- Email: [email protected]
Rapid Emergency Medicine Score as a main predictor of mortality in Vibrio vulnificus–related patients
American Journal of Emergency Medicine 31 (2013) 1037–1041
Contents lists available at SciVerse ScienceDirect
American Journal of Emergency Medicine journal homepage: www.elsevier.com/locate/ajem
Original Contribution
Rapid Emergency Medicine Score as a main predictor of mortality in Vibrio vulnificus–related patients☆,☆☆ Sheng-Hung Kuo MD a, b, Chin-Feng Tsai MD, PhD a, c, d, Chi-Rong Li PhD a, e, Shih-Jei Tsai MD, PhD a, c, d, Wai-Nang Chao MD a, f, g, Khee-Siang Chan MD, PhD a, f, h, Yuan-Ti Lee MD, PhD a, c, d, 1, Ruey-Hong Wong PhD a, i, Chun-Chieh Chen MD, PhD a, c, j, 2, Shiuan-Chih Chen MD, PhD a, c, j,⁎ a
Institute of Medicine, Chung Shan Medical University, Taichung, Taiwan Department of Surgery, Yuan-Lin Kuo Hospital, Changhua, Taiwan c School of Medicine, Chung Shan Medical University, Taichung, Taiwan d Department of Internal Medicine, Chung Shan Medical University Hospital, Taichung, Taiwan e School of Nursing and Institute of Nursing, Chung Shan Medical University, Taichung, Taiwan f Department of Emergency Medicine, Chi Mei Medical Center, Tainan, Taiwan g Department of Surgery, Chi Mei Medical Center, Tainan, Taiwan h Department of Critical Care Medicine, Chi Mei Medical Center, Tainan, Taiwan i School of Public Health, Chung Shan Medical University, Taichung, Taiwan j Department of Family and Community Medicine, Chung Shan Medical University Hospital, Taichung, Taiwan b
a r t i c l e
i n f o
Article history: Received 23 October 2012 Received in revised form 6 February 2013 Accepted 16 March 2013
a b s t r a c t Objectives: Vibrio vulnificus causes potentially life-threatening and rapidly progressing infections. Therefore, the severity-of-illness assessment appears to be important for V vulnificus–infected patients at the time of admission. The aim of our study was to evaluate the performance of the severity-of-illness scoring model on admission in V vulnificus–infected patients. Methods: One hundred seventy-one consecutive patients (mean age: 63.1 ± 12.3 years) with V vulnificus infection who were admitted to a teaching hospital between January 1999 and June 2010 were included in the study. Demographic and clinical characteristics, illness severity on admission, treatment, and outcomes were collected for each patient and extracted for analysis. Logistic regression and receiver operating characteristic curve analyses were performed. Results: The mean Rapid Emergency Medicine Score (REMS) on admission was 6.5 ± 3.0 points. During hospitalization, 68 patients (40%) required intensive care. The overall case-fatality rate was 25%. In multivariate analysis, the presence of underlying liver disease (P = .002), hemorrhagic bullous lesions/ necrotizing fasciitis (P = .012), and higher REMS values on admission (P b .0001) were associated with increased mortality risk; a time span b24 hours between arrival and surgical treatment was associated with a decreased mortality risk (P = .007). Additionally, the area under the receiver operating characteristic (ROC) curve for the REMS in predicting mortality risk was 0.895 (P b .0001). An optimal cut-off REMS ≥ 8 had a sensitivity of 81% and a specificity of 85%, with a 26.6-fold mortality risk (P b .0001) and a 12.5-fold intensive care unit admission risk (P b .0001). Conclusion: The REMS could provide clinicians with an effective adjunct risk stratification tool for V vulnificus– infected patients. © 2013 Elsevier Inc. All rights reserved.
1. Introduction
☆ Funding/Support: None. ☆☆ Competing Interests Declared: None. ⁎ Corresponding author. Faculty of Medicine, School of Medicine and Institute of Medicine, Chung Shan Medical University, Taichung 40201, Taiwan. Tel.: +886 4 24739595x34970; fax: +886 4 23248137. E-mail address: [email protected] (S.-C. Chen). 1 Dr. Yuan-Ti Lee made an equal contribution to this work as the second author. 2 Dr. Chun-Chieh Chen made an equal contribution to this work as the first author. 0735-6757/$ – see front matter © 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ajem.2013.03.030
Vibrio vulnificus is a virulent, Gram-negative, halophilic, rodshaped, motile bacterium that often flourishes in warm estuarine seawaters or brackish environments and is estimated to cause 0.2 to 36 infections per million individuals annually [1-3]. V vulnificus infections can progress rapidly, and without prompt and appropriate management, the host’s condition may sharply deteriorate in 1 to 2 days. Human and animal studies have shown that V vulnificus may enter the host through skin wounds or gastrointestinal mucosal
1038
S.-H. Kuo et al. / American Journal of Emergency Medicine 31 (2013) 1037–1041
invasion and then multiply, producing cytotoxins and enzymes such as capsular polysaccharides, lipopolysaccharides, cytolysins, iron, hemolysins, and metalloproteases [4,5]. The toxic effects that theses toxins/enzymes exert via the bloodstream on skin or soft-tissue lesions appear to progress through three clinical stages: the inflammatory, the bullous, and the gangrenous stage [6,7]. These toxins/enzymes may trigger vigorous septic reactions and worsen the host’s condition, particularly if the person is immunocompromised, such as in the case of individuals with chronic liver disease, thus increasing the mortality risk [1,3,6,8-10]. The reported fatality rate for V vulnificus infections ranged from 10% to 54% [1-3,6,8-14]. Early diagnosis of this infection and the instant recognition of the mortality risk factors appear to be crucial. Several authors have addressed whether hemorrhagic bullous lesions or necrotizing fasciitis, lesions involving two or more extremities, septic shock, and an immunocompromised status, are associated with higher mortality [3,6,8-16]. Recently, several studies revealed that the Acute Physiology and Chronic Health Evaluation (APACHE) II score was a significant predictor of V vulnificus mortality [14,16]. The APACHE II score, initially designed to measure disease severity in patients admitted to intensive care units (ICUs), requires laboratory information obtained in a complex fashion [17,18]; its application at the time of admission does not appear to be feasible. The Rapid Emergency Medicine Score (REMS) and the Mortality in Emergency Department Sepsis (MEDS) score, which are based on parameters available at the time of admission, are often employed to assess disease severity for predicting mortality in patients with infections [19,20]. Therefore, the current study was performed to evaluate the performance of these scoring systems in V vulnificus–infected patients and to identify the predictors of mortality in this group. 2. Methods 2.1. Study subjects and setting For all V vulnificus–infected patients aged N18 years who had been admitted to an academic medical center in Tainan, Taiwan, we collected these study subjects through a computer-aided systemic search of the clinical laboratory and the registration system of the emergency department (ED) from January 1999 to June 2010. V vulnificus infections were diagnosed based on positive results from blood and/or wound cultures. In total, 171 patients were identified via ED admission as having V vulnificus infections and were included in the analysis. This study was conducted with the approval of the institutional review board of this institution. 2.2. Measurements The clinical and laboratory information (demographic and microbiological data, clinical manifestations, treatments, and outcomes) of all the patients included in the study were collected and analyzed. Broad-spectrum empirical antimicrobials were initially administered parenterally after the wound and/or blood specimens had been collected, and the treatment was subsequently modified according to the culture results, as necessary. V vulnificus–infected patients with a history of wound exposure to marine environments, and those with an injury from handling seafood during the week before arrival were regarded as having primary wound infections, while those with sepsis but without any obvious sources of infection were defined as having primary septicemia [1,3,8,11,12]. Illness severity in the first 24-hour period after arrival was retrospectively assessed based on the REMS and the MEDS scoring systems. The definitions and variables of the REMS and the MEDS scoring systems were summarized in Appendix 1 [19,20]. Sepsis, severe sepsis, and septic shock were defined according to the International Sepsis Definitions Conference criteria [21]. Criteria for ICU admission used at the center followed the recom-
