user guide to the economic model

Model User Guide Ver. 1.05
BUILDING THE ECONOMIC CASE FOR TOBACCO CONTROL:
A TOOLKIT TO ESTIMATE ECONOMIC IMPACT OF TOBACCO
USER GUIDE TO THE ECONOMIC MODEL
December 2011
2|Page
Model User Guide Ver. 1.05
Background to the project
The Health Economics Research Group (HERG) at Brunel University, in partnership with London
Health Observatory and Queen’s Medical Centre, University of Nottingham, was commissioned to
develop an economic model to estimate the economic impact of tobacco. The model was intended
to support commissioners, in health and local authorities, to consider the impacts of tobacco on a
range of health and non-health outcomes measured in financial and non-financial terms.
This economic impact of tobacco model builds on previous analyses of tobacco control and smoking
cessation. Previous economic models of tobacco use typically consider the costs and benefits that
are accrued from the use of a particular intervention (e.g. NRT) or alternatively attempt to estimate
the economic impact of smoking to the National Health Service. Whilst many of these analyses are
robust and well developed, arguably, they lack relevance to a commissioning audience as they often
reported aggregated findings over the entire population, taking into account the lifetime impact of
smoking.
This latest model has been developed to support sub national/local commissioners and decision
makers, and incorporates a number of modifications from previous analyses. Crucially, the model
attempts to generate outcomes in the short, medium and long-term, allowing commissioners to
estimate the short-term return on investment in tobacco control initiatives. Whilst longer term
outcomes remain important, it is acknowledged that the current funding environment means that
commissioners may be focussed on returns over a shorter period of time.
Secondly, the model attempts to capture the impacts of tobacco beyond the impact on healthcare
costs by capturing evidence on productivity losses. This will broaden the relevance of the findings to
commissioners and stakeholders outside of healthcare, including local authorities and private sector
employers. Finally, the model also attempts to capture the impact of passive smoking on adults and
children. This has not been attempted in previous analyses and adds an extra dimension to the
current model.
Finally, the model considers the impact of a comprehensive tobacco control programme, coordinated at a sub-national level, in addition to smoking cessation interventions. This not only
reflects best practice, in terms of reducing smoking prevalence and uptake but also reflect current
policy initiatives which recognise the importance of promoting tobacco control alongside cessation.
3|Pa ge
Model User Guide Ver. 1.05
Model features and data
Based on feedback from the stakeholder consultation, the model is built in Microsoft Excel in order
to ensure that it is readily available to commissioners. The model is pre-populated with data on
population statistics and smoking prevalence using data from the Integrated Household Survey, coordinated by the Office of National Statistics. This constitutes the default data on the basis that it is
held to be the most accurate national source of evidence on smoking prevalence.
The model is intended to be an interactive tool, for use by commissioners and planners. As such,
many of the parameters in the model can be amended to reflect local circumstances. These include
local prevalence estimates, wage rates and absenteeism associated with smoking and the underlying
quit rate.
The model assumes that there exists an underlying ‘background’ quit rate occurring due to social
norm changes and self-motivation amongst current smokers to quit, assisted or unassisted. The
assumed default background quit rate is 2%. Note that ‘background quit rate’ refers to ‘overall quit
rate’, and not the rate that is applicable to only those smokers who quit unassisted.
The model allows the user to explore the impact of a collectively commissioned sub-national
tobacco control programme as well as the provision of local tobacco control interventions. Subnational strategies aim to change social norms through a combination of education and awareness,
health promotion activities and facilitating access to higher quality services. Domestically, current
evidence is somewhat limited, although the regional programme co-ordinated across the North-East
of England by FRESH, has resulted in smoking prevalence falling at a more rapid rate than the
national average. International evidence suggests that tobacco control programmes co-ordinated
over a significant sub-national geographic footprint can accelerate the reduction of smoking
prevalence.
On the best available data, we estimate that the presence of a sub-national programme coordinating a package of tobacco control interventions might increase the background quit rate from
2% to 5%. This increase in background quit rate from 2% to 5% is possible because sub-national
programmes tend to increase the number of smokers who are motivated to access Stop Smoking
services, which in turn increases the likelihood of them quitting permanently and also impact on the
number of new smokers, through improved tobacco control. All of this effect is reflected in this
increase from 2% to 5% in background quit rate. So, the model works on the basis of current
smoking pool but allows for the increases in number of smokers both quitting unassisted and
quitting with support from Stop Smoking Services due to sub-national programmes via the new
background quit rate (i.e. 5%). This is so because the evidence base as to how many more smokers
would access Stop Smoking Services if sub-national programmes were implemented is not
4|Pa ge
Model User Guide Ver. 1.05
sufficiently robust to model the effect of sub-national programmes on the number of current
smokers accessing Stop Smoking Services.
The model allows the user to estimate the economic impact of tobacco on their local population
varying the use of smoking cessation interventions and assuming the absence or presence of a subnational tobacco control programme. It is hoped that the outputs can be used to help inform local
investment decisions relating to tobacco control.
5|Pa ge
Model User Guide Ver. 1.05
Opening the model
The model is built in Microsoft Excel, to ensure that it is easily accessible for commissioners.
