An adaptation of the yesterday method to correct for under-reporting of alcohol consumption and estimate compliance with Canadian low-risk drinking guidelines.
Zhao, Jinhui ; Stockwell, Tim ; Thomas, Gerald 等
In November 2011, the Federal, Provincial and Territorial Health
Ministers approved Canada's first national low-risk drinking
guidelines (LRDG). (1) The guidelines recommend that women do not exceed
10 standard drinks (SDs) per week or 3 in any one day, while men should
have no more than 15 per week or 4 in any one day (one Canadian SD =
13.45 g of pure alcohol). (2, 3) This paper seeks to address the extent
of compliance with these guidelines using national survey data for the
three years leading up to their introduction. In particular, we provide
estimates of risky drinking both before and after making adjustments
using a recently published methodology that corrects for under-reporting
in surveys. (4)
In general population surveys, three main methods are used to ask
people to report their alcohol intake. (5) The Quantity-Frequency
approach (QF) asks only two questions: "How much alcohol do you
usually drink?" and "How often do you drink?" The
Graduated-Frequency approach (GF) asks how often people drink specified
amounts of alcohol in one day, usually starting with large amounts and
graduating down to smaller quantities. The Last Seven Day approach (L7D)
asks subjects how much alcohol they drank on each of the last seven
days. However, per capita consumption estimates derived from self-report
surveys usually yield estimates of between 40% and 60% of those derived
from official alcohol sales data. (6) Canadian analyses have found that
some national alcohol and addiction surveys have provided even lower
estimates when compared to sales data. For example, the QF method
yielded 37.0% coverage of alcohol sales data and the L7D approach
yielded 31.9% coverage in the 2004 Canadian Addiction Survey. (7) In the
2008-2010 Canadian Alcohol and Drug Use Monitoring Surveys (CADUMS), the
coverage rates were just 33.8% and 35.6% for the QF and L7D methods
respectively. (4)
One method that yields improved coverage of alcohol sales is the
Yesterday Method, first employed in French surveys of alcohol
consumption. (8) This approach provides higher overall coverage of
sales, (4) particularly when the method is combined with empirically
based estimates of drink size and beverage strength. (9) For example,
when using this approach the 2004 Australian National Drug Strategy
Health Survey (NDSHS) accounted for 79.1% of the sales data while the QF
and GF approaches yielded only 50.5% and 52.4% coverage of sales
respectively. (9)
The Yesterday approach was used in the CADUMS from 2008 to 2010;
several questions asked subjects to report the number of drinks of beer,
wine, spirits, wine cooler and spirit cooler they had yesterday. (10-12)
Respondents were provided with definitions of a "drink" for
each beverage type, i.e., Canadian standard drink sizes. We refer to
this as the Beverage-Specific Yesterday (BSY) approach. We recently
described a new methodology that involves combining information from the
BSY with official sales data to correct for differential under-reporting
by different population groups using the CADUMS. (4) Our approach
revealed that men and women tend to underestimate their consumption to a
similar degree, however, young people and low volume drinkers tend to
under-report to a greater extent than older and high volume drinkers. In
the current study, we apply these adjustments for under-reporting and
use them to develop corrected estimates of compliance with Canada's
new national LRDGs.
The specific aims of this study are: to estimate degree of
noncompliance with Canada's LRDG in different groups of drinkers
after adjusting for degree of under-reporting, and to identify which
types of beverage are more likely to be consumed when LRDGs are
exceeded. (2)
METHOD
Survey sampling
Initiated in April 2008 by Health Canada, the CADUMS was a near
continuous survey on alcohol and other substance use among Canadians.
(10-12) The survey was derived from the 2004 Canadian Addiction Survey
which contained questions on substance use and associated harms. (13)
The CADUMS used random digit dialing to obtain a stratified sample
across all 10 provinces with equal representation of subjects each
month. It is based on a two-stage (telephone household, respondent)
random sample stratified by province. The survey covers population aged
[greater than or equal to] 15 years in the 10 provinces, excluding
permanent residents of institutions, people living in households without
a telephone and people who only have cell phones. The sample size was
16,674 in 2008, 13,082 in 2009 and 13,615 in 2010, representing an
estimated 25,957,435 Canadians aged 15 years and older.
Measures estimating extent of under-reporting alcohol consumption
Table 1 presents the questions used to estimate alcohol consumption
in the CADUMS. The QF approach was used to estimate per capita litres of
pure alcohol consumption by multiplying the typical number of standard
drinks (SDs) reported consumed per day (1 SD = 17.05 mL pure alcohol)
and the number of drinking days in the past year. The BSY method was
used to estimate population per capita alcohol consumption from the
total number of SDs reported consumed "yesterday" across the
whole sample. Weights were added for the day of the week that interviews
were conducted to get equal representation of these in the estimates.