mendations of the American College of Critical Care Medicine and those of the Society of Critical Care Medicine [22]. Case fatality was defined as death during hospitalization. 2.3. Statistical analysis Descriptive data are presented as numbers with percentages for categorical data and means with standard deviations for continuous data. Continuous data were tested using either the Student t test or the Mann-Whitney U test; categorical variables were compared by either the χ 2 test or Fisher exact test (when the expected value was b 5 in one cell), as appropriate. Variables including demographics, clinical features, laboratory results, and treatment modalities were compared between survivors and non-survivors. Significant variables obtained from the univariate analysis were further examined by logistic regression analysis using forward selection to identify significant risk factors for mortality. The Hosmer-Lemeshow test was used to examine the model’s goodness-of-fit. Odds ratios (ORs) and 95% confidence intervals (CIs) were also calculated in the regression model. Missing data was treated with a list-wise deletion approach in analyses [23]. These statistic analyses were conducted using the SAS version 8.2 (SAS Institute, Cary, NC). The discriminatory power of the scoring model was assessed with the ROC curve analysis generated by using the MedCalc Statistical Software, version 9.5 (Broekstraat, Mariakerke, Belgium). The discriminative power, defined as the ability of the model to discriminate between survivors and non-survivors, was assessed by calculating the area under the ROC curve (AUC) and its 95% CI. The optimal cutoff value for predicting V vulnificus mortality was identified as the score giving the best Youden’s index [maximum (sensitivity + specificity − 1)] for the scoring model [24]. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), likelihood ratio positive (LR +), and likelihood ratio negative (LR −) were calculated at the optimal cutoff value for the scoring system assessment. Two-sided P b .05 was considered statistically significant in all analyses. 3. Results 3.1. Demographics, comorbidities, clinical features, therapeutic methods, and outcomes The average age of the 171 patients was 63.1 ± 12.3 years (range, 29-89 years), and 56% were men. Among the 171 V vulnificus–infected patients, 98 patients (25 with a recent history of existing wound exposure to seawater/marine creatures and 73 with a recent injury from handling seafood/fishing) had primary wound infections and 73 patients (56 with a recent history of consuming raw/undercooked seafood and 17 without the relevant information recorded) had primary septicemia. The mean symptom duration prior to admission was 1.4 ± 0.8 days. The most common comorbidity in these patients was diabetes mellitus (30%), followed by liver diseases (29%). The mean REMS and MEDS scores on admission were 6.5 ± 3.0 and 5.3 ± 3.9, respectively. Among the 171 patients, 145 (85%) had skin or soft tissue infections. Of the 145 patents with skin/soft tissue infections, 126 had one-limb involvement and 19 had two limbs affected. Hemorrhagic bullae or necrotizing fasciitis occurred in 102 patients. Treatment options for the 171 patients included antibiotics alone (n = 50) or surgical intervention (incision and drainage, debridement, fasciotomy, and/or amputation) combined with antibiotics (n = 121). Among the 121 patients undergoing surgical intervention, in 105 (87%) these were performed within 24 hours of admission. Sixty-eight patients (40%) required intensive care. The mean duration of hospitalization was 16.8 ± 14.6 days. During hospitalization, 43 patients died, yielding an overall case-fatality rate of 25%. Of the 43 deaths, 27 (63%) occurred within 72 hours after arrival. Demographic
S.-H. Kuo et al. / American Journal of Emergency Medicine 31 (2013) 1037–1041
and clinical information of the 171 patients were summarized in Tables 1 and 2.
When the significant clinical variables obtained from the univariate analyses shown in Tables 1 and 2 were subjected to multivariate analysis, 5 variables fit the logistic regression model and four of them attained statistical significance. An increasing REMS value on admission (P b .0001), the presence of underlying liver disease (P = .002), or hemorrhagic bullous skin lesions/necrotizing fasciitis (P = .012) were associated with an increased mortality risk; the time to surgical treatment after arrival b 24 hours was associated with a decreased mortality risk (P = .007) (Table 3). 3.3. Cutoff value of REMS for predicting mortality risk The AUC estimate for the REMS scoring model was 0.895 (95% CI, 0.840-0.937; P b .0001) for predicting the mortality risk among V vulnificus–infected patients (Fig. 1). Using Youden’s index, the optimal REMS cutoff value for predicting mortality risk was 8. A significantly increased case-fatality rate was observed for REMS ≥ 8, whereas for REMS ≤5, the case-fatality rate was 3% (Fig. 2). At a REMS cutoff ≥ 8, the model had a sensitivity of 81%, a specificity of 85%, PPV of 65%, NPV of 93%, and LR+ of 5.5. After adjusting for covariates, patients with REMS ≥8 had a significantly higher mortality risk compared to those with REMS b8 (adjusted OR, 26.6; 95% CI, 15.5-79.9; P b .0001) (Table 4). Additionally, patients with REMS scores ≥ 8 points had a significantly higher frequency of ICU admission compared with those Table 1 Demographic data, underlying diseases, and clinical features in 171 V vulnificus– infected patients Variable
All patients (n = 171)
Survivors (n = 128)
Non-survivors P (n = 43)
Age (years) Gender, male, No. (%) REMS, mean ± SD (points) MEDS, mean ± SD (points) Duration of symptoms before admission (days) Type of infection, No. (%) Primary septicemia Wound infection Coexisting medical conditionsc Liver disease d Diabetes mellitus Malignancy Immunosuppressive agent/steroid used Chronic renal insufficiency Aplastic anemia No comorbid diseases Signs and symptoms a Fever/chills Skin/soft-tissue lesions involving 2 or more limbs Hemorrhagic bullous cutaneous lesions or necrotizing fasciitis Septic status (n = 171) (1) Sepsis (2) Severe sepsis (3) Septic shock
63.1 ± 12.3 95 (56) 6.5 ± 3.0a 5.3 ± 3.9 1.4 ± 0.8
62.3 ± 11.6 72 (56) 5.4 ± 2.3b 4.2 ± 3.4 1.4 ± 0.8
65.5 ± 13.9 23 (53) 9.7 ± 2.6 8.6 ± 3.5 1.2 ± 0.6
73 (43) 98 (57)
43 (34) 85 (64)
30 (70) 13 (30)
50 (29) 52 (30) 24 (14) 39 (23)
26 36 18 33
24 (56) 16 (37) 6 (14) 6 (14)
b.0001 .263 .986 .110
16 (9) 6 (4) 51 (30)
8 (6) 4 (5) 43 (34)
8 (19) 0 8 (19)
.030 .339 .063
130 (76) 19 (11)
104 (81) 5 (4)
26 (61) 14 (33)
.006 b.0001
102 (60)
69 (54)
33 (77)
.008
a
.143 .755 b.0001 b.0001 .135 b.0001
(20) (28) (14) (26)
b.0001 92 (54) 17 (10) 62 (36)
84 (66) 12 (9) 32 (25)
8 (18) 5 (12) 30 (70)
Data for peripheral oxygen saturation being available in 153 patients. Data for FS being available in 110 patients. c When patients fit into multiple categories, they were counted in each category and expressed as number of patients (%). d Hepatic disorders included chronic hepatitis B, chronic hepatitis C, alcoholic hepatitis, liver cirrhosis, or hepatocellular carcinoma. b
Table 2 Laboratory findings on admission, treatment, and outcomes in 171 patients with V vulnificus infection Variable
3.2. Multivariate analysis of risk factors for mortality
1039
a
Laboratory findings WBC count N1.2 × 104 cells/mm3 or b4 × 103 cells/mm3 Hemoglobin b14 g/dL in males or b12 g/dL in females Serum AST level N40 IU/L Serum creatinine level N1.3 mg/dL Serum albumin level b3.5 mg/dL Bacteremia Treatment method (1) Surgical interventione plus antibiotics (2) Antibiotics alone Antibiotic treatment (1) Penicillin group or first/second-generation cephalosporin with or without an aminoglycoside (2) Third-generation cephalosporin with minocycline (or analogue) or quinolone group Mechanical ventilatory support needed Hemodialysis/hemofiltration needed Vasopressor administration Time to surgical treatment after admission b24 hours Limb amputation needed ICU admission needed Hospital stay, mean ± SD (days)
All patients Survivors Non(n = 171) (n = 128) survivors (n = 43)
P
107 (63)
93 (73)
14 (33)
b.0001
84 (49)
60 (47)
24 (56)
.310
95 (56) 91 (53)
60 (47) 54 (42)
35 (81) 37 (86)
b.0001 b.0001
47 (28) 116 (69)
26 (20) b
77 (62)
b.0001
21 (49) c
39 (93)
d
.0001 b.0001
121 (71)
99 (77)
22 (51)
50 (29)
29 (23)
21 (49)
31 (18)
26 (20)
5 (12)
140 (82)
102 (80)
38 (88)
14 (8)
6 (5)
8 (19)
.008
5 (3)
1 (1)
4 (9)
.015
62 (36) 105 (87)f
32 (25) 93 (94)g
30 (70) 12 (55)h
.201
9 (5) 9 (7) 0 68 (40) 39 (31) 29 (67) 16.8 ± 14.6 20.9 ± 16.6 4.6 ± 4.5
b.0001 b.0001 .114 b.0001 b.0001
AST, aspartate aminotransferase; WBC, white blood cell. a Listed as number of patients (%), except as noted. b Blood cultures being obtained in 167 patients. c Blood cultures being obtained in 125 patients. d Blood cultures being obtained in 42 patients. e Surgical intervention: incision and drainage, debridement, fasciotomy, and/or limb amputation. f Total of 121 patients with skin/soft-tissue infection receiving surgery as denominator. g Ninety-nine patients with skin/soft-tissue infection in the group receiving surgery as denominator. h Twenty-two patients with skin/soft-tissue infection in the group receiving surgery as denominator.
