Skip to main content
NCBI home page
PLOS One logoLink to PLOS One
. 2023 Sep 28;18(9):e0291258. doi: 10.1371/journal.pone.0291258

Bayesian network predicted variables for good neurological outcomes in patients with out-of-hospital cardiac arrest

Kota Shinada 1,*, Ayaka Matsuoka 1, Hiroyuki Koami 1, Yuichiro Sakamoto 1
Editor: Gaetano Santulli2
PMCID: PMC10538776  PMID: 37768915

Abstract

Out-of-hospital cardiac arrest (OHCA) is linked to a poor prognosis and remains a public health concern. Several studies have predicted good neurological outcomes of OHCA. In this study, we used the Bayesian network to identify variables closely associated with good neurological survival outcomes in patients with OHCA. This was a retrospective observational study using the Japan Association for Acute Medicine OHCA registry. Fifteen explanatory variables were used, and the outcome was one-month survival with Glasgow–Pittsburgh cerebral performance category (CPC) 1–2. The 2014–2018 dataset was used as training data. The variables selected were identified and a sensitivity analysis was performed. The 2019 dataset was used for the validation analysis. Four variables were identified, including the motor response component of the Glasgow Coma Scale (GCS M), initial rhythm, age, and absence of epinephrine. Estimated probabilities were increased in the following order: GCS M score: 2–6; epinephrine: non-administered; initial rhythm: spontaneous rhythm and shockable; and age: <58 and 59–70 years. The validation showed a sensitivity of 75.4% and a specificity of 95.4%. We identified GCS M score of 2–6, initial rhythm (spontaneous rhythm and shockable), younger age, and absence of epinephrine as variables associated with one-month survival with CPC 1–2. These variables may help clinicians in the decision-making process while treating patients with OHCA.

Introduction

Out-of-hospital cardiac arrest (OHCA) is a public health concern and a condition with poor prognosis [1, 2]. Accurate prognostic prediction of OHCA is important for appropriate resource allocation for emergency medicine and for providing appropriate information to families [1, 3]. Various prediction models have been attempted for a variety of situations [4], including survival prediction [5–9] and good neurological prognosis [10, 11] for patients in whom return of spontaneous circulation (ROSC) has been achieved or target temperature management therapy has been initiated. In recent years, machine learning models have been developed and validated [12–15], further improving the accuracy of OHCA prognosis prediction.

Concomitantly, the risks of making clinical decisions based solely on prognostic models to determine the course of treatment for patients with OHCA have been discussed. Clinical decisions made according to prognostic models are not always accurate, and there is a risk of withholding treatment in potentially life-saving situations if incorrect decisions are made [16]. Furthermore, to use a predictive model, all components included in the model must be in place at the time the predictive model is used. In other words, if even one of the components of the predictive model is not present, the model may not be usable. Variables that are associated with a favorable prognosis have been reported [17]. However, which variables more directly predict a good prognosis has not been clarified. Bayesian networks build graphical models of causal relationships between events based on uncertain information and calculate the probability that the event they wish to estimate will occur from the given information [18, 19]. Compared with other deep learning methods, Bayesian network allows visualization of the relationships among factors and offers high explanatory potential [20]. This method has been widely applied in medicine, primarily in the fields of cardiac, cancer, psychiatric, and pulmonary diseases [21]. In our facility, we employ BayoLinkS (NTT DATA Mathematical Systems Inc., Tokyo, Japan) to estimate the prognosis of emergency patients and for clinical applications.

In this study, we used a Bayesian network to search for variables associated with the event of good neurological prognosis in adult patients with OHCA who had achieved ROSC.

Materials and methods

Study design and participants

This was a retrospective observational study using the Japan Association for Acute Medicine (JAAM) OHCA registry, a prospective observational data registry kept by JAAM, with participating facilities across Japan. The registry was launched on June 1, 2014 and is still accumulating data. As of January 2023, 99 hospitals from 37 of the 47 prefectures in Japan are included in the registry. JAAM OHCA registry collects data following a patient’s arrival at the hospital (available Japanese item from: http://www.jaamohca-web.com/download/. Accessed 1st August 2022). The data individually entered by the hospital is checked by the JAAM OHCA registry committee of experts in emergency medicine and clinical epidemiology, who also perform data cleansing. Moreover, the data were combined with the pre-hospital data from the All-Japan Utstein Registry of the Fire and Disaster Management Agency [22–24].

Patients not resuscitated in the hospital, not linked to pre-hospital records, exogenous cardiac arrest cases, patients who had not achieved ROSC, and patients aged <18 years were excluded. Moreover, cases with missing appropriate data regarding no flow and low flow time (positive value and <400 minutes, respectively), epinephrine administration, GCS M score, blood gas test, and biochemistry test results were also excluded. One-month survival data is routinely collected in both the Fire and Disaster Management Agency Utstein Registry and Japan Association for Acute Medicine OHCA Registry, and there were no cases with missing information.

This study was approved by the Ethics Committee of the Saga University Hospital (Approval no. 2021-04-R-08) and conforms to the tenets of the Declaration of Helsinki. The need for informed consent was waived owing to the retrospective nature of the study.

Variables and outcome

Fifteen variables were used based on previous studies [11, 25–32]: cause of cardiac arrest, age, sex, presence of bystander CPR, presence of witnesses/no flow time, initial emergency medical services (EMS) rhythm, presence of epinephrine administration, low flow time, motor response in the Glasgow coma scale (GCS M), blood gas test results (pH, lactate, glucose) taken after ROSC from the emergency room to admission to the intensive care unit (ICU), and biochemical test results (creatinine, albumin, potassium) taken after the first hospital arrival. The outcome was one-month survival with Glasgow–Pittsburgh cerebral performance category (CPC) 1–2.

Identification of variables closely associated with one-month survival with CPC1-2

The 2014–2018 and 2019 datasets were used as training and validation data, respectively. The following data were used in the analysis in a non-regressive order: cause of cardiac arrest, age and sex, presence of bystander CPR, presence of witnesses/no flow time (time from witnessing to start of CPR), EMS initial rhythm, presence of epinephrine administration, low flow time (time from start of CPR to ROSC), GCS M, and blood gas test and biochemical test results. Variables involved in one-month survival with CPC 1–2 were selected based on the training data, which were subsequently used in the sensitivity analysis.

