ABSTRACT
Introduction: There are limited reports on the epidemiology of paediatric intensive care unit (PICU) admissions, deaths and organ donation candidacy. We aimed to describe PICU admission characteristics and outcomes, determine risk factors for mortality, and perform an independent assessment of missed organ donation opportunities.
Method:We adopted a clinical audit design recruiting consecutive patients admitted to a single-centre multidisciplinary PICU from June 2020 to December 2023. Clinical characteristics and outcomes of survivors and non-survivors were described. Multivariable regression was performed to identify independent risk factors for mortality. Organ donation candidacy was evaluated by an independent team based on the criteria by Singapore’s National Organ Transplant Unit.
Results: There were 1766 PICU admissions with mean age ± standard deviation of 5.9 ± 6.0 years. Surgical admissions accounted for 707/1766 (40%), while the most common medical admission category was respiratory (416/1766; 23.6%). The majority of 983/1766 (55.7%) had a chronic comorbidity and 312/1766 (17.6%) were dependent on at least 1 medical technology device. Mortality occurred in 99/1766 (5.6%). After adjusting for elective admissions and admission category; comorbidity with adjusted odds ratio (aOR) 95% confidence interval (CI) 3.03 (1.54–5.96); higher Pediatric Index of Mortality 3 (PIM 3) score with aOR 1.06 (95% CI 1.04–1.08); and functional status scale with aOR 1.07 (95% CI 1.00–1.13) were associated with mortality. Among non-survivors, organ donor candidacy was 21/99 (21.2%) but successful organ donation occurred in only 2/99 (2.0%).
Conclusion: In this single-centre audit, comorbidities, PIM 3 score and functional impairment were associated with mortality. Efforts are needed to improve paediatric organ donation rates.
CLINICAL IMPACT
What is New
- This clinical audit identified a substantial population of children with chronic disease and medical technology dependence who were admitted to our paediatric intensive care unit (PICU).
- Presence of chronic disease, higher admission severity score and functional impairment were associated with mortality.
- Among PICU non-survivors, there was a large gap between eligibility and actual organ donation that predominantly stemmed from non-identification of potential organ donors.
Clinical Implications
- Children with chronic disease and functional impairment are vulnerable populations within our PICU, underscoring the need for research in these groups of children to improve outcomes.
- Concrete measures to improve paediatric organ donation rates may include having an independent team screen patients for eligibility when death is anticipated, involving the donor coordinator in family discussions, and establishing an operational workflow for donation after cardiac death.
- Tilford JM, Roberson PK, Lensing S, et al. Differences in pediatric ICU mortality risk over time. Crit Care Med 1998;26:1737-43.
- Choi J, Park E, Choi AY, et al. Incidence and Mortality Trends in Critically Ill Children: A Korean Population-Based Study. J Korean Med Sci 2023;38:e178.
- Burns JP, Sellers DE, Meyer EC, et al. Epidemiology of death in the PICU at five U.S. teaching hospitals*. Crit Care Med 2014;42:2101-8.
- Ishihara T, Tanaka H. Causes of death in critically ill paediatric patients in Japan: a retrospective multicentre cohort study. BMJ Paediatr Open 2019;3:e000499.
- Lyimo RE, Said YH, Kivuyo SL, et al. Mortality and associated factors among children admitted to an intensive care unit in muhimbili national hospital, from the time of admission to three months after discharge: a prospective cohort study. BMC Pediatr 2024;24:170.
- Rashma RP, Remya S, Jayakumar C, et al. Mortality Profile of Children Admitted to Intensive Care Unit of a Tertiary Care Hospital in Kerala, South India Article Information. Int J Med Clin Sci 2018;1:13-6.
- Aryati NMD, Hartawan INB, Wati DK, et al. Mortality Patterns among Critically Ill Children in a Pediatric Intensive Care Unit of a Sanglah Hospital. International Journal of Science and Research 2021;10:1629-34.
- Systems Virtual Pediatric. https://myvps.org. Accessed 10 June 2024.
- Group ANZICS Paediatric Study. https://www.anzics.org/australian-and-new-zealand-paediatric-intensive-care-registry-anzpicr/. Accessed 10 June 2024.
- Network Paediatric Intensive Care Audit. https://www.picanet.org.uk. Accessed 10 June 2024.
- Corkery-Lavender T, Millar J, Cavazzoni E, et al. Patterns of organ donation in children in Australia and New Zealand. Crit Care Resusc 2017;19:296-302.
- Rodríguez Núñez A, Pérez Blanco A, Grupo de Trabajo de la AEP-ONT. National recommendations on pediatric donation. An Pediatr (Engl Ed) 2020;93:134.e1-.e9.
