Development of NOO e(economics)-tool

1
Development of NOO e(economics)-tool

• Jane Wolstenholme
• HERC, University of Oxford
• Jane.wolstenholme_at_dph.ox.ac.uk

2
Collaborators and acknowledgements

• Dr Nick Cavill (National Obesity Observatory)
• Dr Harry Rutter (LSHTM)
• Hywell Dinsdale (National Obesity Observator)
• Funded by Department of Health
• Expert panel
• Alastair Fisher NICE
• Peter Dick, DH
• Ric Fordham, University of East Anglia
• Phil James, IOTF/IOASO
• Carolyn Summerbell, University of Durham
• Louise Woolway, NHS Somerset
• Adrian Coggins, West Essex PCT
• Corinna Hawkes,
• Lesley Manning, Buckinghamshire PCT
• William Hollingworth, University of Bristol
• Carol Weir, Sheffield PCT

3
Background

• Limited evidence on c-e of weight/obesity
  interventions/programmes
• Practical problem of funding being allocated to
  those programmes where evidence base is strong
• Call for decision aid/tool to help make these
  resource allocation decisions in the area of
  obesity/overweight
• Development of National Obesity Observatory (NOO)
  economics/cost-effectiveness tool
• NOO e-tool
• AIM To help the public health community (namely
  commissioners involved with commissioning public
  health interventions) make informed decisions
  about the commissioning of obesity interventions
  through a practical guide/e-tool on the
  cost-effectiveness and cost/economic impact of
  interventions.

4
Objectives

1. To conduct a rapid review of methods and tools
   used by other agencies/analysts when making
   resource allocation decisions related to obesity
   and overweight interventions.
2. Advise on options for approaches for development
   of practical tool To refine and agree a
   recommended approach to providing pragmatic
   cost-effectiveness estimates for obesity
   /overweight programmes/interventions at a local
   level
3. To develop, test and launch a practical tool
   based on this agreed approach

5
Wish List

• The e-tool should address the need for
  information on the cost effectiveness of
  interventions that are likely to be of value in
  preventing/reducing obesity in the local
  population.
• It should bring together the available data,
  evidence and best practice relating to cost
  effectiveness into one resource.
• Be accessible and easy to use
• Transparent and evidence based
• Enable users to input their own data and
  assumptions

6
Overview of research plan

• Rapid review of the literature
• The aim of the rapid review is to provide
  information on what tools are currently being
  used to save the potential of replication.
• Tool development and production
• Starting point for the tool development, two or
  three interventions will be chosen where the
  evidence-base in terms of effectiveness and
  associated costs have been well defined. This
  will provide a benchmark for the tool development
  and define what such a tool could provide in
  terms both the inputs and outputs.
• E-tool testing and refinement
• Use of user friendly transparent platform for
  development and testing (Excel)
• Refinement translate into web-based platform?
• Testing using expert panel and HERC and NOO
  researchers.

7
Rapid review review of reviews

• The literature search resulted in 517 potential
  publications.
• Full text review of 32 reports/reviews
• Resulted in 22 of interest
• Additional 8 reports/toolkits not from literature
  search
• n30 reports/reviews/toolkits

8
Rapid review types of review study

• Reviews of the literature on economic evaluation
  of programmes/interventions aimed at obesity
  overweight n10
• Primary study using effectiveness review to
  inform cost-effectiveness ratios (Wu 2011)
  defined cost per MET hour (ratio of expended
  energy/resting energy based on body size)
• Review of economic evaluation plus model
  development (HTA 2011 (15)44, HTA 2004 (8)21,
  Jacobs van der Bruggen et al. 2009).
• Model/toolkit n8 (Cecchini 2010 (OECD), Carter
  2009, Haby 2006, Forster 2010 (ACE-Obesity),
  Mernagh (NZHTA) 2010, Roux 2008, Galani 2007,
  Bemelmans 2008)

9
Additional reports/toolkits

• Health England Leading Prioritisation (H.E.L.P)
  online tool MATRIX -provides cost-effectiveness,
  impact on health inequalities and reach of 17
  interventions (comprising programmes related to
  alcohol use, mental health, obesity)
• Foresight 2007. (McPherson K, Marsh T, Brown M.
  Foresight Tackling Obesities Future Choices
  Modelling Future Trends in Obesity and the Impact
  on Health. 2007)
• WSIPP (Washington State Institute for Public
  Policy. An Economic Model to Inform Investment
  Decisions made by Central and Local Government.
  2012. Social Research Unit)
• Brunel/Nottingham tobacco control model (HERG
  Brunel University, QMC Nottingham University,
  London Health Observatory. Building the business
  case for Tobacco Control. A toolkit to estimate
  the economic impact of tobacco. 2011).
• Department of Education Family Savings Calculator
  (http//www.c4eo.org.uk/costeffectiveness/supportd
  elivery.aspx )
• NICE Clinical Guideline 43 (NICE. Obesity the
  prevention, identification, assessment and
  management of overweight and obesity in adults
  and children. Clinical Guidedline CG43. 2006).
• ScHARR diabetes prevention model (Gillett M,
  Brennan A, Blake L. Prevention of type2 diabetes
  preventing pre-diabetes among adults in high-risk
  groups. Report on use of evidence from
  effectiveness reviews and cost-effectiveness
  modelling. 2010. NICE Public Health
  Collaborating Centre)

10
Final useful models

• Sassi / OECD
• ACE
• Mernagh (Van Baal)
• Ara HTA 2012 16(5) (updated HTA report on
  interventions for obesity)
• HELP - Matrix
• Smoking Cessation Brunel developed for NICE
• WSIPP
• C4EO

11
Challenges to the development of the e-tool 1)
Scope of cost inputs

• Interventions aimed at tackling overweight and
  obesity have the following economic costs
• To the health sector via the health care system
  for treatment of obesity and its complications,
  intervention costs e.g. equipment, training,
  materials, clinician visits etc, health care
  costs related to diseases and complications
  resulting from obesity/overweight.
• To the individual in terms of time spent
  undertaking lifestyle/behavioural intervention,
  out of pocket expenses (e.g. equipment, clothing
  etc).
• Intersectoral impacts social care, criminal
  justice, voluntary, education, housing,
  transport, environment
• To society in terms of lost workdays
  (absenteeism) and loss of productivity while at
  work (presenteeism) .

