Epidemiol. Infect. (2004), 132, 211–221.

f 2003 Cambridge University Press

DOI : 10.1017/S0950268803001559 Printed in the United Kingdom

Costs of gastroenteritis in The Netherlands

W. E. V A N D E N B R A ND HO F 1*, G. A. D E W IT 2, M. A. S. D E WI T 1
1
A N D Y. T. H. P. V A N DU Y N HO V E N

1
Department of Infectious Diseases Epidemiology, National Institute of Public Health and the Environment
(RIVM), Bilthoven, The Netherlands
2
Department for Health Services Research, National Institute of Public Health and the Environment (RIVM),
Bilthoven, The Netherlands

(Accepted 17 September 2003)

SUMMARY
In order to target the most important cost components of gastroenteritis in The Netherlands
and to indicate which change of policy yields the largest decrease in costs, the cost of illness of
gastroenteritis and the number of Disability Adjusted Life Years (DALYs) in the Dutch
population in 1999 were determined. The costs of gastroenteritis were estimated using data
mainly from a community-based cohort study. For calculating DALYs, data on the number of
deaths due to gastroenteritis were used from Statistics Netherlands. On average, the costs for
gastroenteritis were 77 Euro (E) per case. For all patients in The Netherlands, the costs were
estimated at E345 million (ranging between E252 and E531 million). Indirect costs made up 82 %
of this total. An estimate of costs for patients with campylobacter, salmonella or norovirus
infections was, in total, 10–17% of the costs of gastroenteritis. Gastroenteritis was associated
with a loss of approximately 67 000 DALYs.

INTRODUCTION the total costs of gastroenteritis have not been esti-

mated previously.
In The Netherlands, with a population of 16 million
The objective of this study was to determine the
inhabitants, approximately 4.5 million episodes of
costs of gastroenteritis in The Netherlands in 1999,
gastroenteritis occur every year [1]. Gastroenteritis is
according to the bottom-up method (in which cost
usually self-limiting and in general does not lead to
data of specific age groups are determined and costs
high costs for an individual. However, due to the high
are aggregated over groups to calculate the total
annual number of individuals affected, the total costs
costs). In addition to the costs of gastroenteritis in
can be substantial [2–4]. Estimates of the costs can be
general, we also estimated the respective costs for
used to provide information on the burden of the
gastroenteritis due to the top three of foodborne
disease, to compare the costs of different public health
pathogens: campylobacter, salmonella and norovirus.
interventions, to target the most important cost com-
Furthermore, an estimate was made of the number of
ponents and to indicate which change of policy yields
Disability Adjusted Life Years (DALYs) in the Dutch
the largest decrease in costs [5]. In The Netherlands,
population, in order to provide insight into the public
health burden of gastroenteritis and to compare this
* Author for correspondence : W. E. van den Brandhof, National
Institute of Public Health and the Environment (pb 75), P.O. Box 1,
with the burden of other diseases. The DALYs com-
3720 BA Bilthoven, The Netherlands. bine the effect of morbidity and mortality and has

https://doi.org/10.1017/S0950268803001559 Published online by Cambridge University Press

 212 W. E. van den Brandhof and others

been adopted as a basis for Dutch public health daily about specific gastrointestinal symptoms, bed
policy [6]. rest, absence from work or school (of the patient
or someone taking care of the patient), use of medi-
cation, and use of the health-care system (GP, out-
METHODS patient department, hospital, alternative treatment).
It was clearly stated that only medical care use,
Data sources
absence, or bed rest due to their gastroenteritis epi-
To calculate the cost of illness from the societal per- sode should be reported. All gastroenteritis cases
spective, we needed data on the total number of cases from the CB study who completed the medical diary
of gastroenteritis per age group, the volumes for use were included in the analyses for this study. Gastro-
of resources such as medication, health services and enteritis was defined as : three or more loose stools
the indirect costs, and the actual economic costs of in 24 h, or three or more times vomiting in 24 h, or
each of these items. With this, the average costs per diarrhoea with at least two additional symptoms,
case in a certain age group could be calculated and or vomiting with at least two additional symptoms.
subsequently multiplied by the total number of cases Additional symptoms could be abdominal pain,
in each age group in The Netherlands. Then, the total abdominal cramps, nausea, blood in stool, mucus in
absolute costs of gastroenteritis could be calculated stool, fever, diarrhoea or vomiting.
by summing costs over all age groups. To calculate Since all stool samples in the CB study were tested
DALYs, we needed age-specific data on the number for enteropathogens, with the test results available
of deaths of gastroenteritis patients, duration and to GPs, the frequency of requesting stool tests by the
severity of symptoms and the disability weight for GP could not be determined in this study. However,
gastroenteritis in The Netherlands. We needed the in 2001, GPs of the sentinel network of NIVEL
information per age group, because of differences in completed a questionnaire whenever they ordered
incidence and type of costs (e.g. absence from work) laboratory diagnostics for a gastroenteritis patient.
between age groups. They also reported all consulting cases. These data
Results from a community-based prospective were used to estimate the order frequency of a GP
cohort study (CB study) on gastroenteritis were used for laboratory diagnostics. The age groups on the
to calculate the incidence and the absolute number of NIVEL questionnaire were just slightly different (e.g.
gastroenteritis patients per age group. This study was 10–19 years instead of 12–17 years) from the cat-
performed in The Netherlands in 1999, in cooper- egories of the CB study and were therefore used as
ation with the sentinel general practice (GP) network proxies for our age groups.
of The Netherlands Institute for Health Services To examine whether the number of cases admitted
Research (NIVEL) [1]. This network consists of ap- to a hospital, a rather rare event, in the CB study
proximately 44 practices that cover 1 % of the Dutch differed from the number expected, we used data
population, representative regarding age, gender, from the National Medical Register (NMR) in The
regional distribution, and degree of urbanization. In Netherlands. This registry collects data on hospital
short, the cohort consisted of two consecutive, age- discharge diagnoses from all hospitalized patients in
stratified samples of individuals registered at GPs in The Netherlands, including information on duration
this network. Twenty-seven GPs participated in the of stay. Hospitalizations with a primary or any of the
first cohort and 31 in the second. Persons were invited secondary diagnoses coded with ICD-9 codes 001-009
by mail, and after 3 weeks, a reminder was sent to and 558 (codes related to gastroenteritis) [7, 8] for
non-respondents. At the start of the follow-up, all 1999 were obtained for analyses. The costs for only
participants in the cohort completed a baseline ques- hospital admissions with primary diagnosis gastro-
tionnaire. Cases with gastroenteritis identified in enteritis were calculated separately.
the cohorts during the 6 months of follow-up were
requested to submit stool samples, complete a ques-
Costs
tionnaire on risk factors, and complete a medical
diary for 4 weeks. In retrospect, a research nurse Costs were divided into direct medical costs, including
checked whether a reported GP visit was actually due medication (both over-the-counter and prescriptions),
to gastroenteritis. Consultations for other complaints GP costs, outpatient department costs, hospital in-
were not counted. In the medical diary, cases reported patient costs, and laboratory tests, direct non-medical

