CARDIOVASCULAR JOURNAL OF AFRICA • Vol 24, No 1, January/February 2013
108
AFRICA
costs of the programme, assuming an annual case volume of 100
operations, and including pre-operative evaluation and postopera-
tive care. Cost per year of life saved was calculated based on known
average life expectancy of less than two years for unoperated CHD,
versus an estimated postoperative life expectancy of 40 years, minus
our reported mortality of 7%.
Results
: The value of the donated medical supplies was $190 329
over five years. Total local programme costs averaged $216 600 per
year, making overall costs $2 800 per surgery or $3 011 per survivor.
Cost per year of life saved is $75.28.
Conclusion
: The cost of surgery for CHD falls well within the World
Bank’s definition of highly cost-effective interventions. Cost of
surgery for CHD compares favourably with other interventions such
as BCG vaccination for children, condom distribution to attempt to
prevent HIV, and general/trauma surgery. Future work should attempt
to incorporate disability data and cost of medical care for unoperated
children.
1614: IS IT DANGEROUS TO LIVE ON AN ISOLATED
ISLAND IFYOUAREA CHILDWITH CONGENITAL HEART
DISEASEWHO NEEDS OPERATION?
Inger Fosdal Tevebring
1
, Katharina Hanseus
2
1
Paediatric Clinic Visby Hospital, Sweden
2
Children´s Heart Centre, University Hospital, Lund, Sweden
Background:
Sweden has very good results for paediatric cardiac
surgery with a 30-day mortality of
<
2%. These results depend on a
functioning chain of care from the first suspicion of a congenital heart
defect (CHD) to primary examination, transportation, operation, peri-
operative care and follow up. In order to achieve safe treatment and to
optimise long-term prognosis, all these links have to be efficient and
of high quality all the time and for inhabitants in all geographic loca-
tions. Sweden has nine million inhabitants and 21 counties. The island
of Gotland, with its isolated location in the Baltic Sea is the smallest
county with approximately 57 000 inhabitants and 500 deliveries/year.
The paediatric cardiac service is provided by a paediatrician
with additional training in paediatric cardiology, and a radiologist
performing echocardiography. The hospital collaborates with the
Children’s Heart Centre in Lund for tertiary-level care, mainly paedi-
atric cardiac surgery and interventions. In addition, a senior consult-
ant from Lund visits Visby twice a year. The aim of this study was to
validate the results of paediatric cardiac care in Gotland.
Methods:
Retrospective studies were done of hospital files in Visby
and Lund for children (
<
18 years) born between 1 January 2000
and 31 December 2009, who had undergone treatment for cardiac
conditions.
Results:
Thirty-four children underwent surgery (31) or catheter
treatment (16) (one to five per patient). The median age at operation
was 4.6 months (range four days to 16.9 years). There has been no
surgical acute or late mortality during follow up (three to 18 years).
Forty children visited the centre on a total of 63 occasions: 1.5 visits/
patient (range 1–6); 25% of the visits were cardiac investigations
such as MRI or diagnostic catheterisation.
Conclusion:
Care of children with congenital heart disease can yield
excellent survival rates, even in an isolated area such as the Island of
Gotland. Close collaboration between all health professionals on the
local level and between the local hospital and the tertiary-level centre
are cornerstones for high quality.
1623: MULTINATIONAL, MULTILINGUAL VOLUNTEER
TEAMS OPTIMISE DELIVERY IN PAEDIATRIC CARDIAC
SURGICAL ASSISTANCE MISSIONS
Frank Molloy
1
, Martina Pavanic
1
, Jean Towne
1
, Siarhai Liauchonak
1
,
Pavel Shauchenka
1,3
, Eugene Suslin
1,3
, William Novick
1,2
, Elizabeth
Novick
1
1
International Children’s Heart Foundation, Memphis TN, USA
2
University of Tennessee Health Science Centre, Memphis TN, USA
3
National Children’s Cardiac Surgical Centre, Minsk, Belarus
Background:
The International Children’s Heart Foundation (ICHF),
a US-based non-governmental organisation (NGO), has operated
on over 6 000 children in collaborative educational programmes in
45 cities, in 35 developing countries since 1993. The majority of
programmes receive three or more trips per year in a partnership of
five or more years, which has been shown to be effective in increas-
ing surgical numbers and complexity. W e described the origin and
numbers of our volunteer pool and drew conclusions about the
sustainability of our assistance model.
Methods:
We analysed the ICHF volunteer database from 2012 back
to 1996. Of interest was city of origin, country of origin and volun-
teer participation from previously assisted programmes.
Results:
Over 16 years there were 260 trips, 1 124 volunteers, origi-
nating from 185 departure cities. Analysis of contributing institution
was not possible with accuracy but we can confidently infer that
this volunteer pool came from more than 185 contributing cardiac
centres, since many cities have more than one cardiac programme.
Volunteers were of 42 separate country origins, with 350/1 124
(31%) volunteers as non-US volunteers. Therefore a significant pool
of volunteers from close-proximity Spanish- and Russian-speaking
countries enabled cost efficiency and a working practice in the local
language. Of the 45 assisted sites, 11 eventually became volunteer
contributors to teams, with 66 volunteers from such programmes
contributing the knowledge and skills developed during our mutual
partnership to newer programmes.
Conclusions:
A mixed, multicentre, multinational volunteer pool, as
presented above, enables the delivery of multiple annual assistance
missions without excessive strain on any single centre. Furthermore,
it enables targeted optimisation of volunteers for location and
language skills, and sustains an ongoing partnership with ‘graduated
centres’ by utilising their staff as volunteers.
1624: DATA MINING WITH NATURAL LANGUAGE
PROCESSING FOR A PEDIATRIC CARDIOVASCULAR
PROBLEM LIST
Pouya Hemmati, Robert Bill, Serguie Pakhomov, Genevieve B
Melton, Lee A Pyles
University of Minnesota, USA
Background:
A problem list is a basic building block to ensure
effective medical management of children with cardiac diseases by
facilitating electronic health record (EHR) use. We hypothesised
that problem lists can be generated from discharge summaries using
natural language processing (NLP).
Methods:
We identified concept-unique identifiers (CUIs) for each
diagnostic term using Metamap software with standard settings.
After analysing Metamap output in the paediatric cardiovascular
domain, we extracted noun, verb, prepositional and adjectival phrases
using the Stanford parser, and weighted each phrase by occurrence
frequency and overlap with the Paediatric Cardiac Care Consortium
(PCCC) diagnosis list augmented with acronyms and lexical variants.
The weighted-phrases method was applied to discharge summaries
from 884 patients using the Biomedical Information Collection and
Understanding System (BiomedICUS). A random sample of 100
records was manually analysed to compare output of the weighted-
phrases method to primary cardiac diagnosis.
Results:
The Metamap analysis found 982 terms, of which 345 had
cardiovascular meaning; of these 205 did not map to any CUI. From
this we concluded Metamap was neither sensitive nor specific for
identification of paediatric cardiovascular diagnosis. The weighted-
phrases method found the primary cardiac diagnosis in the highest-
weighted phrase (report line 1) in 87/100 subjects (87% sensitivity).
Although 10 to 20 phrases per patient were available, these added
little to sensitivity beyond the first five. Sensitivity with the five
highest-weighted phrases was 97% for primary diagnosis.
Conclusion:
The low rate of CUI mappings to the PCCC diagnostic
list using Metamap indicates inadequate inclusion of paediatric cardi-
ac diagnoses in standard biomedical terminologies. Our NLP method
using phrases weighted by frequency and overlap with a paediatric
