Andrew T. Costarino

Cytokine elevations in critically ill infants with sepsis and necrotizing enterocolitis

We hypothesized that plasma levels of cytokines such as interleukin-6 and tumor necrosis factor (TNF) are elevated in critically ill infants with sepsis and necrotizing enterocolitis (NEC) and that the magnitude of their elevation is correlated with mortality rate. We measured plasma levels of interleukin-6 and TNF in 62 newborn infants with suspected sepsis or NEC. Eighteen infants had bacterial sepsis, 9 had bacterial sepsis plus NEC, and 15 had NEC but negative culture results. Twenty comparably ill infants with negative results on culture of systemic specimens served as study control subjects. Interleukin-6 levels were five- to tenfold higher in infants with bacterial sepsis plus NEC at the onset of disease than in infants with bacterial sepsis alone, in infants with NEC but negative culture results, and in control infants (p < 0.01). These differences persisted throughout the 48-hour study period. Interleukin-6 levels were also significantly higher in nonsurvivors than in survivors (p < 0.001). In contrast, plasma TNF values were not consistently increased in any of the groups. We conclude that plasma interleukin-6 is a more reliable indicator of bacterial sepsis and NEC than plasma TNF and may identify infants who might benefit from immunotherapeutic strategies.

Correlation of plasma cytokine elevations with mortality rate in children with sepsis

Cytokines are thought to be important endogenous mediators of the host immune response to bacterial infection. We hypothesized that plasma levels of cytokines are elevated in children with sepsis and that the magnitude of elevation of these cytokines is correlated with severity of illness and mortality rate. We determined plasma levels of tumor necrosis factor, interleukin-6, and interleukin-1 in 21 children with sepsis. Plasma samples were collected at presentation and at 12, 24, and 48 hours thereafter. Cytokine levels were elevated in pediatric patients with bacterial sepsis during the first 48 hours after presentation; levels were undetectable in study control subjects. The tumor necrosis factor and interleukin-6 levels (p less than 0.001), as well as levels of interleukin-1 (p = 0.05), were significantly higher in nonsurvivors than in survivors and were independent of severity of illness (pediatric risk of mortality (PRISM) score) at presentation. Elevations of tumor necrosis factor and interleukin-6 were sustained for longer than 24 to 48 hours in nonsurvivors: II-1 concentrations were significantly increased only at time zero. Of 11 children with an interleukin-6 value greater than 2 ng/ml during the first 48 hours, 10 died; only one of 10 not reaching that level died (p less than 0.001). Cytokines were elevated as frequently with gram-positive as with gram-negative infections. We speculate that cytokine determinations may identify children who might benefit from immunotherapeutic interventions.

Adult respiratory distress syndrome in children : associated disease, clinical course, and predictors of death

The adult respiratory distress syndrome is a common cause of respiratory failure; however, its incidence, risk factors, course, and mortality rate for children remain incompletely understood. A 24-month surveillance of pediatric intensive care admissions identified 60 children with adult respiratory distress syndrome constituting 2.7% of such admissions, 8% of total days spent in a pediatric intensive care unit, and 33% of deaths. The mortality rate was 62% (confidence interval, 48.2% to 73.9%). Adult respiratory distress syndrome occurred in approximately 12% of children admitted for sepsis, viral pneumonia, smoke inhalation, or drowning. A low incidence (< 3%) was observed in children admitted with pulmonary contusion or multiple trauma. Ongoing changes in measures of pulmonary gas exchange varied with the magnitude of alveolar injury; no differences were associated with the underlying acute disease or lung injury mechanism. Efficiency of oxygenation differed among outcome groups by the second day after onset of adult respiratory distress syndrome. An alveolar-arterial oxygen tension difference > 420 was the best early predictor of death (sensitivity 80%, specificity 87%, positive predictive value 87%, negative predictive value 80%, and odds ratio 26.7). We conclude that adult respiratory distress syndrome behaves clinically as a single disease regardless of the underlying cause; its course and outcome are dependent on the magnitude of alveolar injury. We speculate that strategies for minimizing secondary lung injury may benefit all patients with adult respiratory distress syndrome.

