.
The 2009 EUPHAS study demonstrated a reduction in mortality with PMX-DHP treatment [11]; however, subsequent findings from three RCTs have refuted its efficacy [8-10]. Although effective subgroups have been explored in recent years (NCT03901807), the impact of the change in recommendation by the SSCG is significant [17,18]. On the other hand, only observational studies or small RCTs were conducted on cytokine-adsorbing hemofilter use for CRRT, and the SSCG did not mention this treatment. Therefore, there was little information for physicians to make informed decisions. The longitudinal changes in PMX-DHP and cytokine-adsorbing hemofilter use observed in our study may be influenced by regional variations in information literacy thresholds. In addition, the implementation of cytokine-adsorbing hemofilters involves lower invasiveness than PMX-DHP, as it merely requires the choice of the hemofilter. The low disadvantage of this treatment might motivate its selection, even if the scientific evidence is limited.
The association of these therapies was observed in cross-sectional analyses. Although both treatments share similarities in cytokine-adsorption, they also have differences: PMX-DHP is primarily used for stabilizing kinetics, whereas CRRT with cytokine-adsorbing hemofilters is used for acute kidney injury patients with hemodynamic instability requiring dialysis. The association of these therapies suggests that physicians may consider them as a single category of cytokine-adsorption therapy, and study results for one treatment might be extrapolated to the other. Additionally, since both therapies involve blood access catheters, they might be performed concurrently. However, the results of PMX-DHP studies [1-3,15] cannot necessarily be applied to CRRT with cytokine-adsorbing hemofilters. The results of this study may provide insights into physicians' interpretations and preferences.
Our study has two strengths. First, it is the first to evaluate the factors associated with cytokine-adsorbing hemofilter use in CRRT, reflecting physicians' thought processes and preferences. Second, it was an all-Japan survey, providing highly comprehensive results. In summary, this study offers valuable insights into the use and perception of cytokine-adsorption therapies in Japan. Despite the limited evidence base, the lower invasiveness and shared characteristics with other cytokine-adsorption treatments influence their use. Future studies should aim to include individual patient data and consider broader geographical contexts to enhance the robustness and generalizability of the findings.
Limitation
Our study also has two limitations. First, the small sample size due to the prefecture-based analysis and the limited number of measured items did not allow for the adjustment of various confounders. Individual patient data would be desirable for more robust analysis and causal inferences. Second, there is no external validity to regions outside of Japan. In Japan, the flow rate of CRRT is often less than in previous studies due to upper insurance limits for dialysate.
Conclusions
This ecological study identified trends and factors associated with cytokine-adsorption therapies in Japan. The longitudinal analysis showed no significant change in cytokine-adsorbing hemofilter use, while PMX-DHP use decreased over seven years. Despite limited evidence supporting the efficacy of cytokine-adsorption therapy, their use continues, likely influenced by clinician preferences and perceptions. Our survey results indicated that CRRT with cytokine-adsorption hemofilters and PMX-DHP tend to be utilized within the same regions, suggesting that region-specific information literacy may influence these treatment choices. This study could provide insights into physicians' interpretations and the factors influencing the use of cytokine-absorption therapies.
Disclosures
Human subjects: All authors have confirmed that this study did not involve human participants or tissue.
Animal subjects: All authors have confirmed that this study did not involve animal subjects or tissue.
Conflicts of interest: In compliance with the ICMJE uniform disclosure form, all authors declare the following:
Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work.
Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work.
Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.
