960
ORIGINAL ARTICLES
48
10
7
4
3
Redo transplants
Chronic rejection
Acute rejection
2
1
Other
Autoimmune hepatitis
Cryptogenic cirrhosis
Neonatal hepatitis
Budd-Chiari syndrome
Alagille syndrome
Congenital hepatic fibrosis
Hepatoblastoma
6
2
3
1
2
1
1
16
Results
Since 1987, 84 orthotopic liver transplants have been performed
on 81 children. All patients have survived the procedure and
currently 60 (74%) patients have survived 3 months to 14 years
post transplant. Fifty-one children have an excellent quality of
life.
The causes and timing of death are as follows: (i) early (b
1 month): sepsis (1), bleeding oesophageal ulcer (1), primary
non-function (1), inferior vena cava (IVC) thrombosis (1),
cerebral oedema (1); (ii) intermediate (> 1 - 6 months): rejection
and associated sepsis (1), portal vein thrombosis with variceal
bleed (1); (iii) late (> 6 months): bacterial, viral and fungal
infections (5), PTLD (7), chronic rejection (2).
Surgical complications
Hepatic artery thrombosis occurred in 2 patients and portal
vein thrombosis in 6 patients. Four patients developed early
portal vein thrombosis (< 6 months). In 2 patients, the portal
vein thrombosis was successfully repaired and the third
presented with an uncontrollable variceal bleed 6 months
post transplant. Attempts at a Meso-Rex shunt and redo
portal anastomosis failed in the fourth patient who remains
clinically stable on enoxaparin sodium (Clexane) therapy.
Two late portal vein thromboses occurred (> 1 year) resulting
in portal hypertension and recurrent GIT bleeds. They were
managed successfully with a Meso-Rex shunt. IVC thrombosis
occurred in 2 patients and this was treated with thrombolytic
therapy, which was successful in 1 patient (6 months post
transplant), but the second patient died from bleeding and
sepsis in the peri-operative period. Bile leaks occurred in 4
patients; in 2 these were successfully managed with revision of
the biliary anastomosis and in the other 2 patients, the biliary
leak was associated with a reduced-size graft and settled
961
ORIGINAL ARTICLES
Medical complications
Tuberculosis (3). Two children developed pulmonary
tuberculosis and 1 patient a pleural effusion. TB drug treatment
was complicated by chronic rejection in 1 patient and TB-druginduced subfulminant liver failure in 1 patient.
De novo hepatitis B (5). Hepatitis B virus infection is
endemic in South Africa with an HBsAg incidence of 3 - 20%
and a HB IgG core Ab incidence of 40 - 80%. Forty per cent of
the potential donors are HB IgG core Ab-positive. Prior to
1996, donors were only screened for HBsAg. Five children
(6 - 14 months post transplant) developed de novo hepatitis B.
All had been HBsAg-negative pre-transplant and the explanted
liver showed no evidence of hepatitis B infection. Three of
the children had received organs from Hep B IgG core Abpositive donors and sera was not available for testing in 2. Two
children have developed mild chronic hepatitis and 2 have
developed severe chronic hepatitis progressing to cirrhosis.
The DNA levels ranged between 1 920 and 4 800 pg/ml
(mean 3 556 pg/ml). All patients were treated initially with
famciclovir and then changed to lamivudine.12-14 Lamivudine
resistance developed in 2 patients and 2 have died as a result of
progressive chronic hepatitis B and complications of cirrhosis.
One died soon after diagnosis from TB-drug-induced liver
failure.
962
Survival
The overall cumulative 1-year and 5-year patient survival is
70% and 79% respectively. Since 1996, with the introduction of
IVI ganciclovir for 3 months and the exclusion of HBV IgG core
Ab-positive donors, the 1-year patient survival is 90% and the
projected 5-year survival is > 80%.
Discussion
As the only established paediatric liver transplant centre within
South Africa, children are referred from all over South Africa.
In the early stages of our programme, children were referred in
a pre-morbid clinical state and frequently died while awaiting
a liver transplant. However, with increasing public awareness
and education of the medical fraternity, children are now
increasingly being referred early on in the course of their liver
disease. Successful outcomes to liver transplantation however,
are dependent on both a committed family and committed
medical support, particularly when the child returns home to a
medical centre in another province.