https://doi.org/10.1017/S0950268803001559 Published online by Cambridge University Press

 Costs of gastroenteritis in The Netherlands 213

costs (transport costs) and indirect non-medical Table 1. Cost vectors in The Netherlands, 1999 (in E)
costs (productivity losses, costs of informal care). All
Costs (E)
costs were calculated per age group (age groups :
<1, 1–4, 5–11, 12–17, 18–64 and 65+ years) and were Direct medical costs
measured in Dutch guilders, but were converted to Prescription charges 5.26
Euros (E) for this study (1 NLG=E0.45, as of GP visit 16.59
Alternative treatment 40.60–50
1 January 2002).
Outpatient department visit 40.84
Hospital admission (per day) 235.97
Stool test* 58.20
Direct medical costs
Direct non-medical costs
Whenever a patient reported taking medication pre- Transport – private transport by 0.11
scribed by a doctor, a surcharge for the pharmacy was car (per km)
added to the price of the medication. Drugs included Indirect costs
antibiotics, oral rehydration solutions (ORS), pain- Average costs of absence from 120.06
work (per day)
killers, anti-diarrhoeic medication and other pre-
Average costs of informal care (per day) 112.43
scribed drugs. Full details and dosage were unknown.
Therefore the guidelines for average daily dosages * Assuming that in general one stool sample is tested for
from the Dutch pharmaco-therapeutical precept were campylobacter, yersinia, salmonella and shigella.
used. For antibiotics the average tariffs for macrolides
were used. The national average costs for different
types of drugs were taken from this precept [9]. Non- transportation by taxi or other public transport and
specified over-the-counter medication was disregarded parking fees. The average costs per kilometre by car,
in the cost calculations. were used in the cost calculation [10].
Each consultation with a GP or specialist in an Transport costs to and from the practice for
outpatient department and admission to a hospital alternative treatment were not calculated since no
generated consultation fees and transport to and from data were available on the average distance to such
the practice or hospital. Guidelines for average costs practices. These costs are likely to be only a very small
of the above components to be used in pharmaco- proportion of total costs and including them would
economic research, as issued by the Dutch National not significantly change the total cost estimate. Other
Health Insurance Board, were used in the cost calcu- transport costs, e.g. transport to a pharmacy, were
lation [10]. not calculated. Direct out-of-pocket expenses (such as
A consultation with a specialist in alternative the purchase of cleaning materials and replacements
treatment generated only consultation fees. For a for items spoilt as a result of the illness) were not
homeopath this was E50, for an acupuncturist this taken into consideration due to lack of reliable data.
was E40.60 per consultation (guidelines from the
Dutch Association of Classic Homeopaths and the
Indirect costs
Dutch Acupuncture Association).
The average costs of laboratory faeces tests for a Productivity losses were calculated for cases over 18
range of enteric pathogens were taken from the years old, using the friction cost method. In the Dutch
guidelines of the National Health Tariffs Authority. guidelines for cost research [10], age and sex are taken
into consideration, but no distinction is made by type
of occupation or education. Time lost from school
Direct non-medical costs
was estimated. However, in the absence of proper cost
The average distance from a Dutch household to their estimates for this, no societal costs were placed upon
GP was estimated to be 1.8 km [10]. We further as- it. For productivity losses due to informal care we
sumed (because of lack of data) that half of the also used the friction cost method. Age and sex of the
patients used a car for transport, while the other half carer could not be taken into account due to lack of
used a bicycle. The average distance from a Dutch data. We therefore used the average costs for a person
household to the nearest general hospital was esti- between 30 and 49 years old.
mated to be 7.0 km [10]. We assumed that all patients All cost vectors that were used in the cost estimate
used a car to go to the hospital. We disregarded are given in Table 1.

