Professional Documents
Culture Documents
Modelling To Support A Future Covid 19 Strategy
Modelling To Support A Future Covid 19 Strategy
Modelling To Support A Future Covid 19 Strategy
23 September 2021
Executive Summary
• Aotearoa New Zealand is on track to vaccinate upwards of 80% of those aged over 12 against
COVID-19 with the Pfizer/BioNTech mRNA vaccine. Recent announcements by Pfizer and BioNTech
suggest that the vaccine may soon be approved for use in children aged 5-11 years. This means it
is possible that Aotearoa could achieve vaccine coverage across the total population of more than
90%.
• We consider how these high rates of coverage might reduce the health burden from COVID-19 if
combined with moderate public health measures to reduce transmission of the virus. Scenarios are
evaluated using the Te Pūnaha Matatini vaccine model, using current data about vaccine
effectiveness with respect to the Delta variant. As the effects of the vaccine on transmission remain
uncertain, we consider three levels of vaccine effectiveness: high, central, and low to illustrate a
range of possibilities.
• Attaining population immunity would significantly reduce health impacts, but this a function of the
particular characteristics of the virus, the structure of the population, the distribution of immunity in
that population, and other measures that can be put in place, all of which can change over time.
Nonetheless, it is not thought to be possible to achieve widespread population immunity to the Delta
variant by vaccination alone, which means that moderate public health measures will be needed to
augment the vaccine programme.
• Here we assume a suite of moderate and sustainable public health measures are maintained, such
as some combination of mask use, ventilation, some density/capacity restrictions for indoor venues,
support for people to isolate, and/or vaccine certificates. We also assume that a test-trace-isolate-
quarantine system is operating either at levels that are high (most likely at low case-loads) or reduced
(most likely at high case-loads). We do not consider the use of stay-at-home orders here (i.e. Alert
Level 3 or 4).
• We find that there are scenarios where, through a combination of high vaccine coverage (including
amongst those aged 5-11) and moderate public health measures, population immunity is achieved,
resulting in very low mortality burden. For example, with 90 per cent vaccine coverage of the
population over the age of 5, a suite of moderate public health measures and an effective test, trace
and isolate system, the modelling suggests there would be around 500 hospitalisations and 50
fatalities from COVID-19 over a one year period.
• There are scenarios where, despite a high vaccination coverage, population immunity is not
achieved, resulting in a disease mortality burden that is an order of magnitude greater. For example,
with 80 per cent vaccine coverage of the population over the age of five and moderate public health
measures, the modelling suggests there would be around 60,000 hospitalisations and 7,000 fatalities
per year from COVID-19. Such outcomes could be mitigated if more restrictive control measures,
akin to current Alert Levels 3 or 4, were utilised.
Introduction
As of 21 September 2021, 79% of Aotearoa New Zealand’s eligible population have either had at least one
dose of the Pfizer-BioNTech vaccine against COVID-19 or are booked to have a dose [1]. Currently this
vaccine is approved in Aotearoa for people over 12 years of age. On 20 September 2021, Pfizer and
BioNTech announced that trials for this vaccine for 5-11 year olds had proved successful, in that the vaccine
had been found to be safe and effective, and that they intended to apply for approval around the world [2].
This raises the possibility that Aotearoa may be able to achieve very high levels of vaccine coverage,
something that would greatly reduce the mortality burden from COVID-19 if combined with moderate public
health interventions.
In order to keep annual mortality from COVID-19 at levels similar to or below those experienced from
seasonal influenza, for instance, the effective reproduction number, 𝑅!"" , of the virus must be kept close to
or below 1 by a combination of vaccination, targeted measures such as case isolation, and other control
measures. Results from both Te Pūnaha Matatini’s SEIR vaccination model [3] and the ESR model [4]
suggest that if the reproduction number is even slightly greater than 1, fatalities would be measured in
thousands, despite high vaccine coverage. As the Delta variant is thought to have a reproductive number
of 𝑅# ~ 5.5 − 6.5, population immunity (𝑅!"" < 1) is likely to be extremely difficult to achieve by vaccination
alone. Thus it is probable that other control measures will be needed on top of vaccination to deal with
COVID-19 outbreaks for the foreseeable future. In this short note we consider scenarios with different
combinations of vaccine effectiveness, vaccine coverage, and additional public health and social measures.