with scores b 8 after adjusting for the same set of covariates (39/54 [72%] vs 29/117 [25%], adjusted OR, 12.5; 95% CI, 5.2-30.2; P b .0001). 4. Discussion In our study, we report that a greater REMS value on admission and the presence of underlying liver disease or hemorrhagic bullous
Table 3 Factors relevant to mortality in V vulnificus–infected patients using multivariate analyses (n = 171) Variable
OR (95% CI)
P
REMS on admission (points) 2.4 (1.6-3.4) b.0001 MEDS on admission (points) 1.3 (0.9-1.6) .077 Liver diseases (presence vs absence) 5.2 (2.8-14.2) .002 Hemorrhagic bullous cutaneous lesions/necrotizing 18.7 (1.9-182.6) .012 fasciitis (presence vs absence) Time to surgery b24 hours after admission (yes vs no) 0.05 (0.005-0.44) .007
1040
S.-H. Kuo et al. / American Journal of Emergency Medicine 31 (2013) 1037–1041
Fig. 1. ROC curve of the REMS for predicting mortality risk in 171 V vulnificus–infected patients. The area under the ROC curve for predicting mortality risk in V vulnificus– infected patients is 0.895 (95% CI = 0.840-0.937; P b 0.0001). The diagonal reference line (green line) indicates no discrimination.
skin lesions/necrotizing fasciitis represent significant risk factors for V vulnificus-related mortality, while prompt surgical intervention within 24 hours after admission confers significant protection against V vulnificus-related mortality. Moreover, we found that the REMS scoring system provides good discriminatory power for predicting the mortality of V vulnificus–infected patients and also identifies an optimal cutoff point of ≥8 in the prediction model. This is, to date, the first report to assess the clinical applicability of the REMS scoring model for V vulnificus–infected patients. Liu et al and other investigators have addressed the link between disease severity and V vulnificus mortality [14,16]. In these studies, disease severity was assessed by using the APACHE II score. However, the APACHE II score is a relatively complex and time-consuming tool that requires more laboratory parameters, whereas the variables of the REMS model are relatively easily available at the time of admission. The REMS scoring system, based on parameters that include mean arterial pressure, heart rate, respiratory rate, peripheral oxygen saturation, consciousness, and age, was developed by Olsson et al and designed to measure disease severity for predicting mortality in medical patients with infections on admission [19]. The practical application of the REMS for V vulnificus–infected patients has not been ascertained yet. The AUC for REMS approximated 0.9 by using the ROC curve analysis, demonstrating that this scoring system had a good
discriminatory ability for predicting mortality in V vulnificus–infected patients at time of admission. Considering the rapidly progressive nature and the potential life-threatening consequences of the infection, the REMS scoring model, having a good sensitivity and specificity (both N 80%), can help clinicians promptly and confidently stratify these patients early in the clinical course. From our results as well as from those reported previously [1-3,6,8-10,13-16], the presence of underlying liver disease and hemorrhagic bullous cutaneous lesions/necrotizing fasciitis can be regarded as significant predictors of V vulnificus mortality. The only protective factor for V vulnificus mortality in the present study was having undergone surgical treatment within 24 hours after admission, a finding in concordance with the suggestions made by Howard et al [11] and Halow et al [7] that early surgical intervention, including incision and drainage, debridement, fasciotomy, and even limb amputation (if necessary), play an essential role in treating V vulnificus–caused skin or soft tissue infections. While these factors were included as covariates in the prediction model, V vulnificus– infected patients with a REMS cutoff score ≥8 carried a 26.6-fold mortality risk and a 12.5-fold risk of ICU admission compared with those having a score b8 after multivariate adjustment. This appears to suggest that the REMS could be an effective means to assess illness severity and to perform risk stratification in these patients at the time of admission. It should be acknowledged that our results are limited by the retrospective design of our study. The uncommon occurrence of V vulnificus infections in humans makes large-scale prospective studies quite difficult to conduct in a measurable period. Because peripheral oxygen saturation is not routinely measured on ED admission in every patient, except for those in severe, critical, or hypoxemic states, the oxygen saturation component of the REMS was not available for all patients in our study. According to Levitt et al, when data for oxygen saturation were not collected, they could be assumed to have a weight of zero; this assumption has been tested and verified [25]. Even with this modification, the REMS for predicting the mortality risk herein exhibited a good discriminative power (AUC of 0.895) among V vulnificus–infected patients. This was comparable to the corresponding performance reported in previous studies that used the REMS scoring model, which found AUC values ranging between 0.735 and 0.910 [19,25,26]. Because these variables were easily available when the patients presented for admission, and before advanced laboratory investigations were performed, a combined system with clinical symptoms/signs and REMS could be valuable by providing clinicians with an immediate, feasible, and reliable adjunct tool to stratify risk in patients clinically suspected to have a V vulnificus infection. The fact that this is a single-center study may also limit the generalization of our results. Despite the majority of laboratory examinations being routinely measured on ED admission, some laboratory data were not available in some patients, and thus weakening our results. However, the number of patients in our series was comparable to the number of V vulnificus–infected patients from previously published reports with study periods of 1 to 13 years [1-3,6-16].
5. Conclusion
Fig. 2. Frequency of the REMS in relation to the case-fatality rate for the 171 V vulnificus–infected patients.
We have learned that the REMS scoring model is an effective risk stratification indicator in V vulnificus–infected patients at the time of admission. The presence of underlying liver disease and hemorrhagic bullous cutaneous lesions/necrotizing fasciitis would increase the mortality risk, but surgical intervention performed promptly, within 24 hours after admission, can be beneficial. In addition to careful history taking and physical examination, the REMS scoring model could help clinicians identify these V vulnificus–infected patients who should receive proper disposition or undergo specific therapeutic interventions, particularly when initial medical resources are limited.
