Long-stay children in intensive care: Long-term functional outcome

and quality of life from a 20-yr institutional study

Poongundran Namachivayam, MRCPCH, FCICM; Anna Taylor, BN, RN, Post Grad. Dip Women’s Hlth;
Terence Montague, FRCPCH; Karen Moran, BN, RN; Joanne Barrie, BN, RN; Carmel Delzoppo, BappSc;
Warwick Butt, FRACP, FCICM

Objective: Long-stay patients (≥28 days) in pediatric intensive were younger compared to survivors (median 3.4 months [interquartile
care units consume a disproportionate amount of resources, and range 0.38–41.5 vs. median 7.6 months, interquartile range 0.6–71.1,
very few studies have reported their outcome. We determined the p = .026], had a higher proportion of comorbid illness (91% vs. 80%,
long-term outcome of these children admitted to intensive care p = .026), and 63% had a preexisting moderate or severe disability
over a 20-yr period (January 1, 1989 to December 31, 2008). compared to 51% of survivors (p = .215). One hundred seventeen
Setting: Pediatric intensive care unit in a university-affiliated of 233 children survived and long-term functional outcome was
tertiary pediatric hospital in Melbourne, Australia favorable (normal, functionally normal, or mild disability) in 27% (63
Methods: Demographic and clinical characteristics were com- of 233) and unfavorable (moderate or severe disability) for 17.2% (40
pared after dividing patients into four groups depending on year of of 233). Outcome status was not known for 6% (14 of 233). Among
admission (1989–1993, 1994–1998, 1999–2003, and 2004–2008). survivors (n = 117), more than 50% (63 of 117) had favorable outcome.
Preadmission health status and long-term functional outcome The quality of life in patients aged >2 yrs at follow up was good in 21%
were evaluated by a modified Glasgow outcome scale. Quality of (40 of 222), moderate in 8% (16 of 222), poor quality in 68% (130 of
life was assessed by using the Health Utilities Index Mark 1. 222, this includes deaths), and very poor in 3% (5 of 222).
Results: Over the 20-yr period, 233 long-stay patients had 269 Conclusions: More than two-thirds of children who stay in inten-
long stay admission episodes to the pediatric intensive care unit, sive care for ≥28 days have an unfavorable outcome (moderate dis-
accounting for 1% (269 of 27,536) of all pediatric intensive care unit ability, severe disability, or death). Long-stay patients in pediatric
admissions and utilized 18.5% (15,740 of 85,032) of occupied bed intensive care utilized a large proportion of resources and this utili-
days. Bed occupancy of long stay patients (as percentage of overall zation has considerably increased with time. Service provision and
pediatric intensive care unit bed occupancy) increased from 8% in policy making should expect worsening of these trends in the future;
1989 to 21% in 2008 (p = .001). Median age at admission was 4.2 its effects on critical care bed availability and overall activity levels
months [interquartile range 0.38–41.5] and median length of stay could be substantial. (Pediatr Crit Care Med 2012; 13:520–528)
was 40 days [interquartile range 32–57]. One hundred sixteen of 233 Key Words: children; long-stay; outcome; pediatric intensive
(49.8%) patients had died at the time of follow-up. Children who died care

A lthough the majority of children

admitted to pediatric intensive
care units (PICUs) require
a stay of only a few days and
have a good outcome, a small proportion
of patients require prolonged periods of
Australia and New Zealand account for
<1.5% of intensive care admissions, but
utilize >20 % of intensive care bed days (1).
A number of previous studies have
reported prevalence, identified risk factors,
and predictors for long stay in intensive
cardiac surgery have similarly identified
risk factors for prolonged length of stay
(8) and have also shown that long stay in
intensive care is an independent predictor
of impaired cognitive function (9). A simi-
lar finding of intensive care length of stay
admission and often utilize considerable care (2–4) and have shown increased length associated with abnormal neurodevelop-
amounts of intensive care resources. Long of stay as a predictor of poor quality of life mental outcome has also previously been
stay patients (length of stay ≥28 days) in (5). Studies have also attempted to report reported in a diverse group of infants and
the long-term outcome of these children children post cardiac surgery (10).
From the Intensive Care Unit (PN, AT, KM, JB, (2, 6, 7). Pollack et al (2), using long stay Critical illness for long stay children
CD, WB), The Royal Children’s Hospital, Melbourne, as >13 days, reported that this subset of might represent only a temporary state;
Australia; Department of Anaesthesia (TM), Our children were significantly younger, had recovery can be prolonged and outcomes
Lady children’s Hospital, Crumlin, Dublin, Ireland;
and Department of Paediatrics (WB), University of
higher prevalence of chronic illness, and vary over time and with illness severity.
Melbourne, Melbourne, Australia. had high PICU mortality (17.4%). At 1-yr- While technology has allowed us to sup-
The authors have not disclosed any potential follow up 58 % had died or were severely port organ systems, they do not guaran-
conflicts of interest. disabled. More recently, Conlon et al (6), tee a recovery to health and a quality of
For information regarding this article, E-mail: siva.
defining long stay as >28 days, concluded life consistent with the family’s or care
namachivayam@[Link]
Copyright © 2012 by the Society of Critical that the mortality rate is high among long provider’s expectations. Evaluation of
Care Medicine and the World Federation of Pediatric stay patients but long-term health-related long-term outcome is important as this
Intensive and Critical Care Societies quality of life is normal for most. Investiga- information will be helpful to expedite
DOI: 10.1097/PCC.0b013e31824fb989 tors evaluating subgroups of infants post services within the intensive care unit,

