American Journal of Emergency Medicine xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
American Journal of Emergency Medicine journal homepage: www.elsevier.com/locate/ajem
Original Contribution
Initial serum lactate level is associated with inpatient mortality in patients with community-acquired pneumonia☆,☆☆,★ Min Hyuk Gwak, MD a, Sion Jo, MD a,⁎, Taeoh Jeong, MD a, Jae Baek Lee, MD, PhD a, Young Ho Jin, MD, PhD a, Jaechol Yoon, MD a, Boyoung Park, MD b a b
Department of Emergency Medicine, Research Institute of Clinical Medicine of Chonbuk National University and Chonbuk National University Hospital, Jeonju-si, Korea National Cancer Control Institute, National Cancer Center, Goyang-si, Kyunggi-do, Korea
a r t i c l e
i n f o
Article history: Received 30 January 2015 Received in revised form 2 March 2015 Accepted 3 March 2015 Available online xxxx
a b s t r a c t Study objective: The aim of the present study was to investigate the prognostic value of the initial serum lactate level in patients with community-acquired pneumonia (CAP). Methods: We collected data on hospitalized adult patients with CAP via the study hospital emergency department between October 2012 and September 2013. Demographics, comorbidities, and physiologic and laboratory variables including initial C-reactive protein (CRP) and serum lactate level were extracted from the electronic medical record. The primary outcome was inpatient mortality. Comparisons between survivors and nonsurvivors were performed, and multivariable logistic regression analyses were constructed as dependent variables of both continuous and categorical varieties. Results: A total of 397 patients were enrolled, and the mortality cases were 46 (11.6%). The mean lactate level was 1.7 ± 1.4 mmol/L and was significantly higher in the nonsurvivor group than in the survivor group (2.4 ± 2.2 mmol/L vs 1.6 ± 1.2 mmol/L). In the multivariable logistic regression model for inpatient mortality constructed using lactate, CRP, and laboratory variables of pneumonia severity index (PSI), lactate and CRP remained as significant factors, but laboratory variables of PSI were not. In other multivariable logistic regression models for the outcome constructed using collected laboratory variables and PSI, lactate remained as a significant factor (adjusted odds ratio, 1.24; 95% confidence interval, 1.01-1.53; P = .042 by continuous variable; adjusted odds ratio of third tertile, 2.60; 95% confidence interval, 1.02-6.66; P = .046 by category variable). C-reactive protein and albumin were also left as significant factors. Conclusions: The initial serum lactate level is independently associated with mortality in hospitalized patients with CAP. However, laboratory variables of PSI or others were not, except CRP and albumin. © 2015 Elsevier Inc. All rights reserved.
1. Introduction 1.1. Background Community-acquired pneumonia (CAP) is a common infectious disease and causes death in a substantial number of patients [1]. Because of the wide spectrum of outcomes, from complete recovery to death, effort to identify risk factors for death in patients with CAP is still in progress. Demographics, comorbidities, physiologic values, and
☆ Conflict of interests: None. ☆☆ Funding and support: None. ★ Author contributions: M.H.G. and S.J. designed this study. M.H.G. supervised the overall data collection process, had full access to all of the data in the study, and takes responsibility for the integrity of the data. B.P. conducted the data analysis. M.H.G. and S.J. wrote the initial draft of the article. All authors provided substantial review and feedback on the final version of the article. S.J. takes responsibility for the article as a whole. All authors have read and approved the submitted manuscript. This article has neither been submitted nor published elsewhere in whole or in part, except as an abstract (if relevant). ⁎ Corresponding author.
some laboratory results have shown an association with mortality in patients with CAP, and mortality prediction models have been designed, including the pneumonia severity index (PSI) and CURB-65 [2-4]. After introducing the PSI and CURB-65, later research documented the association between mortality and some biomarkers, including C-reactive protein (CRP) [5], procalcitonin [6], D-dimer [7], triggering receptor expressed on myeloid cells-1 [8], proadrenomedullin [9], copeptin [10], midregional pro–atrial natriuretic peptide [11], and N-terminal pro–brain natriuretic peptide [12]. Serum lactate level is a well-known prognostic biomarker in patients with sepsis, not only as a lactate clearance indicator [13] but also as an initial serum level [14]. However, in research on CAP, the prognostic role of serum lactate is not well studied. Until now, one abstract from the 2011 European Respiratory Society annual congress discussed the prognostic value of the initial lactate level in patients with CAP [15]. Another article by Mohamed and Ahmed [16] reported the prognostic value of lactate clearance in patients with CAP. However, these studies are limited in their ability to provide universal value because one is an abstract and the other is reported to a regionally restricted journal. To our knowledge, there is no relevant published article in the National
http://dx.doi.org/10.1016/j.ajem.2015.03.002 0735-6757/© 2015 Elsevier Inc. All rights reserved.