S.-H. Kuo et al. / American Journal of Emergency Medicine 31 (2013) 1037–1041
1041
Table 4 Sensitivity, specificity, PPV, NPV, LR, and OR for the REMS cutoff value in predicting the mortality risk among 171 V vulnificus–infected patients REMS (points)
Fatality (n = 43) (%)
Survival (n = 128) (%)
Sen (%) (95% CI)
Spe (%) (95% CI)
PPV (%) (95% CI)
NPV (%) (95% CI)
LR+ (95% CI)
LR− (95% CI)
Adjusted ORb (95% CI)
≥5 ≥6 ≥7 ≥8a
41 41 40 35
86 60 48 19
96 95 93 81
33 53 63 85
32 41 46 65
97 (84-99) 97 (90-99) 96 (89-99) 93 (87-97)
1.4 2.0 2.5 5.5
0.1 0.1 0.1 0.2
11.1 20.7 22.3 26.6
(95) (95) (93) (81)
(67) (47) (38) (15)
(84-99) (83-99) (81-98) (67-92)
(25-42) (44-62) (54-71) (78-91)
(24-41) (31-51) (35-56) (51-77)
(1.1-1.8) (1.7-2.4) (2.1-2.9) (4.7-6.4)
(0.04-0.6) (0.02-0.3) (0.04-0.3) (0.1-0.5)
(4.2-173.8) (5.5-201.9) (8.6-172.6) (15.5-79.9)
Sen = Sensitivity; Spe = Specificity. a Optimal cut-off value estimated according to Youden’s index. b Adjustment for the presence of underlying liver diseases, hemorrhagic bullous cutaneous lesions/necrotizing fasciitis, and surgical intervention b24 hours after admission.
Appendix 1. Definitions of the REMS and the MEDS score REMS scoring system Parameter (1) Age ≥75 years 65-74 years 55-64 years 45-54 years b45 years (2) Mean arterial pressure N159 or b49 mm Hg 130-159 mm Hg 110-129 or 50-69 mm Hg 70-109 mm Hg (3) Pulse rate N179 or b40 beat/ min 140-179 or 40-54 beat/min 110-139 or 55-69 beat/min 70-109 beat/min (4) Respiratory rate N49 or b6 breath/ min 35-49 breath/min 6-9 breath/min 25-34 or 10-11 breath/min 12-24 breath/min (5) Peripheral oxygen saturationa b75% 75-85% 86-89% N89% (6) Glasgow Coma Scale score b5 5-7 8-10 11-13 N13
MEDS scoring system Points assignment 6 5 3 2 0
4 3 2 0 4 3 2 0
Variable (1) Age N65 years Yes No (2) Nursing home resident Yes No (3) Rapid terminal comorbid illness Yes No (4) Lower respiratory tract infection Yes No (5) Bands N5% on a WBC differential Yes No
Points assignment 3 0 2 0
6 0
2 0
3 0
4
(6) Tachypnea or hypoxemia Yes No
3 0
3 2 1
(7) Septic shock Yes No
3 0
0
(8) Platelet count b150 000/mm3 Yes
3
4 3 1 0
No (9) Altered mental status Yes No
0 2 0
4 3 2 1 0
WBC, white blood cell. a When data for peripheral oxygen saturation is not available, it is assumed to have a weight of zero.
References [1] Klontz KC, Lieb S, Schreiber M, et al. Syndromes of Vibrio vulnificus infections. Clinical and epidemiologic features in Florida cases, 1981–1987. Ann Intern Med 1988;109:318–23. [2] Levine WC, Griffin PM. Vibrio infections on the Gulf Coast: results of first year of regional surveillance. Gulf Coast Vibrio Working Group. J Infect Dis 1993;167:479–83. [3] Hsueh PR, Lin CY, Tang HJ, et al. Vibrio vulnificus in Taiwan. Emerg Infect Dis 2004;10:1363–8. [4] Chiang SR, Chuang YC. Vibrio vulnificus infection: clinical manifestations, pathogenesis, and antimicrobial therapy. J Microbiol Immunol Infect 2003;36:81–8. [5] Jones MK, Oliver JD. Vibrio vulnificus: disease and pathogenesis. Infect Immun 2009;77:1723–33. [6] Park SD, Shon HS, Joh NJ. Vibrio vulnificus septicemia in Korea: clinical and epidemiologic findings in seventy patients. J Am Acad Dermatol 1991;24:397–403. [7] Halow KD, Harner RC, Fontenelle LJ. Primary skin infections secondary to Vibrio vulnificus: the role of operative intervention. J Am Coll Surg 1996;183:329–34. [8] Chuang YC, Yuan CY, Liu CY, et al. Vibrio vulnificus infection in Taiwan: report of 28 cases and review of clinical manifestations and treatment. Clin Infect Dis 1992;15:271–6. [9] Shapiro RL, Altekruse S, Hutwagner L, et al. The role of Gulf Coast oysters harvested in warmer months in Vibrio vulnificus infections in the United States, 1988–1996. J Infect Dis 1998;178:752–9. [10] Inoue Y, Ono T, Matsui T, et al. Epidemiological survey of Vibrio vulnificus infection in Japan between 1999 and 2003. J Dermatol 2008;35:129–39. [11] Howard RJ, Bennett NT. Infections caused by halophilic marine Vibrio bacteria. Ann Surg 1993;217:525–30. [12] Hlady WG, Klontz KC. The epidemiology of Vibrio infections in Florida, 1981–1993. J Infect Dis 1996;173:1176–83. [13] Chung PH, Chuang SK, Tsang T, et al. Cutaneous injury and Vibrio vulnificus infection. Emerg Infect Dis 2006;12:1302–3. [14] Liu JW, Lee IK, Tang HJ, et al. Prognostic factors and antibiotics in Vibrio vulnificus septicemia. Arch Intern Med 2006;166:2117–23. [15] Kuo Chou TN, Chao WN, Yang C, et al. Predictors of mortality in skin and soft-tissue infections caused by Vibrio vulnificus. World J Surg 2010;34:1669–75. [16] Chen SC, Chan KS, Chao WN, et al. Clinical outcomes and prognostic factors for patients with Vibrio vulnificus infections requiring intensive care: a 10-yr retrospective study. Crit Care Med 2010;38:1984–90. [17] Knaus WA, Draper EA, Wagner DP, et al. APACHE II: A severity of disease classification system. Crit Care Med 1985;13:818–29. [18] Wong DT, Crofts SL, Gomez M, et al. Evaluation of predictive ability of APACHE II system and hospital outcome in Canadian intensive care unit patients. Crit Care Med 1995;23:1177–83. [19] Olsson T, Terent A, Lind L. Rapid Emergency Medicine score: a new prognostic tool for in-hospital mortality in nonsurgical emergency department patients. J Intern Med 2004;255:579–87. [20] Shapiro NI, Wolfe RE, Moore RB, et al. Mortality in Emergency Department Sepsis (MEDS) score: a prospectively derived and validated clinical prediction rule. Crit Care Med 2003;31:670–5. [21] Levy MM, Fink MP, Marshall JC, et al. 2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference. Crit Care Med 2003;31:1250–6. [22] Egol A, Fromm R, Guntupalli KK, et al. Guidelines for intensive care unit admission, discharge, and triage Task Force of the American College of Critical Care Medicine, Society of Critical Care Medicine. Crit Care Med 1999;27:633–8. [23] Allison PD. Missing Data. In: Lewis-Beck MS, editor. Sage University Papers Series on Quantitative Applications in the Social Sciences, 07–136. Thousand Oaks, CA: Sage; 2001. p. 6–7. [24] Youden WJ. An index for rating diagnostic tests. Cancer 1950;3:32–5. [25] Levitt JE, Bedi H, Calfee CS, et al. Identification of early acute lung injury at initial evaluation in an acute care setting prior to the onset of respiratory failure. Chest 2009;135:936–43. [26] Howell MD, Donnino MW, Talmor D, et al. Performance of severity of illness scoring systems in emergency department patients with infection. Acad Emerg Med 2007;14:709–14.
