OBJECTIVE. Early aggressive resuscitation is accepted best practice for severe pediatric sepsis. Targeting of therapy to individual hemodynamic patterns is recommended, but assessment of patterns is difficult early in the disease process. New technologies enabling earlier hemodynamic assessment in shock may inform choices for vasoactive drugs in fluid-resistant cases.
METHODS. This was a prospective observational study of 30 children with suspected fluid-resistant septic shock (minimum: 40 mL/kg) admitted to the PICU of a tertiary care children's hospital between July 2004 and July 2005. Children were classified according to admission diagnosis (community-acquired sepsis or central venous catheter-associated infection) and assessed within 4 hours after the onset of shock with a noninvasive cardiac output device. Cardiac index and systemic vascular resistance index were measured for all patients. Central venous oxygen saturation was measured for patients with accessible central venous lines at the time of hemodynamic measurements (typically at the superior vena cava-right atrium junction).
RESULTS. Fluid-resistant septic shock secondary to central venous catheter-associated infection was typically “warm shock” (15 of 16 patients; 94%), with high cardiac index and low systemic vascular resistance index. In contrast, this pattern was rarely seen in community-acquired sepsis (2 of 14 patients; 14%), where a normal or low cardiac index was predominant.
CONCLUSIONS. The hemodynamic patterns of fluid-resistant septic shock by the time children present to the PICU are distinct, depending on cause, with little overlap. If these findings can be reproduced, then targeting the choice of first-line vasoactive infusions in fluid-resistant shock (vasopressors for central venous catheter-associated infections and inotropes for community-acquired sepsis) should be considered.
Current guidelines for the care of children with septic shock emphasize the importance of early aggressive fluid resuscitation and titration of subsequent therapy to the hemodynamic pattern for the individual patient (Fig 1). 1 The merit of this early/targeted resuscitation approach is supported by randomized trial data for adults2 and retrospective case series for children.3
Pediatric intensivists frequently describe cases as “warm shock” or “cold shock” on the basis of clinical examination results alone, although this has been shown to be prone to error.4 The lack of more-detailed data for children reflects the difficulties in determining cardiac output (CO) and systemic vascular resistance data early in the clinical course of septic shock. Although the changing hemodynamic patterns seen in pediatric sepsis have been documented,5 the invasive monitoring of CO is not common practice in the pediatric emergency department. Indeed, the interval from first presentation to CO measurement with these techniques typically exceeds the 60-minute time frame recommended in the American College of Critical Care Medicine guidelines for both fluid resuscitation and selection of first- and second-line vasoactive drugs.1 It also exceeds the optimal time frame advocated for early, goal-directed resuscitation in the emergency department.5 Newer, noninvasive, ultrasound techniques offer the opportunity to address such hemodynamic questions at a much earlier stage and in children who have not yet experienced deterioration to the point of requiring the invasive lines or enteral probes that were required previously to measure hemodynamic variables such as CO. Furthermore, the techniques used here allow regular hemodynamic assessment of fully conscious children with sepsis without formal echocardiography.
We hypothesized that the hemodynamic patterns of pediatric septic shock on arrival in the PICU might differ according to cause, with children with central venous catheter (CVC) infections having a different pattern of septic shock, compared with that described by Ceneviva et al.6 Our clinical experience led us to suspect that a much greater proportion of CVC infections would manifest the adult-pattern hemodynamics of warm shock, with vasodilation (low systemic vascular resistance index [SVRI]) and high cardiac index. The null hypothesis for the study was that there would be no difference in cardiac index and SVRI at presentation to the PICU for patients with CVC infection, compared with community-acquired (CA) sepsis.
This was an unblinded observational study of hemodynamics in children with early septic shock. The Central Office for Research Ethics Committees (London, England) advised that, because this was a standard monitoring technique (Doppler ultrasonography) and no interventions were planned, ethical review was not necessary. Informed consent was not required, but all parents received a full explanation of the study.
Consecutive admissions to our tertiary multidisciplinary PICU, between postnatal age of 6 months and age of 17 years, with a presumed diagnosis of fluid-resistant septic shock were assessed for inclusion in the study. Shock was defined according to the International Consensus Conference on Pediatric Sepsis criteria for cardiovascular dysfunction.7 These criteria are that, despite administration of >40 mL/kg isotonic fluid in 1 hour, there is hypotension (<5th percentile for age or systolic blood pressure <2 SD below normal for age), the need for a vasoactive drug to maintain blood pressure in the normal range, or ≥2 of the following: unexplained metabolic acidosis of >5 mEq/L, increased arterial lactate level >2 times the upper limit of the normal range, oliguria (urine output of <0.5 mL/kg per hour), prolonged capillary refill (>5 seconds), or core-peripheral temperature gap of >3°C.