Please cite this article as: Gwak MH, et al, Initial serum lactate level is associated with inpatient mortality in patients with community-acquired pneumonia, Am J Emerg Med (2015), http://dx.doi.org/10.1016/j.ajem.2015.03.002
2
M.H. Gwak et al. / American Journal of Emergency Medicine xxx (2015) xxx–xxx
Center for Biotechnology Information database that shows that serum lactate level is associated with poor outcome in patients with CAP. Because the respiratory tract is a most common site for the origin of sepsis [17], the serum lactate level is thought to be a prognostic factor in CAP. It seems necessary to confirm whether this assumption is true and, if true, to what extent. Therefore, in the present study, we assessed the association between the initial serum lactate level and mortality in patients with CAP. We also evaluated whether the association of serum lactate with mortality remained after controlling for laboratory values in the PSI model and other laboratory variables that are routinely assessed in the emergency department (ED). 2. Methods 2.1. Study design and setting The present study was a retrospective chart review study. This study was approved by the institutional review board of the study hospital, and a waiver for informed consent was obtained. The study hospital was a 1000-bed urban academic tertiary care hospital with 42 licensed ED beds during the study period. In the study hospital, first-line antibiotics for patients with CAP were selected according to the 2007 version of the Infectious Disease Society of America/American Thoracic Society consensus guidelines [18]. 2.2. Selection of study sample We enrolled hospitalized patients with an admission diagnosis of any pneumonia (International Statistical Classification of Diseases, 10th Revision, codes CAP J12.x, J13, J14, J15.x, J16.x, J17.x, and J18.x) who were adults older than 18 years between October 1, 2012, and September 30, 2013.We excluded the following: (1) aspiration pneumonia, (2) hospital-acquired pneumonia, (3) pulmonary tuberculosis, and (4) diagnosed as something other than CAP after admission, such as cancer. All of these exclusions were performed via chart review by a trained physician. The remaining patients were regarded as eligible patients with CAP for the present study. Finally, we excluded those who had a lack of initial laboratory markers. 2.3. Outcome measures The primary outcome measure was inpatient mortality. 2.4. Data collection and processing Demographics, clinical data, physical findings, and laboratory results at ED presentation were collected by trained abstractors following the guidelines recommended by Gilbert et al [19]. Age, sex, comorbidities including malignancy, cerebrovascular disease, congestive heart failure, liver cirrhosis, and chronic kidney disease, current smoking, initial systolic blood pressure, pulse rate, respiratory rate, body temperature, mental status using AVPU (alert, voice, pain, unresponsive) scale, and SpO2 were extracted from electronic medical record. For initial laboratory markers, pH level, PO2, PaCO2, HCO3, SpO2, sodium, potassium, chloride, white blood cell (WBC), hematocrit (Hct), platelet, prothrombin time (PT) international normalized ratio (INR), activated partial thromboplastin time (aPTT), aspartate aminotransferase (AST), alanine aminotransferase, albumin, blood urea nitrogen (BUN), creatinine, glucose, CRP, and lactate were collected. The presence of pleural effusion was determined by radiologic report including chest x-ray or chest computed tomography. The PSI score upon arrival to the ED was calculated using the collected data. Survival status was determined from medical records. Serum lactate levels were primarily measured using arterial blood, but venous blood lactate levels were also permitted. Lactate levels were measured using a Stat Profile Critical Care Xpress Analyzer (Nova Biomedical, Waltham, MA). The measuring range for lactate in
this machine is from 0.3 to 20 mmol/L. This machine was periodically inspected by the manufacturer staff. 2.5. Statistical analysis All continuous data are presented as the mean and SD, and discrete data are presented as both counts and percentages. Patient characteristics were compared between the survivor group and the nonsurvivor group. Comparison of normally distributed data was performed using an independent sample t test. For non-normally distributed data, comparisons were performed using the Mann-Whitney U test. For categorical data, the χ2 test with a Fischer exact test for 2 × 2 tables was used. The results were considered significant at a threshold of P b .05 (2-tailed). Comparison of lactate levels between PSI classes was performed by analysis of variance. Logistic regression analyses were conducted to adjust for the effects of the initial serum lactate level on outcome and for possible confounding factors. In the first multivariable logistic regression model, we evaluated the odds ratios of lactate, CRP, and laboratory variables of PSI (arterial pH, BUN, sodium, glucose, Hct, PO2). Inputting these variables was conducted twice, first as continuous variables and second as categorical variables (lactate and CRP as tertiles, laboratory variables of PSI as dichotomized form of PSI version [arterial pH b7.35, BUN ≥ 30 mg/dL, sodium b130 mmol/L, glucose ≥ 250 mg/dL, Hct b30%, PO2 b 60 mm Hg]). In the second multivariable logistic regression model, we tested the significant association between the main outcome and each laboratory variable by univariate logistic regression, and we conducted a multivariable logistic regression model for trend factors (P b .10) in the univariate analysis. Dependent variables were also input both as continuous and categorical variables (tertiles). In the third model, we additionally input PSI as a predictor variable in the second regression model. Pneumonia severity index was input as a continuous variable (PSI score) and categorical variable (PSI class). The results are presented as odds ratio with a 95% confidence interval (CI). Statistical significance was defined as a 2-sided P b .05. For every logistic regression model, colinearity was tested, and variables that showed colinearity were excluded in the logistic model. The predictive values of lactate and PSI were also tested using the area under the receiver operating characteristic (AUROC) curve analysis. The SEM and P values for the receiver operating characteristic curves and comparisons between them were calculated following the methods of Hanley and McNeil [20,21]. All analyses were conducted using STATA 11.1 (StataCorp LP, College Station, TX) and SAS 9.1 (SAS Institute Inc, Cary, NC). 3. Results 3.1. Characteristics of study subjects During the study period, 557 patients were admitted with an admission diagnosis of any pneumonia. Among these patients, we excluded patients as follows: (1) 95 patients with aspiration pneumonia, (2) 11 patients with hospital-acquired pneumonia, (3) 27 patients with pulmonary tuberculosis, (4) 15 patients who were diagnosed with something other than CAP after admission such as cancer, and (5) 12 patients who did not have initial laboratory findings (lactate 7, CRP 4, glucose 1). Finally, 397 patients with CAP were enrolled in the present study (Figure). Table 1 presents baseline characteristics of enrolled patients. The mean age was 71.4 ± 13.1 years, and 229 patients (57.7%) were male. Cerebrovascular disease was the most frequent comorbidity (22.7%) followed by chronic renal disease (14.1%), neoplastic disease (13.4%), congestive heart failure (5.8%), and chronic liver disease (3.8%). Approximately 80% of the cohort were PSI class III or more, and 18.1% of the cohort was admitted to the intensive care unit.