Statistical analysis

Patient characteristics were analyzed using JMP Pro version 14 (SAS Inc., Cary, NC, USA). Blood test results were divided into three groups using reference values: below reference values, within reference values, and below reference values. The reference values for glucose, creatinine, albumin, and potassium were taken from https://www.jslm.org/books/guideline/2021/GL2021_04.pdf, whereas those for pH and lactate were taken from https://www.acute-care.jp/ja-jp/learning/glossary/bloodgas (both accessed on July 1st, 2023). Except for blood tests, continuous variables were transformed into categorical variables using quartiles. All variables are presented as counts, followed by percentages in parentheses. The comparisons between the training data and the test one were made using the chi-square test. P<0.05 was considered significant. BayoLinkS was used to build and validate the Bayesian network model as well as for the sensitivity analysis.

Results

Of the 57,754 cases enrolled in the study period, 5,340 were included in the analysis; of these, 4,286 and 1,054 cases were used as training and validation sets, respectively (Fig 1). The baseline characteristics and cardiac arrest details are described in Table 1. The training data showed significantly higher levels of low flow time (>39 minutes), lactate (>12.1 mg/dL), creatinine (<0.48 and 0.49–1.08 mg/dL), and albumin (<4.0 g/dL) and significantly lower levels of lactate (5.0–12.0 mg/dL), creatinine (>1.09 mg/dL), and albumin (4.1–5.1 g/dL) than those from the validation data. No significant differences were found for the other items.

Fig 1. Flow diagram of the patient selection procedure.

Fig 1

Open in a new tab

GSC M, motor response in the Glasgow coma scale; JAAM, Japan association for acute medicine; OHCA, out-of-hospital cardiac arrest; ROSC, return of spontaneous circulation.

Table 1. Characteristics of the study population.

Variable All (n = 5,340) Training data (n = 4,286) Test data (n = 1,054) P value
Cause
Cardiac 3,553 (66.5%) 2,835 (66.2%) 718 (68.1%) 0.2294
Noncardiac 1,787 (33.5%) 1,451 (33.9%) 336 (31.9%) 0.2294
Age
<58 years 1,297 (24.3%) 1,039 (24.2%) 258 (24.5%) 0.8727
59–70 years 1,315 (24.6%) 1,078 (25.2%) 237 (22.5%) 0.0727
71–80 years 1,354 (25.4%) 1,078 (25.2%) 276 (26.2%) 0.5018
>81 years 1,374 (25.7%) 1,091 (25.5%) 283 (26.9%) 0.3657
Sex (Female) 1,770 (33.1%) 1,427 (33.3%) 343 (32.5%) 0.6613
Bystander CPR 2,580 (48.3%) 2,068 (48.3%) 512 (48.6%) 0.8635
Bystander defibrillation 376 (7.0%) 308 (7.2%) 68 (6.5%) 0.4209
No flow time / Unwitnessed
0 minutes 1,340 (25.1%) 1,071 (25.0%) 269 (25.5%) 0.7213
1–2 minutes 630 (11.8%) 496 (11.6%) 134 (12.7%) 0.3113
3–7 minutes 877 (16.4%) 703 (16.4%) 174 (16.5%) 0.9261
>8 minutes 937 (17.5%) 767 (17.9%) 170 (16.1%) 0.1899
Unwitnessed 1,556 (29.1%) 1,249 (29.1%) 307 (29.1%) 1.0000
Initial rhythm
Shockable 1,548 (29.0%) 1,260 (29.4%) 288 (27.3%) 0.1976
Pulseless electrical activity 1,632 (30.6%) 1,304 (30.4%) 328 (31.1%) 0.6815
Asystole 1,577 (29.5%) 1,261 (29.4%) 316 (30.0%) 0.7345
Spontaneous rhythm 583 (10.9%) 461 (10.8%) 122 (11.6%) 0.4406
Epinephrine 4,047 (75.8%) 3,251 (75.9%) 796 (75.5%) 0.8410
Physician-staffed EMS 973 (18.2%) 808 (18.9%) 165 (15.7%) 0.0161
Extracorporeal CPR 850 (15.9%) 680 (15.9%) 170 (16.1%) 0.8509
IABP 875 (16.4%) 714 (16.7%) 161 (15.3%) 0.2857
CAG 1,896 (35.5%) 1,519 (35.4%) 377 (35.8%) 0.8575
PCI 926 (17.3%) 752 (17.6%) 174 (16.5%) 0.4404
TTM 1,545 (28.9%) 1,251 (29.2%) 294 (27.9%) 0.4260
Low flow time
<13 minutes 1,255 (23.5%) 984 (23.0%) 271 (25.7%) 0.0622
14–24 minutes 1,312 (24.6%) 1,040 (24.3%) 272 (25.8%) 0.2993
25–38 minutes 1,431 (26.8%) 1,156 (27.0%) 275 (26.1%) 0.5869
>39 minutes 1,342 (25.1%) 1,106 (25.8%) 236 (22.4%) 0.0238
GCS M score
1 4,720 (88.4%) 3,791 (88.5%) 929 (88.1%) 0.7885
2–6 620 (11.6%) 495 (11.6%) 125 (11.9%) 0.7885
pH
<7.349 4,812 (90.1%) 3,874 (90.4%) 938 (89.0%) 0.1851
7.350–7.450 432 (8.1%) 342 (8.0%) 90 (8.5%) 0.5704
>7.451 96 (1.8%) 70 (1.6%) 26 (2.5%) 0.0708
Lactate
<4.9 mg/dL 37 (0.7%) 25 (0.6%) 12 (1.1%) 0.0613
5.0–12.0 mg/dL 117 (2.2%) 73 (1.7%) 44 (4.2%) <0.0001
>12.1 mg/dL 5,186 (97.1%) 4,188 (97.7%) 998 (94.7%) <0.0001
Glucose
<72 mg/dL 271 (5.1%) 224 (5.2%) 47 (4.5%) 0.3472
73–109 mg/dL 259 (4.9%) 207 (4.8%) 52 (4.9%) 0.8730
>110 mg/dL 4,810 (90.1%) 3,855 (89.9%) 955 (90.6%) 0.5653
Creatinine
<0.48 mg/dL 57 (1.1%) 53 (1.2%) 4 (0.4%) 0.0116
0.49–1.08 mg/dL 2,153 (40.3%) 1,764 (41.2%) 389 (36.9%) 0.0117
>1.09 mg/dL 3,130 (58.6%) 2,469 (57.6%) 661 (62.7%) 0.0027
Albumin
<4.0 g/dL 4,797 (89.8%) 3,869 (90.3%) 928 (88.1%) 0.0353
4.1–5.1 g/dL 540 (10.1%) 414 (9.7%) 126 (12.0%) 0.0301
>5.2 g/dL 3 (0.1%) 3 (0.1%) 0 (0.0%) 1.0000
Potassium
<3.5 mmol/L 1,077 (20.2%) 860 (20.1%) 217 (20.6%) 0.7000
3.6–4.8 mmol/L 2,205 (41.3%) 1,795 (41.9%) 410 (38.9%) 0.0809
>4.9 mmol/L 2,058 (38.5%) 1,631 (38.1%) 427 (40.5%) 0.1476
1-month survival with CPC 1–2 1,128 (21.1%) 917 (21.4%) 211 (20.0%) 0.3331
Open in a new tab