12
Costs of health conditions

• Healthcare costs of condition x ( per year)
• Does this change by time since diagnosis?
• Changes by sex and age?
• Effects of comorbidities?

13
Costs of Diabetes Comparison of Ara (2012),
Forster (2011), Van Baal (2008)
14
Challenges to the development of the e-tool 2)
Scope of outcomes

• Economic outcomes from the rapid review ranged
  from
• productivity based-absenteeism
• life years gained and survival
• -combined with measures of quality of
  life/wellbeing in the form of DALYs and QALYs
  (although utility measures tended to come from
  1-2 sources DA from WHO-CHOICE, QA from Macran
  and HSE
• simplistic clinical measures such as MET hours
  (ratio of expended energy/resting energy, based
  on body size), BMI, activity levels, body weight,
  cholesterol level
• Numbers of individuals with chronic disease (CVD,
  stroke, diabetes, cancer)

15
Challenges to the development of the e-tool 3)
Link between intervention impact on clinical and
economic outcome

• How to establish the link between the
  intervention impact on intermediate outcomes and
  long-term quality of life and mortality outcomes
  (LY gained, QALYs DALYs).
• In general this was undertaken using Markov,
  simulation or disease models and using clinical
  outcomes e.g. BMI, potential impact fractions
  (PIFs), MET hours of energy expended, levels of
  overweight etc to predict disease conditions such
  as CVD, cancer, diabetes etc and using the
  diseases as a vehicle for health care costs,
  productivity losses and utility
  measures/survival.
• Requires evidence base to inform modelling risk
  equations

16
RR estimates for diabetes among obese
individualsComparison of NAO (2001), Van Baal
(2008), Forster (2011)
17
RR estimates for diabetes among obese
individualsAra (2012)
18
Challenges to the development of the e-tool 4)
Time horizon

• All the models in the rapid review used a
  lifetime horizon
• Except n1 (Bemelmans 2008) who used a 20 and 80
  year follow up period
• All additional reports/models use lifetime
  horizon apart from Tobacco control toolkit where
  model outputs are split analysis into 3-time
  horizons
• Short-term outcomes (2 years) GP and practice
  nurse consultations, outpatient attendances,
  prescriptions, hospital admissions, numbers of
  people with smoking related disease
• Medium term (10 years) costs of smoking related
  conditions (lung cancer, CHD, COPD, MI and
  stroke), productivity losses.
• Long-term (lifetime) number of deaths and life
  years, treatment costs, QALYs.
• DoE savings calculator SROI(social return on
  investment) over life time or shorter periods
  e.g. 2 yrs
• Commisssioners work to shorter time-horizons and
  want to know how their current investment in
  obesity interventions will impact on costs and
  outcomes in the next 1-2 years

19
Challenges to the development of the e-tool 5)
consider a portfolio of interventions?

• In practice obesity interventions are rarely
  commissioned in isolation.
• All reviews/models from rapid review explore
  interventions in isolation
• Problem is how to model the correlation between
  the interventions and the impact they may
  collectively have on obesity

20
Challenges to the development of the e-tool 6)
Maintenance of impact of intervention ?

• Assumptions need to be made concerning the
  maintenance or decay in impact of intervention
• Cecchini et al, assumed the impact of the
  intervention to disappear once exposure to the
  intervention ends
• ACE-obesity assumed 100 of benefits to be
  maintained over the lifetime of the model
• Mernagh and colleagues assumed a reduction in BMI
  relative to control to decay by 1 per annum over
  the lifespan of the model after the 5th year
  following the initiation of the intervention

21
Challenges to the development of the e-tool 7)
Impact of intervention

• Reduction in BMI (mean? Mean SD? moving from
  obese to overweight?)
• How long before this weight loss is achieved?
  (years? Years months?)
• How long is this weight loss maintained? (years?
  years months?)
• Duration of intervention?
• Uptake period for intervention (number of years
  before full build up)?
• Drop out rate? ( per year?)

22
Additional challenges

• Availability of data to populate the model
  local v national data
• Use of sensitivity/what-if analysis - this was
  used widely in the models found in the rapid
  review, but tended not to be used in toolkits.

23
Systematic approach?2 different approaches?

• NICE time horizon so as to incorporate all
  important costs and effects usually lifetime,
  CUA (CCA and CBA secondary analysis), NHS/PSS or
  public sector perspective (productivity costs
  excluded), use of sensitivity analysis. National
  level evaluation.
• Commissioners 1-2 year time horizon (also
  interested in lifetime impacts), outcomes
  (broader perspective) impact on clinical
  indicators of obesity, BMI, related diseases etc,
  public sector impacts and costs NHS/PSS and other
  sectors, business costs productivity. Use of
  sensitivity and what-if analysis. Local level
  evaluation.

24
Over to you wish list

• Is it the local level you require? Or do you
  need more?
• Are we trying to be too all encompassing?
• What are the key costs/outcomes/benefits you
  require to make your case?