The extent of under-reporting was estimated by comparing the QF and BSY
estimates with official sales data. In the following discussion, partial
adjustment means adjusting the consumption estimated by the QF to the
level of the BSY estimates. To obtain full adjustment, we used official
sales data to further uplift the partially adjusted consumption levels.
Methods used to correct for under-reporting in age-gender groups
Information from both the QF and BSY methods were employed to
estimate adjustments for under-reporting within each age-gender category
using methods described in greater detail elsewhere. (4) First, a
correction factor for frequency (F) of drinking days was developed for
each age-gender group by comparing observed numbers of respondents
reporting drinking the day before with the expected numbers based on
responses to the QF. The F adjustment used the number of drinkers in
each QF frequency category who reported drinking the day before the
interview to estimate a more accurate conversion factor. Specifically,
the proportion of drinkers in each frequency category who drank
"yesterday" was multiplied by 365. This procedure assumes that
respondents a) can remember if they drank the previous day and b) were
truthful about this in the interview. Consistent with other studies,
(14) there was substantial under-reporting of frequency among
individuals reporting less frequent consumption on the QF, with 2 to 3
times more individuals reporting consumption the day before the
interview than would be predicted by their responses on the QF. The only
exception was for individuals reporting consumption "every
day", for whom an actual average drinking frequency of nearer six
days a week was found. The correction factors for frequency for men and
women respectively were 20.43 and 16.72 for less than once a month,
41.79 and 32.76 for once a month, 82.93 and 72.71 for 2-3 times per
month, 115.29 and 107.70 for once per week, 181.42 and 169.91 for 2-3
times a week, 271.95 and 259.01 for 4-6 times a week and 319.85 and
306.34 for every day. Second, unique correction factors were calculated
for typical quantities reported by respondents according to specific
age-gender categories. The quantity or Q adjustment was conducted in two
steps. First, we calculated mean numbers of standard drinks consumed
"yesterday" separately for men and women within each QF
frequency category and compared this with mean "quantity"
calculated directly from BSY responses. It was assumed that the BSY
would provide a more accurate estimate of mean quantity consumed on a
drinking day within these broad categories of drinkers. The second
adjustment to Q involved calculating "uplift" correction
factors for each age and gender group based on i) the extent of
under-reporting of each major beverage class, i.e., beer, wine and
spirits estimated from BSY versus sales data, and ii) the proportions of
each beverage that made up consumption on a typical day for each age and
gender group based on the BSY questions. This adjustment, therefore,
took account of different degrees of under-reporting of each of the main
beverages (beer, wine and spirits) in comparison with official
Statistics Canada sales data (see Table 2) and also the different
pattern of preferences for each beverage type for individuals of
different age and gender based on their BSY responses. This resulted in
upwards adjustments for quantity (Q) ranging from 48.9% for 45 to 64
year old women to 56.1% for 15 to 18 year old women (other adjustment
factors for men and women were between this range). (4)
Measures of compliance with Low Risk Drinking Guidelines
Extent of non-compliance with the LRDG was estimated both with and
without the above adjustment for under-reporting, both for "risky
drinkers" (i.e., those not complying with LRDG) and "risky
drinks" (i.e., drinks consumed on days when LRDG were exceeded).
Risky drinking most relevant to chronic diseases or other long-term
health harm was defined as having 16 or more drinks per week for men and
11 or more drinks per week for women. (2) The percentages of drinkers
exceeding these limits in a typical week were calculated from the last
12 months QF questions (i.e., typical quantity per day times typical
number of drinking days per week). "Risky" drinks for chronic
health harms were counted as all those consumed by these "risky
drinkers". Risky drinking for acute or short-term health harms is
defined in Canada as five or more drinks in one day for men and four or
more for women. (2) All drinkers who reported these levels as typical
for drinking days in the QF were deemed "risky drinkers" for
acute harm. The proportion of drinks consumed on days when these limits
were exceeded was calculated directly from the BSY questions. All drinks
consumed on such days were counted as "risky" for acute harm.
