- Acute Coronary Syndromes
- Refer to a subset of CHD patients with acute myocardial infarction or unstable angina. The common factor is that these patients are usually hospitalized and are eligible for complex, in-hospital interventions.
- Acute myocardial infarction
- The sudden occlusion of one of the arteries feeding the heart muscle, leading ultimately to heart muscle loss and other complications. Usually treated in hospital with complex treatments. Most treatments are aimed to limit the amount of muscle loss, to decrease the immediate risk of death and to prevent recurrence of the myocardial infarction. High case fatality rate, and usually the timescale for the relevant outcomes are measured in days rather than months or year.
- Adherence (proposed name for this functionality is UPTAKE)
- This term refers to a key component of the right side. We modelled the effects of the interventions by a reduction in a relevant hazard using a relative risk reduction associated to an intervention. However, the effect of a drug only is possible if: the drug is prescribed, the drug is available, and the patient adheres to the treatment. Collectively, we refer to these conditional as “uptake” and we proposed to use this term to refer to this functionality, since adherence is only a component of the idea.
- A cardiovascular risk function, used to create a number of “incident” cases to be fed into the Disease module.
- Case Fatality Rate
- An epidemiological term that refers to the risk of dying given the current state and individual is in.
- Chronic Angina
- A condition characterized by chest pain, usually triggered by exertion. These patients are at risk of progressing their coronary artery disease causing an increase in symptoms and increasing their risk of experiencing a heart attack (AMI). Treatment is aimed primarily to improve quality of life and to decrease mortality risk and disease progression.
- Closed Cohort
- A term used in the prototype application to refer to the simulation of a specific population defined with the population editor, with the purpose to follow up this cohort for a period of time. Currently, it follows every individual until he or she reaches any death state.
Please note that the epidemiological definition of this is different.
- In epidemiology, a group of individuals with common characteristics that is follow up in time until censoring (end of the follow up or lost to follow up) or the occurrence of the outcome.
- COMBINED RELATIVE RISK REDUCTION
- A variable that takes into account each individual INTERVENTION RELATIVE RISK REDUCTION that are part of an INTERVENTION SET , alongside its AVAILABILITY and patient UPTAKE relevant to that particular INTERVENTION
- COMPREHENSIVE MODEL
- The comprehensive model is both the incidence and disease module. When we refer to IMPACT2 we refer to the COMPREHENSIVE MODEL.
Why are some terms in lower case, some in CAPITALS?
- A disease of sugar metabolism, with several serious manifestations affecting vision, the cardiovascular system and the kidneys. In CVD is particularly important as a risk factor for CVD (both CHD and stroke) and might affect also case fatality in established CHD.
- Discrete event simulation
- The method used by IMPACT2 to simulate the movement of an individual between disease states. When an event is due, the software draw randomly a value from the densities associated to the edges connected the starting node with nodes down the pathway (this value is a time to transition). The smallest of these tim
- DISEASE CAUSATION MODEL
- The model that represents the causation of the disease of interest, and feed with incident events the DISEASE TREATMENT COMPONENT.
Currently it is implemented in a very limited demonstration purposed only way using the ASSIGN approach
- DISEASE MODEL DATASET
- Disease treatment component
- AKA “the right side”. It refers to all the functionality needed to model the epidemiology of the relevant disease states and the effect of evidence based treatments on relevant outcomes, like risk of death or quality of life.
A description of the RIGHT HAND SIDE MODEL:
We have a MODEL that represents the possible histories that an individual with CHD can take. The MODEL is a GRAPH with NODES representing the possible disease states and EDGES defining the different pathways. A set of PROBABILITY DISTRIBUTIONS are used by the DISCRETE EVENT SIMULATION to move individuals through NODES through the allowed pathways defined by the EDGES. The MODEL also includes all the workings to simulate the INTERVENTIONS EFFECTS on the PROBABILITY DISTRIBUTIONS that drives movements across NODES through allowed EDGES. The INTERVENTIONS EFFECTS are the COMBINED RELATIVE RISK REDUCTION of an INTERVENTION SET. An INTERVENTION SET is a set of INTERVENTIONS that can modify a specific PROBABILITY DISTRIBUTION associated with an EDGE between two NODES. It has a COMBINED RELATIVE RISK REDUCTION that takes into account each individual INTERVENTION RELATIVE RISK REDUCTION and the UPTAKE and patient ADHERENCE relevant to the INTERVENTION.
Then we can have different “localizations” for the MODEL: For example, The ENGLAND AND WALES RIGHT HAND SIDE MODEL will be the MODEL with a MODEL DATASET relevant to the E&W population. The MODEL DATASET include: POPULATION STRUCTURE by age & sex, available INTERVENTIONS, RELATIVE RISK REDUCTIONS, UPTAKE and ADHERENCES, specific to the E&W population.
What makes a model population specific is the MODEL DATASET, not the MODEL.
