Nomogram for the prediction of in-hospital incidence of acute respiratory distress syndrome in patients with acute pancreatitis

Objective: Acute respiratory distress syndrome （ ARDS ） associated with high mortality is the common complication in acute pancreatitis(AP) .The aim of this study was to formulate and validate an individualized predictive nomogram for in-hospital incidence of ARDS in AP patients.

From January 2017 to December 2018, 779 individuals with AP were included in this study . They were randomly distributed into primary cohort (n=560) and validation cohort (n=219). Based on the primary cohort, risk factors were identified by logistic regression model and a nomogram was performed. The nomogram was validated in the primary and validation cohort by the bootstrap validation method. The calibration curve was applied to evaluate the consistency between nomogram and ideal observation.

There were 728 patients in the non-ARDS group and 51 in the ARDS group, with an incidence rate of about 6.55%. Five independent factors including white blood counts(WBC), prothrombin time(PT), albumin(ALB), serum creatinine(SCR) and triglyceride (TG) were associated with in-hospital incidence of ARDS in AP patients. A nomogram was constructed based on the five independent factors with primary cohort of AUC 0.821 and validation cohort of AUC 0.822. Calibration curve analysis indicated that the predicted probability was in accordance with the observed probability in both primary and validation cohorts.

The study developed an intuitive nomogram with easily available laboratory parameters for the prediction of in-hospital incidence of ARDS in patients with AP . The incidence of ARDS for an individual patient can be fast and conveniently evaluated by our nomogram. Key words: Acute pancreatitis, acute respiratory distress syndrome, nomogram 1. Introduction All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint Acute pancreatitis (AP) is one of common illnesses of the gastrointestinal system, of which some can develop into severe acute pancreatitis (SAP) with poor prognosis[1]. SAP is known to be characterized by persistent organ disfunction or pancreatic necrosis, of which the mortality varies from 10% to 50%[2]. Acute respiratory distress syndrome (ARDS) , as a syndrome of bilateral pulmonary infiltrates and inflammation associated with increased permeability of lung epithelium and vascular endothelium[3], occurs in around one-third of SAP patients[4]and accounts for over 50% of deaths in SAP[5]. Research evidence shows that ARDS, to a certain extent, could be preventable, and clinical prognosis may be improved due to appropriate interventions in early phase of ARDS[6]. Therefore, it is significant for the early prediction of in-hospital ARDS incidence in patients with AP . Nomogram, as a statistical model created on the basis of different clinical and biological variables, is widely utilized for the prediction of complication , prognosis and survival in different disorders[7, 8]. And It is also beneficial for physicians to make individualized treatment plans. However, there hasn’t been any reports on nomogram for the early prediction of ARDS in patients with AP. In this study, we analyzed the association of clinical characteristics and laboratory variables with the incidence of ARDS in AP. Furthermore, we formulated an easily applicable nomogram which included several routine laboratory parameters for predicting in-hospital incidence of ARDS in AP patients and validated its predictive capability and accuracy in an independent cohort. 2. Methods Patients This study was a retrospect cohort study. All patients with AP admitted to the Changsha Central Hospital of University of South China and the Second Xiangya Hospital of Central South University between January 2017 and December 2018 were All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint included. Inclusion criteria were set as follows: age ≥ 18 and confirmed diagnosis of AP. Excluded criteria were set as follows: more than 72 hours after onset of symptoms, recurrent pancreatitis, COPD, malignant tumors, chronic kidney disease, acute or chronic heart failure, pregnancy and HIV/ AIDS or presence of other immune-deficiency disorders. Ethics approval and consent to participate The study was approved by the Ethics Committees of Changsha Central Hospital of University of South China (Changsha, China) and the Second Xiangya Hospital of Central South University (Changsha, China). All research was performed in accordance with the relevant guidelines and regulations. Due to retrospective characteristics of the study , informed consent was waived， which was also approved by the Ethics Committee of Changsha Central Hospital of University of South China (Changsha, China) and the Ethics Committee of Second Xiangya Hospital of Central South University (Changsha, China). Definitions A patient was diagnosed with AP when at least two of the following criteria were encountered： 1) symptom of abdominal pain, 2) the level of serum lipase or amylase increased at least three times more than the normal threshold , and 3) image findings of AP on abdominal ultrasonic and/or CT scan[2]. Hypertriglyceridemia associated with AP was defined as follows: the level of triglyceride ≥ 11.3 mmol/L or ≥ 5.65mmol/L accompanied with milky serum[2]. Systemic inflammatory response syndrome(SIRS)was defined when two of following criteria were fulfilled: body temperature38.0 °C, white blood cell count 12*10 9/l or >10% immature forms, heart rate >90/min, and respiratory rate> 20/min or PaCO2< 32 mmHg[9]. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint According to Berlin definition of ARDS[10] , the diagnosis of ARDS was made by an acute hypoxemia, a decrease in the PaO2/FiO2 index below 300 mmHg and bilateral lung infiltration on X-ray/CT scan that was not illuminated totally by fluid overload or cardiac failure. Based on PaO2/FiO2 and PEEP index, ARDS includes 3 categories: mild (200mmHg<PaO2/FiO2 ≤ 300mmHg, PEEP≥ 5cmH2O), moderate (100mmHg < PaO2/FiO2 ≤ 200 mmHg, PEEP ≥ 5cmH2O), and severe (PaO2/FiO2 ≤ 100 mmHg, PEEP≥ 10cmH2O). Arterial blood gas analysis was dynamically performed for patients as well as when patients had the symptom of dyspnea during hospitalization. Data collection Data collected included clinical characteristics and laboratory findings while patients were admitted in ≤ 24h.Collected clinical data were included : age, gender, smoking, hypertension, diabetes mellitus, coronary heart disease, etiology(alcohol, biliary, hypertriglyceridemia),systolic blood pressure(SBP),diastolic blood pressure(DBP), heart rate (HR), respiratory rate(RR), body temperature, oxyhemoglobin saturation (SaO2), white blood cell count(WBC),neutrophil （ N） ,lymphocyte（ L） , hematocrit (HCT), platelet(PLT), fasting blood sugar (FBS), prothrombin time(PT), activated partial thromboplastin time(APTT), albumin (ALB), alanine aminotransferase (ALT), blood urea nitrogen(BUN),serum creatinine (SCR), amylase(AMS), lipase(LPS), lactate dehydrogenase(LDH), calcium, high density lipoprotein, (HDL), low density lipoprotein(LDL), triglyceride(TG), PaO2 and PaCO2. NLR is defined as the ratio of neutrophils to lymphocytes. Incidence of SIRS within 24 hours after admission was calculated. The scores of sequential organ failure assessment (SOFA), bedside index of severity in acute pancreatitis (BISAP),and Ranson were calculated to assess the severity of AP patients in terms of personal characteristics, vital signs and laboratory findings ( Table 1) . Primary clinical outcome was in-hospital incidence of ARDS. Length of stay （ LOS） in ICU and hospital were also compared. (Table 1). All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint Statistical analysis Statistical results were expressed in mean ± standard deviation for normal data, while for non-normal data, interquartile range (IQR)and median were employed. Categorical data were expressed as percentage and number. Student t-test or Rank sum test was applied for comparisons between ARDS and non-ARDS groups. Chi-square or Fishers exact test was applied for categorical data. The original database was randomly distributed into the primary and validation cohort by 7:3 ratio. The primary cohort was applied to develop the model by multivariable logistic regression and the validation cohort was applied to validate the model. The area under receiver operating characteristic curve (AUC) was utilized for evaluating the accuracy of the model. A nomogram was performed on the basis of logistic regression model and we proportionally converted each regression coefficient in the model to a scale of 0 to 100point. The total points were calculated by adding the point of each independent variable, which predicted the individualized incidence of ARDS. The nomogram was validated in the primary and validation cohort by the bootstrap validation method. The calibration curve was applied to evaluate the consistency between nomogram and observation probability. Statistical analysis was performed using SPSS software (version 26) and R software (http://www.R-project.org). Statistical significance was considered to be at two- sided p < 0.05. 3 Results General Characteristics of all patients All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint A total of 817 patients with diagnosis of AP were initially identified and 38 were excluded according to the excluded criteria. Finally, a total of 779 patients were included to analyze in the study.(Figure 1) The general characteristics of the patients were showed in Table 1. There were 728 patients in the non-ARDS group and 51 in the ARDS group, with an incidence rate of about 6.55%. In ARDS group , the number of mild , moderate and severe ARDS were 32, 16 and 3, respectively. Proportion of males in ARDS group was 82.35% while 69.92% in non-ARDS group. Age at the onset, smoking, past medical history( diabetes mellitus, hypertension, coronary heart disease) , etiology(alcohol, biliary), SBP , DBP, RR and SaO2 were not significantly different between patients with and without ARDS. Hypertriglyceridemia(P=0.011), temperature(P=0.025) and heart rate(P<0.001) were significantly different between two groups (both p < 0.05). For ARDS group, ALB was significant lower than that in non-ARDS group(P=0.049). Other laboratory tests such as WBC(P <0.001),N(P<0.001),L(P=0.025), NLR(P=0.001) ， HCT(P=0.006),FBS(P<0.001),PT(P<0.001),APTT(P<0.001),BUN(P=0.017),SCR (P=0.012), LDH(P <0.001),calcium(P=0.024) and TG(P=0.003) were significantly higher in the ARDS group than that in the non-ARDS group. PLT, ALT, AMS, LPS, HDL, LDL, PaO2 and PaCO2 were not significantly different between two groups. The incidence of SIRS within 24 hours after admission was significant higher in ARDS group(P=0.026).