https://doi.org/10.1017/S0950268803001559 Published online by Cambridge University Press

 214 W. E. van den Brandhof and others

Sensitivity analyses campylobacter (n=89) and salmonella cases (n=33)

in the GP study, the proportion of gastroenteritis
A sensitivity analysis was conducted to consider
patients, with campylobacter and salmonella, not
uncertainty in the calculated total number of gastro-
consulting a health-care worker was estimated per age
enteritis cases in The Netherlands and to consider
group [12]. These proportions were then used to cal-
variation in cost components that contributed most to
culate the total costs for all non-consulting patients
the total costs. Using the 95% confidence interval
that could be attributed to campylobacter and
(CI) of the standardized incidences of gastroenteritis
salmonella by age group. The total costs for patients
per age group, a minimum and maximum number of
with campylobacter and salmonella were calculated by
cases per age group were calculated. Furthermore, for
summing the costs of consulting and non-consulting
those cost components that contributed most, 95 %
cases. For norovirus, results from the indirect and
CIs of their point estimates were calculated. With
direct method could be compared, to provide insight
these CIs, the minimum and maximum of the costs
in the dependence of results by the used method.
could be calculated.

Costs for specific pathogens causing gastroenteritis DALYs

In addition to the costs of gastroenteritis in general, The number of DALYs associated with gastroenter-
we also estimated the respective costs for gastro- itis was calculated as follows. We obtained data on
enteritis due to the top three foodborne pathogens : number of deaths with a primary cause of death coded
campylobacter, salmonella and norovirus [1]. The as ICD-9 codes 001-009 and 558 and on the average age
costs of norovirus could be directly estimated from at time of death for 1999 from Statistics Netherlands
the data of the CB study, using the same method as [13]. To calculate the number of Years of Life Lost
for gastroenteritis in total. The number of patients (YLL) in 1999, the remaining statistical life expect-
with campylobacter and salmonella were too small ancy at the time of death was calculated from sur-
in the study (9 and 3 respectively) to use this direct vival tables from 2000 [14] and aggregated over all
method without risking a great uncertainty in the deceased patients with gastroenteritis. The number of
results. Therefore, the costs for campylobacter and Years Lived with Disability (YLD) was calculated as
salmonella (and norovirus) were estimated using an follows. Data from the CB study were used to calcu-
indirect method : cases from the CB study were late the total number of gastroenteritis cases per age
divided into those using the health-care system (GP, group and the average duration of their disease epi-
outpatient department, alternative treatment and/or sode for 1999. The disease weights were taken from a
hospital) (consulting patient) and those not using the Dutch national study on the burden of diseases [6].
health-care system (non-consulting patient). For both This study aimed at determination of the disease
groups, again the total costs for all cases were esti- weight for a large array of diseases, and within dis-
mated per age group by the previously described di- eases, for different stages of severity of disease. For
rect method. From 1996 to 1999, in The Netherlands, infectious diseases, the study resulted in severity
the incidence of gastroenteritis in GPs and the role of weights for a year that includes an episode of gastro-
a broad range of pathogens were studied (GP study) enteritis. Both weights for mild and more severe
[11, 12]. We used the results from this study to esti- gastroenteritis were determined. A disease weight of
mate for consulting patients the proportion that was 0.005 (95 % CI 0.001–0.009) was being attributed to a
due to campylobacter and salmonella by age group. year including a period of mild gastroenteritis (dur-
These proportions were then used to calculate the ation up to 2 weeks), a disease weight of 0.03 (95 % CI
total costs for all consulting cases with symptoms that 0.018–0.039) was being attributed to a year including
could be attributed to campylobacter and salmonella a period with a more severe course of gastroenteritis
by age group. The costs for non-consulting patients (duration 2–4 weeks). These disease weights were
with a campylobacter or salmonella infection were multiplied by the estimated total number of cases with
calculated as follows. From the patients from the CB a short or a longer course of gastroenteritis in The
study who did not consult their GP and did not visit Netherlands. The two figures for YLD were added.
another health-care system, the overall proportion The total number of DALYs related to gastroenteritis
of patients with campylobacter and salmonella was was calculated by summing the number of YLL and
estimated. Then using the age distribution of the YLD of patients with gastroenteritis in 1999.

https://doi.org/10.1017/S0950268803001559 Published online by Cambridge University Press

