Professional Documents
Culture Documents
Abstract
Introduction: Stroke is a major public issue in Singapore, accounting for almost 10,000 admissions annually and
the burden of disease is set to increase with an ageing population.
We therefore seek to examine the factors influencing the cost of acute stroke care in Singapore.
Methods: This is a retrospective analysis of 2,087 discharges with a primary discharge diagnosis of stroke from
a tertiary hospital in Singapore from 1 January 2007 to 31 December 2008. Data including age, gender, length
of hospital stay, components of direct cost and discharge disposition were obtained. A generalised linear model
with a log link function and gamma distribution was used to determine the predictors of total hospital cost.
Results: Mean age was 67.8 12.4 years and 54.5% were males. Mean length of stay was 12.3 16 days and mean
total overall cost was S$6,783. Ward costs accounted for 48% of total cost.
Length of stay strongly correlated to total cost. Being discharged to step-down facilities, death, receiving inpatient
rehabilitation and length of stay significantly incurred higher total cost in multivariate analysis. However, there was
an inverse relationship between age and total cost, possibly as a result of higher costs incurred for radiological,
laboratory investigations and expert care.
Conclusion: Further research is needed to examine factors influencing the cost of treatment particularly for
those being discharged to step-down facilities and receiving inpatient rehabilitation as they have been found
to incur higher total cost. This would impact on the planning of the continuum of healthcare facilities for
stroke management.
hospital stay, a determinant of acute cost7,8, while (whether the patient was discharged home
others reported on factors influencing the cost with or without outpatient follow-up with
of stroke care. However, factors influencing total primary care physician or the specialist, a step-
direct cost can vary from country to country and down rehabilitation facility, that is, community
even from centre to centre internationally916. This hospital or nursing home, transferred to another
observation can be attributed in part to variations hospital for continuation of acute management
in healthcare financing systems, differences or discharged against medical advice) as well
in clinical practices as well as differences as death status.
in study design.
Collection of Cost Data
With 1 in 5 residents projected to be 65 years and Actual cost incurred from each hospitalisation
above in 203017, there is an expected consequent upon discharge was extracted from the financial
increase in the prevalence of stroke. It is therefore database for the list of patients extracted using ICD
timely to examine the factors influencing the cost 9 codes. Hospital bills that were sent to patients
of acute stroke care in Singapore to provide local were not used for computation of cost as the
baseline information to healthcare providers and cost reflected on these bills typically reflect the
policy makers on the factors influencing the cost amount payable by the patient after subsidy by
of acute stroke management. This would be a the government. Patients who seek treatment at
preliminary study into identifying areas for further a public hospital could receive highly affordable
research on the organisation of care for stroke care to varying degrees as a result of government
patients in order to optimise the cost of ischaemic subsidies depending on the type of ward facilities
stroke care in an acute hospital. that they have chosen on admission. Only private
patients pay the full cost incurred.
We seek to examine if demographic factors such
as age, gender and ethnicity as well as if being The perspective of the healthcare provider
discharged to a step-down facility and receiving was taken in this study and only direct costs
inpatient rehabilitation in the acute hospital incurred during each hospitalisation at the
would influence the cost of acute ischaemic stroke full cost without government subsidy was
management in Singapore. included. Direct costs included cost of resources
utilised and services received. These would
Methods include ward charges, investigations, surgery,
Study Design rehabilitative therapy, drugs and doctors fee.
This is a retrospective analysis of patients with a The individual components were grouped as
diagnosis of ischaemic stroke who were discharged follows: (1) ward cost includes total operational
from Singapore General Hospital (SGH), a 1,400- charge, daily fee, facilities charges and ward
bed tertiary public hospital located centrally in charges; (2) radiology investigations includes
Singapore18, between 1 January 2007 and 31 X-rays and magnetic resonance imaging
December 2008. (MRI); (3) laboratory investigations includes
laboratory tests and special investigations;
Patients were identified using the International (4) cost of expert medical care are charges incurred
Classification for Diseases 9th Revision (ICD 9) codes when referrals are made to doctors outside of the
433.X, 434.X, and 436. Patients diagnosed with primary medical team caring for the patient and
transient ischaemic attack (TIA) and intracerebral include doctor fees, surgeon fees and anaesthetist
haemorrhage (ICH) were not included in this fees; (5) cost of services rendered includes ward
analysis. A total of 2,087 ischaemic stroke patients procedures, emergency services, implant fees and
were included in this study. rehabilitation; (6) medication cost includes medical
supply, medicines to take home and consumables;
Collection of Data and (7) miscellaneous cost includes items classified
Administrative databases were used to extract as miscellaneous and others.
demographic data including age, gender, and
ethnicity, length of stay (LOS), data on whether Cost data is presented in the original currency,
inpatient rehabilitation was provided during Singapore dollars (S$). (US $1 equals approximately
the period of hospitalisation, discharge location S$1.44 in 20072008).
