Anthony Slater

Paediatric index of mortality (PIM): a mortality prediction model for children in intensive care

AbstractObjective: To develop a logistic regression model that predicts the risk of death for children less than 16 years of age in intensive care, using information collected at the time of admission to the unit. Design: Three prospective cohort studies, from 1988 to 1995, were used to determine the variables for the final model. A fourth cohort study, from 1994 to 1996, collected information from consecutive admissions to all seven dedicated paediatric intensive care units in Australia and one in Britain. Results: 2904 patients were included in the first three parts of the study, which identified ten variables for further evaluation. 5695 children were in the fourth part of the study (including 1412 from the third part); a model that used eight variables was developed on data from four of the units and tested on data from the other four units. The model fitted the test data well (deciles of risk goodness-of-fit test p=0.40) and discriminated well between death and survival (area under the receiver operating characteristic plot 0.90). The final PIM model used the data from all 5695 children and also fitted well (p=0.37) and discriminated well (area 0.90). Conclusions: Scores that use the worst value of their predictor variables in the first 12–24 h should not be used to compare different units: patients mismanaged in a bad unit will have higher scores than similar patients managed in a good unit, and the bad unit‘s high mortality rate will be incorrectly attributed to its having sicker patients. PIM is a simple model that is based on only eight explanatory variables collected at the time of admission to intensive care. It is accurate enough to be used to describe the risk of mortality in groups of children.

The suitability of the Pediatric Index of Mortality (PIM), PIM2, the Pediatric Risk of Mortality (PRISM), and PRISM III for monitoring the quality of pediatric intensive care in Australia and New Zealand.

Objective: To compare the performance of the Pediatric Index of Mortality (PIM), PIM2, the Pediatric Risk of Mortality (PRISM), and PRISM III in Australia and New Zealand. Design: A two-phase prospective observational study. Phase 1 assessed the performance of PIM, PRISM, and PRISM III between 1997 and 1999. Phase 2 assessed PIM2 in 2000 and 2001. Setting: Ten intensive care units in Australia and New Zealand. Patients: Included in the study were 26,966 patients aged <16 yrs; 1,147 patients died in the intensive care unit. Interventions: None. Measurements and Main Results: Discrimination between death and survival was assessed by calculating the area under the receiver operating characteristic plot for each model. The areas (95% confidence interval) for PIM, PIM2, PRISM, and PRISM III were 0.89 (0.88–0.90), 0.90 (0.88–0.91), 0.90 (0.89–0.91), and 0.93 (0.92–0.94). The calibration of the models was assessed by comparing the number of observed to predicted deaths in different diagnostic and risk groups. Prediction was best using PIM2 with no difference between observed and expected mortality (standardized mortality ratio [95% confidence interval] 0.97 [0.86–1.05]). PIM, PRISM III, and PRISM all overpredicted death, predicting 116%, 130%, and 189% of observed deaths, respectively. The performance of individual units was compared during phase 1, using PIM, PRISM, and PRISM III. There was agreement between the models in the identification of outlying units; two units performed better than expected and one unit worse than expected for each model. Conclusions: Of the models tested, PIM2 was the most accurate and had the best fit in different diagnostic and risk groups; therefore, it is the most suitable mortality prediction model to use for monitoring the quality of pediatric intensive care in Australia and New Zealand. More information about the performance of the models in other regions is required before these results can be generalized.

