Background
Public health economic modelling
Aim
Methods
PRIMEtime
Disease | Data and methods |
---|---|
Coronary heart disease (CHD) | Incidence of CHD estimated from incidence rates of first acute myocardial infarction (derived from Hospital Episode Statistics [38]), adjusted using the proportion of unstable angina among all coronary events in the OXVASC study [39]. Mortality rates from the Office of National Statistics cause-specific death registrations (number of deaths where myocardial infarction was mentioned on the death certificate) [40]. Case fatality rates and baseline prevalence derived using DISMOD IIa [41]. |
Stroke | Incidence of first stroke estimated from the OXVASC study [39] and data from the General Practice Research Database [42]. Mortality rates from the Office of National Statistics cause-specific death registrations [40]. Case fatality rates and baseline prevalence derived using DISMOD II [41]. |
Type two diabetes | Incidence rates from the UK Clinical Practice Research Datalink [43]. Type two diabetes mortality rate ratios and prevalence estimated from the National Diabetes Audit 2011/12 [44]. Case fatality rates derived using DISMOD II [41]. |
Cirrhosis | Incidence rates from a population-based cohort study linking the Clinical Practice Research Datalink and Hospital Episode Statistics [45]. Mortality rates from the Office of National Statistics cause-specific death registrations [40]. Case fatality rates and baseline prevalence derived using DISMOD II [41]. |
Cancers | Incidence rates from Cancer Registrations Statistics, England, 2012 [46]. Mortality rates from the Office of National Statistics cause-specific death registrations [40]. Case fatality rates and baseline prevalence derived using DISMOD II [41]. |
Risk factors | ||||||
---|---|---|---|---|---|---|
Parameter | Exposure parameters | Outcomes | Unit of change | Distribution of uncertainty range | Relative risk (SE) | Notes |
Fruit | CHD [51] | Per 106 g/day fruit | Lognormal | 0.93 (0.019) | ||
Stroke [52] | Per 106 g/day fruit | Lognormal | 0.89 (0.023) | |||
Lung cancer [53] | Per 100 g/day fruit | Lognormal | 0.94 (0.02) | |||
Vegetables | CHD [51] | Per 106 g/day vegetables | Lognormal | 0.89 (0.034) | ||
Lung cancer [53] | Per 100 g/day vegetables | Lognormal | 0.94 (0.025) | |||
Fibre (cereal only) | Mean (SD) g/day from NDNS [49] | CHD [54] | Per 10 g/day fibre (cereal) | Lognormal | 0.91 (0.02) | Cereals only to avoid double counting for fruit and vegetables |
Fibre | Breast cancer [55] | Per 10 g/day fibre | Lognormal | 0.93 (0.027) | ||
Colorectal cancer [56] | Per 10 g/day fibre | Lognormal | 0.90 (0.034) | |||
Stomach cancer [57] | Per 10 g/day fibre | Lognormal | 0.56 (0.12) | |||
Red meat intake | Colorectal cancer [58] | Per 100 g/day red meat | Lognormal | 1.30 (0.06) | ||
Stomach cancer [59] | Per 100 g/day red meat | Lognormal | 1.13 (0.036) | |||
Type two diabetes [60] | Per 100 g/day red meat | Lognormal | 1.20 (0.072) | |||
Processed meat intake | Colorectal cancer [58] | Per 50 g/day processed meat | Lognormal | 1.38 (0.07) | ||