Endemic viral and bacterial infections, particularly TB, CMV,
EBV and hepatitis B have had a significant impact on our
programme12,13,15-17 The recent ability to monitor CMV pp65
Ag and EBV polymerase chain reaction (PCR) has enabled us
to recognise viral infections earlier, treat appropriately and
decrease immunosuppression. The de novo hepatitis B in the 5
children was probably acquired as the result of using hepatitis
B core Ab-positive donors.18 Two children have died as a
result of chronic hepatitis B and 2 have developed lamivudine
resistance. Since 1996, hepatitis B IgG core Ab-positive donors
have been excluded as potential liver donors and no further
cases of de novo hepatitis B have occurred. The hepatitis B IgG
core Ab-positive rate in the donor pool is approximately 40%
and this together with the increasing prevalence of HIV has
significantly decreased the number of potential donors.
ORIGINAL ARTICLES
References
1. Robson SC, Spearman, CW, James MF, et al. Orthotopic liver transplantation at Groote Schuur
Hospital. S Afr Med J 1992; 82: 79-82.
2. Starzl TE, Demetris AJ, van Thiel D. Medical progress: liver transplantation Part I. N Engl J
Med 1989; 321: 1014-1022.
3. Starzl TE, Demetris AJ, van Thiel D. Medical progress: liver transplantation Part II. N Engl J
Med 1989; 321: 1092-1099.
4. Hover PF. Cyclosporin A (Neoral) in paediatric organ transplantation. Pediatr Transplant
1998: 2: 35-39.
5. Ganschon R, Broering DE, Stuerenberg I, Regiers X. First experience with basiliximab in
paediatric liver graft recipients. Pediatr Transplant 2001: 5: 357-358.
6. Whitington PF, Balistreri WF. Liver transplantation in paediatrics: indications,
contraindications and pre-transplant management. J Pediatr 1991: 118: 169-177.
7. Fishman JA, Rubin RH. Infection in organ-transplant recipients. N Engl J Med 1998: 338:
1741-1751.
8. Deen JL, Bumberg DA. Infectious disease considerations in paediatric organ transplantation.
Semin Pediatr Surg 1993: 2: 218-234.
9. Winston DJ, Wirin D, Shaked A, Busuttil W. Randomised comparison of gancyclovr and
high-dose acyclovir for long-term cytomegalovirus prophylaxis in liver-transplant recipients.
Lancet 1995: 346: 69-74.
10. Morgan G, Superina RA. Lymphoproliferative disease after paediatric liver transplantation. J
Pediatr Surg 1994: 29: 1192-1196.
11. McDiarmid SV, Jordan S, Lee GS, et al. Prevention and pre-emptive therapy of posttransplant lymphoproliferative disease in paediatric liver recipients. Transplantation 1998: 66:
1604-1611.
12. Millar AJW, Spearman W, McCulloch M, et al. Paediatric liver transplantation in a
developing country taking stock. Pediatr Transplant 2000; 4: 129.
13. Rayes N, Neuhaus R, Naumann U, et al. Treatment of hepatitis B reinfection or de novo
infection after liver transplantation with famiciclovir how effective is it. Transplant Proc
1998: 31: 481-482.
14. Grellier L, Motimer D, Ahmed M, et al. Lamivudine prophylaxis against reinfection in liver
transplantation for hepatitis B cirrhosis. Lancet 1996: 348: 1212-1215.
15. Goddard EA, Millar AJW, Spearman CWN, McCulloch MI, Kahn D. Cytomegalovirus (CMV)
infection in paediatric liver transplant recipients in a developing country. Pediatr Transplant
2000: 4: 48.
16. Spearman CWN, McCulloch M, Millar AJW, et al. Post-transplant lymphoproliferative
disease (PLTD) in paediatric liver transplant recipients in a developing country. Pediatr
Transplant 2000: 4: 53.
17. Meyers BR, Halpern M, Sheiner P, et al. Tuberculosis in liver transplant patients.
Transplantation 1994: 58: 3: 301-306.
18. Dickson RC, Everhart JE, Lake JR, et al. The National Institute of Diabetes and Digestive
and Kidney Diseases. Liver transplantation database. Transmission of Hepatitis B by
transplantation of livers from donors positive for antibody to hepatitis B core antigen.
Gastroenterology 1997: 113: 1668-1674.
