Difference between revisions of "Impact calculation tool"

Revision as of 12:42, 2 December 2010 (view source) Virpi Kollanus (talk | contribs) (→‎Step-by-step user instructions) ← Older edit		Revision as of 16:20, 2 December 2010 (view source) Anne.knol (talk | contribs) Newer edit →
Line 57:		Line 57:
	==Background information==		==Background information==
−	===Health impact assessment (HIA) and burden of disease (BoD)===	+	=== Environmental health impact assessment===
		+
		+	Environmental problems such as air pollution or radon can affect human health. Several scientific methods have been developed in order to estimate the extent of such environmental health impacts, or the potential impact of policy and plans on these impacts. These methods include [[HIA | health impact assessment]], [[Integrated assessment | integrated assessment]] and  [[Risk assessment | risk assessment]].
		+	When such assessments deal with complex environmental health problems, such as transport or climate change, the effects of different environmental risks often need to be compared. Traditionally, numbers of attributable deaths and disease cases were reported. Such numbers can supply useful information, but cannot always be used to compare dissimilar health effects. For example, effects from air pollution range from aggravation of asthma to premature mortality, while noise exposure is associated with annoyance, sleep disturbance and effects on cognition. Because of the divergence in magnitude, duration and severity of these health effects, summary health measures have been developed. Such measures convert all health effects to a comparable unit, and can thus be very useful for the interpretation and comparison of different (environmental) health problems. This is especially useful for evaluating and comparing different policy options and assessing cost effectiveness of mitigating measures or prevention. Summary health measures in environmental health decision-making include for example Years of Life Lost (YLL), disease burden estimates such as [[Disability-adjusted life years| Disability Adjusted Life Years]] (DALYs) or [[Monetary values for impacts to human health | monetary valuation]].
		+
		+	===Indicators that can be calculated using ICT===
		+
		+	The model output of the ICT model are the following indicators:
		+
		+	'''Number of attributable deaths'''
		+
		+	'''Number of attributable morbidity cases '''
		+
		+	'''Age-specific loss of life-expectancy for target population'''
		+
		+	'''Average loss of life-expectancy for a birth cohort'''
		+
		+	'''Years lost due to mortality (YLL)'''
		+
		+	'''Years lost due to disability (YLD)'''
		+
		+	'''Loss of disability adjusted life years (DALY)'''
−	===Environmental burden of disease (EBD)===
	===Life table modelling===		===Life table modelling===

---

Revision as of 16:20, 2 December 2010

This page is a method. The page identifier is Op_en2909
Moderator:Nobody (see all) Click here to sign up.
Give your opinion to the peer rating of the content of this page. {{ #opasnet_rater: }}
Upload data Show results

Short description

Impact Calculation Tool (ICT) is a modelling tool for quantification of health impacts from environmental exposures. It applies dynamic life table modelling for calculating target population specific mortality and morbidity impacts. The model has been developed in the context of international projects working on environmental health impact assessment (Intarese, Heimtsa) and the Finnish Academy project CLAIH, and is a collaboration between the National Institute of Health and Welfare (THL, Finland), the National Institute for Public Health and the Environment (RIVM, the Netherlands), and Netherlands Environmental Assessment Agency (PBL, the Netherlands).

ICT has been develped to provide answers to questions such as:

What is the burden of disease (BoD) attributable to a given environmental exposure (i.e. environmental burden of disease, EBD) in a given target population?

How much does the burden of disease change if the exposure changes?

ICT allows analysis of health impacts from one environmental exposure in one given population at a time. The follow-up period (time period for which impacts are determined), specific mortality and morbidity impacts analysed, and the target population can be defined according to the needs of the assessment at hand. In one model run, impacts are modelled for a reference, business-as-usual (BAU), and one alternative exposure scenario. Both time discounting and age-weighting can be applied.