Glucose control, organ failure, and mortality in pediatric intensive care

Objective: In ventilated children, to determine the prevalence of hyperglycemia, establish whether it is associated with organ failure, and document glycemic control practices in Australasian pediatric intensive care units (PICUs). Design: Prospective inception cohort study. Setting: All nine specialist PICUs in Australia and New Zealand. Patients: Children ventilated >12 hrs excluding those with diabetic ketoacidosis, on home ventilation, undergoing active cardiopulmonary resuscitation on admission, or with do-not-resuscitate orders. Interventions: None. Measurements and Main Results: All blood glucose measurements for up to 14 days, clinical and laboratory values needed to calculate Paediatric Logistic Organ Dysfunction (PELOD) scores, and insulin use were recorded in 409 patients. Fifty percent of glucose measurements were >6.1 mmol/L, with 89% of patients having peak values >6.1 mmol/L. The median time to peak blood glucose was 7 hrs. Hyperglycemia was defined by area under the glucose-time curve >6.1 mmol/L above the sample median. Thirteen percent of hyperglycemic subjects died vs. 3% of nonhyperglycemic subjects. There was an independent association between hyperglycemia and a PELOD score ≥10 (odds ratio 3.41, 95% confidence interval 1.91–6.10) and death (odds ratio 3.31, 95% confidence interval 1.26–7.7). Early hyperglycemia, defined using only glucose data in the first 48 hrs, was also associated with these outcomes but not with PELOD ≥10 after day 2 or with worsening PELOD after day 1. Five percent of patients received insulin. Conclusions: Hyperglycemia is common in PICUs, occurs early, and is independently associated with organ failure and death. However, early hyperglycemia is not associated with later or worsening organ failure. Australasian PICUs seldom use insulin.

Paediatric index of mortality 3: An updated model for predicting mortality in pediatric intensive care

Objectives: To provide an updated version of the Paediatric Index of Mortality 2 for assessing the risk of mortality among children admitted to an ICU. Design: International, multicenter, prospective cohort study. Setting: Sixty ICUs that accept pediatric admissions in Australia, New Zealand, Ireland, and the United Kingdom. Patients: All children admitted in 2010 and 2011 younger than 18 years old at the time of admission and either died in ICU or were discharged. Patients who were transferred to another ICU were not included. Fifty-three thousand one hundred twelve patient admissions were included in the analysis. Interventions: None. Measurement and Main Results: A revised prediction model was built using logistic regression. Variable selection was based on significance at the 95% level and overall improvement of the model’s discriminatory performance and goodness of fit. The final model discriminated well (area under the curve, 0.88, 0.88–0.89); however, the model performed better in Australia and New Zealand than in the United Kingdom and Ireland (area under the curve was 0.91, 0.90–0.93 and 0.85, 0.84–0.86, respectively). Conclusions: Paediatric Index of Mortality 3 provides an international standard based on a large contemporary dataset for the comparison of risk-adjusted mortality among children admitted to intensive care.

Mortality related to invasive infections, sepsis, and septic shock in critically ill children in Australia and New Zealand, 2002–13: a multicentre retrospective cohort study

BACKGROUND Severe infections kill more than 4·5 million children every year. Population-based data for severe infections in children requiring admission to intensive care units (ICUs) are scarce. We assessed changes in incidence and mortality of severe infections in critically ill children in Australia and New Zealand. METHODS We did a retrospective multicentre cohort study of children requiring intensive care in Australia and New Zealand between 2002 and 2013, with data from the Australian and New Zealand Paediatric Intensive Care Registry. We included children younger than 16 years with invasive infection, sepsis, or septic shock. We assessed incidence and mortality in the ICU for 2002-07 versus 2008-13. FINDINGS During the study period, 97 127 children were admitted to ICUs, 11 574 (11·9%) had severe infections, including 6688 (6·9%) with invasive infections, 2847 (2·9%) with sepsis, and 2039 (2·1%) with septic shock. Age-standardised incidence increased each year by an average of 0·56 cases per 100 000 children (95% CI 0·41-0·71) for invasive infections, 0·09 cases per 100 000 children (0·00-0·17) for sepsis, and 0·08 cases per 100 000 children (0·04-0·12) for septic shock. 260 (3·9%) of 6688 patients with invasive infection died, 159 (5·6%) of 2847 with sepsis died, and 346 (17·0%) of 2039 with septic shock died, compared with 2893 (3·0%) of all paediatric ICU admissions. Children admitted with invasive infections, sepsis, and septic shock accounted for 765 (26·4%) of 2893 paediatric deaths in ICUs. Comparing 2008-13 with 2002-07, risk-adjusted mortality decreased significantly for invasive infections (odds ratio 0·72, 95% CI 0·56-0·94; p=0·016), and for sepsis (0·66, 0·47-0·93; p=0·016), but not significantly for septic shock (0·79, 0·61-1·01; p=0·065). INTERPRETATION Severe infections remain a major cause of mortality in paediatric ICUs, representing a major public health problem. Future studies should focus on patients with the highest risk of poor outcome, and assess the effectiveness of present sepsis interventions in children. FUNDING National Medical Health and Research Council, Australian Resuscitation Outcomes Consortium, Centre of Research Excellence (1029983).

