A New Marker of Sepsis Post Burn Injury?* Jennifer D. Paratz, PhD, FACP1,2,3; Jeffrey Lipman, MBBCh, FCICM, MD1,2; Robert J. Boots, PhD, FCICM1,2; Michael J. Muller, MMedSc, FRACS1,4; David L. Paterson, PhD, FRACP, FRCPA5
Objectives: Accurate diagnosis of sepsis is difficult in patients post burn due to the large inflammatory response produced by the major insult. We aimed to estimate the values of serum N-terminal pro-Btype natriuretic peptide and procalcitonin and the changes in hemodynamic variables as markers of sepsis in critically ill burn patients. Design: Prospective, observational study. Setting: A quaternary-level university-affiliated ICU. *See also p. 2137. 1 Burns, Trauma and Critical Care Research Centre, The University of Queensland, Brisbane, QLD, Australia. 2 Department of Intensive Care Medicine, Royal Brisbane and Women’s Hospital, Brisbane, QLD, Australia. 3 Department of Rehabilitation Sciences, Griffith University, Brisbane, Australia. 4 Professor Stuart Pegg Burn Unit, Royal Brisbane and Women’s Hospital, Brisbane, QLD, Australia. 5 University of Queensland Centre for Clinical Research, Brisbane, QLD, Australia. Supported, in part, by Royal Brisbane and Women’s Hospital Research Foundation. Dr. Paratz consulted for Edwards Lifesciences; is employed by University of Queensland; and received support for travel from Australian and New Zealand Intensive Care Society, South African Burns Society, and Australian Physiotherapy Association. Her institution received grant support from National Health and Medical Research Council (NIHMRC), Office of Health Medical Research, NHMRC Equipment grants, Royal Brisbane Hospital Research Foundation, and Office Medical Research; received provision of equipment from Edwards Lifesciences (equipment lent free of cost to ICU); and lectured for University of Queensland, Australian Catholic University. Dr. Lipman’s institution received grant support from the Royal Brisbane and Women’s Hospital Research Foundation; received provision of equipment from Edwards Lifesciences (equipment lent free of cost to ICU); served as board member for Bayer European Society Council Intensive Care Medicine Advisory Board; consulted for Merk Sharp Dohme (Aust) Pty Ltd, Pfizer Australia, and AstraZeneca; received grant support from AstraZeneca; and lectured for AstraZeneca and Pfizer Australia Pty Ltd. Dr. Boots’ institution received grant support from Royal Brisbane and Women’s Hospital Research Foundation and received provision of equipment from Edwards Lifesciences (equipment lent free of cost to ICU). Dr. Muller’s institution received grant support from Royal Brisbane and Women’s Hospital Research Foundation, received provision of equipment from Edwards Lifesciences (equipment lent free of cost to ICU), and received grant support from Royal Brisbane and Women’s Hospital Research Foundation. Dr. Paterson consulted for Merck, AstraZeneca, and Pfizer; and lectured for AstraZeneca. For information regarding this article, E-mail: [email protected] Copyright © 2014 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: 10.1097/CCM.0000000000000400
Critical Care Medicine
Patients: Fifty-four patients with burns to total body surface area of greater than or equal to 15%, intubated with no previous cardiovascular comorbidities, were enrolled. Interventions: At admission, a FloTrac/Vigileo system was attached and daily blood samples taken from the arterial catheter. Infection surveillance was carried out daily with patients classified as septic/ nonseptic according to American Burns Consensus criteria. Measurements and Main Results: N-terminal pro-B-type natriuretic peptide, procalcitonin, and waveform analysis of changes in stroke volume index and systemic vascular resistance index were measured within the first 24 hours after burn and daily thereafter for the length of the ICU stay or until their first episode of sepsis. Prevalences of stroke volume variation less than 12% (normovolemia) with hypotension (systolic blood pressure < 90 mm Hg) were recorded. Patients with sepsis differed significantly from “no sepsis” for N-terminal pro-B-type natriuretic peptide, systemic vascular resistance index, and stroke volume index on days 3–7. Procalcitonin did not differ between sepsis and “no sepsis” except for day 3. Area under the receiver operating characteristic curves showed excellent discriminative power for B-type natriuretic peptide (p = 0.001; 95% CI, 0.99–1.00), systemic vascular resistance index (p < 0.001; 95% CI, 0.97– 0.99), and stroke volume index (p < 0.01; 95% CI, 0.96–0.99) in predicting sepsis but not for procalcitonin (not significant; 95% CI, 0.29–0.46). A chi-square crosstab found that there was no relationship between hypotension with normovolemia (stroke volume variation < 12%) and sepsis. Conclusions: Serum N-terminal pro-B-type natriuretic peptide levels and certain hemodynamic changes can be used as an early indicator of sepsis in patients with burn injury. Procalcitonin did not assist in the early diagnosis of sepsis. (Crit Care Med 2014; 42:2029–2036) Key Words: burns; natriuretic peptide, brain; procalcitonin; sepsis syndrome; systemic inflammatory response syndrome
D
espite an increased overall survival rate, severe infection remains the main cause of mortality in burn injury (1, 2). Prompt antimicrobial therapy for septic shock demonstrates a mortality benefit in general critical care patients (3); hence, an early diagnosis is important. However, diagnosing sepsis in burn patients has been associated with a number of difficulties. www.ccmjournal.org
2029
Paratz et al
The accepted methods to diagnose sepsis in general intensive care patients (4) are insensitive and nonspecific in burn patients due to the sustained systemic inflammatory state following a burn injury. In an attempt to improve accuracy of diagnosis, a consensus conference was held by the American Burns Association (ABA) in 2007 (5). Table 1 identifies both the triggers to search for infection and confirmation of infection in burn subjects suggested by this conference (5). However, there is inevitably a time delay as confirmation of results from tissue and blood cultures can take 48 hours. In a further attempt to find indicators of sepsis in burns, inflammatory markers have been investigated. Evidence for procalcitonin (PCT), erythrocyte sedimentation rate, C-reactive protein, and interleukin-6 is conflicting (6–11), with PCT the strongest contender (12). B-type natriuretic peptide (BNP) has been shown to be elevated in sepsis in nonburn patients independent of cardiac status (13–15) but has not previously been researched specifically in burn patients. Following increased use of hemodynamic monitoring arterial pressure–based cardiac output (APCO) device (Vigileo, FloTrac, Edwards Lifesciences, Irvine, CA) in our ICU, an audit of retrospective data indicated that stroke volume index Table 1. American Burn Association Criteria for Sepsis Sepsis should be considered when three or more of the following criteria are met: 1. Temperature: > 39°C or < 36.5°C 2. Progressive tachycardia: > 110 beats/min 3. Progressive tachypnea: a. > 25 breaths/min not ventilated b. Minute ventilation > 12 L/min ventilated 4. Thrombocytopenia (not applied until 3 d after initial resuscitation): < 100,000/μL 5. Hyperglycemia (in the absence of preexisting diabetes mellitus) a. Untreated plasma glucose > 200 mg/dL or equivalent mM/L
(SVI) increased and systemic vascular resistance index (SVRI) decreased above those figures expected for the “flow stage” of the burn hemodynamic response during proven infection. Because of the association of BNP with sepsis in general patients mentioned previously, we instituted a prospective study aiming to investigate the efficacy of serum BNP and PCT, changes in predicted values of SVI and SVRI, and prevalences of stroke volume variation (SVV) less than 12% (normovolemia) with hypotension (systolic blood pressure < 90 mm Hg) as criteria for an accurate expedited marker of suspicion of sepsis in burns.
MATERIALS AND METHODS This study was conducted in a quaternary-level universityaffiliated ICU. Approval was granted by both the Hospital and University Human Medical Ethics Committees and consent was obtained from the next of kin or legal guardian of every subject. Subjects Subjects to be included were those with greater than or equal to 15% TBSA burns who were intubated, with an indwelling arterial catheter, previous normal cardiac function, and not expected to be extubated within 72 hours. Subjects were excluded if they had toxic epidermal necrolysis, had a history of cardiovascular disease including coronary artery disease, cardiac failure, or congenital heart problems, had multiple trauma, took longer than 8 hours to reach the tertiary center, or were for compassionate care only. They were resuscitated using the modified Parkland formula (including judicious use of colloid) (16) aiming at the endpoints of urinary output 0.5–2 mL/kg/hr lean body mass, mean arterial pressure greater than 60 mm Hg, intra-abdominal pressure less than 25 mm Hg, hemoglobin less than 16 mg/dL and more than 8 mg/dL, and Na less than 155 mEq/L. Although a number of patients underwent part of their resuscitation in a smaller hospital or in transit, personnel were in contact with tertiary center staff regarding fluid resuscitation. All fluids received were recorded, and it was stated Yes/No whether patients had exceeded the Parkland formula. All patients received the usual standard of care:
b. > 7 U of insulin/hr IV drip c. Significant resistance to insulin (> 25% increase in insulin requirement over 24 hr)
6. Inability to continue enteral feedings > 24 hr a. Abdominal distension
b. High gastric residuals (residuals two times feeding rate) c. Uncontrollable diarrhea (> 2,500 mL/d) Documented infection when any of these criteria are met: 1. Positive culture 2. Pathologic tissue source 3. Clinical response to antimicrobials Reproduced with permission from Greenhalgh et al (5).