The QF method is typically used to estimate individual drinking patterns
over the past year but suffers from substantial under-reporting. The BSY
cannot be used to measure individual consumption patterns over time, but
can estimate beverage preferences, typical drinking quantities and
typical frequencies more accurately for subgroups of drinkers. The QF
approach was used to estimate compliance with LRDG both before and after
adjustment for under-reporting once the more accurate data from the BSY
questions and official sales data were factored in to the estimates. We
present both partially and fully adjusted estimates of LRDG compliance
for different age and gender groups. Partial adjustment uses estimates
of typical drinking quantity (Q) and frequency (F) for population
subgroups based on BSY responses. Full adjustment incorporates
additional Q uplift factors to match official Statistics Canada data on
age 15+ per capita sales of beer, wine and spirits. (4) The latter
calculation first required estimates of the proportion of pure alcohol
consumed as beer, wine and spirits for each age-gender subgroup
calculated from the BSY responses.
Statistical analysis
The entire pooled sample across the three years is included in our
analysis to maximize statistical power. (15) The estimates are based on
the weighted sample and the expansion weights are recalculated and were
rescaled to the sample size. (15) Multivariate logistic regression
analysis was performed to examine the differences in percent of risky
drinkers among specified age and sex groups, and z-tests were used to
examine the differences in percent of risky drinks among age and sex
groups. (16) Statistical analyses were completed using SAS SURVEYMEANS,
SURVEYFREQ, SURVEYLOGISTIC and TABULATE procedures because these
procedures analyze sample survey data taking into account the sample
design effect. (17)
RESULTS
Per capita alcohol consumption estimate using the Yesterday Method
Table 2 presents per capita alcohol consumption by types of
beverage and percentage of beer, wine and spirits estimated from the
surveys compared to sales data. Per capita consumption of beer
(including wine and spirit coolers), wine and spirits estimated using
the BSY approach was substantially lower than recorded sales data.
Reported spirits consumption in the surveys was 0.75 L, which accounted
for only 34% of official spirits sales. The reporting of beer and wine
consumption was more accurate in the surveys, however, covering 51% and
62% of official sales respectively. The per capita consumption estimates
from sales data were outside the 95% CIs for the survey estimates,
indicating significant under-reporting of all beverage types.
Risky drinkers and "risky drinks" for chronic harm
Table 3 presents the estimates of the percentage of both risky
drinkers (those exceeding LRDGs) and risky drinks (drinks consumed on
days and weeks when LRDGs were exceeded) for chronic harm using the QF
method, i.e., exceeding or in excess of the weekly LRDG limits. Without
any adjustments for under-reporting, 6.8% of drinkers exceeded the
weekly LRDG and, collectively, contributed 42.3% of total alcohol
consumption. After the final uplift adjustments for under-reporting of
alcohol consumption (i.e., uplifting self-reported consumption of each
beverage type to sales data reflecting gender- and age-specific beverage
preferences), it was estimated that 27.3% of Canadian drinkers exceed
the weekly LRDG limits, between them accounting for 68.3% of total
Canadian alcohol consumption. Wald chi-square tests in multivariate
logistic regression of risky drinkers showed a significantly higher
percent of risky drinkers, either unadjusted or adjusted with F values
and/or Q uplift factors among those aged 18/19-24 years than among other
age groups (p < 0.001); and a significantly higher percent among men
than women (p < 0.001). Z-tests also suggested a significantly higher
percent of risky drinks among those aged 18/19-24 than among the other
four age groups (p < 0.001); a significantly higher percent among
those aged 15-17/18 than among those aged 25 years or older (p <
0.001); and a significantly higher percent among men than women (p <
0.001).
Risky drinkers and risky drinks for acute harm
Table 4 presents estimated percentages of risky drinkers and risky
drinks for acute harm for the three major categories of beverage. Before
adjusting for under-reporting, just over half of all alcohol consumed
(53.3%) was estimated to have been consumed on a day when LRDGs for
avoiding acute harm were exceeded and 16.7% of drinkers regularly
exceeded these limits. After full adjustment for under-reporting, it was
estimated that 38.6% of all Canadian drinkers exceeded the daily LRDG
limits (i.e., 3 for women and 4 for men) on a typical drinking day,
collectively accounting for 80.5% of all alcohol consumed. Despite the
higher thresholds for men, there were more male risky drinkers than
female (40.1% versus 37%). Across all age groups, spirits were more
likely to be consumed on risky drinking days (92.1% of all spirit
consumption), than beer (83% of all beer consumption) or wine (60.6% of
all wine consumption).