Please cite this article as: Gwak MH, et al, Initial serum lactate level is associated with inpatient mortality in patients with community-acquired pneumonia, Am J Emerg Med (2015), http://dx.doi.org/10.1016/j.ajem.2015.03.002
M.H. Gwak et al. / American Journal of Emergency Medicine xxx (2015) xxx–xxx
3
Table 1 Baseline characteristics of enrolled patients All
Figure. Diagram of the study.
There were 46 mortality cases (11.6%) among the enrolled patients. There were significant differences between survivors and nonsurvivors in demographic characteristics, comorbidities, and physiologic and laboratory findings (Table 1). The PSI score was higher in the nonsurvivor group (117.9 ± 33.2) than in the survivor group (95.1 ± 31.8). Among 85 patients of PSI class I and II, 2 (2.4%) died. The length of hospital stay was higher in the nonsurvivor group (25.9 ± 30.5 days) than in the survivor group (18.4 ± 14.6 days). The mean lactate level for the enrolled patients was 1.7 ± 1.4 mmol/ L. The lactate level was significantly higher in nonsurvivors (2.4 ± 2.2 mmol/L vs 1.6 ± 1.2 mmol/L, P b .01).The mean lactate level was 1.0 ± 0.4 mmol/L for PSI class I, 1.1 ± 0.5 for PSI class II, 1.5 ± 1.1 for PSI class III, 1.8 ± 1.4 for PSI class IV, and 2.5 ± 2.0 for PSI class V (P b .001). 3.2. Logistic regression analysis results To investigate the association between inpatient mortality and lactate, CRP, and laboratory variables of PSI, multivariable logistic regression analysis was performed (Table 2). Dependent variables were input as both continuous and categorical. When inputted as continuous variables, serum lactate (adjusted odds ratio [AOR], 1.27; 95% CI, 1.051.55; P = .016) and CRP (AOR, 1.01; 95% CI, 1.00-1.01; P = .01) remained as significant factors for inpatient mortality. Laboratory variables of PSI were not significantly associated with inpatient mortality. When inputted as categorical variables, the third tertile of serum lactate (AOR, 3.02; 95% CI, 1.29-7.11; P = .011) and CRP (AOR, 4.63; 95% CI, 1.74-12.29; P = .002) remained as significant factors for inpatient mortality. Among the laboratory variables of PSI, only arterial pH remained as a significant factor. To investigate the association between inpatient mortality and collected laboratory variables, we performed multivariable logistic regression analysis (Table 3). Serum lactate, CRP, WBC, Hct, platelet, PT INR, aPTT, sodium, potassium, chloride, AST, alanine aminotransferase, albumin, BUN, creatinine, glucose, pH, PaO2, PaCO2, and HCO3were tested by univariate logistic regression analysis, and trend factors (P b .1) were chosen for dependent variables of multivariable logistic regression analysis. Dependent variables were input in both forms as continuous and categorical. When inputted as continuous variables, serum lactate CRP, WBC, Hct, sodium, AST, albumin, BUN, and pH showed a trend in the univariate regression analysis. Among these, serum lactate (AOR, 1.27; 95% CI, 1.03-1.56; P = .024), CRP (AOR, 1.01; 95% CI, 1.00-1.01; P = .035), and albumin (AOR, 0.37; 95% CI, 0.15-0.64; P = .002) were significant factors in the multivariable logistic regression model. When inputted as tertile variables, serum lactate, CRP, Hct, PT INR, aPTT, sodium, chloride, albumin, BUN, pH, and HCO3 tertiles showed a trend in the