2030
www.ccmjournal.org
• Mechanical ventilation initially using the modes of synchronized intermittent mandatory ventilation or BiLevel initially, followed by pressure support ventilation with positive end-expiratory pressure or inspired oxygen as the patient’s sedation requirements lessened. • Inhalation injury diagnosed by bronchoscopy for those suspected of a lung injury. • Sedation and analgesia administered according to routine guidelines, including a mixture of fentanyl, k etamine, morphine, and midazolam. • Escharotomies as required. • Early debridement and grafting with autografts, allografts, or both as soon as initial resuscitation is complete.
Intensive care clinical staff were blinded to FloTrac results (i.e., cardiac output, systemic vascular resistance, and SVV) but accessed heart rate, arterial blood pressure, arterial blood gases, urine output, fluid balance, electrolytes, lactate, and central September 2014 • Volume 42 • Number 9
Clinical Investigations
venous pressure (i.e., the usual variables to provide resuscitation and management in burn patients). All usual demographics were recorded and are listed in Table 1. Hemodynamic Indices An APCO device (Vigileo, FloTrac, Edwards Lifesciences) was attached to an existing peripheral artery catheter to record hemodynamic variables. This third-generation model of semiinvasive machine has shown good correlation with more invasive methods and is capable of tracking hemodynamic changes and trends (19, 20). We considered utilization of the Vigileo to be more acceptable from an ethical point of view than the insertion of a pulmonary artery catheter. SVI, SVRI, and SVV were downloaded continuously, together with all other observations, that is, arterial blood pressure, heart rate, respiratory rate, Spo2, and central venous pressure, to the Integrated Clinical Information Program (ICIP) (Phillips Healthcare, Andover, MA). A program using purpose-written software automatically calculated the daily highest, lowest, mean, and median for all these variables. The variables are described as percentage of normal SI and SVRI for age, height and gender, and prevalence of hypotension (systolic blood pressure < 90 mm Hg) with normovolemia (i.e., SVV < 12%). SVV% is a measure that provides information on relative preload responsiveness. SVV is calculated as percentage change between the maximal and minimal stroke volumes divided by the average of the minimum and maximum over a respiratory cycle or other period of time. This measure can be useful in differentiating the cause of hypotension between hypovolemia or a distributive cause and was therefore included in the outcome measures. SVV was recorded only when the subject was on full mechanical ventilation with a tidal volume of greater than 8 mL/kg and no arrhythmias. The SVV at which the burn patient is fluid responsive has been shown to be 13.6% (21); therefore, the figure of less than 12% was chosen to ensure normovolemia. Blood Samples Blood samples were taken once daily from indwelling arterial catheters for measurements of PCT and BNP. They were then centrifuged, separated, and plasma stored at –80°C and analyzed in bulk at the end of the study. Serum levels of the inactive cleavage product of BNP, N-terminal fragment (NT-proBNP), were determined by electrochemiluminescence immunoassay (ECLIA) (Elecsys proBNP II Assay, Cobas, Henningsdorf, Germany). The coefficient of variation was 2%, with a lower limit of detection of 5 pg/mL. NT-proBNP was measured as it has higher plasma levels and a longer half-life and, therefore, slower fluctuations than BNP, but NT-proBNP will be referred to by the term “BNP” throughout the remainder of the article. PCT was measured by ECLIA (Elecsys BRAHMS PCT, Cobas), with a coefficient of variation of 2% and a lower detection rate of less than or equal to 0.02 ng/mL. Diagnosis of Sepsis The following criteria by the research staff were independent and additional to infection surveillance by the clinical staff. For the purposes of diagnosis of sepsis in this study, all triggers Critical Care Medicine
(heart rate, temperature, and respiratory rate [if not fully sedated and fully ventilated], platelets, prevalence of feeding intolerance, and insulin requirements) as per Table 1 were continuously screened by the researchers. If these variables indicated infection, tissue diagnosis and blood cultures were requested if the clinical staff had not already ordered these. Blood cultures consisted of three separate sets from separate venipuncture sites with a minimum of 20-mL volume (22). To satisfy the criteria for sepsis, three or more criteria were presented in Table 1, with either a positive tissue or blood culture diagnosis. We purposely did not use the criteria of response to antimicrobial therapy in our diagnosis. Additionally, the organ suspected of being the source of infection was required to satisfy the modified criteria for Centers for Disease Control and Prevention as per the ABA (5). When all results were available, the patients were scored as Yes/No for the presence of sepsis for each day. Clinical staff had the responsibility for prescription of antibiotics, based on their clinical judgment of whether an infection was present. Statistics The analyses were performed using SPSS 19.0 software (SPSS, Chicago, IL). Patients satisfied the criteria for sepsis on a Yes/ No basis, with this criteria applied every day. Overall values for BNP, PCT, SVI, and SVRI were compared between patients described as septic/nonseptic for each day using a one-way analysis of variance for time. As group allocation, that is, whether sepsis criteria were met, altered each day, a t test compared factors between sepsis and nonsepsis for each day. A chisquare crosstab investigated whether BNP levels were related to exceeding the recommended Parkland formula. A chi-square crosstab investigated whether significantly more patients with SVV less than 12% and systolic arterial blood pressure less than 90 mm Hg satisfied the criteria for sepsis. The sensitivity, specificity, and positive likelihood ratio for BNP, PCT, SVI, and SVRI in diagnosing sepsis were calculated at the best cutoff value (Table 2) To determine the predictive ability of these four factors, receiver operating characteristic (ROC) curves were constructed and the areas under the curve were calculated with 95% CIs. Factors were then entered into a binary logistic regression to detect which factors were independent markers of sepsis (23). A p value of less than 0.05 was considered to be statistically significant in all tests.
RESULTS Fifty-four patients were included in the study, with clinical characteristics included in Table 3. Figure 1 indicates the flow of patients and reasons for exclusion. Although it may appear unusual that subjects with less than or equal to 20% burns were admitted to intensive care, all these subjects had inhalation or airway injuries, necessitating an endotracheal tube. There were 23 of 54 patients with sepsis confirmed. Day refers to the actual day post burn not the day of admission to the tertiary burn unit. Only the first episode of sepsis for each patient was recorded for the purposes of this study. Following resolution of this first episode, no further pathology tests or hemodynamic www.ccmjournal.org
2031
Paratz et al
Table 2. Sensitivity, Specificity, Positive Likelihood Ratios, Negative Likelihood Ratios, and Diagnostic Odds Ratio for Each Factor as a Marker of Sepsis—For the First Episode of Sepsis Factor
B-Type Natriuretic Peptide
Procalcitonin
Stroke Volume Index
Systemic Vascular Resistance Index
Sensitivity
95.9% (86.2–99.1%)
20.2% (12.1–34.3%)
94.3% (84.2–98.4%)
86.2% (73.3–93.4%)
Specificity
99.4% (98.1–99.8%)
64.3% (59.1–69.2%)
94.2% (91.3–96.5%)
97.1% (94.6–98.2%)
Positive likelihood ratios
165.5 (41.5–659.7)
0.57 (0.321–1.01)
Negative likelihood ratios
40.04 (0.01–0.16)
1.24 (1.06–1.46)
0.07 (0.02–0.19)
0.15 (0.08–0.29)
0.53 (0.23–1.22)
0.46 (0.22–0.94
166.5 (62.1–445.2)
Diagnostic odds ratio
4030.1 (554.5–29293.3)
14.68 (9.47–22.1)
24.6 (13.97–43.47)
Cutoff values are taken as B-type natriuretic peptide > 1,200 (pg/mL); procalcitonin 1.4 (ng/mL); stroke volume index ≥ 160% of predicted for age, height, and gender; systemic vascular resistance index ≤ 50% of predicted for age, height, and gender.
Table 3.
Demographics and Patient Details for Patients Both With and Without Sepsis
Factor
Sepsis (n = 23)
Age (yr)a
42.65 (12.8)
No Sepsis (n = 31)
37.68 (10.4)
All Subjects (n = 54)
40.16 (18–60)
16:7
29:2
45:9
TBSA %a
46.2 (15.3)
42.0 (16.7)
44.1 (16.0)
Total body surface area full thickness %b
21.8 (11.0)
20.8 (7.4)
Baux indexa
96.1 (18.6)
89.8 (18.5)
93.0 (18.5)
17 (74%)
7 (23%)
24 (44%)
9 (8–11)
9 (7–11)
Gender, male:female
b
Inhalation injurya
9 (7–11)
Admission Sequential Organ Failure Assessment (17)c
1.1 (0.9–1.5)
Admission shock index (18)c Severity of burn injury injury (Belgian Outcome in Burn Injury) (24)c LOS (ICU) (d)a LOS (hospital) (d)
a
Mortality
3 (1–6)
20.86(9.2)
1.1 (0.85–1.5) 3 (1–6)
1.1 (0.85–1.5) 3 (1–6)
21.2 (4.6)
17.0 (5.4)
19.1 (5.0)
42.9 (7.6)
37.7 (9.4)
40.8 (8.5)
1
0
1
Depression
5
2
7
Asthma
3
2
5
Hepatitis C
2
2
4
Comorbidities
Total volume (0–24 hr) (mL/kg/%TBSA)
9.65 (3.2)
Total % of ICU stay on vasoactive drugs
22%
15%
17.5%
Abdominal hypertension
2
1
3
Escharotomy/fasciotomy
4
5
9
No. of grafting procedures/patient
6.5 (3.2)
5.1 (2.9)
5.5 (3.0)
Time to first excision (d)
1.9 (0.8)
1.8 (0.7)
1.85 (0.7)
b
9.88 (3.7)
9.77 (3.5)
TBSA = total body surface area, LOS = length of stay. a The subjects who developed sepsis were significantly different at the p < 0.01 level. b The subjects who developed sepsis were significantly different at the p < 0.05 level. c Median (range). Values are mean and sd or number and percentage unless otherwise specified.
