Discussion on the prediction model of early judgment of critical risk in emergency patients

doi:10.21203/rs.3.rs-4370377/v1

Download PDF

Research Article

Discussion on the prediction model of early judgment of critical risk in emergency patients

https://doi.org/10.21203/rs.3.rs-4370377/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Objective: To establish a predictive model for early assessment of critical risk in emergency patients and evaluate its clinical clinical benefits.

Method: Clinical data of 3859 patients who visited the emergency department at Hebei General Hospital from November 2021 to December 2021 were selected. The enrolled patients were randomly divided into a training set (2703 cases) and a validation set (1156 cases) in a 7:3 ratio. In the training set, a predictive model was established based on the results of multivariate logistic stepwise regression analysis. At the same time, risk levels were divided, and the predictive efficacy and clinical benefits of the predictive model were verified.

Result: There was no statistically significant difference in clinical data between the training and validation set (P>0.05), indicating comparability between the two groups. Multivariate logistic stepwise regression analysis showed that Gender, age, HR, R, SBP, SPO2, consciousness, pupil, mental state, and pain score were independent risk factors for early assessment of critical risk (P<0.05), and a predictive model was established based on this. Using conditional inference tree, critical risks are classified into low risk (P≤0.075), medium risk (0.075＜P≤0.656), and high risk (P>0.656). Furthermore, the prediction model was internally validated in both the training and validation sets, with a training set AUC of 0.926 [95% CI (0.912-0.939), P<0.001] and a validation set AUC of 0.911 [95% CI (0.886-0.936), P<0.001], indicating good discrimination of the prediction model; The calibration curve of the training set fits the standard curve, while the calibration curve of the validation set model slightly deviates from the standard curve. In the Hosmer Limeshow test, the results of the two groups were P=0.180 and P=0.546, respectively, indicating good calibration of the predicted model; The DCA curve suggests that both groups of people can achieve good clinical benefits.

Conclusion: Establishing a predictive model for early assessment of emergency critical risk is helpful for early identification and intervention of emergency critical patients.

emergency critical risk

predictive model

risk level

early identification

Emergency medicine is a clinical secondary discipline that pays special attention to timeliness. Emergency doctors need to make use of limited medical resources to complete the emergency evaluation and initial treatment of emergency patients in a short time. At present, emergency overcrowding and emergency detention has become a global problem, which also restricts the development of emergency care in our country^[1]. In order to shorten the stay time of emergency department, the "five-level triage method" is mostly used abroad at the present stage, such as American triage severity index (ESI), Canadian injury and urgency scale (CTAS) and Manchester triage system (MTS)^[2–5]. However, most hospitals in China adopt the triage model of "four levels and three zones". Although the problems have been improved to some extent, more research parameters are included in these triage models. Subjective judgment parameters occupy a considerable number, which can easily lead to excessive triage or insufficient triage in emergency situations^[6–7]. Therefore, this study collects 14 vital signs of emergency patients, establishes a visual prediction model for judging the critical risk of emergency patients in the early stage, and evaluates the predictive efficiency and clinical effectiveness of the model, so as to improve the efficiency of triage and evaluation. It provides a convenient and accurate scientific basis for early clinical intervention to reduce the time of emergency stay.

1.1 Research object

The clinical data of 5000 patients in the emergency department of Hebei Provincial people's Hospital from November 2021 to December 2021 were analyzed retrospectively. 3859 patients were selected according to the inclusion criteria and exclusion criteria. Inclusion criteria: (1) age ≥ 18 years old; (2) complete and accurate data. Exclusion criteria: (1) the purpose of medical treatment is to prescribe drugs, examination forms and hospitalization certificates, and patients with no diagnosis and treatment process; (2) patients who die before emergency. This study has been approved by the Ethics Committee of our hospital and accords with the regulations of medical ethics.

1.2 Research methods

1.2.1 Outcome evaluation method

(1)The triage standard of emergency department in our hospital refers to the Chinese Emergency Triage Scale^[7]. (2) the triage nurses made the initial evaluation, and the evaluation results were recorded in the emergency electronic medical record system. After the initial evaluation, the receiving physician carried out the second evaluation, and confirmed and modified the evaluation results in the system. All patients were divided into grade Ⅰ, gradeⅡ, grade Ⅲ and grade Ⅳ, among which gradeⅠand Ⅱwere critically ill patients.