(Table 1) The incidence of ventilation and in-hospital mortality in ARDS group were 37.25%(19/51) and 5.882%(3/51), respectively. LOS in ICU and hospital (all P <0.001) were also significantly higher in the ARDS group. Scores of SOFA, Ranson and BISAP (all P<0.001) were significantly higher in the ARDS group than that in the non-ARDS group. (Table 1) Baseline characteristics of primary and validation cohort All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint The baseline characteristics of patients in the primary cohort(n=560) and validation cohort(n=219) are shown in Table 2. Proportion of males in primary and validation cohorts were respectively 68.57% and 76.25%(P=0.378). The main etiologies of AP in primary cohort were hypertriglyceridemia(40.17%), alcohol(19.64%) and biliary (18.21%),while in validation cohort the etiologies were hypertriglyceridemia(39.72%), biliary(23.74%) and alcohol(17.8%). The incidence of ARDS were respectively 6.446% and 6.829% in two groups(P=0.416). Except serum amylase and lipase, there were no significant difference between the primary and validation cohort in other laboratory tests. Comparison of variables between ARDS and non-ARDS groups in primary cohort Variables including HR(P<0.001), temperature(P=0.027), WBC(P<0.001), N(P<0.001), L(P=0.018), NLR(P=0.004), FBS(P<0.001), PT(P<0.001), APTT(P=0.01), ALB (P<0.001), SCR(P<0.001),LDH(P<0.001), Calcium(P=0.003), HDL(P=0.034), TG (P<0.001) and SIRS(P=0.045) were significantly different between two groups in primary cohort(Ta ble 3). Multivariate logistic regression The model with five independent variables were identified by multivariate logistic regression analysis in primary cohort ( Table 4 ): WBC (Odds Ratio(OR)=1.115,95%CI ： 1.031-1.206,P=0.0064),PT(OR=1.412,95%CI:1.147-1.74,P=0.0011),ALB(OR= 0.925, 95%CI:0.863-0.993,P=0.0302),SCR(OR=1.007,95%CI:1.001-1.015,P=0.0348)and TG (OR=1.086, 95%CI:1.045-1.129, P<0.001). Nomogram All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint Nomogram was performed on the basis of the significantly independent and predictive variables identified by multivariate logistic regression. WBC, PT, ALB, SCR and TG were included in the predictive nomogram, which was a graphical version of the statistical model that illuminated the link between the predictive variables and the probability of in-hospital incidence of ARDS ( Figure 2) .The higher the score calculated in the nomogram, the higher the likelihood of incidence of ARDS. For example, clinical data of two patients in our study were analyzed. First patient with PT of 12.8(32points), WBC of 38.48(32points), TG of 13.6 (10points), ALB of 46(30points) and SCR of 166 (8 points) scored a total of 112 points. Therefore, the patient had approximate over 50% probability of developing into ARDS in hospital. Actually, he was transferred to ICU due to ARDS. Another patient with PT of 11.1(27points), WBC of 16.65(13points), TG of 26.95(15points), ALB of 47.6 (28points)and SCR of 53(3 points) scored a total of 86 points. The in-hospital incidence of ARDS was less than 0.1 and the patient recovered totally in less than one week without ARDS. Validation of the predictive accuracy of nomogram in primary and validation cohort In primary cohort, the nomogram displayed a good accuracy in eval uating the risk of ARDS in AP patients with an AUC=0.821 （ 95%CI:0.756-0.886） ( Figure 3A), which was more than 0.8 demonstrating a good discrimination[11]. The calibration curve showed that the predicted probability was in accordance with the observed probability(Figure 3B). In validation cohort, the nomogram also verified its accuracy with an AUC=0.823 （ 95%CI:0.707-0.937） (Figure 4A ). The calibration curve also demonstrated that there was relatively close consistency between the predicted probability and the observed probability(Figure 4B). 4 Discussion All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint AP patients varied in severity of disease, clinical course and outcomes. 20-30% of AP developed into severe complications such as acute kidney injury and ARDS. ARDS associated with high mortality was the common complication in AP[12]. Early identification of ARDS in AP patients may contribute to identifying the patients who need to be monitored and ameliorating the clinical outcomes[13]. The aim of this study was to formulate and validate an individualized predictive nomogram for in-hospital incidence of ARDS in AP patients. In the study, we analyzed and evaluated the predictive capability of clinical and laboratory parameters by logistic regression. We found several different predictors incorporating WBC, PT, SCR, ALB and TG that were independently associated with the increased risk of ARDS. Therefore, predictive nomogram was constructed. In our study, WBC counts and neutrophil counts in AP patients with ARDS were significantly higher than those without ARDS. Evidence showed that WBC as a biomarker associated with systemic inflammatory response could be potential predictors in various disorders such as cerebral vascular disease, cancer and pulmonary dysplasia[14-16]. In both gallstone AP and hypertriglyceridemia /i3 induced AP, WBC was also an effective indicator for predicting severity and prognosis, which was consistent with our study[17]. This phenomenon can be partly explained by that the etiology including gallstone and hypertriglyceridemia accounted for over 50% AP patients in our study. Due to the increased WBC and neutrophils in circulation, their toxic mediators can lead to lung tissue injury with an increased permeability of lung epithelium and endothelium, which finally results in respiratory failure such as ARDS[18]. The dynamic changes of coagulation such as prolonged PT and APTT were identified risk factors and related to the worse outcome of AP[19] 18. Several studies have illuminated that systemic inflammation reaction and coagulative dysfunction played a significant role in the development of AP, which may lead to pancreatic necrosis, acute kidney injury and ARDS[20, 21]. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint Elevated SCR as an indicator of acute kidney injury (AKI) was associated with pancreatic necrosis, organ failure and mortality in AP as well[22, 23].Early change especially in the first 24 hours after admission in SCR level was an effective predictor of disease severity in AP[24]. In the early phase of AP, inflammatory mediators and hypoperfusion in kidney contributed to the progress of AKI, which on the contrary also increased inflammatory cytokines such as IL-6 and aggravated lung injury[25]. In the ARDS group of our study, levels of serum albumin were significantly lower than those without ARDS. Hypoalbuminemia has been found to be linked with the morbidity and mortality in many different disorders such as sepsis and AKI[26, 27].In AP patients, partly due to decreased enteral energy intake and increased tissue protein catabolism , low serum albumin was an independent indicator of worse clinical outcome[28, 29]. The lower the level of serum albumin, the higher the risk of persistent organ failure[30]. Albumin as an osmotic pressure maintainer may improve hemodynamic stabilization and decrease alveolar- capillary permeability and inflammatory reaction which can ameliorate oxygenation in patients with ARDS[31]. Hypertriglyceridemia is one of the risky factors of AP as well as in cerebral cardiovascular disease and all-cause mortality[32, 33]. Some previous reports revealed that the incidence of hypertriglyceridemia /i3 induced AP varied from 10-30% in different countries[34, 35]. In China, the incidence of hypertriglyceridemia associated AP has been increased in recent years[36]. In our study, hypertriglyceridemia accounted for around 1/3 in the etiologies of AP. Evidence showed that an increased TG level was related to the severity and clinical outcomes of AP, including pancreatic necrosis, organ failure, ICU admission, and mortality[37]. Research illuminated that TG of excess amounts in circulation were hydrolyzed into high levels of free fatty acids(FFA) by pancreatic lipase[38] . FFA can impair vascular endothelium in microcirculation and result in an increment in viscosity which may cause hypoperfusion, inflammation, tissue ischemia, and organ dysfunction eventually[39]. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint As far as we know, this is the first study to develop a nomogram for predicting the in-hospital incidence of ARDS in AP patients. For each AP patient, our nomogram enables physicians to directly and conveniently calculate a numeric probability of ARDS. It also helps physicians for clearly communicating with a patient via individual nomogram evaluation and making individualized treatment plans. In our study, there are still several limitations. First, hospitals in this study were tertiary hospitals in urban city, which may lead to the difference of clinical characteristics in AP compared to the patients in hospitals which located in the suburb of city and countryside or other areas. It could partly explained that why the etiologies and variables associated with ARDS in AP in our study differed from some other researches[40]. When applying the proposed nomogram to a larger population or other centers, caution must be used. Multicenter research with larger samples needs to be explored for validating our model. Second, the different models in three subgroups of ARDS couldn’t be performed in this study due to its relatively small samples. For better applying the model in different patient, studies should be carried out by expanding the sample size in the future. Third, our study was retrospective and there might be patient selection biases, which was an inevitable limitation in these types of studies. Moreover, due to its retrospective nature and part of data missing(such as C-reactive protein, procalcitonin), not all the variables potentially associated ARDS in AP can be evaluated and analyzed for constructing the model. Further prospective research should be done for comprehensively analyzing the association between clinical characteristics and ARDS in AP. Researches should also focus on the management including some drugs or related interventions (such as glucocorticoids, early ventilation support, blood purification, etc.) in patients with early-stage of ARDS in AP. 5 Conclusion In conclusion, an intuitive nomogram with easily available laboratory parameters for the prediction of in-hospital incidence of ARDS in patients with AP was performed. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint The incidence of ARDS for an individual patient can be fast and conveniently evaluated by the nomogram. List of abbreviations ARDS=acute respiratory distress syndrome; AP=acute pancreatitis; SAP=severe acute pancreatitis; ICU=Intensive care unit; LOS=length of stay; SBP=systolic blood pressure; DBP=diastolic blood pressure; HR=heart rate; RR=respiratory rate, SaO2 = oxyhemoglobin saturation; WBC=white blood cell count; N=neutrophil; L=lymphocyte; HCT=hematocrit; PLT=platelet; FBS=fasting blood sugar; PT=prothrombin time; APTT=activated partial thromboplastin time; ALB= albumin; ALT=alanine aminotransferase; BUN=blood urea nitrogen; SCR=serum creatinine; AMS=amylase; LPS=lipase; LDH=lactate dehydrogenase; HDL=high density lipoprotein; LDL=low density lipoprotein; TG=triglyceride; SOFA=sequential organ failure assessment; BISAP=bedside index of severity in acute pancreatitis; SIRS= systemic inflammatory response syndrome, AUC= area under receiver operating characteristic curve, OR= odds ratio, AKI=acute kidney injury, FFA=free fatty acid. Declarations Acknowledgments No Funding This manuscript was supported by the Sanitation and Health Committee Foundation of Hunan Province, China (NO.20200075). Ethics approval and consent to participate The study was approved by institutional review board of Changsha Central Hospital of University of South China and the Second Xiangya Hospital of Central South University. Due to retrospective characteristics of the study, informed consent was waived. Author Contributions The manuscript writing and patient’s data recording were done by Ning Ding. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint Cuirong Guo, Kun Song, Changluo Li ， Yang Zhou and Guifang Yang assisted in information collection. Xiangping Chai analyzed and interpreted the patients' general indices. The final manuscript was read and ratified by all authors. Availability of data and materials Datasets used and/or analyzed in the present study were availed by the corresponding author on reasonable request. Disclosure Statement There are no real or apparent conflicts of interest to disclose. Consent for publication Not applicable