DISCUSSION
This study explored and made adjustment for the extensive
under-reporting of alcohol consumption in the CADUMS. Self-report
surveys are commonly used to assess levels of risky drinking, especially
since the advent of the first national drinking guidelines in Canada
introduced in late 2011. (2) The ability to make corrections for
under-reporting was made possible in the CADUMS by the inclusion of
questions about whether or not respondents consumed any alcohol the day
before the interview. A newly published methodology has shown that the
information in this Yesterday method can be used to correct for
under-reporting in both drinking frequencies and drinking quantities.
(4) Furthermore, the different degrees of under-reporting of beer, wine
and spirits compared with sales data and the different preferences among
population subgroups for these beverage types, can be used to further
uplift estimates based on self-reported consumption. (4) After applying
corrections for under-reporting, estimated non-compliance with weekly
LRDG limits increased from 6.8% to 27.3% among drinkers, and from 42.3%
to 68.3% with respect to drinks; and drinking above the daily LRDG
limits increased from 16.7% to 38.6% among drinkers, and from 53.3% to
80.5% with respect to drinks. It should be noted that the 4 drinks per
day limit for men and the 3 drinks per day limit for women is
recommended "on occasion" with the usual daily limit set one
drink lower for both sexes (3 for men and 2 for women).
After correcting for under-reporting in CADUMS, we estimated the
proportion of all alcohol consumed by different subgroups of drinkers
that took place on days or weeks when LRDG were exceeded. This
proportion was especially high among young people, being over 92% (acute
harm) and 75% (chronic harm) for those under legal drinking age and over
91% (acute harm) and 80% (chronic harm) for adults aged 18/19 to 24
years compared with 80.5% (acute harm) and 68.3% (chronic harm) for the
population as a whole.
Our study found that young people aged 15-24 had a significantly
higher rate of drinking risky drinks for both chronic and acute harms
compared to other age groups, particularly seniors. Further, beer and
spirits were most likely to be drunk on risky drinking days among young
people. Thus government pricing and taxation policies for controlling
consumption and harm might usefully be targeted towards beer and spirits
for which the highest costs are incurred and toward younger drinkers who
tend to prefer these categories of beverages. Australian research
employing the Yesterday method has also identified spirits and regular
strength beer as being the beverages most likely to be consumed on risky
drinking occasions. (9) This research also found that lower strength
beer (< 3.8% by volume) was the least likely to be drunk when LRDGs
are exceeded, suggesting that encouraging consumption of lower strength
beverages may be an effective approach for reducing alcohol-related harm
and costs.
Some limitations of our analysis should be kept in mind. Purchasing
and consumption are not the same phenomenon, as other sources of alcohol
consumption such as home production and goods purchased outside Canada
are possible. (5) Second, it is important to acknowledge the low
response rates in the CADUMS (43.5% in 2008, 44.7% in 2009 and 44.3% in
2010) especially for younger people with cellphones, an increasing
problem with population surveys in developed countries. (7, 9, 18, 19)
Since younger and heavier drinkers may be more likely to be excluded
from the survey, (19) the estimates of drinks consumed on risky drinking
days are likely to be conservative.
CONCLUSION
In summary, after applying systematic corrections for
underreporting, there were substantial increases in estimates of the
proportion of the Canadian population at risk for both long- and
short-term harm from their drinking. Further, it was estimated that the
great majority of alcohol consumption occurs on days and weeks when
national low risk drinking guidelines are exceeded. We recommend
inclusion of the detailed Yesterday method in future national alcohol
surveys (as was done in the Australian NDSHS surveys) so that more
accurate empirical estimates of patterns of risky alcohol consumption
can be obtained from Canadians. We caution, however, that these
adjustments for under-reporting may only be appropriate at the aggregate
level and cannot be applied reliably to individual survey responses.
La traduction du resume se trouve a la fin de l'article.
REFERENCES
(1.) Canadian Centre on Substance Abuse. Canada's Low-Risk
Alcohol Drinking Guidelines. Ottawa, ON: CCSA, 2011. Available at:
http://www.ccsa.ca/ Resource%20Library/2012-Canada-Low-Risk-Alcohol-Drinking-GuidelinesBrochure-en.pdf (Accessed May 14, 2014).
(2.) Butt P, Beirness D, Gliksman L, Paradis C, Stockwell T.
Alcohol and health in Canada: A summary of evidence and guidelines for
low-risk drinking. Ottawa: CCSA, 2011. Available at:
http://www.ccsa.ca/2011%20CCSA%
20Documents/2011-Summary-of-Evidence-and-Guidelines-for-Low-Risk%
20Drinking-en.pdf (Accessed September 18, 2012).
