  |
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Journal of the New Zealand Medical Association,
11-March-2005, Vol 118 No 1211
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Estimating the impact of the next influenza pandemic on
population health and health sector capacity in New Zealand
There have been 31 influenza pandemics reported since the
first pandemic described in 15801 and the major
influenza pandemics of the 20th century (1918,
1957, and 1968) all reached New Zealand. Furthermore, New Zealand was unable to
avoid any of the three pandemic waves that comprised the 1918
pandemic.2 In this pandemic, between a third to
a half of the entire population suffered illness and there were an estimated
8250 deaths (0.74% of the population).3 Its
impact on Maori was particularly severe.3,4
Another influenza pandemic is considered to be highly
likely5or even
inevitable.6 In 1997 and 2003, new strains of
influenza virus emerged that had the potential to become pandemic, though their
circulation may have been stopped by the mass slaughter of
poultry.7,8 There is also concern that
influenza could be genetically modified and used as a
bioweapon.9
New Zealand has undertaken influenza pandemic planning and
undertaken simulation exercises.10–12 The
pandemic plan has even been updated in the light of the experience with
SARS.13 However, there has been relatively
little work to estimate the likely impact of pandemic influenza on population
health. Therefore this article aims to provide estimates for the impact of
pandemic influenza on health and the New Zealand health sector through the use
of a publicly available software package and model (FluAid).
MethodsModelling
assumptions and
software—The
US Centers for Disease Control and Prevention (CDC) developed software package,
FluAid14 utilises a relatively simple
deterministic model. The output of the model is the number of deaths,
hospitalisations, and illness requiring medical consultations for a single wave
of pandemic influenza. The model assumes no effective public health
interventions to control disease spread (such as use of an appropriate vaccine
or widespread use of antiviral drugs). Specific details on the FluAid software
and the various assumptions in the model are detailed on the CDC
website15 and other
documents.16,17
Given the lack of relevant New Zealand data, the
default values used in FluAid were used for the proportion of the population in
the ‘high-risk’ category for each age group, the mortality rates,
the hospitalisation rates, and the rates of illness. The model’s
parameters were based on the available data (mostly from North America and some
from Europe) from the 1957 pandemic and subsequent non-pandemic
data.16 Individuals categorised as
“high-risk” are those who have a pre-existing medical condition
(e.g. diabetes) that makes them more susceptible to developing medical
complications due to influenza. The proportions in the “high-risk”
category used in the model were 6.4% for 0-18-year-olds, 14.4% for
19–64-year-olds, and 40% for those aged 65 years and older. The output
values from the model were for ‘most likely,’ ‘minimum,’
and ‘maximum’ values (each for mortality, hospitalisation, and
illness requiring medical consultations).
Population data
sources—The population data used for national level calculations
were Statistics New Zealand population estimates for
2004.18 District health board (DHB) level
calculations used 2001 Census data.19
Time
distribution—The
FluAid model does not consider the time frame of the epidemic within an affected
region. The length of influenza epidemics is highly
variable20,21 but for this analysis the first
pandemic wave was assumed to span 8 weeks with a distribution pattern in which
80% of cases occurred in a 3-week period around the peak of the epidemic (as per
the pattern for the deaths in the second pandemic wave of the 1918 pandemic for
Auckland, Wellington, Christchurch, and
Dunedin3). This distribution is very similar to
that of a recently published stochastically simulated influenza
epidemic22 (i.e. both models had 32% of the
clinical cases occurring in the ‘peak week’ of the pandemic
wave).
Health sector capacity
data—Data outputs from the model were used to estimate the demand
on the health sector. The supply side was based on the most up-to-date data on:
total number of public hospital beds in the country (12,484 in 2002),
non-pandemic year mortality levels (28,224 deaths for the most recent year ie,
1999 – giving a weekly average of 543 deaths per week), total number of
general practitioners (2917 in 2002) and total number of registered nurses
working in primary healthcare (3394 in
2003).23
ResultsMortality—The
model predicts 1600 to 3700 deaths as most likely from the first wave of
pandemic influenza that has incidence rates of 15% and 35%, respectively (Table
1). Most (83%) of these deaths, would be among those with high-risk conditions
(with 42% aged 19-64 years and 41% aged 65 years and over).
Hospitalisations—The
model predicts likely hospitalisations at between 6900 and 16,200 with a full
range of 2500 to 20,800 (Table 2). Of these hospitalisations, 71% would be of
people without high-risk conditions
(55% would be aged 19–64 years and 12% would be 65+). Only 10% and 19% of
hospitalisations would be contributed by those with high-risk conditions in the
19–64 and 65+ age groups respectively.
Illness requiring medical
consultations—The
model predicts 325,000 to 759,000 medical consultations with a full range of
254,000 to 1.1 million (Table 3). Of these consultations, most (85%) would be
among those without high-risk
conditions (28% aged 0-18 years, 50% aged 19-64 years, and 7% aged 65+
years).