 Costs of gastroenteritis in The Netherlands 215

RESULTS another baby was admitted to the hospital for 10

days. One elderly person (72 years old) went to the
Gastroenteritis cases and duration of illness
emergency room and was admitted to the hospital
Of the 4860 participants in the cohort, 1052 case epi- where he remained for 10 days (Table 2).
sodes were observed : 720 persons had one episode of According to the NMR of hospital discharge diag-
gastroenteritis, 129 had two, 18 had three and 5 had noses in The Netherlands, 17 978 patients with gas-
four. This yielded an overall standardized incidence troenteritis as primary or secondary diagnosis were
of gastroenteritis (standardized for age, gender and admitted to the hospital in 1999. Of these, 14, 22, 5, 2,
cohort) of 283/1000 person-years, with an estimated 30 and 27 % were for individuals aged <1, 1–4, 5–11,
total of 4.5 million cases per year (Table 4). The 12–17, 18–64 and 65+ years. The median number of
incidence was the highest in children between 1 and days admitted to the hospital was 5 days. In total
4 years and <1 year old (900/1000 and 740/1000 approximately 4/1000 patients (0.4%) with gastroen-
person-years respectively). teritis were hospitalized in The Netherlands. In the
Of these cases, 774 (74 %) participated in the case- CB study, 0.5 % of the cases were admitted to the
control component, which was found to be largely hospital (3/646) (Tables 2 and 3). Of the total hospi-
representative of all observed cases in the cohort [1]. talizations, 12 669 (70 %) had gastroenteritis as their
Of these, 646 (83 %) completed a medical diary. The primary diagnosis (Tables 3 and 4), corresponding
medical diary was completed relatively more often with a hospitalization rate of 0.3% of all community
for children (89 % of 0–4 years old, completion by cases.
a parent) than for cases in older age groups (70 % of In 2001, the GPs from the sentinel network
cases older than 18 years). Sixty-three of the 646 reported 1306 consultations for gastroenteritis, by
patients (10 %) had no complaints of diarrhoea, but age group. For these 1306 consultations, 177 (14 %)
they did have complaints of vomiting. patients’ stool samples were sent to a laboratory for
Duration of illness was on average 12 days (median testing on enteropathogens. Relatively more stool
8 days, range 1–101 days). The cases in the oldest samples were taken from adult cases ; for 21% of
age group (65 years and older) suffered the longest patients older than 65 years a stool sample was taken.
(median 22 days, range 1–56 days) and the 18–64 Seventy per cent of all stool samples were taken from
years group suffered the shortest (median 3 days, patients older than 19 years (Table 2).
range 1–28 days). In total, 296 cases (46 %) were In total, direct medical costs were approximately
confined to bed for an average of 2.6 days (median E61 000 (Table 4), with an estimated average of E14
2 days, range 1–14 days). per case using the hospital data from the CB study.
Average costs were highest for those 65 years or older
(Table 3). Calculating the costs for hospital ad-
Direct medical costs
missions based on the data from the NMR, the total
Forty-two cases (7 %) used antibiotics, 48 (7 %) used average costs amounted to E13 per person. However,
ORS, 28 (4 %) used anti-diarrhoeic medication, 218 the costs were distributed differently over the age
(34 %) used painkillers, 40 (6 %) used other prescribed groups. Average costs were still highest for those 65
medication and 31 cases (5 %) used other medication years or older (Table 3). The costs due to hospital
without a prescription from a GP. Patients 12–17 admissions contributed most to the direct medical
years old, and patients older than 65 years used rela- costs (Tables 3 and 4).
tively more medication. In total, 65 GP-confirmed
consultations were reported (standardized consul- Direct non-medical costs
tation rate of 5 %). Young children were seen rela-
Transport costs were relatively small ; for all cases
tively more often by their GP. For one baby a
with gastroenteritis in The Netherlands seeking
homeopath was consulted, and one person 69 years
medical care they amounted to E80 000, with a negli-
old visited an acupuncturist five times because of his
gible average per case (Tables 3 and 4).
gastroenteritis complaints. A specialist in an out-
patient department (a paediatrician) was consulted
Indirect costs
three times (0.5 % of 646). One baby visited a pae-
diatrician and was admitted to the hospital 1 day In total 146 cases (23 %) in the CB study reported
later, where he stayed for 2 days. Furthermore, absence from school (including day-care centres)

https://doi.org/10.1017/S0950268803001559 Published online by Cambridge University Press

 216 W. E. van den Brandhof and others

Table 2. Use of resources, per age group (%)

Age (years)

<1 1–4 5–11 12–17 18–64 65+ Total

(n=184) (n=253) (n=119) (n=15) (n=51) (n=24) (n=646)

Associated with direct medical costs

Total medication 38 25 16 53 33 54 29
GP visit 17 9 6 7 2 8 5*
Alternative treatment 0.5 0 0 0 0 21 0.9
Outpatient department visit 0.5 0.8 0 0 0 0 0.5
Hospital admission CB 1.1 0 0 0 0 4 0.5
Hospital admission NMR# 2 0.6 0.1 0.2 0.2 1 0.4
Stool tests (% of all cases)$ 1.2 0.9 0.5 0.5 0.3 2 0.6
Stool tests (% of GP patients)$ 8 10 8 8 17 21 14
Associated with direct non-medical costs
Transport· 18 10 6 7 2 13 11
Associated with indirect costs
Absence from work 0 0 0 0 14 0 1.1
Absence of other person 14 12 18 7 6 0 13
for informal care

* Standardized for age and cohort.

