Review Paper
Prevalence Studies of Substance-Related Disorders:
A Systematic Review of the Literature
Julian M Somers, MSc, PhD1, Elliot M Goldner, MD, MHSc2, Paul Waraich, MD, MHSc1, Lorena Hsu, MSc3
| |
This is the third in a series of papers that present systematic reviews of the prevalence and incidence of psychiatric disorders drawn from studies published between the years January 1, 1980, and December 31, 2000. The series discusses the implications of these epidemiologic findings for mental health policy and practice.
Objective: To present the results of a systematic review of literature published between January 1, 1980, and December 31, 2000, that reports epidemiologic estimates of substance-related disorders.
Method: We conducted a literature search of substance-related epidemiologic studies, using medline and HealthSTAR databases and applying a set of predetermined inclusion and exclusion criteria to identify relevant studies. We extracted and analyzed prevalence and incidence data for heterogeneity.
Results: A total of 19 prevalence studies of substance-related disorders met inclusion criteria for this review. Heterogeneity analyses revealed significant variability across 1-year and lifetime prevalence of both alcohol and other substance use disorders. The corresponding 1-year and lifetime pooled rates were 6.6 per 100 and 13.2 per 100, respectively, for alcohol use disorders and 2.4 per 100 and 2.4 per 100, respectively, for other substance use disorders. We observed variability among countries and also among regions within the same country. In contrast to other drug problems, alcohol use disorders were substantially more common, were more likely to occur among male subjects, and were more likely to be associated with abuse symptoms. For other drugs, dependence was consistently more prevalent than abuse.
Conclusions: Studies using rigorous and comparable methodologies report significant variability in rates of alcohol and other substance use disorders. These data suggest that different policies and regional practices are associated with variability in rates of disorders. Policy-makers and health planners require regular, regionally sensitive estimates of prevalence rates to respond effectively to unique patterns of need in their constituencies.
(Can J Psychiatry 2004;49:373-384)
Click here for author affiliations.
|
Clinical Implications
-
Prevalence rates for substance use disorders appear to vary markedly among samples drawn from different countries or regions.
-
Population-level factors are implicated as playing an important role in the prevalence and pattern of different substance use disorders within large samples.
-
Informed health planning requires the undertaking of planned comparative studies among regions, as well as updated 1-year prevalence estimates and estimates of the incidence of substance use disorders.
Limitations
-
Subtle differences among studies may have contributed an unknown degree of variance to the observed rates.
-
Cultural factors may influence the validity of self-reported substance use information in a manner that biases these data.
-
We did not evaluate several factors that could influence prevalence rates, including the legal status of substances, cost and availability of drugs, mortality associated with substance use, systematic underreporting, prevention programs, social norms, economic conditions, and psychiatric comorbidity.
|
Key Words: substance-related disorders, alcohol abuse,
alcohol dependence, drug abuse, drug dependence,
epidemiology, prevalence
Résumé : Études de prévalence des troubles liés à une substance : une analyse méthodique de la documentation
|
The worldwide use of alcohol and other drugs is associated with a substantial burden of illness (1). Historically, trends in substance use and associated harms vary considerably among regions and across time. Various environmental factors have been identified as mediators of these differences, including economic conditions, cultural traditions, historical practices, social policies, and access to drugs. Individual differences are also operative in conferring risk of harm or in providing protection against the development of substance-related problems. The findings of contemporary epidemiologic investigations may help to determine the relevance of different factors within a biopsychosocial framework and may help to identify opportunities to diminish the burden of illness associated with substance use. In turn, this information could assist in guiding health policy decisions and could add constructively to the knowledge base of clinicians and researchers. Differences in prevalence rates across geographic regions or populations may help to identify variables that contribute to or protect against the development of substance use disorders. Period prevalence rates (for example, 1-year prevalence) may assist planners in allocating and distributing resources according to the needs of a cohort and in deploying resources in a targeted, rather than inclusive, manner. Incidence rates can be interpreted in association with local risk and protective factors and can provide a means of assessing the strength of relations between substance use problems and their putative causes. In theory, one can envision maps detailing the relative role of environmental factors in the formation of substance use problems. These maps may be studied in concert with individual gene maps to clarify patterns of interaction that result in substance abuse or substance dependence among particular individuals (2).