520 Pediatr Crit Care Med 2012 Vol. 13, No. 5

assist in the counseling of families, and individual patient details in the intensive care Health Utilities Index: Mark 1 were defined
can also identify problems after intensive database were recorded and coded according to (Appendix questionnaire) as follows: 1) mobil-
care that can be solved in partnership guidelines produced by the Australian and New ity/physical activity; 2) self-care/role activity;
with other specialties. Zealand Pediatric Intensive Care Registry (15). 3) social/emotional function; and 4) health
The following data were recorded for the study: problems. The different levels within each cat-
In two previous studies from Mel-
age at admission, length of stay, principal di- egory were assigned a numerical value and an
bourne looking at cohorts of all consecu- agnosis, preadmission health status, reason for overall health state utility value (HSUV) was
tive admissions to intensive care in 1982 admission, comorbidities, treatment modali- calculated. The HSUV for all possible health
and 1995, we reported that the long-term ties used, number of other admissions to in- states lies between 1.00 and −0.21, where 1.00
outcome for ≥90% children at follow-up tensive care for each patient, and outcome (at is healthy, 0.00 is dead, and negative values re-
was favorable (11, 12) and these children discharge from PICU and long-term outcome). flect a state “worse than dead”. Four outcome
were likely to lead an independent exis- Subjects were divided into six major admis- categories were assigned: good (HSUV 1–0.7);
tence. More recently, in another study sion categories: cardiovascular, respiratory, moderate (HSUV 0.69–0.3); poor (HSUV
(13) comparing trends in intensive care neurological, accident/tuma, postoperative 0.29–0); and very poor (HSUV <0). Children
outcomes over three decades, we noted (non cardiac), and other [Link] presence <2 yrs of age at the time of follow-up were not
of comorbid illness was also identified using assigned a HSUV score. Outcome evaluation
an increase in the proportion of survivors
International Classification of Diseases, Ninth was performed after at least a minimum of 6
with moderate or severe disability from Revision, Clinical Modification diagnostic months post discharge from intensive care. A
8.4% in 1982 to 17.9 % in 2005–2006. codes and previously reported in literature 16. child is considered “lost to follow-up” if they
The purpose of this paper is to 1) study The nine major categories for comorbid illness are unable to be contacted, refused subsequent
the demographic and clinical profile of are: cardiovascular, respiratory, gastrointes- consent for interview, are an overseas patient,
long-stay patients; and 2) evaluate the tinal, renal, neuromuscular, hematology and or do not speak English. The relationship be-
long-term functional outcome and qual- immunodeficiency, metabolic, malignancy, tween age, diagnostic category, preadmission
ity of life among survivors. and other congenital or genetic defect. For health status, and “timing of outcome assess-
children who required more than one long ment” with outcome was studied and present-
stay admission, only details from the first ad- ed in the Results section.
METHODS mission were used in the study. For children
who died, the following data were collected: Statistical Analysis
Subjects and Design timing of death in relation to the admission,
place of death (PICU, ward, or home), and We divided children into four groups
The study population included all children mode of death (elective withdrawal for poor (based on the year of admission) to compare
who stayed ≥28 days between January 1, 1989, prognosis, brain death, death despite full treat- demographic and clinical characteristics and
and December 31, 2008, in a tertiary PICU ment, and other). mortality rate. Continuous variables are pre-
at Royal Children’s Hospital in Melbourne, sented as medians with interquartile ranges
Australia. Children who stayed in intensive Preadmission Health Status and (IQR) and categorical variables as percentages.
care for ≥28 days were classified as long stay pa- Continuous variables were compared with
Outcome Assessment Mann-Whitney U test (for two groups) or
tients in line with criteria set by the Australian
and New Zealand Pediatric Intensive Care Preadmission health status (before the on- Kurskal-Wallis test (for >2 groups), and cat-
Registry (14). Potential patients were identi- set of illness leading to PICU admission) and egorical variables were compared with chi-
fied from the database maintained in the PICU. long-term outcome were assessed by research square test. Survival curves were estimated
Data pertaining to the study and long-term assistants (T.M., A.T., K.M., J.B.) trained in with the use of Kaplan-Meier product-limit
outcome was collected in a prospective man- outcome evaluation. Preadmission health sta- method, and survival distributions between
ner. The PICU in Royal Children’s Hospital is tus and long-term functional outcome were groups were compared with the log rank test.
a 19-bed multidisciplinary unit with a catch- evaluated using a standardized questionnaire Data on subjects were censored at the time of
ment population of about 5.5 million adults to obtain a modified Glasgow Outcome Score last follow up or on January 1, 2011. Analysis
and children. The unit admits critically ill chil- (MGOS) (11). Outcome assessment was con- was performed using SigmaPlot version 11
dren from the states of Victoria, Tasmania, and ducted by means of telephone interview with (Systat software, San Jose, CA). All reported
southern New South Wales. Royal Children’s the child’s parent or guardian. Children who p values are two-sided; those under .05 are
Hospital, which has all major pediatric sub- reached adulthood at the time of evaluation considered to be statistically significant.
specialties, is the Australian national center were interviewed directly. The functional out-
for heart transplantation, and provides pedi- come obtained by the MGOS divides children RESULTS
atric extracorporeal life support (ECLS) for into the following categories: normal, func­
children from Victoria and for children from tionally normal (physically and intellectually
other states. This study was originally started normal) but requiring medication or medical Characteristics of the Study
10 yrs ago and was approved under the Royal supervision, mild disability but likely to lead Population
Children’s Hospital ethics committee “rapid an independent existence, moderate disability
approval pathway” process. In view of the and dependent on care, severe disability and During the study period, 233 patients
length of the study, the human research and totally dependent on care (including persistent stayed in intensive care for ≥28 days. The
ethics committee recently confirmed that the vegetative state), and death. Infants <1 month demographic and clinical characteris-
review has been carried out according to the of age at the time of PICU admission were not tics during the four 5-yr periods of study
current principles governing human research. included in preadmission health status assess- (1989–1993, 1994–1998, 1999–2003, and
ment. Secondly, quality of life in our study was
2004–2008) are presented in Table 1. The
Data Collection assessed by using the Health Utilities Index
Mark 1 by Torrance et al (17). More extended median age at admission for all patients
Data for this study were collected from versions of the Health Utilities Index (Mark 2 was 4.2 months [IQR 0.38–41.5 months]
multiple sources, which included the pediat- and Mark 3) have been developed, but were not and 60% (140 of 233) were male. Infants
ric intensive care database, medical records, used in our study as Mark 1 was the standard <1 month old at admission accounted
and cardiology departmental database. The in the 1980s and 1990s. Four attributes in the for 34% (79 of 233) of study population.