The work here is based on Te Pūnaha Matatini’s vaccination model [3]. This model is age-structured, but
otherwise does not capture population heterogeneity in vaccination or contact rates. This means that while
𝑅!"" could be less than one, on average, across the population, there may still exist communities with 𝑅!"" >
1 in which a significant outbreak could occur. It is also likely that 𝑅!"" would vary seasonally and is likely to
be higher during school and university terms and higher in winter than in summer months. Thus, estimates
of fatalities made here may be optimistic. Furthermore, we have used baseline assumptions about the
effectiveness of the vaccine [5,7] that do not consider immune escape or waning immunity.
Nonetheless the results here demonstrate the considerable benefits of achieving high vaccination coverage
in the coming months. It suggests that a combination of high levels of vaccination within the community, a
strong test-trace-isolate-quarantine (TTIQ) system and moderate public health measures may be enough to
attain population immunity, which would greatly reduce the need for stay-at-home orders and tight border
restrictions in 2022.
Lower Higher
Effectiveness Central
Effectiveness Effectiveness
Against infection 70% 50% 90%
Against transmission given infection 50% 40% 50%
Against severe disease given infection 80% 80% 80%
Overall transmission reduction 85% 70% 95%
Overall protection against severe disease 94% 90% 98%
Table 1. Vaccine effectiveness parameters chosen to reflect estimates of the effectiveness of the Pfizer-BioNTech
mRNA vaccine after 2 doses [5].
• Simulated outbreaks are seeded with an average of 1 case per day arriving at the border and entering
the community. This approximately represents a situation where current tight border restrictions are
relaxed, but strong border controls remain in place to limit the number of infectious travel-related
cases entering the community Results are relatively insensitive to this assumption when 𝑅!"" > 1,
but will be sensitive to the border settings and corresponding in flow of infected travellers for
scenarios where 𝑅!"" < 1.
• Age-stratified hospitalisation rates are as in [10] with a hazard ratio of 2.26 representing the increased
severity of the Delta variant [6].
• The risk of fatality given hospitalisation is as in 103]; see also [6].
• Mean length of hospital stay = 8 days [11].
Key caveats
• Although this model is age-structured, it neglects other forms of heterogeneity in the population. In
particular, it neglects homophily in vaccine coverage, whereby people who are unvaccinated also
tend to have contacts who are unvaccinated. Thus a high average coverage that allowed for
population immunity with other measures may still leave small groups of people or communities
below the required threshold, making them vulnerable to outbreaks and health impacts.
• There is still a high degree of uncertainty about the effectiveness of vaccines against the Delta
variant, particularly for breakthrough infection and subsequent propensity to transmit. Furthermore
immunity may wane over time. Our central vaccine effectiveness against infection and severe illness
is comparable to that reported by Public Health England on 17 September 2021 for the Delta variant
[7].
• Model outputs are sensitive to the assumed value of 𝑅# = 6 and assumed reductions in transmission
due to baseline public health measures (17%), partial TTIQ (10%) and full TTIQ (20%) [7]. These are
based on a combination of empirical data and modelling assumptions and are subject to uncertainty.
In addition, the effect of such control measures may be different in future as a result of changes in
behaviour in an environment where the population is highly vaccinated.