1. Tenne r S , Bail li e J, DeW it t J , Vege S S: American Col le ge of G as troen tero logy Gu i del i nes : Mana g emen t of Acut e Panc reatit is (v ol 10 8 , pg 14 00 , 20 13) . A me rican Journa l o f Gastroen tero logy 201 4 , 109 ( 2) :302 - 302. 2. Banks P A , Bol l en TL , Derv eni s C , G ooszen H G , Jo hn son CD , Sa rr M G , Tsio tos G G , Vege SS , Acute P an c reat i t i s C : Clas s if i c ation o f ac u te pancrea ti tis-2012 : rev i sio n of the A t lan ta c las s if i ca t io n and d ef in i t io ns by inter n at iona l conse n sus . Gut 2 013 , 62(1) :10 2 -111 . 3. Abe T , Madotto F , Pha m T , Nagat a I, U chid a M, Ta m iy a N, Kura hashi K , Bellan i G , Laffey J G, Inv es t iga tors L-S et a l: Epidem iology and pat te rns o f tra c heos to m y pr ac t ice in pa t ie nts w ith acu t e respira tor y d is tre s s s ynd ro me in ICU s acro s s 50 count r ies . C r iti c al C ar e 2018 , 22 . 4. Fei Y , Gao K, Li W - q: Pred i ct ion and evaluation o f the se v er ity of acut e res p i rat ory dis t ress s ynd ro me f ol low ing s e vere a c ute pa ncreat it is using an ar ti fic i a l ne ural network algor ith m mo del . H pb 201 9, 21 (7 ):891 -897. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint 5. Domber nowsk y T , K ri stensen M O , Ry sg a ard S , Gluu d LL, Novovic S : Risk facto r s f or and impac t of r es pi ra tory fa i lure o n mor t ality in the e a r ly pha s e o f acute pan creat it is . Pancreatology 2016, 16(5) :756 -760. 6. Reil ly JP, Chr is ti e JD: Is I t Possib l e to P r event ARD S? Jama -Journa l o f the A merican Medica l A ssociation 2016 , 31 5(22 ):24 03 - 2 405. 7. Berard i G, Mo r ise Z , Sposi to C, Iga rash i K , Panetta V , S i mone lli I, K i m S, G oh B KP, Kubo S, Tanak a S e t a l : Developmen t o f a nomogra m to pred ict ou tcome a f ter l i v er r e s e c t i o n f o r h e p at o c el lu l ar c a r ci n oma in C hi ld -P u g h B c i r rh o s is . Journ a l of Hepato l ogy 202 0 , 72(1) :75- 8 4. 8. Ga ndag l i a G , P lou s sard G, Va ler io M , M attei A, Fi or i C , Fo ss a ti N, St abi l e A, B e a uval JB, Ma la vau d B , R oumigu ie M e t a l : A N ov el N omog ra m to Iden tify C an d idates fo r Extend ed Pel v ic Ly mp h Node D issec t io n A mo ng P a t i ents w ith Cl inica l ly Loca li zed Prosta te C an ce r Diagn os ed wi th Ma gnetic Resonan ce I magi ng- targeted and Systemat ic Biop si es . European U ro logy 2 019, 75 ( 3) :506-514 . 9 . R a i t h E P , U d y A A , B a i l e y M , M c G l ou g h l i n S , M a c I s a a c C , B e l l om o R , P i l c h e r D V , An z i c s, Core : Prognos t ic Accur a cy o f th e S O FA S core , SIR S C r iteri a, and qS OFA Score f o r In-H ospi tal Mo rt alit y Among Ad ults W ith Suspe c ted I nfe ct i on Ad mi tt e d to the Intensive Care U nit . Ja ma-J ou rna l of the A meri ca n M edical Ass oc iation 2 017 , 317(3) :2 90 -300 . 10. Ranier i VM , Rub enfeld GD , Th ompso n BT, Fer guson ND, C a l dw ell E , Fan E, Campor ota L , S lutsky AS , Force AD T : Acute R es pi ra tory D ist ress Syndr ome The Berl in Def init ion . J ama- J our nal of t he Ameri c an Medi c al As s oc i ati on 20 12 , All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint 307(23 ) :2526 -2533. 11. Zhang ZH : Mode l b u i ldi n g s t r ategy for log is t ic r e g ress i on : pu rpo se fu l se le c ti on . Ann als of T ransla tion al Medici n e 2016, 4 (6 ). 12. Wi rt z TH, P u eng el T , Buendg en s L, L ue d de T , T r a utw ein C , K och A: Diagnosis and Therapy o f sever e acute Panc reati ti s in I ntensive C a re Med ic in e . Deutsc he Medizin is ch e Woc hen sch rift 202 0, 145 ( 1 2):850- 8 60 . 13. Fan E , B r od ie D, S lutsky AS: Acute R e spir ator y D ist res s S yn dro me Adv a nces in Diagnosis and T rea tmen t . J ama-Jou rna l of the A merican Medica l Ass o ciation 201 8 , 319(7) :6 98 -710 . 14. Liu S , L iu XQ, Chen SY, Xia o YX, Zh ua ng WD : Neutrophi l- lymphoc y te rat io predict s the o u tco me of in tr a cereb ra l h e mo rrhage A me ta-a nal y si s. M e dic ine 2019 , 98(26 ). 15. Sun YY, C h en CE , Zh an g XX , Weng XC, Sheng A Q, Zh u YK , C hen S J, Zheng X X, Lu CS : High Neu trophi l- to-L y mpho c yt e Ra ti o Is a n E a r ly P red ict o r of B ron ch opulmo nary Dyspl as ia . Fr o n ti er s i n Pe di a tri c s 2019 , 7 . 16. Hayama T , H a s higuchi Y, Okada Y , O no K , Nem o to K, Sh imada R , Oz aw a T , Toy o da T, Ts uc hiya T, Iinu ma H e t a l : Si gnif i can c e of t he 7th p osto pe ra tive day neutrop h i l- to- l y mpho c yte ratio in c olo rec tal can c er . Inte rnation al Jour nal of C o l o re cta l Diseas e 2020, 35(1 ):1 19-12 4 . 17. Huang L , Ch en C Y , Y an g LJ , Wan R , H u GY: N e utroph il- to- lymp h ocy te rat io can spe ci fical l y p redict the s eve rity o f hy p ert rig l yce r i dem ia -i ndu ce d acu te panc reati ti s compa red w ith wh ite blo od ce l l . Jou rnal o f C lin i ca l Labo rato ry A na lysis 20 19, 33(4) . 18. Zeman s RL , Ma tt h ay MA : What dr iv es neutroph ils to the a l veoli in ARD S? Tho rax All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint 2017 , 72 (1 ): 1-3 . 19. Liu CN, Z h ou XF, L ing LQ, Chen S , Zho u J: P r ed ic t io n o f m o r t a lit y a n d or ga n f a il ur e bas ed o n c oa gulat ion and f ibr in o lysis m arke r s i n p at ien ts w ith acu te pa nc reat i t i s A ret rosp ec ti v e s tud y . Medic in e 2019 , 98(2 1). 