Health impacts can be quantified using different approaches depending on the type of input data available for the analysis:

Modelling approach 1: Life table method

User defines exposure or health outcome scenarios and provides exposure-response relationships, background mortality and, if needed, morbidity data for the population and health outcomes of interest. The model then calculates the mortality or morbidity risks attributable to the exposure in different scenarios, makes population projections for the future using dynamic lifetables, and based on these derives the different health impact indicators.

Modelling approach 2: Calculation of EBD from total BoD

User defines exposure scenarios and provides exposure-response relationships and total BoD data (e.g. from WHO) for the health outcomes of interest. The model then calculates the fraction of BoD caused by the risk factor. In this approach, the assessment time frame is limited to the time frame of the total BoD data provided.

Input data requirements vary between different modelling approaches, but may include

• Age-specific population data
• Age-specific baseline mortality incidence and morbidity incidence/prevalence data
• Birth rate
• Exposure levels
  • Business-as-usual (BAU) scenario
  • Alternative scenario
  • Reference scenario (no exposure or natural background exposure)
• Exposure-response functions for the health endpoints of interest (relative risk or absolute risk)
• Severity weights and durations for the morbidity endpoints

The model itself does not contain a database or default values for the required input data, but functions more like a calculation shell. All key input data has to be provided by the user, and needs to be preprocessed to be in the correct form.

Model outputs vary between the different modelling approaches, but include

• Loss of disability adjusted life years (DALY)
  • Years lost due to mortality (YLL)
  • Years lost due to disability (YLD)
• Age-specific loss of life-expectancy for target population
• Average loss of life-expectancy for a birth cohort
• Number of attributable deaths
• Number of attributable morbidity cases

ICT runs in Analytica software, which enables probabilistic modelling using Monte Carlo simulation and, therefore, advanced uncertainty analysis. Probabilistic modelling of uncertainty requires the key model inputs to be defined in terms of probability distributions. As a result, the uncertainty in the model outputs can be viewed with basic statistical descriptors, probability bands, probability density functions, and cumulative probability density functions.

Full use of the Analytica programme requires a software licence. However, ICT can also be run with a free Analytica player, which can be downloaded from lumina.com. The player allows the user to view to the model contents and calculation specifics, to input data for key parameters, and to calculate results and run probabilistic uncertainty analysis. ICT contains a simple user interface, which enables these functions without advanced knowledge of Analytica or the model technicalities. Analytica also has several built-in functions for conducting sensitivity analyses for both deterministic and probabilistic analyses. However, these functions are not incorporated into the user interface and their use requires advanced model editing.

Background information

Environmental health impact assessment

Environmental problems such as air pollution or radon can affect human health. Several scientific methods have been developed in order to estimate the extent of such environmental health impacts, or the potential impact of policy and plans on these impacts. These methods include health impact assessment, integrated assessment and risk assessment. When such assessments deal with complex environmental health problems, such as transport or climate change, the effects of different environmental risks often need to be compared. Traditionally, numbers of attributable deaths and disease cases were reported. Such numbers can supply useful information, but cannot always be used to compare dissimilar health effects. For example, effects from air pollution range from aggravation of asthma to premature mortality, while noise exposure is associated with annoyance, sleep disturbance and effects on cognition. Because of the divergence in magnitude, duration and severity of these health effects, summary health measures have been developed. Such measures convert all health effects to a comparable unit, and can thus be very useful for the interpretation and comparison of different (environmental) health problems. This is especially useful for evaluating and comparing different policy options and assessing cost effectiveness of mitigating measures or prevention. Summary health measures in environmental health decision-making include for example Years of Life Lost (YLL), disease burden estimates such as Disability Adjusted Life Years (DALYs) or monetary valuation.

