1. This document summarizes the methods and key findings of the Global Burden of Diseases, Injuries, and Risk Factors Study 2010. It describes how risk factors were selected, exposures estimated, effect sizes determined, and attributable burden calculated.
2. The top 25 risk factors accounted for over 50% of global disease burden in 2010. Leading risks included high blood pressure, tobacco smoking, dietary risks, high BMI, and air pollution. There was significant regional and national variation in risk factors.
3. Key findings showed a shift from communicable to noncommunicable disease risks. Understanding diet and developing interventions for high BMI and glucose were highlighted as research priorities given the growing burden.
Bhawanipatna Call Girls 📞9332606886 Call Girls in Bhawanipatna Escorts servic...
Comparative risk assessment
1. Comparative risk assessment
June 18, 2013
Stephen Lim
Associate Professor of Global Health
Global Burden of Diseases, Injuries, and Risk Factors
Study 2010: workshop on methods and key findings
2. Outline
1. Methods for estimating Burden of Disease attributable
to risk factors
2. Summary of key findings
2
3. Outline
1. Methods for estimating Burden of Disease
attributable to risk factors
2. Summary of key findings
3
5. Calculating risk factor burden
1. Select risk-outcome pairs
2. Estimate exposure distributions to each risk factor in the
population
3. Choose a counterfactual exposure distribution: theoretical
minimum risk exposure distribution (TMRED)
4. Estimate cause effect sizes: relative risk per unit of exposure
for each risk-outcome pair
5. Compute attributable burden, including uncertainty
5
6. Risk-outcome pair inclusion criteria
1. Likely importance of a risk factor to disease burden or policy
2. Availability of sufficient data and methods to enable
estimation of exposure distributions by country for at least
one of the study periods
3. Sufficient evidence for causal effect (convincing or probable
evidence) and to estimate outcome-specific effect sizes
4. Evidence to support generalizability of effect sizes to
populations other than those included in epidemiological
studies
6
7. GBD 2010: risks quantified
Unimproved water and sanitation
Unimproved water
Unimproved sanitation
Air pollution
Ambient particulate matter pollution
Household air pollution from solid fuels
Ambient ozone pollution
Other environmental risks
Residential radon
Lead exposure
Child and maternal undernutrition
Suboptimal breastfeeding
Nonexclusive breastfeeding
Discontinued breastfeeding
Childhood underweight
Iron deficiency
Vitamin A deficiency
Zinc deficiency
Tobacco smoking and secondhand smoke
Tobacco smoking
Secondhand smoke
Alcohol and other drugs
Alcohol use
Drug use (opioids, cannabis, amphetamines)
Physical inactivity and low physical
activity
Physiological chronic disease risks
High fasting plasma glucose
High total cholesterol
High systolic blood pressure
High body mass index
Low bone mineral density
Sexual abuse and violence
Childhood sexual abuse
Intimate partner violence
7
8. GBD 2010: risks quantified (cont’d)
Dietary risk factors
Diet low in fruits
Diet low in vegetables
Diet low in whole grains
Diet low in nuts/seeds
Diet low in milk
Diet high in unprocessed red meat
Diet high in processed meat
Sugar-sweetened beverages
Diet low in fiber
Diet low in calcium
Diet low in seafood omega-3
Diet low in polyunsaturated fatty acid (PUFA)
Diet high in trans fatty acids
Diet high in sodium
Occupational exposures
Asbestos
Arsenic
Benzene
Beryllium
Cadmium
Chromium
Diesel
Formaldehyde
Nickel
PAHs
Secondhand smoke
Silica
Sulfuric acid
Asthmagens
Particulates and gases
Noise
Occupational injury
Low back pain
8
9. Calculating risk factor burden
1. Select risk-outcome pairs
2. Estimate exposure distributions to each risk factor in the
population
3. Choose a counterfactual exposure distribution: theoretical
minimum risk exposure distribution (TMRED)
4. Estimate cause effect sizes: relative risk per unit of exposure
for each risk-outcome pair
5. Compute attributable burden, including uncertainty
9
12. 12
PM2.5 (µg per m3)
Example: ambient PM pollution (cont’d)
• Satellite-based measures of aerosol optical depth (AOD)
• TM5 chemical transport models
• Cross-walk to ground-based PM2.5 sensor data
13. Calculating risk factor burden
1. Select risk-outcome pairs
2. Estimate exposure distributions to each risk factor in the
population
3. Choose a counterfactual exposure distribution:
theoretical minimum risk exposure distribution (TMRED)
4. Estimate cause effect sizes: relative risk per unit of exposure
for each risk-outcome pair
5. Compute attributable burden, including uncertainty
13
14. Calculating risk factor burden
1. Select risk-outcome pairs
2. Estimate exposure distributions to each risk factor in the
population
3. Choose a counterfactual exposure distribution: theoretical
minimum risk exposure distribution (TMRED)
4. Estimate cause effect sizes: relative risk per unit of
exposure for each risk-outcome pair
5. Compute attributable burden, including uncertainty
14
15. Risk-outcome effect sizes
1. Recent or new systematic reviews and meta-analyses
2. New/updated effect size estimates conducted for:
• Water and sanitation
• Dietary risk factors
• Air pollution: integrated exposure response (IERs)
3. Examined validity of single dietary risk factor effect sizes:
• Dietary pattern studies, e.g., Mediterranean diet