1.2.2 Predict factors

According to the real situation of clinical work and literature research at home and abroad, predictive indicators were included: Gender, age, body temperature, heart rate(HR), respiratory rate(R), systolic blood pressure(SBP), finger pulse oxygen(SPO2), pain score, consciousness, pupil, mental state, tip, urine volume, hyperhidrosis. Among them, Gender, consciousness, pupil, mental state, tip, urine volume and hyperhidrosis were classified according to the index recommended by experts. Body temperature, HR, R, SBP and SPO2 were classified according to the national early warning score (NEWS) standard. Age and pain score were evaluated by continuous value^[8] ( Table 1).

Table 1

Dependent and independent variables and their assignments
Variate	Mutator method
Dependent variables	critical patients (0 = no,1 = yes)
Independent variables
Gender	1 = female, 2 = male
Age(years)	continuous value
Temperature (℃)	0=“36.1–38”,1=“35.1–36或38.1–39”,2=“≥39.1”,3=“≤35”
Heart rate (bpm)	0=“51–90”,1=“41–50或91–110”,2=“111–130”,3=“≤40或≥131”
Respiratory rate(bpm)	0=“12–20”,1=“9–11”,2=“21–24”,3=“≤8或≥25”
Systolic blood pressure (mmHg)	0=“111–219”,1=“101–110”,2=“91–100”,3=“≤90或≥220”
Finger pulse oxygen (%)	0=“≥96”,1=“94–95”,2=“92–93”3=“≤91”
Consciousness	1=“conscious state”,2=“unconscious”
Pupil (size/light reflex)	1=“normal”,2=“abnormal”
Mental state	1=“normal”,2=“abnormal”
Pain score	continuous value
Tip	1=“worm”,2=“cool”
Urine volume	1=“normal”,2=“ abnormal”
Hyperhidrosis	1=“no”,2=“yes”

1.2.3 Data quality control

The data of this study come from the emergency electronic medical records of the information system of our hospital. There are two entrants, one examiner and one quality controller, in which the examiner is the attending physician and the quality controller is the senior physician. According to the forecast index collection table, the data are entered separately by two people, and the consistency, repeatability and logicality of the two input data are verified by the examiner. If any doubt arises during the verification process, submit it to the quality controller at any time to control the authenticity and accuracy of the data.

1.2.4 Statistical method

3859 patients after screening were randomly divided into training set (n = 2703) and verification set (n = 1156) according to the proportion of 7:3. SPSS 26.0 and R 4.3.1 (https://www.r-proj.ect.org/) software were used for statistical analysis. The measurement data are expressed by mean ± standard deviation, and the counting data are expressed by frequency (percentage). T-test was used to compare the measurement data, and χ 2 test was used to compare the counting data. The independent risk factors for early judgment of critical risk of emergency patients were screened out by multi-factor logistic regression analysis of the training set, and the prediction model was established. Through the method of conditional inference tree, the prediction probability of the model in the training set is divided into three risk levels: low, medium and high, so as to establish a risk stratification system to judge the critical degree of emergency patients in the early stage. After that, the discrimination, calibration and clinical validity of the prediction model are verified in the training set and verification set: the discrimination degree is evaluated by the area under the subject working characteristic curve (AUC). The AUC of 0.50–0.70 is low differentiation, 0.71–0.90 is medium differentiation, and > 0.90 is high differentiation. The calibration degree was evaluated by calibration graph method and HosmerLemeshow test. The more suitable the predictive calibration curve and the standard curve were, the better the calibration degree of the predictive model was. The clinical validity was evaluated by drawing decision curve analysis (DCA). P < 0.05 indicates that the difference is statistically significant. (Fig. 1)

2.1 Baseline

A total of 3859 emergency patients were included in this study, which were divided into training set (n = 2703) and verification set (n = 1156). There were 612 critically ill patients (15.9%), 1970 males (51.0%) and 1889 females (49.0%). The average age was (50.61 ± 18.99) years. There was no significant difference in Gender, age, body temperature, HR, R, SBP, SPO2, pain score, consciousness, pupil, mental state, tip, urine volume and hyperhidrosis between the two groups (P > 0.05). There was comparability between the two groups (Table 2).