Author Contributions
Concept and design: Shohei Ono
Acquisition, analysis, or interpretation of data: Shohei Ono, Keiki Shimizu
Drafting of the manuscript: Shohei Ono
Critical review of the manuscript for important intellectual content: Shohei Ono, Keiki Shimizu
Supervision: Keiki Shimizu
References
- 1.C-reactive protein apheresis as anti-inflammatory therapy in acute myocardial infarction: results of the CAMI-1 study. Ries W, Torzewski J, Heigl F, et al. Front Cardiovasc Med. 2021;8:591714. doi: 10.3389/fcvm.2021.591714. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Hemofiltration with the Seraph(®) 100 Microbind(®) Affinity filter decreases SARS-CoV-2 nucleocapsid protein in critically ill COVID-19 patients. Kielstein JT, Borchina DN, Fühner T, Hwang S, Mattoon D, Ball AJ. Crit Care. 2021;25:190. doi: 10.1186/s13054-021-03597-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Hemoadsorption with CytoSorb®. Poli EC, Rimmelé T, Schneider AG. Intensive Care Med. 2019;45:236–239. doi: 10.1007/s00134-018-5464-6. [DOI] [PubMed] [Google Scholar]
- 4.Endotoxin and cytokine reducing properties of the oXiris membrane in patients with septic shock: a randomized crossover double-blind study. Broman ME, Hansson F, Vincent JL, Bodelsson M. PLoS One. 2019;14:0. doi: 10.1371/journal.pone.0220444. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Biocompatibility and cytotoxic evaluation of new sorbent cartridges for blood hemoperfusion. Pomarè Montin D, Ankawi G, Lorenzin A, Neri M, Caprara C, Ronco C. Blood Purif. 2018;46:187–195. doi: 10.1159/000489921. [DOI] [PubMed] [Google Scholar]
- 6.In vitro comparison of the adsorption of inflammatory mediators by blood purification devices. Malard B, Lambert C, Kellum JA. Intensive Care Med Exp. 2018;6:12. doi: 10.1186/s40635-018-0177-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Broad adsorption of sepsis-related PAMP and DAMP molecules, mycotoxins, and cytokines from whole blood using CytoSorb® sorbent porous polymer beads. Gruda MC, Ruggeberg KG, O'Sullivan P, et al. PLoS One. 2018;13:0. doi: 10.1371/journal.pone.0191676. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Effect of targeted polymyxin B hemoperfusion on 28-day mortality in patients with septic shock and elevated endotoxin level: the EUPHRATES randomized clinical trial. Dellinger RP, Bagshaw SM, Antonelli M, et al. JAMA. 2018;320:1455–1463. doi: 10.1001/jama.2018.14618. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.The effect of a novel extracorporeal cytokine hemoadsorption device on IL-6 elimination in septic patients: a randomized controlled trial. Schädler D, Pausch C, Heise D, et al. PLoS One. 2017;12:0. doi: 10.1371/journal.pone.0187015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Early use of polymyxin B hemoperfusion in patients with septic shock due to peritonitis: a multicenter randomized control trial. Payen DM, Guilhot J, Launey Y, et al. Intensive Care Med. 2015;41:975–984. doi: 10.1007/s00134-015-3751-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Early use of polymyxin B hemoperfusion in abdominal septic shock: the EUPHAS randomized controlled trial. Cruz DN, Antonelli M, Fumagalli R, et al. JAMA. 2009;301:2445–2452. doi: 10.1001/jama.2009.856. [DOI] [PubMed] [Google Scholar]
- 12.Continuous hemodiafiltration with a cytokine-adsorbing hemofilter in patients with septic shock: a preliminary report. Shiga H, Hirasawa H, Nishida O, et al. Blood Purif. 2014;38:211–218. doi: 10.1159/000369377. [DOI] [PubMed] [Google Scholar]
- 13.The impacts of continuous veno-venous hemofiltration on plasma cytokines and monocyte human leukocyte antigen-DR expression in septic patients. Peng Z, Pai P, Hong-Bao L, Rong L, Han-Min W, Chen H. Cytokine. 2010;50:186–191. doi: 10.1016/j.cyto.2010.02.005. [DOI] [PubMed] [Google Scholar]