- Ministry of Health, Singapore. National Organ Transplant Unit. https://www.liveon.gov.sg. Accessed 12 June 2024.
- Straney L, Clements A, Parslow RC, et al. Paediatric index of mortality 3: an updated model for predicting mortality in pediatric intensive care*. Pediatr Crit Care Med. 2013;14:673-81.
- Wong JJ, Hornik CP, Mok YH, et al. Performance of the Paediatric Index of Mortality 3 and Paediatric Logistic Organ Dysfunction 2 Scores in Critically Ill Children. Ann Acad Med Singap 2018;47:285-90.
- Edwards JD, Houtrow AJ, Vasilevskis EE, et al. Chronic conditions among children admitted to U.S. pediatric intensive care units: their prevalence and impact on risk for mortality and prolonged length of stay*. Crit Care Med 2012;40:2196-203.
- Feudtner C, Feinstein JA, Zhong W, et al. Pediatric complex chronic conditions classification system version 2: updated for ICD-10 and complex medical technology dependence and transplantation. BMC Pediatr 2014;14:199.
- Pollack MM, Holubkov R, Glass P, et al. Functional Status Scale: new pediatric outcome measure. Pediatrics 2009;124:e18-28.
- Ministry of Health, Singapore. Manual on Organ Donation and Transplantation. 5th ed. Singapore: Ministry of Health; 2018.
- Moynihan KM, Alexander PMA, Schlapbach LJ, et al. Epidemiology of childhood death in Australian and New Zealand intensive care units. Intensive Care Med 2019;45:1262-71.
- Sands R, Manning JC, Vyas H, et al. Characteristics of deaths in paediatric intensive care: a 10-year study. Nurs Crit Care 2009;14:235-40.
- Lee OJ, Jung M, Kim M, et al. Validation of the Pediatric Index of Mortality 3 in a Single Pediatric Intensive Care Unit in Korea. J Korean Med Sci 2017;32:365-70.
- Pollack MM, Patel KM, Ruttimann UE. PRISM III: An updated Pediatric Risk of Mortality score. Crit Care Med 1996;24:743-52.
- O’Brien S, Nadel S, Almossawi O, et al. The Impact of Chronic Health Conditions on Length of Stay and Mortality in a General PICU. Pediatr Crit Care Med 2017;18:1-7.
- Black CK, Termanini KM, Aguirre O, et al. Solid organ transplantation in the 21(st) century. Ann Transl Med 2018;6:409.
- Legislation Division of the Attorney-General’s Chambers of Singapore. Singapore Statues Online. HUMAN ORGAN TRANSPLANT ACT 1987 2020 REVISED EDITION. https://sso.agc.gov.sg/Act/HOTA1987?Timeline=Off. Accessed 25 July 2024.
- Legislation Division of the Attorney-General’s Chambers of Singapore. Singapore Statutes Online. MEDICAL (THERAPY, EDUCATION AND RESEARCH) ACT 1972 2020 REVISED EDITION. https://sso.agc.gov.sg/Act/MTERA1972. Accessed 25 July 2024.
- Kwek TK, Lew TW, Tan HL, et al. The transplantable organ shortage in Singapore: has implementation of presumed consent to organ donation made a difference? Ann Acad Med Singap 2009;38:346-8.
- Charles E, Scales A, Brierley J. The potential for neonatal organ donation in a children’s hospital. Arch Dis Child Fetal Neonatal Ed 2014;99:F225-9.
- Hawkins KC, Scales A, Murphy P, et al. Current status of paediatric and neonatal organ donation in the UK. Arch Dis Child 2018;103:210-5.
- de Vries EE, Snoeijs MG, van Heurn E. Kidney donation from children after cardiac death. Crit Care Med 2010;38:249-53.
- Miñambres E, Rubio JJ, Coll E, et al. Donation after circulatory death and its expansion in Spain. Curr Opin Organ Transplant 2018;23:120-9.
- Kwon JH, Blanding WM, Shorbaji K, et al. Waitlist and Transplant Outcomes in Organ Donation After Circulatory Death: Trends in the United States. Ann Surg 2023;278:609-20.
- British Transplantation Society. Transplantation from deceased donors after circulatory death. British Transplantation Society Guidelines. UK: Working Party of The British Transplantation Society; 2013.
- Rao V, Dhanani S, MacLean J, et al. Effect of organ donation after circulatory determination of death on number of organ transplants from donors with neurologic determination of death. CMAJ 2017;189:E1206-11.
- Tan J, Khalil MAM, Kee T, et al. Deceased donor kidney transplant policies in Asia – implications on practice and recommendations for the future. The Lancet Regional Health – Southeast Asia 2024;21:100312.
- Thong WY, Chong PH, Koh PL, et al. First pediatric organ donation after circulatory determination of death in Singapore: Facing challenges in the absence of a local practice guideline. Pediatr Transplant 2020;24:e13740.