2.2 Construction of forecasting model by multi-factor logistic stepwise regression analysis

Gender, age, temperature, HR, R, SBP, SpO2, pain score, consciousness, pupil, mental state, tip, urine volume and hyperhidrosis were taken as dependent variables, and multivariate logistic stepwise regression analysis was carried out (Table 1). The results showed that the independent risk factors for early judgment of critical risk in emergency patients were Gender, age, HR, R, SBP, SpO2, consciousness, pupil, mental state and pain score (P < 0.05) (Table 3).

Table 2

Clinical characteristics of training set and validation set
Parameters	Totality (N = 3859)	Training set (N = 2703)	Validation set (N = 1156)	t/χ²	P-value
Age,(years,‾x ± s)	50.61 ± 18.99	50.42 ± 19.03	51.08 ± 18.89	0.991	0.496
Pain score,(‾x ± s)	1.82 ± 2.36	1.84 ± 2.37	1.78 ± 2.32	0.829	0.403
Illness degree,n(%)				0.004	0.949
mildly	3247(84.1)	2275(84.2)	972(84.1)
critical	612(15.9)	428(15.8)	184(15.9)
Gender,n(%)				1.132	0.287
Female	1889(49.0)	1308(48.4)	581(50.3)
Male	1970(51.0)	1395(51.6)	575(49.7)
Temperature,(℃,n(%))				1.199	0.753
36.1–38.0	2888(74.8)	2030(75.1)	858(74.2)
35.1–36.0或38.1–39.0	856(22.2)	591(21.9)	265(22.9)
≥ 39.1	56(1.5)	38(1.4)	18(1.6)
≤ 35.0	59(1.5)	44(1.6)	15(1.3)
Heart rate,(bpm,n(%))				1.652	0.648
51–90	2651(68.7)	1840(68.1)	811(70.2)
41–50或91–110	861(22.3)	616(22.8)	245(21.2)
111–130	270(7.0)	192(7.1)	78(6.7)
≤ 40或≥131	77(2.0)	55(2.0)	22(1.9)
Respiratory rate, (bpm,n(%))				5.341	0.148
12–20	2509(65.0)	1746(64.6)	763(66.0)
9–11	186(4.8)	141(5.2)	45(3.9)
21–24	940(24.4)	668(24.7)	272(23.5)
≤ 8或≥25	224(5.8)	148(5.5)	76(6.6)
Systolic blood pressure, (mmHg,n(%))				0.457	0.928
111–219	3324(86.1)	2334(86.3)	990(85.6)
101–110	285(7.4)	197(7.3)	88(7.6)
91–100	163(4.2)	111(4.1)	52(4.5)
≤ 90 或≥220	87(2.3)	61(2.3)	26(2.2)
Finger pulse oxygen, (%,n(%))				5.089	0.165
≥ 96	3403(88.2)	2389(88.4)	1014(87.7)
94–95	239(6.2)	173(6.4)	66(5.7)
92–93	60(1.6)	43(1.6)	17(1.5)
≤ 91	157(4.1)	98(3.6)	59(5.1)
Consciousness,n(%)				0.047	0.829
conscious state	3617(93.7)	2532(93.7)	1085(93.9)
unconscious	242(6.3)	171(6.3)	71(6.1)
Pupil,n(%)				0.467	0.494
normal	3775(97.8)	2647(97.9)	1128(97.6)
abnormal	84(2.2)	56(2.1)	28(2.4)
Mental state, n(%)				0.727	0.394
normal	3146(81.5)	2213(81.9)	933(80.7)
abnormal	713(18.5)	490(18.1)	223(19.3)
Tip, n(%)				1.004	0.316
warm	3827(99.2)	2678(99.1)	1149(99.4)
cool	32(0.8)	25(0.9)	7(0.6)
Urine volume, n(%)				0.387	0.534
normal	3763(97.5)	2633(97.4)	1130(97.8)
abnormal	96(2.5)	70(2.6)	26(2.2)
Hyperhidrosis,n(%)				1.634	0.201
yes	3797(98.4)	2655(98.2)	1142(98.8)
no	62(1.6)	48(1.8)	14(1.2)