(3.) Stockwell T, Butt P, Beirness D, Gliksman L, Paradis C. The
basis for Canada's new low-risk drinking guidelines: A relative
risk approach to estimating hazardous levels and patterns of alcohol
use. Drug Alcohol Rev 2012;31:12634. PMID: 21954872. doi:
10.1111/j.1465-3362.2011.00342.x.
(4.) Stockwell T, Zhao JH, Macdonald S. Who under-reports their
alcohol consumption in telephone surveys and by how much? An application
of the 'yesterday method' in a national Canadian substance use
survey. Addiction 2014;109:1657-66. PMID: 24825591. doi:
10.1111/add.12609.
(5.) Stockwell T, Chikritzhs T. International guide for monitoring
alcohol consumption and related harm. Geneva, Switzerland: World Health
Organization, 2000. Available at: http://whqlibdoc.who.int/hq/2000/
who_msd_msb_00.4.pdf (Accessed April 27, 2011).
(6.) Pernanan K. Validity of survey data on alcohol use. In:
Gibbins RJ, Israel Y, Kalant H, Popham RE, Schmidt W Smart RG (Eds.),
Research Advances in Alcohol and Drug Problems. New York, NY: Wiley and
Sons, 1974;355-74.
(7.) Stockwell T, Zhao JH, Thomas G. Should alcohol policies aim to
reduce total alcohol consumption? New analyses of Canadian drinking
patterns. Addict Res Theory 2009;17:135-51.
(8.) Knibbe RA, Bloomfield K. Alcohol consumption estimates in
surveys in Europe: Comparability and sensitivity for gender differences.
Subst Abus 2001;22:23-38. PMID: 12466667.
(9.) Stockwell T, Zhao J, Chikritzhs T, Greenfield TK. What did you
drink yesterday? Public health relevance of a recent recall method used
in the 2004 Australian National Drug Strategy Household Survey.
Addiction 2008;103:919-28. PMID: 18482414. doi:
10.1111/j.1360-0443.2008.02219.x.
(10.) Health Canada. Canadian Alcohol and Drug Use Monitoring
Survey 2008: Microdata User Guide. Ottawa: Health Canada, 2009.
Available at: http://
prod.library.utoronto.ca/datalib/codebooks/cstdli/cadums/2008/cadumstechnical-guide-2008-final-eng.pdf (Accessed December 14, 2010).
(11.) Health Canada. Canadian alcohol and drug use monitoring
survey 2009: Microdata user guide. Ottawa: Health Canada, 2010.
Available at: http://
datalib.chass.utoronto.ca/codebooks/cstdli/cadums09/cadums%20final%20
technical%20guide%202009%20-en.pdf (Accessed March 29, 2011).
(12.) Health Canada. Canadian alcohol and drug use monitoring
survey 2010: Microdata user guide. Ottawa: Health Canada, 2011.
Available at: odesi.ca/
documentation/CADUMS/2010/cadums_2010_gid_eng.pdf (Accessed June 18,
2013).
(13.) Canadian Centre on Substance Abuse. Canadian Addiction Survey
2004: Microdata eGuide. Ottawa: CCSA, 2005.
(14.) Ye Y, Bond JC, Cherpitel CJ, Borges G, Monteiro M, Vallance
K. Evaluating recall bias in a case-crossover design estimating risk of
injury related to alcohol: Data from six countries. Drug Alcohol Rev
2013;32:512-18. PMID: 23574580. doi: 10.1111/dar.12042.
(15.) Thomas S, Wannell B. Combining cycles of the Canadian
Community Health Survey. Health Reports 2009;20(1):35-38.
(16.) Pagano M, Gauvreau K. Principles of Biostatistics (2nd ed.).
Pacific Grove, CA: Duxbury, 2000.
(17.) SAS Institute. SAS/STAT 9.3 user's guide. Cary, NC: SAS
Institute Inc., 2011.
(18.) Caetano R. Non-response in alcohol and drug surveys: A
research topic in need of further attention. Addiction 2001;96:1541-45.
PMID: 11784451.
(19.) Zhao JH, Stockwell T, MacDonald S. Non-response bias in
alcohol and drug population surveys. Drug Alcohol Rev 2009;28:648-57.
PMID: 19930019. doi: 10.1111/j.1465-3362.2009.00077.x.
(20.) Statistics Canada. Table 183-0019--Volume of sales of
alcoholic beverages in litres of absolute alcohol and per capita 15
years and over, fiscal years ended March 31, annual. Ottawa: Statistics
Canada, 2014. Available at:
http://cansim2.statcan.gc.ca/cgi-win/cnsmcgi.pgm (Accessed August 25,
2014).