Indicators that can be calculated using ICT

The model output of the ICT model are the following indicators:

Number of attributable deaths

Number of attributable morbidity cases

Age-specific loss of life-expectancy for target population

Average loss of life-expectancy for a birth cohort

Years lost due to mortality (YLL)

Years lost due to disability (YLD)

Loss of disability adjusted life years (DALY)

Life table modelling

Benefits of life table modelling

The advantage of dynamic life table modelling in health impact assessment is that it enables to predict impacts in a real life population over time as the population structure and risk level changes. Life table models take into account effect of competing causes of death. The concept of competing causes of deaths refer to the fact, that changing the risk due to a certain cause of death will affect the mortality from other causes if their risk level stays the same. Thus, a life table model gives the net change in the life years saved or lost over time. By taking competing causes of death into account it is also possible to avoid the risk of over-estimating the overall mortality impact when evaluating impacts from multiple mortality endpoints for a single exposure or combined mortality effects from multiple exposures.

Life table modelling provides most benefits in terms of estimating the net total impacts in a real life population when used in a direct way, i.e. comparing the life tables and life years predicted for different scenarios. However, the indirect use, i.e. when life table modelling is used to determine age-conditional life expectancy in a real life population, which can be further multiplied with attributable deaths to derive YLL, can be more preferable in some situations. This is, for example, in cases where impacts are modelled for a short follow-up period (one or few years), but the aim is to estimate total loss of life years due to the attributable deaths. A simplified solution would be to use age-conditional life expectancy data for the current population. However, if the aim is to model impacts due to an existing risk factor, this approximation would lead to underestimation of YLL because it ignores that in the (theoretical) absence of the risk life expectancy would, in fact, be a fraction higher. In many cases this difference would be negligible, but could in some cases be of importance. This source of bias is avoided when applying life table modelling in the impact assessment, because the model also predicts the impact of the risk factor to the current life-expectancy in the target population.

Model description

Modelling approach 1: Life table method

Short overview

Mortality impacts

Mortality impacts are modelled using the dynamic life table method. First, total mortality risk is determined for the reference and alternative exposure scenarios using the population and baseline mortality data, exposure levels and exposure-response functions. Based on the total mortality risk and population input data, the future population structure is then projected for each scenario using life table methodology. These population projections are applied in determining the age-conditional life expectancies in the different scenarios. YLL (years of life lost due to mortality) can subsequently be calculated directly from the life tables as the difference in the life years lived by the projected populations in different scenarios, or indirectly based on the age-specific attributable deaths and age-specific life expectancies in a given scenario.

Morbidity impacts

First, morbidity risk attributable to the exposure is modelled for each scenario based the population and baseline morbidity data, exposure level and exposure-response functions. Number of attributable morbidity cases is then calculated using the attributable morbidity risk and the modelled population projections for each scenario, and YLD (years of life lost due to disease) subsequently based on the attributable cases, severity weight and duration.

Life tables for different exposure scenarios

Total mortality risk in different exposure scenarios

if Starting point for the assessment: Exposure scenarios

The total mortality risk in different exposure scenarios (and in different follow-up years in case the exposure level in the scenario varies through the assessment follow-up period) is determined based on input data on:

• Exposure level in different scenarios
• Fraction of the exposed population (the rest of the population is assumed to expose to the reference scenario level)
• Exposure-response functions for the mortality endpoints of interest
• Population specific mortality data

For those mortality endpoints for which the exposure-response is defined as a relative risk (RR), the modeling steps are:

1) The model calculates what the current average population mortality risk would be in the absence of exposure to the risk factor of interest (the so-called baseline risk). This is done based on the attributable fraction (AF) method as follows:

BMR = MRcurrent – MRcurrent*AF

• BMR = Mortality risk in the absence of the risk factor
• MR_current = Measured population mortality risk which includes the current exposure to the risk factor
• AF = Fraction of risk attributable to the exposure of interest

AF = (RR’-1) / RR’

• RR’ = Relative risk adjusted to the exposure level

RR’= Exp( ln(RR) / Unit * Exposure )

• RR = Relative risk per unit of exposure
• Unit = Exposure unit
• Exposure = Current exposure level

Average adjusted risk for the total population is calculated based on the fraction of exposed population. The non-exposed population is assumed to expose to the reference scenario level, which usually corresponds to zero or a so-called natural background exposure level.