Table 3

logistic regression analysis results of training set
	B	OR(95%CI)	P-value
Intercept	-7.035		<0.001
Age (years)	0.043	1.044(1.035–1.053)	<0.001
Pain score	0.073	1.076(1.013–1.141)	0.016
Gender (ref = Male)		1.00
Female	0.881	2.413(1.774–3.303)	<0.001
Heart rate (bpm) (ref = 51–90)		1.00
41–50 and 91–110	0.748	2.114 (1.511–2.952)	<0.001
111–130	1.213	3.365(1.998–5.584)	<0.001
≤ 40 and ≥ 131	2.237	9.369(4.015–21.414)	<0.001
Respiratory rate (bpm) (ref = 12–20)		1.00
9–11	0.558	1.747 (0.997–3.006)	0.047
21–24	1.220	3.389(2.357–4.887)	<0.001
≤ 8 and ≥ 25	2.339	10.368(5.958–17.921)	<0.001
Systolic blood pressure(mmHg) (ref = 111–219)		1.00
101–110	0.755	2.128(1.205–3.679)	0.008
91–100	1.533	4.631(2.483–8.478)	<0.001
≤ 90 and ≥ 220	2.051	7.778(3.585–16.612)	<0.001
Finger pulse oxygen(%) (ref = ≥ 96)		1.00
94–95	0.667	1.949(1.161–3.225)	0.01
92–93	1.054	2.870(1.163–7.239)	0.024
≤ 91	1.776	5.905(2.877–12.543)	<0.001
Consciousness (ref = conscious state)		1.00
unconscious	1.941	6.966(4.128–11.885)	<0.001
Pupil(ref = normal)		1.00
abnormal	1.817	6.152(1.809–22.890)	0.005
Mental state(ref = normal)		1.00
abnormal	2.342	10.403(7.530-14.462)	<0.001

2.3 Establishment of risk stratification and nomograph

On the basis of the above prediction model, the corresponding prediction probability is calculated for each research object in the training set (Fig. 2). It can be seen that the prediction probability obviously shows the characteristics of positive skewness distribution. that is, with the gradual increase of the critical risk of emergency patients, the corresponding number is getting smaller and smaller.

According to the prediction probability of the training set, the critical risk of emergency patients is stratified by conditional inference tree (Fig. 3). Through two-level segmentation, with the cutoff values of 0.075 and 0.656, the prediction probability is divided into four levels. In order to facilitate the practical application, the two grades from 0.075 to 0.656 are merged and finally divided into three risk levels: low risk (P ≤ 0.075), medium risk (0.075 < P ≤ 0.656) and high risk (P > 0.656). The actual critical risk rate and average prediction probability of emergency patients corresponding to the training set and verification set were consistent in the three levels (Fig. 4).

According to the risk stratification results made by the conditional inference tree, the above prediction model is visualized as a line chart with risk stratification (Fig. 5). The 10 risk factors and their corresponding scores are shown in the line chart. The total score can be obtained by adding the scores corresponding to the specific values of the 10 factors, and then the vertical line is drawn with the total score. The corresponding risk grade and prediction probability are the model prediction results of the critical risk of emergency patients.

2.4 Validation

2.4.1 Discrimination

The ROC curves of the training set and the verification set were drawn. The AUC of the training set was 0.926[95%CI(0.912–0.939), P < 0.001], the critical value was 0.127, the sensitivity was 84.8%, and the specificity was 84.0%. The AUC of the validation set is 0.911[95%CI(0.886–0.936), P < 0.001], the critical value is 0.105, the sensitivity is 81.3%, and the specificity is 87.0%. The AUC of the prediction model is greater than 0.90 in both groups, and the prediction accuracy is good (Fig. 6).

2.4.2 Calibration

In the training set, the calibration curve of the model fits the standard curve, and in the verification set, the calibration curve of the model is slightly off the standard curve, which indicates that the calibration degree of the prediction model is good (Fig. 7). The results of Hosmer-Lemeshow test in training set and verification set are χ2 = 11.389, P = 0.180 and χ2 = 6.909, P = 0.546, P > 0.05, the difference is not statistically significant, indicating that the prediction model has good calibration.

2.4.3 Clinical effectiveness

The clinical validity of the prediction model was evaluated by DCA curve (Fig. 8). The abscissa is the threshold probability, the ordinate is the net income, and the horizontal straight line assumes that all patients are not critical patients and do not accept intervention, and the net income is 0; Oblique straight line assumes that all patients are critically ill and receive intervention, and the net income is a backslash with a negative slope. In the training set and verification set, the net benefit of patients is higher than the other two extreme curves, and it is suggested that the prediction model has clinical validity in a long range.