- Kee T, Shridhar Ganpathi I, Sivathasan C, et al. Solid Organ Transplantation in Singapore. Transplantation 2018;102:1397-400.
- Ministry of Health, Singapore. Manual on Organ Donation and Transplantation. 6th ed. Singapore: Ministry of Health; 2024.
- Ministry of Health, Singapore. Circular No. 07/2024: Mainstreaming of kidney donation after circulatory death (DCD) in Singapore. 13 February 2024.
With advancements in medical care, mortality rates in critically ill patients have decreased substantially.1,2 Contemporary studies from developed countries report paediatric intensive care mortality rates of 2–3%3,4 in 2014 to 2019, whereas mortality in developing countries can be as high as 50%5 as of 2024, indicating large disparities in resources, expertise and infrastructure between high- and low-resource countries. Other evident differences include the causes of death—sepsis, multiorgan failure and pneumonia are leading causes of mortality in developing countries,6,7 whereas, cardiovascular disease (including congenital heart disease, cardiomyopathy and myocarditis) and neuromuscular disease (including encephalitis/ encephalopathy) are more prevalent causes of death in developed countries.3,4 Paediatric intensive care unit (PICU) registries are a useful means to track the epidemiology of PICU admissions and mortality characteristics. For example, the Virtual Pediatric Systems registry in the US,8 the Australian and New Zealand Paediatric Intensive Care Registry (ANZPICR)9 and the Pediatric Intensive Care Audit Network (PICANet)10 in the UK are established PICU registries with audit, quality improvement and research aims. There is currently a lack of such registry initiatives in other parts of the world including Asia. As a result, reliable year-on-year admission characteristics and mortality data are lacking.
As most inpatient paediatric deaths occur in the PICU, identification of potential organ donors is most actively conducted in there, which is a pragmatic representation of the potential paediatric organ donor pool.11 Besides being affected by prevailing mortality rates, organ donation in children is also diminished due to medical suitability (e.g. higher prevalence of inborn errors of metabolism and genetic conditions compared to adults) and parental values and beliefs.12 Singapore legislation requires parents or legal guardians of minors to give consent during an already stressful period for organ donation.13 There are limited reports on organ donation rates and missed donation opportunities in PICUs in Singapore and globally. By leveraging a recently established paediatric ICU registry, this study aimed to (1) describe the epidemiology of PICU admissions, (2) compare characteristics of survivors and non-survivors to identify risk factors for poor outcome, and (3) explore perimortem characteristics to independently determine the number of potential organ donors over the study period.
METHOD
Study design
This study employed an audit design, extracting data from the Singapore Pediatric Intensive Care Registry (SG-PedIC). We chose to perform a clinical audit from the SG-PedIC Registry as our primary aim was to describe the epidemiology of PICU admissions, deaths and organ donation candidacy to inform regarding the current standard of care delivered in our PICU. By adopting an audit design, all consecutive PICU admissions were included. The registry routinely and prospectively collects data on patient demographics, clinical characteristics, PICU therapies and outcomes from a single-centre PICU at KK Women’s and Children’s Hospital. This is a tertiary, university-affiliated paediatric centre in Singapore with a multidisciplinary PICU comprising 16 beds (including medical, surgical, oncology, neurosurgical and cardiothoracic surgical subspecialties). It is also the largest paediatric hospital in Singapore, serving the majority of the local paediatric population. As this study was deemed an audit, ethics approval was waived. No generative artificial intelligence (AI) or AI-assisted technologies were used.
All consecutive patients admitted to the PICU between July 2020 and December 2023 (3.5-year period) were included in this study. Patients’ characteristics including age, sex, weight, race, presence of comorbidities, admission details, Paediatric Index of Mortality 3 (PIM 3) score, PICU therapies and outcomes were extracted using a standardised case report form. The PIM 3 score that has been validated at our centre was calculated on admission to the PICU.14,15 Comorbidities were defined according to the list of chronic complex conditions and the main system affected was reported.16,17 The functional status scale was scored on admission to the PICU.18 Technology dependence was defined according to the list provided by the ANZPICR.9 Non-survivors were defined as patients who demised during in-patient admission or who were discharged directly home from the PICU for terminal extubation. Perimortem characteristics and cause of death were examined in non-survivors. In addition to routinely collected registry data, we independently identified potential organ donors based on published criteria; age (at least 2 months and above); and absence of any malignancy, disseminated infection, hepatitis C and HIV (Supplementary Table S1).19 Patients who were diagnosed with or suspected of having inborn errors of metabolism (IEM) or other genetic conditions were excluded. Once a patient was deemed a potential organ donor, perimortem case notes (up to 72 hours before terminal event) were reviewed to determine if the medical team identified organ donation eligibility and if the topic of organ donation was broached with the patient’s parents/legal guardian. Reasons for and against organ donation were identified and summarised.