Pediatr Crit Care Med 2012 Vol. 13, No. 5521

Table 1.  Demographic and clinical characteristics compared between the four 5-yr periods

All Patients 1989–1993 1994–1998 1999–2003 2004–2008

Variable (N = 233) (N = 34) (N = 39) (N = 66) (N = 94) pa

Male sex – n (%) 140 (60) 22 (65) 18 (46) 43 (65) 57 (60.5) .24
Age – months .99
 Median 4.2 3.6 4 4.4 4.5
  Interquartile range (0.38–41.5) (0.5–23.2) (1.4–18.1) (0.3–44.4) (0.2–71.2)
Length of stay – days
 Median 40 37.6 43 41.4 40 .88
  Interquartile range (32–57) (31–60) (31.7–67) (32.7–56.7) (32.5–53.5)
No of long-stay admissions 269 47 46 73 103 —
Average no of admissions for long-stay 4.1 (978 of 233 7.1 (243 of 34 5.1 (200 of 39 3.9 (258 of 66 2.9 (277 of 94 < .001
patientsb (average/patient) patients) patients) patients) patients) patients)
Diagnostic category – no. (%) .008
 Cardiovascular 128 (55) 15 21 36 56
 Respiratory 43 (18) 9 9 12 13
 Neurological 21 (9) 5 6 5 5
  Other Medical 32 (14) 3 2 12 15
 Accident/trauma 7 (3) 2 1 — 4
  Postoperative (non-cardiac) 2 (1) — — 1 1
Long-term follow up, with deaths – n (%) 219 (94) 32 (94) 38 (97) 63 (95) 86 (91) —
Prior moderate-severe disabilityc 86/150 (57%) 11/23 (48%) 22/30 (73%) 23/40 (57%) 30/57 (53%) .21
Comorbid illness – n (%) 200 (86) 27 (79) 37 (95) 58 (88) 78 (83) .20
Intervention – no. (%) < .001
  Ventilated, no inotrope 35 (15) 9 (26.5) 12 (30.8) 8 (12.1) 6 (6.4)
  Ventilated, with inotrope 198 (85) 25 (73.5) 27 (69.2) 58 (87.9) 88 (93.6)
Tracheostomy – no. (%) 64 (28) 7 11 16 30 .55
Requirement for extracorporeal life 55 (24) 6 3 13 33 .004
support – no. (%)
Overall pediatric intensive care unit — 472 of 7351 (6.4%) 396 of 6514 (6%) 383 of 6823 (5.6%) 256 of 6848 (3.7%) < .001
Mortality rate – no. (%)
Long-stay patient mortality — 23 of 47 (49%) 21 of 46 (46%) 31 of 73 (42%) 41 of 103 (40%) .27
rate – n (%)

p values were calculated by Mann-Whitney U test, Kurskal-Wallis test, or chi-square analyses; bincludes both long stay and non-long stay admissions;
a

infants <1 month old not assessed for pre-admission health status.
c

Median length of stay in PICU for all Bed occupancy (occupied bed days) with cardiovascular illness from 44%
patients was 40 days [IQR 32–57 days] and is the length of stay computed in hours (15 of 34) in 1989–1993 to 59 % (56 of 94)
there was no difference in the length of and converted into days after dividing in 2004–2008. Long-term follow-up data
stay during the four time periods. Among by 24. The overall PICU bed occupancy (including deaths) for study patients were
children <1 month of age at admission, increased from an average of 3,912 days available for 94% of patients (219 of 233),
the median length of stay was much the per year in 1989–1993 to 5198 days per and >90% of patients during each study
same at 38 days [IQR 33–48 days]. year in 2004–2008 (p ≤ .001). Simi- period had outcome information. Among
Long stay patients during each time larly, the bed occupancy for long stay children in whom preadmission health
period were prone to PICU readmis- patients also showed a significant (p = status was recorded, 57% (86 of 150) had
sion and subsequent further episodes .014) increase (515 bed occupancy days moderate or severe disability before the
of long stay. In 1989–1993, an average per year in 1989–1993 and 1,036 in index admission to PICU. This estimate
of 1.38 admissions (34 patients need- 2004–2008). Long-stay patients utilized varied between 48% (11 of 23) in 1989–
ing 47 admissions) was noted. This ratio 8% of 4,363 PICU bed days in 1989 (Fig. 1993 to 53% (30 of 57) in 2004–2008
dropped during subsequent years and in 1), and in 2008 this increased to 21% of (p = .21).
2004–2008, an average of 1.09 admis- 5,263 PICU bed days, an increase of162 % The therapeutic intervention needed
sions (94 patients needing 103 admis- (p ≤ .001). Also, the proportion of long- was calculated based on the need for
sions) were needed. The total number stay admissions (expressed as a percent of mechanical ventilation and inotrope
of admissions per patient during each all PICU admissions) increased from 0.7% support. Of the 233 study patients, 35
period (both long stay and non-long stay) of 1,522 admissions in 1989 to 1.5% of (15%) were ventilated (without need-
was also estimated and there was a sig- 1,199 admissions in 2008 (p = .048). ing inotropes) and 198 (85%) needed
nificant downward trend: in 1989–1993, The major diagnostic category was both mechanical ventilation and inotope
34 patients accounted for 243 admission children with cardiovascular conditions infusion. In 1989–1993, 74% needed
episodes (7.1 admission episodes/patient) that accounted for 55% (128 of 233) of both ventilation and inotrope support
and in 2004–2008, 94 patients needed patients. Children with respiratory ill- and this increased to 94% in 2004–2008
277 admissions (2.9 admission episodes/ ness accounted for 18% of admissions, (p = .001). Similarly, the need for tracheos-
patient) to PICU (p ≤ .001). However, neurological illness 9%, and other medi- tomy and requirement for ECLS was also
changes in PICU admission, discharge, cal conditions accounted for 14% of studied. Overall, 28% (65 of 233) of long-
and transfer criteria over time could have patients. There was a trend toward an stay patients received a tracheostomy
contributed to these differences. increase in admissions among children and there was no significant variation