963
ORIGINAL ARTICLES
964
Division of Paediatric Medicine, Red Cross War Memorial Childrens Hospital and
School of Child and Adolescent Health, University of Cape Town
Tony Westwood, FCP (SA) MD
Methods
Information was collected on all deaths occurring under the
services of the Red Cross War Memorial Childrens Hospital
(RCCH) from 1999 to 2003. The RCCH is a 288-bed state
teaching hospital with tertiary and regional functions. It is a
referral hospital but also houses 24-hour trauma and emergency
units, both of which have overnight inpatient beds. It is the
only paediatric inpatient facility for the central health districts
of Cape Town. Its tertiary services, including an intensive care
ORIGINAL ARTICLES
Results
There were 1 978 deaths registered at the RCCH. The annual
number rose from 367 in 1999 to 414 in 2003, an 11.4% increase
(Table I). One hundred and thirty-nine deaths occurred outside
the hospital (at home, dead upon arrival at the hospital, or
died at a primary care facility but registration took place at
RCCH). In addition, 172 deaths occurred in the emergency
room, leaving a total of 1 667 inpatient deaths. The absolute
number of inpatient deaths as well as the rate of death per
1 000 admissions increased from 303 deaths (15.9 per 1 000
admissions) in 1999 to 357 deaths (18.4 per 1 000 admissions) in
2002. The odds ratio of the death rate in 2002 compared with
1999 was 1.16 (95% condence interval (CI): 1.0 - 1.4, p = 0.052).
The rate dropped to 17.4 per 1 000 admissioins in 2003.
Female inpatients died at a greater rate than males (18.4
versus 17.6 deaths per 1 000 admissions, p = 0.007). Over
the 5 years, a female patient admitted to the hospital was
approximately 14% more likely to die than a male patient.
The absolute number of deaths declined as age rose (Table I).
More than half of all the deaths occurred in children who
came from three very low-income health districts in Cape Town,
viz. Khayelitsha (26.5%), Nyanga (20.1%) and Langa (11%).
The most common locations for deaths within the hospital
were the ICU (37%) and the medical wards (36%). The vast
majority (96.4%) of the deaths outside hospital occurred at
home.
GBD group I diseases accounted for the greatest proportion
of deaths (58.6%), followed by non-communicable diseases
(29.1%), and injuries (7.9%). Ill-dened deaths made up 4.3% of
the deaths (86 cases). Included in this category were 7 cases of
sudden infant death syndrome.
Overall the mortality prole was similar for males and
females. The prole of death for each of the age groups was
distinct (Fig. 1).
The group I causes of death were predominantly infectious
diseases (89.5%), with perinatal conditions contributing
5.7% and malnutrition 4.8%. Of the deaths attributed to
malnutrition, 60.7% were due to kwashiorkor.
<1
1-5
>5
Total
1999
2000
2001
2002
2003
Total
215 (58.6)
242 (61.7)
237 (60.0)
247 (60.2)
242 (58.6)
1 183 (59.8)
83 (22.6)
94 (24.0)
97 (24.6)
95 (23.2)
98 (23.7)
467 (23.6)
69 (18.8)
56 (14.3)
61 (15.4)
68 (16.6)
73 (17.7)
327 (16.5)
367
392
395
409
413*
1 977*
965
ORIGINAL ARTICLES
100
N = 1 183
N = 467
Neoplasms
12.2%
N = 327
Cardiovascular diseases
6.8%
90
70
Respiratory diseases
2.1%
60
Genito-urinary diseases
1.4%
I l-defined
Group III
Group II
Group I
50
Endocrine/metabo ic disorders
1.0%
30
20
10
<1
1-5
>5
Septicaemia
6.2%
Diarrhoeal diseases
8.2%
Meningitis/meningococcaemia
5.4%
Tuberculosis
4.0%
Myocarditis
2.6%
Hepatitis
1.3%
abnormalities 3.0
Respiratory abnorma
Other infection s
1.2%
n
s
966
HIV/AIDS
60.3%
Fig. 2. Infectious causes of death at the Red Cross Childrens Hospital from
1999 to 2003 (N = 1 038).