Most users might want to:
Use a different MODEL DATASET (for example, instead of the E&W MODEL DATASET, he or she wants to use the LIVERPOOL PCT MODEL DATASET)
Tweak some of the variables in the MODEL DATASET (for example, only considered one intervention, and modify the uptake and adherence)
We want to modify both the MODEL and the MODEL DATASET:
For a closed cohort simulation we want to prepare a MODEL DATASET based on an existing cohort, for validation purposes.
We want to modify the MODEL, by for example, exploring a better set of PROBABILITY DISTRIBUTIONS or by adding or deleting some EDGES and NODES.
But the description of the epidemiology of the disease is not population specific; we can say that is “universal”.
- Downstream risk factors
- Risk factors more proximally related to risk, also know as “biological risk factor”. There are major (see risk factor entry) and minor (a host of biological measures, some causally linked to CVD)
- A part of a graph. In IMPACT2, an edge is a transition allowed between states, to represent a disease pathway. You can add or delete edges in a graph using the graph editor.
Every edge has an associated “density” that is used to estimate the transition time for an individual from a node to another in the discrete event simulation
- A set of NODES and EDGES representing disease pathways
- HEALTHY POPULATION MODEL DATASET
- The model dataset for the right hand side module.
- Heart Failure
- A condition where the heart cannot accomplish its function as a pump efficiently. Usually the result of loss of heart muscle due to various causes. We assume that in 50% of the cases, the cause is CHD, and only these cases are taken into account in our simulation.
It is a disease state with a high case fatality and has two subgroups: Heart Failure in the community (those patients who has never been admitted for this into a hospital) and Heart Failure hospitalized. We use this classification as a rough measure of severity of the condition (each subgroup has a different case fatality). The timescales for outcomes is measured here in years (usually 5 years)
- High risk individuals or targeted interventions
- A primary prevention intervention that targets a specific subset of the population. The criteria could be any attribute (age, gender, risk factor profile, and usually a criteria is 10 year CVD risk as estimated with a risk function like ASSIGN
- IMPACT2 Disease
- The name of a disease policy model. For example, the current development is IMPACT2CHD. If we modify the DISEASE MODEL adding for example a state for STROKE, we can call this new “model”, IMPACT2CVD. In time, this nomenclature will highlight the generic nature of the modelling environment for non-communicable disease.
- The MODEL DATASET for the left hand side module.
- An epidemiological term that describes the occurrence of a new event in the simulation. It is generally expressed as number of “new cases” /population at risk taking into account time. It can be expressed as a density (number of new cases/person-time) or in cumulative form (number of new cases/population in a given period of time).
It could be a tricky concept, since it is not always easy to define what a new case is. For the simulation, a practical definition of new case could be “the first visit to a given state”.
For example, a healthy individual that develops is or her first Acute myocardial infarction (AMI) is an “incident AMI”. If this patient form the AMI state enters the heart failure state for the first time, is considered an “incident heart failure patient”. However, a patient that re-enters the AMI state after an AMI, is not an incident AMI.
It also been used to define the event of an individual developing any of the CVD starting states coming from the left side.
- Drug, procedure or policy decision that might have an effect on incidence (left side) or hazards (right side)
- INTERVENTION SET
- A group of interventions that and user can decide to work with.
- INTERVENTIONS EFFECTS
- is a set of INTERVENTIONS that can modify a specific PROBABILITY DISTRIBUTION associated with an EDGE between two NODES. These are the COMBINED RELATIVE RISK REDUCTION of an INTERVENTION SET
Life expectancy is a statistical measure of the average life span (average length of survival) of a specified population. It most often refers to the expected age to be reached before death for a given human population (by nation, by year of birth, or by other demographic variables).
It is usually calculated using a life table
A life table (also called a mortality table or actuarial table) is a table which shows, for a person at each age, what the probability is that they die before their next birthday. From this starting point, a number of statistics can be derived and thus also included in the table:
the probability of surviving any particular year of age remaining life expectancy for people at different ages the proportion of the original birth cohort still alive estimates of a cohort's longevity characteristics.
- Specific data to make the MODEL relevant to a specific population, by modifying the MODEL DATASET
- The possible histories that an individual with CHD can take. The MODEL is a GRAPH with NODES representing the possible disease states and EDGES defining the different pathways. A set of PROBABILITY DISTRIBUTIONS are used by the DISCRETE EVENT SIMULATION to move individuals through NODES through the allowed pathways defined by the EDGES
- MODEL DATASET
- Variables that describe the particulars of a specific population. It includes data on the population structure by age and gender, available interventions, relative risk reductions, uptake and adherences relevant for a particular population in the case of the right hand side module, and population structure and risk factors distribution for the left hand side module.
Most of the time however, the user will use default datasets provided by us, and he or she can tweak these defaults.