Type two diabetes [60] | Per 50 g/day processed meat | Lognormal | 1.57 (0.1) | |||
Serum cholesterol | CHD [62] | Per -1 mmol/l total cholesterol | Lognormal | < 49 years: 0.44 (0.034) 50–59 years: 0.58 (0.034) 60–69 years: 0.72 (0.018) 70–79 years: 0.82 (0.015) 80+ years: 0.85 (0.21) | ||
Stroke [62] | Per -1 mmol/l total cholesterol | Lognormal | < 59 years: 0.90 (0.037) 60–69 years: 1.02 (0.027) 70–79 years: 1.04 (0.025) 80+ years: 1.06 (0.031) | |||
Systolic blood pressure | CHD [63] | Per -20 mmHg systolic blood pressure | Lognormal | < 49 years: 0.49 (0.042) 50–59 years: 0.50 (0.015) 60–69 years: 0.54 (0.0094) 70–79 years: 0.60 (0.013) 80+ years: 0.67 (0.023) | ||
Stroke [63] | Per -20 mmHg systolic blood pressure | Lognormal | < 49 years: 0.36 (0.057) 50–59 years: 0.38 (0.034) 60–69 years: 0.43 (0.024) 70–79 years: 0.50 (0.02) 80+ years: 0.67 (0.03) | |||
Body mass index | CHD [64] | Per 5 kg/m2 BMI | Lognormal | 35–59 years: 1.50 (0.039) 60–69 years: 1.40 (0.031) 70–79 years: 1.31 (0.033) 80–89 years: 1.30 (0.055) | ||
Stroke [64] | Per 5 kg/m2 BMI | Lognormal | 35–59 years: 1.76 (0.075) 60–69 years: 1.49 (0.056) 70–79 years: 1.33 (0.056) 80–89 years: 1.10 (0.083) | |||
Diabetes [64] | Per 5 kg/m2 BMI | Lognormal | BMI 15–25: 0.96 (0.25) BMI 25–50: 2.16 (0.067) | |||
Pancreas cancer [65] | Per 5 kg/m2 BMI | Lognormal | 1.10 (0.016) | |||
Colorectal cancer [64] | Per 5 kg/m2 BMI | Lognormal | Men: 1.24 (0.016) Women: 1.09 (0.019) | |||
Breast cancer [64] | Per 5 kg/m2 BMI | Lognormal | Women 60+ years: 1.12 (0.018) | |||
Kidney cancer [64] | Per 5 kg/m2 BMI | Lognormal | Men: 1.24 (0.039) Women: 1.34 (0.034) | |||
Liver cancer [64] | Per 5 kg/m2 BMI | Lognormal | 1.47 (0.078) | |||
Liver cirrhosis [64] | Per 5 kg/m2 BMI | Lognormal | BMI 15–25: 0.73 (0.016) BMI 25–50: 1.79 (0.077) | |||
Diabetes | Prevalence | CHD [66] | If have diabetes compared to no diabetes | Lognormal | Men: 1.85 (0.063) Women: 2.63 (0.076) | |
Stroke [67] | If have diabetes compared to no diabetes | Lognormal | Men: 1.83 (0.067) Women: 2.28 (0.085) | |||
Physical activity | Mean (SD) MET hours per week from Active People’s Survey [68] | CHD [69] | 11.25METhr/wk. increase | Normal | −0.204 (0.027) | Parameter is the beta value of the 0.25 transformation of the relative risk, unadjusted for BMI. |
Stroke [69] | 11.25METhr/wk. increase | Normal | −0.195 (0.041) | Parameter is the beta value of the 0.25 transformation of the relative risk, unadjusted for BMI. | ||
Diabetes [69] | 11.25METhr/wk. increase | Normal | −0.240 (0.023) | Parameter is the beta value of the 0.25 transformation of the relative risk, unadjusted for BMI. | ||
Colorectal cancer [70] | 11.25METhr/wk. increase | Normal | −0.080 (0.059) | Parameter is the beta value of the 0.25 transformation of the relative risk, unadjusted for BMI. | ||
Breast cancer [70] | 11.25METhr/wk. increase | Normal | −0.053 (0.023) | Parameter is the beta value of the 0.25 transformation of the relative risk, unadjusted for BMI. |
Risk factors operating through intermediate variables | ||||||
---|---|---|---|---|---|---|