Results for an epidemic with baseline public health measures and TTIQ
The results presented here are for a 12-month period following the start of an outbreak. We have not
included the effects of individual public health measures in this study. Instead we assume that a moderate
suite of measures are in place and that these can be sustained for a long period of time. Such measures
may include a combination of interventions such as:
• Improved air filtration and ventilation requirements, or where these cannot be met, mandatory mask
use or some density or capacity restrictions for indoor venues
• Vaccine passports
• Support for people to isolate
• Rapid-testing at workplaces and schools
1. Baseline public health measures (17% reduction in transmission) and limited TTIQ (10% reduction
in transmission), giving a combined 25% reduction in 𝑅!"" .
2. Baseline public health measures (17% reduction in transmission) and full TTIQ (20% reduction in
transmission), giving a combined 33% reduction in 𝑅!"" .
The first of these could be interpreted as a scenario where measures to contain outbreaks are relaxed or fail
and a mitigated epidemic wave spreads through the population. High case numbers would mean that TTIQ
is less effective and may rely primarily on widespread testing and isolation of symptomatic individuals and
confirmed cases. In the second case, we assume that a fully effective TTIQ system can be maintained when
case numbers are sufficiently low. Here, effective performance is set to be similar to that observed early in
the Auckland August 2021 outbreak, based on the time cases were isolated or quarantined relative to the
time of symptom onset. Future work will explicitly model the performance of the TTIQ system under different
caseloads, but it is important to note that in general it will be easier to keep 𝑅!"" supressed when overall
case numbers are low.
We consider a range of scenarios that combine a mix of vaccine effectiveness assumptions, vaccine
coverage levels, and the two TTIQ performance tiers described above. This model does not attempt to
capture in detail the effect of border controls. In scenarios where 𝑅!"" > 1, border controls have less of an
effect on outcomes (unless very stringent), but they become important for determining cases when 𝑅!"" < 1.
The scenarios here best represent situations with moderate border controls in place. In scenarios where
𝑅!"" < 1, real outcomes will be highly variable depending on the nature of each chain of infection and will be
sensitive the number of border-related cases, which are not modelled here. Nonetheless, provided
outbreaks are only seeded into communities where 𝑅!"" < 1, the mortality burden is expected to be lower
than or comparable to seasonal influenza. Note that these scenarios ignore the effects of waning immunity,
which may well be significant over the timescales considered here, but which could be mitigated with
vaccine boosters.
Figure 1. Average new daily SARS-CoV-2 infections, number of COVID-19 patients in hospital, and cumulative COVID-
19 deaths over a 365 day period for different levels of vaccine coverage (from 70% to 95%) in: over 12s with limited
TTIQ (first row), over 5s with limited TTIQ (second row), over 12s with full TTIQ (third row), over 5s with full TTIQ (fourth
row). Results shown are for the central vaccine effectiveness assumptions.
However if vaccination coverage is increased to 95% and/or into the 5-11 age group then the results suggest
that moderate public health measures can result in 𝑅!"" < 1, leading to considerably better outcomes. A
combination of high levels of vaccination within the community, a strong test-trace-isolate-quarantine
system, and moderate public health measures would greatly reduce the need for stay-at-home orders and
workplace closures.
The results suggest that it would be valuable to develop a suite of moderate public health interventions that
can be sustained for a long period of time if required. Interventions such as mandated mask use, ventilation
requirements and some density or capacity restrictions for indoor venues, rapid-testing at workplaces and
schools, and support for symptomatic people, confirmed cases and close contacts to isolate should be
investigated. Effort should also be given to strengthening the test-trace-isolate-quarantine system, which
would be expected to operate for extended periods with low to moderate levels of cases in the community,
possibly punctuated by larger outbreaks in parts of the community where vaccine coverage is lower due to
reduced uptake or waning immunity. Such outbreaks may require significant surge capacity and may need
to be accompanied by more stringent public health measures. However, the higher the vaccine coverage
achieved, the less such measures will be needed.