- 14.Procalcitonin and proinflammatory cytokine clearance during continuous venovenous haemofiltration in septic patients. Dahaba AA, Elawady GA, Rehak PH, List WF. Anaesth Intensive Care. 2002;30:269–274. doi: 10.1177/0310057X0203000302. [DOI] [PubMed] [Google Scholar]
- 15.Continuous renal replacement therapy with the adsorptive oXiris filter may be associated with the lower 28-day mortality in sepsis: a systematic review and meta-analysis. Wang G, He Y, Guo Q, et al. Crit Care. 2023;27:275. doi: 10.1186/s13054-023-04555-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Effect of CRRT with oXiris filter on hemodynamic instability in surgical septic shock with AKI: a pilot randomized controlled trial. Feng J, Zhang S, Ai T, Wang L, Gao Y, Li W, Zhu M. Int J Artif Organs. 2022;45:801–808. doi: 10.1177/03913988221107947. [DOI] [PubMed] [Google Scholar]
- 17.Surviving sepsis campaign: International Guidelines for Management of Sepsis and Septic Shock: 2016. Rhodes A, Evans LE, Alhazzani W, et al. Intensive Care Med. 2017;43:304–377. doi: 10.1007/s00134-017-4683-6. [DOI] [PubMed] [Google Scholar]
- 18.Surviving sepsis campaign: international guidelines for management of sepsis and septic shock 2021. Evans L, Rhodes A, Alhazzani W, et al. Intensive Care Med. 2021;47:1181–1247. doi: 10.1007/s00134-021-06506-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Extracorporeal haemoadsorption: does the evidence support its routine use in critical care? Supady A, Brodie D, Wengenmayer T. Lancet Respir Med. 2022;10:307–312. doi: 10.1016/S2213-2600(21)00451-3. [DOI] [PubMed] [Google Scholar]
- 20.Methodology series module 7: ecologic studies and natural experiments. Setia MS. Indian J Dermatol. 2017;62:25–28. doi: 10.4103/0019-5154.198048. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Ecologic studies in epidemiology: concepts, principles, and methods. Morgenstern H. Annu Rev Public Health. 1995;16:61–81. doi: 10.1146/annurev.pu.16.050195.000425. [DOI] [PubMed] [Google Scholar]
- 22.Challenging terrains: socio-spatial analysis of Primary Health Care Access Disparities in West Virginia. Hong I, Wilson B, Gross T, Conley J, Powers T. Appl Spat Anal Policy. 2023;16:141–161. doi: 10.1007/s12061-022-09472-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Spatial clusters distribution and modelling of health care autonomy among reproductive-age women in Ethiopia: spatial and mixed-effect logistic regression analysis. Tesema GA, Yeshaw Y, Kasie A, Liyew AM, Teshale AB, Alem AZ. BMC Health Serv Res. 2021;21:74. doi: 10.1186/s12913-020-06052-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Spatial autocorrelation: an overlooked concept in behavioral ecology. Valcu M, Kempenaers B. Behav Ecol. 2010;21:902–905. doi: 10.1093/beheco/arq107. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Spatial autocorrelation analysis of health care hotspots in Taiwan in 2006. Tsai PJ, Lin ML, Chu CM, Perng CH. BMC Public Health. 2009;9:464. doi: 10.1186/1471-2458-9-464. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Ministry of Health, Labour and Welfare. NDB Open Data (in Japanese) [ Jan; 2024 ]. 2024. https://translation.mhlw.go.jp/LUCMHLW/ns/tl.cgi/https https://translation.mhlw.go.jp/LUCMHLW/ns/tl.cgi/https
- 27.e-Stat. [ Oct; 2023 ]. 2024. https://www.e-stat.go.jp/en/stat-search/files https://www.e-stat.go.jp/en/stat-search/files
- 28.An analytical process of spatial autocorrelation functions based on Moran's index. Chen Y. PLoS One. 2021;16:0. doi: 10.1371/journal.pone.0249589. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.New approaches for calculating Moran's index of spatial autocorrelation. Chen Y. PLoS One. 2013;8:0. doi: 10.1371/journal.pone.0068336. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Spatial relationships between polychaete assemblages and environmental variables over broad geographical scales. Benedetti-Cecchi L, Iken K, Konar B, et al. PLoS One. 2010;5:0. doi: 10.1371/journal.pone.0012946. [DOI] [PMC free article] [PubMed] [Google Scholar]