Statistical analysis
The primary outcome of this audit was PICU mortality. We calculated the standardised mortality ratio by the ratio of the observed deaths to expected deaths based on the PIM 3 score at admission. Characteristics and PICU therapies of PICU survivors and non-survivors were compared using the Student’s t-test or chi-square test for continuous or categorical outcomes, respectively. Multivariable logistic regression was performed to identify independent risk factors for PICU mortality, and adjusted for baseline patient characteristics with significant univariate associations and clinical significance. Analysis was performed on STATA software version 15.1 (StataCorp, College Station, TX, US). All tests were two-tailed and P value <0.05 was accepted as statistically significant.
RESULTS
A total of 1766 patients were admitted to the PICU during the study period. The mean age of patients was 5.9 (±6.0) years (Table 1). Among them, 997/1766 (56.5%) were male, 946/1766 (53.6%) were Chinese and 514/1766 (29.1%) were elective admissions. Surgical admissions accounted for 707/1766 (40%) of total admissions, while the most common medical admission category was respiratory at 416/1766 (23.6%). The main sources of admission were the general inpatient ward/intermediate care unit at 550/1766 (31.1%), operating theatre at 541/1766 (30.6%) and emergency department at 487/1766 (27.6%). The majority of the patients who were admitted also had preexisting comorbidities at 983/1766 (55.7%), with cardiovascular and neurological/neuromuscular comorbidities being most prevalent in our study population.
Table 1. Comparison of characteristics of survivors and non-survivors admitted to the paediatric intensive care.
Patient characteristics | Total (n=1766) | Survivors (n=1667) |
Non-survivors (n=99) | P value |
Age, mean (SD), years | 5.9 (6.0) | 5.9 (5.9) | 5.6 (6.0) | 0.647 |
Age group, no. (%) | 0.209 | |||
Infant (1 month to <1 year) |
544 (30.8) | 506 (30.3) | 38 (38.3) | |
Child (1 year to <10 years) |
739 (41.9) | 706 (42.3) | 33 (33.3) | |
Adolescent (10 to <18 years) |
445 (25.2) | 418 (25.0) | 27 (27.2) | |
Adult (>18years) |
37 (2.1) | 36 (2.1) | 1 (1.0) | |
Weight, mean (SD), kg | 35.8 (568.9) | 36.6 (585.5) | 21.8 (22.9) | 0.801 |
PIM 3 score, mean (SD), % | 2.7 (7.3) | 2.3 (5.2) | 12.1 (22.5) | <0.001 |
Male, no. (%) | 997 (56.5) | 946 (56.8) | 52 (52.5) | 0.403 |
Race, no. (%) | 0.147 | |||
Chinese | 946 (53.6) | 901 (54.1) | 44 (44.4) | |
Malay | 521 (29.5) | 485 (29.1) | 36 (36.3) | |
Indian | 147 (8.3) | 135 (8.1) | 12 (12.1) | |
Others | 150 (8.5) | 144 (8.6) | 7 (7.0) | |
Any comorbidities, no. (%) | 983 (55.7) | 918 (55.2) | 65 (65.6) | 0.042 |
Neurological/ neuromuscular | 297 (16.8) | 279 (16.7) | 18 (18.1) | 0.709 |
Cardiovascular | 303 (17.1) | 279 (16.7) | 24 (24.2) | 0.054 |
Respiratory | 104 (5.8) | 96 (5.7) | 8 (8.0) | 0.340 |
Renal | 27 (1.5) | 25 (1.5) | 2 (2.0) | 0.682 |
Gastrointestinal | 84 (4.7) | 80 (4.8) | 4 (4.0) | 0.730 |
Haematology/immunology | 47 (2.6) | 45 (2.7) | 2 (2.0) | 0.683 |
Metabolic | 66 (3.7) | 63 (3.7) | 3 (3.0) | 0.703 |
Genetic/congenital | 185 (10.4) | 171 (10.2) | 14 (14.1) | 0.220 |
Neoplastic | 180 (10.1) | 169 (10.1) | 11 (11.1) | 0.756 |
Technology dependent, no. (%) | 312 (17.6) | 288 (17.3) | 24 (24.2) | 0.077 |
Elective, no. (%) | 514 (29.1) | 506 (30.4) | 9 (9.0) | <0.001 |
Admission category, no. (%) | <0.001 | |||
Cardiac surgical | 274 (15.5) | 262 (15.7) | 12 (12.1) | |
Cardiac medical | 131 (7.4) | 118 (7.0) | 13 (13.1) | |
Trauma | 48 (2.7) | 40 (2.4) | 8 (8.0) | |
Respiratory | 416 (23.6) | 388 (23.3) | 28 (28.2) | |
Neurological | 148 (8.3) | 139 (8.3) | 9 (9.0) | |
Surgical (non-cardiac) | 385 (21.8) | 379 (22.7) | 6 (6.0) | |
Other medical diagnoses | 361 (20.4) | 338 (20.3) | 23 (23.2) | |