2) The mortality risk in different exposure scenarios is then determined based on the baseline mortality risk and exposure scenario specific exposure level as follows:

MRscenario + BMR * RR’scenario

• MR_scenario = Mortality risk in a given scenario

RR’scenario = Exp( ln(RR) / Unit * Exposurescenario)

For mortality endpoints for which the exposure-response is defined as an absolute risk (AR):

The model calculates the absolute difference in the average population mortality risk in different scenarios compared to the current risk level based on the exposure difference, fraction of population exposed and exposure-response function as follows:

∆risk = AR / Unit * ∆exposure * Pop_frac

• AR = absolute risk per unit of exposure
• Unit = Exposure unit
• ∆exposure = Absolute difference between the current exposure level (i.e. BAU scenario in the beginning of the assessment) and the exposure level in a given scenario
• Pop_frac = Fraction of exposed population

Changes in the risk for different causes of mortality are then summed up to yield the net total mortality risk in different exposure scenarios.

Population projection

Population structure for the whole population is projected to the future in 5 year age-categories and 5-year-periods, and is based on:

• Population input data
• Birth rate input data
• Modelled total mortality risk for different exposure scenarios.

The population is projected for a 100 year period beginning from the assessment start year. If the assessment follow-up period is shorter than 100 years, the mortality risk for the rest of the 100-year period is assumed to stay on the level it is during the last follow-up year.

The modelling steps are:

1) Population age-structure in the beginning of the assessment is defined by the population input data.

2) The annual survival rate for each age-category is calculated as follows:

SR = (2 – MR) / (2 + MR)

• MR = Annual total mortality risk in a given age-category

3) Survival rate from a given age-category (x) to the next age-category (y) during a 5-year-period is then calculated as follows:

SRp = SRx5 + (SRx4 * SRy) + (SRx3 * SRy2) + (SRx2 * SRy3) + (SRx * SRy4)

• SR_x = Annual survival rate in the younger age-category (x)
• SR_y = Annual survival rate in the older age-category (y)

4) Number of population in each age-category in the beginning of each 5-year-period is then calculated as follows:

Populationage x, period x = Populationage x-1, period x-1 * SRp, age x-1

Caculation of health impact indicators

Age-adjusted life-expectancy

Life-expectancy for a birth cohort

Attributable deaths

Years of life lost due to mortality (YLL)

Attributable disability cases

Years of life lost due to disability (YLD)

Modelling approach 2: Calculation of EBD from total BoD

Step-by-step user instructions

Analytica software and ICT model technicalities

To be able to run the Impact Calculation Tool, you need to install Analytica software to your computer. Analytica installer can be downloaded from lumina.com. The full use of the software requires a license. However, a free Analytica player is also available. The free Analytica player lets you review and run the ICT model without having to purchase a license. With the Player edition, you can change designated inputs, run the model, view results, and examine selected model diagrams and variables. However, it does not let you make changes other than to selected inputs, or save models.

To use Analytica software, you need the following minimum configuration:

• Intel 486-66 MHz or better (Pentium 500 MHz+ or AMD Athlon recommended).
• 30 MB disk space
• 256 MB RAM (2 GB recommended for large models)
• 8-bit color display
• Windows XP, Server 2003, Server 2008, Vista, or Windows 7.

It helps to have a faster CPU, and, especially, more RAM for large models. Analytica will benefit from up to 3 GB RAM.

The Impact calculation Tool -model can be downloaded from this link:

File:Impact Calculation Tool.ana

The latest model version is listed on the very top of the page.

The Impact Calculation tool can be operated simply from the user interface. To find detailed instructions on how to view the model and it's outputs, go to

Help → User guide → Chapter 1: Examining a Model and Chapter 2: Result Tables and Graphs

Running the Impact Calculation Tool

Defining the assessment boundaries

Starting point for the assessment

• Exposure scenarios: health impacts are calculated based on alternative exposure scenarios, background mortality/morbidity data and exposure-response functions.
• Health outcome scenarios: changes in life-expectancy and years of life lost are calculated based on alternative scenarios on attributable deaths and morbidity cases.
• Total burden of disease: environmental burden of disease (EBD) is calculated based on exposure scenarios, total burden of disease (BoD) data and exposure-response functions.