2032
www.ccmjournal.org
September 2014 • Volume 42 • Number 9
Clinical Investigations
(p < 0.001) and SVI significantly lower (p < 0.001) in the first 2 days post burn compared to days 3–7. However, those patients who reached the criteria for sepsis had a significantly lower SVRI (p < 0.001) and higher SVI (p < 0.001) on days 3–5 than patients who did not reach the criteria. BNP was significantly lower on days 1 and 2 (p < 0.001) compared with days 3–7 in all patients. However, in those who reached the sepsis criteria, BNP was significantly higher from days 3 to 7. There was no correlation between levels of BNP and exceeding the recommended Parkland formula in the first 48 hours. A chi-square crosstab indicated that there was no relationship between hypotension with normovolemia (SVV < 12%) and prevalence of sepsis. The overall sensitivity and specificity for each factor are described in Table 2. Optimum cutoff values are listed in Table 2. A receiver operating characteristic (ROC) curve was constructed for the factors and is illustrated in Figure 4. Logistic regression models were constructed to investigate the association between the potential markers and sepsis. First, we investigated each factor individually and then included all significant factors (BNP, SVRI, and SVI) in a multivariable model with adjustment for severity of burn (24). These factors were entered into a binary logistic regression using the backward conditional method, with sepsis (Yes = 1, No = 0) as the dependent variable. The overall model was significant at the p less than 0.001 level according to the model c2 statistic (Table 2). The model predicted 99.2% of the responses correctly. The overall result was that BNP was an independent marker of sepsis (Table 5).
Figure 1. Flowchart of patients through the study.
Figure 2. Source of infection for the septic episode.
DISCUSSION
Most Commonly Isolated Bacterial Pathogens Table 4.
Bacteria
Gram negative
No. of Pathogens Isolated From Positive Blood Cultures or Tissue Culture
21
Pseudomonas aeruginosa
8
Klebsiella species
6
Acinetobacter baumannii
2
Serratia marcescens
1
Escherichia coli
2
Gram positive
3
Enterococcus species
1
Staphylococcus aureus
2
data were recorded for the study. However, outcome data, for example, length of stay and mortality, were recorded. Figure 2 indicates sites of infection and Table 4 organisms identified. The mean level for each day for patients with episodes of sepsis and no sepsis is illustrated for BNP (Fig. 3A), PCT (Fig. 3B), SVRI (Fig. 3C), and SVI (Fig. 3D). SVRI was significantly higher Critical Care Medicine
Early, accurate diagnosis of sepsis is vital in thermal injury both for a mortality advantage potentially achieved by timely initiation of empiric antibiotics and to decrease inappropriate antibiotic use increasing the risk of antimicrobial resistance. This study found a high degree of both positive and negative predictive ability for BNP and the hemodynamic values of stroke volume index (SI) and SVRI but not for PCT or hypotension with normovolemia. BNP is a well-known biomarker of cardiac failure (25). High levels have been well documented and accepted to differentiate dyspnea in the emergency setting, specify levels of severity, and provide prognosis in cardiac failure (26, 27). However, increases in BNP have been well documented in other clinical conditions including sepsis (14, 28, 29). At first, this was attributed to cardiac failure in conjunction with sepsis (29) and then later to the accompanying myocardial depression that occurs in 50% of patients with septic shock (29) considered to be functional rather than structural (30–32) with adequate perfusion of coronary arteries. More recently, investigators (33) have claimed that high levels in sepsis are related to an alteration in the BNP clearance pathway specifically the neutral endopeptidase 24.11 with a parallel increase in BNP. It is being increasingly recognized that elevated levels of BNP are present in subjects without myocardial dysfunction including marathon runners at the completion of a race (34) and healthy volunteers given an infusion of LPS to induce sepsis (35). www.ccmjournal.org
2033
Paratz et al
Figure 3. N-terminal pro-B-type natriuretic peptide (NT-proBNP) % of predicted for days 1–7 (a), procalcitonin (PCT) over days 1–7 (b), systemic vascular resistance index (SVRI) % of predicted for days 1–7 (c), and stroke volume index (SVI) % of predicted for days 1–7 (d). Number of patients with sepsis (S) and those not reaching the criteria for sepsis (NS) differed each day, apart from days 1 and 2 on which no subjects had sepsis. Number of patients was as follows: day 1 (NS) n = 54, day 2 (NS) n = 54, day 3 (S) n = 4 and (NS) n = 50, day 4 (S) n = 11 and (NS) n = 43, day 5 (S) n = 16 and (NS) n = 37, day 6 (S) n = 12 and (NS) n = 31, and day 7 (S) n = 5 and (NS) n = 28. * indicates that on that particular day patients reaching sepsis criteria were significantly different from those not reaching sepsis criteria: *p < 0.05, ** p < 0.01.
Figure 4. a, Area under the curve for B-type natriuretic peptide (BNP), procalcitonin, and stroke index (SI). b, Area under the curve for systemic vascular resistance index (SVRI). This is shown separately in (a) as values become more negative with sepsis. ROC = receiver operating characteristic.
Irrespective of whether sepsis was present or not, there were interesting results for BNP over the course of the burn. We had predicted that we would see high values of BNP on the first 2034
www.ccmjournal.org
2 days when the left ventricle was depressed and very large quantities of fluid were administered. However, BNP values were very low (66.2 ± 57.6/mL) at this time, and BNP rose when the heart September 2014 • Volume 42 • Number 9
Clinical Investigations
Table 5. Backward Conditional Binary Logistic Regression for Independent Factors to Predict Sepsis (Dependent Variable = SEPSIS) Ba
Wald χ2b
Significance
OR (Exp β)
95% CI for Exp β
–0.005
9.86
p = 0.002
0.99
0.992–0.998
Systemic vascular resistance index
0.29
3.46
p = 0.063
1.34
0.98–1.80
Stroke volume index
0.11
2.47
p = 0.12
1.12
0.97–1.28
Factor
B-Type natriuretic peptide
Model χ2c
194.42
OR = odds ratio, Exp β = predicted change in odds for unit increase in predictor. a B = regression coefficient. b This statistic tests the unique contribution of each predictor and eliminates overlap between predictors. c Model χ2 = how well the model predicts the outcome variable.
became hyperdynamic, that is, at 72 hours, fluid resuscitation was completed, and a diuresis was occurring. Papp et al (36) investigated BNP during fluid resuscitation in burns and found similar results. They claimed that the cessation of capillary leakage and increasing filling pressures lead to atrial and ventricular dilatation, showing an improved response to fluid therapy. PCT is a peptide precursor of the hormone calcitonin and rises in a response to a proinflammatory stimulus, especially of bacterial origin. In cases of general sepsis, it has high predictive ability (37). We did not find PCT useful as a marker of sepsis. High levels were related to the day of burn—the levels of PCT were significantly higher (Fig. 3b) on the first 2 days postburn and then decreased. Previous studies investigating PCT as a marker of sepsis in burn patients have had conflicting results (8–11, 38) but can be criticized in terms of utilization of the American College of Chest Physicians/Society of Critical Care Medicine guidelines, which was intended for a general ICU population (12). Only two studies (8, 38) have previously used the ABA Consensus Criteria guidelines (5). Reynolds et al (39) have recently reported that in general intensive care patients, PCT is high in the first few days after admission and has a strong association with shock independent of the presence of infection. This was similarly found in our study. A high sensitivity and specificity was also shown by SVRI. It has been shown (40) that the normal hemodynamic response in burns consists of high SVRI, that is, vasoconstriction and low SVI during the first 48–72 hours during active fluid resuscitation followed by a dramatic alteration with low SVRI, that is, vasodilatation and a high SVI. Our results illustrated this “ebb and flow response” (Fig. 3, c and d). In cases of proven sepsis on days 3–7, SVRI was significantly lower than in cases without sepsis. As low SVRI is a feature of sepsis, it is hypothesized that this further decrease in SVRI was due to a synergistic effect with potentially increased levels of BNP exerting a vasodilatory effect via nitric oxide (NO) production. An increased NO production in sepsis can have both beneficial and deleterious effects. Interestingly, SVRI was an item in the Baltimore Sepsis scale (41), a scoring system for sepsis in burns, which is now rarely used. SVV less than 12% with hypotension did not demonstrate adequate sensitivity or specificity for sepsis. Hypotension can be due to many circumstances in the burn patient, and Critical Care Medicine
distributive shock may have been due, on occasions, to medication as well as sepsis. On a number of occasions (n = 8) when the results of tissue diagnosis or blood cultures became available, according to our criteria infection was not present, yet antibiotics had been prescribed by the clinical staff. Conversely, there were also occasions when a tissue diagnosis or blood culture was positive yet that patient had not received antibiotics (n = 4). For this reason, the response to antibiotics was not recorded as a component of the diagnostic criteria of sepsis. The ABA consensus criteria (5) were used for diagnosis in this study. A recent article (42) has further investigated the validity for this scoring system against bacteremia. It was found that the triggers of heart rate, maximum temperature, thrombocytopenia, hyperglycemia, and feed intolerance were all significant in multivariate analysis. The article recommended that the search for further biomarkers be continued. The prospective markers that we have identified, that is, NT-proBNP, SVI, and SVRI, are easy to apply and occur early in the course of infection. BNP can be measured by point-of-care devices (43). Likewise, APCO devices are noninvasive and form part of routine care in most critical care units (44). However, it is important to have an appreciation of “normal” values at various stages for SVRI and SVI in burn patients in order to avoid an over diagnosis of sepsis. Limitations of this study included the fact that only the first episode of sepsis and the early stage of intensive care admission were recorded. Further work is required to elucidate normal values for BNP and hemodynamics for the entire duration of the intensive care stay.