Parameter | Exposure parameters | Outcomes | Unit of change | Distribution of uncertainty range | Change in value of outcome (SE) | Notes |
Total fat | % of total energy [49] | Total serum cholesterol (mmol/l) [71] | Per 1% energy total fat | Normal | 0.020 (0.005) | |
Saturated fat | % of total energy [49] | Total serum cholesterol (mmol/l) [71] | Per 1% energy saturated fat | Normal | 0.052 (0.003) | |
Monounsaturated fatty acids | % of total energy [49] | Total serum cholesterol (mmol/l) [71] | Per 1% energy MUFA | Normal | 0.005 (0.003) | |
Polyunsaturated fatty acids | % of total energy [49] | Total serum cholesterol (mmol/l) [71] | Per 1% energy PUFA | Normal | −0.026 (0.004) | |
Dietary cholesterol | mg/day [49] | Total serum cholesterol (mmol/l) [71] | Per 1 g/day dietary cholesterol | Normal | 0.0007 (0.0001) | |
Salt consumption | g/day [49] | Systolic blood pressure (mmHg) [72] | Per 100 mmol/24 h urinary sodium | Normal | 5.80 (1.71) | Grams of salt consumed per day converted into urinary sodium excretion. |
Total energy | kJ/day [49] | BMI (kg/m2) [73] | Details of equations describing the relationship between energy intake and body weight for men and women can be found in Christiansen and Garby [74]. |
Risk factors’ theoretical minimum risk | ||||||
---|---|---|---|---|---|---|
Parameter | Exposure parameters | Outcomes | Unit of change | Distribution of uncertainty range | Minimum risk value (SE) | Notes |
Body mass index (kg/m2) [61] | – | – | – | Normal | 21 (1) | |
Systolic blood pressure (mmHg) [61] | – | – | – | Normal | 115 (6) | |
Total cholesterol (mmol/L) [61] | – | – | – | Normal | 3.8 (0.6) | |
Vegetable intake (g/day) [50] | – | – | – | Normal | 400 (30) | |
Fruit intake (g/day) [50] | – | – | – | Normal | 300 (30) | |
Fibre intake (g/day) [50] | – | – | – | Normal | 30 (3) | |
Red meat intake (g/day) [50] | – | – | – | Normal | 14.3 (1.43) | |
Processed meat intake (g/day) [50] | – | – | – | Normal | 0 | |
Physical activity (METhrs/wk) [74] | – | – | – | Normal | 133 (13.3) |
Mediation factors | ||||||
---|---|---|---|---|---|---|
Parameter | Exposure parameters | Outcomes | Unit of change | Distribution of uncertainty range | Change in value of outcome (SE) | Notes |
Ischaemic stroke mediation | BMI (kg/m2) | Systolic blood pressure (mmHg) [74] | Normal | 0.65 (0.04) | ||
Fruit intake (g/day) | Systolic blood pressure (mmHg) [74] | Normal | 0.42 (0.17 | |||
Vegetable intake (g/day) | Systolic blood pressure (mmHg) [74] | Normal | 0.54 (0.2) | |||
Fruit intake (g/day) | Total cholesterol (mmol/L) [74] | Normal | 0.027 (0.017) | |||
Vegetable intake (g/day) | Total cholesterol (mmol/L) [74] | Normal | 0.047 (0.026) | |||
Ischaemic heart disease mediation | BMI (kg/m2) | Systolic blood pressure (mmHg) [74] | Normal | 0.31 (0.016) | ||
Fruit intake (g/day) | Systolic blood pressure (mmHg) [74] | Normal | 0.39 (0.15) | |||
Vegetable intake (g/day) | Systolic blood pressure (mmHg) [74] | Normal | 0.47 (0.21) | |||
Fruit intake (g/day) | Total cholesterol (mmol/L) [74] | Normal | 0.008 (0.0057) | |||