Characteristics of the study population including fifteen predictor variables and an outcome were described. All categorical variables are shown as n (%). CAG, coronary angiography; CPC, cerebral performance category; CPR, cardiopulmonary resuscitation; EMS, emergency medical services; GCS M, motor response in the Glasgow coma scale; IABP, intra-aortic balloon pumping; PCI, percutaneous coronary intervention; TTM, target temperature management

Four variables, including GCS M, initial rhythm, age, and absence of epinephrine were chosen for one-month survival with CPC 1–2 in the training model (Fig 2). The estimated probabilities for each combination are presented in S1 Table.

Fig 2. Bayesian network by training set (2014–2018).

Fig 2

Open in a new tab

CPC, cerebral performance category; GCS M, motor response in the Glasgow coma scale.

The results of the sensitivity analysis are shown in Table 2. The estimated probabilities increased in the following order: GCS M score: 2–6; epinephrine: non-administered; initial rhythm: spontaneous rhythm and shockable; and age: <58 and 59–70. In contrast, they decreased in the following order: initial rhythm: asystole; age: 71–80 and >81; epinephrine: administered; initial rhythm: pulseless electrical activity; and GCS M score: 1. The validation analysis showed a sensitivity of 75.4% and a specificity of 95.4% (Table 3).

Table 2. Probability analysis.

Rank Age Initial rhythm Epinephrine GCS M Probability value Gap of probability values
1   2–6 0.72 0.53
2 Non-administered 0.62 0.43
3 Spontaneous rhythm 0.48 0.29
4 Shockable 0.45 0.26
5 <58 0.37 0.19
6 59–70 0.25 0.06
7 0.19 0.00
8 71–80 0.15 -0.03
9 1 0.13 -0.06
10 Pulseless electrical activity 0.10 -0.09
11 Administered 0.07 -0.11
12 >81 0.07 -0.12
13   Asystole     0.03 -0.16
Open in a new tab

GCS M, motor response in the Glasgow coma scale

Table 3. The Bayesian model validation result.

  Predict good outcome Predict poor outcome Sensitivity Specificity Positive predictive value Negative predictive value
Good outcome 159 39 75.4% 95.4% 80.3% 94.0%
Poor outcome 52 804
Open in a new tab

Good outcome: one-month survival with CPC 1–2; Poor outcome: none of one-month survival with CPC 1–2

Discussion

We used a Bayesian network to identify variables associated with good neurological prognosis in adult patients with OHCA who had achieved ROSC and visualize the relationships among the variables. The variables included GCS M score after ROSC, initial rhythm, age, and absence of epinephrine, all of which have been used as components of previous OHCA prognostic variable exploration studies and predictive models (S2 Table).

Some of the predictive models that have been developed and studied to date are highly accurate and have been tested for practicality [4]. For example, the NULL-PLEASE score reported in 2017 [9] has been frequently validated as a prognostic model for OHCA, suggesting that it may perform better than other models [4, 33]. Modifications of the NULL-PLEASE have also been attempted to create models with fewer components [34]. However, the results do not always indicate a good prognosis. Previously, Kjetil et al. argued that a high degree of accuracy is required when considering predictive models for OHCA; however, clinical decisions based solely on predictive models also carry the risk of overlooking potentially life-saving situations [16]. Therefore, we believe that it is important to encourage clinicians to make comprehensive judgments by specifying the priority of variables. This study identified four variables that can be adapted to patients after ROSC and lead to a good neurological prognosis. Factors leading to the four variables were also identified from the Bayesian network model. Knowledge of these favorable prognostic variables may help clinicians to decide which tests and treatments to offer to patients and effectively communicate with their families.

Several reports have suggested that epinephrine increases the likelihood of ROSC; however, it does not affect survival in the long term and it may also worsen neurological prognosis [35, 36]. In this study, the absence of epinephrine was linked to survival with a good neurological prognosis. The first choice of treatment for shockable rhythm is defibrillation, and if ROSC is achieved immediately upon defibrillation, epinephrine is not administered. The prognosis is more favorable in patients with good responsiveness to defibrillation and short time to ROSC.

There are several limitations to this study. First, the Bayesian network analysis is a method unsuitable for continuous variables and the fact that only nominal variables were associated with a good prognosis in this study may be owing to the choice of analysis. Although the variables in this study were chosen from the existing literature, bias may be present in the selection of variables. Important underlying variables in addition to the variables used in the analysis are possible. Furthermore, the researchers specified the order of the nodes, which may have restricted the causal relationship [37]. Second, a total of 8,464 cases were excluded owing to missing data, and thus, the results may not be conclusive for the general population. Possible treatments and treatment protocols may differ depending on the participating facilities. The timing of the GCS M observation and blood sampling may differ among the patients, and the individual variables were not observed at a consistent time. In addition, the blood test measurements used in this study were a mixture of those taken immediately after ROSC and those taken on admission to the ICU after undergoing various treatments. Blood test results can change significantly before and after the treatment of cardiac arrest. Therefore, it may be desirable to standardize the timing of blood tests in all patients. Making comparisons with previously reported prognostic models was also difficult owing to the differences in collectible variables. Last, the interval between CRP initiation and ROSC was relatively short in some patients, making it difficult to evaluate whether these patients had a cardiac arrest. The possibility that some patients were erroneously diagnosed as patients with OHCA cannot be completely ruled out.

Conclusions

Using a Bayesian network, four variables, GCS M score of 2–6 after ROSC, initial rhythm (spontaneous rhythm and shockable), younger age, and absence of epinephrine were shown to be potentially closely associated with good neurological survival. These variables may help clinicians in their overall decision-making.

Supporting information

S1 Table. Estimated probability for one month survival with CPC 1–2.

(XLSX)

Click here for additional data file. (13.3KB, xlsx)
S2 Table. Variables in previous studies.

(XLSX)

Click here for additional data file. (10.8KB, xlsx)

Acknowledgments

The authors would like to acknowledge Editage (www.editage.com) for English language editing.