Admission source, no. (%) | <0.001 | |||
Operating theatre | 541 (30.6) | 531 (31.9) | 10 (10.1) | |
Emergency department | 487 (27.6) | 460 (27.6) | 27 (27.2) | |
General ward/intermediate care unit | 550 (31.1) | 517 (31.0) | 33 (33.3) | |
Neonatal department | 44 (2.4) | 39 (2.3) | 5 (5.0) | |
Inter-hospital transfers | 89 (5.0) | 71 (4.2) | 18 (18.1) | |
Others | 53 (3.0) | 47 (2.8) | 6 (6.0) | |
Admission functional status scale, mean (SD) | 8.3 (3.8) | 8.3 (3.7) | 9.4 (4.5) | 0.003 |
Mental status | 1.1 (0.6) | 1.1 (0.6) | 1.3 (0.8) | 0.007 |
Sensory functioning | 1.2 (0.6) | 1.1 (0.6) | 1.3 (0.6) | 0.037 |
Communication | 1.2 (0.7) | 1.2 (0.7) | 1.4 (0.7) | 0.099 |
Motor functioning | 1.4 (1.0) | 1.4 (1.0) | 1.7 (1.2) | 0.001 |
Feeding | 1.6 (0.9) | 1.5 (0.9) | 1.8 (1.0) | 0.029 |
Respiratory status | 1.6 (1.2) | 1.6 (1.2) | 1.8 (1.4) | 0.083 |
PIM 3: Pediatric Index of Mortality 3
Respiratory support was the most common PICU therapy required by patients (invasive mechanical ventilation in 777/1766 [44.0%] and non-invasive respiratory support in 560/1766; 31.7%), whereas cardiovascular support was required in 484/1766 (27.4%) (Table 2). Non-survivors had increased frequency of use of all PICU support therapies, except non-invasive respiratory support. The mean length of PICU stay of our study population was 7 (±26.6) days, though 137/1766 (7.8%) patients had a prolonged PICU stay of more than 2 weeks. Patients admitted for more than 2 weeks were sicker (PIM 3 score mean standard deviation [SD] 5.4 [± 11.0]% vs. 2.6 [± 6.9]%; P<0.001); more likely to have an underlying chronic condition (100/137 [73.0%] vs. 884/1626 [54.4%]; P≤0.001); be technology dependent (48/137 [35.0%] vs 264/1629 [16.2%]; P<0.001); and have a poorer functional state (functional status scale, mean [SD] 10.8 [±5.1] vs 8.2 [±3.6]; P<0.001). When excluding these long stayers, the mean length of PICU stay was 3.6 (±2.5) days.
Table 2. Comparison of intensive care therapies in survivors and non-survivors.
Intensive care therapies | Total (n=1766) |
Survivors (n=1667) |
Non-survivors (n=99) |
P value |
Inotrope, no. (%) | 484 (27.4) | 402 (24.1) | 82 (82.8) | <0.001 |
Adrenaline | 295 (16.7) | 220 (13.2) | 75 (75.8) | <0.001 |
Noradrenaline | 247 (14.0) | 186 (11.2) | 61 (61.6) | <0.001 |
Milrinone | 242 (13.7) | 212 (12.7) | 30 (30.3) | <0.001 |
Vasopressin | 62 (3.5) | 19 (1.1) | 43 (43.4) | <0.001 |
Dopamine | 32 (1.8) | 21 (1.3) | 11 (11.1) | <0.001 |
Dobutamine | 12 (0.7) | 7 (0.4) | 5 (5.1) | <0.001 |
Continuous renal replacement therapy, no. (%) | 43 (2.4) | 20 (1.2) | 23 (23.2) | <0.001 |
Continuous veno-venous haemodialysis | 42 (2.4) | 19 (1.1) | 23 (23.2) | <0.001 |
Peritoneal dialysis | 4 (0.2) | 2 (0.1) | 2 (2.0) | <0.001 |
ECMO, no. (%) | 59 (3.3) | 28 (1.6) | 31 (31.3) | <0.001 |
Non-invasive respiratory support, no. (%) | 560 (31.7) | 533 (31.9) | 27 (27.2) | 0.329 |
Continuous positive airway pressure | 455 (25.8) | 443 (26.6) | 12 (12.1) | 0.001 |
Bilevel positive airway pressure | 326 (18.5) | 306 (18.4) | 20 (20.2) | 0.646 |
High flow nasal cannula | 47 (2.7) | 46 (2.8) | 1 (1.0) | 0.293 |
Mechanical ventilation, no. (%) | 777 (44.0) | 684 (41.0) | 93 (93.9) | <0.001 |
Duration | 3.0 (12.0) | 2.4 (10.5) | 13.6 (24.3) | <0.001 |
High frequency ventilation | 51 (2.8) | 29 (1.7) | 22 (22.2) | <0.001 |
Nitric oxide | 42 (2.3) | 26 (1.5) | 16 (16.1) | <0.001 |
Intracranial pressure monitor, no. (%) | 45 (2.5) | 38 (2.2) | 7 (7.0) | 0.003 |
ICU stay, mean (SD), days | 7.0 (26.6) | 6.4 (26.5) | 15.6 (27.8) | <0.001 |
Prolonged ICU stay, no. (%) | 137 (7.8) | 114 (6.8) | 23 (23.2) | <0.001 |
ECMO: extracorporeal membrane oxygenation; ICU: intensive care unit
Categorical variables are presented as counts, no. (%).