Another inclusion criterion was that hemodynamic assessment with a noninvasive device, the ultrasound cardiac output monitor (USCOM; Beaver Medical, Northampton, England), was performed by a single operator (Dr Brierley) within 4 hours after the onset of septic shock. A single operator was selected to minimize operator variability because, although reliable readings can be obtained after ∼20 USCOM readings,8 Dr Brierley was the only formally trained and assessed operator throughout the time of study. Exclusions included patients already in the PICU developing sepsis, recurrent sepsis (eg, incompletely resolved or continuing episode of sepsis), repeat PICU admission during the study period, and hemodynamic assessment at >4 hours.
Admission demographic data, time of diagnosis of shock, acute and underlying diagnosis, first-contact pediatric index of mortality score, and pediatric logistic organ dysfunction scores were recorded in addition to hemodynamic variables. An admission inotrope score was calculated as described by Wernovsky et al9 and modified by Skippen and Krahn.10 We added vasopressin to this score, with vasopressin at 0.0003 to 0.002 U per kg per minute (0.3–2 mU per kg per minute) being equivalent to 0.3 to 2 μg/kg per minute norepinephrine. This is used to quantify inotropic and vasopressor support and is calculated as the sum of all vasoactive agents corrected for potency, as follows: 1 × dopamine (μg/kg per minute) + 1 × dobutamine + 15 × milrinone + 100 × epinephrine + 100 × norepinephrine + 1000 × vasopressin (expressed as U/kg per minute).
Doppler ultrasonography is a simple noninvasive method of assessing blood flow, as described by Light,11 and the Doppler technique is an accepted method of deriving CO values noninvasively.12 In this study, the hemodynamic parameters were measured by using a dedicated Doppler ultrasound device, the USCOM. This device projects a 3.3-MHz, continuous-wave, Doppler signal generated from a handheld transcutaneous probe at either the pulmonary or aortic valve, from the left parasternal or suprasternal position, respectively. The reflected signal is received back by the probe and analyzed by the base unit, which, with accurate signal processing, generates a real-time hemodynamic display. Data obtained with the USCOM correlate well with the accepted standard hemodynamic methods in animal studies,13 adults,14 and neonates.15 Formal validation in children is in progress.16
The child's height and weight are used to calculate aortic or pulmonary valve cross-sectional area from a preprogrammed algorithm. Blood flow is then measured by placing the probe on the suprasternal notch (aortic) or over the fourth intercostal space (pulmonary) (Fig 2). Small manipulations are then made17 to achieve the optimal signal, which is obtained when the transducer beam is directed parallel to the flow through the valve. The velocity of transvalvular blood flow is automatically plotted against time, and the velocity-time integral is calculated. Parameters are automatically calculated from the flow signal and algorithm of valve size and are displayed in real time, from beat to beat. All study cases underwent hemodynamic assessment 4 times within 2 minutes, twice at the left and right outflow tracts. The 4 results were then averaged. The USCOM calculates the systemic vascular resistance from the cardiac index and the arterial and central venous pressures according to the standard equation, as follows: systemic vascular resistance = [(mean arterial pressure − central venous pressure)/CO] × 80.
All children underwent standard hemodynamic monitoring, namely, invasive arterial blood pressure monitoring, central venous pressure monitoring, and, where possible, central venous oxygen saturation assessment by sampling from neck lines at the time of hemodynamic assessment (this was not performed in 1 CVC and 3 CA cases because all lumens were being used for inotropes/medication). Blinding was impossible because of the design of the study, and the operator knew the presumptive diagnosis.
Continuous variables were compared with Student's t test (normally distributed data) or the Mann-Whitney test (non–normally distributed data). Categorical variables were analyzed with Fisher's exact test. In all cases, P < .05 was considered significant.
A total of 910 children were admitted during the period of the study, of whom 45 had a primary diagnosis of fluid-resistant septic shock at arrival in the PICU. All children were undergoing standard hemodynamic monitoring (electrocardiography and invasive arterial blood pressure monitoring) and were receiving artificial ventilation at the time of the study.