September 2006, Vol. 96, No. 9 SAMJ
ORIGINAL ARTICLES
9.00
8.00
7.00
6.00
5.00
.00
3.00
2.00
1.00
r
be
be
em
em
ec
D
ob
N
ov
ct
O
em
er
r
be
st
gu
Se
pt
Au
Ju
ly
ne
Ju
M
ay
ril
Ap
ry
br
ua
ar
Fe
nu
Ja
M
ar
ch
0.00
Month
Discussion
This study has shown the value of keeping routine data on
deaths in a hospital. The data were systematically recorded
over 5 years, allowing an analysis that can inform managers,
planners and clinicians. For example, the demonstration that
there is a tripling of in-hospital mortality from diarrhoea in the
summer months requires investigation. The magnitude of the
effect of HIV/AIDS on childhood in-hospital death demands
a response in terms of health resource provision. However,
although this study used both direct and underlying cause of
death, which allowed more detail than the Burden of Disease
studies for South Africa1 and for Cape Town,12 this method is
still limited by the lack of detail on other factors (e.g. health
service factors) that contributed to the deaths. Nevertheless
if simple data such as those collected here were routinely
recorded, valuable information on trends and patterns of
hospital-related deaths would be available. Indeed it is notable
that ill-dened causes of death were considerably fewer than
in other South African death certication studies,7,13 reecting
the potential value of hospital-related mortality data. Simple
guidelines on lling in the second page of the death notication
form have been supplied to most new medical staff at the
RCCH since the late 1990s. These have been reinforced by the
clerical staff responsible for collating the forms.
While the RCCHs large referral base complicates the drawing
of conclusions on the relevance of these data to child mortality
in general, comparison of these gures with a study of
mortality in Cape Town in 200112 shows that 1 in 6 of the 1 845
deaths under the age of 14 years in metropolitan Cape Town in
2001 were registered at the RCCH. Similarly the preponderance
of deaths from the metropolitan areas that refer children with
primary- and secondary-level health problems to the RCCH
enables conclusions to be drawn regarding common causes of
death in childhood in these largely low-income communities.
The leading causes of death in this study are congruent with
the South African and Cape Town Burden of Disease studies,12,14
supporting the potential utility of the RCCH data in drawing
conclusions on mortality in the surrounding population. The
differing rates of perinatal (lower) and congenital (higher)
causes of death reect the RCCHs position in the health care
system.
Causing at least 1 in 3 deaths, HIV/AIDS stands out as by
far the most common cause of death. This gure is likely to
be an underestimate of HIVs role in causing death as to some
extent doctors remained reluctant to certify it as the cause of
death.9,15 The predominance of HIV parallels the ndings of
the Burden of Disease studies of the South African Medical
967
ORIGINAL ARTICLES
968
ORIGINAL ARTICLES
969
ORIGINAL ARTICLES
Results
970
Outcome
Follow-up for survivors was conducted prospectively at the
neurosurgical clinic with a minimum follow-up period of 1
year; the median duration of follow-up was 35 months (range
1 - 6 years). All survivors had a favourable outcome: 6 patients
had a Glasgow Outcome Score (1 to 5) of 4, and 5 patients
had a score of 5. One patient developed bone flap sepsis after
the bone had been replaced, which required removal of the
flap. Another demonstrated slight subsidence of the flap at
follow-up. Asymptomatic subdural hygromas were observed
in 2 patients. Long-term behavioural disturbances and mild
cognitive deficits were relatively common (Table I).
Discussion
The use of decompressive craniectomy for the relief of
intracranial hypertension was originally described by Kocher
in 1905. Since then, its use in trauma has been surrounded
by controversy. The debate is largely centred on the issue of
whether it is of clinical benefit in TBI, and therefore, whether
its use justifies the additional effort and risk that the operation
may entail. There are also concerns about potential adverse
consequences of cranial decompression, such as the possibility
of increasing cerebral oedema beneath the bone flap, which
may further exacerbate the problem of raised ICP and
compromise tissue perfusion.
ORIGINAL ARTICLES
Age
Initial presentation
Initial imaging
MVA pedestrian,
GCS 8/15
11
Blunt assault,
GCS 14/15
12
11
MVA pedestrian,
GCS 5T/15
Fall from height, GCS
8/15, left pupil fixed
11
10
6.5
11
12
MVA pedestrian,
shocked, hypoxic,
GCS 6T/15
Deterioration
Right frontal
contusion, cisterns
open
Temporal contusion,
cisterns compressed
Swelling, cisterns
effaced
EDH
ICP uncontrolled
Cisterns open
Gunshot wound,
GCS 13/15
Repeat CT scan
Outcome
Nil
Good outcome.
Behavioural problems.