Table 1. Predicted number of deaths nationally in a
future New Zealand influenza pandemic (based on modelling with FluAid)
*These incidence rates are for clinical illness of a
severity that causes some measurable economic impact, such as one-half day of
work lost, or a visit to a doctor.
Table 2. Predicted number of hospitalisations
nationally in a future New Zealand influenza pandemic (based on modelling with
FluAid)
Health sector
capacity—In
the peak week of the epidemic, it is estimated that influenza deaths would
exceed the usual (non-pandemic year) weekly average by 2.2 times (35% incidence
scenario) (Table 4). Also, in this peak week, it is estimated that 42% of all
public hospital beds would be required for pandemic influenza cases, at least
for some proportion of the week. If the average length of hospital stay is half
a week per influenza case, then only half this proportion of all beds would be
required for influenza cases (i.e. 21% of beds in the peak week of the
epidemic).
Table 3. Predicted number of medical consultations*
nationally in a future New Zealand influenza pandemic (based on modelling
with FluAid)
* In the FluAid model this category was described as
“outpatient-based visits”, but in the New Zealand context this would
generally equate to primary care consultations ie, with general
practitioners.
Table 4. Predicted time distribution of the health
impact and demand on services from influenza (assuming a 35% incidence
rate—‘most likely’ estimates) in a future New Zealand
influenza pandemic (based on modelling with FluAid)
It is estimated that 83 influenza consultations per general
practitioner (GP) would occur during the peak week. But if only 50% to 75% of
GPs were working during this week (e.g. due to illness or caring for relatives),
then the average weekly caseload would rise to 125 to 166 people. If half of the
consultations for influenza were seen by a registered nurse working in primary
healthcare, then during the peak week these nurses would be consulted by 36
people per week. But if only 50% to 75% of such nurses were working during this
week, then the average weekly caseload would rise to 54 to 72 people.
DHB
impact—The impact at the DHB level is entirely related to the age
structure and size of the population (since the FluAid model does not address
differences in rural versus urban risk of infection). The impact in terms of
numbers would be highest in Canterbury for deaths and hospitalisations (413 and
1776 respectively) and in Waitemata for consultations (80,706) (Table
5).
Table 5. Predicted health impact at the district health
board (DHB) level (assuming a 35% incidence rate—‘most likely’
estimates and based on 2001 Census data) in a future New Zealand influenza
pandemic (based on modelling with FluAid)
DiscussionMortality
impact—The modelling results suggest a potentially large death toll
from the next influenza pandemic in the range of 1600 to 3700 deaths. This
outcome would make it the worst internal demographic event for New Zealand since
the 1918 pandemic. The upper figure in this range gives a mortality rate of 91
per 100,000 (albeit for a single wave), which compares to 745 per 100,000 in New
Zealand from the 1918 pandemic (i.e. a time when antibiotics were not available
and medical care was much less advanced). It is also less than the 1918 pandemic
mortality rate for the United States of 218 per 100,000; but more than United
States rates for the 1957 Asian flu pandemic (22 per 100,000) and the 1968 Hong
Kong flu pandemic (14 per 100,000).24
Although very high, this death toll would still be less than
the Ministry of Health’s estimates for annual deaths attributable to diet
(around 8500 deaths), tobacco (5000), deprivation (4800), and cholesterol
(4700).25 Even so, pandemic influenza may cause
a disproportionately high number of years of life lost due to the relatively
high proportion of deaths in the 19–64 year old age group (compared to
these other causes where the premature deaths are generally in older age
groups). Furthermore, these deaths would be concentrated over a short period
(assumed to be 8 weeks in this modelling).
The high concentration of deaths among those with high-risk
conditions in the over 18-year age group (83% of the total) would suggest the
importance of targeting available preventive measures (in terms of anti-virals
and appropriate vaccinations, when these become available). But in fact some
particular strains of pandemic influenza (such as the 1918 strain) may still
have a severe impact on mortality among healthy young people (an issue not
adequately addressed by this particular model).
Hospitalisations—The
‘most likely’ range of hospitalisations attributable to pandemic
influenza was between 6900 and 16,200 (Table 2). It is likely that these levels
would overwhelm current hospital capacity for much of the epidemic time period
(especially for the 35% incidence scenario). Indeed, some New Zealand hospitals
already suffer from capacity problems during winter months of non-pandemic
influenza years. Rapid action at the start of the epidemic could free up
hospital beds and resources (e.g. cancelling of elective procedures and early
discharge to community care).
Other contingency planning by DHBs and hospitals could also
facilitate lower hospital admission rates (e.g. through strengthening primary
care response capacity both now and during the crisis phase). The use of
stockpiled antiviral medication for these health workers would help to reduce
worker absenteeism rates as might plans to care for the ill dependents of health
staff to reduce absenteeism.