2 0 . A k b a l E, De mi rc i S , Koc a k E , Kokl u S , B as a r O, T u na Y: Al t er a ti o n s o f pl a t e l e t f u n cti o n and coagulat ion pa ra meter s du rin g acute pan creati t is . Blood Coagulation & Fibrino lysis 2 013 , 24(3 ):2 4 3-246 . 21. Gomerc ic C , Gelsi E , Van G ys el D , Fr in A C, Ou v r ier D , Tonoh o uan M, A ntune s O , Lomb a rdi L , De Gal lea n i L, Vanbierv liet G e t a l : As sessmen t o f D -Di mer s for the Early Predict ion o f Co mpli ca t i o n s i n A c ut e Pa nc r e a ti t i s . P anc reas 20 16, 45(7) :9 80 -985 . 22. Papac hr is tou GI , M uddana V , Yad av D , Whi tc o mb DC : Increas ed Se ru m Crea ti n i ne Is A s s o c i at e d Wi th P a nc re at ic N ec ro si s in A c ut e P a n c r e ati t i s . Amer i c an J o u r nal o f Ga s troe ntero logy 2010, 10 5(6) :1451 -1452. 2 3 . W a n J H , S h u W Q , H e W H , Z h u Y , Z h u Y , Z e n g H , L i u P , X i a L , L u N H : Ser um Cr e at inine Level and APAC HE- I I Sco re w it hin 24h of Ad mission Ar e Ef fec tive f o r Predict ing Pe rsisten t Organ Fa i l ure in Ac u te Panc rea titi s . Ga stroen tero logy R esea rch and Pra c tice 2019 . 24. Lipinski M , Rydz e w ski A, R y dz ewsk a G: Early changes in ser u m c rea tinine lev e l and est i ma ted glom eru l ar fil tra tion rate pred i ct pa nc reat i c ne crosis and mo rta l ity in a cute panc rea ti tis C reat inine a nd e G FR in ac u te p a ncrea tit is. Pa nc r e at ol o g y 2 013 , 13(3) :20 7 -211 . 25. Seeley EJ: Updates in the Managemen t of Ac u te Lun g I nju ry: A Fo c us o n the Ove rlap All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint Between AK I and ARD S . Adva nce s in Ch ronic K i dney Dis ease 2013 , 20 (1 ):14 -20 . 26. Holder AL , Gup ta N, L u la j E , Furg iue le M, H idalgo I, J o nes MP , Jolly T , Gen n is P, Birnba u m A : P redi c tor s o f ea r ly progr essi on to severe se p si s or shock a m ong emergency de pa rt ment pat ients w ith no nsev ere sep si s . In te rnationa l Jou rna l of Emerg enc y Medici ne 2016, 9 . 27. Wied e r m ann C J: Hypoa l bum i ne mi a and th e R isk o f Acut e K idney In j ury in S epsis . Critica l C a re Med i cine 2 0 19, 47 (4 ) : E 3 7 7 -E3 7 8 . 28. Hong WD , Li n SH , Z ipp i M, G eng W J, St ock S , B as har at Z , Chen g BC , Pa n JY , Z hou MT: S e r um A lbu m in Is In dependent ly Assoc ia ted w i th Pers i s tent O rgan Fa ilur e in Acute P an c rea ti t i s. Ca nadian Jou rna l of Gastroen te rology and H epa tolo gy 2 017 , 2017 . 29. Yue W , L iu Y, Ding W , Ji a ng W, Huang J , Zhang J , L iu J : The p r edic t i ve v a lue of the preal b um in- to - fibr inoge n rat io i n pa tie n ts with acute pa nc r e a t it is . In ternationa l Jou rnal of C linic al P ra ct ice 2015 , 69 (10 ):11 21-1128. 30. Li SK, Zha n g Y S, Li MJ, X ie C , Wu HS : Serum a l bu m in, a good i nd icator o f per s is t en t organ fa ilu re in acute pan c rea ti t i s (vo l 17 , 59, 2017 ) . Bm c Gas troente ro log y 2017, 17. 31. Uhlig C, S i lv a PL , D e c ker t S, Schmi tt J , de Ab reu M G: A lbu m in v e r s us c r y s t a ll o id sol u tions in p at ien ts w i th the a cute res pi r at ory d ist res s syndro me : a sy s te matic re view and meta -ana l ys i s . Critica l Care 2014 , 18 (1) . 3 2 . C a r r R A , R e j ow s k i B J , C o t e G A , P i t t H A , Z y r om s k i N J : S y ste matic review o f hyp ert r iglycer idemia - ind uced acute pancrea titi s: A mor e v iru l en t e tiolo gy? Pancreatology 2016, 16(4) :469 -476. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint 33. Nordes tga ard B G : Tr i g lyc e ride -Ric h L i po prote i ns and A ther os c lero tic C a r d iov as c ul ar Diseas e New In s igh ts F rom Ep ide mi olog y , G en et i cs , and Bio logy . C ircula tion Researc h 2016 , 118(4 ):5 47-56 3 . 34. Charle s w or th A , S te ger A, C rook MA: Ac ute pan c reat it i s as soc i a ted w ith se ve r e hyp ert r iglycer idaemi a; A r e t r o specti ve c o hort study . Inte rna tional J ourna l of Surgery 2015 , 23 :2 3-27. 35. Lepp anie mi A , Tolonen M , Ta rasconi A, Sego v ia-L oh se H , G a mbe r in i E , K irkpat ric k AW, Ball C G, Par ry N , Sar te lli M , Wolbrink D e t a l : 2019 WSES gu id e lin e s for the mana ge men t o f se v ere ac u te pancrea ti t i s . World Journal of E mergency Su rgery 20 19 , 14. 36. Zhu Y , Pa n XL , Z e ng H , H e W H, X ia L , L i u P, Zhu Y , C he n YX , L v NH : A Study on the Etio lo gy , Sev er i ty, a nd Mo rta lity of 3260 Pati en ts With Acut e Pa n crea t it i s Acc o rdi n g to the Rev i sed A tlanta C l as sifi ca t ion in Jian gx i, Ch ina O v er an 8-Yea r Pe riod . P a nc reas 2017 , 46 (4 ): 504-509. 37. Gu o YY , Li H X , Zhang Y , H e WH: Hype r tr i g lyc e r idemia -indu c ed Acu te Pancr eati ti s : Progres s on Disease Mecha n is ms and Treat ment Mo da l it ie s . Disc ove ry Med icine 2019 , 27 (14 7) :10 1- 109. 3 8 . Ewa l d N, Ha r dt PD, Kl oe r HU: Sev ere h y per tr ig ly ce ridem ia and panc rea t it is: prese n tat ion and ma nage ment . Cu rre nt Opi ni on i n Li pi dol og y 2009, 20(6 ):49 7-504. 39. Valdivie l so P, R am ire z-Buen o A, E wa ld N : Cur rent kn ow l edg e of hy per tr ig l yce ri d emi c panc rea ti tis . Euro pean Journ a l of Intern a l Me d ic in e 2014 , 25(8 ):68 9-694. 