Continuous variables are presented as mean (standard deviation).
There were a total of 99/1766 (5.6%) deaths in this population (mortality rate for each calendar year ranged 5.1–6.8% with no statistical difference between years). The standardised mortality ratio was 2.1. There were no differences in demographics between survivors and non-survivors. However, compared to survivors, non-survivors were more likely to have an underlying comorbidity (65/99 [65.6%] vs 918/1667 [55.2%]; P=0.042), higher PIM 3 (mean [SD] 12.1% [±22.5] vs 2.3% [±5.2], P<0.001), and higher admission functional status scale (mean [SD] 9.4 [± 4.5] vs 8.3 [± 3.7]; P=0.003). There were also differences in admission categories and source of admission between survivors and non-survivors (Table 1). In the multivariable regression model, after adjusting for elective admissions and admission category, the presence of comorbidities (adjusted odds ratio [aOR] 3.03, 95% confidence interval [CI] 1.54–5.96), PIM 3 (aOR 1.06, 95% CI 1.04–1.08) and admission functional status scale (aOR 1.07, 95% CI 1.00–1.13) were independently associated with PICU mortality (Table 3). Among the non-survivors, the main causes of death include central nervous system infarction/ haemorrhage/hypertension (17/99; 17.1%) and pneumonia/chronic lung disease (15/99; 15.1%) (Table 4). There were also 30/99 (30.3%) of patients who required coroner’s referral. Most deaths occurred on maximal support 38/99 (39.1%) while brain death occurred in 6/99 (6.2%) of the patients. Majority (51/99; 52.0%) of patients who died had existing limitation of care orders.
Table 3. Multivariable logistic regression for paediatric intensive care mortality.
Covariate | Univariate | Multivariate | ||
OR (95% CI) | P value | aOR (95% CI) | P value | |
Comorbidities | 1.55 (1.01–2.38) | 0.043 | 3.03 (1.54–5.96) | 0.001 |
PIM 3 score | 1.06 (1.04–1.08) | <0.001 | 1.06 (1.04–1.08) | <0.001 |
Admission functional status scale | 1.06 (1.02–1.11) | 0.004 | 1.07 (1.00–1.13) | 0.039 |
Elective | 0.23 (0.11–0.46) | <0.001 | 0.42 (0.14–1.31) | 0.134 |
Admission category Ref: Surgical (non-cardiac) |
||||
Cardiac surgery | 2.89 (1.07–7.81) | 0.036 | 5.46 (1.55–19.26) | 0.008 |
Cardiac medical | 6.96 (2.59–18.71) | <0.001 | 3.23 (0.77–13.60) | 0.110 |
Trauma | 12.63 (4.17–38.24) | <0.001 | 11.09 (2.02–60.94) | 0.006 |
Respiratory | 4.56 (1.87–11.13) | 0.001 | 2.72 (0.78–9.53) | 0.117 |
Neurological | 4.09 (1.43–11.70) | 0.009 | 4.30 (1.07–17.35) | 0.040 |
Other medical diagnoses | 4.30 (1.73–10.68 | 0.002 | 2.99 (0.83–10.76) | 0.094 |
aOR: adjusted odds ratio; CI: confidence interval; OR: odds ratio; PIM 3: Pediatric Index of Mortality 3
Table 4. Perimortem characteristics of non-survivors.
