Differences in organ dysfunctions between neonates and older children: a prospective, observational, multicenter study

Multiple organ dysfunction syndrome (MODS) is a major problem in the pediatric intensive care unit (PICU) [1, 2]. Several studies have shown that mortality increased with the number of organ dysfunctions (ODs) in critically ill children. Two MODS scores have been developed to describe and quantify the severity of OD in critically ill children: the pediatric logistic organ dysfunction (PELOD) score and the pediatric multiple organ dysfunction score (P-MODS) [3‐5]. The neonatal multiple organ dysfunction (NEOMOD) score provides information on ODs influencing mortality during the first 28 days of life among critically ill premature babies [6].

There are several physiological and immunological characteristics that may differentiate the neonatal population [7‐11]. For example, neonates are especially vulnerable to sepsis and nosocomial infections [12], which represent a well-known cause of MODS [13]. Human and animal studies have shown differences in organ response to injury between neonates and adults [14, 15]. Recently, Typpo and colleagues [16] described the epidemiology of MODS at day 1 in patients older than 1 month across PICUs in the US. There was differential PICU mortality based on age. Infants had the highest overall PICU mortality compared with other age groups, and their increased mortality was supposed to be linked to the increased incidence of MODS [16]. No similar study has considered neonates as an independent age group in comparison with the rest of the pediatric population. The aim of the present study was to determine whether there are differences in mortality, OD incidence estimated by the PELOD score, and OD contribution to mortality, between neonates on one side and older children on the other side.

There were 2,021 consecutive admissions between September 1998 and February 2000. We excluded 215 patients for the following reasons: admission to the PICU for scheduled procedures normally cared for in another hospital location (117), prematurity (55), PICU length of stay of less than 4 hours (13), incomplete records (13), age (12), still cared for in the PICU at the end of the study (2), palliative care (2), and transfer to another PICU (1). Therefore, we enrolled 1,806 patients, including 171 (9.5%) neonates and 1,635 (90.5%) older children (525 infants: 1 month to less than 1 year; 853 children: 1 to less than 12 years; 257 adolescents: at least 12 years). Neonates were more frequently ventilated and had a higher PRISM (pediatric risk of mortality) score than older children. Organ systems of primary dysfunction on admission were different between the two strata. The population characteristics are presented in Table 1. Neurological primary dysfunction was more frequent in older children compared with neonates, whereas cardiovascular and gastrointestinal primary dysfunctions were more frequent in neonates compared with older children. Trauma, cancer, and allergic/immunologic diseases were more frequent in older children, whereas congenital diseases were more frequent in neonates (Table 1).

The global case-fatality rate was 6.4% (115 deaths). Mortality was higher among neonates compared with older patients (14.6% versus 5.5%, P < 10⁻⁷). Given different age subgroups of older children (infant: 1 month to 1 year), toddler and preschool (2 to 5 years), school-age child (6 to 12 years), and adolescent and young adult (13 to 18 years), mortality rate was not different (6.6%, 4.6%, 4.8%, 5.8%, respectively; P = 0.48). Kaplan-Meier analysis at PICU discharge showed that the mortality rate was significantly higher in neonates compared with older children (log-rank P = 0.003). MODS was also significantly more frequent in neonates compared with older children (75.4% versus 50.9%, P < 10⁻⁴).

This study showed that the incidence rate and severity of MODS and mortality rate were significantly higher in neonates compared with older children. The median and mean dPELOD scores were significantly higher in neonates from days 1 to 4 after admission to the PICU. Cardiovascular, respiratory, and renal dysfunctions were also significantly more frequent in neonates from days 1 to 4. Only three ODs were statistically related to mortality in neonates, whereas all ODs contributed significantly to mortality among older children.

There are a few studies on the incidence of MODS in the PICU. Proulx and colleagues [19] reported an incidence rate of 18%, but no distinction was made between neonates and older children. In the study by Typpo and colleagues [16] (n = 44,693 patients; neonates excluded), the incidence of MODS on day 1 was 18.6%, and all ODs contributed to mortality. We found that MODS was more frequent in neonates; this suggests that a stratification for neonates versus older children might be useful in clinical trials in which MODS is considered an outcome measure.

Mortality rate of neonates in the present study, in which prematures were excluded, was higher (14.6%) than mortality rate in neonates of all birth weights, admitted to Canadian neonatal intensive care units (4%) [20]. In our study, the significant mortality difference found between neonates and older children could be attributed to the higher incidence of MODS and a higher PRISM score among neonates compared with older children. This may reflect different diseases leading to PICU admission among neonates compared with older children (Table 1). This difference in mortality might also be attributed to different physiological processes among neonates [21] and the high frequency and variety of congenital anomalies [22].

Even though physiology does not change abruptly, studies have shown differences in organ response to injury between neonates and adults [14, 15]. In neonates with MODS, there is an early and prominent microvascular failure, characterized by a generalized capillary leak and anasarca, followed by renal and hepatic dysfunctions, while pulmonary dysfunction is the first to develop in human and animal adult MODS [14, 15].