CONCLUSIONS Diagnosing sepsis in patients post burn injury remains a major problem in critical care units. This study indicated that the variables of NT-proBNP and the hemodynamic values of SVRI and SVI may assist in providing a high suspicion of sepsis prior to results of the blood culture. Further work is required to confirm these findings.
REFERENCES
1. Mann EA, Baun MM, Meininger JC, et al: Comparison of mortality associated with sepsis in the burn, trauma, and general intensive www.ccmjournal.org
2035
Paratz et al care unit patient: A systematic review of the literature. Shock 2012; 37:4–16 2. D’Avignon LC, Hogan BK, Murray CK, et al: Contribution of bacterial and viral infections to attributable mortality in patients with severe burns: An autopsy series. Burns 2010; 36:773–779 3. Kumar A, Roberts D, Wood KE, et al: Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock. Crit Care Med 2006; 34:1589–1596 4. Levy MM, Fink MP, Marshall JC, et al: 2001 SCCM/ESICM/ACCP/ ATS/SIS International Sepsis Definitions Conference. Crit Care Med 2003; 31:1250–1256 5. Greenhalgh D, Saffle JR, Holmes JH, et al: American Burns Association consensus conference to define sepsis and infection in burns. J Burns Care Res 2007; 28:776–790 6. Lavrentieva A, Kontakiotis T, Lazaridis L, et al: Inflammatory markers in patients with severe burn injury. What is the best indicator of sepsis? Burns 2007; 33:189–194 7. Jeschke MG, Finnerty CC, Kulp GA, et al: Can we use C-reactive protein levels to predict severe infection or sepsis in severely burned patients? Int J Burns Trauma 2013; 3:137–143 8. Lavrentieva A, Papadopoulou S, Kioumis J, et al: PCT as a diagnostic and prognostic tool in burn patients. Whether time course has a role in monitoring sepsis treatment. Burns 2012; 38:356–363 9. von Heimburg D, Stieghorst W, Khorram-Sefat R, et al: Procalcitonin—A sepsis parameter in severe burn injuries. Burns 1998; 24:745–750 10. Barati M, Alinejad F, Bahar MA, et al: Comparison of WBC, ESR, CRP and PCT serum levels in septic and non-septic burn cases. Burns 2008; 34:770–774 11. Sachse C, Machens HG, Felmerer G, et al: Procalcitonin as a marker for the early diagnosis of severe infection after thermal injury. J Burn Care Rehabil 1999; 20:354–360 12. Mann EA, Wood GL, Wade CE: Use of procalcitonin for the detection of sepsis in the critically ill burn patient: A systematic review of the literature. Burns 2011; 37:549–558 13. Wang F, Wu Y, Tang L, et al: Brain natriuretic peptide for prediction of mortality in patients with sepsis: A systematic review and metaanalysis. Crit Care 2012; 16:R74 14. Shor R, Rozenman Y, Bolshinsky A, et al: BNP in septic patients without systolic myocardial dysfunction. Eur J Intern Med 2006; 17:536–540 15. Paratz J, Lipman J: BNP: Not just a heart throb. Anaesth Analg 2013; 1:59–61 16. Dulhunty JM, Boots RJ, Rudd MJ, et al: Increased fluid resuscitation can lead to adverse outcomes in major-burn injured patients, but low mortality is achievable. Burns 2008; 34:1090–1097 17. Vincent JL, de Mendonça A, Cantraine F, et al: Use of the SOFA score to assess the incidence of organ dysfunction/failure in intensive care units: Results of a multicenter, prospective study. Crit Care Med 1998; 26:1793–1800 18. Allgöwer M, Buri C: Shock index. Deutsche Medizinische Wodenschrift 1967; 46:1–10 19. Broch O, Renner J, Gruenewald M, et al: Comparison of thirdgeneration semi-invasive arterial waveform analysis with thermodilution in patients undergoing coronary surgery. Sci World J 2012; 2012:451081 20. Mutoh T, Ishikawa T, Kobayashi S, et al: Performance of third-generation FloTrac/Vigileo system during hyperdynamic therapy for delayed cerebral ischemia after subarachnoid hemorrhage. Surg Neurol Int 2012; 3:99 21. Lavrentieva A, Kontakiotis T, Kaimakamis E, et al: Evaluation of arterial waveform derived variables for an assessment of volume resuscitation in mechanically ventilated burn patients. Burns 2013; 39:249–254 22. Weinstein MP: Current blood culture methods and systems: Clinical concepts, technology, and interpretation of results. Clin Infect Dis 1996; 23:40–46
2036
www.ccmjournal.org
23. Field A: Discovering Statistics Using SPSS for Windows. Fourth Edition. London, Sage, 2003 24. Belgian Outcome in Burn Injury Study Group: Development and validation of a model for prediction of mortality in patients with acute burn injury. Br J Surg 2009; 96:111–117 25. Levin ER, Gardner DG, Samson WK: Natriuretic peptides. N Engl J Med 1998; 339:321–328 26. Doust JA, Pietrzak E, Dobson A, et al: How well does B-type natriuretic peptide predict death and cardiac events in patients with heart failure: Systematic review. BMJ 2005; 330:625 27. Januzzi JL, van Kimmenade R, Lainchbury J, et al: NT-proBNP testing for diagnosis and short-term prognosis in acute destabilized heart failure: An international pooled analysis of 1256 patients: The International Collaborative of NT-proBNP Study. Eur Heart J 2006; 27:330–337 28. Pirracchio R, Salem R, Mebazaa A: Use of B-type natriuretic peptide in critically ill patients. Biomark Med 2009; 3:541–547 29. Hoffmann U, Brueckmann M, Bertsch T, et al: Increased plasma levels of NT-proANP and NT-proBNP as markers of cardiac dysfunction in septic patients. Clin Lab 2005; 51:373–379 30. Rudiger A, Singer M: Mechanisms of sepsis-induced cardiac dysfunction. Crit Care Med 2007; 35:1599–1608 31. Court O, Kumar A, Parrillo JE, et al: Clinical review: Myocardial depression in sepsis and septic shock. Crit Care 2002; 6:500–508 32. Krishnagopalan S, Kumar A, Parrillo JE, et al: Myocardial dysfunction in the patient with sepsis. Curr Opin Crit Care 2002; 8:376–388 33. Pirracchio R, Deye N, Lukaszewicz AC, et al: Impaired plasma B-type natriuretic peptide clearance in human septic shock. Crit Care Med 2008; 36:2542–2546 34. Scharbag J, Herrman M, Urhausen A, et al: Independent elevations of N-terminal pro-brain natriuretic peptide and cardiac troponins in endurance athletes after prolonged strenuous exercise. Am Heart J 2005; 150:1128–1134 35. Vila G, Resl M, Stelzeneder D, et al: Plasma NT-proBNP increases in response to LPS administration in healthy men. J Appl Physiol (1985) 2008; 105:1741–1745 36. Papp A, Uusaro A, Parviainen I, et al: Myocardial function and haemodynamics in extensive burn trauma: Evaluation by clinical signs, invasive monitoring, echocardiography and cytokine concentrations. A prospective clinical study. Acta Anaesthesiol Scand 2003; 47:1257–1263 37. Castelli GP, Pognani C, Meisner M, et al: Procalcitonin and C-reactive protein during systemic inflammatory response syndrome, sepsis and organ dysfunction. Crit Care 2004; 8:R234–R242 38. Bognar Z, Foldi V, Rezman B, et al: Extravascular lung water index as a sign of developing sepsis in burns. Burns 2010; 36:1263–1270 39. Reynolds SC, Shorr AF, Muscedere J, et al: Longitudinal changes in procalcitonin in a heterogeneous group of critically ill patients. Crit Care Med 2012; 40:2781–2787 40. Holm C, Mayr M, Tegeler J, et al: A clinical randomized study on the effects of invasive monitoring on burn shock resuscitation. Burns 2004; 30:798–807 41. Meek M, Munster AM, Winchurch RA, et al: The Baltimore Sepsis Scale: Measurement of sepsis in patients with burns using a new scoring system. J Burn Care Rehabil 1991; 12:564–568 42. Hogan BK, Wolf SE, Hospenthal DR, et al: Correlation of American Burn Association Sepsis Criteria with the presence of bacteremia in burned patients admitted to the intensive care unit. J Burn Care Res 2012; 33:371–378 43. Ishida J, Suzuki T, Aizawa K, et al: Comparison of analytical performance of two single-step measurement devices of B-type natriuretic peptide. Int Heart J 2012; 53:320–323 44. Lavrentieva A, Palmieri T: Determination of cardiovascular parameters in burn patients using arterial waveform analysis: A review. Burns 2011; 37:196–202
September 2014 • Volume 42 • Number 9