Contents lists available at SciVerse ScienceDirect
American Journal of Emergency Medicine journal homepage: www.elsevier.com/locate/ajem
Original Contribution
Rapid Emergency Medicine Score as a main predictor of mortality in Vibrio vulnificus–related patients☆,☆☆ Sheng-Hung Kuo MD a, b, Chin-Feng Tsai MD, PhD a, c, d, Chi-Rong Li PhD a, e, Shih-Jei Tsai MD, PhD a, c, d, Wai-Nang Chao MD a, f, g, Khee-Siang Chan MD, PhD a, f, h, Yuan-Ti Lee MD, PhD a, c, d, 1, Ruey-Hong Wong PhD a, i, Chun-Chieh Chen MD, PhD a, c, j, 2, Shiuan-Chih Chen MD, PhD a, c, j,⁎ a
Institute of Medicine, Chung Shan Medical University, Taichung, Taiwan Department of Surgery, Yuan-Lin Kuo Hospital, Changhua, Taiwan c School of Medicine, Chung Shan Medical University, Taichung, Taiwan d Department of Internal Medicine, Chung Shan Medical University Hospital, Taichung, Taiwan e School of Nursing and Institute of Nursing, Chung Shan Medical University, Taichung, Taiwan f Department of Emergency Medicine, Chi Mei Medical Center, Tainan, Taiwan g Department of Surgery, Chi Mei Medical Center, Tainan, Taiwan h Department of Critical Care Medicine, Chi Mei Medical Center, Tainan, Taiwan i School of Public Health, Chung Shan Medical University, Taichung, Taiwan j Department of Family and Community Medicine, Chung Shan Medical University Hospital, Taichung, Taiwan b
a r t i c l e
i n f o
Article history: Received 23 October 2012 Received in revised form 6 February 2013 Accepted 16 March 2013
a b s t r a c t Objectives: Vibrio vulnificus causes potentially life-threatening and rapidly progressing infections. Therefore, the severity-of-illness assessment appears to be important for V vulnificus–infected patients at the time of admission. The aim of our study was to evaluate the performance of the severity-of-illness scoring model on admission in V vulnificus–infected patients. Methods: One hundred seventy-one consecutive patients (mean age: 63.1 ± 12.3 years) with V vulnificus infection who were admitted to a teaching hospital between January 1999 and June 2010 were included in the study. Demographic and clinical characteristics, illness severity on admission, treatment, and outcomes were collected for each patient and extracted for analysis. Logistic regression and receiver operating characteristic curve analyses were performed. Results: The mean Rapid Emergency Medicine Score (REMS) on admission was 6.5 ± 3.0 points. During hospitalization, 68 patients (40%) required intensive care. The overall case-fatality rate was 25%. In multivariate analysis, the presence of underlying liver disease (P = .002), hemorrhagic bullous lesions/ necrotizing fasciitis (P = .012), and higher REMS values on admission (P b .0001) were associated with increased mortality risk; a time span b24 hours between arrival and surgical treatment was associated with a decreased mortality risk (P = .007). Additionally, the area under the receiver operating characteristic (ROC) curve for the REMS in predicting mortality risk was 0.895 (P b .0001). An optimal cut-off REMS ≥ 8 had a sensitivity of 81% and a specificity of 85%, with a 26.6-fold mortality risk (P b .0001) and a 12.5-fold intensive care unit admission risk (P b .0001). Conclusion: The REMS could provide clinicians with an effective adjunct risk stratification tool for V vulnificus– infected patients. © 2013 Elsevier Inc. All rights reserved.
1. Introduction
☆ Funding/Support: None. ☆☆ Competing Interests Declared: None. ⁎ Corresponding author. Faculty of Medicine, School of Medicine and Institute of Medicine, Chung Shan Medical University, Taichung 40201, Taiwan. Tel.: +886 4 24739595x34970; fax: +886 4 23248137. E-mail address: [email protected] (S.-C. Chen). 1 Dr. Yuan-Ti Lee made an equal contribution to this work as the second author. 2 Dr. Chun-Chieh Chen made an equal contribution to this work as the first author. 0735-6757/$ – see front matter © 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ajem.2013.03.030
Vibrio vulnificus is a virulent, Gram-negative, halophilic, rodshaped, motile bacterium that often flourishes in warm estuarine seawaters or brackish environments and is estimated to cause 0.2 to 36 infections per million individuals annually [1-3]. V vulnificus infections can progress rapidly, and without prompt and appropriate management, the host’s condition may sharply deteriorate in 1 to 2 days. Human and animal studies have shown that V vulnificus may enter the host through skin wounds or gastrointestinal mucosal
1038
S.-H. Kuo et al. / American Journal of Emergency Medicine 31 (2013) 1037–1041
invasion and then multiply, producing cytotoxins and enzymes such as capsular polysaccharides, lipopolysaccharides, cytolysins, iron, hemolysins, and metalloproteases [4,5]. The toxic effects that theses toxins/enzymes exert via the bloodstream on skin or soft-tissue lesions appear to progress through three clinical stages: the inflammatory, the bullous, and the gangrenous stage [6,7]. These toxins/enzymes may trigger vigorous septic reactions and worsen the host’s condition, particularly if the person is immunocompromised, such as in the case of individuals with chronic liver disease, thus increasing the mortality risk [1,3,6,8-10]. The reported fatality rate for V vulnificus infections ranged from 10% to 54% [1-3,6,8-14]. Early diagnosis of this infection and the instant recognition of the mortality risk factors appear to be crucial. Several authors have addressed whether hemorrhagic bullous lesions or necrotizing fasciitis, lesions involving two or more extremities, septic shock, and an immunocompromised status, are associated with higher mortality [3,6,8-16]. Recently, several studies revealed that the Acute Physiology and Chronic Health Evaluation (APACHE) II score was a significant predictor of V vulnificus mortality [14,16]. The APACHE II score, initially designed to measure disease severity in patients admitted to intensive care units (ICUs), requires laboratory information obtained in a complex fashion [17,18]; its application at the time of admission does not appear to be feasible. The Rapid Emergency Medicine Score (REMS) and the Mortality in Emergency Department Sepsis (MEDS) score, which are based on parameters available at the time of admission, are often employed to assess disease severity for predicting mortality in patients with infections [19,20]. Therefore, the current study was performed to evaluate the performance of these scoring systems in V vulnificus–infected patients and to identify the predictors of mortality in this group. 2. Methods 2.1. Study subjects and setting For all V vulnificus–infected patients aged N18 years who had been admitted to an academic medical center in Tainan, Taiwan, we collected these study subjects through a computer-aided systemic search of the clinical laboratory and the registration system of the emergency department (ED) from January 1999 to June 2010. V vulnificus infections were diagnosed based on positive results from blood and/or wound cultures. In total, 171 patients were identified via ED admission as having V vulnificus infections and were included in the analysis. This study was conducted with the approval of the institutional review board of this institution. 2.2. Measurements The clinical and laboratory information (demographic and microbiological data, clinical manifestations, treatments, and outcomes) of all the patients included in the study were collected and analyzed. Broad-spectrum empirical antimicrobials were initially administered parenterally after the wound and/or blood specimens had been collected, and the treatment was subsequently modified according to the culture results, as necessary. V vulnificus–infected patients with a history of wound exposure to marine environments, and those with an injury from handling seafood during the week before arrival were regarded as having primary wound infections, while those with sepsis but without any obvious sources of infection were defined as having primary septicemia [1,3,8,11,12]. Illness severity in the first 24-hour period after arrival was retrospectively assessed based on the REMS and the MEDS scoring systems. The definitions and variables of the REMS and the MEDS scoring systems were summarized in Appendix 1 [19,20]. Sepsis, severe sepsis, and septic shock were defined according to the International Sepsis Definitions Conference criteria [21]. Criteria for ICU admission used at the center followed the recom-