# Based on data from the National Medical Registry, 1999 (total of registered hospitalizations at NMR divided by total
estimated number of cases per age group). Number of admissions by age group (median duration of stay) : <1 year, 2581
(5 days) ; 1–4 years, 3979 (4 days) ; 5–11 years, 891 (4 days) ; 12–17 years, 339 (4 days) ; 18–64 years, 5403 (5 days) ; 65+ years,
4785 (11 days).
$ Based on the questionnaires completed by GPs from the sentinel network, 2001.
· Transport to GP (E0.40 per visit), outpatient department and hospital (E1.54 per visit).

or work due to illness. Of these, 65 were younger than 1 and 11 years had relatively low costs per case
5 years, 71 were children going to school, 3 were (Table 3). Using the costs for hospital admissions
adults going to school and only 7 were adults with based on data from the NMR, the highest costs per
jobs. These adults were absent from work for an case were found in cases between 18 and 64 years old
average of 5 days (range 2–10 days). Eighty-two cases (Table 3).
(13 %), of which 79 were children, reported absence For the 4.5 million case episodes in the Dutch
from work by a carer in the same household. This population in 1999, the societal costs were estimated
was on average for 2 days (range 1–12 days) (Table 2). at E345 million (Table 4). Using the data from the
Indirect costs were in total E284 million. Despite a NMR did not significantly change this estimate.
small proportion of cases being absent from work
(1 %), the costs for absence from work were higher
Sensitivity analyses
than costs for informal care, because of the high
number of cases between 18 and 64 years old and the Confidence intervals were calculated per age group
higher costs per day (Table 4). The average costs were for the incidence rate, the number of GP consul-
E63.5 per case, 82% of the total average costs per case tations, the number of hospital admissions, the num-
(using the hospital data from the CB study) (Table 3). ber of days absent from work and the number of days
of informal care. With these CIs, the minimum and
maximum of the total number of cases and of related
Total costs
costs for these cost vectors were estimated per age
On average, the total costs per case were E77. In- group. Varying the incidence rates led to a range in
dividuals 18 years and older had relatively higher the total costs of E249–439 million. Of the four major
costs per person, with the highest costs being for cases cost components, varying the number of hospital ad-
over 65 years (E114 per case). Children aged between missions had the greatest impact on the total costs,

https://doi.org/10.1017/S0950268803001559 Published online by Cambridge University Press

 Costs of gastroenteritis in The Netherlands 217

Table 3. Average costs (in E) per case in a prospective population-based cohort study in The Netherlands,
December 1998 to December 1999, by age group

Age (years)

Items contributing to <1 1–4 5–11 12–17 18–64 65+ Total

costs (E) (n=184) (n=253) (n=119) (n=15) (n=51) (n=24) (n=646)

Medication 2.0 1.2 0.8 2.3 2.0 4.6 1.9

GP visit 2.8 1.5 1.0 1.1 0.3 1.4 0.8
Alternative treatment 0.3 0 0 0 0 8.5 0.8
Outpatient department visit 0.2 0.3 0 0 0 0 0.1
Hospital admission CB 15.4 0 0 0 0 98.3 9.6
Hospital admission NMR* 27.2 6.1 1.5 2.9 4.6 46.6 8.9
Stool tests 0.7 0.5 0.3 0.3 0.2 1.0 0.4
Direct medical sub-total# 21.4 3.5 2.1 3.7 2.5 113.7 13.5
Direct medical sub-total$ 33.2 9.6 3.6 6.6 7.1 62.1 12.9
Transport costs 0.1 0 0 0 0 0.1 0.0
Direct non-medical sub-total 0. 1 0 0 0 0 0.1 0.0
Absence by case 0 0 0 0 84.9 0 45.1
Absence of other person for 31.2 27.1 33.1 89.9 8.8 0 18.3
informal care
Indirect sub-total 31.2 27.1 33.1 89.9 93.7 0 63.5
Total# 52.6 30.7 35.1 93.7 96.2 113.8 77.0
Total$ 64.5 36.8 36.7 96.6 100.8 62.1 76.4

* Based on data from the National Medical Registry, 1999. Patients with both a primary or secondary diagnosis. Total costs
for patients with only gastroenteritis as a primary diagnosis were estimated at E71.9 per case.
# Including costs from hospital admissions based on data from the CB study.
$ Including costs from hospital admissions based on data from the National Medical Registry, 1999.

leading to a range of E327–436 million. Varying all consulting cases with a campylobacter, salmonella
four cost vectors at the same time, led to a range in the and norovirus infection were E4.8, E2.8 and E2.1
total costs of E252 million (if all four minimum esti- million respectively. Of the cases in the CB study who
mates were taken) to E531 million (for the maximum did not consult the health-care system, 1.4, 0.5 and
estimates) (Table 5). 17.4% had a campylobacter, salmonella and noro-
virus infection, leading to the total costs of these
patients of E4.4, E1.3 and E44.0 million respectively.
Costs for specific pathogens causing gastroenteritis
The total costs of campylobacter, salmonella and
Using the direct method, the costs of norovirus in- norovirus infections (both consulting and non-
fections were estimated to be 6% of the total costs consulting) were E9, E4 and E46 million ; 3, 1 and
of gastroenteritis. In order to calculate the costs of 13 % of the total costs respectively. Using the costs of
campylobacter, salmonella and norovirus using the norovirus estimated with the direct method and the
indirect method, the costs for cases from the CB study costs of campylobacter and salmonella estimated with
were divided into costs for cases using the health-care the indirect method, the costs of these three pathogens
system [in total E62 million (18 %), average E148 per combined were 10 % of the total costs of gastroen-
case] and costs for cases not using the health-care teritis (Table 6).
system [in total E281 million (81 %), average E70 per
case]. Eighty-five per cent of the costs for cases using
DALYs
the health-care system were direct medical costs. In
10 % of the cases in the GP study a campylobacter In 1999, 241 persons died of gastroenteritis, with
infection was detected, in 4 % a salmonella infection relatively the highest percentage in cases 65 years and
and in 5 % a norovirus infection [1]. Taking the older (51 deaths/100 000 persons). As a result of this
age distribution into account, the total costs for premature death, 2563 years of life were lost.