Received: August 25, 2014
Accepted: February 20, 2015
Jinhui Zhao, PhD, [1] Tim Stockwell, PhD, [1, 2] Gerald Thomas, PhD
[1, 3]
Author Affiliations
[1.] Centre for Addictions Research of British Columbia, University
of Victoria, Victoria, BC
[2.] Department of Psychology, University of Victoria, Victoria, BC
[3.] BC Ministry of Health, Victoria, BC
Correspondence: Dr. Jinhui Zhao, Centre for Addictions Research of
British Columbia, University of Victoria, PO Box 1700 STN CSC, Victoria,
BC V8Y 2E4, Tel: 250-472-5935,
[email protected]
Acknowledgements: This study was supported by the Centre for
Addictions Research of BC Endowment Fund held by the University of
Victoria.
Conflict of Interest: None to declare.
Table 1. Questions and answer options for alcohol consumption in
the 2008-2010 Canadian Alcohol and Drug Use Monitoring Surveys
Method Question Answer options
1) QF--quantity- How often did you ... Less than once a
frequency approach drink alcoholic month
beverages during the
past 12 months? ... Once a month
... 2 to 3 times a
month
... Once a week
... 2 to 3 times a
week
... 4 to 6 times a
week
... Every day
... Never
During the past 12 ... number of drinks
months, on those in a typical day
days when you drank,
how many drinks did
you usually have?
2) L7D--last seven Starting with ... Number of drinks
day approach yesterday, that is yesterday, day
<> and including up before yesterday.
to 4 AM this
morning, how many
drinks did you have?
3) BSY--beverage- How many regular ... Number of drinks
specific yesterday bottles of beer did for beer yesterday
approach you drink yesterday,
including up to 4 AM
this morning?
(Include light beer
but not
dealcoholized beer)
How many glasses of ... Number of drinks
wine did you drink for wine yesterday
yesterday, including
up to 4 AM this
morning? (142 mL or
5 oz. glass of wine,
including ciders and
sparkling wines)
How many drinks of ... Number of drinks
cocktails, spirits for spirits
or hard liquor did yesterday
you drink yesterday,
including up to 4 AM
this morning?
(Straight or mixed
drink with one and a
half ounces (43 mL)
of hard liquor)
How many regular ... Number of drinks
bottles of wine for wine cooler
cooler did you drink yesterday
yesterday, including
up to 4 AM this
morning? (A regular
bottle (341 mL or 12
ounce) of cooler,
but not
dealcoholized
cooler)
How many regular ... Number of drinks
bottles of spirit for spirit cooler
cooler did you drink yesterday
yesterday, including
up to 4 AM this
morning? (A regular
bottle (341 mL or 12
oz.) of cooler, but
not dealcoholized
cooler)
Table 2. Estimated proportions of per capita consumption from
different alcoholic beverage types (litres of ethanol-person-
year) based on either weighted Beverage-Specific Yesterday (BSY)
responses or official Statistics Canada estimates based on sales
data
Beverage Statistics Statistics BSY CADUMS L
Canada L Canada % of ethanol/
ethanol/ total L year
year
([dagger])
Beer 4.20 51.22 2.14 ([+ or -] 0.18)
Wine 1.80 21.95 1.11 ([+ or -] 0.07)