Number 397 (100) Age 71.4 ± 13.1 Male sex, n (%) 229 (57.7) Comorbidity, n (%) Neoplastic disease 53 (13.4) Congestive heart failure 23 (5.8) Cerebrovascular disease 90 (22.7) Chronic renal disease 56 (14.1) Chronic liver disease 15 (3.8) Current smoking 24 (6.1) Physiology Systolic blood pressure, 120.4 ± 22.1 mm Hg Diastolic blood pressure, 73.5 ± 12.9 mm Hg Pulse rate, beats/min 92.7 ± 21.2 Respiratory rate, breaths/min 19.7 ± 2.8 Body temperature, °C 37.0 ± 0.9 95.0 ± 6.3 SpO2, % Mental status Alert 369 (92.9) Verbal 15 (3.8) Pain 7 (1.8) Unresponsive 6 (1.5) Laboratory and radiographic findings at ED Variables that belong to PSI Pleural effusion 141 (35.5) pH 7.44 ± 0.07 BUN (mg/dL) 22.4 ± 15.6 Sodium (mmol/L) 134.2 ± 5.5 Glucose (mg/dL) 147.7 ± 65.4 Hct (%) 34.5 ± 6.0 89.6 ± 31.8 PaO2 (mm Hg) Variables that do not belong to PSI WBC (×103/μL) 11.8 ± 6.6 242.0 ± 111.3 Platelet (×103/μL) PT INR 1.21 ± 0.35 aPTT (s) 34.6 ± 8.9 Potassium (mmol/L) 3.9 ± 0.7 Chloride (mmol/L) 105.9 ± 5.4 AST (IU/L) 49.9 ± 113.3 ALT (IU/L) 36.2 ± 79.6 Creatinine (mg/dL) 1.2 ± 1.3 Albumin (g/dL) 3.6 ± 0.6 31.4 ± 10.1 PaCO2 (mm Hg) HCO3 (mmol/L) 21.2 ± 4.3 CRP (mg/L) 103.2 ± 75.5 Lactate (mmol/L) 1.7 ± 1.4 PSI 97.7 ± 32.8 PSI class I 20 (5.0) II 65 (16.4) III 91 (22.9) IV 158 (39.8) V 63 (15.9) ICU admission (%) 72 (18.1) Length of hospital stay (d) 19.2 ± 17.3
Survivor
Nonsurvivor
351 (88.4) 70.9 ± 13.5 195 (55.6)
46 (11.6) 74.5 ± 10.0 34 (73.9)
P
43 (12.3) 21 (6.0) 80 (22.8) 44 (12.5) 12 (3.4) 23 (6.6)
10 (21.7) 2 (4.4) 10 (21.7) 12 (26.1) 3 (6.5) 1 (2.2)
121.4 ± 22.3
112.8 ± 19.2
.01
74.0 ± 12.9
70.0 ± 12.1
.05
92.5 ± 21.3 19.7 ± 2.7 37.0 ± 0.9 95.7 ± 5.3
94.6 ± 20.0 20.1 ± 3.8 36.9 ± 0.8 89.8 ± 10.0
.53 .36 .51 b.01 .18
331 (94.3) 9 (2.6) 7 (2.0) 4 (1.1)
38 (82.6) 6 (13.0) 0 2 (4.3)
127 (36.2) 7.44 ± 0.07 21.9 ± 15.6 134.4 ± 5.4 147.0 ± 64.2 34.7 ± 5.9 89.8 ± 31.4
14 (30.4) 7.42 ± 0.10 26.5 ± 14.9 132.7 ± 6.1 153.5 ± 74.3 32.8 ± 6.3 88.2 ± 35.7
11.6 ± 5.9 241.6 ± 105.3 1.20 ± 0.0.37 34.4 ± 8.9 3.9 ± 0.7 106.0 ± 5.3 45.0 ± 94.8 33.7 ± 69.9 1.2 ± 1.2 3.6 ± 0.5 31.2 ± 9.7 21.2 ± 4.1 98.1 ± 74.8 1.6 ± 1.2 95.1 ± 31.8
13.8 ± 10.5 245.0 ± 151.0 1.25 ± 0.16 36.6 ± 8.8 3.9 ± 0.9 105.2 ± 6.2 86.8 ± 203.7 55.5 ± 131.7 1.2 ± 2.4 3.2 ± 0.4 32.8 ± 12.8 21.0 ± 5.8 142.2 ± 70.1 2.4 ± 2.2 117.9 ± 33.2
19 (5.4) 64 (18.2) 82 (23.4) 139 (39.6) 47 (13.4) 42 (12.0) 18.4 ± 14.6
1 (2.17) 1 (2.17) 9 (19.6) 19 (41.3) 16 (34.8) 30 (65.2) 25.9 ± 30.5
.09 .02 .08 1.00 .87 .01 .40 .34
.59 .05 .06 .05 .52 .04 .75 .03 .84 .37 .11 .48 .32 .02 .08 .89 b.01 .32 .82 b.01 b.01 b.01 .21
b.01 b.01
Abbreviations: ALT, alanine aminotransferase; ICU, intensive care unit.