522 Pediatr Crit Care Med 2012 Vol. 13, No. 5

 of 233) subjects had comorbid illness and
the proportion varied between 79% (27
of 34) in 1989–1993 to 83% (78 of 94) in
2004–2008 (p = .20).The PICU mortality
rate decreased from 6.4% (472 of 7,351)
in 1989–1993 to 3.7% (256 of 6,848) in
2004–2008 (p ≤ .001). The PICU mortality
rate for long stay patients averaged 40%
or more during all four study periods
(p = .27).

Functional Outcome and Quality

of Life
Long-term outcome status (including
deaths) was available for 94% (219 of 233)
of the study population. One hundred six-
teen of 233 (49.8%) children died either
during or following their long stay admis-
sion. Of the survivors, outcome status
was favorable in 27% (63 of 233), who are
likely to lead an independent existence:
of which 5 (2.1%) were normal, 25 were
functionally normal (10.7%), and 33 had
mild disability (14.2%). 17.2% (40 of 233)
children had an unfavorable outcome
and were likely to survive dependant on
Figure 1.  Trends in long stay patient bed occupancy and long stay admissions during the study period. care by others for activities of daily liv-
PICU, pediatric intensive care unit. ing, moderate disability in 20 (8.6%) and
severe disability in 20 (8.6%). The out-
come status for 14 (6%) survivors is not
known. Among survivors (n = 117), >50%
(63 of 117) had favorable outcome.
Age at Admission, Diagnostic Cat­
egory, and Outcome. The association
between age at admission and diagnostic
category is shown in Figure 2. Of the 80
infants the in 0–1 month age group, 70
(87.5%) had a cardiac diagnoses. With
increasing age, the number of children
with a cardiac diagnosis decreased and
in children >5 yrs of age at admission it
was 35% (16 of 45). Children in the “other
medical” category contributed to 2.3%
(2 of 80) in the 0–1-month age group and
this increased to 38% (17 of 45) in children
>5 yrs of age. The association between
admission category and long-term out-
come is presented in Table 2. More than
50% of children in each of the diagnostic
categories died following their long stay
admission, except those admitted with a
neurological illness–24% (5 of 21) died.
However, among children who survived
Figure 2. Relation between age and diagnostic groups. Long stay children with accident/trauma with neurological illness 24% (5 of 21)
(n = 7) and noncardiac post operative patients (n = 2) not shown. had severe disability at long-term assess-
ment and this proportion was higher than
(p = .55) in the percentage of children was a significant increase (p = .004) in the that of other diagnostic categories.
who had a tracheostomy over the four number of long stay patients supported Outcome as a Function of Pread­
time periods of study. Children who were with ECLS over the period of study–6 mission Health Status. The relation-
supported with ECLS accounted for 24% of 34 (18%) in 1989–1993 and 33 of 94 ship between preadmission health status
(55 of 233) of the study patients and there (35%) in 2004–2008. Overall, 86% (200 (n = 143) and the long-term outcome of