2.5 Visualization application of prediction model

According to the prediction model for early judgment of critical risk of emergency patients, and in order to facilitate medical staff to use nomogram, a simple web operation interface (https://hbghemergency85988120.shinyapps.io/emergency120/)is established to calculate the exact probability of whether emergency patients are critical patients. Take the emergency patient as an example: the patient's gender is 68 years old, HR 95bpm, R 22bpm, SBP 105mmHg, SpO2 95%, consciousness is clear, pupils are normal, mental state is poor, and pain score is 4. According to the prediction model with risk grade, it is predicted that her risk of becoming a critical patient is 0.869 > 0.656, which is a high-risk stratification.

Emergency medicine is different from other disciplines in the urgency of time, and the emergency department especially emphasizes the urgency of time^[9], which requires early identification and early intervention to determine the hidden dangers that threaten life and deal with them in time. In recent years, the phenomenon of emergency crowding is particularly prominent^{[1, 9–10]}. How to identify critically ill patients at an early stage under high load and improve the accuracy of triage and evaluation is particularly important. At present, the emergency prediction of critically ill patients mostly adopts vital sign assessment method, NEWS, emergency pre-examination and triage expert consensus, adult five-level emergency triage system and so on^[11–12]. In each scoring table, those with fewer indicators are convenient to use, but their accuracy may be poor, while those with more indicators improve their accuracy, but the indicators are not easy to obtain, which prolongs the triage time. Moreover, if there are more subjective indicators, it will also affect the accuracy of triage by medical staff and information systems, leading to over-triage or insufficient triage. Therefore, this study includes the predictive indicators that can be obtained in the first time in clinic, and establishes a visual prediction model for early judgment of critical risk of emergency patients, which shortens the time of emergency triage and evaluation and improves the accuracy of triage.

3.1 Risk factors of prediction model

At present, in the research on the related factors of early identification of critical emergency patients, the factors involved in identification include: temperature, HR, R, SBP, SpO2, pain score, consciousness, pupil, mental state, blood sugar, blood routine, creatinine, blood potassium, etc., but there is no consensus yet^{[3, 8, 13–14]}. In this study, through multivariate logistic regression analysis, it is found that the risk factors for early judgment of critical risk of emergency patients are gender, age,HR, R, SBP, SpO2, consciousness, pupil, mental state and pain score, while temperature, tip, urine volume and hyperhidrosis are not independent risk factors.

The general health status of the elderly in China declines with the increase of age, and age is also an independent risk factor for various common diseases (such as diabetes, coronary heart disease, hypertension, stroke, etc.)^[15–16]. Middle-aged and elderly patients account for a large proportion of emergency patients. Chung HS and other studies show that the number of emergency elderly critically ill patients and rescue operations is more than that of other adult groups^[17–19]. The average age of emergency patients in this study is (50.61 ± 18.99) years, and age is also one of the risk factors in the prediction model (OR 1.044,95%CI:1.035–1.053), and the degree of critical risk is higher with the increase of age, which is consistent with the above study. Gender in this study showed that the proportion of male and female patients was similar, and the proportion of male patients was slightly higher. The results showed that male patients (OR 2.413,95%CI:1.774–3.303). The risk of critical illness is higher than that of women. Engebretsen S other studies also show that the mortality rate of men is higher than that of women, and Engebretsen S also suggest that the rate of men staying in ICU is higher than that of women, which may be related to the fact that men are still the main group of social productivity and the high-risk group of sudden serious diseases (such as cardiovascular and cerebrovascular diseases)^[20–21].

HR, R, SBP, SpO2 and consciousness, as commonly used vital signs, are widely used as one of the important indicators for clinical observation of disease changes, prediction of critical illness and mortality^{[2–4, 8]}. In the course of this study, we tried to substitute HR, R,SBP and SPO2 into continuous values in the model construction, but in clinical practice, when the first three factors are in low value and high value, they will prompt the severity of the disease, which belongs to nonlinear indicators. There is little difference in the risk degree of SPO2 in the range of 96%-100% in clinic, so these four factors are classified and assigned according to NEWS score, and they are recognized as two-category factors. Substituting into the model construction, the results show that HR, R, SBP, SPO2 and consciousness are the risk factors for early judgment of emergency critical risk, which is consistent with the research results of Kim S and Simbawa JH^{[13, 21]}. In addition to evaluating the nervous system by observing consciousness, this study also shows that pupil size and abnormal light reflection are one of the risk factors of the prediction model (OR 6.152,95%CI:1.809–22.890). Pupil size and reflex are important monitoring parameters for patients with consciousness disorder. Minami Y and others think that pupil measurement in emergency is simpler and faster than EEG and imaging in early prediction of patients' consciousness level, but the accuracy of direct examination with flashlight is lower than that of automatic pupillometer^[22–23]. so the availability and accuracy of tools for measuring pupils in actual emergency work need to be further evaluated^[24].