Mortality endpoints

If the starting point for the assessment is 'Exposure scenarios' or 'Health outcome scenarios', then the list of mortality endpoints needs to cover ALL causes of death. However, total mortality should not be divided into more sub-categories than needed to distingish the mortality endpoints of interest.

If the starting point for the assessment is 'Total burden of disease', then the list should cover only the mortality endpoints of interest.

Morbidity endpoints

List morbidity endpoints included in the assessment.

Sex-specific assessment

Assessment start year

If starting point for the assessment is 'Exposure scenarios' or 'Health outcome scenarios'

Follow-up period

If starting point for the assessment is 'Exposure scenarios' or 'Health outcome scenarios'

Number of follow-up years should be 1 or any number divisible by five.

Note! If follow-up period is changed after inputting data, some input tables may have to be updated.

Assessment starting point: Exposure / health outcome scenarios

Input data

Target population data

• Population: annual total population
• Birth rate: annual number of births
• Baseline mortality: annual number of deaths in the start year in bisness-as-usual (BAU) scenario
• Baseline morbidity: annual number of morbidity cases in the start year in bisness-as-usual (BAU) scenario. Note! This is only needed for those morbidity endpoints for which relative risk exposure-response function is used.

If starting point for the assessment is 'Exposure scenarios':

Exposure data

Select whether or not you want to define follow-up year or age-specific exposure levels

Type in or copy-paste exposure data into the correct table.

• BAU: business-as-usual exposure level, i.e. the exposure level related to the baseline mortality and morbidity data
• Alternative: alternative exposure level, can be below or above BAU level
• Reference: reference exposure level, can be set to zero or impact threshold level

Define the fraction of total population that is assumed to be exposed to the defined level. The rest of the population is assumed to be exposed to the reference level. When age and/or year specific exposure levels are defined, the fraction of exposed population is assumed to be the same in all age categories and/or follow-up years.

Exposure-response functions

Select if acute or chronic mortality impacts are assessed. Chronic ERF should reflect changes in annual mortality rates, whereas acute ERF reflect changes in daily mortality rates.

Select whether or not you want to define age-specific exposure-response functions for mortality endpoints.

Type in or copy-paste exposure-response functions into the correct table.

ERFs can be in the form of relative risk (RR) or absolute risk (AR). However, only one ERF should be defined for any given mortality endpoint. RRs are defined on the 'Relative risk' -sheet of the table. If there is no RR, value 1 should be listed.

ARs are defined on the 'Absolute risk' -sheet of the table. If there is no AR, value 0 should be listed.

Note! ERFs should not overlap in regard to the mortality endpoints.

Note! If Age-specific mortality ERF = 'Yes', then all ERFs (including both age-specific and general) should be inputted in the 'Age-specific mortality ERF' -table.

Define the exposure unit for which the ERF is given.

If Type of mortality effect = 'Acute', then define the assumed loss of life for a premature death attributable to the exposure.

If starting point for the assessment is 'Health outcome scenarios':

Health outcome data

Define the fraction of deaths caused by the risk factor of interest in the business-as-usual (BAU) scenario.

Select how the change in mortality is defined

• Percentage = percentage increase/decrease in mortality compared to BAU scenario, e.g.

      5% increase = 5
      5% decrease = -5.

• No. of cases = absolute increase/decrease in mortality compared to BAU scenario, e.g.

      10 more deaths = 10
      10 less deaths = -10

Select whether or not you want to define follow-up year -specific change in mortality.

Type in or copy-paste the data into the correct the table.

Disability data

Duration

Select whether or not you want to define age-specific durations of morbidity.

Note! If Age-specific duration of morbidity = 'Yes' then durations for all morbidity endpoints (whether age-specific or not) should be inserted in the 'Age-specific duration of morbidity' -table.

Severity weight

Select whether or not you want to define age-specific severity weights for morbidity endpoints.