In our study, cardiovascular dysfunction significantly contributed to neonatal mortality. In neonates, cardiomyocyte differs from that in adults because of structural differences, functional alterations in proliferative activity, and excitation-contraction coupling [23]. Cardiac physiology is also quite different: the capacity to increase stroke volume is lower in neonates. These physiological abnormalities, coupled with the fact that the neonatal left ventricular myocardium already functions at a higher baseline contractile state, and the high dependence of left ventricle systolic performance on afterload increase the susceptibility of neonates to sudden cardiac deterioration in the setting of shock and vasoconstriction [23]. Severe congenital cardiac diseases might also explain why the hazard ratio of death attributable to cardiac dysfunction is so high in neonates.

Respiratory dysfunction did not significantly contribute to mortality in neonates but was significantly more frequent in neonates during the first 4 days only. This high frequency of respiratory dysfunction may explain the higher percentage of ventilated neonates compared with that of older children. The contribution of respiratory dysfunction to mortality could have been diminished by recent management development such as extracorporeal membrane oxygenation in newborns [24].

Renal dysfunction was significantly more frequent in neonates during the first 4 days but did not significantly contribute to mortality. This might be explained by the good efficacy of supportive treatment in most cases of neonatal acute renal failure [25]. Factors associated with neonatal mortality in case of renal dysfunction include multiorgan failure, hypotension, need for vasopressors, hemodynamic instability, and need for mechanical ventilation and dialysis [26]. This probably means that death in neonates with renal failure is seldom caused primarily by renal diseases [27].

Hepatic dysfunction significantly contributed to mortality in the two strata (odds ratios [ORs] 3.02, 95% confidence interval [CI] 1.01 to 9.1 in neonates and 2.00, 95% CI 1.12 to 3.60 in older children). However, hepatic dysfunction in neonates and older children had a relatively low incidence (22.8% and 16.8%, respectively). In the study by Tantaleán and colleagues [1] carried out on 276 patients (including 37 newborns) admitted to the PICU, hepatic dysfunction was infrequent (5.8%) and associated with the highest risk of mortality (OR 7.33, 95% CI 1.99 to 26.9) [1]. In the study by Typpo and colleagues [16], hepatic dysfunction had the lowest incidence (0.9%) and the OR of mortality (3.7, 95% CI 2.7 to 5.1) was close to that of the other ODs (from 2.8 [95% CI 2.5 to 3.2] for cardiovascular dysfunction to 5.5 [95% CI 4.7 to 6.5] for respiratory dysfunction).

Neurological dysfunction significantly contributed to mortality in neonates and older children. In the study by Flori and colleagues [28], which included children and neonates who were more than 36 weeks of gestational age and meeting the 1994 American European Consensus Committee definition of acute lung injury (n = 328 admissions), neurological dysfunction contributed independently to an increased risk of death (OR 12.58, 95% CI 6.78 to 23.31). In our study, the ORs of 1.118 in neonates and 1.156 in older children corresponded to a variation of 1 point for the neurological OD score. In patients with severe neurological dysfunction (corresponding to 20 points), ORs were 9.31 in neonates (1.118²⁰) and 18.16 in older children (1.156²⁰). These values were close to the OR of 12.58 reported by Flori and colleagues [28].

Incidence of hematological dysfunction of the entire stay was the same in neonates and older children (17%). Hematological dysfunction was a significant contributor to mortality in older children only (Table 2). Similarly, in children excluding neonates, Johnston and colleagues [29] also showed that hematological dysfunction was a significant contributor to mortality (OR 3.10, 95% CI 2.78 to 3.46).

A limitation of this prospective study is the time elapsed since the period when data were collected (1998, 2000): there is a risk that the case mix of patients in the PICU has changed over this period. A second limitation is the possibility of a false association between the PELOD score and death rate in neonates. In fact, the AUC in this group was acceptable but lower (0.78) than in the older children group (0.93), suggesting a differential performance (discrimination) of the model between the two groups. Another limitation is that we considered only intensive care unit mortality and not hospital mortality. However, in hospital post-intensive care, mortality of critically ill children is not frequent. A study from a PICU, typical of US units, showed that among 341 survivors only three children (0.9%) died in the hospital after discharge from intensive care [30]. Otherwise, the PELOD score has been criticized, mainly because it does not assign risk on a continuous scale [31].

The incidence and severity of MODS and mortality rate were significantly higher in neonates than in older children. Three ODs - neurological, cardiovascular, and hepatic dysfunctions - significantly contributed to mortality in neonates, whereas all ODs were significantly associated with mortality in older children. In our hands, the PELOD and dPELOD scores were higher in neonates and risk of death with similar PELOD scores tended to be higher in neonates than older children. These data suggest that an updated version of the PELOD score should take this into account; also, they suggest that it might be a good strategy to consider these two strata in randomized clinical trials involving critically ill children.