Sodium restriction versus daily maintenance replacement in very low birth weight premature neonates: A randomized, blind therapeutic trial

To test the hypothesis that restriction of sodium intake during the first 3 to 5 days of life will prevent the occurrence of hypernatremia and the need for administration of large fluid volumes, we prospectively and randomly assigned 17 babies (mean +/- SD: 850 +/- 120 gm; 27 +/- 1 weeks of gestation) to receive in blind fashion either daily maintenance sodium or salt restriction with physician-prescribed parenteral fluid intake. Maintenance-group infants received 3 to 4 mEq of sodium per kilogram per day; restricted infants received no sodium supplement other than with such treatments as transfusion. Sodium balance studies conducted for 5 days demonstrated that maintenance salt intake resulted in a daily sodium balance near zero, whereas sodium-restricted infants continued to excrete urinary sodium at a high rate, which promoted a more negative balance (average daily sodium balance -0.30 +/- 1.78 SD in maintenance group vs -3.71 +/- 1.47 mEq/kg per day in restriction group; p less than 0.001). Care givers tended to prescribe daily increases in parenteral fluids for the salt-supplemented infants, perhaps because serum sodium concentrations were elevated in these infants after the first day of the study (p less than 0.001). Hypernatremia developed in two sodium-supplemented infants (greater than 150 mEq/L), and hyponatremia developed in two sodium-restricted infants (less than 130 mEq/L); however, the restricted infants were more likely to have normal serum osmolality (p less than 0.05). Both groups of infants produced urine that was neither concentrated nor dilute, with a high fractional excretion of sodium; renal failure was not observed. The mortality rate was not affected, but the incidence of bronchopulmonary dysplasia was significantly less in the sodium-restricted babies (p less than 0.02). We conclude that in tiny premature infants, a fluid regimen that restricts sodium may simplify parenteral fluid therapy targeted to prevent hypernatremia and excessive administration of parenteral fluids.

Cytokine elevations in infants with bacterial and aseptic meningitis.

We sought to determine whether the detection of cytokines, produced during the inflammatory response, would aid in the diagnosis of meningitis in young infants. We measured cerebrospinal fluid (CSF) and plasma levels of interleukin-6 (IL-6) and tumor necrosis factor (TNF) in 62 infants less than 6 months of age whose condition was evaluated for meningitis. Twenty infants had culture-proved meningitis, 22 had aseptic meningitis, and 20 control infants had no evidence of meningitis. The CSF IL-6 levels were elevated in all 20 infants with bacterial meningitis and in 9 of 22 infants with aseptic meningitis but were undetectable in all control subjects. Furthermore, CSF IL-6 levels were 10 times greater in infants with bacterial versus aseptic meningitis (p < 0.001). Levels of TNF in CSF were detected in 12 of 20 infants with bacterial meningitis and were undetectable in infants with aseptic meningitis and in control infants (p < 0.02). Plasma IL-6 and TNF levels were unreliable for the detection of meningitis in this patient population. We conclude that the presence of IL-6 in the CSF reliably identifies infants with meningitis and that the presence of CSF TNF is a highly specific indicator of bacterial meningeal inflammation.

The impact of adverse patient occurrences on hospital costs in the pediatric intensive care unit.