mendations of the American College of Critical Care Medicine and those of the Society of Critical Care Medicine [22]. Case fatality was defined as death during hospitalization. 2.3. Statistical analysis Descriptive data are presented as numbers with percentages for categorical data and means with standard deviations for continuous data. Continuous data were tested using either the Student t test or the Mann-Whitney U test; categorical variables were compared by either the χ 2 test or Fisher exact test (when the expected value was b 5 in one cell), as appropriate. Variables including demographics, clinical features, laboratory results, and treatment modalities were compared between survivors and non-survivors. Significant variables obtained from the univariate analysis were further examined by logistic regression analysis using forward selection to identify significant risk factors for mortality. The Hosmer-Lemeshow test was used to examine the model’s goodness-of-fit. Odds ratios (ORs) and 95% confidence intervals (CIs) were also calculated in the regression model. Missing data was treated with a list-wise deletion approach in analyses [23]. These statistic analyses were conducted using the SAS version 8.2 (SAS Institute, Cary, NC). The discriminatory power of the scoring model was assessed with the ROC curve analysis generated by using the MedCalc Statistical Software, version 9.5 (Broekstraat, Mariakerke, Belgium). The discriminative power, defined as the ability of the model to discriminate between survivors and non-survivors, was assessed by calculating the area under the ROC curve (AUC) and its 95% CI. The optimal cutoff value for predicting V vulnificus mortality was identified as the score giving the best Youden’s index [maximum (sensitivity + specificity − 1)] for the scoring model [24]. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), likelihood ratio positive (LR +), and likelihood ratio negative (LR −) were calculated at the optimal cutoff value for the scoring system assessment. Two-sided P b .05 was considered statistically significant in all analyses. 3. Results 3.1. Demographics, comorbidities, clinical features, therapeutic methods, and outcomes The average age of the 171 patients was 63.1 ± 12.3 years (range, 29-89 years), and 56% were men. Among the 171 V vulnificus–infected patients, 98 patients (25 with a recent history of existing wound exposure to seawater/marine creatures and 73 with a recent injury from handling seafood/fishing) had primary wound infections and 73 patients (56 with a recent history of consuming raw/undercooked seafood and 17 without the relevant information recorded) had primary septicemia. The mean symptom duration prior to admission was 1.4 ± 0.8 days. The most common comorbidity in these patients was diabetes mellitus (30%), followed by liver diseases (29%). The mean REMS and MEDS scores on admission were 6.5 ± 3.0 and 5.3 ± 3.9, respectively. Among the 171 patients, 145 (85%) had skin or soft tissue infections. Of the 145 patents with skin/soft tissue infections, 126 had one-limb involvement and 19 had two limbs affected. Hemorrhagic bullae or necrotizing fasciitis occurred in 102 patients. Treatment options for the 171 patients included antibiotics alone (n = 50) or surgical intervention (incision and drainage, debridement, fasciotomy, and/or amputation) combined with antibiotics (n = 121). Among the 121 patients undergoing surgical intervention, in 105 (87%) these were performed within 24 hours of admission. Sixty-eight patients (40%) required intensive care. The mean duration of hospitalization was 16.8 ± 14.6 days. During hospitalization, 43 patients died, yielding an overall case-fatality rate of 25%. Of the 43 deaths, 27 (63%) occurred within 72 hours after arrival. Demographic
S.-H. Kuo et al. / American Journal of Emergency Medicine 31 (2013) 1037–1041
and clinical information of the 171 patients were summarized in Tables 1 and 2.
When the significant clinical variables obtained from the univariate analyses shown in Tables 1 and 2 were subjected to multivariate analysis, 5 variables fit the logistic regression model and four of them attained statistical significance. An increasing REMS value on admission (P b .0001), the presence of underlying liver disease (P = .002), or hemorrhagic bullous skin lesions/necrotizing fasciitis (P = .012) were associated with an increased mortality risk; the time to surgical treatment after arrival b 24 hours was associated with a decreased mortality risk (P = .007) (Table 3). 3.3. Cutoff value of REMS for predicting mortality risk The AUC estimate for the REMS scoring model was 0.895 (95% CI, 0.840-0.937; P b .0001) for predicting the mortality risk among V vulnificus–infected patients (Fig. 1). Using Youden’s index, the optimal REMS cutoff value for predicting mortality risk was 8. A significantly increased case-fatality rate was observed for REMS ≥ 8, whereas for REMS ≤5, the case-fatality rate was 3% (Fig. 2). At a REMS cutoff ≥ 8, the model had a sensitivity of 81%, a specificity of 85%, PPV of 65%, NPV of 93%, and LR+ of 5.5. After adjusting for covariates, patients with REMS ≥8 had a significantly higher mortality risk compared to those with REMS b8 (adjusted OR, 26.6; 95% CI, 15.5-79.9; P b .0001) (Table 4). Additionally, patients with REMS scores ≥ 8 points had a significantly higher frequency of ICU admission compared with those Table 1 Demographic data, underlying diseases, and clinical features in 171 V vulnificus– infected patients Variable
All patients (n = 171)
Survivors (n = 128)
Non-survivors P (n = 43)
Age (years) Gender, male, No. (%) REMS, mean ± SD (points) MEDS, mean ± SD (points) Duration of symptoms before admission (days) Type of infection, No. (%) Primary septicemia Wound infection Coexisting medical conditionsc Liver disease d Diabetes mellitus Malignancy Immunosuppressive agent/steroid used Chronic renal insufficiency Aplastic anemia No comorbid diseases Signs and symptoms a Fever/chills Skin/soft-tissue lesions involving 2 or more limbs Hemorrhagic bullous cutaneous lesions or necrotizing fasciitis Septic status (n = 171) (1) Sepsis (2) Severe sepsis (3) Septic shock
63.1 ± 12.3 95 (56) 6.5 ± 3.0a 5.3 ± 3.9 1.4 ± 0.8
62.3 ± 11.6 72 (56) 5.4 ± 2.3b 4.2 ± 3.4 1.4 ± 0.8
65.5 ± 13.9 23 (53) 9.7 ± 2.6 8.6 ± 3.5 1.2 ± 0.6
73 (43) 98 (57)
43 (34) 85 (64)
30 (70) 13 (30)
50 (29) 52 (30) 24 (14) 39 (23)
26 36 18 33
24 (56) 16 (37) 6 (14) 6 (14)
b.0001 .263 .986 .110
16 (9) 6 (4) 51 (30)
8 (6) 4 (5) 43 (34)
8 (19) 0 8 (19)
.030 .339 .063
130 (76) 19 (11)
104 (81) 5 (4)
26 (61) 14 (33)
.006 b.0001
102 (60)
69 (54)
33 (77)
.008
a
.143 .755 b.0001 b.0001 .135 b.0001
(20) (28) (14) (26)
b.0001 92 (54) 17 (10) 62 (36)
84 (66) 12 (9) 32 (25)
8 (18) 5 (12) 30 (70)
Data for peripheral oxygen saturation being available in 153 patients. Data for FS being available in 110 patients. c When patients fit into multiple categories, they were counted in each category and expressed as number of patients (%). d Hepatic disorders included chronic hepatitis B, chronic hepatitis C, alcoholic hepatitis, liver cirrhosis, or hepatocellular carcinoma. b
Table 2 Laboratory findings on admission, treatment, and outcomes in 171 patients with V vulnificus infection Variable
3.2. Multivariate analysis of risk factors for mortality
1039
a
Laboratory findings WBC count N1.2 × 104 cells/mm3 or b4 × 103 cells/mm3 Hemoglobin b14 g/dL in males or b12 g/dL in females Serum AST level N40 IU/L Serum creatinine level N1.3 mg/dL Serum albumin level b3.5 mg/dL Bacteremia Treatment method (1) Surgical interventione plus antibiotics (2) Antibiotics alone Antibiotic treatment (1) Penicillin group or first/second-generation cephalosporin with or without an aminoglycoside (2) Third-generation cephalosporin with minocycline (or analogue) or quinolone group Mechanical ventilatory support needed Hemodialysis/hemofiltration needed Vasopressor administration Time to surgical treatment after admission b24 hours Limb amputation needed ICU admission needed Hospital stay, mean ± SD (days)
All patients Survivors Non(n = 171) (n = 128) survivors (n = 43)
P
107 (63)
93 (73)
14 (33)
b.0001
84 (49)
60 (47)
24 (56)
.310
95 (56) 91 (53)
60 (47) 54 (42)
35 (81) 37 (86)
b.0001 b.0001
47 (28) 116 (69)
26 (20) b
77 (62)
b.0001
21 (49) c
39 (93)
d
.0001 b.0001
121 (71)
99 (77)
22 (51)
50 (29)
29 (23)
21 (49)
31 (18)
26 (20)
5 (12)
140 (82)
102 (80)
38 (88)
14 (8)
6 (5)
8 (19)
.008
5 (3)
1 (1)
4 (9)
.015
62 (36) 105 (87)f
32 (25) 93 (94)g
30 (70) 12 (55)h
.201
9 (5) 9 (7) 0 68 (40) 39 (31) 29 (67) 16.8 ± 14.6 20.9 ± 16.6 4.6 ± 4.5
b.0001 b.0001 .114 b.0001 b.0001
AST, aspartate aminotransferase; WBC, white blood cell. a Listed as number of patients (%), except as noted. b Blood cultures being obtained in 167 patients. c Blood cultures being obtained in 125 patients. d Blood cultures being obtained in 42 patients. e Surgical intervention: incision and drainage, debridement, fasciotomy, and/or limb amputation. f Total of 121 patients with skin/soft-tissue infection receiving surgery as denominator. g Ninety-nine patients with skin/soft-tissue infection in the group receiving surgery as denominator. h Twenty-two patients with skin/soft-tissue infection in the group receiving surgery as denominator.