Vegetable intake (g/day) | Total cholesterol (mmol/L) [74] | Normal | 0.012 (0.01) |
Parameter | Data and methods |
---|---|
English population | From Office for National Statistics census data, no uncertainty estimated [75]. |
Mortality rates | Extracted from the Human Mortality Database, no uncertainty estimated [76]. |
Health sector costs | Disease specific costs derived from NHS England programme budgeting data [77] and unrelated disease costs estimated using NHS England cost curves [78]. Using the same approach as Blakely et al., [34] health sector costs are assumed to be “moderately uncertain”, and therefore uncertainty is estimated using a generic multiplication factor across all health sector costs with a gamma distribution based on a normal distribution (mean 1, SD 0.1). |
Societal costs | Disease specific and unrelated productivity, social care, and wider societal costs estimated using a Department of Health tool published as a supplementary file in Claxton et al [79]. As with health sector costs, uncertainty estimated using a generic multiplication factor across all societal costs with a gamma distribution based on a normal distribution (mean 1, SD 0.1). |
Utilities | Baseline mean EQ-5D utility scores and disease specific decrements and their standard errors taken from Sullivan et al., with adjustments made for age and number of chronic conditions [80]. |
PRIMEtime CE
Stakeholder category | Stakeholder |
---|---|
(i) National governmental organisations | National governmenta,b |
Department of Healtha,b | |
National Institute for Health and Care Excellencea,b | |
Public Health Englandb | |
Physical activity and diet responsibility dealb | |
(ii) Local governmental organisations | Local Government Associationb |
Thames Valley Public Health England Centreb | |
Oxfordshire County Council Public Health Departmentb | |
(iii) Charitable organisations | The Wellcome Trust (project funder) |
UK Health Foruma,b | |
British Heart Foundationa,b | |
Food Ethics Councila,b | |
Consensus Action on Salt and Hypertension / Action on Sugara,b | |
World Obesity Federationa,b | |
Diabetes UKa,b | |
Sustaina,b | |
Blood Pressure UKa,b | |
(iv) Health professional and academic organisations | Association of Directors of Public Healthb |
Academy of Medical Royal Collegesa,b | |
Faculty of Public Healtha,b | |
International Society for Physical Activity and Healthb | |
International Society of Behavioural Nutrition and Physical Activityb | |
The Nutrition Societya,b | |
Association for the Study of Obesitya,b | |
(v) Patients and the public | Members of the publica |
Patients with chronic diseasea |
Adding physical activity as a risk factor
Estimating healthcare and social care costs
Modelled disease | Programme budgeting category | Programme budgeting expenditure | Specialised services expenditure | Primary care expenditure | Total NHS England disease costs | Annual cost per prevalent case (£) |
---|---|---|---|---|---|---|
Step 1 | Step 2 | Step 3 | ||||
Ischaemic heart disease | 10a Coronary Heart Disease | 953,743 | 41,818 | 485,056 | 1,480,617 | 1905 |