Perimortem characteristics | No. (%) n=99 |
Limitation of care order | 51 (52.0) |
Any CPR in ICU | 37 (37.3) |
CPR as terminal event | 20 (20.8) |
Mode of death | |
Brain death | 6 (6.2) |
Death with maximal support | 38 (39.1) |
Death with therapy limited but not withdrawn | 23 (23.7) |
Death with therapy withdrawn (but not brain dead) | 30 (30.9) |
Cause of death | |
CNS infarction/haemorrhage/hypertension | 17 (17.1) |
Pneumonia/chronic lung disease | 15 (15.1) |
Cardiac failure/congenital heart disease | 10 (10.1) |
Sepsis/multiorgan dysfunction | 10 (10.1) |
Pulmonary hypertension | 5 (5.0) |
Malignancy | 4 (4.0) |
Others | 8 (8.0) |
Coroner’s referral | 30 (30.3) |
Organ donation | 2 (2.1) |
CNS: central nervous system; CPR: cardiopulmonary resuscitation; ICU: intensive care unit
Among non-survivors, 78/99 (76.8%) were deemed ineligible for organ donation— the most common reasons for ineligibility were disseminated infection (26/78; 33.33%) and presence or suspicion of IEM/genetic conditions (21/78; 26.9%) (Fig. 1). Of the remaining 21 patients who were independently deemed to be potential organ donors, 4 met brainstem death criteria but only 1 went on to donate organs after brainstem death certification. Six patients had severe neurological prognosis with incomplete brainstem death criteria—among these only 1 went on to donate their tissue after cardiac death. The medical team did not consider the others 11/21 (52.4%) who were potential tissue donors.
Fig. 1. Breakdown of the organ donation process.
DISCUSSION
In this single-centre audit of PICU admissions, we report a mortality rate of 5.6% and a standardised mortality ratio of 2.1. We identified the presence of comorbidities, higher PIM 3 score and functional impairment to be independently associated with mortality. The most common causes of death were neurological and respiratory. Of those who died, approximately 21% were eligible to donate their organs/tissue but only 2% went on to successfully donate their organs/tissue.
At 5.6%, the mortality rate reported in this study is slightly higher than those reported from developed countries including the US (1.9–3.4%),3 Australia/New Zealand (2.6%),20 UK (5.1%)21 and Japan (2.1%).4 The number of predicted deaths based on the PIM 3 score was lower, yielding a standardised mortality ratio of 2.1. This may indicate that PICU management and quality of care fell behind expected norms or may be contributed by deaths occurring in patients with existing limitation of care orders. However, there are additional considerations in interpreting the standardised mortality ratio. The PIM 3 score was derived from critically ill children admitted to PICUs from 2010–2011 in Australia, New Zealand, Ireland and the UK via the ANZPICR and PICANet registries, and its performance in predicting mortality may require recalibration in different regions.15,22 Our hospital also runs an active rapid response team (intra-hospital) and transport service (inter-hospital), which provides advice to referring physicians on stabilising potential PICU candidates before transfer—this may artificially lower the PIM 3 score, which is calculated based on variables obtained once the patient is admitted to the PICU. Physiological variables for the calculation of PIM 3 may also be unavailable for some children (e.g. partial pressure of arterial oxygen), and given that PIM 3 relies on default imputation in which missing observations are given a value considered “normal” for that variable, this could have artificially lowered the score.14 Lastly, though it is possible that a different severity of illness scores (e.g. PRISM23) may yield a more accurate predicted mortality rate in our unit, it was beyond the scope of the current study to compare the predictive ability of these commonly used scores in the PICU.
Higher mortality was observed in patients with existing chronic comorbidities, which is consistent with previous reports.24 These patients may have existing organ dysfunction and/or reduced cardiopulmonary reserve, potentially increasing their risk of death16 or may be admitted with existing limitation of care orders. Though slightly different from comorbidity per se, we observed that patients who had a higher functional status scale score (functionally impaired) also had a higher mortality rate. These patients were more likely to have a prolonged PICU stay, although the difference between these 2 groups of patients may not be clinically apparent. However, patients who were technology dependent (e.g. required long-term assisted feeding, home non-invasive respiratory support or long-term vascular device) did not have a significantly higher mortality rate. With advances in medical care, there is likely an increasing population of patients with chronic diseases, functional impairment and technology dependence who may be admitted to PICU.24 Data from this audit indicates that this population is vulnerable in the PICU, and further research is required to study their needs.