This review integrates the findings of high-quality epidemiologic studies that used comparable methodologies to estimate the prevalence and incidence of substance use disorders in different jurisdictions. We interpreted results as they concern the above-noted areas of ongoing investigation, with the objective of achieving evidence-based health planning.
Methods
In the first article of this series, we present a detailed description of the methods used in this review (3). In brief, we searched the Medline and HealthSTAR databases for relevant epidemiologic studies, using the key indexing terms epidemiology, prevalence, and incidence, which we exploded and combined with the terms mental disorders, substance-related disorders, alcohol abuse or dependence, and drug abuse or dependence. We limited the search to English-language studies published between January 1, 1980, and December 31, 2000. We also searched reference lists of all relevant primary and review articles identified.
We selected only community surveys using probability sampling techniques and having sample sizes of 450 or more. Studies were eligible for inclusion if they examined age ranges covering the adult population. We included only those using recognized diagnostic criteria and case identification based either on standardized instruments or on clinician diagnosis. We extracted prevalence data including overall sex-specific and age-specific rates from eligible studies.
To summarize the estimated rates and to elucidate any observed differences among rates, we conducted qualitative analyses of variables related to study population, sample characteristics, and diagnostic assessment. As well, we pooled each set of rates according to a Bayesian approach to metaanalysis, using the Fastpro software program (version 1.7) by Eddy and Hasselblad. Readers interested in a more detailed discussion of this approach should refer to Eddy and others (4). We calculated the pooled or best-estimate of effect values using Jeffrey’s prior (4) and a hierarchical model. We then analyzed each of the pooled rates for heterogeneity, using chi-square tests according to Fleiss’ method (5). For pooled rates demonstrating heterogeneity, we employed a systematic method to explore variables of interest (for example, geographic location) that might have contributed to the variation in rates. We then pooled rates from individual studies according to the variables found to be contributing to heterogeneity.
Results
Description of Studies
The initial electronic search identified 30 prevalence studies and 4 incidence studies potentially meeting inclusion criteria, in addition to 8 review papers. In reviewing reference lists, we identified an additional 9 potential prevalence studies (but no incidence studies), for which we obtained full-text articles.
Of the 39 prevalence studies for which we reviewed full-text articles, a total of 15 studies were excluded: 7 studies did not meet eligibility criteria and 8 presented duplicate data. Thus, 24 prevalence studies of substance-related disorders met inclusion criteria (6–29), resulting in a total of 19 unique primary investigations of substance use disorders. We present the 7 studies not fulfilling inclusion criteria and their reasons for exclusion in Table 1 (30–36). Of the 4 potential incidence studies, 2 did not meet full eligibility criteria (37–39). We document their reasons for exclusion in Table 1. We excluded incidence data from the review because we could only identify 2 appropriate incidence studies (38,40).
|
Table 1 Substance-related disorder studies excluded at article review
stage
|
|
Study reference
|
Reason for exclusion
|
Comments
|
|
Prevalence studies
|
|
|
|
Weisner and others (33)
|
Unclear whether case definition meets criteria for substance-related disorder
|
Assesses problem drinkers and drug users
|
|
Bucholz and others (34)
|
Prevalence data not reported
|
Data taken from ECA study
|
|
Fournier and Kovess (36)
|
Unclear whether diagnostic criteria used
|
Comparison of mail and telephone interview
|
|
Hughes (35)
|
Case definition does not meet criteria for substance-related disorder
|
Assesses illicit drug use only
|
|
Stefansson and others (32)
|
Does not include entire adult age group
|
Presents rates applicable to subjects aged 55 to 57 years
|
|
Surtees and Sashidharan (31)
|
Limited to population of women
|
Comparison of 2 community samples
|
|
Dilling and Weyerer (30)
|
Uses ICD-8 diagnostic criteria
|
Community sample
|
|
Incidence studies
|
|
|
|
Sejda and others (39)
|
Assesses treated population
|
Those who sought treatment for first time at health care or other facility
dealing with drug addiction
|
|
Gfroerer and Brodsky (37)
|
Case definition does not meet criteria
|
Assesses illicit drug use only
|
|
ECA = Epidemiologic Catchment Area
|
Prevalence Rates
We present findings for the 17 studies reporting 1-year and (or) lifetime prevalence rates only (Tables 2 to 4) because relatively fewer studies reported data for 6-month prevalence (7,9,15,22,24,28) and point prevalence (24,27,28). Also, we present findings separately for alcohol abuse or dependence, alcohol abuse, alcohol dependence, drug abuse or dependence, drug abuse, and drug dependence because these were the diagnostic categories for which prevalence rates were most commonly reported. We carried out data analyses only when 3 or more rates were reported.
