|
| |
Alcohol Abuse and Dependence in the General Population
From: Columbia University
| By:
Deborah Hasin |
EDITOR'S INTRODUCTION |
People have been drinking ever since the fermentation of alcohol was discovered, thousands of years ago. Drinking has spawned powerful, if ambivalent, cultural reactions, as well as widespread health consequences that are often not fully appreciated. Physicians can easily identify severe alcoholism, but far too little is known about the impact and prevalence of alcohol abuse and dependence in the population as a whole. 
Deborah Hasin, professor of clinical public health at the Mailman School of Public Health at Columbia University, seeks to better define--and thereby understand and treat--alcohol disorders. Hasin's ongoing study of heavy drinkers in Essex County, New Jersey, has already provided new insights into the role of alcohol in America, including the identification of alcohol dependence and abuse as independent disorders (without the commonly assumed stepping-stone relationship of alcohol abuse to alcohol dependence). |
 iagnosing alcoholism is relatively easy when considering people with severe problems in a treatment setting. These people come to treatment with disability in every area of their social and occupational functioning. Their physicians are never in doubt about their alcoholism diagnosis, because it is so obvious. But can physicians accurately diagnose individuals in the general population when a diagnosis of alcohol dependence can be made in the presence of only three diagnostic criteria? Are we really diagnosing a valid disorder? This question is important, because very few people diagnosed with alcohol dependence or alcohol abuse ever receive treatment. In fact, less than 25 percent enter some kind of a treatment facility. |
In the early 1980s, the prevalence of alcoholism in the general population was thought to fairly closely follow the distribution of heavy drinking. I felt it was important to determine whether alcoholism in the general population represented a real disorder or whether the reported rates of alcoholism were, rather, just a manifestation of heavy drinking among young people, who then matured out of it as they assumed normative adult roles. If the latter is true, it is not accurate to call such cases of "alcoholism" a psychiatric condition. Many major areas of research in the alcohol field depend on the answers to these questions--including studies of the genetics, etiology and chronicity of alcohol abuse and dependence. |
The ancient culture of alcohol
Alcohol and drinking have been with people for thousands of years, as exemplified by a Sumerian recipe for beer found by archaeologists. Maps have been found showing ancient Middle East wine-producing centers that existed as far back as 1000 BCE. Much of the production of alcohol then centered on grapes and wine. Today, when we talk about alcohol we use the consumption categories of beer, wine and hard liquor. |
Drinking alcohol provides a source of pleasure for many people, as it has for thousands of years. Alcohol is involved in many cultural interactions--courtship, celebrations and special occasions of all kinds. Alcohol also has strong commercial aspects; revenues from producing and taxing alcoholic beverages play an important part in its societal role. |
International comparisons of alcohol consumption
Per-capita rates of alcohol consumption vary by culture and nation. Luxembourg stands first in a 1994 ranking of 51 countries in terms of per-capita alcohol consumption worldwide. Northern European countries also rank high on this list, along with some Eastern European nations and the former Soviet Union. Middle Eastern countries are at the bottom of the list. The United States ranks almost exactly in the middle at 25. Because it is composed of many groups from different countries, the US population does not have a homogeneous pattern of drinking. |
Alcohol in the United States
Alcohol obviously has a downside and has been viewed with some ambivalence in America. National and local legislation has reflected this, resulting in numerous changes in the legal status of alcohol in this country.  Further, there have been marked shifts in levels of per-capita alcohol consumption over time. A pair of engravings of George Washington reflects the changes in attitudes and habits. An 1848 engraving depicts Washington celebrating a happy occasion with his men. In his hand is a glass and on the table is a bottle containing alcohol. After the initial publication of this engraving, the temperance movement (a movement strongly advocating complete abstinence from alcohol) began to gain considerable strength. The movement stirred up popular feelings against the drinking of any alcohol, advocating pledges of abstinence from its members. |
The second engraving, produced in the 1860s during the temperance movement, was altered. Someone removed the glass from Washington's hand and the bottle from the table. In the modified picture, Washington stands among his men with his hand in the same position but empty, looking a bit awkward. On the table, a hat appears in place of the bottle of alcohol. What was once seen as acceptable drinking behavior had become questionable because of changes in public sentiment, and so the picture of our first president was altered accordingly. |
Time trends in US alcohol consumption
 It is possible to see the two images of George Washington as a visual representation of more quantitative trends in per-capita alcohol consumption in the United States. In the early days of this country, people (including children) drank a surprising amount of alcohol. The consumption rate decreased sharply with the start of the temperance movement and reached its low point during Prohibition. |
Stereotypes from movies and other sources about the illegal production of alcohol and the existence of speakeasies during Prohibition have caused many people to assume that drinking did not decrease during that period. Widespread opposition to Prohibition and continued drinking in America did eventually lead to the repeal of Prohibition. Nevertheless, there was a very sharp decrease in per-capita alcohol consumption during the Prohibition years, and this is notable from a public health standpoint. |