Mental state is also a risk factor in the prediction model of this study (OR 10.403,95%CI:7.530-14.462), and its OR value is the highest among the 10 risk factors. In critical diseases such as pneumonia, acute myocardial infarction, shock, etc., it is often manifested as listlessness, but at present, domestic and foreign studies have not mentioned the role of mental state in assessing the risk of emergency critical patients, so mental state should be highly concerned in clinical work.

Pain is one of the common reasons for emergency treatment, and this study also shows that pain is one of the risk factors for predicting the critical risk of emergency treatment (OR 1.076,95%CI:1.013–1.141). Pain can cause changes in heart rate and blood pressure, and it is also a prominent manifestation of the aggravation of some diseases. Giusti GD and Hämäläinen J also think that pain is the fifth vital sign, but they also point out that there are great difficulties and diversities in the emergency assessment of acute pain^[25–26]. The difference in pain scores between doctors and patients, the crowded emergency environment and the use of analgesic drugs all affect the pain assessment, which is also a place that needs close attention in the future.

In addition, temperature, tip, urine volume and hyperhidrosis are not included in the risk factors in this study. Kim S and Kushimoto S think that hypothermia is positively related to critically ill patients with sepsis, but hyperpyrexia and normal temperature have little effect on evaluation^{[21, 27]}. This study does not show any correlation between temperature segmentation and critical risk judgment, which may be because the accuracy of axillary temperature measurement is affected by environmental temperature, emotional agitation and patient cooperation, which reduces the severity of emergency. Moreover, hyperhidrosis, tip and temperature are also greatly influenced by environmental factors and subjective factors. At present, there is no literature report that hyperhidrosis and tip are included in the study at home and abroad, so it is necessary to refine the evaluation criteria of hyperhidrosis and tip in the further study. Urine volume is an index to evaluate circulation and renal function, but it is not an index to judge critical risk in chronic renal failure, which may be the reason why urine volume is not included as a risk factor in this study.

3.2 Visual prediction model

At present, there is a lack of relevant research on the prediction model of critical risk for emergency patients at an early stage in clinic. The nomogram can simply predict the individual risk and has high clinical practical value^[28]. In this study, while making nomogram, conditional reasoning tree is also used to stratify the critical risk of emergency patients. In this study, it can be seen that when the model predicts that the patient's critical risk is higher than 0.656, if intervention measures are taken immediately, the patient will benefit from the clinic. If it is lower than or equal to 0.075, the patient can temporarily not take intervention measures and continue to observe dynamically. If it is between the two, it is necessary to closely observe the condition and adjust the intervention plan at any time, which will reasonably allocate resources to some extent and solve the emergency congestion problem.

3.3 Limitations

This study is a retrospective analysis of a single center. If the patient data of multiple centers are further included, the sample size is increased and external verification is carried out, the model can be further strengthened.

Congestion in emergency department is a common problem in emergency department at present. It is very difficult to triage and evaluate quickly and accurately in a crowded environment, shorten the treatment time and reduce the missed diagnosis and misdiagnosis of critically ill patients. At present, some triage parameters and systems are complex and difficult to apply, while others are too simple to affect the accuracy. In this study, the independent risk factors for early judging the critical risk of emergency patients are gender, age, HR, R, SBP, SpO2, consciousness and pupil. This prediction model has good discrimination, calibration and clinical effectiveness in both the training set and the validation set of emergency patients, which improves the ability of early identification and early intervention of critical emergency patients, and the risk factors in this study are very easy to be obtained early in clinic, especially suitable for primary medical and health institutions.

Ethics approval and consent to participate

This study was in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the declaration of Helsinki. This study was conducted with approval from the Ethics Committee of Hebei General Hospital. A written informed consent was obtained from all participants.

Availability of data and materials

Additional data and materials may be requested from the corresponding author on reasonable request.

Competing interests

The authors declare that they have no competing interests.