Other work on influenza pandemic reaching New Zealand
suggests that the demands for critical care beds and for mechanical ventilation
may also exceed current capacity under some pandemic
scenarios.26 Planning can identify additional
beds that could be utilised when critical care services reach capacity. However,
whatever planning is put in place it is likely that some difficult decisions
will be required in limiting hospital care to those where it would most likely
affect final health outcomes.
A potential upstream approach to limiting the burden of
hospitalisations and deaths includes reducing the prevalence of chronic diseases
known to increase the risk of adverse sequelae of influenza infection. This
would suggest a stronger focus by the health sector on promoting tobacco
control, improving nutrition and increasing physical activity levels.
Illness requiring
consultations—The
estimated number of medical consultations attributable to pandemic influenza was
huge, with an upper limit of 1.1 million consultations. During the peak of the
epidemic, the numbers could strain the resources of GPs and primary care nurses
in some areas (eg, for the 35% incidence scenario). This workload would be
particularly acute in those parts of the country that are relatively
under-served by GPs (eg, the West Coast27). The
workload problem could potentially be reduced through public education on
appropriate home care for those with influenza and by providing information on
when to seek medical attention (e.g. as detailed on a CDC
websit28). Indeed, this approach could be
promoted now so that the public build up further knowledge and experience about
when medical consultation is really required for influenza-like illness.
Similarly, public education could encourage the use of the existing free
telephone ‘Healthline’ service (that provides access to a registered
nurse) so that the need for face-to-face consultations with GPs and nurses is
reduced. Contingency planning could also address the issues of recalling GPs and
nurses back from leave or retirement, and even utilising medical students (as
done in 1918 in New Zealand29). The
co-ordinated efforts of volunteers were also very valuable for providing home
care during 1918 in New Zealand.3,30 If
employers relaxed requirements for medical certificates associated with
influenza-like illness then this might also reduce the demand for these more
“administrative” type of medical consultations.
Limitations with the
modelling—The uncertainties associated with pandemic influenza mean
that any modelling of its future impact is relatively crude. Modelling using the
FluAid model also has a number of specific limitations (e.g. it is entirely
deterministic and does not include any stochastic elements). Results from this
model could be substantial underestimates of the health impact of the next
pandemic for the following reasons:
In
contrast, the results could be overestimates of the health impact of the next
pandemic for the following reasons:
Another limitation with this
modelling is that it does not consider any differential risks for adverse
outcomes among particular population groups such as Maori and Pacific Peoples.
Even in non-pandemic years Maori and Pacific Peoples have relatively higher
rates of hospitalisations and deaths for respiratory infections (primarily
pneumonia/influenza).34 This difference may be
due to higher prevalence rates of high-risk conditions (e.g. diabetes and
chronic lung disease) but other factors might also be relevant (e.g. higher
levels of disease transmission in situations of over-crowded
housing).35 Similarly, the model does not
consider rurality—despite past evidence for pandemic influenza having less
impact in rural settings.3
Further
research—This modelling could be further refined to address some of
the limitations detailed above. Nevertheless, such refinements are unlikely to
address some of the fundamental uncertainties around the basic biological
characteristics of a new emergent strain of natural or engineered pandemic
influenza. Further work could also be done to access the impact of pandemic
influenza on the economy and society (e.g. as done in the
US16). Such impacts could in turn have indirect
affects on health outcomes. For example, a major downturn in tourism associated
with a pandemic could impact on health via increased unemployment rates and
poverty levels.
Summary—This
modelling work has a number of limitations and so these results could still
substantially over- or under-estimate the impact of the next influenza pandemic.
Nevertheless, the potentially severe impact of pandemic influenza on population
health and health sector capacity provides a strong case for health authorities
to intensify preparatory efforts and to strengthen health sector infrastructure.
Author information:
Nick Wilson, Senior Lecturer (Public Health), Department of Public Health,
Wellington School of Medicine and Health Sciences, Otago University, Wellington;
Osman Mansoor, Public Health Physician, Public Health Consulting Ltd,
Wellington, Michael Baker, Senior Lecturer (Public Health), Department of Public
Health, Wellington School of Medicine and Health Sciences, Otago University,
Wellington
Acknowledgements:
The work was part-funded by the New Zealand Ministry of Health but the views
presented do not necessarily represent Ministry of Health policy. Peter Dunn and
an anonymous reviewer provided helpful comments on the draft.
Correspondence: Dr
Nick Wilson, Department of Public Health, Wellington School of Medicine &
Health Sciences, PO Box 7343 Wellington South; email: nwilson@actrix.gen.nz
References:
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| | Current
issue | Search journal |
Archived issues | Classifieds
| Hotline (free ads) Subscribe | Contribute | Advertise | Contact Us | Copyright | Other Journals |
|