Alcohol-Attributable Fractions (AAF)
Prevalence of Alcohol Consumption
Alcohol-Attributable Deaths (AAD)
General Information on ARDI
Why are the ARDI estimates important?
The Alcohol-Related Disease Impact (ARDI) application generates
estimates of alcohol-attributable deaths (AAD) and years of potential life lost
(YPLL) due to alcohol consumption. These estimates provide vital information
to better understand the health consequences of excessive alcohol use in the
United States. In addition to estimating the national health effects of
alcohol consumption, the ARDI application can produce state estimates of AAD and YPLL. Such
state-specific analyses are needed because the prevalence of excessive alcohol
use, particularly binge drinking, is known to vary substantially by location. State-specific
estimates of alcohol-related health outcomes can better focus
discussions of evidence-based public health strategies (e.g., increasing
alcohol excise taxes and screening for alcohol misuse in clinical settings)
aimed at preventing consequences associated with excessive alcohol use.
Who is the intended audience for ARDI?
The primary audience for ARDI is state governments,
particularly state health departments and state substance abuse agencies
interested in determining the health impact of excessive alcohol use in their
state for policy and informational purposes. In addition, academic
researchers, particularly from schools of public health, will also benefit from
using the ARDI application to estimate alcohol-attributable deaths (AAD) and years of
potential life lost (YPLL) for research and analysis purposes.
The secondary audience includes those organizations involved with
alcohol-related treatment and prevention programs, as well as state health policy organizations.
These organizations include, but are not limited to, health-related nonprofit
organizations, primary care associations, advocacy groups, as well as local
boards of health, and city and county health departments.
Can ARDI be used to evaluate the effectiveness of public health programs?
ARDI is designed to estimate the health effects of excessive alcohol consumption over
a specified period of time (e.g., over a 5-year period). Therefore,
these estimates are not intended to be used to evaluate the effectiveness
of public health programs or policies aimed at reducing alcohol consumption.
Furthermore, these estimates are subject to year-to-year variations, which although
reduced by using multiple years of data, are still subject to anomalies in the
collection of mortality data that may not reflect changes in actual alcohol
consumption resulting from public health programs.
Can ARDI be used to study trends over time in
alcohol-attributable deaths (AAD) or years of potential life lost (YPLL)?
The ARDI application is used to assess AAD or YPLL over a specified period of time,
usually 5 years. The application is not set up to examine trends in AAD or YPLL
over time mainly due to the year-to-year variations in these estimates that may
not be due to alcohol consumption.
Can ARDI be used to compare my state’s alcohol-related
outcomes to other states or national estimates?
The estimates provided in ARDI are the total number of
alcohol-attributable deaths (AAD) or years of potential life lost (YPLL) for
the location specified. To accurately compare states to each other or
to national estimates, the AAD and YPLL must be adjusted appropriately to
reflect differences in demographics between locations (e.g., regional
differences in average age of the population). ARDI does not report adjusted
AAD or YPLL; therefore, the estimates provided in the ARDI reports should
not be compared between locations.
Can I obtain a breakdown of alcohol-attributable
deaths (AAD) and years of potential life lost (YPLL) by race and ethnicity from
No, the current version of ARDI is only able to stratify AAD
and YPLL by sex and age.
How can you have an alcohol-attributable death (AAD) or years of
potential life lost (YPLL) less than one (i.e., <1)?
ARDI assigns a value of less than one when there was exactly one
death from a specific cause that is not 100% alcohol-attributable (i.e., a condition
with an alcohol-attributable fraction (AAF) less than one). Under these circumstances,
the number of deaths from this cause equals the AAF for this cause.
However, it is important to note that while the reported value is shown
as less than one, ARDI includes the actual value (i.e., the AAF for this condition)
in calculations of YPLL.
Why are beneficial effects associated with alcohol
consumption reported for some causes but not for others?