Data Availability

The data are owned by a third party. Data are available from the JAAM-OHCA registry committee (contact via http://www.jaamohca-web.com/) for researchers who meet the criteria for access to confidential data.

Funding Statement

The authors received no specific funding for this work.

References

  • 1.Kim HS, Park KN, Kim SH, Lee BK, Oh SH, Jeung KW, et al. Prognostic value of OHCA, C-GRApH and CAHP scores with initial neurologic examinations to predict neurologic outcomes in cardiac arrest patients treated with targeted temperature management. PLoS ONE. 2020;15(4):e0232227. doi: 10.1371/journal.pone.0232227 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Wu JZ, Chiu WC, Wu WT, Chiu IM, Huang KC, Hung CW, et al. Clinical validation of cardiac arrest hospital prognosis (CAHP) score and MIRACLE2 score to predict neurologic outcomes after out-of-hospital cardiac arrest. Healthcare (Basel). 2022;10(3):578. doi: 10.3390/healthcare10030578 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Chae MK, Lee SE, Min YG, Park EJ. Initial serum cholesterol level as a potential marker for post cardiac arrest patient outcomes. Resuscitation. 2020;146:50–55. doi: 10.1016/j.resuscitation.2019.11.003 [DOI] [PubMed] [Google Scholar]
  • 4.Gue YX, Adatia K, Kanji R, Potpara T, Lip GYH, Gorog DA. Out-of-hospital cardiac arrest: A systematic review of current risk scores to predict survival. Am Heart J. 2021;234:31–41. doi: 10.1016/j.ahj.2020.12.011 [DOI] [PubMed] [Google Scholar]
  • 5.Glober NK, Lardaro T, Christopher S, Tainter CR, Weinstein E, Kim D. Validation of the NUE rule to predict futile resuscitation of out-of-hospital cardiac arrest. Prehosp Emerg Care. 2021;25(5):706–711. doi: 10.1080/10903127.2020.1831666 [DOI] [PubMed] [Google Scholar]
  • 6.Gue YX, Sayers M, Whitby BT, Kanji R, Adatia K, Smith R, et al. Usefulness of the NULL-PLEASE score to predict survival in out-of-hospital cardiac arrest. Am J Med. 2020;133(11):1328–1335. doi: 10.1016/j.amjmed.2020.03.046 [DOI] [PubMed] [Google Scholar]
  • 7.Ji C, Brown TP, Booth SJ, Hawkes C, Nolan JP, Mapstone J, et al. Risk prediction models for out-of-hospital cardiac arrest outcomes in England. Eur Heart J Qual Care Clin Outcomes. 2021;7(2):198–207. doi: 10.1093/ehjqcco/qcaa019 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Wong XY, Ang YK, Li K, Chin YH, Lam SSW, Tan KBK, et al. Development and validation of the SARICA score to predict survival after return of spontaneous circulation in out of hospital cardiac arrest using an interpretable machine learning framework. Resuscitation. 2022;170:126–133. doi: 10.1016/j.resuscitation.2021.11.029 [DOI] [PubMed] [Google Scholar]
  • 9.Ahmad R, Lumley S, Lau YC. NULL-PLEASE: a new “Futility score” in the management of survivors of out-of-hospital cardiac arrest. Resuscitation. 2016;106(Suppl 1):e83. doi: 10.1016/j.resuscitation.2016.07.201 [DOI] [Google Scholar]
  • 10.Nishioka N, Kobayashi D, Kiguchi T, Irisawa T, Yamada T, Yoshiya K, et al. Development and validation of early prediction for neurological outcome at 90 days after return of spontaneous circulation in out-of-hospital cardiac arrest. Resuscitation. 2021;168:142–150. doi: 10.1016/j.resuscitation.2021.09.027 [DOI] [PubMed] [Google Scholar]
  • 11.Seewald S, Wnent J, Lefering R, Fischer M, Bohn A, Jantzen T, et al. CaRdiac Arrest Survival Score (CRASS)—A tool to predict good neurological outcome after out-of-hospital cardiac arrest. Resuscitation. 2020;146:66–73. doi: 10.1016/j.resuscitation.2019.10.036 [DOI] [PubMed] [Google Scholar]
  • 12.Cheng CY, Chiu IM, Zeng WH, Tsai CM, Lin CHR. Machine learning models for survival and neurological outcome prediction of out-of-hospital cardiac arrest patients. Biomed Res Int. 2021;2021:9590131. doi: 10.1155/2021/9590131 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Kwon JM, Jeon KH, Kim HM, Kim MJ, Lim S, Kim KH, et al. Deep-learning-based out-of-hospital cardiac arrest prognostic system to predict clinical outcomes. Resuscitation. 2019;139:84–91. doi: 10.1016/j.resuscitation.2019.04.007 [DOI] [PubMed] [Google Scholar]
  • 14.Lo YH, Siu YCA. Predicting survived events in nontraumatic out-of-hospital cardiac arrest: A comparison study on machine learning and regression models. J Emerg Med. 2021;61:683–694. doi: 10.1016/j.jemermed.2021.07.058 [DOI] [PubMed] [Google Scholar]
  • 15.Seo DW, Yi H, Bae HJ, Kim YJ, Sohn CH, Ahn S, et al. Prediction of neurologically intact survival in cardiac arrest patients without pre-hospital return of spontaneous circulation: machine learning approach. J Clin Med. 2021;10:1089. doi: 10.3390/jcm10051089 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Sunde K, Kramer-Johansen J, Pytte M, Steen PA. Predicting survival with good neurologic recovery at hospital admission after successful resuscitation of out-of-hospital cardiac arrest: the OHCA score. Eur Heart J. 2007;28(6):773; author reply 773–774. doi: 10.1093/eurheartj/ehl536 [DOI] [PubMed] [Google Scholar]
  • 17.Al-Dury N, Ravn-Fischer A, Hollenberg J, Israelsson J, Nordberg P, Strömsöe A, et al. Identifying the relative importance of predictors of survival in out of hospital cardiac arrest: A machine learning study. Scand J Trauma Resusc Emerg Med. 2020;28(1):60. doi: 10.1186/s13049-020-00742-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Hwang S, Boyle LN, Banerjee AG. Identifying characteristics that impact motor carrier safety using Bayesian networks. Accid Anal Prev. 2019;128:40–45. doi: 10.1016/j.aap.2019.03.004 [DOI] [PubMed] [Google Scholar]
  • 19.Kopacheva E. Predicting online participation through Bayesian network analysis. PLoS ONE. 2021;16(12):e0261663. doi: 10.1371/journal.pone.0261663 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Kyrimi E, Neves MR, McLachlan S, Neil M, Marsh W, Fenton N. Medical idioms for clinical Bayesian network development. J Biomed Inform. 2020;108:103495. doi: 10.1016/j.jbi.2020.103495 [DOI] [PubMed] [Google Scholar]
  • 21.McLachlan S, Dube K, Hitman GA, Fenton NE, Kyrimi E. Bayesian networks in healthcare: Distribution by medical condition. Artif Intell Med. 2020;107:101912. doi: 10.1016/j.artmed.2020.101912 [DOI] [PubMed] [Google Scholar]