NOTE: the model includes ‘macros’ and as such the user will have to enable content prior to starting
the model. This can be done by clicking the button in the header highlighted in the screenshot
below.
This will then open the model. This may take a short time during which you will see a message
stating ‘Initialising Model.’
6|Page
Model User Guide Ver. 1.05
Introduction page
The pop-up box has sheet tabs (highlighted in the figure below) which take the user to introductions,
model structure, inputs and disclaimer. The first tab on the pop-up box is the introduction page.
This provides an overview of the model as well as details on how this should be cited.
7|Page
Model User Guide Ver. 1.05
Model structure
The second sheet tab takes the user to the model structure. The model is a simple ‘Markov’ model
which includes three states:- smokers, former smokers and death.
The model considers the population of a specific region/area. The population is divided into current
smokers, former smokers and non-smokers, who may enter the model if they take up smoking (‘new
smokers’). The risk of developing any of the smoking related outcomes specified in the model varies
according to which state an individual is in – i.e. a current smoker will have the highest risk of
developing an adverse outcome, followed by a former smoker, followed by a non-smoker.
Individuals can move between each of the health states each year. That is, in any given year an
individual might start smoking, continue to smoke, stop smoking and become a former smoker or
die. The model follows the cohort over an entire lifetime but also reports short and medium term
outcomes as described below.
8|Pa ge
Model User Guide Ver. 1.05
Input screen
The input screen allows the user to specify what data is to be used in the model. This includes the
population under consideration, smoking prevalence and the use of smoking cessation services. In
most instances these inputs are populated automatically but the user can over—ride these default
inputs.
Selecting the population: the user can select their local population, categorised as top tier (subnational), second tier or third tier. Population data for these are already included in the model along
with estimates of smoking prevalence. In the example below, we have randomly selected Salford in
the North-West which results in the population of 184,000 and a smoking pool (i.e. current smokers)
of just over 53,000.
9|Page
Model User Guide Ver. 1.05
Use of NHS Stop Smoking Services: The model automatically calculates how many current smokers
will access smoking cessation therapy provided by NHS Stop Smoking Services. Cessation therapy is
categorised as monotherapy NRT, combination NRT, Buproprion or Varenicline all of which are
provided in combination with behavioural support.
The default allocation of smokers to each of these services is based on the best available evidence
on current use at a national level. However, the user can easily over-ride these values and
incorporate local values where these are known, either by incorporating actual numbers of service
users or using the up/down buttons in the table.
Use of other cessation therapies: the model makes an assumption about the proportion of current
smokers who access cessation services without recourse to NHS Stop Smoking Services. Based on
the best available evidence at a national level, this is assumed to be 17.2% of all smokers. This
default figure can be changed to reflect local uptake where this is known. However, given the
difficulties in estimating access to non-NHS services, this functionality is currently locked in the
model and can only be changed by users with a password. (Details on how to obtain a password are
provided later in the guide).
The final row in the table summarises access to smoking cessation therapy provided by the NHS and
privately.
10 | P a g e
Model User Guide Ver. 1.05
Running the model
Once the user is satisfied with the input values, simply click ‘Run Model’ to generate the outputs.
The model takes around one minute to run on most computers, during which time you will see a
screen which reports on the progress of the model run.
11 | P a g e
Model User Guide Ver. 1.05
Outputs
The outputs are reported at three intervals
1. Short-term, which summarises the outcomes over two years
2. Medium term, which summarises the outcomes over ten years
3. Long-term, which summarises the lifetime costs and outcomes.
Furthermore, three scenarios are also included in the analysis:
•
•
•
Scenario 1 assumes that there are no NHS Stop Smoking Services made available and no subnational tobacco control programmes in place so any quitting results from self-motivation.
Scenario 2 assumes that NHS Stop Smoking Services are in place, with smokers allocated to
these services according to the user-defined inputs. However, this scenario assumes that
there are no sub-national tobacco control programmes in place.
Scenario 3 assumes that NHS Stop Smoking Services are in place and furthermore these are
complemented by a sub-national tobacco control programme.
12 | P a g e
Model User Guide Ver. 1.05
Short-term outputs
The short-term outputs focus mainly on healthcare resource use and the associated costs.
Outcomes included in the model include GP consultations, prescriptions and admissions to
secondary care. In addition to this, passive smoking episodes, quitters and productivity losses are
also reported.
The final two columns of the table report the ‘savings’ that accrue from moving from Scenario 1 to
Scenario 2 or 3. The example below, based on Salford in the North West, shows that NHS Stop
Smoking Services result in a cumulative saving to the NHS over two years of over £300,000
compared to no intervention. The introduction of a sub-national tobacco control programme leads
to a further saving to the NHS of almost £500,000 and results in an additional 2,000 quitters over
two years. The tobacco control programme has benefits which reach beyond the health service
notably, resulting in productivity gains of over £250,000 compared to the NHS Stop Smoking Service
alone.