https://doi.org/10.1017/S0950268803001559 Published online by Cambridge University Press

 218 W. E. van den Brandhof and others

Table 4. Estimated costs (rE1000) for The Netherlands of gastroenteritis, by age group

Age (years)

<1 1–4 5–11 12–17 18–64 65+ Total

Incidence rate (per 1000 person-years)* 740 900 481 157 234 194 283
Estimated total number of cases 147 799 698 802 663 629 173 750 2 379 018 413 401 4 476 399
Medication 296 841 528 401 4 733 1 889 8 687
GP visit 413 1 054 648 192 774 572 3 653
Alternative treatment 40 0 0 0 0 3 497 3 537
Outpatient department visit 33 226 0 0 0 0 258
Hospital admission CB 2 274 0 0 0 0 40 645 42 920
Hospital admission NMR# 4 025 4 270 1 027 508 10 849 19 277 39 956
Stool tests 105 347 188 53 454 418 1 565
Direct medical sub-total$ 3 162 2 469 1 363 646 5 961 47 020 60 620
Direct medical sub-total· 4 913 6 738 2 390 1 154 16 810 25 652 57 657
Transport costs 9 17 8 2 10 34 81
Direct non-medical sub-total 9 17 8 2 10 34 81
Absence by case 0 0 0 0 201 971 0 201 971
Absence of other person for 4 606 18 943 21 945 15 628 20 978 0 82 100
informal care
Indirect sub-total 4 606 18 943 21 945 15 628 222 949 0 284 070
Total$ 7 777 21 429 23 316 16 276 228 919 47 055 344 771
Total· 9 527 25 699 24 343 16 784 239 769 25 686 341 808

* Standardized for age, gender and cohort. Standardized for the Dutch population, 1999.
# Based on data from the National Medical Registry, 1999. Both patients with a primary or secondary diagnosis. Total costs
for patients with only gastroenteritis as a primary diagnosis were estimated at E20 million.
$ Including costs from hospital admissions based on data from the CB study.
· Including costs from hospital admissions based on data from the National Medical Registry, 1999.

Table 5. Estimated ranges of the total costs (rE1000) for the cases with more severe gastroenteritis 50 332
for The Netherlands of gastroenteritis, by age group (0.03r37 % of the estimated total number of cases).
The total number of YLD was therefore 64 326 (95 %
5th 95th
CI 32 998–90 620). The total number of DALYs was
percentile percentile
thus 66 889 (95 % CI 35 561–93 183).
Varying the incidence rate* 249 223 439 448
Varying the cost vectors
GP consultations 342 494 352 182 DISCUSSION
Hospital admissions 326 789 435 909
In this study, the average costs per patient with
Absence from work 302 482 376 960
Informal care 314 350 400 088 gastroenteritis were E77. The total costs of patients
with gastroenteritis in The Netherlands were esti-
Varying all four cost vectors 251 802 530 825
mated at E345 million in 1999. The combined costs
* 95 % CI of incidence per 1000 person-years : <1 year, of campylobacter, salmonella and norovirus were
606–875 ; 1–4 years, 766–1034 ; 5–11 years, 389–575 ; 12–17 approximately 10–17 % of the total costs of gastroen-
years, 95–219 ; 18–64 years, 169–298 ; 65+ years, 126–262. teritis. In 1999, gastroenteritis was associated with a
loss of approximately 67 000 DALYs.
In the CB study, duration of illness was known for Even with relatively low costs per case (E77), the
619 persons. A total of 387 (63 %) cases had com- total costs are high because of a high incidence of gas-
plaints less than 2 weeks (on average 5.2 days) and for troenteritis in The Netherlands (283/1000 person-
232 (37 %) cases the complaints lasted longer than 2 years). Varying this incidence per age group resulted
weeks (on average 23.7 days). The number of YLD in a range of the costs of E249–439 million. However,
for the cases with mild gastroenteritis were 13 993 this range should probably be smaller, since it is un-
(0.005r63 % of the estimated total number of cases), likely that the incidences are either all in the highest

https://doi.org/10.1017/S0950268803001559 Published online by Cambridge University Press