40. Zhou M -T, C he n C -S , Chen B - C, Zh an g Q-Y , And e rsson R : Acu te l ung in jury and All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint AR DS in acute panc rea t i tis: Me cha n isms and po t ent ia l i n terv en t ion. Wo rld Jo urna l of Ga s troe ntero logy 2010, 16 ( 1 7) :2094 -2099. Figure legends Figure 1. Flow diagram of patients enrollment and study design. Figure 2. Nomogram indicating the risk of ARDS in AP. Abbreviations: white blood cell count(WBC) ， prothrombin time(PT), albumin(ALB), serum creatinine(SCR), triglyceride(TG). Figure 3. V alidation of nomogram in primary cohort. (A) Discrimination: AUC of the ROC curve was 0.821 (95%CI: 0.756-0.886). (B) Nomogram calibration curve. The blue line indicates perfect prediction by an ideal model. The orange line indicates actual performance of the model. Figure 4. Validation of nomogram in validation cohort. (A) Discrimination: AUC of the ROC curve was 0.823 (95%CI: 0.707-0.937). (B) Nomogram calibration curve. The blue line indicates perfect prediction by an ideal model. The orange line indicates actual performance of the model. Table 1 General characteristics of 779 patients with AP non-ARDS(n=728) ARDS(n=51) P-value Demographics Male, n (%) 509(69.92%) 42(82.35%) 0.059 Age, median( IQR)(years) 45(36-55) 41(35-51) 0.617 Diabetes mellitus, n(%) 111(15.25%) 8(15.69%) 0.933 Hypertension, n(%) 134(18.41%) 9(17.65%) 0.892 Coronary heart disease, n(%) 23(3.16%) 2(3.92%) 0.766 Smoking, n(%) 262(35.99%) 23(45.1%) 0.067 Etiology Alcohol, n(%) 137(18.82%) 12(23.53%) 0.409 Hypertriglyceridemia, n(%) 283(38.87%) 29(56.86%) 0.011 Biliary, n(%) 148(20.33%) 6(11.76%) 0.138 Vital signs SBP, median (IQR)(mmHg) 130(121-142) 132(120-144) 0.342 DBP, median (IQR)(mmHg) 81(75-89.75) 82(75-90) 0.919 HR, median(IQR)(beats/min) 80(74-88) 99(80-116) <0.001 RR, median (IQR)(beats/min) 18(17-19) 18(17-20) 0.125 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint SaO2,median(IQR)(%) 96(96-97) 98(94-98) 0.225 Temperature (°C ) 36.7(36.5-36.9) 36.9(36.8-37.3) 0.025 Clinical outcomes Ventilation, n(%) In-hospital mortality, n(%) LOS in hospital , median(IQR)(days) 0(0%) 0(0%) 7(5-9) 19(37.25%) 3(5.882%) 12(8-19) <0.001 <0.001 <0.001 LOS in ICU, median (IQR)(days) 0(0-0) 6(4-8) <0.001 Laboratory findings WBC, median (IQR)(^109/l) 12.32(9.27-15.15) 15.425(12.36-18.20) <0.001 N, median (IQR)(^109/l) 9.76(6.65-12.51) 12.46(9-14.84) <0.001 L, median (IQR)(^109/l) 1.49(1.01-2.16) 1.19(0.8-2.02) 0.025 NLR, median (IQR) 6.25(3.61-10.41) 11.24(6.20-19.92) 0.001 HCT, median (IQR) 0.43(0.40-0.47) 0.46(0.41-0.48) 0.006 PLT, median (IQR)(^109/l) 209(167-250) 211(174-253.25) 0.808 FBS, median (IQR)(mmol/l) 7.4(6.4-9.5) 9.8(7.77-13.45) <0.001 PT, median (IQR)(s) 11.4(10.7-12.2) 12.75(11.7-14.42) <0.001 APTT, median (IQR)(s) 27.8(24.3-31.75) 32.1(26.52-36.12) <0.001 ALB, median (IQR)(g/l) 43.7(40.2-46.9) 39.9(36.2-46.6) 0.049 ALT, median (IQR)(u/l) 33(20-71.12) 40(22.9-87) 0.683 BUN, median (IQR)(mmol/l) 4.7(3.67-5.99) 4.9(4.33-6.61) 0.017 SCR, median (IQR)(umol/l) 66.95(54.22-77) 76(59-123) 0.012 AMS, median (IQR)(u/l) 310.5(113.25-954.75) 369(114-1150) 0.932 LPS, median (IQR)(u/l) 385(124-859.5) 404(221-876) 0.570 LDH, median (IQR)(u/l) 228(180-304.75) 398(219-543) <0.001 Calcium, median (IQR)(mmol/l) 2.4(2.3-2.5) 2.34(2.07-2.51) 0.024 HDL, median (IQR)(mmol/l) 0.92(0.65-1.3) 0.61(0.46-0.91) 0.089 LDL ,median (IQR)(mmol/l) 2.18(1.39-2.96) 1.48(1.09-2.438) 0.928 TG, median (IQR)(mmol/l) 5.59(1.43-17.77) 17.16(2.56-26.69) 0.003 PaO2, median (IQR)(mmHg) 78.5(75.3.2-82.5) 77.1(74.6-81.8)- 0.068 PaCO2,median (IQR)(mmHg) 38.8(36-41.5) 37.4(36.3-41.6) 0.085 SIRS,n(%) 276(37.9%) 26(50.98%) 0.026 SOFA 0(0-1) 2(1-3) <0.001 RANSON 1(0-2) 4(2-5) <0.001 BISAP 1(0-2) 2(2-3) <0.001 Abbreviations: ICU, Intensive care unit; LOS, length of stay; SBP, systolic bl ood pressure; DBP, diastolic blood pressure; HR, heart rate; RR, respiratory rate ,SaO2, oxyhemoglobin saturation; WBC, white blood cell count; N, neutrophil; L, lymphocyte; HCT, hematocrit; PLT, platelet; FBS, fasting blood sugar; PT, prothrombin time; APTT, activated partial thromboplastin time; ALB, albumin; ALT, alanine aminotransferase; BUN, blood urea nitrogen; SCR, serum creatinine; AMS, amylase; LPS, lipase; LDH, lactate dehydrogenase; HDL, high density li poprotein; LDL, low All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint density lipoprotein; TG, triglyceride; SOFA, Sequential organ failure assessment; BISAP, bedside index of severity in acute pancreatitis; SIRS, systemic inflammatory response syndrome; NLR is defined as the ratio of neutrophils to lymphocytes. Table 2 Comparison primary cohort and validation cohort Primary cohort (n=560) Validation cohort (n=219) P-value Demographics Male, n (%) 384(68.57%) 167(76.25%) 0.378 Age, median( IQR)(years) 45(36-55) 45(36-57) 0.094 Diabetes, n (%) 89(15.89%) 30(13.69%) 0.001 Hypertension, n (%) 106(18.92%) 37(16.89%) 0.133 Coronary heart disease, n(%) 22(3.92%) 3(1.36%) 0.04 Smoking n(%) 199(35.53%) 86(39.27%) 0.402 Etiology Alcohol, n (%) 110(19.64%) 39(17.80%) 0.02 Hypertriglyceridemia, n (%) 225(40.17%) 87(39.72%) 0.081 Biliary, n (%) 102(18.21%) 52(23.74%) 0.148 Vital signs SBP, median (IQR)(mmHg) 130(120-142) 132(122-142) 0.76 DBP, median (IQR)(mmHg) 81(74-90) 82(75-89) 0.258 HR, median (IQR)(beats/min) 80(74-90) 80(75-90) 0.418 RR, median (IQR)(beats/min) SaO2 ,median(IQR)(%) Temperature(°C) ARDS, n (%) 18(17-19) 96(96-97) 36.7(36.5-36.9) 36(6.446%) 18(17-20) 97(96-97) 36.7(36.6-37) 15(6.829%) 0.422 0.356 0.525 0.416 Laboratory findings WBC, median (IQR)(^109/l) 12.49(9.42-15.4) 12.55(9.43-15.52) 0.13 N, median (IQR)(^109/l) 9.91(6.82-12.76) 10.16(6.83-12.58) 0.517 L, median (IQR)(^109/l) 1.44(0.98-2.1) 1.50(1.04-2.27) 0.445 NLR, median (IQR) 6.7413.683-11.415) 6.21(3.729-9.626) 0.606 HCT, median (IQR) 0.43(0.4-0.46) 0.44(0.41-0.48) 0.073 PLT, median (IQR)(^109/l) 208(167-248) 212(170.25-254.75) 0.222 FBS, median (IQR)(mmol/l) 7.5(6.4-9.7) 7.7(6.4-10) 0.094 PT, median (IQR)(s) 11.5(10.8-12.3) 11.4(10.7-12.3) 0.268 APTT, median (IQR)(s) 28.1(25.2-32.5) 27.15(23.72-31.2) 0.147 ALB, median (IQR)(g/l) 43.5(40-46.8) 43.7(40-46.9) 0.072 ALT, median (IQR)(u/l) 33(20-68) 38(21.15-81.25) 0.986 BUN, median (IQR)(mmol/l) 4.7(3.7-5.99) 