13 | P a g e
Model User Guide Ver. 1.05
Medium term outcomes
The medium term outcomes focus mainly on episodes of smoking related morbidity and their costs
over a 10 year period. The model estimates the number of smoking attributable cases of lung cancer,
CHD, COPD, MI and stroke and estimates the cost of managing these. Once again, the addition of a
tobacco control programme on top of NHS Stop Smoking Services results in a significant saving,
estimated to be in excess of £4M over 10 years, using our example of Salford in the North-West.
14 | P a g e
Model User Guide Ver. 1.05
Long-term outcomes
Finally, the long-term outcomes in the model focus on life expectancy, quality adjusted life
expectancy (reported as a quality adjusted life years or QALYs) and lifetime treatment costs. The
number of deaths attributable to smoking in the selected population is reported, along with the
mean number of life years accrued by smokers in the population. Note, the life years does not
equate to the life expectancy of the cohort but rather reflects the average number of years lived
recognising the baseline age distribution of the population under consideration. Treatment costs
reflect the lifetime costs of treating smoking attributable conditions – these exclude any costs not
attributable to smoking.
The pop-up table reports summary findings for each of the main outcomes at each interval. Further
details are available by clicking on the ‘Export Data’ button which takes the user to the Excel
worksheets used to generate the outputs. These include counts of admissions as well as costs and
are presented in a way intended to be suitable for export to other packages, for example Microsoft
Word.
15 | P a g e
Model User Guide Ver. 1.05
Accessing the model
The model is made freely available for use and can be downloaded at
http://www.brunel.ac.uk/about/acad/herg/research/tobacco. Passwords to access protected
aspects of the model are available from the contact listed below.
Version control
Note, some of the model inputs are time limited (e.g. population statistics) and may be updated
when new data become available. As such, it is the user’s responsibility to ensure that they are
using the latest version of the model. All versions made available for download will be clearly
marked with a version number.
Referencing the model
The Health Economics Research Group at Brunel University developed this work in partnership with
London Health Observatory and Queen's Medical Centre, University of Nottingham. This work was
funded by Smoke Free North West, Fresh Smoke Free North East and Smokefree South West. Inputs
from the Steering Group and stakeholders are acknowledged. For details on how to use this model,
please refer to accompanying Report and User Guide. Also, read the disclaimer information before
you use this model.
The following citation is recommended:
Trapero-Bertran M, Pokhrel S, Trueman P. An economic model of tobacco control version 1. Tobacco
Free Futures, Fresh Smoke Free North East & Smokefree South West. December 2011.
Disclaimer information
This toolkit when available on the web can be freely accessed and used for building business cases
for tobacco control. This is primarily intended for commissioners, public health professionals, subnational managers and policy-makers. The toolkit has been developed using the current best
practice and available data, and while doing so, a number of assumptions have been made. It is the
responsibility of the users to understand the workings of the model and accept the assumptions
underlying it. The Project team and the sponsors, therefore, accept no liability for any adverse
consequence arising from the use of this toolkit. The users are expected to acknowledge the source
in any form of verbal and written communications. The accompanying report describes the
underlying principles of this toolkit and therefore the users are strongly recommended to read the
report and the User Guide beforehand.
16 | P a g e
Model User Guide Ver. 1.05
Acknowledgement
We would like to thank the members of the Steering Group: Ailsa Rutter, Andrea Crossfield, Fiona
Andrews, Adam Lester-George, Paula Wheeler, Mike Lavender, Martyn Willmore and David Phillips,
for their guidance throughout the project.
We are also thankful to all those practitioners and professionals who participated in our Stakeholder
Engagement Events and provided their feedback on various issues they would have liked to be
addressed in this work.
We particularly thank the following who attended a round table discussion in May 2011 to discuss the
development of this toolkit and who have endorsed the need for a tool capable of assessing the
impact of tobacco control at a sub-national level: Deborah Arnott, Martin Dockrell, Howard Reed,
Robert West, Ann McNeill, Mike Robinson, Lesley Owen and Matthew Glover for their valuable inputs.
We are thankful to Matthew Glover who reviewed the model internally.
Project Team
Brunel University- Paul Trueman; Subhash Pokhrel; Marta Trapero-Bertran
University of Nottingham- Tim Coleman
London Health Observatory- Bobbie Jacobson; Paul Deponte
South West Public Health Observatory- Adam Lester-George
Request to the users
This toolkit may be subject to continuous improvement. If any problem is encountered or
inconsistency is found, please report it to the project team by email.
Contact details for further information
Any further information on the model can be obtained from Dr Subhash Pokhrel at Brunel University.
Dr Subhash Pokhrel ([email protected]).
17 | P a g e
Model User Guide Ver. 1.05
Supplements:
1. MS Excel/Web-based toolkit: Economic model of tobacco control
2. User Guide: Economic model of tobacco control (this document)
3. Tobacco Control Economic Toolkit Background and Rational Report
4. Tobacco Control Economic Toolkit Technical Report
Recommended citation:
Trapero-Bertran M, Pokhrel S, Trueman P. An economic model of tobacco control version 1. Tobacco
Free Futures, Fresh Smoke Free North East & Smokefree South West. December 2011.
18 | P a g e