Fifteen children were excluded because, when they were admitted, assessment could not be performed within 4 hours after the onset of septic shock (Fig 3). The 30 remaining children were classified according to the previous presence of an indwelling CVC, into a CVC group (N = 16) or a CA septic shock group (N = 14). Table 1 shows the baseline medical characteristics of these 2 groups, whereas Table 2 shows admission physiologic data. The groups were comparable with respect to age, gender, and weight.
All children had central venous pressure of >8 mmHg at the time of assessment; although this is debatable as a measure of adequacy of volume resuscitation, it was the level used in the randomized, controlled trial of early-goal resuscitation.2 Measured central venous pressure was equivalent in the 2 groups (Table 2). No differences in electrolyte levels at the time of hemodynamic assessment were seen. Specifically, the ionized calcium levels showed no significant difference between the groups (Table 2), being >1.0 mmol/L for all children, with 3 children in each group having received standard calcium correction during transfer to the PICU because of borderline low ionized calcium levels (0.7–1.0 mmol/L). Cardiac index estimates showed low levels of variability, with no consistent discrepancy between aortic and pulmonary measurements even at high cardiac index levels, where this has been described for adult patients14 (Fig 4).
The null hypotheses was rejected because 94% of the CVC cases (15 of 16 cases) showed high cardiac index and low SVRI, in comparison with only 14% of the CA cases (2 of 14 cases; P < .001, Fisher's exact test) (Figs 5 and 6). The CVC cases differed from the CA cases in several other respects. Twelve patients with CA sepsis were transferred directly to the PICU from local emergency departments, which they had attended from home; the remaining 2 patients were transferred at the onset of fluid-resistant shock from pediatric wards to which they had been admitted shortly before deterioration. Twelve patients with CVC-related sepsis were admitted directly from wards at Great Ormond Street Hospital for Children; the other 4 patients were transferred at the onset of fluid-resistant septic shock, immediately before hemodynamic measurements, from a large local tertiary center without PICU facilities.
Admission pediatric index of mortality scores were significantly higher in the CVC group (presumably reflecting the high rate of serious underlying diseases in these patients), and the degree of organ failure (pediatric logistic organ dysfunction score) was slightly less. There was a trend toward having been in shock for less time in the CVC group (median difference: 29 minutes). However, patients in the CA group had received significantly more resuscitation, in terms of both fluid resuscitation volumes (69 mL/kg vs 55 mL/kg; P = .03) and more-intensive inotrope and vasopressor therapy (Table 3). Importantly, both groups demonstrated residual shock at presentation, by virtue of their reduced central venous oxygen saturation values (measured at the time of hemodynamic assessment, when possible), although these were lower in the CA group (Figs 4 and 7). These observations suggest that the observed hemodynamic differences are not simply a reflection of the resuscitation delivered before assessment.
Infectious investigations are shown in Table 4. No specific type of infective organism was linked to the hemodynamic patterns seen.
One child each in the CA and CVC groups died while in the PICU. There were 3 other deaths before hospital discharge in the CVC group, reflecting the severity of the patients' underlying conditions after recovery from this episode of septic shock.
On discharge review of admission diagnoses, all 30 cases were considered by the admitting clinician, who was blinded to the hemodynamic assessment, to have been correctly diagnosed with fluid-resistant septic shock. One patient in the CVC group had a discharge diagnosis of pneumonia rather than CVC infection. Interestingly, this was the only patient who did not exhibit the high cardiac index/low SVRI pattern in the initial examination.