Stroke | 10b Cerebrovascular disease | 689,876 | 55,443 | 29,475 | 774,794 | 843 |
Type two diabetes | 04a Diabetes | 1,071,537 | 25,577 | 959,716 | 2,056,831 | 444 |
Breast cancer | 02f Cancer, breast | 472,192a | N/A | N/A | 472,192 | 573 |
Colon cancer | 02c Cancer, lower GI | 248,315 | 84,919 | 20 | 333,253 | 810 |
Lung cancer | 02d Cancer, lung | 98,250 | 33,599 | 0b | 131,849 | 904 |
Stomach cancer | 02b Cancer, upper GI | 32,794 | 11,215 | 0b | 44,008 | 535 |
Liver cancer | 02b Cancer, upper GI | 16,990 | 5810 | 0b | 22,801 | 1532 |
Kidney cancer | 02 h Cancer, urological | 25,145 | 8599 | 7833 | 41,577 | 618 |
Pancreatic cancer | 02b Cancer, upper GI | 42,133 | 14,409 | 0b | 56,542 | 3074 |
Liver disease | 13c Hepatobiliary | 59,702 | 4543 | 2963 | 67,209 | 314 |
Estimating utilities
Age | Male | Female | Age | Male | Female |
---|---|---|---|---|---|
0 | 1.000 | 1.000 | 51 | 0.800 | 0.799 |
1 | 1.000 | 1.000 | 52 | 0.800 | 0.799 |
2 | 1.000 | 1.000 | 53 | 0.799 | 0.798 |
3 | 1.000 | 1.000 | 54 | 0.799 | 0.798 |
4 | 1.000 | 1.000 | 55 | 0.799 | 0.798 |
5 | 1.000 | 1.000 | 56 | 0.799 | 0.798 |
6 | 1.000 | 1.000 | 57 | 0.798 | 0.797 |
7 | 1.000 | 1.000 | 58 | 0.798 | 0.797 |
8 | 1.000 | 1.000 | 59 | 0.798 | 0.797 |
9 | 1.000 | 1.000 | 60 | 0.776 | 0.775 |
10 | 0.916 | 0.916 | 61 | 0.776 | 0.775 |
11 | 0.916 | 0.916 | 62 | 0.776 | 0.775 |
12 | 0.916 | 0.916 | 63 | 0.775 | 0.774 |
13 | 0.916 | 0.916 | 64 | 0.775 | 0.774 |
14 | 0.915 | 0.915 | 65 | 0.775 | 0.774 |
15 | 0.915 | 0.915 | 66 | 0.774 | 0.773 |
16 | 0.915 | 0.915 | 67 | 0.774 | 0.773 |
17 | 0.914 | 0.914 | 68 | 0.774 | 0.773 |
18 | 0.914 | 0.914 | 69 | 0.774 | 0.773 |
19 | 0.914 | 0.914 | 70 | 0.725 | 0.724 |
20 | 0.907 | 0.906 | 71 | 0.725 | 0.724 |
21 | 0.907 | 0.906 | 72 | 0.725 | 0.724 |
22 | 0.907 | 0.906 | 73 | 0.724 | 0.723 |
23 | 0.906 | 0.905 | 74 | 0.724 | 0.723 |
24 | 0.906 | 0.905 | 75 | 0.724 | 0.723 |
25 | 0.906 | 0.905 | 76 | 0.724 | 0.723 |
26 | 0.906 | 0.905 | 77 | 0.723 | 0.722 |
27 | 0.905 | 0.904 | 78 | 0.723 | 0.722 |
28 | 0.905 | 0.904 | 79 | 0.723 | 0.722 |
29 | 0.905 | 0.904 | 80 | 0.659 | 0.658 |
30 | 0.881 | 0.880 | 81 | 0.659 | 0.658 |
31 | 0.881 | 0.880 | 82 | 0.658 | 0.657 |
32 | 0.881 | 0.880 | 83 | 0.658 | 0.657 |
33 | 0.880 | 0.879 | 84 | 0.658 | 0.657 |
34 | 0.880 | 0.879 | 85 | 0.658 | 0.657 |
35 | 0.880 | 0.879 | 86 | 0.657 | 0.656 |
36 | 0.880 | 0.879 | 87 | 0.657 | 0.656 |
37 | 0.879 | 0.878 | 88 | 0.657 | 0.656 |
38 | 0.879 | 0.878 | 89 | 0.656 | 0.655 |
39 | 0.879 | 0.878 | 90 | 0.656 | 0.655 |
40 | 0.839 | 0.838 | 91 | 0.656 | 0.655 |
41 | 0.839 | 0.838 | 92 | 0.656 | 0.655 |
42 | 0.839 | 0.838 | 93 | 0.655 | 0.654 |
43 | 0.838 | 0.837 | 94 | 0.655 | 0.654 |
44 | 0.838 | 0.837 | 95 | 0.655 | 0.654 |
45 | 0.838 | 0.837 | 96 | 0.655 | 0.654 |
46 | 0.838 | 0.837 | 97 | 0.654 | 0.653 |
47 | 0.837 | 0.836 | 98 | 0.654 | 0.653 |
48 | 0.837 | 0.836 | 99 | 0.654 | 0.653 |
49 | 0.837 | 0.836 | 100 | 0.653 | 0.652 |
50 | 0.800 | 0.799 |
PRIMEtime CE disease outcome | PRIMEtime CE ICD-10 codes | Equivalent ICD-9 codes | Available utility decrement from Sullivan et al. [80]a | Utility value used in PRIMEtime CE (SE) |