The purpose of this report is to provide a broad guide to the relative benefits of different levels of vaccine
coverage in terms of reducing potential health impacts of COVID-19 and therefore lessening the need for
stringent restrictions to reduce transmission of SARS-CoV-2. Decisions about the nature and timing of
changes in public health measures and border controls should be informed both by more detailed modelling
based on up-to-date national and international data and by specific public health considerations. These
include: current case numbers and transmission rates; vaccine coverage in specific communities and
regions, including Māori and Pasifika vaccination rates; regional and national health system capacity and
demands and contact tracing system capacity; seasonal factors including, for example, school/university
terms and faster transmission in winter months; epidemiological characteristics of recent and current
outbreaks.
Future modelling work will focus on specific public health interventions in more detail, including how these
interventions interact with health system and contact tracing system capacity, border controls, and the
frequency at which more stringent interventions may be needed. When 𝑅!"" < 1, the health outcomes are
more moderate, but do become sensitive to the particular choice of border restrictions. Future work will
investigate the relationship between outcomes and border restrictions.
Acknowledgements
This work was funded by the New Zealand Ministry of Business, Innovation and Employment and Te Pūnaha
Matatini, Centre of Research Excellence in Complex Systems.
References
[1] Ministry of Health, COVID-19: Vaccine data. Retrieved 22 September 2021.
https://www.health.govt.nz/our-work/diseases-and-conditions/covid-19-novel-coronavirus/covid-19-data-
and-statistics/covid-19-vaccine-data
[2] Pfizer Media Release, “Pfizer And BioNTech Announce Positive Topline Results From Pivotal Trial Of
Covid-19 Vaccine In Children 5 To 11 Years”. 20 September 2021. https://www.pfizer.com/news/press-
release/press-release-detail/pfizer-and-biontech-announce-positive-topline-results
[3] N. Steyn, M. Plank, R. Binny, S. C. Hendy, A. Lustig, and K. Ridings, “A COVID-19 Vaccination Model for
Aotearoa New Zealand” Te Pūnaha Matatini Technical Report June 2021.
https://www.tepunahamatatini.ac.nz/2021/06/30/a-covid-19-vaccination-model-for-aotearoa-new-
zealand/
[4] T. Nguyen et al, “COVID-19 vaccine strategies for Aotearoa New Zealand: a mathematical modelling
study” Lancet Regional Health - Western Pacific 15, 100256 (2021).
[5] Scientific Pandemic Influenza Group on Modelling, SPI-M-O: Summary of further modelling of easing
restrictions – Roadmap Step 4. 2021. Retrieved 21 Sept 2021.
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/99351
0/S1287_SPI-M-O_Summary_Roadmap_step_4.pdf
[6] K. A. Twohig et al, “Hospital admission and emergency care attendance risk for SARS-CoV-2 delta
(B.1.617.2) compared with alpha (B.1.1.7) variants of concern: a cohort study” Lancet Infectious Diseases.
Published Online August 27, 2021.
[7] Public Health England. COVID-19 vaccine surveillance report: Week 37. Retrieved 21 Sept 2021.
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/10184
16/Vaccine_surveillance_report_-_week_37_v2.pdf
[8] Doherty Institute. “Doherty Institute Modelling Report for National Cabinet”. Retrieved 21 Sept 2021.
https://www.doherty.edu.au/uploads/content_doc/DohertyModelling_NationalPlan_and_Addendum_2021
0810.pdf
[9] Office for National Statistics. Technical article: Updated estimates of the prevalence of post-acute
symptoms among people with coronavirus (COVID-19) in the UK, 26 April 2020 to 1 August 2021.
Retrieved 21 Sept 2021.
https://www.ons.gov.uk/releases/technicalarticleupdatedestimatesoftheprevalenceofpostacutesymptomsa
mongpeoplewithcoronaviruscovid19intheuk26april2020to1august2021
[10] R. Verity et al, “Estimates of the severity of coronavirus disease 2019: a model-based analysis”. The
Lancet Infectious Diseases, 2020. 20(6): p. 669-677.
[11] N. Steyn et al, “Māori and Pacific people in New Zealand have a higher risk of hospitalisation for
COVID-19”. New Zealand Medical Journal 134(1538): 28-43.