Sepsis, defined as the systemic response to infection,18 remains a significant cause of morbidity and death in the pediatric population.3 Severe sepsis causes a release of inflammatory mediators and an associated redistribution of intravascular volume, together with depression of myocardial function. Treatment of the resultant septic shock consists of antimicrobial agents, aggressive fluid resuscitation, and titration of appropriate vasoactive drugs, depending on the cardiovascular physiologic features.1
The cardiovascular responses to severe sepsis are complex, variable, and critical to survival. They include both cardiac dysfunction and altered vascular tone, which in adults typically manifests as a hemodynamic pattern of high CO, low systemic vascular resistance, and hypotension refractory to vasopressors.19 Although data are limited for children, there is a suggestion that a greater proportion have a pattern with low CO.6
In adults with severe sepsis and septic shock, even with the elevated CO described above, depression of myocardial function is a consistent finding.20 The mechanisms underlying this cardiac dysfunction are complex, with circulating and locally produced myocardial depressant substances being postulated.19,21 Many pathways have been implicated in this myocardial depression, including nitric oxide, cyclic guanosine monophosphate, interleukin 6, tumor necrosis factor, and lipid polysaccharide.22 The mechanisms of the vascular changes in septic shock also are multifactorial but again nitric oxide is implicated, together with deficiency of vasopressin.23
Clinical practice guidelines are, of necessity, generic.1 However, the ability to obtain accurate hemodynamic data immediately at the presentation of a severe disease process known to respond to early aggressive treatment3 offers new opportunities to rationalize therapy for individual patients. Specifically, the rapid determination of hemodynamic status offered by truly noninvasive hemodynamic devices would allow tailoring of vasoactive drugs in what is known to be a dynamic situation, with children being known to change between different hemodynamic patterns as their illness progresses.6
Our observations are novel, because they demonstrate that the manifestation of septic shock in children is cause dependant, with CVC infection leading to the “adult-type” pattern with high CO and low systemic vascular resistance, in comparison with CA infections. There are several possible explanations for this pattern in CVC infections. The intravascular location of the CVC tip may allow more-direct contact of endotoxin or lipoteichoic acid with the endothelium, in comparison with other infective sites, where these and other pathogen-associated molecules may be mostly met by epithelial cells or tissue macrophages. Alternatively, a low level of colonization may somehow precondition the vasculature and, as increased organism load occurs, vascular failure supervenes. Importantly, as discussed below, the different types of organisms responsible for CVC infections, compared with CA infections, and the known cellular mechanisms of these infections in inducing septic shock may be important.
The relatively large proportion of all children with septic shock with a primary CVC infection in our study needs additional comment. The PICU in question is a tertiary referral center for a large number of children's services, many off-site. Sixty-three percent of patients with CVC infections had immunosuppression, and all except 1 had semipermanent, tunneled, neck lines for continued therapy; the remaining child had a dedicated hemodialysis line. Data on CVC infection rates in our PICU were published,24 and rates remain low.25 We presented our analysis on admission diagnoses because of the reduced potential for introducing bias and the fact that this is the situation that confronts those caring for children with septic shock.
Because most children with CVC infections were in the hospital before shock occurred and thus had regular nursing observation, it was to be anticipated that this study would have an inherent lead-time bias. There is evidence of a trend for this effect; however, the timing alone may underestimate the reality that children with CA shock are diagnosed only as having shock at admission to the ED, regardless of their prehospital condition. Lead-time bias may indeed underlie the observed hemodynamic differences between the groups, but we suggest that this does not reduce the value of this study, because our observations were made at the point at which decisions about the choice of further hemodynamic support are made. The study can be criticized for potential observer bias but is supported by other factors, such as increased inotrope scores, fluids given at assessment, and the fact that both admission pediatric index of mortality and pediatric logistic organ dysfunction scores were consistent with findings and outcomes.
Another issue is the potential confounding influence of the known response to sepsis caused by different organisms. Gram-positive infections are largely mediated by Toll-like receptor 2, whereas Gram-negative infections are known to be mediated by Toll-like receptor 4.26 More children in the CVC group had Gram-positive infections; furthermore, the CVC group had a higher incidence of immunosuppression, either disease related or iatrogenic. The burden of this clinical situation is well described.27
One finding needs further exploration, namely, the prevalence of low central venous oxygen saturation values after appropriate primary resuscitation,1 despite restoration of traditional end points of resuscitation. This demonstrates “cryptic shock”28 in these children or the inadequacy of simple clinical end points in determination of shock reversal. Although this concept has been described in adult resuscitation, our study confirms its prevalence in resuscitated children with sepsis and demonstrates that it occurs at all levels of CO measured and is independent of the cause of shock.