After Prohibition, per-capita consumption showed a more or less steady gradual increase until the beginning of the 1980s, when alcohol rates started to fall. That decrease remained fairly steady until the last few years, when some evidence emerged that it is now tapering out to a flatter curve. |
The figures concerning time trends in alcohol consumption are based on aggregated records of alcohol sales. Like most data sources of this type, various factors exist that may detract from the accuracy of the information. To confirm findings in such a situation, epidemiologists attempt to identify different sources of data on the same phenomenon. If markedly different types of data on a given phenomenon lead to the same conclusions, then the validity of the findings of all data sources is supported. |
Yearly death rates from alcohol-related liver cirrhosis provide such an alternative source of data when investigating large-scale changes in time in per-capita alcohol consumption.  These rates follow a more or less similar pattern to per-capita alcohol consumption. Sharp decreases in the death rates typically occur during the same periods as low per-capita consumption rates. Also in parallel to the per-capita alcohol consumption rates, death rates from alcohol-related liver cirrhosis generally increased until the late 1970s and have declined since. Note that similar trends are found when population subgroups are examined separately. |
Societal costs of alcohol abuse in the US
The costs associated with alcohol abuse in the United States are considerable, not just in terms of health care but also in terms of productivity. Given the increased funding allocated for drug problems versus alcohol problems in this country, it is important to note that the costs associated with alcohol problems are larger than those associated with drug problems across all categories except crime. |
Defining alcohol abuse and dependence
Specific criteria for diagnosing mental disorders, including alcohol use disorders, are found in the Diagnostic and Statistical Manual of Mental Disorders (DSM), published by the American Psychiatric Association. These criteria provide a basis for making research and clinical diagnoses in a systematic, replicable way. The American Psychiatric Association revises and updates these criteria periodically to reflect new findings from clinical, biological and epidemiological research. In the version published in 1987, DSM-III-R, the criteria for diagnosing alcohol use disorders changed from previous editions. Additional smaller changes were made in the most recent edition, DSM-IV, published in 1994. These criteria are the ones used today. |
DSM-IV establishes seven criteria for alcohol dependence:
- Tolerance
- Withdrawal, or drinking to relieve or avoid withdrawal
- Alcohol taken in larger amounts or more often than intended
- Persistent desire or unsuccessful attempts to cut down or control drinking
- A great deal of time spent getting alcohol, drinking or getting over its effects
- Important social, occupational or recreational activities given up or reduced in order to drink
- Continued drinking despite persistent physical or psychological problems.
|
To receive a diagnosis of DSM-IV alcohol dependence, three or more of these criteria must be met. |
Some debate remains about aspects of these criteria. For example, what is the appropriate role of tolerance and withdrawal? Should they be required in order to get the diagnosis? Do they characterize an important subtype? Can the diagnosis of alcohol dependence be made in an individual who does not manifest these phenomena? |
DSM-IV alcohol abuse is based on four criteria:
- Recurrent drinking that results in a failure to fulfill major role obligations at work, school or home
- Recurrent drinking in physically hazardous situations (e.g., driving)
- Recurrent alcohol-related legal problems
- Continued drinking despite persistent social or interpersonal problems.
|
To receive a diagnosis of DSM-IV alcohol abuse, one or more of these criteria must be met in the absence of DSM-IV alcohol dependence. Because diagnosis of abuse was very rare according to similar criteria in DSM-III-R, an effort was made to broaden the criteria for abuse in DSM-IV and narrow the net for dependence somewhat by raising the threshold needed to receive a dependence diagnosis. |
Prevalence of alcohol abuse and dependence in the US
Data from three major national surveys provide basic information about the prevalence of current and lifetime alcohol use disorders in the United States. The Epidemiologic Catchment Area (ECA) survey conducted in 1980 was the first large-scale psychiatric epidemiologic study in which DSM criteria were used to define cases in the general population. In the ECA, approximately 20,000 Americans were surveyed in five catchment areas in the US. The ECA served as a major source of psychiatric epidemiologic data in the United States for many years. Findings from the ECA indicated that the prevalence for any current alcohol use disorder was 5 percent, while the prevalence of all individuals who had ever had the disorder (lifetime prevalence) was 13.5 percent. While this is quite a high figure, subsequent surveys have found even higher prevalences. |
Using the guidelines established in DSM-III-R, the 1992 National Comorbidity Survey (NCS) survey of around 8,000 Americans found a current prevalence of 9.7 percent and a lifetime prevalence of 23.5 percent. |
The 1992 National Longitudinal Alcohol Epidemiologic Survey (NLAES) study was a national survey of approximately 42,000 Americans. NLAES has been a very important national source of information on alcohol use disorders and associated conditions based on DSM-III-R and DSM-IV criteria. NLAES data indicated that the current (last 12 months) prevalence of alcohol dependence was 7.4 percent and that the lifetime prevalence was 18.2 percent. Had the actual prevalence of the disorder increased since the 1980 ECA survey? Or was the higher prevalence due to the fact that different instruments were used in the later surveys? The answers to these questions remain controversial. |
Essex County study