  • 22.Hatakeyama T, Kiguchi T, Sera T, Nachi S, Ochiai K, Kitamura T, et al. Physician’s presence in pre-hospital setting improves one-month favorable neurological survival after out-of-hospital cardiac arrest: A propensity score matching analysis of the JAAM-OHCA Registry. Resuscitation. 2021;167:38–46. doi: 10.1016/j.resuscitation.2021.08.010 [DOI] [PubMed] [Google Scholar]
  • 23.Irisawa T, Matsuyama T, Iwami T, Yamada T, Hayakawa K, Yoshiya K, et al. The effect of different target temperatures in targeted temperature management on neurologically favorable outcome after out-of-hospital cardiac arrest: A nationwide multicenter observational study in Japan (the JAAM-OHCA registry). Resuscitation. 2018;133:82–87. doi: 10.1016/j.resuscitation.2018.10.004 [DOI] [PubMed] [Google Scholar]
  • 24.Oskarsson A, Fowler BA. Effects of lead inclusion bodies on subcellular distribution of lead in rat kidney: the relationship to mitochondrial function. Exp Mol Pathol. 1985;43(3):397–408. doi: 10.1016/0014-4800(85)90076-0 [DOI] [PubMed] [Google Scholar]
  • 25.Maupain C, Bougouin W, Lamhaut L, Deye N, Diehl JL, Geri G, et al. The CAHP (Cardiac Arrest Hospital Prognosis) score: A tool for risk stratification after out-of-hospital cardiac arrest. Eur Heart J. 2016;37(42):3222–3228. doi: 10.1093/eurheartj/ehv556 [DOI] [PubMed] [Google Scholar]
  • 26.Nishikimi M, Matsuda N, Matsui K, Takahashi K, Ejima T, Liu K, et al. A novel scoring system for predicting the neurologic prognosis prior to the initiation of induced hypothermia in cases of post-cardiac arrest syndrome: the CAST score. Scand J Trauma Resusc Emerg Med. 2017;25(1):49. doi: 10.1186/s13049-017-0392-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Kiehl EL, Parker AM, Matar RM, Gottbrecht MF, Johansen MC, Adams MP, et al. C-GRApH: A validated scoring system for early stratification of neurologic outcome after out-of-hospital cardiac arrest treated with targeted temperature management. J Am Heart Assoc. 2017;6(5):e003821. doi: 10.1161/JAHA.116.003821 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Lim HJ, Ro YS, Kim KH, Park JH, Hong KJ, Song KJ, et al. The ED-PLANN score: A simple risk stratification tool for out-of-hospital cardiac arrests derived from emergency departments in Korea. J Clin Med. 2021;11(1):174. doi: 10.3390/jcm11010174 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Adrie C, Cariou A, Mourvillier B, Laurent I, Dabbane H, Hantala F, et al. Predicting survival with good neurological recovery at hospital admission after successful resuscitation of out-of-hospital cardiac arrest: the OHCA score. Eur Heart J. 2006;27(23):2840–2845. doi: 10.1093/eurheartj/ehl335 [DOI] [PubMed] [Google Scholar]
  • 30.Nishikimi M, Ogura T, Nishida K, Takahashi K, Nakamura M, Matsui S, et al. External validation of a risk classification at the emergency department of post-cardiac arrest syndrome patients undergoing targeted temperature management. Resuscitation. 2019;140:135–141. doi: 10.1016/j.resuscitation.2019.05.028 [DOI] [PubMed] [Google Scholar]
  • 31.Pareek N, Kordis P, Beckley-Hoelscher N, Pimenta D, Kocjancic ST, Jazbec A, et al. A practical risk score for early prediction of neurological outcome after out-of-hospital cardiac arrest: MIRACLE2. Eur Heart J. 2020;41(47):4508–4517. doi: 10.1093/eurheartj/ehaa570 [DOI] [PubMed] [Google Scholar]
  • 32.Shida H, Matsuyama T, Iwami T, Okabayashi S, Yamada T, Hayakawa K, et al. Serum potassium level on hospital arrival and survival after out-of-hospital cardiac arrest: The CRITICAL study in Osaka, Japan. Eur Heart J Acute Cardiovasc Care. 2020;9(4_suppl):S175–S183. doi: 10.1177/2048872619848883 [DOI] [PubMed] [Google Scholar]
  • 33.Tsuchida T, Ono K, Maekawa K, Wada T, Katabami K, Yoshida T, et al. Simultaneous external validation of various cardiac arrest prognostic scores: a single-center retrospective study. Scand J Trauma Resusc Emerg Med. 2021;29(1):117. doi: 10.1186/s13049-021-00935-w [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Potpara TS, Mihajlovic M, Stankovic S, Jozic T, Jozic I, Asanin MR, et al. External validation of the dimple NULL-PLEASE clinical score in predicting outcome of out-of-hospital cardiac arrest. Am J Med. 2017;130(12):1464.e13-1464.e21. doi: 10.1016/j.amjmed.2017.05.035 [DOI] [PubMed] [Google Scholar]
  • 35.Jacobs IG, Finn JC, Jelinek GA, Oxer HF, Thompson PL. Effect of adrenaline on survival in out-of-hospital cardiac arrest: A randomised double-blind placebo-controlled trial. Resuscitation. 2011;82(9):1138–1143. doi: 10.1016/j.resuscitation.2011.06.029 [DOI] [PubMed] [Google Scholar]
  • 36.Olasveengen TM, Sunde K, Brunborg C, Thowsen J, Steen PA, Wik L. Intravenous drug administration during out-of-hospital cardiac arrest: A randomized trial. JAMA. 2009;302(20):2222–2229. doi: 10.1001/jama.2009.1729 [DOI] [PubMed] [Google Scholar]
  • 37.Kim J, Kim YJ, Han S, Choi HJ, Moon H, Kim G. Effect of prehospital epinephrine on outcomes of out-of-hospital cardiac arrest: A Bayesian network approach. Emerg Med Int. 2020;2020:8057106. doi: 10.1155/2020/8057106 [DOI] [PMC free article] [PubMed] [Google Scholar]

Decision Letter 0

Gaetano Santulli

2 May 2023

PONE-D-23-03455Bayesian network predicted variables for good neurological outcomes in patients with out-of-hospital cardiac arrestPLOS ONE

Dear Dr. Shinada,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Jun 16 2023 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

  • A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

  • A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

  • An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Gaetano Santulli

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at 

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and 

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Partly

Reviewer #3: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: I Don't Know

Reviewer #3: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Major Point

The authors discussed about OHCA registry with Bayesian network.