4.905(3.76-6.42) 0.175 SCR, median (IQR)(umol/l) 66(54-57) 69(57.85-79.8) 0.06 AMS, median (IQR)(u/l) 323(113.5-946) 284(116.25-1200.35) 0.001 LPS, median (IQR)(u/l) 398.5(132.75-862.25) 369(126.75-868.25) 0.004 LDH, median (IQR)(u/l) 230(180.75-313) 233(184.5-328) 0.217 Calcium, median (IQR)(mmol/l) 2.39(2.29-2.49) 2.41(2.29-2.51) 0.827 HDL, median (IQR)(mmol/l) 0.91(0.63-1.28) 0.905(0.58-1.29) 0.193 LDL, median (IQR)(mmol/l) 2.15(1.33-2.9) 2.115(1.40-3.12) 0.191 TG, median (IQR)(mmol/l) 5.88(1.43-18.6) 5.635(1.53-21.00) 0.699 PaO2, median (IQR)(mmHg) 78.4(75.5-81.6) 77.9(75.2-82.1) 0.185 PaCO2, median (IQR)(mmHg) 38.5(36.5-41.4) 38.2(36.8-41.5) 0.108 SIRS,n(%) 210((37.5%) 92(42.01%) 0.134 Abbreviations: SBP, systolic blood pressure; DBP, diastolic blood pressure; HR, heart rate; RR, respiratory rate,SaO2, oxyhemoglobin saturation; WBC, white blood cell count; N, neutrophil; L, All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint lymphocyte; HCT, hematocrit; PLT, platelet; FBS, fasting blood sugar; PT, prothrombin time; APTT, activated partial thromboplastin time; ALB, albumin; ALT, alanine aminotransferase; BUN, blood urea nitrogen; SCR, serum creatinine; AMS, amylase; LPS, lipase; LDH, lactate dehydrogenase; HDL, high density lipoprotein; LDL, low density lipoprotein; TG, triglyceride; SOFA, Sequential organ failure assessment; BISAP, bedside index of severity in acute pancreatitis; SIRS, systemic inflammatory response syndrome; NLR is defined as the ratio of neutrophils to lymphocytes. Table 3 Comparison of variables between ARDS and non-ARDS groups in primary cohort non-ARDS(n=524) ARDS(n=36) P-value Demographics Male, n (%) 355(67.75%) 20(85.55%) 0.11 Age, median( IQR)(years) 46(36-55)- 43(34-50.75) 0.375 Diabetes mellitus, n(%) 84(16.03%) 5(13.88%) 0.726 Hypertension, n(%) 99(18.89%) 7(19.44%) 0.97 Coronary heart disease, n(%) 20(3.81%) 2(5.55%) 0.663 Smoking, n(%) 185(35.3%) 14(38.89%) 0.368 Etiology Alcohol, n(%) 102(19.46%) 8(22.22%) 0.705 Hypertriglyceridemia, n(%) 206(39.31%) 19(52.77%) 0.13 Biliary, n(%) 98(18.7%) 4(11.11%) 0.254 Vital signs SBP, median (IQR)(mmHg) 130(120-142) 132(116.5-145) 0.448 DBP, median (IQR)(mmHg) 81(74-90) 80.5(75.25-90) 0.964 HR, median(IQR)(beats/min) 80(74-88) 99.5(80.25-119.5) <0.001 RR, median (IQR)(beats/min) 18(17-19) 19(19-20) 0.224 SaO2,median(IQR)(%) 97(96-97) 96(94-98) 0.306 Temperature (°C ) 36.7(36.5-36.9) 36.9(36.8-37.3) 0.027 Laboratory findings WBC, median (IQR)(^109/l) 12.28(9.35-15.14) 15.55(12.663-19.37) <0.001 N, median (IQR)(^109/l) 9.68(6.675-12.525) 13.01(9.23-17.84) <0.001 L, median (IQR)(^109/l) 1.48(0.99-2.11) 1.12(0.81-1.67) 0.018 NLR, median (IQR) 6.31(3.61-10.81) 12.34(6.78-19.33) 0.004 HCT, median (IQR) 0.43(0.4-0.46) 0.45(0.41-0.48) 0.083 PLT, median (IQR)(^109/l) 208(167-247.5) 212(171-253.75) 0.949 FBS, median (IQR)(mmol/l) 7.4(6.4-9.42) 10(7.72-13.55) <0.001 PT, median (IQR)(s) 11.4(10.8-12.2) 12.7(11.7-14.8) <0.001 APTT, median (IQR)(s) 28(25.02-32.1) 32.3(26.7-36.1) 0.01 ALB, median (IQR)(g/l) 43.7(40.3-46.8) 39.6(34.6-46.45) <0.001 ALT, median (IQR)(u/l) 32.7(20-64) 40(21-129.5) 0.297 BUN, median (IQR)(mmol/l) 4.69(3.67-5.96) 4.82(4.34-6.85) 0.103 SCR, median (IQR)(umol/l) 65.1(53.8-75) 78.5(59.5-121) <0.001 AMS, median (IQR)(u/l) 318(113-899) 438.5(118-1219) 0.752 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint LPS, median (IQR)(u/l) 397(126-843.75) 442.5(200.25-920.5) 0.31 LDH, median (IQR)(u/l) 226(178.75-302.25) 403(222.5-551.25) <0.001 Calcium, median (IQR)(mmol/l) 2.39(2.29-2.49) 2.31(2.14-2.48) 0.003 HDL, median (IQR)(mmol/l) 0.92(0.65-1.29) 0.67(0.43-0.96) 0.034 LDL ,median (IQR)(mmol/l) 2.18(1.36-2.91) 1.515(1.128-2.33) 0.431 TG, median (IQR)(mmol/l) 5.74(1.41-17.09) 13.21(2.48-25.06) <0.001 PaO2, median (IQR)(mmHg) 78.1(75-82.6) 76.1(73.6-80.5)- 0.087 PaCO2,median (IQR)(mmHg) 38.5(36-41.2) 37.2(36.1-40.6) 0.132 SIRS,n(%) 193(36.83%) 17(47.22%) 0.045 Abbreviations: SBP, systolic blood pressure; DBP, diastolic blood pressure; HR, heart rate; RR, respiratory rate,SaO2, oxyhemoglobin saturation; WBC, white blood cell count; N, neutrophil; L, lymphocyte; HCT, hematocrit; PLT, platelet; FBS, fasting blood sugar; PT, prothrombin time; APTT, activated partial thromboplastin time; ALB, albumin; ALT, alanine aminotransferase; BUN, blood urea nitrogen; SCR, serum creatinine; AMS, amylase; LPS, lipase; LDH, lactate dehydrogenase; HDL, high density lipoprotein; LDL, low density lipoprotein; TG, triglyceride; SOFA, Sequential organ failure assessment; BISAP, bedside index of severity in acute pancreatitis; SIRS, systemic inflammatory response syndrome; NLR is defined as the ratio of neutrophils to lymphocytes. Table 4 Multivariate logistic regression analysis of primary cohort Variables 95% CI for Odds Ratio B S.E. Wald P value Odds Ratio Lower Upper WBC 10.90% 4.00% 7.447 0.0064 1.115 1.031 1.206 PT 34.56% 10.63% 10.576 0.0011 1.412 1.147 1.740 ALB -7.70% 3.55% 4.700 0.0302 0.925 0.863 0.993 SCR 0.75% 0.36% 4.455 0.0348 1.007 1.001 1.015 TG 8.26% 1.96% 17.721 <0.001 1.086 1.045 1.129 Abbreviations: white blood cell count(WBC), prothrombin time(PT), albumin(ALB), serum creatinine (SCR), triglyceride(TG). All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint AP cases in database(n=817) AP cases included in the study(n=779) Primary cohort(n=560) .Risk factors for in-hospital ARDS .Development of Nomogram Validation cohort(n=219) .V alidation of nomogram Patients excluded(n=38): .COPD(n=9) .pregnancy(n=2) .Cancer(n=16) .others(n=11) All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint (A) (B) All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint (A ) ( B) All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted October 5, 2020. ; https://doi.org/10.1101/2020.02.09.20019513doi: medRxiv preprint