Pediatr Crit Care Med 2012 Vol. 13, No. 5523

Table 2.  The relationship between reason for admission to pediatric intensive care unit and long-term survivors following their long-stay admis-
functional outcome (n = 210) sion. Of these, 63% (65 of 103) survived
with a favorable outcome (normal, func-
Outcome, n (%)
tionally normal, or mild disability) and
Normal and 37% (38 of 103) survived with unfavorable
Functionally Moderate Severe outcome (moderate or severe disability).
Admission Category Normal Mild Disability Disability Disability Death The proportion of children with favorable
and unfavorable outcome did not change
Cardiovascular (n = 117) 17 (14.5) 16 (14) 13 (11) 6 (5) 65 (55.5) (p = .995) in relation to the timing of
Respiratory (n = 41) 4 (10) 3 (7) 2 (5) 5 (12) 27 (66) outcome assessment (Fig. 3) after dis-
Neurological (n = 21) 3 (14) 7 (33) 1 (5) 5 (24) 5 (24)
Other Medical (n = 31) 7 (23) 2 (6) 3 (10) 2 (6) 17 (55)
charge from PICU. Thus the proportion
Total (n = 210) 31 (15) 28 (13.2) 19 (9) 18 (8.5) 114 (54.3) of favorable and unfavorable outcome was
similar whether outcome was assessed at
0–2, 2–5, 5–10 or 10–20 yrs after PICU
Table 3.  The relationship between preadmission health status long-term functional outcome (n = 143) discharge. However, because each patient
was only assessed once, individual out-
Outcome, n (%) come may have changed overtime.
Quality of Life. The Health State
Normal and Utilities Index was used to assess qual-
Functionally Mild Moderate Severe
ity of life in children ≥2 yrs of age at the
Preadmission Health Status Normal Disability Disability Disability Death
time of outcome assessment. Informa-
Normal + functionally 7 (19.4) 8 (22.2) 3 (8.4) 6 (16.7) 12 (33.3) tion was available for 191 of 222 (86%)
  normal (n = 36) eligible children who could be assessed.
Mild disability (n = 25) 4 (16) 4 (16) 2 (8) 1 (4) 14 (56) According to this measure, 40 (21%) had
Moderate disability (n = 68) 7 (10.3) 9 (13.2) 7 (10.3) 4 (5.9) 41 (60.3) a good quality of life (HSUV, 1.00–0.7), 16
Severe disability (n = 14) 1 (7.2) 1 (7.2) — 8 (57.1) 4 (28.5)
(8%) had moderate quality (HSUV 0.69–
Total (n = 143) 19 (13.3) 22 (15.4) 12 (8.4) 19 (13.3) 71 (49.6)
0.3), and 130 (68%) children (includ-
ing deaths) had a poor quality (HSUV,
0.29–0). Five children (3%) had very poor
quality of life (HSUV, <0). The association
Table 4.  The relationship between therapeutic intervention and long-term functional outcome (n = 217)
between quality of life and functional
Outcome, n (%) outcome among 75 eligible children who
had both measures ascertained is shown
Normal and in Table 5. More than 85% of children
Functionally Mild Moderate Severe who had poor quality of life also had
Therapeutic Intervention Normal Disability Disability Disability Death severe functional disability and all sub-
jects (n = 5) with very poor quality of life
Ventilated, no inotrope (n = 32) 5 (15.6) 9 (28.1) 3 (9.4) 5 (15.6) 10 (31.3)
Ventilated, with inotrope (n = 185) 25 (13.5) 23 (12.4) 17 (9.2) 14 (7.6) 106 (57.3) were severely disabled.
Mortality Characteristics of Long-
Stay Patients. One hundred sixteen
children of 233 (49.8%) with long stay
children evaluated by the MGOS is listed had a combination of severe brain injury in intensive care died. Fifty-four of 116
in Table 3. Infants <1 month of age on (3), cervical cord injury (3), and syn- (46.5%) were discharged from PICU after
admission to PICU were not evaluated drome association (3). Three of these five their initial long-stay admission but died
for preadmission health status. Fifty-six children had severe disability on pread- subsequently. The median age at admis-
percent (14 of 25) of children with mild mission health evaluation. A higher pro- sion for children who died was 3.4 months
disability and 60% (41 of 68) of those with portion of children in the “ventilated, (IQR, 0.2–22.5). This was significantly
moderate disability prior to admission to with inotropes” group died than those lower (p = .026) than that of children
PICU and who subsequently had a long in the “ventilated, no inotropes” group who survived their long stay admission
stay admission died. Although only 28.5% (57.3% vs. 31.3%, p = .01). Functional (7.6 months [IQR 0.6–71]). Ninety-one
(4 of 14) children admitted with prior outcome was known for 52 of 55 children percent (106 of 116) of children who died
severe disability died, the majority of chil- who required ECLS during their long stay had a comorbid illness compared to 80%
dren (57%, 8 of 14) in this group contin- admission: 69.2% (36 of 52) died, 7.7% (94 of 117) who survived (p = .026). The
ued to have severe disability at long-term (4 of 52) had moderate disability, 9.6% (5 median length of stay for children who
evaluation. of 52) had mild disability, and 13.5% (7 died (40 days [IQR 31–57]) was similar
Therapeutic Intervention and Out­ of 52) were either normal or functionally to that of children who survived (40 days
come. Association between main mode of normal. [IQR 33–58]). Among children in whom
intensive care therapy (mechanical ven- Outcome in Relation to “Timing of preadmission health status was available,
tilation and vasoactive agents) is shown Outcome Assessment”. Outcome was 51% (39 of 76) of those with moderate or
in Table 4. Among children in the “venti- evaluated at a median of 4 yrs (IQR 1.4– severe disability were alive and 63% (44 of
lated, no inotrope” group, 15.6% (5 of 32) 7.6) after discharge from PICU. Long- 70) had died (p = .215). Of 116 children
had severe disability: these five children term outcome was known for 103 of 117 who died, cause of death was identified as

524 Pediatr Crit Care Med 2012 Vol. 13, No. 5

 (9.1 months) and for the cardiology group
was 63.7 months (1,911 days).