Hypothermia for traumatic brain injury in children - a Phase II randomized controlled trial

Objectives:To perform a pilot study to assess the feasibility of performing a phase III trial of therapeutic hypothermia started early and continued for at least 72 hours in children with severe traumatic brain injury. Design:Multicenter prospective randomized controlled phase II trial. Setting:All eight of the PICUs in Australia and New Zealand and one in Canada. Patients:Children 1–15 years old with severe traumatic brain injury and who could be randomized within 6 hours of injury. Interventions:The control group had strict normothermia to a temperature of 36–37°C for 72 hours. The intervention group had therapeutic hypothermia to a temperature of 32–33°C for 72 hours followed by slow rewarming at a rate compatible with maintaining intracranial pressure and cerebral perfusion pressure. Measurements and Main Results:Of 764 children admitted to PICU with traumatic brain injury, 92 (12%) were eligible and 55 (7.2%) were recruited. There were five major protocol violations (9%): three related to recruitment and consent processes and two to incorrect temperature management. Rewarming took a median of 21.5 hours (16–35 hr) and was performed without compromise in the cerebral perfusion pressure. There was no increase in any complications, including infections, bleeding, and arrhythmias. There was no difference in outcomes 12 months after injury; in the therapeutic hypothermia group, four (17%) had a bad outcome (pediatric cerebral performance category, 4–6) and three (13%) died, whereas in the normothermia group, three (12%) had a bad outcome and one (4%) died. Conclusions:Early therapeutic hypothermia in children with severe traumatic brain injury does not improve outcome and should not be used outside a clinical trial. Recruitment rates were lower and outcomes were better than expected. Conventional randomized controlled trials in children with severe traumatic brain injury are unlikely to be feasible. A large international trials group and alternative approaches to trial design will be required to further inform practice.

Pandemic H1N1 in children requiring intensive care in Australia and New Zealand during winter 2009

OBJECTIVE: To describe in detail the pediatric intensive care experience of influenza A, particularly pandemic H1N1-09, in Australia and New Zealand during the 2009 Southern Hemisphere winter and to compare the pediatric experience with that of adults. METHOD: This was an inception-cohort study of all children who were admitted to intensive care with confirmed influenza A during winter 2009 at all general ICUs and PICUs in Australia and New Zealand. RESULTS: From June 1 through August 31, 2009, 107 children (20.0 per million [95% confidence interval: 16.1–23.8]) with influenza A, including 83 (15.5 per million [95% confidence interval: 12.1–18.9]) with H1N1-09 were admitted to ICUs. Fifty-two percent (39 of 75) of children with H1N1-09 had 1 or more comorbidity, most commonly neurologic (20%). Most (48 of 83 [58%]) presented with pneumonia. Thirteen of 83 (16%) had neurologic presentations. Eighty percent of the children with H1N1-09 required ventilation. Mortality was lower than in adults: 6 of 83 (7%) vs 114 of 668 (17%) (P = .02). The median length of stay for children with H1N1-09 was 5 days. Children with H1N1-09 occupied 773 bed-days (147 per million children) and 5.8% of specialist PICU beds. Presentation with septic shock or after cardiac arrest and the presence of 1 or more comorbidities were risk factors for severe disease. CONCLUSIONS: H1N1-09 caused a substantial burden on pediatric intensive care services in Australia and New Zealand. Compared with adults, children more commonly had nonrespiratory presentations and required ventilation more often but had a lower mortality rate.

Five-Year Survival of Children With Chronic Critical Illness in Australia and New Zealand.