|
Table 2 One-year prevalence rates of substance-related disorders
|
|
|
|
|
|
|
Prevalence rate (per 100 persons)
|
|
Study details
|
Study site
|
n
|
Response
rate (%)
|
Case-finding method
|
Alcohol abuse and (or)
dependence
|
Drug abuse
and (or)
dependence
|
Any substance abuse and (or) dependence
|
|
Hall and others (20)
|
Australia - national
|
10 641
|
78.0
|
Census; CIDI/ICD-10; lay interviewers; method of diagnosis unclear
|
6.5a
3.0b
3.5c
|
2.2a
0.2b
2.0c
|
—
|
|
Bijl and others (6)
|
Netherlands - national
|
7 146
|
69.7
|
Census; CIDI/DSM-III-R; lay interviewers, algorithm diagnosis
|
—a
4.6b
3.7c
|
—a
0.5b
0.8c
|
—
|
|
Farrell and others (41)
|
Great Britain - national
|
10 108
|
79.4
|
DIS, US National Alcohol Survey; lay interviewers
|
—a
—b
5.0c
|
—a
—b
2.0c
|
—
|
|
Grant (17)
|
US - national
|
42 862
|
90.0
|
Census; AUDADIS/DSM-IV; type of interviewers not clear
|
—a
—b
—c
|
—a
—b
0.48c
|
—
|
|
Offord and others (13)
|
Ontario
|
8 116
|
67.4
|
Census; UM-CIDI/DSM-III-R; lay interviewers; algorithm diagnosis
|
4.4a
—b
—c
|
—a
—b
—c
|
5.2
|
|
Ross (19)
|
Ontario
|
8 116
|
67.4
|
Census; UM-CIDI/DSM-III-R; lay interviewers; algorithm diagnosis
|
—d
1.5b
2.9c
|
—a
—b
—c
|
—
|
|
Kessler and others (11)
|
US (NCS) - national
|
8 098
|
82.6
|
Census; UM-CIDI/DSM-III-R; clinical reinterview and diagnosis
|
—a
2.5b
7.2c
|
—a
0.8b
2.8c
|
11.3
|
|
Bourdon and others (22)
|
US (ECA) - 5 sites, mainly urban
|
20 291
|
68.0–
80.0
|
Census; DIS/DSM-III; lay interviewers; algorithm diagnosis
|
—a
—b
—c
|
—a
—b
—c
|
7.5
|
|
Hwu and others (10)
|
Taiwan
|
11 004
|
95.0
|
Census; DIS-CM/DSM-III; lay interviewers; method of diagnosis unclear
|
—
|
—
|
—
|
|
|
Metropolitan
Taipei
|
5 005
|
|
|
—a
1.3b
0.66c
|
—a
0.02b
0.04c
|
—
|
|
|
Small towns
|
3 004
|
|
|
—a
4.7b
1.1c
|
—a
0.0b
0.13c
|
—
|
|
|
Rural
villages
|
2 995
|
|
|
—a
3.4b
0.6c
|
—a
0.0b
0.0c
|
—
|
|
Oakley- Browne and others (28)
|
New Zealand - area of Christchurch, mostly urban
|
1 498
|
70.0
|
Census; DIS/DSM-III; lay interviewers; algorithm diagnosis
|
9.3a
—b
—c
|
2.3a
—b
—c
|
10.5
|
|
Bland and others (24)
|
Edmonton
metropolitan
|
3 258
|
71.6
|
Census; DIS/DSM-III; lay interviewers; algorithm diagnosis
|
7.9a
—b
—c
|
2.6a
—b
—c
|
9.1
|
|
|
|
|
|
Best-estimate (95%CI)
|
6.6 (5.1–8.3)a
|
2.4 (2.2–2.6)a
|
8.4 (6.4–10.9)
|
|
|
|
|
|
|