Objectives: Accurate diagnosis of sepsis is difficult in patients post burn due to the large inflammatory response produced by the major insult. We aimed to estimate the values of serum N-terminal pro-Btype natriuretic peptide and procalcitonin and the changes in hemodynamic variables as markers of sepsis in critically ill burn patients. Design: Prospective, observational study. Setting: A quaternary-level university-affiliated ICU. *See also p. 2137. 1 Burns, Trauma and Critical Care Research Centre, The University of Queensland, Brisbane, QLD, Australia. 2 Department of Intensive Care Medicine, Royal Brisbane and Women’s Hospital, Brisbane, QLD, Australia. 3 Department of Rehabilitation Sciences, Griffith University, Brisbane, Australia. 4 Professor Stuart Pegg Burn Unit, Royal Brisbane and Women’s Hospital, Brisbane, QLD, Australia. 5 University of Queensland Centre for Clinical Research, Brisbane, QLD, Australia. Supported, in part, by Royal Brisbane and Women’s Hospital Research Foundation. Dr. Paratz consulted for Edwards Lifesciences; is employed by University of Queensland; and received support for travel from Australian and New Zealand Intensive Care Society, South African Burns Society, and Australian Physiotherapy Association. Her institution received grant support from National Health and Medical Research Council (NIHMRC), Office of Health Medical Research, NHMRC Equipment grants, Royal Brisbane Hospital Research Foundation, and Office Medical Research; received provision of equipment from Edwards Lifesciences (equipment lent free of cost to ICU); and lectured for University of Queensland, Australian Catholic University. Dr. Lipman’s institution received grant support from the Royal Brisbane and Women’s Hospital Research Foundation; received provision of equipment from Edwards Lifesciences (equipment lent free of cost to ICU); served as board member for Bayer European Society Council Intensive Care Medicine Advisory Board; consulted for Merk Sharp Dohme (Aust) Pty Ltd, Pfizer Australia, and AstraZeneca; received grant support from AstraZeneca; and lectured for AstraZeneca and Pfizer Australia Pty Ltd. Dr. Boots’ institution received grant support from Royal Brisbane and Women’s Hospital Research Foundation and received provision of equipment from Edwards Lifesciences (equipment lent free of cost to ICU). Dr. Muller’s institution received grant support from Royal Brisbane and Women’s Hospital Research Foundation, received provision of equipment from Edwards Lifesciences (equipment lent free of cost to ICU), and received grant support from Royal Brisbane and Women’s Hospital Research Foundation. Dr. Paterson consulted for Merck, AstraZeneca, and Pfizer; and lectured for AstraZeneca. For information regarding this article, E-mail: [email protected] Copyright © 2014 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: 10.1097/CCM.0000000000000400
Critical Care Medicine
Patients: Fifty-four patients with burns to total body surface area of greater than or equal to 15%, intubated with no previous cardiovascular comorbidities, were enrolled. Interventions: At admission, a FloTrac/Vigileo system was attached and daily blood samples taken from the arterial catheter. Infection surveillance was carried out daily with patients classified as septic/ nonseptic according to American Burns Consensus criteria. Measurements and Main Results: N-terminal pro-B-type natriuretic peptide, procalcitonin, and waveform analysis of changes in stroke volume index and systemic vascular resistance index were measured within the first 24 hours after burn and daily thereafter for the length of the ICU stay or until their first episode of sepsis. Prevalences of stroke volume variation less than 12% (normovolemia) with hypotension (systolic blood pressure < 90 mm Hg) were recorded. Patients with sepsis differed significantly from “no sepsis” for N-terminal pro-B-type natriuretic peptide, systemic vascular resistance index, and stroke volume index on days 3–7. Procalcitonin did not differ between sepsis and “no sepsis” except for day 3. Area under the receiver operating characteristic curves showed excellent discriminative power for B-type natriuretic peptide (p = 0.001; 95% CI, 0.99–1.00), systemic vascular resistance index (p < 0.001; 95% CI, 0.97– 0.99), and stroke volume index (p < 0.01; 95% CI, 0.96–0.99) in predicting sepsis but not for procalcitonin (not significant; 95% CI, 0.29–0.46). A chi-square crosstab found that there was no relationship between hypotension with normovolemia (stroke volume variation < 12%) and sepsis. Conclusions: Serum N-terminal pro-B-type natriuretic peptide levels and certain hemodynamic changes can be used as an early indicator of sepsis in patients with burn injury. Procalcitonin did not assist in the early diagnosis of sepsis. (Crit Care Med 2014; 42:2029–2036) Key Words: burns; natriuretic peptide, brain; procalcitonin; sepsis syndrome; systemic inflammatory response syndrome
D
espite an increased overall survival rate, severe infection remains the main cause of mortality in burn injury (1, 2). Prompt antimicrobial therapy for septic shock demonstrates a mortality benefit in general critical care patients (3); hence, an early diagnosis is important. However, diagnosing sepsis in burn patients has been associated with a number of difficulties. www.ccmjournal.org
2029
Paratz et al
The accepted methods to diagnose sepsis in general intensive care patients (4) are insensitive and nonspecific in burn patients due to the sustained systemic inflammatory state following a burn injury. In an attempt to improve accuracy of diagnosis, a consensus conference was held by the American Burns Association (ABA) in 2007 (5). Table 1 identifies both the triggers to search for infection and confirmation of infection in burn subjects suggested by this conference (5). However, there is inevitably a time delay as confirmation of results from tissue and blood cultures can take 48 hours. In a further attempt to find indicators of sepsis in burns, inflammatory markers have been investigated. Evidence for procalcitonin (PCT), erythrocyte sedimentation rate, C-reactive protein, and interleukin-6 is conflicting (6–11), with PCT the strongest contender (12). B-type natriuretic peptide (BNP) has been shown to be elevated in sepsis in nonburn patients independent of cardiac status (13–15) but has not previously been researched specifically in burn patients. Following increased use of hemodynamic monitoring arterial pressure–based cardiac output (APCO) device (Vigileo, FloTrac, Edwards Lifesciences, Irvine, CA) in our ICU, an audit of retrospective data indicated that stroke volume index Table 1. American Burn Association Criteria for Sepsis Sepsis should be considered when three or more of the following criteria are met: 1. Temperature: > 39°C or < 36.5°C 2. Progressive tachycardia: > 110 beats/min 3. Progressive tachypnea: a. > 25 breaths/min not ventilated b. Minute ventilation > 12 L/min ventilated 4. Thrombocytopenia (not applied until 3 d after initial resuscitation): < 100,000/μL 5. Hyperglycemia (in the absence of preexisting diabetes mellitus) a. Untreated plasma glucose > 200 mg/dL or equivalent mM/L
(SVI) increased and systemic vascular resistance index (SVRI) decreased above those figures expected for the “flow stage” of the burn hemodynamic response during proven infection. Because of the association of BNP with sepsis in general patients mentioned previously, we instituted a prospective study aiming to investigate the efficacy of serum BNP and PCT, changes in predicted values of SVI and SVRI, and prevalences of stroke volume variation (SVV) less than 12% (normovolemia) with hypotension (systolic blood pressure < 90 mm Hg) as criteria for an accurate expedited marker of suspicion of sepsis in burns.
MATERIALS AND METHODS This study was conducted in a quaternary-level universityaffiliated ICU. Approval was granted by both the Hospital and University Human Medical Ethics Committees and consent was obtained from the next of kin or legal guardian of every subject. Subjects Subjects to be included were those with greater than or equal to 15% TBSA burns who were intubated, with an indwelling arterial catheter, previous normal cardiac function, and not expected to be extubated within 72 hours. Subjects were excluded if they had toxic epidermal necrolysis, had a history of cardiovascular disease including coronary artery disease, cardiac failure, or congenital heart problems, had multiple trauma, took longer than 8 hours to reach the tertiary center, or were for compassionate care only. They were resuscitated using the modified Parkland formula (including judicious use of colloid) (16) aiming at the endpoints of urinary output 0.5–2 mL/kg/hr lean body mass, mean arterial pressure greater than 60 mm Hg, intra-abdominal pressure less than 25 mm Hg, hemoglobin less than 16 mg/dL and more than 8 mg/dL, and Na less than 155 mEq/L. Although a number of patients underwent part of their resuscitation in a smaller hospital or in transit, personnel were in contact with tertiary center staff regarding fluid resuscitation. All fluids received were recorded, and it was stated Yes/No whether patients had exceeded the Parkland formula. All patients received the usual standard of care:
b. > 7 U of insulin/hr IV drip c. Significant resistance to insulin (> 25% increase in insulin requirement over 24 hr)
6. Inability to continue enteral feedings > 24 hr a. Abdominal distension
b. High gastric residuals (residuals two times feeding rate) c. Uncontrollable diarrhea (> 2,500 mL/d) Documented infection when any of these criteria are met: 1. Positive culture 2. Pathologic tissue source 3. Clinical response to antimicrobials Reproduced with permission from Greenhalgh et al (5).