---|---|---|---|---|
Ischaemic heart disease | I20-I25 | 410–414 | 410: −0.063 411: −0.087 412: −0.037 413: −0.085 414: −0.063 | Incident case: −0.071 (0.024) Prevalent: −0.070 (0.015) |
Stroke | I60-I69 | 430–438 | 433: −0.035 435: − 0.033 436: − 0.117 437: − 0.031 438: − 0.073 | Incident: − 0.094 (0.019) Prevalent: − 0.046 (0.031) |
Type two diabetes | E11, E14 | 250.×0 | 250: − 0.071 | − 0.071 (0.005) |
Breast cancer | C50 | 174, 175 | 174: − 0.019 | − 0.019 (0.014) |
Colon cancer | C18-C20 | 153, 154.0, 154.1 | 153: − 0.067 | − 0.067 (0.017) |
Lung cancer | C34 | 162.2–162.9 | 162: − 0.119 | −0.119 (0.043) |
Stomach cancer | C16 | 151 | 151: −0.071 | −0.071 (0.105) |
Liver cancer | C22 | 155 | 155: −0.093 | −0.093 (0.044) |
Kidney cancer | C64 | 189.0 | 189: −0.048 | −0.048 (0.041) |
Pancreatic cancer | C25 | 157 | 195: −0.086 | −0.086 (0.027) |
Liver disease | K70, K74 | 571.0–571.3, 571.5, 571.6, 571.9 | 571: −0.083 | −0.083 (0.031) |
Results
Estimating the effects and costs of the interventions and model validation
Variable | Value or setting |
---|---|
Annual discount rate for health outcomes | 1.5% |
Annual discount rate for costs | 1.5% |
Economic perspective | NHS England and social care costs for both modelled and unrelated diseases |
Intervention costs | Costs to government and to industry where appropriate |
Time horizon | 10 years |
Sensitivity analysis | Explanation of what is changed compared to the primary analysis |
---|---|
Changing the time horizon. | Time horizon changing from 10 years in the primary analysis to 1 year, 5 years, 20 years, and 100 years (lifetime of the cohort). |
Analysing results from an NHS perspective. | Estimating cost effectiveness using the change in NHS costs and intervention costs only (without any societal costs). |
Analysing results from a social care perspective. | Estimating cost effectiveness using the change in social care and intervention costs only (without any NHS costs). |
Including social care costs and productivity. | Adding an economic estimate of changes to productivity arising from the intervention. |
Including all wider societal costs. | Including an economic estimate of the intervention on all wider societal costs (including productivity and social care costs). |
Using a discount rate of 3.5%. | Changing the discount rate for costs and outcomes from 1.5 to 3.5%. |
No disease costs estimated for diseases not explicitly modelled by PRIMEtime CE (unrelated disease costs). | Removing from the model any NHS and social care costs estimated to accrue due to diseases that are not explicitly modelled by PRIMEtime CE. |
No cancer included in the model | Cancer removed from the model so that only IHD, stroke, type two diabetes, and liver cirrhosis are included. |
Only including diseases directly related to the risk factor affected. | For a diet intervention, only IHD and stroke are included in the model, and for a physical activity intervention, IHD, stroke, type two diabetes, breast cancer, and colorectal cancer are included. |