DISCUSSION
This study shows that more than
two-thirds (67%) of children who stay in
intensive care for ≥28 days have an unfa-
vorable outcome (moderate disability,
severe disability, or death). A very high
proportion of long stay patients (86%)
have pre existing comorbid illness, and
the health care resource utilization for
long stay patients in pediatric intensive
care has increased considerably over the
last two decades as reflected in their PICU
bed occupancy--8% in 1989 to 21% in
2008. One of the strengths of our study
is that it involves a large cohort of long
stay patients in intensive care over a
20 yr period, with 94% outcome status
known for the entire cohort and among
survivors outcome ascertained for 88%
(103 of 117). Although it is difficult to
achieve high rates of outcome assessment
(because of frequent address changes,
privacy legislation, families maintain-
ing anonymity), it is important to follow
up every child in order to achieve high
outcome assessment rates (18), as this
enhances the reliability of results and
minimizes uncertainty. Many of the out-
come calls in our study were as a result
of multiple attempts to contact families.
Secondly, dividing our cohort into groups
based on year of admission was helpful
Figure 3. Relationship between timing of outcome assessment and outcome among survivors. for comparing trends and the long-term
Favorable outcome = normal, functionally normal, and mild disability. Unfavorable outcome = moder- follow-up has highlighted the very high
ate and severe disability. mortality and morbidity in this group of
children.
Table 5.  Relationship between quality of life and functional outcome among survivors (n = 75). Children This study evaluated the long-term
< 2 yrs of age at time of follow-up were not assigned a quality of life score functional outcome and quality of life of
long stay patients admitted to our PICU
Outcome, n (%)
over a 20-yr period. Two previous studies
Normal and of long stay patients have given us varying
Functionally Mild Moderate Severe results. A 2004 study by van der Heide et
Quality of Life Normal Disability Disability Disability al (7) was a case control study of long stay
patients involving only 19 subjects and
Good (n = 40) 22 (55) 17 (42.5) 1 (2.5) — reported no difference in mortality and
Moderate (n = 16) 1 (6.2) 7 (43.8) 7 (43.8) 1 (6.2) functional outcome compared to controls.
Poor (n = 14) — — 2 (14.3) 12 (85.7)
Very poor (n = 5) — — — 5 (100) Conlon et al (6) looked at long stay chil-
dren over an 8-yr period and reported high
mortality rates while the quality of life for
follows: elective withdrawal for poor prog- the hospital ward, and four in a different the majority of survivors was normal.
nosis (n = 80), cardiovascular collapse hospital. The place of death of four chil- However, the study population included
(n = 23), respiratory arrest (n = 2), elec- dren is not known. Figure 4 shows Kaplan preterm neonates and a 64.8% response
tive withdrawal for brain death (n = 3), Meier curves for survival for the differ- rate to the survey. In our study, the inclu-
and cause not known in eight (four died ent admission categories. Log rank test sion of long stay patients over two decades
at home and four in other hospitals). showed a significant survival difference and higher outcome assessment rates
Place of death: 93 children died in inten- (p = .036) among the different categories. allowed for meaningful comparisons and
sive care unit, seven died at home, four The median (50th percentile) survival potentially greater generalization to long
in the emergency department, four in for the respiratory group was 275 days stay patients in other units.

Pediatr Crit Care Med 2012 Vol. 13, No. 5525

 to determine predictors of poor outcome
would be a prospective study from mul-
tiple intensive care units using a standard
definition of long stay (22)
Families of children with prolonged
intensive care stays experience consider-
able burdens. Evidence from adult litera-
ture suggests that depression, financial
hardships, decline in physical health, and
high stress levels are very common among
families of adults with chronic critical ill-
ness (23, 24). Although evidence from
pediatric literature is limited, it is reason-
able to suggest that similar patterns or
more family burdens exist among parents
and families of long stay pediatric inten-
sive care patients. Parents of infants who
received surgery for hypoplastic left heart
syndrome have reported socioeconomic
difficulties, traumatic stress, and post-
traumatic stress disorder years after the
admission (A. M. Cantwell-Bartl, personal
communication, 2011). Traumatic stress
and post-traumatic stress disorder is
common among parents in PICU (25, 26)
and may persist well after discharge. More
studies looking at the impact on families
of long stay patients and their physical
and psychological outcome are needed.
One of the major implications of our
Figure 4.  Kaplan-Meier curves for survival time among patients in the different diagnostic groups. findings is the acceptance that long-stay
Log-rank test showed statistical significance between the survival curves (p = .036). p = .01 between patients are an integral part of intensive
respiratory and neurology groups and p = .08 between cardiac and neurology groups for survival.
care and their numbers along with resource
utilization will continue to increase in
A recent study looking at national esti- during the same time we recorded a sig- the future. As more resources are being
mates of comorbid illness in the United nificant increase in the number of long spent on long-stay patients, our models
States showed that 41% of children stay patients. One possible explanation for of care have changed to accommodate
admitted to critical care in 2006 had a this trend could be that patients with seri- this. When we plan strategy of care, we
comorbid condition. Presence of a comor- ous illness who would have been allowed recognize 15%–20 % of our bed days will
bid condition was significantly associated to die in previous epochs are kept alive be occupied by long-stay patients. We also
with a longer stay, increased mortality, now and many of these children continued have long-stay patient coordinators who
and greater hospital resource use (19). to have late mortality or experienced dis- work closely with the different team mem-
These children were also at increased risk ability. Clearly an increase in technology bers, families and other support services
of repeated admissions to intensive care (e.g., ECLS) and also family and society and this has been very instrumental in
(20). More than 85% of subjects in our expectations correspond to these changes. effective coordination of the care process.
study had a comorbid illness and the pro- Early identification of potential long stay
portion was significantly higher among patients would help patients and families Study Limitations
those who died. Successful weaning understand the implications of disease
from mechanical ventilation and other and potential outcomes, it would allow for Firstly, our study is from a single
intensive care therapies and even subse- proper planning of therapies and also help institution and the genralizability of our
quent discharge from PICU for a long stay to rationalize resource utilization. Length findings to other institutions could be
patient does not always guarantee long of stay is a marker of cost-efficient utiliza- questioned. However, previous studies in
term survival, as most of these children tion in intensive care units and a main tar- literature have recorded similar findings
have comorbid illness and many with get for cost containment in health care (3). (2, 6) in PICU outcomes among long stay
ongoing organ dysfunction and are prone However, in spite of different predictors of patients. While wide variations in out-
to recurrent hospitalizations. Almost half long stay being identified (21), their use- come may be noted among different insti-
of all deaths in our study population (54 fulness in making management decisions tutions, we believe our observed trends
of 116) involved children who were dis- is very limited. This possibly relates to would be a reasonable approximation of
charged successfully following their long multiple factors such as variations in case pediatric intensive care performance at
stay admission, but died subsequently. mix and illness severity, physician experi- least in the western society.
The overall PICU mortality has declined ence, prognostic uncertainty, and also the Second, outcome was assessed at
substantially over the last 20 yrs and expectations of the family. The ideal study only a single time point and there were