2030
www.ccmjournal.org
• Mechanical ventilation initially using the modes of synchronized intermittent mandatory ventilation or BiLevel initially, followed by pressure support ventilation with positive end-expiratory pressure or inspired oxygen as the patient’s sedation requirements lessened. • Inhalation injury diagnosed by bronchoscopy for those suspected of a lung injury. • Sedation and analgesia administered according to routine guidelines, including a mixture of fentanyl, k etamine, morphine, and midazolam. • Escharotomies as required. • Early debridement and grafting with autografts, allografts, or both as soon as initial resuscitation is complete.
Intensive care clinical staff were blinded to FloTrac results (i.e., cardiac output, systemic vascular resistance, and SVV) but accessed heart rate, arterial blood pressure, arterial blood gases, urine output, fluid balance, electrolytes, lactate, and central September 2014 • Volume 42 • Number 9
Clinical Investigations
venous pressure (i.e., the usual variables to provide resuscitation and management in burn patients). All usual demographics were recorded and are listed in Table 1. Hemodynamic Indices An APCO device (Vigileo, FloTrac, Edwards Lifesciences) was attached to an existing peripheral artery catheter to record hemodynamic variables. This third-generation model of semiinvasive machine has shown good correlation with more invasive methods and is capable of tracking hemodynamic changes and trends (19, 20). We considered utilization of the Vigileo to be more acceptable from an ethical point of view than the insertion of a pulmonary artery catheter. SVI, SVRI, and SVV were downloaded continuously, together with all other observations, that is, arterial blood pressure, heart rate, respiratory rate, Spo2, and central venous pressure, to the Integrated Clinical Information Program (ICIP) (Phillips Healthcare, Andover, MA). A program using purpose-written software automatically calculated the daily highest, lowest, mean, and median for all these variables. The variables are described as percentage of normal SI and SVRI for age, height and gender, and prevalence of hypotension (systolic blood pressure < 90 mm Hg) with normovolemia (i.e., SVV < 12%). SVV% is a measure that provides information on relative preload responsiveness. SVV is calculated as percentage change between the maximal and minimal stroke volumes divided by the average of the minimum and maximum over a respiratory cycle or other period of time. This measure can be useful in differentiating the cause of hypotension between hypovolemia or a distributive cause and was therefore included in the outcome measures. SVV was recorded only when the subject was on full mechanical ventilation with a tidal volume of greater than 8 mL/kg and no arrhythmias. The SVV at which the burn patient is fluid responsive has been shown to be 13.6% (21); therefore, the figure of less than 12% was chosen to ensure normovolemia. Blood Samples Blood samples were taken once daily from indwelling arterial catheters for measurements of PCT and BNP. They were then centrifuged, separated, and plasma stored at –80°C and analyzed in bulk at the end of the study. Serum levels of the inactive cleavage product of BNP, N-terminal fragment (NT-proBNP), were determined by electrochemiluminescence immunoassay (ECLIA) (Elecsys proBNP II Assay, Cobas, Henningsdorf, Germany). The coefficient of variation was 2%, with a lower limit of detection of 5 pg/mL. NT-proBNP was measured as it has higher plasma levels and a longer half-life and, therefore, slower fluctuations than BNP, but NT-proBNP will be referred to by the term “BNP” throughout the remainder of the article. PCT was measured by ECLIA (Elecsys BRAHMS PCT, Cobas), with a coefficient of variation of 2% and a lower detection rate of less than or equal to 0.02 ng/mL. Diagnosis of Sepsis The following criteria by the research staff were independent and additional to infection surveillance by the clinical staff. For the purposes of diagnosis of sepsis in this study, all triggers Critical Care Medicine
(heart rate, temperature, and respiratory rate [if not fully sedated and fully ventilated], platelets, prevalence of feeding intolerance, and insulin requirements) as per Table 1 were continuously screened by the researchers. If these variables indicated infection, tissue diagnosis and blood cultures were requested if the clinical staff had not already ordered these. Blood cultures consisted of three separate sets from separate venipuncture sites with a minimum of 20-mL volume (22). To satisfy the criteria for sepsis, three or more criteria were presented in Table 1, with either a positive tissue or blood culture diagnosis. We purposely did not use the criteria of response to antimicrobial therapy in our diagnosis. Additionally, the organ suspected of being the source of infection was required to satisfy the modified criteria for Centers for Disease Control and Prevention as per the ABA (5). When all results were available, the patients were scored as Yes/No for the presence of sepsis for each day. Clinical staff had the responsibility for prescription of antibiotics, based on their clinical judgment of whether an infection was present. Statistics The analyses were performed using SPSS 19.0 software (SPSS, Chicago, IL). Patients satisfied the criteria for sepsis on a Yes/ No basis, with this criteria applied every day. Overall values for BNP, PCT, SVI, and SVRI were compared between patients described as septic/nonseptic for each day using a one-way analysis of variance for time. As group allocation, that is, whether sepsis criteria were met, altered each day, a t test compared factors between sepsis and nonsepsis for each day. A chisquare crosstab investigated whether BNP levels were related to exceeding the recommended Parkland formula. A chi-square crosstab investigated whether significantly more patients with SVV less than 12% and systolic arterial blood pressure less than 90 mm Hg satisfied the criteria for sepsis. The sensitivity, specificity, and positive likelihood ratio for BNP, PCT, SVI, and SVRI in diagnosing sepsis were calculated at the best cutoff value (Table 2) To determine the predictive ability of these four factors, receiver operating characteristic (ROC) curves were constructed and the areas under the curve were calculated with 95% CIs. Factors were then entered into a binary logistic regression to detect which factors were independent markers of sepsis (23). A p value of less than 0.05 was considered to be statistically significant in all tests.
RESULTS Fifty-four patients were included in the study, with clinical characteristics included in Table 3. Figure 1 indicates the flow of patients and reasons for exclusion. Although it may appear unusual that subjects with less than or equal to 20% burns were admitted to intensive care, all these subjects had inhalation or airway injuries, necessitating an endotracheal tube. There were 23 of 54 patients with sepsis confirmed. Day refers to the actual day post burn not the day of admission to the tertiary burn unit. Only the first episode of sepsis for each patient was recorded for the purposes of this study. Following resolution of this first episode, no further pathology tests or hemodynamic www.ccmjournal.org
2031
Paratz et al
Table 2. Sensitivity, Specificity, Positive Likelihood Ratios, Negative Likelihood Ratios, and Diagnostic Odds Ratio for Each Factor as a Marker of Sepsis—For the First Episode of Sepsis Factor
B-Type Natriuretic Peptide
Procalcitonin
Stroke Volume Index
Systemic Vascular Resistance Index
Sensitivity
95.9% (86.2–99.1%)
20.2% (12.1–34.3%)
94.3% (84.2–98.4%)
86.2% (73.3–93.4%)
Specificity
99.4% (98.1–99.8%)
64.3% (59.1–69.2%)
94.2% (91.3–96.5%)
97.1% (94.6–98.2%)
Positive likelihood ratios
165.5 (41.5–659.7)
0.57 (0.321–1.01)
Negative likelihood ratios
40.04 (0.01–0.16)
1.24 (1.06–1.46)
0.07 (0.02–0.19)
0.15 (0.08–0.29)
0.53 (0.23–1.22)
0.46 (0.22–0.94
166.5 (62.1–445.2)
Diagnostic odds ratio
4030.1 (554.5–29293.3)
14.68 (9.47–22.1)
24.6 (13.97–43.47)
Cutoff values are taken as B-type natriuretic peptide > 1,200 (pg/mL); procalcitonin 1.4 (ng/mL); stroke volume index ≥ 160% of predicted for age, height, and gender; systemic vascular resistance index ≤ 50% of predicted for age, height, and gender.
Table 3.
Demographics and Patient Details for Patients Both With and Without Sepsis
Factor
Sepsis (n = 23)
Age (yr)a
42.65 (12.8)
No Sepsis (n = 31)
37.68 (10.4)
All Subjects (n = 54)
40.16 (18–60)
16:7
29:2
45:9
TBSA %a
46.2 (15.3)
42.0 (16.7)
44.1 (16.0)
Total body surface area full thickness %b
21.8 (11.0)
20.8 (7.4)
Baux indexa
96.1 (18.6)
89.8 (18.5)
93.0 (18.5)
17 (74%)
7 (23%)
24 (44%)
9 (8–11)
9 (7–11)
Gender, male:female
b
Inhalation injurya
9 (7–11)
Admission Sequential Organ Failure Assessment (17)c
1.1 (0.9–1.5)
Admission shock index (18)c Severity of burn injury injury (Belgian Outcome in Burn Injury) (24)c LOS (ICU) (d)a LOS (hospital) (d)
a
Mortality
3 (1–6)
20.86(9.2)
1.1 (0.85–1.5) 3 (1–6)
1.1 (0.85–1.5) 3 (1–6)
21.2 (4.6)
17.0 (5.4)
19.1 (5.0)
42.9 (7.6)
37.7 (9.4)
40.8 (8.5)
1
0
1
Depression
5
2
7
Asthma
3
2
5
Hepatitis C
2
2
4
Comorbidities
Total volume (0–24 hr) (mL/kg/%TBSA)
9.65 (3.2)
Total % of ICU stay on vasoactive drugs
22%
15%
17.5%
Abdominal hypertension
2
1
3
Escharotomy/fasciotomy
4
5
9
No. of grafting procedures/patient
6.5 (3.2)
5.1 (2.9)
5.5 (3.0)
Time to first excision (d)
1.9 (0.8)
1.8 (0.7)
1.85 (0.7)
b
9.88 (3.7)
9.77 (3.5)
TBSA = total body surface area, LOS = length of stay. a The subjects who developed sepsis were significantly different at the p < 0.01 level. b The subjects who developed sepsis were significantly different at the p < 0.05 level. c Median (range). Values are mean and sd or number and percentage unless otherwise specified.