univariate regression analysis. Among them, only the third tertile of lactate (AOR, 2.79; 95% CI, 1.10-7.10; P = .031) and albumin (AOR, 0.13; 95% CI, 0.03-0.54; P = .005) were significant factors in the multivariable logistic regression model. Furthermore, to investigate the association between inpatient mortality and collected laboratory variables and PSI, multivariable logistic regression analysis was performed in the same way (Table 4). When inputted as continuous variables including PSI cores, serum lactate (AOR, 1.24; 95% CI, 1.01-1.53; P = .042), CRP (AOR, 1.01; 95% CI, 1.00-1.01; P = .025), and albumin (AOR, 1.36; 95% CI, 0.17-0.76; P = .008) were significant factors for inpatient mortality. The PSI scores showed a trend (P = .67). When inputted as tertile variables including PSI classes, only the third tertile of lactate (AOR, 2.60; 95% CI, 1.02-6.66; P = .046)
Please cite this article as: Gwak MH, et al, Initial serum lactate level is associated with inpatient mortality in patients with community-acquired pneumonia, Am J Emerg Med (2015), http://dx.doi.org/10.1016/j.ajem.2015.03.002
4
M.H. Gwak et al. / American Journal of Emergency Medicine xxx (2015) xxx–xxx
Table 2 Multivariable logistic regression analysis for inpatient mortality and lactate, CRP, and laboratory variables of PSI AOR By continuous variables Lactate CRP Arterial pH BUN Sodium Glucose Hct PaO2 By categorical variables Lactate 1st tertile 2nd tertile 3rd tertile CRP 1st tertile 2nd tertile 3rd tertile Arterial pH b7.35 BUN ≥30 mg/dL Sodium b130 mmol/L Glucose ≥250 mg/dL Hct b30% PaO2 b60 mm Hg
95% CI
Table 3 Multivariable logistic regression analysis for inpatient mortality for collected laboratory variables
P
1.27 1.01 0.02 1.00 1.00 1.00 0.97 1.00
1.05-1.55 1.00-1.01 0.00-1.74 0.97-1.02 0.90-1.01 1.00-1.00 0.92-1.03 0.99-1.01
.016 .001 .086 .655 .131 .808 .303 .780
Reference 2.19 3.02
0.90-5.33 1.29-7.11
.083 .011
Reference 2.51 4.63 3.34 0.92 1.30 0.59 1.77 1.65
0.89-7.07 1.74-12.29 1.35-8.31 0.41-2.03 0.59-2.90 0.17-1.98 0.87-3.60 0.66-4.12
.081 .002 .009 .827 .517 .391 .117 .287
and albumin (AOR, 0.15; 95% CI, 0.03-0.64; P = .010) were significant factors in the multivariable logistic regression model. Pneumonia severity index classes were not associated with inpatient mortality. Area under the receiver operating characteristic of the lactate was 0.654 (95% CI, 0.571-0.735) and that of the PSI score was 0.688 (95% CI, 0.608-0.768). There was no significant difference between them (P = .418). 4. Discussion In the present study, the initial serum lactate level was independently associated with inpatient mortality in patients with CAP in multivariable logistic regression models, which included laboratory variables and PSI scores. Compared with laboratory variables that included PSI, CRP, and others, the initial serum lactate level is superior when used to predict inpatient mortality. Current CAP guidelines recommend that physicians consider the site and level of care according to CAP severity [18,22]. The PSI [2] and CURB65 [3], 2 well-known severity measuring tools, are mortality prediction models derived from large cohorts. However, considering their discriminative values of 0.82 for PSI and 0.79 for CURB-65 reported by a recent systematic review [4], a more accurate mortality prediction tool for patients with CAP seems to be necessary. In addition, in the present study, the AUROC of the PSI score was 0.68. Thus, effort to find prognostic factors and determine their magnitude of impact is important. Predetermined factors such as age, sex, and comorbidities are already included in some CAP risk scores such as the PSI and PIRO (predisposition, insult, response, and organ dysfunction) [23]. However, newly introduced laboratory variables are not included in the mortality prediction model. Instead, they are still being tested for their prognostic performance in patients with CAP. Hyperlactatemia is a common laboratory finding in critically ill patients and is known to be associated with poor outcomes. Khosravani et al [24] reported that hyperlactatemia, defined by a serum lactate level N 2 mmol/L, was an independent factor for mortality in a large cohort of critically ill adult patients. Other studies support that the serum lactate level is an indicator for death in critically ill patients with sepsis [14,25]. These findings imply that the implementation of serum lactate in risk scoring systems may improve the prognostic performance for patients with sepsis. Actually, Chen et al [26] recently reported that
By continuous variables Lactate CRP WBC Hct Sodium AST Albumin BUN pH By categorical variables (as tertiles) Lactate 1st tertile 2nd tertile 3rd tertile CRP 1st tertile 2nd tertile 3rd tertile Hct 1st tertile 2nd tertile 3rd tertile PT INR 1st tertile 2nd tertile 3rd tertile aPTT 1st tertile 2nd tertile 3rd tertile Sodium 1st tertile 2nd tertile 3rd tertile Chloride 1st tertile 2nd tertile 3rd tertile Albumin 1st tertile 2nd tertile 3rd tertile BUN 1st tertile 2nd tertile 3rd tertile pH 1st tertile 2nd tertile 3rd tertile HCO3 1st tertile 2nd tertile 3rd tertile
AOR
95% CI
P
1.27 1.01 1.01 1.00 0.98 1.00 0.37 0.99 0.02
1.03-1.56 1.00-1.01 0.96-1.05 0.94-1.06 0.92-1.04 1.00-1.00 0.15-0.64 0.97-1.02 0.00-1.76
.024 .035 .813 .962 .537 .374 .002 .492 .086
Reference 1.69 2.79
0.65-4.39 1.10-7.10
.278 .031
Reference 1.19 2.49
0.36-3.90 0.72-8.60
.775 .150
0.86 0.69
0.37-2.01 0.26-1.83
.724 .450
0.68 1.82
0.23-2.04 0.60-5.53
.496 .286
1.62 0.56
0.52-5.04 0.15-2.06
.406 .382
0.62 0.81
0.25-1.53 0.30-2.16
.302 .668
0.62 0.61
0.23-1.67 0.23-1.60
.342 .312
0.66 0.13
0.29-1.51 0.03-0.54
.328 .005
0.88 1.13
0.33-2.36 0.43-2.99
.803 .803
0.71 0.54
0.27-1.83 0.22-1.37
.476 .197
0.58 1.32
0.22-1.55 0.52-3.34
.279 .563
For each model, dependent variables that showed a trend by univariate logistic regression analysis (P b .1) were chosen.