2.8 (2.0–3.8)b
|
0.18(0.05–0.42)b
|
|
|
|
|
|
|
|
2.4 (1.3–3.8)c
|
0.58 (0.23–1.2)c
|
|
|
aTotal; bAbuse; cDependence; dExisting follow-up publication based on
same data; — = not reported
AUDADIS = Alcohol Use Disorders and Associated Disabilities Interview;
CIDI = Composite International Diagnostic Interview; DIS = diagnostic
interview schedule; NCS = National Comorbidity Survey; UM-CIDI = University
of Michigan Composite International Diagnostic Interview
|
|
Table 3 Lifetime prevalence rates of substance-related disorders
|
|
|
|
|
|
|
Prevalence rate (per 100 persons)
|
|
Study details
|
Study site
|
n
|
Response rate (%)
|
Case-finding method
|
Alcohol abuse and (or)
dependence
|
Drug abuse
and (or)
dependence
|
Any substance abuse and (or) dependence
|
|
Bijl and others (6)
|
Netherlands - national
|
7 146
|
69.7
|
Census; CIDI/DSM-III-R; lay interviewers, algorithm diagnosis
|
—a
11.7b
5.5c
|
—
1.5
1.8
|
—
|
|
Fournier and others (26)
|
Montreal-city
|
893
|
63.6
|
Telephone survey; CIDIS/DSM-III-R; lay interviewers; algorithm diagnosis
|
11.5
—
—
|
5.4
—
—
|
14.1
|
|
Grant (17)
|
US - national
|
42 862
|
90.0
|
Census; AUDADIS/DSM-IV; type of interviewer not clear
|
—a
—b
—c
|
—a
—b
2.91c
|
—
|
|
Ross (19)
|
Ontario
|
8 116
|
67.4
|
Census; UM-CIDI/DSM-III-R; lay interviewers; algorithm diagnosis
|
12.0a
6.1b
5.9c
|
—a
—b
—c
|
—
|
|
Kessler and others (11)
|
US (NCS) - national
|
8 098
|
82.6
|
Census; UM-CIDI/DSM-III-R; clinical reinterview and diagnosis
|
—
9.4
14.1
|
—
4.4
7.5
|
26.6
|
|
Canino and others (21)
|
Puerto Rico - entire island-nation
|
912
|
92.9
|
Census; DIS/DSM-III; lay interviewers; algorithm diagnosis
|
—a
—b
—c
|
1.2a
—b
—c
|
—
|
|
Chen and others (8)
|
Hong Kong - national
|
7 229
|
77.8
|
Census; DIS-III-CM/DSM-III; lay interviewers; algorithm diagnosis
|
4.5d
3.0d
1.6d,e
|
0.24d
—
—
|
—
|
|
Wittchen and others (15)
|
Germany - former West Germany
|
483
|
73.5
|
Census; DIS/DSM-III and ICD-9; clinical interview and diagnosis
|
13.0a
—b
—c
|
1.2a
—b
—c
|
13.5
|
|
Bourdon and others (22)
|
US (ECA) - 5 sites, mainly urban
|
20 291
|
68.0–
80.0
|
Census; DIS/DSM-III; lay interviewers; algorithm diagnosis
|
13.5a
—b
—c
|
6.1a
—b
—c
|
16.7
|
|
Wells and others (14)
|
New Zealand - area of Christchurch, mostly urban
|
1 498
|
70.0
|
Census; DIS/DSM-III; lay interviewers; algorithm diagnosis