Despite advances in medical therapy, organ transplant remains the definitive, life-saving treatment for patients with irreversible or end-stage organ disease.25 In Singapore, organ donation is governed by the Human Organ Transplant Act26 and the Medical (Therapy, Education and Research) Act (MTERA),27 and is mainly applied to adults aged 21 years old and above.28 Paediatric organ donation may be applied for via MTERA, but it requires consent from the patient’s parents/guardian and is often hindered by cultural and personal beliefs.12 Here, we demonstrated that while 21% of our patients were deemed potential organ/tissue donors, up to 60% of these patients were not approached by the medical team. A single-centre retrospective study conducted in 2014 in the UK29 has similarly shown that up to 54% of their neonatal patients who died were potential donors. A more recent multicentre retrospective study conducted in 2019 in the UK30 had also reported that the majority of potential paediatric organ donors were not approached by the medical team. Both studies agree that with proper identification of potential organ donors and by approaching families respectfully, this niche group of patients may potentially augment nationwide donor pools. One concrete way to improve identification of organ donors is to have an independent team of medical staff screen patients for eligibility when death is anticipated, similar to how potential donors are screened in non-paediatric hospitals in Singapore. This differs from the current practice, in which transplant eligibility is determined by the medical team that is also in charge of the patient’s treatment (a suggested workflow is provided in Supplementary Fig. S1). Once eligibility is established, family discussions should include the donor coordinator, who is a resource person on organ donation. Organ donation may be beneficial to families by providing an option to give meaning to the loss of a child’s life, and if this is aligned with the values of the family, the managing team should incorporate their decision into the care plans.12 Ideally, discussions on organ donation should be part of end-of-life care for dying patients of all ages.12,30 Nonetheless, such conversations remain a challenge to clinicians, especially when families experience a sudden demise and such discussions may seem insensitive and even offensive, if done at the wrong time.
Organ donation after circulatory death (DCD)31 involves initiating organ donation after circulatory death is confirmed and has been shown to increase donor pools in countries such as Spain,32 US,33 UK34 and Canada.35 This is a potential way of increasing the paediatric pool of organ donors as well as for adults.36 Though the current legislation (i.e. MTERA) in Singapore allows for DCD and the first paediatric DCD in Singapore was reported in 2016,37 national guidelines with an operational framework for carrying out DCD have only just been established following a pilot38 and the guidelines were published in 2024.39,40 Given that the largest proportion of cases with missed opportunities for tissue/organ donation in this audit were cases of DCD, we anticipate that donation rates may improve with the publication of this guideline/framework and with experience. Organ transplant in Singapore remains a sensitive issue in a relatively conservative culture. More work needs to be done to increase the awareness of the benefits of organ donation for both recipient and donor among the general public, in order to effect a change in beliefs and values even prior to an ICU encounter.38
Though this is to our knowledge the first audit of a large database of critically ill children in Singapore, there are several limitations. The SG-PedIC registry was initially established in a single centre but has only recently become a national-level registry that includes the only other restructured hospital admitting paediatric patients in Singapore, the National University Hospital, in 2024. As such, national-level data were not available to be included in this report and data from this single-centre audit may have limited generalisability. Another limitation of this study was that cause of death was not captured in the registry in approximately 30% of deaths that were referred to the coroners. In determining organ donation eligibility, we have taken a simplistic approach that may have reduced the number of eligible donors—we did not consider organ-specific criteria (Supplementary Table S1) and we excluded all malignancies even though some malignancies may still be considered for organ donation. Lastly, despite minimal missing quantitative data, qualitative data of communication details on organ donation between healthcare providers and patients’ families may have been missed. As such, more granular details on the reasons for not considering organ donation candidacy by medical staff and the reasons for declining organ donation by parents were not available.
CONCLUSION
This single-centre audit of a multidisciplinary PICU reports a mortality rate of 5.6% and identifies the presence of comorbidities, higher PIM 3 score and functional impairment to be associated with mortality. To improve outcomes of the PICU, research is required to study the needs of patients with chronic diseases and functional impairment. We also found that close to 60% of potential organ donors were missed by the medical team indicating that more effort is needed to improve paediatric organ/tissue donation rates at our unit, which may include independent screening for eligibility, involvement of the donor coordinator in family discussions and establishing an operational workflow for donation after cardiac death.
Table S1. Summary of general and organ specific screening criteria.
Fig. S1. Suggested workflow and timeframe to identify potential organ donors and initiate organ donation processes.
REFERENCES
As this study was deemed an audit, ethics approval was waived.
An abstract of the current work has been presented at the Asia Pacific Intensive Care Symposium 2024. The authors declare they have no affiliations or financial involvement with any commercial organisation with a direct financial interest in the subject or materials discussed in the manuscript. The authors disclose no direct funding for this audit study. JJMW was supported by the Paediatric ACP/Nurturing Clinician Scientist Scheme (PAEDS ACP NCSS FY 2020), NMRC Research Training Fellowship (FLWP20nov_0002) and NMRC Transition Award (TA23jul-0001).
Dr Judith Ju-Ming Wong, Children’s Intensive Care Unit, KK Women’s and Children’s Hospital, 100 Bukit Timah Road, Singapore 229899. Email: [email protected]