- A part of a graph. In IMPACT2, a node is a disease state. You can add or delete nodes in a graph using the graph editor.
- A disease characterized by increase in body mass, distribution of fat. It is associated to increase CVD risk and cancer risk. It is usually defined using a measurement called Body Mass Index (weight/(height squared))
- PCT (Primary Care Trust)
- An administrative division of the NHS, responsible of delivering health promotion, preventative, curative and rehabilitative services to a defined population, usually smaller than a region. For example, the Liverpool PCT provides services to the people living in the Liverpool City Council Area.
- Population Disease Causation Component
- AKA LEFT HAND SIDE MODULE (alternatives: Incidence module, Incidence generator, Primary prevention model, ?)This module provides “incident” CVD cases from healthy people in the population at different levels of risk. AKA as the “left side”
- Population editor
- A functionality in the prototype that allow us to input the age and gender of a population to use in the closed cohort simulation
- Population level intervention
- A primary prevention intervention (acting on the left side) that has effects on the entire population. Usually this type of interventions shift the distribution of a risk factor towards more healthy levels (eg: shifting to lower values for blood pressure by a small amount).
- POPULATION STRUCTURE
- Usually refers to the age and gender distribution of a population being simulated. In the COMPREHENSIVE MODEL, is the population structure of the healthy population, since the population structure for the right hand side module is composed of incident events feeding the starting states generated by the PRIMARY PREVENTION MODULE.
In a CLOSED COHORT SIMULATION, is a user entered population structure.
- The proportion of individuals with an attribute existing in a population/population at risk for the event.
In more practical terms, prevalent cases in a period of time include those cases that are “new” in the period (incident cases) and cases that already had the event of interest before the period of interest.
- Primary Prevention
- Every intervention aimed to reduce CVD incidence (the first visit to any CVD state from the left side)
- Quality of Life
- A construct that captures the impact of disease on several aspects of well-being. In modelling, usually this construct is represented by a quality of life weight associated to a given disease state.
- Quality of Life adjusted years (QALYs)
- A mortality based measure combining length of life and quality of life. Essentially, a year spent in a disease state is less valuable for an individual than a year spent in complete health. It is the result of weighting the duration in a given state by the appropriate quality of life weight
- Relative Risk
- The ratio between the incidence of the outcome in the treated group over the incidence of the outcome in the comparison group.
- Relative Risk Reduction
- 1-Relative Risk
- Risk Factor
- An attribute of an individual that is associated to an increased risk of an event. In CVD epidemiology, the major risk factors for a CVD event (it may be sudden death, a heart attack, a stroke, chronic angina, etc) are: blood pressure, blood lipids (fats), smoking, and diabetes. We call them major risk factors because the causal link is proven, they are frequent in most population, together they explain almost 90% of the risk and they are modifiable. Diet and lifestyle also are risk factors, acting more upstream in the causation chain
- RISK FUNCTION BASED INCIDENCE GENERATOR
- A simple implementation of the left hand side module, to generate incident events to feed the STARTING STATES in the right hand side module.
Currently, the risk function used is ASSIGN.
The plan is to develop a Bayesian Network approach to generate incident events.
- Smoking has a proven causal relationship with mortality, and also has a dose gradient (the more you smoke, the higher the risk) although usually is treated as an ordinal variable (no smoking, ex smoker, current smoker) or by a variable combining
- A class of blood lipids lowering drugs.
- Systolic blood pressure (SBP)
- One of the variables defining blood pressure. SBP has a strong linear relationship with CVD mortality, causally proven.
- Take up (term needs a new name)
- In the prototype application, this term refers to the proportion of the healthy population in the “left side” that could experience the effect of the primary prevention intervention(s) selected
- Total Cholesterol
- One of the measurements used to asses blood lipids. It has a strong linear relationship with CVD mortality, causally proven. There are other subfractions of blood lipids that can be used to quantify the association between lipids and mortality (LDL, HDL, ratios)
- These terms is used in the current project to refer to any intervention that can be used to treat established disease (any of the states in the right side).
- Unstable Angina
- Refers to an episode of chest pain that prompt the admission of the patient to the hospital with the aim of preventing the development of an acute myocardial infarction and to assess the patient risk of developing a heart attack in the future and to decrease mortality risk.
- Upstream risk factors
- Risk factors that act more distally in the causal chain, usually by acting on the downstream factors, although they have direct effects on risk as well. Their role in causation is accepted, although the evidence is not so strong as per the downstream risk factors. Examples are socio-economics determinants, diet, physical activity, obesity.
- Expert: an individual that can modify any part of the policy model. Usually a member of the IMPACT development team
Advanced: user that might access more functionality like running closed cohort simulations, adding new interventions, making more sophisticated comparisons or downloading simulation outputs for further analysis
Intermediate: User that will manipulate localization variables .
Basic: user that will be happy to use model defaults and pre specified simulations.