Slightly sunken flap
Cisterns obliterated Bone flap sepsis, mild
cognitive deficits,
special schooling
Increased swelling, Good outcome
compressed
obliterated
Increased swelling, Good outcome,
cisterns obliterated behavioural change
GCS 5T/15
Good outcome. CN VI
palsy unchanged
Good outcome, no
motor deficits
Died day 3 post-injury
MVA = motor vehicle accident; GCS = Glasgow Coma Score; CN = cranial nerve; EDH = extradural haematoma; PCA = posterior cerebral artery; VF = visual field; DAI = diffuse
axonal injury.
Brain entrapment
Another concern about the operation is the potential for
brain entrapment, which may occur when the oedematous
971
ORIGINAL ARTICLES
972
Fig. 1. A: Initial axial CT scan of patient 6 demonstrating a patent quadrageminal cistern and no signicant brain swelling. B: CT scan of the same patient
after clinical deterioration showing worsening brain swelling with cisternal obliteration; ICP baseline was 40 mmHg with spikes to 90 mmHg despite medical therapy. C: Axial CT scan after bifrontal craniectomy showing re-appearance of the basal cisterns. D: Lateral CT surview demonstrating the extent of the
craniectomy.
ORIGINAL ARTICLES
ICP reduction
The mechanical benefit of decompressive craniectomy is well
documented in the experimental and clinical literature. Because
the cranium represents a non-compliant system and cerebral
perfusion pressure is adversely affected by rising ICP, the
removal of a large bone flap expands the available volume
for brain swelling, increases the compliance of the system
and compensatory capacity of the cerebrospinal fluid axis,
and improves cerebral perfusion.6,12,13 It is also worth noting
the additional benefit obtained by expanding the rigid dura
mater. Yoo et al.8 demonstrated this graphically with an 85%
immediate reduction in ICP after craniectomy: 50% occurred
when the bone flap was removed and a further 35% when
the dura mater was opened and expanded. Other clinical
reports have also confirmed the 2-step reduction in ICP, almost
doubling the benefit with the dural augmentation.10
The mechanical benefits of craniectomy may be even greater
in children, as it has been demonstrated that children have
a lower pressure-volume index than adults, i.e. ICP is more
sensitive to changes of intracranial volume.14
Medical therapy
Although medical therapy is often effective, particularly in
patients who have a less aggressive course of elevated ICP,
the treating physician must be aware of the limitations of the
available options. External ventricular drainage is effective in
reducing ICP by releasing CSF from the ventricular system.
However, in patients with swollen brains the ventricles are
usually effaced and accurate placement of a catheter is often
impossible, the amount of CSF that can be drained is limited,
and the risk of infection with prolonged drainage must be
considered. Mannitol may also be effective in ICP reduction,
but usually for a limited period of time only. Repeated doses
may be associated with extravasation into the brain with
reversal of its effect15 and renal injury has been documented
even without elevated serum osmolarity.16 Hypertonic saline
may be safer than mannitol but fewer data are available on
its use. Barbiturates reduce ICP by decreasing metabolism,
but prolonged weaning from ventilation after discontinuation
and hypotension are common with its use, particularly with
the high doses required to achieve burst suppression on the
EEG.17 Ward et al.18 reported hypotensive episodes in more than
50% of subjects on barbiturates compared with 7% of controls.
Clinical outcome
Although TBI is a heterogenous condition, a number of factors
can be used to predict clinical outcome. The Glasgow Coma
Score is perhaps the most common of these. Marshall et al. 27
analysed the outcome of 746 patients with severe TBI from
the Traumatic Coma Data Bank (TCDB), including all patients
with an initial GCS of 8 or less, or who had deteriorated to
this level within 24 hours. Overall mortality was 36%. They
973
ORIGINAL ARTICLES
CT features
I
II
III
IV
Evacuated mass lesion
Non-evacuated mass
lesion
974
Interpretation of results
Although the outcome was favourable in this series of patients,
these results cannot necessarily be extrapolated across the
spectrum of head injury in children. With the exception of the
one patient who died, all patients in this series were selected
because the primary injury was thought to be salvageable,
but had been compounded by the secondary contribution of
raised ICP. This is key to understanding the effectiveness of
the operation despite the poor preoperative condition. In these
patients, uncontrolled raised ICP rather than the primary injury
was considered to be the prime determinant of mortality. This
situation should be clearly distinguished from patients who
present with a devastating primary brain injury that is unlikely
to be consistent with meaningful outcome. These patients
often present with posturing and pupillary abnormalities at
ORIGINAL ARTICLES
975