To the readers of this paper, I guess that the explanation of machine learning and difference between Bayesian network and deep learning are insufficient.

Will this paper have more meanings more than that authors have used Bayesian network.

Please explain this point in detail more.

Reviewer #2: In this analysis, Shinada et al. aimed at finding and validating a subset of criteria for the prediction of survival at 1-month with good neurological outcome in OHCA patients. From a pool of fifteen variables selected by the authors based on previous publications, predictors of survival with a good neurological outcome were identified using a Bayesian network.

The authors found that 4 variables, namely age, initial rhythm, presence of epinephrine, and GCS M were associated with one-month survival with CPC 1–2.

The topic is interesting and the database large and apparently of good quality.

However, there are several major issues to be resolved:

1) the Registry and the methods of data collection should be better described: how many centers participate to the Japan Association for Acute 56 Medicine (JAAM) OHCA registry? What is the population of the area served by the facilities participating to the registry? Are the participating centers able to provide all the necessary treatments to OHCA patients? Please provide a more detailed description of the system as recommended by the Utstein criteria (https://doi.org/10.1161/CIR.0000000000000144) should be provided.

2) how was the clinical endpoint assessed? Was a study-specific follow-up conducted? or was survival and GCS assessed solely based on health records? How many patients had complete follow-up information?

3) The description of the Bayesian method should be much more detailed. Was a biostatistician involved?

4) It is a major limitation that continuous variables (eg. pH, glucose, creatinine) were categorized into quartiles resulting in cut-offs that have no relationship with validated cut-offs used in clinical activity. Probably a second method should be used to identify predictors of good neurological outcome.

5) A definition of intrinsic causes of cardiac arrest should be provided. Do the authors mean medical non traumatic causes?

6) The authors should provide the references, on which the selection of the 15 variables was based. It is a major limitation that the variables were selected based on previous literature and authors' judgement.

6) The results should be contextualized within the existing literature. In particular the authors should compare the findings of the current analysis with other available risk assessment tools (variables selected, sensitivity and specificity).

7) Additional information on baseline characteristics and interventions (e.g. CPR provided by emergency medical service, mechanical CPR, coronary angiography, PCI, therapeutic hypothermia), timing of death (KM curves) according to the Utstein criteria (https://doi.org/10.1161/CIR.0000000000000144) should be provided

8) English language should be improved. Examples:

- Line 99: “with a 100 dominance level set at 5%”: what do you mean?

- Line 107: “the patient background information”: i suggest "baseline characteristics and cardiac arrest details"

- Line 149: exacerbate the neurological prognosis: "worsen" or "affect" would be more appropriate

9) Please provide the measure units of the laboratory values.

10) Conclusions should be more informative: rather than "age, presence of prehospital adrenaline, GCS M after ROSC, and initial rhythm" i suggest to report older age, absence of prehospital adrenaline, type of rhythm (shockable vs. non shockable).

Reviewer #3: Comments

This study examined factors associated with good neurological outcome in patients with out-of-hospital cardiac arrest using a Bayesian network. Prediction of prognosis after resuscitation of patients with out-of-hospital cardiac arrest is a critical issue for treatment selection and explanation to patients' families. The strengths of this study are using large-scale data from the Japan Association for Acute Medicine OHCA registry and the All-Japan Utstein Registry of the Fire and Disaster Management Agency and using a Bayesian network.

On the other hand, the novelty of this study is somewhat weak, as several similar studies have been reported. In addition, although the authors identified variables that are particularly important in predicting the prognosis of out-of-hospital cardiac arrest, the significance of this identification is difficult to convey to the reader.

Major comments

41-43 Please describe the problems that should be improved in the prognostic and predictive scores of OHCA reported so far. Then, please describe your motivation for conducting this study to address these problems.

116 What is the basis or criteria for setting the values of No flow time and GCS M cutoffs in Table.1, please describe details in the statistic.

151-155 The authors discussed that some cases with ROSC after a relatively short CPR time were included; the prognosis may be better if ROSC does not require adrenaline, and it may also include patients in whom CPR was initiated because they were wrongly assessed to have an OHCA. However, no data are presented to show the relationship between low flow time and adrenaline administration. Also, data are not provided on how many patients were incorrectly evaluated for OHCA.

157-164 The authors discussed that the four variables identified using the Bayesian network may help clinicians in making decisions on the prognosis of OHCA and the overall treatment strategy. Please describe how you would like clinicians to use these four variables to make decisions.

Minor comments

67-69 It is better to have a light explanation of the meaning of no flow and low flow at the beginning because it is difficult for some readers to understand.

76 Please list citations for the 15 variables.

78 Please standardize the terminology to either adrenaline or epinephrine.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2023 Sep 28;18(9):e0291258. doi: 10.1371/journal.pone.0291258.r002

Author response to Decision Letter 0

12 Jul 2023

Reply to Reviewer 1’s comments

>We appreciate the reviewer’s comments. We have tried to incorporate the reviewer’s suggestions as much as possible, but welcome any additional comments that the reviewer may have.

The authors discussed about OHCA registry with Bayesian network.

To the readers of this paper, I guess that the explanation of machine learning and difference between Bayesian network and deep learning are insufficient.

Will this paper have more meanings more than that authors have used Bayesian network.

Please explain this point in detail more.

>We thank the reviewer for the suggestion. We have added an explanation in the Introduction regarding the significance of using Bayesian networks in this analysis, as well as a comparison with other deep learning methods (lines 39–53).

Replies to Reviewer 2’s comments

>We appreciate the reviewer’s comments. We have tried to incorporate the suggestions as much as possible, but welcome any additional comments that the reviewer may have.

1) the Registry and the methods of data collection should be better described: how many centers participate to the Japan Association for Acute 56 Medicine (JAAM) OHCA registry? What is the population of the area served by the facilities participating to the registry? Are the participating centers able to provide all the necessary treatments to OHCA patients? Please provide a more detailed description of the system as recommended by the Utstein criteria (https://doi.org/10.1161/CIR.0000000000000144) should be provided.