Funding

Funding information is not available.

Authors' contributions

Xurui Li and Jianguo Li designed the research. Jian Lv and Hongling Li collected and analyzed the data. All authors wrote the manuscript. Jianguo Li, Xurui Li, Hui Guo and Qian Zhao discussed the results and revised the final manuscript. All authors read and approved the final manuscript.

Acknowledgements

We acknowledged the Medical Records Department for collecting the data of the patients.

Savioli G, Ceresa IF, Gri N, et al. Emergency Department Overcrowding: Understanding the Factors to Find Corresponding Solutions[J]. J Pers Med, 2022, 12(2). DOI: 10.3390/jpm12020279.
Davis S, Ju C, Marchandise P, et al. Impact of Pain Assessment on Canadian Triage and Acuity Scale Prediction of Patient Outcomes[J]. Ann Emerg Med, 2022, 79(5):433-440. DOI: 10.1016/j.annemergmed.2022.01.014.
Kuriyama A, Urushidani S, Nakayama T. Five-level emergency triage systems: variation in assessment of validity[J]. Emerg Med J, 2017, 34(11):703-710. DOI: 10.1136/emermed-2016-206295.
Zaboli A, Ausserhofer D, Pfeifer N, et al. Acute abdominal pain in triage: A retrospective observational study of the Manchester triage system's validity[J]. J Clin Nurs, 2021, 30(7-8):942-951. DOI: 10.1111/jocn.15635.
Zhao J, He L, Zhao Y, et al. A Review of the Roles and Implementation of Pediatric Emergency Triage Systems in China and Other Countries[J]. Med Sci Monit, 2023, 29:e941582. DOI: 10.12659/MSM.941582.
Mistry B, Stewart De Ramirez S, Kelen G, et al. Accuracy and Reliability of Emergency Department Triage Using the Emergency Severity Index: An International Multicenter Assessment[J]. Ann Emerg Med, 2018, 71(5):581-587.e3. DOI: 10.1016/j.annemergmed.2017.09.036.
Zhiting G, Jingfen J, Shuihong C, et al. Reliability and validity of the four-level Chinese emergency triage scale in mainland China: A multicenter assessment[J]. Int J Nurs Stud, 2020, 101:103447. DOI: 10.1016/j.ijnurstu.2019.103447.
Physicians RCO. National Early Warning Score (NEWS)[J]. 2012 .
Sartini M, Carbone A, Demartini A, et al. Overcrowding in Emergency Department: Causes, Consequences, and Solutions-A Narrative Review[J]. Healthcare (Basel), 2022, 10(9). DOI: 10.3390/healthcare10091625.
Badr S, Nyce A, Awan T, et al. Measures of Emergency Department Crowding, a Systematic Review. How to Make Sense of a Long List[J]. Open Access Emerg Med, 2022, 14:5-14. DOI: 10.2147/OAEM.S338079.
Chien CY, Chaou CH, Yeh CC, et al. Using mobility status as a frailty indicator to improve the accuracy of a computerised five-level triage system among older patients in the emergency department[J]. BMC Emerg Med, 2022, 22(1):86. DOI: 10.1186/s12873-022-00646-0.
Hodgson NR, Poterack KA, Mi L, et al. Association of Vital Signs and Process Outcomes in Emergency Department Patients[J]. West J Emerg Med, 2019, 20(3):433-437. DOI: 10.5811/westjem.2019.1.41498.
Simbawa JH, Jawhari AA, Almutairi F, et al. The Association Between Abnormal Vital Signs and Mortality in the Emergency Department[J]. Cureus, 2021, 13(12):e20454. DOI: 10.7759/cureus.20454.
Zhu A, Liu X, Zhang J. Identifying a Clinical Risk Triage Score for Adult Emergency Department[J]. Clin Nurs Res, 2021, 30(8):1135-1143. DOI: 10.1177/10547738211003273.
Tu Q, Hyun K, Hafiz N, et al. Age-Related Variation in the Provision of Primary Care Services and Medication Prescriptions for Patients with Cardiovascular Disease[J]. Int J Environ Res Public Health, 2022, 19(17). DOI: 10.3390/ijerph191710761.
Wahabi H, Esmaeil S, Zeidan R, et al. Effects of Age, Metabolic and Socioeconomic Factors on Cardiovascular Risk among Saudi Women: A Subgroup Analysis from the Heart Health Promotion Study[J]. Medicina (Kaunas), 2023, 59(3). DOI: 10.3390/medicina59030623.