with scores b 8 after adjusting for the same set of covariates (39/54 [72%] vs 29/117 [25%], adjusted OR, 12.5; 95% CI, 5.2-30.2; P b .0001). 4. Discussion In our study, we report that a greater REMS value on admission and the presence of underlying liver disease or hemorrhagic bullous
Table 3 Factors relevant to mortality in V vulnificus–infected patients using multivariate analyses (n = 171) Variable
OR (95% CI)
P
REMS on admission (points) 2.4 (1.6-3.4) b.0001 MEDS on admission (points) 1.3 (0.9-1.6) .077 Liver diseases (presence vs absence) 5.2 (2.8-14.2) .002 Hemorrhagic bullous cutaneous lesions/necrotizing 18.7 (1.9-182.6) .012 fasciitis (presence vs absence) Time to surgery b24 hours after admission (yes vs no) 0.05 (0.005-0.44) .007
1040
S.-H. Kuo et al. / American Journal of Emergency Medicine 31 (2013) 1037–1041
Fig. 1. ROC curve of the REMS for predicting mortality risk in 171 V vulnificus–infected patients. The area under the ROC curve for predicting mortality risk in V vulnificus– infected patients is 0.895 (95% CI = 0.840-0.937; P b 0.0001). The diagonal reference line (green line) indicates no discrimination.
skin lesions/necrotizing fasciitis represent significant risk factors for V vulnificus-related mortality, while prompt surgical intervention within 24 hours after admission confers significant protection against V vulnificus-related mortality. Moreover, we found that the REMS scoring system provides good discriminatory power for predicting the mortality of V vulnificus–infected patients and also identifies an optimal cutoff point of ≥8 in the prediction model. This is, to date, the first report to assess the clinical applicability of the REMS scoring model for V vulnificus–infected patients. Liu et al and other investigators have addressed the link between disease severity and V vulnificus mortality [14,16]. In these studies, disease severity was assessed by using the APACHE II score. However, the APACHE II score is a relatively complex and time-consuming tool that requires more laboratory parameters, whereas the variables of the REMS model are relatively easily available at the time of admission. The REMS scoring system, based on parameters that include mean arterial pressure, heart rate, respiratory rate, peripheral oxygen saturation, consciousness, and age, was developed by Olsson et al and designed to measure disease severity for predicting mortality in medical patients with infections on admission [19]. The practical application of the REMS for V vulnificus–infected patients has not been ascertained yet. The AUC for REMS approximated 0.9 by using the ROC curve analysis, demonstrating that this scoring system had a good
discriminatory ability for predicting mortality in V vulnificus–infected patients at time of admission. Considering the rapidly progressive nature and the potential life-threatening consequences of the infection, the REMS scoring model, having a good sensitivity and specificity (both N 80%), can help clinicians promptly and confidently stratify these patients early in the clinical course. From our results as well as from those reported previously [1-3,6,8-10,13-16], the presence of underlying liver disease and hemorrhagic bullous cutaneous lesions/necrotizing fasciitis can be regarded as significant predictors of V vulnificus mortality. The only protective factor for V vulnificus mortality in the present study was having undergone surgical treatment within 24 hours after admission, a finding in concordance with the suggestions made by Howard et al [11] and Halow et al [7] that early surgical intervention, including incision and drainage, debridement, fasciotomy, and even limb amputation (if necessary), play an essential role in treating V vulnificus–caused skin or soft tissue infections. While these factors were included as covariates in the prediction model, V vulnificus– infected patients with a REMS cutoff score ≥8 carried a 26.6-fold mortality risk and a 12.5-fold risk of ICU admission compared with those having a score b8 after multivariate adjustment. This appears to suggest that the REMS could be an effective means to assess illness severity and to perform risk stratification in these patients at the time of admission. It should be acknowledged that our results are limited by the retrospective design of our study. The uncommon occurrence of V vulnificus infections in humans makes large-scale prospective studies quite difficult to conduct in a measurable period. Because peripheral oxygen saturation is not routinely measured on ED admission in every patient, except for those in severe, critical, or hypoxemic states, the oxygen saturation component of the REMS was not available for all patients in our study. According to Levitt et al, when data for oxygen saturation were not collected, they could be assumed to have a weight of zero; this assumption has been tested and verified [25]. Even with this modification, the REMS for predicting the mortality risk herein exhibited a good discriminative power (AUC of 0.895) among V vulnificus–infected patients. This was comparable to the corresponding performance reported in previous studies that used the REMS scoring model, which found AUC values ranging between 0.735 and 0.910 [19,25,26]. Because these variables were easily available when the patients presented for admission, and before advanced laboratory investigations were performed, a combined system with clinical symptoms/signs and REMS could be valuable by providing clinicians with an immediate, feasible, and reliable adjunct tool to stratify risk in patients clinically suspected to have a V vulnificus infection. The fact that this is a single-center study may also limit the generalization of our results. Despite the majority of laboratory examinations being routinely measured on ED admission, some laboratory data were not available in some patients, and thus weakening our results. However, the number of patients in our series was comparable to the number of V vulnificus–infected patients from previously published reports with study periods of 1 to 13 years [1-3,6-16].
5. Conclusion
Fig. 2. Frequency of the REMS in relation to the case-fatality rate for the 171 V vulnificus–infected patients.
We have learned that the REMS scoring model is an effective risk stratification indicator in V vulnificus–infected patients at the time of admission. The presence of underlying liver disease and hemorrhagic bullous cutaneous lesions/necrotizing fasciitis would increase the mortality risk, but surgical intervention performed promptly, within 24 hours after admission, can be beneficial. In addition to careful history taking and physical examination, the REMS scoring model could help clinicians identify these V vulnificus–infected patients who should receive proper disposition or undergo specific therapeutic interventions, particularly when initial medical resources are limited.