Alcohol consumption has been shown to reduce the risk of
death from some causes (e.g., cholethiases). For these causes, the relative
risk estimates included in ARDI are less than one. When these estimates are
used to calculate indirect alcohol-attributable fractions (AAF), the result is
a negative AAF. When this negative AAF is then multiplied by the total number
of deaths for that condition, the resulting number of deaths is negative. This
indicates that there are net lives saved from alcohol use at a particular
consumption level for these causes.
Why are alcohol-attributable deaths (AAD) and years
of potential life lost (YPLL) among persons younger than 21 years of age described as
being caused by “exposure” to alcohol?
The term “exposure” is used to describe AAD and YPLL for
persons younger than 21 years of age because deaths in this age group may result from
an individual’s own drinking, or from the second-hand effects of someone else’s
drinking (e.g., deaths from riding in a vehicle with an alcohol-impaired
driver). Some causes of death in ARDI specifically affect infants and children
(e.g., child maltreatment, fetal alcohol syndrome, and low birth weight, etc.)
are entirely the result of another person’s (e.g., the infant’s mother’s)
drinking. Therefore, while all the AAD and YPLL included in the younger than 21
reports are attributable to alcohol exposure, many of them are actually due to
someone else’s alcohol consumption.
How often are the data updated in ARDI?
Default data on deaths by cause, life expectancy, and
prevalence of alcohol consumption are updated periodically to reflect newly
available mortality estimates. Risk estimates and alcohol-attributable
fractions (AAF) are re-examined periodically as new scientific estimates become
Alcohol-Attributable Fractions (AAF)
What are alcohol-attributable fractions (AAF)?
Alcohol-attributable fractions (AAF) are used to express the extent to which
alcohol consumption contributes to a health
outcome. In ARDI, AAF measure the total proportion of deaths from various
causes that are directly or indirectly attributable to alcohol consumption.
How are alcohol-attributable fractions (AAF)
Information on the calculations of the AAF used in ARDI can
be found in the Methods section.
Why is the Fatal Accident
Reporting System (FARS) used to obtain alcohol-attributable fractions (AAF) for
motor-vehicle crash deaths?
The FARS, which is administered by the National Highway
Traffic and Safety Administration (http://www.nhtsa.dot.gov),
provides annual estimates of alcohol involvement for all traffic crashes that
occurred on United States roadways in a given year. The FARS protocol for
determining alcohol involvement in a crash also makes it possible to calculate
age-specific AAF. Therefore, FARS is considered the best and most timely source
of AAF for motor-vehicle crash deaths.
Prevalence of Alcohol Consumption
How does ARDI determine the
prevalence of alcohol consumption to calculate indirect estimates of
alcohol-attributable fractions (AAF)?
The prevalence of alcohol consumption, measured as the
average number of drinks consumed per day, is estimated using self-reported
information on alcohol consumption from the CDC’s Behavioral Risk Factor
Surveillance System (BRFSS), www.cdc.gov/BRFSS.
The BRFSS includes three questions on the core survey about alcohol use in the
past 30 days: the frequency of drinking days, the quantity of drinks consumed
on drinking days, and the frequency of binge drinking. The total quantity and
frequency of drinking is combined to form an estimate of the average amount of
alcohol consumed per day. More detailed information regarding
the calculation of average daily alcohol consumption and the cutpoints used in calculating indirect
AAF are explained in the ARDI Methods section under Prevalence Data.
Can prevalence data from other sources besides the CDC’s
Behavioral Risk Factor Surveillance System (BRFSS) be used to determine
the prevalence of alcohol consumption?
The Custom Data section of ARDI allows users to enter their own mortality and prevalence data. Prevalence
data can be used from surveys other than the BRFSS as long as questions regarding both the quantity and
frequency of alcohol consumption are used and the survey is administered to a representative sample.
Why are there two different
prevalence cutpoints for alcohol consumption listed in ARDI, and how do I know
which one is used to calculate alcohol-attributable fractions (AAF) for a given
ARDI uses several meta-analyses to obtain the risk estimates
for several chronic causes of death included in the application. Prevalence
cutpoints for defining levels of alcohol consumption are determined by the
authors of these analyses and included in the Methods section.