Chung HS, Namgung M, Lee DH, et al. Validity of the Korean triage and acuity scale in older patients compared to the adult group[J]. Exp Gerontol, 2023, 175:112136. DOI: 10.1016/j.exger.2023.112136.
Lin CF, Lin PC, Hu SY, et al. Comprehensive Geriatric Assessment and Clinical Outcomes in the Older People at the Emergency Department[J]. Int J Environ Res Public Health, 2021, 18(11). DOI: 10.3390/ijerph18116164.
Narongsanoi P, Banharak S, Panpanit L, et al. Factors related to changes in severity among adult and older adult patients at an internal medicine department clinic: an embedded mixed-method study[J]. Sci Rep, 2024, 14(1):3914. DOI: 10.1038/s41598-024-54266-8.
Engebretsen S, Bogstrand ST, Jacobsen D, et al. Characteristics, management and outcome of critically ill general medical patients in the Emergency Department: An observational study[J]. Int Emerg Nurs, 2021, 54:100939. DOI: 10.1016/j.ienj.2020.100939.
Kim S, Kang H, Cho Y, et al. Emergency department utilization and risk factors for mortality in older patients: an analysis of Korean National Emergency Department Information System data[J]. Clin Exp Emerg Med, 2021, 8(2):128-136. DOI: 10.15441/ceem.20.098.
Minami Y, Mishima S, Oda J. Prediction of the level of consciousness using pupillometer measurements in patients with impaired consciousness brought to the emergency and critical care center[J]. Acute Med Surg, 2020, 7(1):e537. DOI: 10.1002/ams2.537.
Monai E, Favaretto C, Salvalaggio A, et al. Pupillary dynamics predict long-term outcome in a cohort of acute traumatic brain injury coma patients[J]. Ann Clin Transl Neurol, 2023, 10(10):1854-1862. DOI: 10.1002/acn3.51879.
Olsen MH, Jensen HR, Ebdrup SR, et al. Automated pupillometry and the FOUR score - what is the diagnostic benefit in neurointensive care?[J]. Acta Neurochir (Wien), 2020, 162(7):1639-1645. DOI: 10.1007/s00701-020-04381-y.
Giusti GD, Reitano B, Gili A. Pain assessment in the Emergency Department. Correlation between pain rated by the patient and by the nurse. An observational study[J]. Acta Biomed, 2018, 89(4-S):64-70. DOI: 10.23750/abm.v89i4-S.7055.
Hämäläinen J, Kvist T, Kankkunen P. Acute Pain Assessment Inadequacy in the Emergency Department: Patients' Perspective[J]. J Patient Exp, 2022, 9:23743735211049677. DOI: 10.1177/23743735211049677.
Kushimoto S, Abe T, Ogura H, et al. Impact of Body Temperature Abnormalities on the Implementation of Sepsis Bundles and Outcomes in Patients With Severe Sepsis: A Retrospective Sub-Analysis of the Focused Outcome Research on Emergency Care for Acute Respiratory Distress Syndrome, Sepsis and Trauma Study[J]. Crit Care Med, 2019, 47(5):691-699. DOI: 10.1097/CCM.0000000000003688.
Kong X, Li J, Cai Y, et al. A modified TNM staging system for non-metastatic colorectal cancer based on nomogram analysis of SEER database[J]. BMC Cancer, 2018, 18(1):50. DOI: 10.1186/s12885-017-3796-1.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Discussion on the prediction model of early judgment of critical risk in emergency patients

Status:

Version 1

Abstract

Figures

Instruction

Materials and methods

1.1 Research object

1.2 Research methods

1.2.1 Outcome evaluation method

1.2.2 Predict factors

1.2.3 Data quality control

1.2.4 Statistical method

Results

2.1 Baseline

2.2 Construction of forecasting model by multi-factor logistic stepwise regression analysis

2.3 Establishment of risk stratification and nomograph

2.4 Validation

2.4.1 Discrimination

2.4.2 Calibration

2.4.3 Clinical effectiveness

2.5 Visualization application of prediction model

Discussion

3.1 Risk factors of prediction model

3.2 Visual prediction model

3.3 Limitations

Conclusion

Declarations

References

Additional Declarations

Status:

Version 1