526 Pediatr Crit Care Med 2012 Vol. 13, No. 5

variations on the timing of outcome and families and the healthcare provider. 14. Alexander J, Slater A: Australian and New

assessment. Differences between time This important information must be con- Zealand Paediatric Intensive Care Registry
of PICU discharge and timing of out- sidered when talking with families and 2008
15. Slater A, Shann F, McEniery J; ANICS Study
come assessment could potentially result also for future planning of critical care
Group: The ANZPIC registry diagnostic
in variations in measured outcome. services. codes: A system for coding reasons for admit-
Improvement in quality of life overtime ting children to intensive care. Intensive Care
tends to occur as disease process changes ACKNOWLEDGMENTS Med 2003; 29:271–277
and caregivers get more accustomed to 16. Feudtner C, Christakis DA, Connell FA: Pedi-
patients needs (27, 28). Having more than We thank Professor Trevor Duke for atric deaths attributable to complex chronic
one outcome measurement for every reading the manuscript and providing conditions: A population-based study of
patient can give us information regard- useful suggestions. Washington State, 1980–1997. Pediatrics
ing temporal changes in functional status 2000; 106:205–209
and quality of life with time. However, the REFERENCES 17. Torrance GW, Boyle MH, Horwood SP: Appli-
cation of multi-attribute utility theory to
process of locating a family and gaining 1. Report of the Australian and New Zealand measure social preferences for health states.
an opportunity to communicate can often Paediatric Intensive Care Registry 2009, 2009 Oper Res 1982; 30:1043–1069
be very difficult, especially with privacy 2. Pollack MM, Wilkinson JD, Glass NL: Long- 18. Butt WW: Outcome assessment–what out-

legislation, frequent address changes, stay pediatric intensive care unit patients: come do we want? Pediatr Crit Care Med
and families willingness to maintain Outcome and resource utilization. Pediatrics 2009; 10:133–4; discussion 135
anonymity. 1987; 80:855–860 19. Odetola FO, Gebremariam A, Davis MM:

Thirdly, availability and utilization 3. Marcin JP, Slonim AD, Pollack MM, et al: Comorbid illnesses among critically ill hospi-
Long-stay patients in the pediatric intensive talized children: Impact on hospital resource
of rehabilitation and social services can
care unit. Crit Care Med 2001; 29:652–657 use and mortality, 1997–2006. Pediatr Crit
also influence outcome and quality of
4. Fiser DH, Tilford JM, Roberson PK: Relation- Care Med 2010; 11:457–463.
life overtime. In the small minority of ship of illness severity and length of stay to 20. Dosa NP, Boeing NM, Ms N, et al: Excess

children who need extensive rehabilita- functional outcomes in the pediatric inten- risk of severe acute illness in children with
tion services, individualized programs sive care unit: A multi-institutional study. chronic health conditions. Pediatrics 2001;
designed to achieve specific goals in Crit Care Med 2000; 28:1173–1179 107:499–504
areas like self care, mobility, and com- 5. Morrison AL, Gillis J, O’Connell AJ, et al: 21. Slonim AD, Marcin JP, Pollack MM: Long-stay
munication with active participation of Quality of life of survivors of pediatric inten- patients: Are there any long-term solutions?
families and caregivers can have signifi- sive care. Pediatr Crit Care Med 2002; 3:1–5 Crit Care Med 2003; 31:313–314
cant impact on measured outcomes. Our 6. Conlon NP, Breatnach C, O’Hare BP, et al: 22. Carson SS, Bach PB: The epidemiology and
Health-related quality of life after prolonged costs of chronic critical illness. Crit Care Clin
institution has a rehabilitation service
pediatric intensive care unit stay. Pediatr Crit 2002; 18:461–476
and this would have impacted on out-
Care Med 2009; 10:41–44 23. Van Pelt DC, Milbrandt EB, Qin L, et al:

comes measured. Future studies looking 7. van der Heide P, Hassing MB, Gemke RJ: Informal caregiver burden among survivors
at long-term outcome after discharge Characteristics and outcome of long-stay of prolonged mechanical ventilation. Am J
from intensive care should also sample patients in a paediatric intensive care unit: Respir Crit Care Med 2007; 175:167–173
the potential effects of these services on A case-control study. Acta Paediatr 2004; 24. Douglas SL, Daly BJ: Caregivers of long-term
outcomes. 93:1070–1074 ventilator patients: Physical and psychologi-
Fourth, one could question the reli- 8. Brown KL, Ridout DA, Goldman AP, et al: cal outcomes. Chest 2003; 123:1073–1081
ability of MGOS in assessing functional Risk factors for long intensive care unit stay 25. Balluffi A, Kassam-Adams N, Kazak A, et al:
outcome in infants and children. In view after cardiopulmonary bypass in children. Traumatic stress in parents of children
of the diversity of age, generic tools with Crit Care Med 2003; 31:28–33 admitted to the pediatric intensive care unit.
9. Newburger JW, Wypij D, Bellinger DC, et al: Pediatr Crit Care Med 2004; 5:547–553
broad classifications were required to
Length of stay after infant heart surgery is 26. Bronner MB, Peek N, Knoester H, et al:

provide global assessments of functional related to cognitive outcome at age 8 years. J Course and predictors of posttraumatic stress
outcome. MGOS is a practical and cost- Pediatr 2003; 143:67–73 disorder in parents after pediatric intensive
effective method of obtaining health 10. Limperopoulos C, Majnemer A, Shevell MI, care treatment of their child. J Pediatr Psy­
information and also has wide recogni- et al: Predictors of developmental disabilities chol 2010; 35:966–974
tion with high clinical relevance (29) after open heart surgery in young children 27. Knoester H, Bronner MB, Bos AP, et al: Qual-
along with a very high interrater reliabil- with congenital heart defects. J Pediatr 2002; ity of life in children three and nine months
ity when used by experienced staff (30). 141:51–58 after discharge from a paediatric intensive
11. Butt W, Shann F, Tibballs J, et al: Long-term care unit: A prospective cohort study. Health
outcome of children after intensive care. Crit Qual Life Outcomes 2008; 6:21
CONCLUSION
Care Med 1990; 18:961–965 28. Ambuehl J, Karrer A, Meer A, et al: Quality of
The number of children requiring pro- 12. Taylor A, Butt W, Ciardulli M: The functional life of survivors of paediatric intensive care.
longed intensive care admission and their outcome and quality of life of children after Swiss Med Wkly 2007; 137:312–316
admission to an intensive care unit. Intensive 29. Wilson JT, Pettigrew LE, Teasdale GM:

resource utilization has increased con-
Care Med 2003; 29:795–800 Structured interviews for the Glasgow
siderably over the last two decades. More
13. Namachivayam P, Shann F, Shekerdemian Outcome Scale and the extended Glasgow
than two-thirds of children requiring ≥28 L, et al: Three decades of pediatric inten- Outcome Scale: Guidelines for their use. J
days stay in PICU have an unfavorable sive care: Who was admitted, what hap- Neurotrauma 1998; 15:573–585
outcome (moderate or severe disability pened in intensive care, and what happened 30. Epstein AM: The outcomes movement–will it
or death). This major finding has impor- afterward. Pediatr Crit Care Med 2010; get us where we want to go? N Engl J Med
tant implications both for the children 11:549–555 1990; 323:266–270

Pediatr Crit Care Med 2012 Vol. 13, No. 5527

APPENDIX: OUTCOME 1. Integration aide those health problems that your
ASSESSMENT QUESTIONNAIRE 2.  Special needs school child experienced prior to admis-
3. Rehabilitation (including phys- sion to hospital? Yes/No. If yes,
1. Mobility/Physical Activity iotherapy, occupational therapy, please specify.
a. Does your child have any limitations speech therapy) c. Does your child require regular fol-
in regard to walking/running/jump- 4. None low-up by a specialist doctor? Yes/
ing? Yes/No. If yes, please specify. d. If your child attends school, how No. If yes, how often?
b. Does your child experience any unex- much school would your child miss d.  Does your child require regular
pected breathlessness/tiredness when throughout the year? medication? Yes/No If yes, please
playing with other children or exer- 1. Less than 1 week specify
cising? Yes/No. If yes, please specify. 2.  Between 1 and 2 weeks e. Does your child experience any pain/
c. Is help required from other people, 3.  Between 2 and 4 weeks discomfort on a regular basis? Yes/
or from mechanical aids (wheel- 4.  More than a month No. Specify location and frequency.
chair, frame), for your child to move e. Parent’s perception of child’s ability
around? Yes/No. If yes, please specify. f. Does your child have any vision
to cope with school level. problems? Yes/No. If yes, please
d. Parent’s perceptions of child’s phys-
ical abilities in relation to other specify.
3. Social/Emotional Function   Does your child need to wear glasses?
children of a similar age. a. During an average day is your child Yes/No
generally
g. Does your child have any hearing
2. Self Care/Role Activity 1. Happy
difficulties or do you have any con-
Please circle the most appropriate 2. Anxious
cerns regarding your child’s hear-
a. How much help does your child 3. Depressed
ing (exclude selective deafness)?
need to eat/dress/bathe/toilet (as age 4. Aggressive
appropriate) Yes/No
b. Does your child have any problems
1. No help in making and maintaining friend-  Does your child require a technical
2.  A small amount of help ships? Yes/No aide (hearing aide, cochlear implant,
3.  A moderate amount of help c. Parents concerns with respect to other) to hear? Yes/No
4. Totally dependent upon help child’s behavior. Please specify.  Did your child have any hearing dif-
from another person ficulties prior to admission to hospi-
b. Which of the following does your 4. Health problems tal? Yes/No
child attend? a. Has your child developed any new h. Does your child have any scars that
1. Kindergarten health problems since your admis- have healed poorly, or have any
2.  School Year sion to hospital? Yes/No If yes, please other physical problems that cause
3. None specify. your child, or you, concern? If yes,
c. Does your child need any of the fol- b. Has there been any change in the please specify.
lowing support? management, or the severity of i.  Additional comments:

528 Pediatr Crit Care Med 2012 Vol. 13, No. 5