the modified Acute Physiology and Chronic Health Evaluation (APACHE) II, Sequential Organ Failure Assessment (SOFA), and Mortality in Emergency Department Sepsis (MEDS) by arterial lactate level showed better prognostic value than original score systems in patients with sepsis in the ED. Because the respiratory tract is a most common site for the origin of sepsis [17], the serum lactate level is supposed to be a prognostic factor in CAP. Actually, the present study revealed that the serum lactate level is independently associated with mortality in patients admitted with CAP. Therefore, further work may include the modification of the PSI using the serum lactate level. However, we did not try to modify the PSI using the serum lactate level. Considering laboratory variables in the PSI were not associated with mortality in the multivariable logistic
Please cite this article as: Gwak MH, et al, Initial serum lactate level is associated with inpatient mortality in patients with community-acquired pneumonia, Am J Emerg Med (2015), http://dx.doi.org/10.1016/j.ajem.2015.03.002
M.H. Gwak et al. / American Journal of Emergency Medicine xxx (2015) xxx–xxx Table 4 Multivariable logistic regression analysis including laboratory variables
By continuous variables Lactate CRP WBC Hct Sodium AST Albumin BUN pH PSI By categorical variables (as tertiles) Lactate 1st tertile 2nd tertile 3rd tertile CRP 1st tertile 2nd tertile 3rd tertile Hct 1st tertile 2nd tertile 3rd tertile PT INR 1st tertile 2nd tertile 3rd tertile aPTT 1st tertile 2nd tertile 3rd tertile Sodium 1st tertile 2nd tertile 3rd tertile Chloride 1st tertile 2nd tertile 3rd tertile Albumin 1st tertile 2nd tertile 3rd tertile BUN 1st tertile 2nd tertile 3rd tertile pH 1st tertile 2nd tertile 3rd tertile HCO3 1st tertile 2nd tertile 3rd tertile PSI class I II III IV V
AOR
95% CI
P
1.24 1.01 1.00 1.01 1.00 1.00 0.36 0.98 0.15 1.01
1.01-1.53 1.00-1.01 0.96-1.05 0.94-1.07 0.94-1.07 1.00-1.00 0.17-0.76 0.96-1.01 0.00-23.39 1.00-1.03
.042 .028 .832 .829 .924 .296 .008 .182 .459 .067
5
C-reactive protein was also shown to have an independently significant association with mortality in the present study. C-reactive protein is not only associated with mortality in sepsis [28] but also in CAP [5]. In addition, modification of the PSI using the CRP showed increased prognostic performance in CAP [29]. However, the authors did not show how to modify the PSI using the serum CRP level. Albumin was also an independent factor for mortality in the present study. Some studies have already showed the prognostic value of albumin in CAP [29,30]. Authors reported that modification of risk scores by albumin increased the predictive value more than original risk scores. Given those reports, CRP and albumin are preferable candidates as variables for a mortality prediction model of patients with CAP. 5. Limitations
Reference 1.70 2.60
0.65-4.47 1.02-6.66
.283 .046
Reference 1.42 3.01
0.42-4.78 0.81-11.14
.575 .099
Reference 0.94 0.73
0.40-2.25 0.27-1.98
.897 .536
Reference 0.62 1.60
0.20-1.89 0.51-5.01
.403 .421
Reference 1.52 0.52
0.47-4.94 0.13-2.00
.486 .340
Reference 0.68 1.00
0.26-1.77 0.36-2.82
.432 .997
Reference 0.67 0.62
0.24-1.86 0.23-1.64
.441 .335
Reference 0.71 0.15
0.31-1.64 0.03-0.64
.424 .010
The initial serum lactate level is independently associated with mortality in hospitalized patients with CAP. However, laboratory variables of the PSI or others were not, except for CRP and albumin. Future mortality prediction models for CAP should include the lactate level.