Table 2. Relationship between baseline characteristics and total charges of hospitalization for ischaemic stroke patients.
(All values are expressed in Singapore Dollars.)
Mean Median
Std
N Total Total Min Max p-value Test
Deviation
Charges Charges
Age Group
<65 years 807 6,234.3 3,777.1 7,714.5 494.8 116,340 0.002 MWU
65 years 1,270 7,129.1 4,659.1 8,607.8 246.0 119,041
Gender
Male 1,138 7,137.2 4,318.1 9,359.6 246.0 119,041 0.454 MWU
Female 949 6,358.4 4,105.1 6,751.4 598.2 56,762.1
Ethnicity
Chinese 1,648 6,890.1 4,313.6 8,623.4 246.0 119,041
Malay 209 6,334.5 4,313.6 5,693.6 989.6 44,475.0 0.274 KW
Indian 156 6,108.4 3,710.1 7,622.1 454.9 54,495.6
Others 74 7,089.7 4,502.1 8,082.4 829.2 56,048.7
Discharge Destination
Discharged home 1,712 5,615.4 3,490.4 7,051.3 246.0 119,041
Discharged to step-down 0.001 KW
facilities 285 11,990 9,122.6 10,063.1 989.7 81,416.6
Death 90 12,508 6,544.5 13,528.4 1029.9 61,496.3
Patients Received Inpatient
Acute Hospital Rehabilitation
0.001 MWU
Yes 211 11,776 9,878.2 6,365.3 2360.2 34,900.8
No 1,876 6,221.5 3,735.4 8,288.2 246.0 11,9041
Total 2,087 6,783.1 4237.5 8283.4 246.0 119041
MWU Mann-Whitney U test; KW Kruskal-Wallis test
Table 3. Relationship between baseline characteristics and hospital Length of Stay (LOS) in days.
Median Std
N Mean LOS Min Max p-value Test
LOS Deviation
Age Group
<65 years 807 10.6 4 14.6 1 117 0.001 MWU
65 years 1,270 13.4 8 16.7 1 190
Gender
Male 1,138 12.6 6 17.4 1 190 0.727 MWU
Female 949 11.6 7 14.0 1 141
Ethnicity
Chinese 1,648 12.8 6 16.7 1 190
Malay 209 11.2 6 11.6 1 62 0.019 KW
Indian 156 10.0 4 14.5 1 105
Others 74 8.4 5 9.4 1 52
Discharge Destination
Discharged home 1,712 9.8 5 13.2 1 190
Discharged to step-down 0.001 KW
facilities 285 26.4 20 21.7 1 141
Death 90 14.9 8 18.4 1 113
Patients Received Inpatient
Acute Hospital Rehabilitation
0.001 MWU
Yes 211 26.6 23 12.7 2 72
No 1,876 10.7 5 15.5 1 190
Total 2,087 12.3 6 16.0 1 190
MWU Mann-Whitney U test; KW Kruskal-Wallis test
Length of stay was found to be strongly correlated Those who were <65 years were likely to incur 6%
to total hospital charges (R2=0.764). more total charges than those who were 65 years
[Cost Ratio (CR) = 1.060; p = 0.007]. Those who
Additionally, LOS was also significantly influenced underwent inpatient rehabilitation incurred 11.9%
by age group (p=0.01), ethnicity (p=0.019), (CR = 1.119; p = 0.003) more in total charges than
discharge destination (p=0.01) and whether those who did not. Patients who died and those
patient underwent inpatient rehabilitation were discharged to step-down facility incurred
(p=0.01). In particular, those who were discharged 75.8% (CR = 1.758; p < 0.001) and 17.3% (CR =
to a step-down facility stayed an average of 26.4 1.173; p < 0.001) more in terms of total hospital
days compared to 9.8 days for those who were charges respectively as compared to those who
discharged home. Those who underwent inpatient were discharged home.
rehabilitation in the acute hospital also stayed an
average of 26.6 days as compared to 10.7 days for An additional 4.8% (CR = 1.048; p < 0.001) of total
those who did not (Table 3). charges was also incurred with each additional day
of stay in the acute hospital (Table 4, overleaf ).