 Costs of gastroenteritis in The Netherlands 219

Table 6. Estimated costs (rE1000) for including non-infectious and chronic diseases [17, 18],
The Netherlands for specific pathogens causing or on specific pathogens [19–22]. Therefore, their re-
gastroenteritis, by age group sults are difficult to compare with ours. A recent study
in England (the IID study), similar to ours, estimated
% of total
the total costs of gastroenteritis at E1039 million [2].
costs for
gastroenteritis When the number of inhabitants of England is taken
Costs (E344 771) into account, these costs are almost equal to ours :
about E22 per inhabitant.
Direct method
The use of ICD codes in mortality statistics and
Norovirus 19 579 6
Direct costs* 4 192 1 hospital discharge diagnoses has some limitations,
Indirect costs 15 386 4 because the ICD codes are not very specific. Most
Indirect method hospitalizations and deaths were found for the rather
Patients using the 62 147 18 non-specific code 558 (other non-infectious gastroen-
health-care system teritis and colitis). Exclusion of this ICD code would
Direct costs* 53 023 15 produce a fall in the number of hospitalizations from
Indirect costs 9 124 3
17 978 to 5906, the number of deaths from 241 to 40
Patients not using the 280 921 81
health-care system and the number of DALYs from 67 000 to 65 000.
Direct costs* 6 070 2 This code (558), however, strongly correlates with,
Indirect costs 274 851 80 for example, the incidence of laboratory-confirmed
Campylobacter 9 209 3 rotavirus [8]. To obtain more reliable estimates of the
Salmonella 4 044 1 true number of hospitalizations due to gastroenteritis,
Norovirus 46 059 13 specific hospital studies into the use of ICD codes are
* Including transport costs.
needed.
For the estimates of the costs for specific pathogens
it was assumed that within the groups of cases con-
or all in the lowest percentiles for all age groups at the sulting and not consulting the health-care system,
same time. A variation in the major four cost com- patients have the same costs, regardless of their aeti-
ponents also lead to substantial variation in the costs. ology. However, for example, duration of illness is on
For example, the costs for hospital admissions range average longer in patients with a campylobacter in-
between E25 000 and E134 000 in total. fection when compared to patients with a norovirus
We consider the costs (and DALYs) estimate a infection [23]. Therefore, costs of absence from work
minimum because of several reasons. In our study we and informal care are probably higher for patients
did not measure any cases with complicated gastro- with a campylobacter infection. Also duration of
enteritis, leading to haemolytic uraemic syndrome, hospital admission might vary between pathogens.
Guillain–Barré syndrome, sepsis, or arthritis. Neither This might explain the difference in the costs of noro-
costs of hospitalizations and deaths, nor costs of virus infection calculated with the direct and the in-
treatment and long-term health consequences for direct method. The costs calculated with the indirect
these relatively expensive conditions are included in method are probably too high for cases with a noro-
our calculations. Also, we disregarded several cost virus infection and too low for cases with campylo-
vectors, such as non-specified over-the-counter medi- bacter and/or salmonella infection. Consequently, the
cation, direct out-of-pocket expenses, taxi use and pathogen-specific costs should be considered as rough
parking fees, although we believe that the effect of estimates. Although often regarded as a mild disease,
these aspects on the total would be limited. In order to the costs of a norovirus infection are substantial (E20
quantify the costs of informal care more precisely million, calculated with the direct method), mainly
more information on carers should be obtained in due to absence from work by a carer (79 % of the total
further studies (e.g. who was the carer, and what was costs).
his/her work). The number of DALYs calculated by us (67 000),
This is the first cost-of-illness study of gastroenter- indicates that the public health burden of gastroen-
itis in The Netherlands. There have been studies in teritis is substantial. In the Dutch Public Health
other countries, but mainly focusing on only food- Status and Forecasts (PHSF), published in 1997,
borne illness [3, 4, 15, 16], on gastrointestinal diseases the number of DALYs for a large number of diseases

https://doi.org/10.1017/S0950268803001559 Published online by Cambridge University Press