|
18.9a
—b
—c
|
5.7a
—b
—c
|
21.0
|
|
Hwu and others (10)
|
Taiwan
|
11 004
|
75.0
|
Census; DIS-CM/DSM-III; lay interviewers; method of diagnosis unclear
|
—
|
—
|
—
|
|
|
Metropolitan
Taipei
|
5 005
|
|
|
—a
3.4b
1.5c
|
—a
0.02b
0.08c
|
—
|
|
|
Small towns
|
3 004
|
|
|
—a
8.0b
1.8c
|
—a
0.0b
0.2c
|
—
|
|
|
Rural
villages
|
2 995
|
|
|
—a
6.3b
1.2c
|
—a
0.0b
0.0c
|
—
|
|
Bland and others (25)
|
Metropolitan Edmonton
|
3 258
|
71.6
|
Census; DIS/DSM-III; lay interviewers; algorithm diagnosis
|
18.0a
—b
—c
|
6.9a
—b
—c
|
20.6
|
|
Lee and others (12)
|
Korea - Dong, Seoul (urban) and Eub, Myeon (rural)
|
5 100
|
81.8
|
Census; DIS/DSM-III; lay interviewers; algorithm diagnosis
|
22.0a*
12.1b
9.9c
|
0.74a
—b
—c
|
31.8
|
|
Canino and others (7)
|
Puerto Rico - entire island-nation
|
1 513
|
91.0
|
Census; DIS/DSM-III; lay interviewers; algorithm diagnosis
|
12.6a
4.4b
8.2c,e
|
—a
—b
—c
|
—
|
|
|
|
|
|
Best-estimate (95%CI)
|
13.2 (9.7–17.2)a
|
2.4 (0.95–4.2)a
|
20.2 (15.8–25.1)
|
|
|
|
|
|
|
6.5 (4.8–8.8)b
|
0.29 (0.05–0.93)b
|
|
|
|
|
|
|
|
4.0 (2.3–6.4)c
|
0.76 (0.20–1.9)c
|
|
|
aTotal; bAbuse; cDependence; dOverall rate calculated from raw data (only sex- and age-specific rates reported); eSum of 2, categories alcohol dependence and alcohol abuse and dependence; *Prevalence figure differs from that reported in Lee and others(12) as we were informed by the authors of an
error in the reported estimate; — = not reported
CIDIS = Composite International Diagnostic Interview Schedule; DIS-CM
= Diagnostic Interview Schedule–Composite Manual
|
|
Table 4 Sex-specific 1-year and lifetime prevalence rates of substance-related
disorders
|
|
|
|
Prevalence rate (per 100 persons)
|
|
|
|
Alcohol abuse or dependence
|
Alcohol abuse
|
Alcohol
dependence
|
Drug abuse or dependence
|
Drug abuse
|
Drug
dependence
|
|
Study details
|
Study site
|
M
|
F
|
M
|
F
|
M
|
F
|
M
|
F
|
M
|
F
|
M
|
F
|
|
1-year prevalence
|
|
Hall and others (20)
|
Australia
|
9.4
|
3.7
|
4.3
|
1.8
|
5.2
|
1.8
|
3.2
|
1.3
|
0.3
|
0.1
|
2.9
|
1.2
|
|
Bijl and others (6)
|
Netherlands
|
—
|
—
|
7.3
|
1.8
|
6.1
|
1.1
|
—
|
—
|
0.6
|
0.3
|
1.0
|
0.7
|
|
Jenkins and others (6)
|
UK
|
—
|
—
|
—
|
—
|
7.5
|
2.1
|
—
|
—
|
—
|
—
|
2.9
|
1.5
|
|
Grant (17)
|
US
|
—
|
—
|
—
|
—
|
—
|
—
|
—
|
—
|
—
|
—
|
0.61
|
0.35
|
|
Ross (19)
|
Ontario
| |