Generalised Linear Model (GLM)
The Generalised Linear Model (GLM) with log link Subgroup analysis on the effect of age group and
function and gamma distribution was used to sub-components of total cost.
estimate the total charges of acute hospitalisation.
Possible confounding variables such as age Further subgroup analysis was performed using
group, LOS, discharge destination, whether the above model to examine if there were any
patient underwent inpatient rehabilitation differences in cost incurred for those <65 years
were entered into the model. These variables compared with those aged 65years for the
were selected based on the results from the respective components of total hospitalisation
bivariate analysis. cost that might account for the inverse relationship
Table 4. Factors influencing total charges of hospitalisation using Generalised Linear Model (GLM)
with gamma distribution and log link function.
between total cost and age group. There are a rehabilitation and the discharge destination after
total of 7 cost components that contribute to adjusting for potential confounding factors. With
total cost, namely: (1) ward costs; (2) radiological increasing spending on healthcare both locally20
investigations; (3) laboratory investigations; and internationally21,22, our findings suggest that
(4) cost of expert care; (5) cost of services rendered; it is timely to examine in greater detail, factors
(6) medications; and (7) miscellaneous. Seven GLMs, influencing care transition and cost of care
one for each cost component, were constructed particularly for patients requiring discharge to
and analysed. The same possible confounding step-down facilities as well as those receiving
variables such as age group, LOS, discharge inpatient rehabilitation in the acute hospital. Length
destination, whether patient underwent inpatient of stay and delivery of care at the appropriate
rehabilitation were entered into the model. The facilities could be optimised through improved
reference age group in each model was the group planning and organization of the healthcare
that was 65years. delivery system.
It was found that those <65years were likely to incur Venketasubramanian et al reported in 2000 that
28.5% (CR = 1.285; p < 0.001) more for radiological the mean cost per discharge was S$7,547 with
investigations, 13.8% (CR = 1.138; p<0.001) an average length of stay of 17 days, whereas
(CR = 1.151; p = 0.027) more for expert medical care. the mean cost per discharge for our study was
S$6,783 and average length of stay was 12.3 days6.
Discussion The higher cost reported by Venketasubramanian
The number of hospitalisations as a result of et al could be a result of including the cost of
ischaemic stroke has been on the increase. In management of haemorrhagic strokes; strokes
20072008, the mean total cost of hospitalisation as a result of subarachnoid haemorrhage and
in our study was S$6,783 and the average length of intraparenchyma haemorrhage cost 4 times
stay was 12.3 days. Ward costs accounted for 48.1% and 2 times more respectively than those from
of the total cost. Length of stay accounted for non-lacunar infarcts in the former study (mean
76% of the variability in total cost. Length of stay cost per discharge were S$28,539 and S$14,398
was significantly different between patients with respectively versus S$7,476). The average length
different discharge destinations and dependent on of stay could have also been longer in the earlier
whether inpatient rehabilitation was provided. We study for the same reason; the mean length of
found that total cost was significantly influenced stay were 2 times and 1.5 times more respectively
by age, LOS, whether the patient received inpatient for strokes from subarachnoid haemorrhage and
23. Mamoli A, Censori B, Casto L, Sileo C, Cesana B, Camerlingo 27. van Exel J, Koopmanschap MA, van Wijngaarden JDH,
M. An analysis of the costs of ischemic stroke in an Italian Scholte op Riemer WJM. Costs of stroke and stroke
stroke unit. Neurology. 1999;53(1):1126. services: Determinants of patient costs and a comparison
24. Reed SD, Blough DK, Meyer K, Jarvik JG. Inpatient costs, of costs of regular care and care organised in stroke
length of stay, and mortality for cerebrovascular events services. Cost Eff Resour Alloc. 2003;1:2.
in community hospitals. Neurology. 2001;57(2):30514. 28. Ministry of Health. Parliamentary QA. Updates on clusters
25. Gubitz G, Phillips S, Aguilar E. Discharge disposition of [Internet]. Singapore: Ministryof Health. 2007 [cited
patients on an acute stroke unit. J Stroke Cerebrovasc Dis. 2010 Oct 27]. Available from: http://www.moh.gov.sg/
1999;8(5):3305. mohcorp/parliamentaryqa.aspx?id=23840.
26. Anderson C, Ni Mhurchu CN, Brown PM, Carter K. Stroke
rehabilitation services to accelerate hospital discharge
and provide home-based care: an overview and cost
analysis. Pharmacoeconomics. 2002;20(8):53752.