 220 W. E. van den Brandhof and others

and disorders were estimated [24]. The number of REFERENCES

DALYs calculated by us for gastroenteritis is in the 1. Wit MAS de, Koopmans MPG, Kortbeek LM, et al.
same order as the number of DALYs associated Sensor, a population-based cohort study on gastroen-
with dementia, heart failure, colon cancer and traffic teritis in the Netherlands, incidence and etiology. Am J
accidents. In the PHSF of 1997, the number of Epidemiol 2001 ; 154 : 666–674.
DALYs of gastroenteritis was estimated at 6900. This 2. Roberts JA, Cumberland P, Sockett PN, et al. The
study of infectious intestinal disease in England : socio-
difference was because the incidence used was only
economic impact. Epidemiol Infect 2003 ; 130 : 1–11.
based on patients consulting their GP. In a study by 3. Buzby JC, Roberts T. Economic costs and trade
Havelaar et al. [25] the mean health burden of cam- impacts of microbial foodborne illness. World Health
pylobacter-associated illness in the Dutch population Statist Quart 1997 ; 50 : 57–66.
in the period 1990–1995 was estimated at 1400 DA- 4. Scott WG, Scott HM, Lake RJ, Baker MG. Economic
LYs (90 % CI 900–2000). Campylobacteriosis is the cost to New Zealand of foodborne infectious disease.
NZ Med J 2000 ; 113 : 281–284.
cause of gastroenteritis in 2 % of all gastroenteritis 5. Koopmanschap MA. Cost-of-illness studies. Useful for
cases [1]. If the health burden would be proportionally health policy ? Pharmacoeconomics 1998 ; 14 : 143–148.
subdivided among the related pathogens, the health 6. Maas PJ van der, Kramers PGN, eds. The Dutch Public
burden of campylobacteriosis according to our study Health Status and Forecasts Report 1997. Weighted
would yield a similar estimate, i.e. 1287 DALYs (95 % health and life expectancy [in Dutch]. Bilthoven/
Utrecht : National Institute of Public Health and the
CI 660–1812). However, of the estimated DALYs in
Environment/Elsevier/de tijdstroom, 1997.
the Havelaar et al. study, 38% are for complications 7. World Health Organization (WHO). International
such as Guillain–Barré syndrome and arthritis, which classification of diseases, 9th revision. Geneva : WHO,
we did not include in our estimates. 1977.
The number of DALYs was highly dependent on 8. Wit MAS de, Koopmans MPG, Blij JF van der,
the YLD. However, in general with common diseases, Duynhoven YTHP van. Hospital admissions for rota-
virus infection in the Netherlands. Clin Infect Dis 2000 ;
the YLD is strongly influenced by the weight factors 31 : 698–704.
used for the mild and more severe illness. Changing 9. Loenen AC van, ed. Pharmaco-therapeutical precept [in
this factor slightly (e.g. from 0.005 to 0.01) causes a Dutch]. Amstelveen : College voor zorgverzekeringen,
substantial change in the total YLD (from 64 000 to 2002.
78 000), because of the high absolute number of cases. 10. Oostenbrink JB, Koopmanschap MA, Rutten FFH.
Guidelines for cost research. Methods and guideline
Consequently, the usefulness and validity of DALYs
prices for economic evaluations in health care [in Dutch].
for common, short-term and relatively mild infectious Amstelveen : College voor zorgverzekeringen, 2000.
diseases such as gastroenteritis should be evaluated 11. Wit MAS de, Koopmans MPG, Kortbeek LM,
thoroughly. Leeuwen WJ van, Bartelds AIM, Duynhoven YTHP
In conclusion, in 1999, gastroenteritis was associ- van. Gastroenteritis in sentinel general practices, the
ated with considerable costs to society and loss of Netherlands. Emerg Infect Dis 2001 ; 1 : 82–91.
12. Wit MAS de, Kortbeek LM, Koopmans MPG, et al.
DALYs in The Netherlands. Because of increasing Comparison of gastroenteritis cases in a general practice
trends in consultations for gastroenteritis in GPs since based-study and a community-based study. Epidemiol
2000 [26], as well as in salmonellosis and campylo- Infect 2001 ; 127 : 389–397.
bacteriosis in recent years, it is anticipated that 13. Statistics Netherlands. Primary causes of death. (http://
current costs are even higher. statline.cbs.nl), accessed September 2002.
14. Statistics Netherlands. Life expectancies prognoses
2002. (http://statline.cbs.nl), accessed September 2002.
15. Todd ECD. Preliminary estimates of costs of foodborne
ACKNOWLEDGEMENTS disease in the United States. J Food Prot 1989 ; 52 :
595–601.
The authors thank the participating general prac- 16. Withington SG, Chambers ST. The cost of campylo-
titioners of the sentinel network and the NIVEL for bacteriosis in New Zealand in 1995. NZ Med J 1997 ;
their cooperation in the data collection. In addition, 110 : 222–224.
they also thank Nancy Hoeymans for information on 17. Beck IT. Disproportion of economic impact, research
achievements and research support in digestive diseases
the calculation of DALYs and Dr Marcel Peeters
in Canada. Clin Invest Med 2001 ; 24 : 12–36.
for kindly providing the average costs of laboratory 18. Sandler RS, Everhart JE, Donowitz M, et al. The
faeces tests from the guidelines of the National Health burden of selected digestive diseases in the United
Tariffs Authority. States. Gastroenterology 2002 ; 122 : 1500–1511.

https://doi.org/10.1017/S0950268803001559 Published online by Cambridge University Press

 Costs of gastroenteritis in The Netherlands 221

19. Powell M, Ebel E, Schlosser W. Considering uncer- in the Netherlands and the related economic costs
tainty in comparing the burden of illness due to food- [in Dutch]. Infectieziekten Bulletin 2000 ; 11 : 4–8.
borne microbial pathogens. Int J Food Microbiol 2001 ; 23. Chin J, ed. Control of communicable diseases manual,
69 : 209–215. 17th edn. Washington : American Public Health As-
20. Frühwirth M, Berger K, Ehlken B, Moll-Schüler I, sociation, 2000.
Brösl S, Mutz I. Economic impact of community- 24. Ruwaard D, Kramers PGN, eds. The Dutch Public
and nosocomially acquired rotavirus gastroenteritis in Health Status and Forecasts Report 1997. Overview [in
Austria. Pediatr Infect Dis J 2001 ; 20 : 184–188. Dutch]. Bilthoven/Utrecht : National Institute of Public
21. Morales RA, McDowell RM. Economic consequences Health and the Environment/Elsevier/de tijdstroom,
of Salmonella enterica serovar enteriditis infection 1997.
in humans and the U.S. egg industry. In : Saeed AM, ed. 25. Havelaar AH, Wit MAS de, Koningsveld R van,
Salmonella enterica serovar enteriditis in humans and Kempen E van. Health burden in the Netherlands
animals. Epidemiology, pathogenesis and control. due to infection with thermophylic Campylobacter
Ames : Iowa State University Press, 1999 : 271–290. spp. Epidemiol Infect 2000 ; 125 : 505–522.
22. Pelt W van, Giessen AW van de, Leeuwen WJ van, 26. Bartelds AIM. Continuous morbidity registration
et al. Origin, extent and costs of human salmonellosis. sentinel network the Netherlands, 2001. NIVEL,
Part 2 : estimate of the extent of human salmonellosis Utrecht, 2002.

https://doi.org/10.1017/S0950268803001559 Published online by Cambridge University Press