>We thank the reviewer for the suggestion. We have added a statement regarding the number of participating facilities (lines 61–62). However, it was difficult to confirm the population sample for the entire region. The registry includes a variety of hospitals and there may be differences in the care that can be provided. We have also added a statement stating the same in the limitations paragraph (lines 177).

2) how was the clinical endpoint assessed? Was a study-specific follow-up conducted? or was survival and GCS assessed solely based on health records? How many patients had complete follow-up information?

>We thank the reviewer for the question. There is no follow-up specific to the outcomes of this study. In Japan, one-month survival data is routinely collected in both the Fire and Disaster Management Agency Utstein Registry and Japan Association for Acute Medicine OHCA Registry. GCS and one-month survival was recorded by the respective centers; there were 5 missing cases regarding GCS M (Figure 1) and no cases with missing information on one-month survival.

3) The description of the Bayesian method should be much more detailed. Was a biostatistician involved?

>We have added a part on the Bayesian network method (lines 46–53). No biostatistician was involved in the study.

4) It is a major limitation that continuous variables (eg. pH, glucose, creatinine) were categorized into quartiles resulting in cut-offs that have no relationship with validated cut-offs used in clinical activity. Probably a second method should be used to identify predictors of good neurological outcome.

>We thank the reviewer for the suggestion. We have reanalyzed the blood test results by dividing them into categorical variables with clinically relevant values (lines 98–102).

5) A definition of intrinsic causes of cardiac arrest should be provided. Do the authors mean medical non traumatic causes?

>We thank the reviewer for pointing this out. We have replaced "intrinsic except cardiogenic" with "non-cardiogenic" because extrinsic cardiac arrest was excluded in the first place (Table 1).

6) The authors should provide the references, on which the selection of the 15 variables was based. It is a major limitation that the variables were selected based on previous literature and authors' judgement.

>We thank the reviewer for pointing this out. We have added some references (line 78) as per the reviewer’s suggestion. We have also mentioned the possibility of bias due to variable selection in the limitations (line 172–174).

6) The results should be contextualized within the existing literature. In particular the authors should compare the findings of the current analysis with other available risk assessment tools (variables selected, sensitivity and specificity).

>We thank the reviewer for the comment. We have added text on the comparison with previous studies for the selected variables in Table S2 . NULL-PLEASE scores are reported as excellent in previous systematic reviews (Gue YX, et al. Out-of-hospital cardiac arrest: A systematic review of current risk scores to predict survival. Am Heart J. 2021;234:31-41.). We attempted to compare the findings with the NULL-PLEASE score; however, the NULL-PLEASE score included variables that were not collected in the OHCA registry, making actual comparison difficult. We have added the lack of comparison with existing scores to the limitations paragraph (lines 182–184).

7) Additional information on baseline characteristics and interventions (e.g. CPR provided by emergency medical service, mechanical CPR, coronary angiography, PCI, therapeutic hypothermia), timing of death (KM curves) according to the Utstein criteria (https://doi.org/10.1161/CIR.0000000000000144) should be provided

>We thank the reviewer for pointing this out. We have added as much additional information as we could find: bystander defibrillation, physician-staffed EMS, ECPR, IABP, CAG, PCI, and TTM (Table 1). It was difficult to draw a KM curve for timing of death because of missing data.

8) English language should be improved. Examples:

- Line 99: “with a 100 dominance level set at 5%”: what do you mean?

- Line 107: “the patient background information”: i suggest "baseline characteristics and cardiac arrest details"

- Line 149: exacerbate the neurological prognosis: "worsen" or "affect" would be more appropriate

>We thank the reviewer for pointing this out.

- We have corrected the first statement to "P<0.05 was considered significant." (line 104-105)

- Corrected as suggested (lines 110–111).

- Corrected as suggested (line 154).

9) Please provide the measure units of the laboratory values.

>We thank the reviewer for the comment. We have added the unit of measure (Table 1).

10) Conclusions should be more informative: rather than "age, presence of prehospital adrenaline, GCS M after ROSC, and initial rhythm" i suggest to report older age, absence of prehospital adrenaline, type of rhythm (shockable vs. non shockable).

>We thank the reviewer for the comment. We have revised the conclusions as suggested (lines 25–26 and 189–190)

Reply to Reviewer 3’s comments

>We appreciate the reviewer’s comments. We have tried to incorporate the suggestions as much as possible, but welcome any additional comments the reviewer may have.

41-43 Please describe the problems that should be improved in the prognostic and predictive scores of OHCA reported so far. Then, please describe your motivation for conducting this study to address these problems.

>We thank the reviewer for the suggestion. We have added a description in the Introduction (lines 39–53).

116 What is the basis or criteria for setting the values of No flow time and GCS M cutoffs in Table.1, please describe details in the statistic.

>We thank the reviewer for pointing this out. Blood test items with reference values were divided by the reference value, but variables without reference values, such as the NFT and GCS M, were divided by quartiles. We have added corresponding text in the revised manuscript (lines 98–103).

151-155 The authors discussed that some cases with ROSC after a relatively short CPR time were included; the prognosis may be better if ROSC does not require adrenaline, and it may also include patients in whom CPR was initiated because they were wrongly assessed to have an OHCA. However, no data are presented to show the relationship between low flow time and adrenaline administration. Also, data are not provided on how many patients were incorrectly evaluated for OHCA.

>We thank the reviewer for the comment. As pointed out by the reviewer, there are no data to support this relationship; therefore, we have deleted the relevant text from the manuscript.

157-164 The authors discussed that the four variables identified using the Bayesian network may help clinicians in making decisions on the prognosis of OHCA and the overall treatment strategy. Please describe how you would like clinicians to use these four variables to make decisions.

>We thank the reviewer for the advice. We have added a part in the manuscript as per the suggestions (lines 163–169).

Minor comments

67-69 It is better to have a light explanation of the meaning of no flow and low flow at the beginning because it is difficult for some readers to understand.

>We thank the reviewer for the comment. We have added a brief explanation in the revised manuscript (lines 90–92).

76 Please list citations for the 15 variables.

>We thank the reviewer for pointing this out. We have added citations as suggested (line 78).

78 Please standardize the terminology to either adrenaline or epinephrine.

>We thank the reviewer for the comment. We have replaced “adrenaline” with "epinephrine" in each instance.