Reference 0.80 0.89
0.29-2.20 0.32-2.49
.669 .819
References
Reference 0.87 0.58
0.31-2.36 0.22-1.52
.771 .267
Reference 0.60 1.38
0.22-1.62 0.54-3.54
.316 .497
Reference 0.41 2.05 1.94 2.79
0.02-7.59 0.21-20.25 0.21-18.06 0.27-28.39
.547 .541 .561 .386
For each model, dependent variables that showed a trend by univariate logistic regression analysis (P b .1) were chosen.
regression model that included the serum lactate level (Table 2), there could be various ways to put serum lactate level in the PSI. (1) Modification of the PSI plus serum lactate level or (2) modification of the PSI including serum lactate level and deleting other laboratory variables may be suggested. (3) It may be plausible to test another scoring system that includes the serum lactate level, such as a modified early warning score in patients with CAP [27].
There are some limitations in the present study. First, the present study represents a retrospective observational study performed in a single tertiary hospital. Demographic and regional differences may result in different cohorts of hospitalized patients with CAP, and this may lead to different statistical results. Second, we did not include some laboratory variables that were not routine laboratory evaluations in the study hospital, such as procalcitonin or ESR and others. Third, the study hospital was an urban academic tertiary care hospital, and the visiting patients are generally in severe conditions, which results in high proportion of high-risk CAP (PSI IV-V) and higher mortality in lowrisk CAP. The characteristics of the study site and patients might be limitation on generalizability of the results, but the authors expect that the results could be applied to patients with CAP visiting tertiary care or general hospitals that have similar characteristics with ours. Further large studies including these laboratory variables that have been shown to be associated with mortality in CAP are warranted. Fourth, the admission diagnosis was not specific in the study hospital. Although a trained physician picked out patients with CAP among all patients with pneumonia, there may have been some errors. 6. Conclusions
[1] Welte T, Suttorp N, Marre R. CAPNETZ – Community-Acquired Pneumonia Competence Network. Infection 2004;32(4):234–8. [2] Fine MJ, Auble TE, Yealy DM, Hanusa BH, Weissfeld LA, Singer DE, et al. A prediction rule to identify low-risk patients with community-acquired pneumonia. N Engl J Med 1997;336(4):243–50. [3] Lim WS, van der Eerden MM, Laing R, Boersma WG, Karalus N, Town GI, et al. Defining community acquired pneumonia severity on presentation to hospital: an international derivation and validation study. Thorax 2003;58(5):377–82. [4] Kwok CS, Loke YK, Woo K, Myint PK. Risk prediction models for mortality in community-acquired pneumonia: a systematic review. Biomed Res Int 2013;2013: 504136. http://dx.doi.org/10.1155/2013/504136 [Epub 2013 Oct 21]. [5] Chalmers JD, Singanayagam A, Hill AT. C-reactive protein is an independent predictor of severity in community-acquired pneumonia. Am J Med 2008;121(3):219–25. http://dx.doi.org/10.1016/j.amjmed.2007.10.033. [6] Horie M, Ugajin M, Suzuki M, Noguchi S, Tanaka W, Yoshihara H, et al. Diagnostic and prognostic value of procalcitonin in community-acquired pneumonia. Am J Med Sci 2012;343(1):30–5. http://dx.doi.org/10.1097/MAJ.0b013e31821d33ef. [7] Querol-Ribelles JM, Tenias JM, Grau E, Querol-Borras JM, Climent JL, Gomez E, et al. Plasma d-dimer levels correlate with outcomes in patients with communityacquired pneumonia. Chest 2004;126(4):1087–92. [8] Tejera A, Santolaria F, Diez ML, Alemán-Valls MR, González-Reimers E, MartínezRiera A, et al. Prognosis of community acquired pneumonia (CAP): value of triggering receptor expressed on myeloid cells-1 (TREM-1) and other mediators of the inflammatory response. Cytokine 2007;38(3):117–23 [Epub 2007 Jul 30]. [9] Christ-Crain M, Morgenthaler NG, Stolz D, Müller C, Bingisser R, Harbarth S, et al. Pro-adrenomedullin to predict severity and outcome in community-acquired pneumonia. Crit Care 2006;10(3):R96 [Epub 2006 Jun 28]. [10] Kolditz M, Halank M, Schulte-Hubbert B, Bergmann S, Albrecht S, Höffken G. Copeptin predicts clinical deterioration and persistent instability in community-
Please cite this article as: Gwak MH, et al, Initial serum lactate level is associated with inpatient mortality in patients with community-acquired pneumonia, Am J Emerg Med (2015), http://dx.doi.org/10.1016/j.ajem.2015.03.002
6