2032
www.ccmjournal.org
September 2014 • Volume 42 • Number 9
Clinical Investigations
(p < 0.001) and SVI significantly lower (p < 0.001) in the first 2 days post burn compared to days 3–7. However, those patients who reached the criteria for sepsis had a significantly lower SVRI (p < 0.001) and higher SVI (p < 0.001) on days 3–5 than patients who did not reach the criteria. BNP was significantly lower on days 1 and 2 (p < 0.001) compared with days 3–7 in all patients. However, in those who reached the sepsis criteria, BNP was significantly higher from days 3 to 7. There was no correlation between levels of BNP and exceeding the recommended Parkland formula in the first 48 hours. A chi-square crosstab indicated that there was no relationship between hypotension with normovolemia (SVV < 12%) and prevalence of sepsis. The overall sensitivity and specificity for each factor are described in Table 2. Optimum cutoff values are listed in Table 2. A receiver operating characteristic (ROC) curve was constructed for the factors and is illustrated in Figure 4. Logistic regression models were constructed to investigate the association between the potential markers and sepsis. First, we investigated each factor individually and then included all significant factors (BNP, SVRI, and SVI) in a multivariable model with adjustment for severity of burn (24). These factors were entered into a binary logistic regression using the backward conditional method, with sepsis (Yes = 1, No = 0) as the dependent variable. The overall model was significant at the p less than 0.001 level according to the model c2 statistic (Table 2). The model predicted 99.2% of the responses correctly. The overall result was that BNP was an independent marker of sepsis (Table 5).
Figure 1. Flowchart of patients through the study.
Figure 2. Source of infection for the septic episode.
DISCUSSION
Most Commonly Isolated Bacterial Pathogens Table 4.
Bacteria
Gram negative
No. of Pathogens Isolated From Positive Blood Cultures or Tissue Culture
21
Pseudomonas aeruginosa
8
Klebsiella species
6
Acinetobacter baumannii
2
Serratia marcescens
1
Escherichia coli
2
Gram positive
3
Enterococcus species
1
Staphylococcus aureus
2
data were recorded for the study. However, outcome data, for example, length of stay and mortality, were recorded. Figure 2 indicates sites of infection and Table 4 organisms identified. The mean level for each day for patients with episodes of sepsis and no sepsis is illustrated for BNP (Fig. 3A), PCT (Fig. 3B), SVRI (Fig. 3C), and SVI (Fig. 3D). SVRI was significantly higher Critical Care Medicine
Early, accurate diagnosis of sepsis is vital in thermal injury both for a mortality advantage potentially achieved by timely initiation of empiric antibiotics and to decrease inappropriate antibiotic use increasing the risk of antimicrobial resistance. This study found a high degree of both positive and negative predictive ability for BNP and the hemodynamic values of stroke volume index (SI) and SVRI but not for PCT or hypotension with normovolemia. BNP is a well-known biomarker of cardiac failure (25). High levels have been well documented and accepted to differentiate dyspnea in the emergency setting, specify levels of severity, and provide prognosis in cardiac failure (26, 27). However, increases in BNP have been well documented in other clinical conditions including sepsis (14, 28, 29). At first, this was attributed to cardiac failure in conjunction with sepsis (29) and then later to the accompanying myocardial depression that occurs in 50% of patients with septic shock (29) considered to be functional rather than structural (30–32) with adequate perfusion of coronary arteries. More recently, investigators (33) have claimed that high levels in sepsis are related to an alteration in the BNP clearance pathway specifically the neutral endopeptidase 24.11 with a parallel increase in BNP. It is being increasingly recognized that elevated levels of BNP are present in subjects without myocardial dysfunction including marathon runners at the completion of a race (34) and healthy volunteers given an infusion of LPS to induce sepsis (35). www.ccmjournal.org
2033
Paratz et al
Figure 3. N-terminal pro-B-type natriuretic peptide (NT-proBNP) % of predicted for days 1–7 (a), procalcitonin (PCT) over days 1–7 (b), systemic vascular resistance index (SVRI) % of predicted for days 1–7 (c), and stroke volume index (SVI) % of predicted for days 1–7 (d). Number of patients with sepsis (S) and those not reaching the criteria for sepsis (NS) differed each day, apart from days 1 and 2 on which no subjects had sepsis. Number of patients was as follows: day 1 (NS) n = 54, day 2 (NS) n = 54, day 3 (S) n = 4 and (NS) n = 50, day 4 (S) n = 11 and (NS) n = 43, day 5 (S) n = 16 and (NS) n = 37, day 6 (S) n = 12 and (NS) n = 31, and day 7 (S) n = 5 and (NS) n = 28. * indicates that on that particular day patients reaching sepsis criteria were significantly different from those not reaching sepsis criteria: *p < 0.05, ** p < 0.01.
Figure 4. a, Area under the curve for B-type natriuretic peptide (BNP), procalcitonin, and stroke index (SI). b, Area under the curve for systemic vascular resistance index (SVRI). This is shown separately in (a) as values become more negative with sepsis. ROC = receiver operating characteristic.
Irrespective of whether sepsis was present or not, there were interesting results for BNP over the course of the burn. We had predicted that we would see high values of BNP on the first 2034
www.ccmjournal.org
2 days when the left ventricle was depressed and very large quantities of fluid were administered. However, BNP values were very low (66.2 ± 57.6/mL) at this time, and BNP rose when the heart September 2014 • Volume 42 • Number 9
Clinical Investigations
Table 5. Backward Conditional Binary Logistic Regression for Independent Factors to Predict Sepsis (Dependent Variable = SEPSIS) Ba
Wald χ2b
Significance
OR (Exp β)
95% CI for Exp β
–0.005
9.86
p = 0.002
0.99
0.992–0.998
Systemic vascular resistance index
0.29
3.46
p = 0.063
1.34
0.98–1.80
Stroke volume index
0.11
2.47
p = 0.12
1.12
0.97–1.28
Factor
B-Type natriuretic peptide
Model χ2c
194.42
OR = odds ratio, Exp β = predicted change in odds for unit increase in predictor. a B = regression coefficient. b This statistic tests the unique contribution of each predictor and eliminates overlap between predictors. c Model χ2 = how well the model predicts the outcome variable.
became hyperdynamic, that is, at 72 hours, fluid resuscitation was completed, and a diuresis was occurring. Papp et al (36) investigated BNP during fluid resuscitation in burns and found similar results. They claimed that the cessation of capillary leakage and increasing filling pressures lead to atrial and ventricular dilatation, showing an improved response to fluid therapy. PCT is a peptide precursor of the hormone calcitonin and rises in a response to a proinflammatory stimulus, especially of bacterial origin. In cases of general sepsis, it has high predictive ability (37). We did not find PCT useful as a marker of sepsis. High levels were related to the day of burn—the levels of PCT were significantly higher (Fig. 3b) on the first 2 days postburn and then decreased. Previous studies investigating PCT as a marker of sepsis in burn patients have had conflicting results (8–11, 38) but can be criticized in terms of utilization of the American College of Chest Physicians/Society of Critical Care Medicine guidelines, which was intended for a general ICU population (12). Only two studies (8, 38) have previously used the ABA Consensus Criteria guidelines (5). Reynolds et al (39) have recently reported that in general intensive care patients, PCT is high in the first few days after admission and has a strong association with shock independent of the presence of infection. This was similarly found in our study. A high sensitivity and specificity was also shown by SVRI. It has been shown (40) that the normal hemodynamic response in burns consists of high SVRI, that is, vasoconstriction and low SVI during the first 48–72 hours during active fluid resuscitation followed by a dramatic alteration with low SVRI, that is, vasodilatation and a high SVI. Our results illustrated this “ebb and flow response” (Fig. 3, c and d). In cases of proven sepsis on days 3–7, SVRI was significantly lower than in cases without sepsis. As low SVRI is a feature of sepsis, it is hypothesized that this further decrease in SVRI was due to a synergistic effect with potentially increased levels of BNP exerting a vasodilatory effect via nitric oxide (NO) production. An increased NO production in sepsis can have both beneficial and deleterious effects. Interestingly, SVRI was an item in the Baltimore Sepsis scale (41), a scoring system for sepsis in burns, which is now rarely used. SVV less than 12% with hypotension did not demonstrate adequate sensitivity or specificity for sepsis. Hypotension can be due to many circumstances in the burn patient, and Critical Care Medicine
distributive shock may have been due, on occasions, to medication as well as sepsis. On a number of occasions (n = 8) when the results of tissue diagnosis or blood cultures became available, according to our criteria infection was not present, yet antibiotics had been prescribed by the clinical staff. Conversely, there were also occasions when a tissue diagnosis or blood culture was positive yet that patient had not received antibiotics (n = 4). For this reason, the response to antibiotics was not recorded as a component of the diagnostic criteria of sepsis. The ABA consensus criteria (5) were used for diagnosis in this study. A recent article (42) has further investigated the validity for this scoring system against bacteremia. It was found that the triggers of heart rate, maximum temperature, thrombocytopenia, hyperglycemia, and feed intolerance were all significant in multivariate analysis. The article recommended that the search for further biomarkers be continued. The prospective markers that we have identified, that is, NT-proBNP, SVI, and SVRI, are easy to apply and occur early in the course of infection. BNP can be measured by point-of-care devices (43). Likewise, APCO devices are noninvasive and form part of routine care in most critical care units (44). However, it is important to have an appreciation of “normal” values at various stages for SVRI and SVI in burn patients in order to avoid an over diagnosis of sepsis. Limitations of this study included the fact that only the first episode of sepsis and the early stage of intensive care admission were recorded. Further work is required to elucidate normal values for BNP and hemodynamics for the entire duration of the intensive care stay.