• Randomized controlled feeding studies, e.g., DASH, OMNI Heart
15
17. Calculating risk factor burden
1. Select risk-outcome pairs
2. Estimate exposure distributions to each risk factor in the
population
3. Choose a counterfactual exposure distribution: theoretical
minimum risk exposure distribution (TMRED)
4. Estimate cause effect sizes: relative risk per unit of exposure
for each risk-outcome pair
5. Compute attributable burden, including uncertainty
17
18. Population attributable fractions
• Continuous risk factors:
• Categorical risk factors:
• Joint effects of risk factor cluster:
18
R
r
rPAFPAF
1
)1(1
19. Limitations
• Few risks for major communicable diseases
• Exclusion of risk-outcomes based on insufficient
data
• Limited exposure distribution data
• Potential for residual confounding, especially in
the absence of intervention studies
• Uncertainty about generalizability of effect sizes
across populations
• Approximation of joint effects of risk factor
clusters
19
20. Outline
1. Methods for estimating Burden of Disease attributable
to risk factors
2. Summary of key findings
20
21. Burden of Disease attributable to 25 leading risk factors
as a percentage of global DALYs, both sexes, 2010
Residential radon
Ambient ozone pollution
Low bone mineral density
Unimproved water source
Childhood sexual abuse
Zinc deficiency
Vitamin A deficiency
Lead exposure
Unimproved sanitation
Intimate partner violence
Drug use
High total cholesterol
Suboptimal breastfeeding
Iron deficiency
Occupational risks
Physical inactivity and low physical activity
Ambient particulate matter pollution
Childhood underweight
High fasting plasma glucose
High body-mass index
Alcohol use
Household air pollution from solid fuels
Tobacco smoking
High blood pressure
Dietary risks
0 2 4 6 8 10
DALYs (%)
Cancer
Cardiovascular and circulatory
diseases
Chronic respiratory diseases
Cirrhosis
Digestive diseases
Neurological disorders
Mental and behavioural disorders
Diabetes, urogenital, blood, and
endocrine
Musculoskeletal disorders
Other non-communicable diseases
HIV/AIDS and tuberculosis
Diarrhoea, lower respiratory infections,
& other common infectious diseases
Neglected tropical diseases and malaria
Maternal disorders
Neonatal disorders
Nutritional deficiencies
Other communicable diseases
Transport injuries
Unintentional injuries
Intentional injuries
War and disaster
Global, 2010
Leading risk factors, percent of total DALYs
21
22. Global risk factor ranks and percentage change with 95% UI
for all ages and sexes combined, 1990 and 2010
22
25. Key risk factor messages
• Dramatic shift away from communicable disease risks in
children toward noncommunicable disease risks in adults
• Global rise in high BMI and glucose emphasizes research
priorities given the absence of effective interventions
• More nuanced understanding of the role of diet in
preventing chronic disease
• Considerable variation in risk factor burden by region and
country
• In much of sub-Saharan Africa, the leading risks continue
to be those associated with poverty
25
Hinweis der Redaktion
Chris (and others) presenting the overall flow chart for estimating the global burden of diseases, injuries and risk factors. I will be covering four steps shown in the bottom left hand corner of the flow chart outlined in green. The first step in computing risk factor burden, however, is not shown on this diagram and that is the selection of risk-outcome pairs to be included in the quantification of risk burden. An example of a risk outcome pair is systolic blood pressure and its effects on ischemic heart disease. Once risk outcome pairs are selected, the next steps are to estimate the current exposure distribution to each risk factor. For blood pressure this would be the mean and standard deviation of systolic blood pressure in the population. The theoretical minimum risk exposure distribution to which the current exposure will be compared to, and the relative risk per exposure unit for each of the risk-outcome pairs. These three steps allow us to calculate the fraction of the disease burden for each of the outcomes that is currently attributable to the risk factor, namely the population attributable fraction or PAF for each risk-outcome pair.PAFs are then multiplied by the corresponding YLL and YLDs for the specific outcome to determine the YLLs, YLDs and DALYs attributable to the risk factor. Uncertainty in the estimation of risk attributable burden is computed by generating a 1,000 draws of each of the corresponding inputs.
For the selection of risk outcome pairs, we used a set of four criteria to guide these choices.
The four risk inclusion criteria for GBD 2010 were :2. For example, while household surveys and censuses routinely collect information on water and sanitation, information on hygiene exposure is extremely limited and it was not subsequently included in the GBD20103. Importantly, we also require sufficient epidemiological evidence to estimate outcome-specific effect sizes. For example, there is a large body of evidence documenting the effect of maternal education on child mortality but the literature is predominantly focused on all-cause child mortality outcomes. 4. For example, we did not include the effects of intimate partner violence on HIV burden as longitudinal evidence is only available from South Africa and there is uncertainty about how the effect sizes might be applied to other populations which may have very different transmission dynamics.