Objective:Outcomes for children with chronic critical illness are not defined. We examined the long-term survival of these children in Australia and New Zealand. Design:All cases of PICU chronic critical illness with length of stay more than 28 days and age 16 years old or younger in Australia and New Zealand from 2000 to 2011 were studied. Five-year survival was analyzed using Kaplan-Meir estimates, and risk factors for mortality evaluated using Cox regression. Setting:All PICUs in Australia and New Zealand. Patients:Nine hundred twenty-four children with chronic critical illness. Intervention:None. Measurements and Main Results:Nine hundred twenty-four children were admitted to PICU for longer than 28 days on 1,056 occasions, accounting for 1.3% of total admissions and 23.5% of bed days. Survival was known for 883 of 924 patients (95.5%), with a median follow-up of 3.4 years. The proportion with primary cardiac diagnosis increased from 27% in 2000–2001 to 41% in 2010–2011. Survival was 81.4% (95% CI, 78.6–83.9) to PICU discharge, 70% (95% CI, 66.7–72.8) at 1 year, and 65.5% (95% CI, 62.1–68.6) at 5 years. Five-year survival was 64% (95% CI, 58.7–68.6) for children admitted in 2000–2005 and 66% (95% CI, 61.7–70) if admitted in 2006–2011 (log-rank test, p = 0.37). After adjusting for admission severity of illness using the Paediatric Index of Mortality 2 score, predictors for 5-year mortality included bone marrow transplant (hazard ratio, 3.66; 95% CI, 2.26–5.92) and single-ventricle physiology (hazard ratio, 1.98; 95% CI, 1.37–2.87). Five-year survival for single-ventricle physiology was 47.2% (95% CI, 34.3–59.1) and for bone marrow transplantation 22.8% (95% CI, 8.7–40.8). Conclusions:Two thirds of children with chronic critical illness survive for at-least 5 years, but there was no improvement between 2000 and 2011. Cardiac disease constitutes an increasing proportion of pediatric chronic critical illness. Bone marrow transplant recipients and single-ventricle physiology have the poorest outcomes.

Monitoring outcome in paediatric intensive care

How effectively do we care for children admitted to our intensive care units? How do we answer this question? A number of statistical models have been developed that allow the outcome of a group of intensive care patients to be assessed after adjustment for known risk factors. It is becoming increasingly common for these techniques to be used routinely to monitor, or benchmark, the performance of intensive care units (ICU). Monitoring of ICU performance is often initiated by clinicians, however, government and other health care funding agencies now aim to ensure that the health outcomes in institutions they fund meet acceptable standards. It is important that clinicians working in the area understand the benefits and limitations of the methods used to assess ICU performance.

Burden of disease and change in practice in critically ill infants with bronchiolitis.

Bronchiolitis represents the most common cause of non-elective admission to paediatric intensive care units (ICUs). We assessed changes in admission rate, respiratory support, and outcomes of infants <24 months with bronchiolitis admitted to ICU between 2002 and 2014 in Australia and New Zealand. During the study period, bronchiolitis was responsible for 9628 (27.6%) of 34 829 non-elective ICU admissions. The estimated population-based ICU admission rate due to bronchiolitis increased by 11.76 per 100 000 each year (95% CI 8.11–15.41). The proportion of bronchiolitis patients requiring intubation decreased from 36.8% in 2002, to 10.8% in 2014 (adjusted OR 0.35, 95% CI 0.27–0.46), whilst a dramatic increase in high-flow nasal cannula therapy use to 72.6% was observed (p<0.001). We observed considerable variability in practice between units, with six-fold differences in risk-adjusted intubation rates that were not explained by ICU type, size, or major patient factors. Annual direct hospitalisation costs due to severe bronchiolitis increased to over USD30 million in 2014. We observed an increasing healthcare burden due to severe bronchiolitis, with a major change in practice in the management from invasive to non-invasive support that suggests thresholds to admittance of bronchiolitis patients to ICU have changed. Future studies should assess strategies for management of bronchiolitis outside ICUs. Changing thresholds to admit bronchiolitis patients to PICU have had a major impact on cost and resource utilisation http://ow.ly/AVA630a08rx