ObjectivesTo study the influence of adverse patient occurrences defined as airway complication (AC), vascular complication (VC), and infectious complication (IC) on intensive care unit (ICU) costs and length of stay (LOS). DesignRetrospective, cohort study SettingAn urban, tertiary care children’s hospital in Philadelphia, PA. PatientsAll children admitted to a multidisciplinary pediatric ICU during the fiscal year 1994. InterventionsNone Measurements and Main ResultsDemographic data, diagnoses, Pediatric Risk of Mortality scores, LOS, and deaths were recorded. Hospital charges were converted into costs by using cost-to-charge ratios. There were 23 AC, 35 VC, and 40 IC events. Multiple regression in models adjusting for age, Pediatric Risk of Mortality score, mortality, chronic disease, and diagnosis by using AC, VC, and IC indicator variables was performed on the dependent variables of LOS and total costs. IC was associated with an increase in total costs of

Nonoliguric hyperkalemia in the premature infant weighing less than 1000 grams

50,361.89 (p < .001), an increased LOS of 15.6 days (p < .001), and no significant increase in daily costs. There were no significant increases in costs or LOS seen with the AC and VC variables. In a matched analysis, the total costs attributable to IC averaged

Hospital costs of pediatric intensive care

32,040 per patient. ConclusionsThe occurrence of complications in the pediatric ICU is associated with meaningful increases in LOS and hospital costs. ICs are more important predictors of costs than ACs or VCs. Continuing efforts should be made to minimize adverse occurrences to improve patient care and reduce costs.

Continuous arteriovenous hemofiltration/dialysis improves pulmonary gas exchange in children with multiple organ system failure

Eighteen very low birth weight premature infants born before 28 weeks gestation and weighing less than 1000 gm were evaluated prospectively for disturbances in serum electrolyte concentrations and for renal glomerular and tubular functions. Clinically symptomatic hyperkalemia resulting in significant electrocardiographic dysrhythmias developed in eight of these infants; 10 babies remained normokalemic. Peak serum potassium concentration ranged from 6.9 to 9.2 mEq/L in the hyperkalemic group; all potassium values in the normokalemic group were less than 6.6 mEq/L. Indices of renal glomerular function and urine output were similar in both groups; no infant had oliguria. Serum creatinine concentrations were the same in both groups (1.04 +/- 0.16 SD mg/dl in normokalemic vs 1.19 +/- 0.24 mg/dl in hyperkalemic infants, beta less than 0.2 at alpha = 0.05), and glomerular filtration rates did not differ significantly (6.29 +/- 1.78 ml/min/1.73 m2 in normokalemic vs 5.70 +/- 1.94 ml/min/1.73 m2 in hyperkalemic infants, beta less than 0.2 at alpha = 0.05). In contrast, indicators of tubular function revealed a significantly larger fractional excretion of sodium in hyperkalemic infants: 13.9 +/- 5.4% versus 5.6 +/- 0.9% in normokalemic control subjects (p less than 0.001). Hyperkalemic infants also had a tendency toward lower urine concentrations of potassium, although there was no significant difference in their net potassium excretion in comparison with that in the normokalemic group. We speculate that hyperkalemia in the tiny baby is in part the result of immature distal tubule function with a compromise in ability to regulate potassium balance.

Airway pressure release ventilation in pediatrics

OBJECTIVE To characterize hospital costs of pediatric intensive care and to determine which demographic and disease characteristics are associated with cost. DESIGN Prospective cohort study. SETTING A 20-bed pediatric intensive care unit (PICU) in an urban university-affiliated teaching childrens hospital. PATIENTS All children (n = 1,376) admitted to the multidisciplinary PICU during the fiscal year 1994. INTERVENTIONS None. MEASUREMENTS AND MAIN RESULTS Demographics, diagnoses, organ failure, Pediatric Risk of Mortality score, length of stay (LOS), and outcome were recorded. All hospital charges were obtained. The actual hospital costs were calculated by two separate methods. First, we converted the itemized patient charges into costs, using corresponding cost-to-charge ratios for each charge. In addition, we examined all direct and indirect expenses for the PICU. Univariate and multivariate regression analyses were used to determine the correlates to cost. The study population was similar to that of other studies of pediatric intensive care. The PICU was 86% efficient. The total cost for PICU care was