Spirits 2.20 26.83 0.75 ([+ or -] 0.11)
Total 8.20 100.00 4.00 ([+ or -] 0.24)
Beverage CADUMS BSY % % BSY/
of total L Statistics
Canada
Beer 53.53 50.98
Wine 27.74 61.65
Spirits 18.73 34.06
Total 100.00 48.78
([dagger]) Annual litres per youth/adult aged 15+ for years 2008,
2009 and 2010. (20)
Table 3. Weighted estimates of % risky drinkers and "risky
drinks" for long-term harm as per Canada's Low-Risk Alcohol
Drinking Guidelines by age and sex using unadjusted and adjusted
estimates from QF from the 2008-2010 CADUMS
Age and sex Number of
drinkers
Age, years
([section])
15-17/18 1486
18/19-24 3154
25-44 9113
45-64 12,272
65+ 5782
Sex
Male 13,855
Female 17,952
Total 31,807
% Risky drinkers /All drinkers ([+ or -] 95% CI)
Age and sex Non-adjusted Partially adjusted
([dagger])
Age, years
([section])
15-17/18 5.96 ([+ or -] 3.16) 10.66 ([+ or -] 3.63)
18/19-24 *** 10.86 ([+ or -] 2.21) *** 19.92 ([+ or -] 2.76)
25-44 5.39 ([+ or -] 0.90) 9.19 ([+ or -] 1.16)
45-64 6.88 ([+ or -] 0.85) 7.90 ([+ or -] 0.90)
65+ 7.53 ([+ or -] 1.31) 6.87 ([+ or -] 1.22)
Sex
Male *** 8.87 ([+ or -] 0.91) *** 12.66 ([+ or -] 1.08)
Female 4.75 ([+ or -] 0.61) 6.60 ([+ or -] 0.74)
Total 6.84 ([+ or -] 0.56) 9.68 ([+ or -] 0.66)
% Risky drinkers -All
drinkers ([+ or -] 95% CI)
Age and sex Fully adjusted
([double dagger])
Age, years
([section])
15-17/18 25.62 ([+ or -] 4.62)
18/19-24 *** 41.32 ([+ or -] 3.42)
25-44 26.74 ([+ or -] 1.68)
45-64 25.76 ([+ or -] 1.41)
65+ 22.31 ([+ or -] 1.96)
Sex
Male *** 29.80 ([+ or -] 1.42)
Female 24.81 ([+ or -] 1.21)
Total 27.34 ([+ or -] 0.94)
% Risky drinks/All drinks ([+ or -] 95% CI)
Age and sex Non-adjusted Partially adjusted
([dagger])
Age, years
([section])
15-17/18 *** 50.68 ([+ or -] 1.46) *** 63.36 ([+ or -] 1.41)
18/19-24 *** 54.79 ([+ or -] 0.70) *** 68.93 ([+ or -] 0.65)
25-44 37.29 ([+ or -] 0.45) 46.17 ([+ or -] 0.46)
45-64 41.08 ([+ or -] 0.44) 42.71 ([+ or -] 0.44)
65+ 41.83 ([+ or -] 0.71) 38.16 ([+ or -] 0.70)
Sex
Male *** 45.57 ([+ or -] 0.32) *** 51.84 ([+ or -] 0.32)
Female 35.04 ([+ or -] 0.45) 38.79 ([+ or -] 0.46)
Total 42.25 ([+ or -] 0.26) 47.72 ([+ or -] 0.27)
% Risky drinks-All drinks
([+ or -] 95% CI)
Age and sex Fully adjusted
([double dagger])
Age, years
([section])
15-17/18 *** 75.38 ([+ or -] 0.61)
18/19-24 *** 80.90 ([+ or -] 0.30)
25-44 66.68 ([+ or -] 0.25)
45-64 65.21 ([+ or -] 0.26)
65+ 60.32 ([+ or -] 0.46)
Sex
Male *** 69.50 ([+ or -] 0.18)
Female 65.80 ([+ or -] 0.26)
Total 68.28 ([+ or -] 0.15)
([dagger]) The QF estimates were adjusted with revised Q and F values
estimated from typical BSY responses for each age and gender
group.
([double dagger]) QF estimates were further adjusted with Q uplift
factors calculated for each age and gender group based on typical
beverage preferences and under-reporting of beverage- specific
consumption by BSY method compared with sales data.
([section]) Underage drinking varied by province = 15-17 years in
Alberta, Manitoba and Quebec, 15-18 elsewhere in Canada.
Wald chi-square test or z-test: *** p < 0.001.