Note! If Age-specific severity weight = 'Yes' then weights for all morbidity endpoints (whether age-specific or not) should be inserted in the 'Age-specific severity weight' -table.

Time discount rate

Time discount rate: define the annual discount rate when calculating future loss of life.

Age weighting

Select whether or not to apply age weights in calculating the loss of life.

• No = No age weighting applied
• Yes = Life lost at a given age is weighed according to the following formula: Cxe^(bx), where x = age in years, C = 0.1658, b = 0.04. This is the formula used in the Global Burden of Disease studies. For further information, see: Murray 1994. Quantifying the burden of disease: the technical basis for disability-adjusted life years. Bulletin of the World Health Organization 72, 429-445.

Health impact outcomes

Specify how outcomes are calculated

Select the outcome of interest:

• EBD = Health impact caused by the given exposure level in different scenarios
• Change in BoD = Increase/decrease in health impact in the alternative scenario in relation to BAU

Select the approach for calculating YLL:

• Life tables: YLL is calculated based on the projected life tables and the difference in years lived in different exposure scenarios. Life years lost due to a given premature mortality case are divided into the follow-up years, i.e. the total life lost due to a given premature mortality case is NOT allocated to the year during which the death occurs, but to all follow-up years during which the life would have taken place.
• Attributable deaths: YLL is calculated based on year and age-specific attributable deaths and age-specific life-expectancies, i.e. the total life lost due to a given premature mortality case is allocated to the year during which the death takes place.

If approach for YLL calculation is 'Attributable deaths', select the life-expectancy to be used in the calculation of YLL.

• Target population = Age-conditional life-expectancy in the target population
• Standard life-expectancy, i.e. the life-expectancy in the japanese population (the theoretical optimum life-expectancy)

Impacts on life-expectancy

• Life expectancy in birth cohort: average life-expectancy in a birth cohort assumed to expose to through lifetime. Note! If year specific exposure levels are defined the number of follow-up years should be 100 or more.
• Loss of life expectancy for birth cohort: average loss of life expectancy compared to the reference exposure level
• Age-specific life expectancy: average age-specific life-expectancy for the follow-up period
• Loss of age-specific life expectancy: average loss of age-specific life expectancy compared to the reference exposure level

Attributable mortality and disability counts

• Attributable mortality: number of deaths attributable to the exposure
• Attributable morbidity: number of morbidity cases attributable to the exposure

Burden of disease measures

• YLL: loss of life years due to premature mortality attributable to the exposure
• YLD: loss of life years due to morbidity attributable to the exposure
  • Note! Total life lost due to a given morbidity case is allocated to the year during which the morbidity case occurs.
• DALY: loss of disability adjusted life years attributable to the exposure

Assessment starting point: Total burden of disease

Input data

Total burden of disease data

BoD data should be divided into years lost due to mortality (YLL) and morbidity (YLD)

Exposure data

Select whether or not you want to define age-specific exposure levels

• BAU: business-as-usual exposure level, i.e. the exposure level related to the total burden of disease data.
• Alternative: alternative exposure level, can be below or above BAU level
• Reference: reference exposure level, can be set to zero or impact threshold level

Define the fraction of total population that is assumed to be exposed to the defined level. The rest of the population is assumed to be exposed to the reference level. When age-specific exposure levels are defined, the fraction of exposed population is assumed to be the same in all age categories.

Exposure-response function

Select whether or not you want to define age-specific exposure-response functions for mortality endpoints.

ERFs should be in the form of relative risk (RR). If there is no RR, value 1 should be listed.

Note! If Age-specific mortality ERF = 'Yes', then all ERFs for all mortality endpoints (including both age-specific and general) should be inputted in the 'Age-specific mortality ERF' -table.

Define the exposure unit for which the ERF is given.

Health impact outcomes

• YLL: loss of life years due to premature mortality attributable to the exposure
• YLD: loss of life years due to morbidity attributable to the exposure
• DALY: loss of disability adjusted life years attributable to the exposure

See also

• File:Impact Calculation Tool.ana