CONCLUSIONS Diagnosing sepsis in patients post burn injury remains a major problem in critical care units. This study indicated that the variables of NT-proBNP and the hemodynamic values of SVRI and SVI may assist in providing a high suspicion of sepsis prior to results of the blood culture. Further work is required to confirm these findings.
REFERENCES
1. Mann EA, Baun MM, Meininger JC, et al: Comparison of mortality associated with sepsis in the burn, trauma, and general intensive www.ccmjournal.org
2035
Paratz et al care unit patient: A systematic review of the literature. Shock 2012; 37:4–16 2. D’Avignon LC, Hogan BK, Murray CK, et al: Contribution of bacterial and viral infections to attributable mortality in patients with severe burns: An autopsy series. Burns 2010; 36:773–779 3. Kumar A, Roberts D, Wood KE, et al: Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock. Crit Care Med 2006; 34:1589–1596 4. Levy MM, Fink MP, Marshall JC, et al: 2001 SCCM/ESICM/ACCP/ ATS/SIS International Sepsis Definitions Conference. Crit Care Med 2003; 31:1250–1256 5. Greenhalgh D, Saffle JR, Holmes JH, et al: American Burns Association consensus conference to define sepsis and infection in burns. J Burns Care Res 2007; 28:776–790 6. Lavrentieva A, Kontakiotis T, Lazaridis L, et al: Inflammatory markers in patients with severe burn injury. What is the best indicator of sepsis? Burns 2007; 33:189–194 7. Jeschke MG, Finnerty CC, Kulp GA, et al: Can we use C-reactive protein levels to predict severe infection or sepsis in severely burned patients? Int J Burns Trauma 2013; 3:137–143 8. Lavrentieva A, Papadopoulou S, Kioumis J, et al: PCT as a diagnostic and prognostic tool in burn patients. Whether time course has a role in monitoring sepsis treatment. Burns 2012; 38:356–363 9. von Heimburg D, Stieghorst W, Khorram-Sefat R, et al: Procalcitonin—A sepsis parameter in severe burn injuries. Burns 1998; 24:745–750 10. Barati M, Alinejad F, Bahar MA, et al: Comparison of WBC, ESR, CRP and PCT serum levels in septic and non-septic burn cases. Burns 2008; 34:770–774 11. Sachse C, Machens HG, Felmerer G, et al: Procalcitonin as a marker for the early diagnosis of severe infection after thermal injury. J Burn Care Rehabil 1999; 20:354–360 12. Mann EA, Wood GL, Wade CE: Use of procalcitonin for the detection of sepsis in the critically ill burn patient: A systematic review of the literature. Burns 2011; 37:549–558 13. Wang F, Wu Y, Tang L, et al: Brain natriuretic peptide for prediction of mortality in patients with sepsis: A systematic review and metaanalysis. Crit Care 2012; 16:R74 14. Shor R, Rozenman Y, Bolshinsky A, et al: BNP in septic patients without systolic myocardial dysfunction. Eur J Intern Med 2006; 17:536–540 15. Paratz J, Lipman J: BNP: Not just a heart throb. Anaesth Analg 2013; 1:59–61 16. Dulhunty JM, Boots RJ, Rudd MJ, et al: Increased fluid resuscitation can lead to adverse outcomes in major-burn injured patients, but low mortality is achievable. Burns 2008; 34:1090–1097 17. Vincent JL, de Mendonça A, Cantraine F, et al: Use of the SOFA score to assess the incidence of organ dysfunction/failure in intensive care units: Results of a multicenter, prospective study. Crit Care Med 1998; 26:1793–1800 18. Allgöwer M, Buri C: Shock index. Deutsche Medizinische Wodenschrift 1967; 46:1–10 19. Broch O, Renner J, Gruenewald M, et al: Comparison of thirdgeneration semi-invasive arterial waveform analysis with thermodilution in patients undergoing coronary surgery. Sci World J 2012; 2012:451081 20. Mutoh T, Ishikawa T, Kobayashi S, et al: Performance of third-generation FloTrac/Vigileo system during hyperdynamic therapy for delayed cerebral ischemia after subarachnoid hemorrhage. Surg Neurol Int 2012; 3:99 21. Lavrentieva A, Kontakiotis T, Kaimakamis E, et al: Evaluation of arterial waveform derived variables for an assessment of volume resuscitation in mechanically ventilated burn patients. Burns 2013; 39:249–254 22. Weinstein MP: Current blood culture methods and systems: Clinical concepts, technology, and interpretation of results. Clin Infect Dis 1996; 23:40–46
2036
www.ccmjournal.org
23. Field A: Discovering Statistics Using SPSS for Windows. Fourth Edition. London, Sage, 2003 24. Belgian Outcome in Burn Injury Study Group: Development and validation of a model for prediction of mortality in patients with acute burn injury. Br J Surg 2009; 96:111–117 25. Levin ER, Gardner DG, Samson WK: Natriuretic peptides. N Engl J Med 1998; 339:321–328 26. Doust JA, Pietrzak E, Dobson A, et al: How well does B-type natriuretic peptide predict death and cardiac events in patients with heart failure: Systematic review. BMJ 2005; 330:625 27. Januzzi JL, van Kimmenade R, Lainchbury J, et al: NT-proBNP testing for diagnosis and short-term prognosis in acute destabilized heart failure: An international pooled analysis of 1256 patients: The International Collaborative of NT-proBNP Study. Eur Heart J 2006; 27:330–337 28. Pirracchio R, Salem R, Mebazaa A: Use of B-type natriuretic peptide in critically ill patients. Biomark Med 2009; 3:541–547 29. Hoffmann U, Brueckmann M, Bertsch T, et al: Increased plasma levels of NT-proANP and NT-proBNP as markers of cardiac dysfunction in septic patients. Clin Lab 2005; 51:373–379 30. Rudiger A, Singer M: Mechanisms of sepsis-induced cardiac dysfunction. Crit Care Med 2007; 35:1599–1608 31. Court O, Kumar A, Parrillo JE, et al: Clinical review: Myocardial depression in sepsis and septic shock. Crit Care 2002; 6:500–508 32. Krishnagopalan S, Kumar A, Parrillo JE, et al: Myocardial dysfunction in the patient with sepsis. Curr Opin Crit Care 2002; 8:376–388 33. Pirracchio R, Deye N, Lukaszewicz AC, et al: Impaired plasma B-type natriuretic peptide clearance in human septic shock. Crit Care Med 2008; 36:2542–2546 34. Scharbag J, Herrman M, Urhausen A, et al: Independent elevations of N-terminal pro-brain natriuretic peptide and cardiac troponins in endurance athletes after prolonged strenuous exercise. Am Heart J 2005; 150:1128–1134 35. Vila G, Resl M, Stelzeneder D, et al: Plasma NT-proBNP increases in response to LPS administration in healthy men. J Appl Physiol (1985) 2008; 105:1741–1745 36. Papp A, Uusaro A, Parviainen I, et al: Myocardial function and haemodynamics in extensive burn trauma: Evaluation by clinical signs, invasive monitoring, echocardiography and cytokine concentrations. A prospective clinical study. Acta Anaesthesiol Scand 2003; 47:1257–1263 37. Castelli GP, Pognani C, Meisner M, et al: Procalcitonin and C-reactive protein during systemic inflammatory response syndrome, sepsis and organ dysfunction. Crit Care 2004; 8:R234–R242 38. Bognar Z, Foldi V, Rezman B, et al: Extravascular lung water index as a sign of developing sepsis in burns. Burns 2010; 36:1263–1270 39. Reynolds SC, Shorr AF, Muscedere J, et al: Longitudinal changes in procalcitonin in a heterogeneous group of critically ill patients. Crit Care Med 2012; 40:2781–2787 40. Holm C, Mayr M, Tegeler J, et al: A clinical randomized study on the effects of invasive monitoring on burn shock resuscitation. Burns 2004; 30:798–807 41. Meek M, Munster AM, Winchurch RA, et al: The Baltimore Sepsis Scale: Measurement of sepsis in patients with burns using a new scoring system. J Burn Care Rehabil 1991; 12:564–568 42. Hogan BK, Wolf SE, Hospenthal DR, et al: Correlation of American Burn Association Sepsis Criteria with the presence of bacteremia in burned patients admitted to the intensive care unit. J Burn Care Res 2012; 33:371–378 43. Ishida J, Suzuki T, Aizawa K, et al: Comparison of analytical performance of two single-step measurement devices of B-type natriuretic peptide. Int Heart J 2012; 53:320–323 44. Lavrentieva A, Palmieri T: Determination of cardiovascular parameters in burn patients using arterial waveform analysis: A review. Burns 2011; 37:196–202
September 2014 • Volume 42 • Number 9