Therefore, the prevalence cutpoints that are used to
calculate the indirect AAF for a particular cause correspond to the
meta-analysis that was used to identify risk estimates for these causes. Most
of the chronic causes of death used relative risk estimates from the
meta-analysis performed by English et al. (1995) and Ridolfo and Stevenson
(2001); however, for three causes (ischemic stroke, hemorrhagic stroke and
prostate cancer) the relative risk estimates are drawn from a meta-analysis by Corrao
et al. (1999) and Bagnardi et al. (2001).
Alcohol-Attributable Deaths (AAD)
How were the alcohol–related causes of death
included in ARDI selected?
The US Centers for Disease Control and Prevention (CDC) convened a scientific work group comprised of
experts on alcohol and health to guide the development of the ARDI application. One of the group’s tasks was
to select alcohol-related causes of death based upon ones that were previously examined in meta-analyses.
Some causes (e.g., tuberculosis, pneumonia, and hepatitis C) were not included in this version of ARDI because
suitable pooled relative risk estimates or alcohol-attributable fractions (AAF) were not available for them at the
time the work group convened.
How do I know if a cause of death is alcohol-attributable?
ARDI currently includes a list of 54 causes of death with
enough scientific evidence to show alcohol-attribution. The ICD codes
associated with these deaths are found at Alcohol-Related ICD Codes. ARDI
uses the underlying cause of death listed on death certificates to determine if the death was alcohol-attributable.
calculate the number of deaths due to binge drinking?
Of the nearly 88,000 average annual alcohol-attributable deaths reported in ARDI for 2006–2010, more than 50% are due to binge drinking. This is because the studies that were used to obtain alcohol-attributable fractions for the acute causes of death (e.g., injuries) included in the ARDI application defined a death as being alcohol-attributable if the decedent, or another person who was responsible for a death (e.g., the driver of a vehicle in a fatal motor vehicle crash), had a blood alcohol concentration (BAC) greater than or equal to 0.10 g/dL at the time of death, and this BAC level is greater than the level used by the National Institute on Alcohol Abuse and Alcoholism to define binge drinking (i.e., greater than or equal to 0.08 g/dL).
Can I use ARDI to estimate
alcohol-attributable deaths (AAD) and years of potential life lost (YPLL) for
different years — that is, for years other than those already included in ARDI?
Yes, in the Custom Data section, you can enter death data
and prevalence data for any given year or a range of years; however, both the
death data and the prevalence data should be for the same time period.
Can I use ARDI to estimate alcohol-attributable
deaths (AAD) and years of potential life lost (YPLL) for cities, counties, or
other geographic areas?
Through the Custom Data section in ARDI, users can enter
data on total deaths and the prevalence of alcohol consumption for other
geographic areas besides state-level. The Custom Data section allows reports
to be produced estimating alcohol-attributable deaths (AAD) and years of
potential life lost (YPLL) for the specified locations.
Is there a
minimum population size needed to estimate alcohol-attributable deaths (AAD) or
years of potential life lost (YPLL) in the Custom Data section of ARDI?
There is no minimum population size needed to run ARDI.
However, the smaller the sample size, the more variable and less reliable are
the results. In general, it is recommended that analyses be limited to
populations of 500,000 persons or more.
Can I use different alcohol-attributable
fractions (AAF) than those included in ARDI?
In the Custom Data section of ARDI, users can change the AAF
for the acute causes of death if the information is available for the
geographic area specified by both deaths and prevalence of alcohol consumption.
Who do I contact if I am
having problems with the site?
Please contact us through the online form.
How do users cite ARDI as a
reference/resource in publications?
Please use the following citation for ARDI if using the data in publication:
Centers for Disease Control and Prevention. Alcohol Related Disease Impact (ARDI) application, 2013.
Available at www.cdc.gov/ARDI.
Can I use ARDI to calculate economic impacts related
to alcohol use?
No. This feature is not available in the current version of
the ARDI application.
Where can I find more information on the health and social impacts of alcohol consumption?
More information on the health and social impacts of alcohol consumption is available at the
CDC's Alcohol and Public Health Web site (http://www.cdc.gov/alcohol).