Based on these criteria we included:
-
This flow chart provides a summary of the exposure estimation process, including the types of data sources used which ranged from household surveys, administrative data and censuses as well as trade sales and consumption data and as I will show in a moment satellite imagery. Similar to the previous presentations, we make a number of corrections for representativeness and selection and importantly for risk factors, cross walk between different definitions of risk exposure so that the measure of exposure matches best with the effect size estimates. For example, for computing the burden due to high sodium consumption, we cross walk between dietary based measures of sodium consumption and urinary sodium as a gold standard. We utilize a range of statistical procedures that generate predictions based on time, space and covariates to produce exposures by risk, age, sex, year and country.
A good example of this estimation process is for ambient particulate matter pollution. This map shows the availability of data on particulate matter measures as PM2.5 from ground-based monitoring stations. As you can see data are largely restricted to cities and are unavailable for many populations globally, for example for most parts of Africa.
To estimate exposure distributions for all populations globally, we used sattelite-based measures of aerosol
The third steps is to choose a counterfactual exposure distribution. The choice of the TMRED was guided by the epidemiological literature in terms of how
WASH and seafood omega-3s are posters
For ambient air pollution, as we saw in the previous map there are large populations such as those in East Asia that are exposed to high levels – greater than 80 ug per cubic meter - of PM2.5. The epidemiological studies of the health effects of ambient PM2.5 are largely restricted to North American and European populations with lower levels of exposure. To quantify the health effects of high ambient PM2.5 exposure we integrated evidence across different sources of PM2.5 as shown in this figure. On the y-axis we have the relative risk of lung cancer and on the x-axis we have log-transformed PM2.5. The red circles indicate the various ambient PM2.5 epidemiological studies, green household or indoor air pollution and the blue circles various categories of cigarette consumption. By fitting non-linear functions to this data, we are then able to estimate the health effects of PM2.5 exposure for populations with higher levels of ambient PM exposure, that is, largely between the ambient PM studies shown in red and the green household air pollution studies.
If we focus on the top 25 risk factors and risk factor clusters, in 2010, the cluster of dietary risks were the leading risk factors in terms of global disability adjusted life years, accounting for almost a tenth of global disability adjusted life years, followed by high blood pressure, tobacco smoking, including second hand smoke and HAP. The colors on this figure indicate the underlying cause attributable to the risk factor. For example, the effects of high blood pressure are primarily via cardiovascular disease while the effects of alcohol use are across a more diverse range of outcomes including cancer, cardiovascular disease, injuries, and communicable diseases. Many of the leading risk factors such as high BMI, high fasting plasma glucose, physical inactivity, and high total cholesterol have effects on primarily non-communicable disease. Household air pollution and ambient particulate matter pollution which have effects on both communicable disease and non-communicable disease were ranked fourth and 9th, respectivelyThe leading communicable disease risk factors in 2010 were childhood underweight ranked 8th and accounting for more than 3% of total health burden with iron deficiency and suboptimal breastfeeding accounting for more than 2% of health burden.
The picture in 2010 reflects a dramatic shift away from communicable disease risk factors towards non-communicable disease risk factors as shown in this arrow diagram. This diagram depicts the leading risk factors in 1990 on the left and the leading risk factors in 2010 on the right. Risks are color coded according to the cluster of risk factors. For example, red are the maternal and child undernutrition risks and blue are the physiological risks for chronic disease. The lines connect the same risk. Numbers in the right hand most column represent the % change in the risk factor between 1990 and 2010. In 1990 childhood underweight was the leading risk factor accounting for almost 8% of total health burden in 1990; by 2010 it had more than halved and was the 8th ranked risk factor. Similar declines are present for other communicable disease risks as denoted by the dotted lines. These include suboptimal breastfeeding, unimproved sanitation and water as well as micronutrient deficiencies such as Vitamin A and Zinc. Solid lines denote whether the risk increased in rank. For example, high blood pressure was previously the 5th leading risk in 1990 and the 2nd leading risk in 2010. Overall, the burden of non-communicable disease risk factors has increased with the two of the other more notable being high body mass index for which the burden increased by more than 80% and high fasting plasma glucose that increased by more than 50%.
The global results mask considerable variation by region. This heatmap shows the top 25 global risk factors as rows ordered by the global rank and I have included just the Asian regions as columns ordered by the mean age at death, a marker of the epidemiological transition. Cells are shaded according to the rank of the risk factor in the corresponding region with dark red indicating the 5 leading risk factors and green indicating ranks 21-25 or greater. A number of the important patterns to note are:The cluster of dietary risk factors for chronic disease, SBP and Tobacco are generally among the top 5 ranked risk factors for all regions outside of SSA. Alcohol is a leading risk factor in Southern sub-Saharan Africa, Eastern Europe and Latin AmericaHousehold air pollution is an important cause of disease burden in many parts of Asia and sub-Saharan Africa. Ambient particularly matter pollution is the 4th leading risk factor in East Asia. Despite declines, the cluster child and maternal undernutrition risk factors remain the leading risk factors in Western, Central and Eastern sub-Saharan Africa