Table 1 Summary of the six principal CHD policy models

CHD Policy Model (Weinstein and Goldman)

State transition Markov Model

USA, Men and Women aged 35–84

Smoking, total cholesterol, DBP and weight to estimate CHD risk using Framingham Equations

Angina, AMI, sudden death, post MI, CABG, PTCA

Specific treatments considered in different studies eg statins, aspirin, beta-blockers etc

Number of deaths prevented, LYG, CHD incidence (number of arrests, angina, AMI), CHD prevalence, CHD mortality, cost per life year

In the initial model none. Subsequently papers reported one way sensitivity analysis

Model was calibrated using 1986 mortality data.

Validity: Model estimates were compared with 1990 observed-92–98% fit reported.

First policy model rather basic.

Steadily refined since then.

Many papers in high impact journals

PREVENT (Gunning-Scheppers)

Cell based

Netherlands; Denmark, England

Depending on the purpose aged <65

Smoking, cholesterol, hypertension, obesity, physical activity, alcohol

None

Number of deaths prevented, life years gained

One way, different scenarios

Not checked

Mainly a primary prevention model. Developed and adopted in several different populations.

CHD Life Expectancy Model (Grover et al)

Life table analysis-Markov model from 1998 onwards

Canada, Adult men and women, age group not clear

Smoking, total cholesterol, DBP, glucose intolerance, age

Did not consider CHD disease categories but treatments can be considered for primary prevention

Years of life saved, cost per life year saved, years of life without CHD symptoms

One-way

Calibrated

This model uses hypothetical cohorts of participants. In most of the papers, time and the specific population are not clear.

CHD Policy Analysis (Sanderson and Davies)

Micro simulation

England and Wales,

Up to 85 years. Men and women

Smoking, cholesterol, systolic blood pressure

Angina (stable and unstable), AMI, postMI, CABG, PTCA None

Deaths prevented, morbidity prevented, CHD & non-cardiac deaths, unstable angina admissions, investigations, angiograms, PTCA, CABG

No validation reported

Separate risk factor and treatment components. Future model may include secondary prevention treatments. No sensitivity analyses yet. Model fit appears better for men than women.

IMPACT (Capewell, Critchley and Unal)

Spread-sheet

Scotland, England & Wales, New Zealand.

Initially men and women aged 45–84. IMPACT Model for England and Wales includes 25–84

Initially smoking, cholesterol, blood pressure – then also obesity, diabetes and physical activity and deprivation

This model is comprehensive and considers all principal CHD categories and over 20 specific CHD treatments

Deaths prevented or postponed, life years gained.

Multi way sensitivity analysis using Analysis of extremes method.

Estimated falls in CHD mortality were compared with observed falls over specific time period stratified by age and sex.

Considers all major effective treatments available for CHD and all major risk factors.

Data quality adequate, used trial and meta-analyses: National population statistics and results from representative studies

Global Burden of Disease (Murray and Lopez)

Population attributable risk method

World divided into eight geographic regions

M-F all ages

Malnutrition, poor water, unsafe sex, alcohol, tobacco occupation, hypertension, physical activity, illicit drugs, and air pollution

None

Disability adjusted life years (DALYs)

Multi-way sensitivity analysis-discounting and age weighting

None

A comprehensive and global model for WHO strategies. Well documented and described. CHD is included, and modelled as caused by tobacco use, hypertension and physical inactivity, and reduced by alcohol.

Data quality: Extremely variable depending on the region