Attachment

Submitted filename: Response to Reviewers.docx

Click here for additional data file. (28KB, docx)

Decision Letter 1

Gaetano Santulli

28 Jul 2023

PONE-D-23-03455R1Bayesian network predicted variables for good neurological outcomes in patients with out-of-hospital cardiac arrestPLOS ONE

Dear Dr. Shinada,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Sep 11 2023 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

  • A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

  • A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

  • An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Gaetano Santulli, MD

Academic Editor

PLOS ONE

Journal Requirements:

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

Reviewer #3: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: I Don't Know

Reviewer #3: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Previous comment is replied appropriately about the explanation of machine learning and difference between Bayesian　network and deep learning. I think that this paper is acceptable.

Reviewer #2: I thank the authors for addressing the majority of my concerns.

I have few additional comments:

1) Line 24: I suggest to replace “non-cardiac arrest” with “spontaneous rhythm” or “ROSC”; the label “non-cardiac arrest” as type of rhythm is very confusing.

2) Table 1: I suggest to replace “cardiogenic and non cardiogenic cause” with “cardiac or noncardiac” in agreement with the Utstein guidelines (https://doi.org/10.1161/CIR.0000000000000144)

3) The authors answered to a previous comment: “There is no follow-up specific to the outcomes of this study. In Japan, one-month survival data is routinely collected in both the Fire and Disaster Management Agency Utstein Registry and Japan Association for Acute Medicine OHCA Registry. GCS and one-month survival was recorded by the respective centers; there were 5 missing cases regarding GCS M (Figure 1) and no cases with missing information on one-month survival.”

This information should be reported in the methods.

4) Please mention in the discussion previous studies that assessed the sensitivity and specificity of other scores, such as the NULL PLEASE

Reviewer #3: The author generally answered all questions somewhat adequately and the arguments were easy to understand.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2023 Sep 28;18(9):e0291258. doi: 10.1371/journal.pone.0291258.r004

Author response to Decision Letter 1

23 Aug 2023

August 24 2023

Emily Chenette

Editor-in-Chief

PLoS One

Dear Editor-in-Chief:

We thank you for considering our paper titled “Bayesian network predicted variables for good neurological outcomes in patients with out-of-hospital cardiac arrest,” manuscript ID, PONE-D-23-03455R1, for publication in PLoS One. We are hereby re-submitting the revised version of our manuscript.

We are grateful for the feedback provided by the reviewers, which has helped improve the quality of our paper. In accordance with the valuable comments, we have made corrections and additions to the text and tables.

The corrections and additions in the text are highlighted in yellow in the revised manuscript.

We have tried to incorporate your suggestions as much as possible; however, if you have any further suggestions, please let us know.

Thank you for your constructive comments.

Sincerely,

Kota Shinada

Department of Emergency and Critical Care Medicine,

Faculty of Medicine, Saga University

5-1-1 Nabeshima

Saga City, Saga Prefecture 849-8501, Japan

Phone number: +81-952-34-3160

Fax number: +81-952-34-1061

Email address: st9137@cc.saga-u.ac.jp  

Reply to Reviewer 1’s comments

Previous comment is replied appropriately about the explanation of machine learning and difference between Bayesian　network and deep learning. I think that this paper is acceptable.

Response: Thank you for your confirmation and comments.

Reply to Reviewer 2’s comments

I thank the authors for addressing the majority of my concerns.

I have few additional comments:

Response: Thank you very much for your confirmation and additional constructive comments.

1) Line 24: I suggest to replace “non-cardiac arrest” with “spontaneous rhythm” or “ROSC”; the label “non-cardiac arrest” as type of rhythm is very confusing.

Response: We thank the reviewer for the suggestion. We have accordingly made corrections to the lines. (Lines 24, 26, 139, and 196 and tables 1, 2, and S1)

2) Table 1: I suggest to replace “cardiogenic and non cardiogenic cause” with “cardiac or noncardiac” in agreement with the Utstein guidelines (https://doi.org/10.1161/CIR.0000000000000144)

Response: We thank the reviewer for the suggestion. We have accordingly made the corrections. (Table 1)

3) The authors answered to a previous comment: “There is no follow-up specific to the outcomes of this study. In Japan, one-month survival data is routinely collected in both the Fire and Disaster Management Agency Utstein Registry and Japan Association for Acute Medicine OHCA Registry. GCS and one-month survival was recorded by the respective centers; there were 5 missing cases regarding GCS M (Figure 1) and no cases with missing information on one-month survival.”

This information should be reported in the methods.

Response: We thank the reviewer for pointing this out. We have included this in the Methods section. (Lines 72-74)

4) Please mention in the discussion previous studies that assessed the sensitivity and specificity of other scores, such as the NULL PLEASE

Response: We thank the reviewer for the suggestion. Citing a systematic review, we have mentioned the predictive performance of the NULL-PLEASE score in the Discussion section. However, it was difficult to mention the sensitivity and specificity as they were not mentioned in most previous studies, although AUC was mentioned.

Reply to Reviewer 3’s comments

The author generally answered all questions somewhat adequately and the arguments were easy to understand.

Response: Thank you for your confirmation and comments.

Attachment

Submitted filename: Response_to_Reviewers.docx

Click here for additional data file. (19.8KB, docx)

Decision Letter 2

Gaetano Santulli

25 Aug 2023

Bayesian network predicted variables for good neurological outcomes in patients with out-of-hospital cardiac arrest

PONE-D-23-03455R2

Dear Dr. Shinada,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Gaetano Santulli, MD

Academic Editor

PLOS ONE

Acceptance letter

Gaetano Santulli

18 Sep 2023

PONE-D-23-03455R2

Bayesian network predicted variables for good neurological outcomes in patients with out-of-hospital cardiac arrest

Dear Dr. Shinada:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Professor Gaetano Santulli

Academic Editor

PLOS ONE

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    S1 Table. Estimated probability for one month survival with CPC 1–2.

    (XLSX)

    Click here for additional data file. (13.3KB, xlsx)
    S2 Table. Variables in previous studies.

    (XLSX)

    Click here for additional data file. (10.8KB, xlsx)
    Attachment

    Submitted filename: Response to Reviewers.docx

    Click here for additional data file. (28KB, docx)
    Attachment

    Submitted filename: Response_to_Reviewers.docx

    Click here for additional data file. (19.8KB, docx)

    Data Availability Statement

    The data are owned by a third party. Data are available from the JAAM-OHCA registry committee (contact via http://www.jaamohca-web.com/) for researchers who meet the criteria for access to confidential data.

    Articles from PLOS ONE are provided here courtesy of PLOS

    RESOURCES