M.H. Gwak et al. / American Journal of Emergency Medicine xxx (2015) xxx–xxx
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
acquired pneumonia. Respir Med 2012;106(9):1320–8. http://dx.doi.org/10.1016/j. rmed.2012.06.008 [Epub 2012 Jun 23]. Prat C, Lacoma A, Dominguez J, Papassotiriou J, Morgenthaler NG, Andreo F, et al. Midregional pro-atrial natriuretic peptide as a prognostic marker in pneumonia. J Infect 2007;55(5):400–7 [Epub 2007 Sep 10]. Jeong KY, Kim K, Kim TY, Lee CC, Jo SO, Rhee JE, et al. Prognostic value of N-terminal pro-brain natriuretic peptide in hospitalised patients with community-acquired pneumonia. Emerg Med J 2011;28(2):122–7. http://dx.doi.org/10.1136/emj.2009. 089383 [Epub 2010 May 29]. Nguyen HB, Rivers EP, Knoblich BP, Jacobsen G, Muzzin A, Ressler JA, et al. Early lactate clearance is associated with improved outcome in severe sepsis and septic shock. Crit Care Med 2004;32(8):1637–42. Mikkelsen ME, Miltiades AN, Gaieski DF, Goyal M, Fuchs BD, Shah CV, et al. Serum lactate is associated with mortality in severe sepsis independent of organ failure and shock. Crit Care Med 2009;37(5):1670–7. http://dx.doi.org/10.1097/CCM.0b013e31819fcf68. Chalmers JD, Singanayagam A, Akram A, Mandal P, Choudhury G, Smith M, et al. Lactate is an independent marker of severity in hospitalised patients with communityacquired pneumonia. ERJ 2011;38:1463 [http://erj.ersjournals.com/content/38/ Suppl_55/p1463.meeting_abstract]. Mohamed KAE, Ahmed DA. Prognostic value of lactate clearance in severe community acquired pneumonia. Egypt J Chest Dis Tuberc 2014. http://dx.doi.org/10.1016/j.ejcdt. 2014.05.11 [http://www.sciencedirect.com/science/article/pii/S0422763814001149]. Martin GS. Sepsis, severe sepsis and septic shock: changes in incidence, pathogens and outcomes. Expert Rev Anti Infect Ther 2012;10(6):701–6. http://dx.doi.org/10. 1586/eri.12.50. Mandell LA, Wunderink RG, Anzueto A, Bartlett JG, Campbell GD, Dean NC, et al. Infectious Diseases Society of America/American Thoracic Society consensus guidelines on the management of community-acquired pneumonia in adults. Clin Infect Dis 2007;44(Suppl. 2):S27–72. Gilbert EH, Lowenstein SR, Koziol-McLain J, Barta DC, Steiner J. Chart reviews in emergency medicine research: Where are the methods? Ann Emerg Med 1996; 27(3):305–8.
[20] Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143(1):29–36. [21] Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983;148(3): 839–43. [22] Lim WS, Baudouin SV, George RC, Hill AT, Jamieson C, Le Jeune I, et al. BTS guidelines for the management of community acquired pneumonia in adults: update 2009. Thorax 2009;64(Suppl. 3):1–55. [23] Rello J, Rodriguez A, Lisboa T, Gallego M, Lujan M, Wunderink R. PIRO score for community-acquired pneumonia: a new prediction rule for assessment of severity in intensive care unit patients with community-acquired pneumonia. Crit Care Med 2009;37(2):456–62. http://dx.doi.org/10.1097/CCM.0b013e318194b021. [24] Khosravani H, Shahpori R, Stelfox HT, Kirkpatrick AW, Laupland KB. Occurrence and adverse effect on outcome of hyperlactatemia in the critically ill. Crit Care 2009; 13(3):R90. [25] Shapiro NI, Howell MD, Talmor D, Nathanson LA, Lisbon A, Wolfe RE, et al. Serum lactate as a predictor of mortality in emergency department patients with infection. Ann Emerg Med 2005;45(5):524–8. [26] Chen YX, Li CS. Arterial lactate improves the prognostic performance of severity score systems in septic patients in the ED. Am J Emerg Med 2014;32(9):982–6. [27] Jo S, Lee JB, Jin YH, Jeong TO, Yoon JC, Jun YK, et al. Modified early warning score with rapid lactate level in critically ill medical patients: the ViEWS-L score. Emerg Med J 2013;30(2):123–9. [28] Gradel KO, Jensen TG, Kolmos HJ, Pedersen C, Vinholt PJ, Lassen AT. Does C-reactive protein independently predict mortality in adult community-acquired bacteremia patients with known sepsis severity? APMIS 2013;121(9):835–42. [29] Lee JH, Kim J, Kim K, Jo YH, Rhee J, Kim TY, et al. Albumin and C-reactive protein have prognostic significance in patients with community-acquired pneumonia. J Crit Care 2011;26(3):287–94. [30] Viasus D, Garcia-Vidal C, Simonetti A, Manresa F, Dorca J, Gudiol F, et al. Prognostic value of serum albumin levels in hospitalized adults with community-acquired pneumonia. J Infect 2013;66(5):415–23.
Please cite this article as: Gwak MH, et al, Initial serum lactate level is associated with inpatient mortality in patients with community-acquired pneumonia, Am J Emerg Med (2015), http://dx.doi.org/10.1016/j.ajem.2015.03.002