S.-H. Kuo et al. / American Journal of Emergency Medicine 31 (2013) 1037–1041
1041
Table 4 Sensitivity, specificity, PPV, NPV, LR, and OR for the REMS cutoff value in predicting the mortality risk among 171 V vulnificus–infected patients REMS (points)
Fatality (n = 43) (%)
Survival (n = 128) (%)
Sen (%) (95% CI)
Spe (%) (95% CI)
PPV (%) (95% CI)
NPV (%) (95% CI)
LR+ (95% CI)
LR− (95% CI)
Adjusted ORb (95% CI)
≥5 ≥6 ≥7 ≥8a
41 41 40 35
86 60 48 19
96 95 93 81
33 53 63 85
32 41 46 65
97 (84-99) 97 (90-99) 96 (89-99) 93 (87-97)
1.4 2.0 2.5 5.5
0.1 0.1 0.1 0.2
11.1 20.7 22.3 26.6
(95) (95) (93) (81)
(67) (47) (38) (15)
(84-99) (83-99) (81-98) (67-92)
(25-42) (44-62) (54-71) (78-91)
(24-41) (31-51) (35-56) (51-77)
(1.1-1.8) (1.7-2.4) (2.1-2.9) (4.7-6.4)
(0.04-0.6) (0.02-0.3) (0.04-0.3) (0.1-0.5)
(4.2-173.8) (5.5-201.9) (8.6-172.6) (15.5-79.9)
Sen = Sensitivity; Spe = Specificity. a Optimal cut-off value estimated according to Youden’s index. b Adjustment for the presence of underlying liver diseases, hemorrhagic bullous cutaneous lesions/necrotizing fasciitis, and surgical intervention b24 hours after admission.
Appendix 1. Definitions of the REMS and the MEDS score REMS scoring system Parameter (1) Age ≥75 years 65-74 years 55-64 years 45-54 years b45 years (2) Mean arterial pressure N159 or b49 mm Hg 130-159 mm Hg 110-129 or 50-69 mm Hg 70-109 mm Hg (3) Pulse rate N179 or b40 beat/ min 140-179 or 40-54 beat/min 110-139 or 55-69 beat/min 70-109 beat/min (4) Respiratory rate N49 or b6 breath/ min 35-49 breath/min 6-9 breath/min 25-34 or 10-11 breath/min 12-24 breath/min (5) Peripheral oxygen saturationa b75% 75-85% 86-89% N89% (6) Glasgow Coma Scale score b5 5-7 8-10 11-13 N13
MEDS scoring system Points assignment 6 5 3 2 0
4 3 2 0 4 3 2 0
Variable (1) Age N65 years Yes No (2) Nursing home resident Yes No (3) Rapid terminal comorbid illness Yes No (4) Lower respiratory tract infection Yes No (5) Bands N5% on a WBC differential Yes No
Points assignment 3 0 2 0
6 0
2 0
3 0
4
(6) Tachypnea or hypoxemia Yes No
3 0
3 2 1
(7) Septic shock Yes No
3 0
0
(8) Platelet count b150 000/mm3 Yes
3
4 3 1 0
No (9) Altered mental status Yes No
0 2 0
4 3 2 1 0
WBC, white blood cell. a When data for peripheral oxygen saturation is not available, it is assumed to have a weight of zero.
References [1] Klontz KC, Lieb S, Schreiber M, et al. Syndromes of Vibrio vulnificus infections. Clinical and epidemiologic features in Florida cases, 1981–1987. Ann Intern Med 1988;109:318–23. [2] Levine WC, Griffin PM. Vibrio infections on the Gulf Coast: results of first year of regional surveillance. Gulf Coast Vibrio Working Group. J Infect Dis 1993;167:479–83. [3] Hsueh PR, Lin CY, Tang HJ, et al. Vibrio vulnificus in Taiwan. Emerg Infect Dis 2004;10:1363–8. [4] Chiang SR, Chuang YC. Vibrio vulnificus infection: clinical manifestations, pathogenesis, and antimicrobial therapy. J Microbiol Immunol Infect 2003;36:81–8. [5] Jones MK, Oliver JD. Vibrio vulnificus: disease and pathogenesis. Infect Immun 2009;77:1723–33. [6] Park SD, Shon HS, Joh NJ. Vibrio vulnificus septicemia in Korea: clinical and epidemiologic findings in seventy patients. J Am Acad Dermatol 1991;24:397–403. [7] Halow KD, Harner RC, Fontenelle LJ. Primary skin infections secondary to Vibrio vulnificus: the role of operative intervention. J Am Coll Surg 1996;183:329–34. [8] Chuang YC, Yuan CY, Liu CY, et al. Vibrio vulnificus infection in Taiwan: report of 28 cases and review of clinical manifestations and treatment. Clin Infect Dis 1992;15:271–6. [9] Shapiro RL, Altekruse S, Hutwagner L, et al. The role of Gulf Coast oysters harvested in warmer months in Vibrio vulnificus infections in the United States, 1988–1996. J Infect Dis 1998;178:752–9. [10] Inoue Y, Ono T, Matsui T, et al. Epidemiological survey of Vibrio vulnificus infection in Japan between 1999 and 2003. J Dermatol 2008;35:129–39. [11] Howard RJ, Bennett NT. Infections caused by halophilic marine Vibrio bacteria. Ann Surg 1993;217:525–30. [12] Hlady WG, Klontz KC. The epidemiology of Vibrio infections in Florida, 1981–1993. J Infect Dis 1996;173:1176–83. [13] Chung PH, Chuang SK, Tsang T, et al. Cutaneous injury and Vibrio vulnificus infection. Emerg Infect Dis 2006;12:1302–3. [14] Liu JW, Lee IK, Tang HJ, et al. Prognostic factors and antibiotics in Vibrio vulnificus septicemia. Arch Intern Med 2006;166:2117–23. [15] Kuo Chou TN, Chao WN, Yang C, et al. Predictors of mortality in skin and soft-tissue infections caused by Vibrio vulnificus. World J Surg 2010;34:1669–75. [16] Chen SC, Chan KS, Chao WN, et al. Clinical outcomes and prognostic factors for patients with Vibrio vulnificus infections requiring intensive care: a 10-yr retrospective study. Crit Care Med 2010;38:1984–90. [17] Knaus WA, Draper EA, Wagner DP, et al. APACHE II: A severity of disease classification system. Crit Care Med 1985;13:818–29. [18] Wong DT, Crofts SL, Gomez M, et al. Evaluation of predictive ability of APACHE II system and hospital outcome in Canadian intensive care unit patients. Crit Care Med 1995;23:1177–83. [19] Olsson T, Terent A, Lind L. Rapid Emergency Medicine score: a new prognostic tool for in-hospital mortality in nonsurgical emergency department patients. J Intern Med 2004;255:579–87. [20] Shapiro NI, Wolfe RE, Moore RB, et al. Mortality in Emergency Department Sepsis (MEDS) score: a prospectively derived and validated clinical prediction rule. Crit Care Med 2003;31:670–5. [21] Levy MM, Fink MP, Marshall JC, et al. 2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference. Crit Care Med 2003;31:1250–6. [22] Egol A, Fromm R, Guntupalli KK, et al. Guidelines for intensive care unit admission, discharge, and triage Task Force of the American College of Critical Care Medicine, Society of Critical Care Medicine. Crit Care Med 1999;27:633–8. [23] Allison PD. Missing Data. In: Lewis-Beck MS, editor. Sage University Papers Series on Quantitative Applications in the Social Sciences, 07–136. Thousand Oaks, CA: Sage; 2001. p. 6–7. [24] Youden WJ. An index for rating diagnostic tests. Cancer 1950;3:32–5. [25] Levitt JE, Bedi H, Calfee CS, et al. Identification of early acute lung injury at initial evaluation in an acute care setting prior to the onset of respiratory failure. Chest 2009;135:936–43. [26] Howell MD, Donnino MW, Talmor D, et al. Performance of severity of illness scoring systems in emergency department patients with infection. Acad Emerg Med 2007;14:709–14.