Table 4. Weighted estimates of the percentage of drinks consumed
on risky drinking days by types of beverage for short-term harm
regarding Canada's Low-Risk Alcohol Drinking Guidelines in five
age groups, men and women in Canada, 2008-2010
Age and sex % Risky drinkers
([+ or -] 95% CI)
([dagger])
Based on the
yesterday
drinks
Age, years
([section])
15-17/18 34.47 ([+ or -] 4.97)
18/19-24 *** 41.44 ([+ or -] 3.44)
25-44 17.65 ([+ or -] 1.35)
45-64 10.08 ([+ or -] 0.99)
65+ 4.99 ([+ or -] 1.03)
Sex
Male *** 18.79 ([+ or -] 1.23)
Female 14.46 ([+ or -] 1.05)
Total 16.66 ([+ or -] 0.61)
Uplifted with Q
to the level
of sales
Age, years
([section])
15-17/18 45.00 ([+ or -] 5.27)
18/19-24 *** 59.86 ([+ or -] 3.40)
25-44 39.11 ([+ or -] 1.85)
45-64 35.10 ([+ or -] 1.55)
65+ 27.01 ([+ or -] 2.11)
Sex
Male *** 40.14 ([+ or -] 1.54)
Female 36.97 ([+ or -] 1.35)
Total 38.58 ([+ or -] 1.02)
% Risky drinks ([+ or -] 95% CI) ([dagger])
Age and sex Total Beer
Based on the
yesterday
drinks
Age, years
([section])
15-17/18 * 74.47 ([+ or -] 2.74) *** 65.60 ([+ or -] 4.60)
18/19-24 *** 80.05 ([+ or -] 1.22) *** 81.45 ([+ or -] 1.47)
25-44 55.29 ([+ or -] 0.97) 54.85 ([+ or -] 1.28)
45-64 45.74 ([+ or -] 1.00) 49.65 ([+ or -] 1.50)
65+ 28.03 ([+ or -] 1.30) 33.16 ([+ or -] 2.80)
Sex
Male *** 55.30 ([+ or -] 0.70) * 57.63 ([+ or -] 0.90)
Female 48.66 ([+ or -] 1.08) 54.01 ([+ or -] 1.91)
Total 53.33 ([+ or -] 0.59) 56.97 ([+ or -] 0.81)
Uplifted with Q
to the level
of sales
Age, years
([section])
15-17/18 *** 92.38 ([+ or -] 1.21) *** 92.02 ([+ or -] 1.83)
18/19-24 *** 91.27 ([+ or -] 0.60) *** 90.67 ([+ or -] 0.77)
25-44 82.14 ([+ or -] 0.52) 83.80 ([+ or -] 0.66)
45-64 78.02 ([+ or -] 0.58) 79.92 ([+ or -] 0.84)
65+ 65.56 ([+ or -] 0.96) 68.65 ([+ or -] 1.93)
Sex
Male *** 82.22 ([+ or -] 0.37) ** 83.39 ([+ or -] 1.62)
Female 76.62 ([+ or -] 0.64) 81.36 ([+ or -] 1.05)
Total 80.54 ([+ or -] 0.32) 83.02 ([+ or -] 0.43)
% Risky drinks ([+ or -] 95% CI) ([dagger])
Age and sex Wine Spirit
Based on the
yesterday
drinks
Age, years
([section])
15-17/18 6.15 ([+ or -] 8.96) *** 90.42 ([+ or -] 2.67)
18/19-24 *** 62.26 ([+ or -] 5.58) *** 81.20 ([+ or -] 2.26)
25-44 32.75 ([+ or -] 2.15) 73.43 ([+ or -] 1.76)
45-64 26.66 ([+ or -] 1.61) 62.36 ([+ or -] 1.97)
65+ 13.69 ([+ or -] 2.01) 39.19 ([+ or -] 3.12)
Sex
Male 25.95 ([+ or -] 1.58) 67.79 ([+ or -] 1.32)
Female *** 29.38 ([+ or -] 1.58) 67.82 ([+ or -] 1.84)
Total 27.73 ([+ or -] 1.12) 67.80 ([+ or -] 1.07)
Uplifted with Q
to the level
of sales
Age, years
([section])
15-17/18 16.85 ([+ or -] 12.97) *** 95.20 ([+ or -] 1.34)
18/19-24 *** 79.06 ([+ or -] 3.28) *** 95.66 ([+ or -] 0.81)
25-44 62.19 ([+ or -] 1.56) 92.86 ([+ or -] 0.71)
45-64 63.46 ([+ or -] 1.23) 92.25 ([+ or -] 0.75)
65+ 46.68 ([+ or -] 2.05) 83.94 ([+ or -] 1.62)
Sex
Male *** 61.58 ([+ or -] 1.23) 91.79 ([+ or -] 0.53)
Female 59.76 ([+ or -] 1.19) * 92.73 ([+ or -] 0.70)
Total 60.64 ([+ or -] 0.86) 92.11 ([+ or -] 0.43)
([dagger]) Risky drinking for acute harms consisted for men of
having 5 or more drinks and for women 4 or more drinks in a
typical day in past year (QF approach).
([double dagger]) Risky drinks overall and by beverages for acute
harms consisted of the alcohol consumed "yesterday".
Self-reported consumption yesterday was uplifted to the level of
recorded alcohol sales in Canada (8.2 L in 2008, 2009 and 2010).
([section]) Underage drinking varied by province = 15-17 years
old in Alberta, Manitoba and Quebec, 15-18 elsewhere in Canada.
Wald chi-square test or z-test: * p < 0.05; ** p < 0.01;
*** p < 0.001.