Current best-practice guidelines describe patients as having warm or cold shock on the basis of clinical signs1 and recommend adjustment of treatment following these allocations, although the clinical ability to judge hemodynamic parameters is known to be poor.4 With the understanding that pathogenesis may be a factor that leads to a predominance of one hemodynamic pattern, current evidence does not allow any recommendation to be made regarding the choice of vasoactive agents in these scenarios. Specifically, any suggestion that children with a CVC in the setting of fluid-resistant septic shock should be treated with vasoconstrictors alone, in the absence of adequate hemodynamic measurements, is flawed. This is well illustrated for the patient with a suspected CVC infection who was shown subsequently not to have CVC sepsis, who might have received vasoconstrictors when, although her SVRI was low, her underlying impaired cardiac function required inotropic support rather than vasoconstriction. Indeed, it is important to highlight that the only proven target to optimize in such children, given our present knowledge base, remains central venous oxygen saturation. A more realistic consequence of our findings is that a vasoconstrictor, such as norepinephrine, is a realistic first-line agent if rapid, noninvasive, hemodynamic assessment is available and the classic low SVRI/high cardiac index pattern is demonstrated for a child with an indwelling CVC in the setting of fluid-resistant septic shock. A possible future management standard might be to use a bolus dose of a long-acting vasoconstrictor, such as terlipressin,29 allowing CVC removal in a hemodynamically optimal state and potentially avoiding deterioration, as seen in children whose infected CVC is removed after additional central vascular access is inserted, enabling continuation of vasoactive infusions. This latter course of action exposes the child to prolonged bacteremia and possibly a longer time in clinical shock, which is known to be directly related to death.3
New noninvasive techniques are allowing a more rational approach to the early-goal management of septic shock in children. In this study, patients who presented with CVC-related sepsis predominantly demonstrated a pattern of elevated cardiac index with low systemic vascular resistance, whereas those who presented with CA sepsis predominantly exhibited a low cardiac index but variable systemic vascular resistance, with 9 (64%) of 14 patients exhibiting low or normal systemic vascular resistance at the time of assessment. Vasopressors may be more appropriate earlier in the course of CVC-associated sepsis, compared with CA infections, as long as hemodynamic data are available.
- Accepted December 20, 2007.
- Address correspondence to Joe Brierley, MA, Pediatric and Neonatal Intensive Care Units, Great Ormond Street Hospital for Children, Great Ormond St, London, England WC1N 3JH. E-mail:
Financial Disclosure: Dr Brierley received travel expenses from Beaver Medical to speak at a conference.
What's Known on This Subject
Both warm and cold shock have been observed in pediatric septic shock. Outcomes worsen exponentially as shock persists. Guidelines recommend that therapy be tailored to individual hemodynamics, and targeting a central venous oxygen saturation of >70% may offer an advantage.
What This Study Adds
Noninvasive, ultrasound, cardiac output measurement is practical and reproducible in the PICU. Cases of pediatric septic shock exhibit central venous desaturation even with high cardiac output. Distinct patterns of shock are seen in cases with different causes and times to assessment.
- ↵Han Y, Carcillo J, Dragotta M, et al. Early reversal of pediatric-neonatal septic shock by community physicians is associated with improved outcome. Pediatrics.2003;112 (4):793– 799
- ↵Tibby SM, Hatherill M, Marsh MJ, Murdoch IA. Clinicians' abilities to estimate cardiac index in ventilated children and infants. Arch Dis Child.1997;77 (6):516– 518
- ↵Ceneviva G, Paschall JA, Maffei F, Carcillo JA. Hemodynamic support in fluid-refractory pediatric septic shock. Pediatrics.1998;102 (2). Available at: www.pediatrics.org/cgi/content/full/102/2/e19
- ↵Wernovsky G, Wypij D, Jonas RA, et al. Postoperative course and hemodynamic profile after the arterial switch operation in neonates and infants: a comparison of low-flow cardiopulmonary bypass and circulatory arrest. Circulation.1995;92 (8):2226– 2235
- ↵Haites NE, McLennan FM, Mowat DH, Rawles JM. Assessment of cardiac output by the Doppler ultrasound technique alone. Br Heart J.1985;53 (2):123– 129
- ↵Heerman W, Doyle T, Churchwell K, Taylor M. Accuracy of non-invasive cardiac output monitoring (USCOM). Crit Care Med.2006;34 (12 suppl 2):A61
- ↵USCOM. USCOM users manual. Available at: www.uscom.com.au/support/index.html. Accessed August 10, 2008
- ↵Finkel MS, Oddis CV, Jacobs TD, et al. Negative inotropic effects of cytokines on the heart mediated by nitric oxide. Science.1992;257 (5068):387– 389
- ↵Schumacher K, Brierly J. Sustained reduction in infection rates using heparin-bonded central venous lines in intensive care. Arch Dis Child.2008;93 :A23
- Copyright © 2008 by the American Academy of Pediatrics