When I began my research career, essentially no data were available on whether diagnoses of alcohol abuse and dependence were valid when made in untreated community residents. The question is important, because much scientific work on the causes of alcoholism depends on diagnoses made in untreated individuals--for example, family genetic studies of relatives of alcoholics. In addition, policy planning and the delivery of treatment services also may be shaped to some extent by epidemiologic data on service needs. If the DSM criteria improperly identify some individuals as cases who do not actually have the medical condition of alcoholism or alcohol dependence, then research results may be incorrect. I used principles from psychiatry and psychometric theory to investigate the validity of the alcohol use diagnoses in the general population. |
Rather than using the full community survey approach required when the goal of the research is to determine the rates of a disorder in a given area, I investigated the validity questions in a sample of untreated community residents who were screened for higher-than-average drinking. Their drinking placed them at higher-than-average risk for alcohol-use disorders. The study was designed to allow efficient and cost-effective investigation of the validity issues without requiring interviews from thousands of unscreened participants who were uninformative about the research issues because they never drank very much. Random digit dialing was used in Essex County, New Jersey, to select approximately 7,000 households for screening. After screening randomly selected residents of these households for higher-than-average drinking, a pool of 965 participants were identified and interviewed. These individuals all had at least one occasion in the 12 months prior to screening when they consumed five or more drinks. From this original pool, 91 percent were re-interviewed approximately one year later. |
Cross-sectional Essex County validity findings
My group and I first examined the validity issues cross-sectionally after data collection was complete. Cross-sectional research is based on data from one point in time--in this case, the baseline interview for the study. Our initial test involved a study of convergent case identification based on different measures of the alcohol use disorders. Just as the per-capita alcohol consumption figures on time trends in drinking were supported by similar findings in alcohol cirrhosis rates (see above), a measurement or psychiatric diagnosis made by one system is supported if it is also made by several other systems. When this happens, convergence or agreement between systems suggests a common valid concept underlying all the systems. We found excellent agreement among diagnostic systems for case identification of subjects with alcohol dependence. In contrast, agreement across diagnostic systems for alcohol abuse was very poor, suggesting discrepancies between systems on the underlying concept of abuse and the absence of a generalized concept. |
"External criterion variables" were investigated next. I hypothesized that if the diagnostic category of alcohol dependence was valid in this screened community sample, then alcohol-related adverse conditions should be found more often in individuals with a diagnosis of alcohol dependence than among individuals with a diagnosis of abuse or without a diagnosis at all. |
The results consistently confirmed the hypothesis. Individuals in the sample with a dependence diagnosis were significantly more likely to be suicidal than their counterparts with abuse or no diagnosis. Alcohol-dependent subjects were more likely to have been the victims of physical abuse or assault. They were more likely to suffer blackouts--blocks of time during intoxication when activities cannot be remembered even though the drinker was conscious. Subjects in the sample with an alcohol dependence diagnosis were more likely to skip meals, a problem leading to nutritional deficits in some alcoholics. |
In addition, a significantly increased number of people diagnosed with dependence had received some form of alcohol treatment compared with those with a diagnosis of abuse or with no diagnoses. Further, subjects with DSM-IV alcohol dependence were significantly more likely to have a positive family history of alcohol problems (e.g., first-degree relatives with alcohol problems) than subjects with either DSM-IV alcohol abuse or no alcohol disorder. The study of external criterion variables indicated validity for the alcohol dependence diagnosis but not for alcohol abuse, consistent with the findings on convergent case identification described above. |
Longitudinal Essex County validity findings
In studying validity, longitudinal research is used to investigate predictive validity and stability of conditions over time. A question often raised about the relationship between abuse and dependence is whether abuse is simply a prodromal condition to dependence. Many clinicians working with alcoholics feel that most cases of alcoholism start as alcohol abuse. They therefore assume that most cases of alcohol abuse lead to dependence. This question cannot be studied in a clinical sample in which all subjects already have alcohol dependence. It must be studied epidemiologically among individuals who have abuse, in order to determine whether dependence develops or not. In such research, the likelihood of cases of abuse becoming cases of dependence at a later point is studied. An ancillary question is whether cases of dependence are likely to stay true to type (e.g., dependent on alcohol) or if they are likely later to become cases of alcohol abuse. |
The findings from the Essex County study strongly supported the fact that abuse is not simply a prodromal condition to dependence. Subjects in the Essex County study with abuse at baseline were not likely to move into dependence by the one-year follow-up. In contrast, most subjects with abuse at baseline remitted from abuse entirely by their follow-up interview. Further, most subjects with dependence were likely to stay chronic and not become cases of abuse. The findings provide support for the fact that the diagnosis of alcohol abuse is not simply a prodromal condition to dependence, and that dependence and abuse are distinct conditions. |
Note that the Essex County results replicated secondary analyses on the same question that I had undertaken several years before. (In secondary analyses, data collected for a different purpose is used to investigate a question unanticipated initially by the original researchers.) My Essex County results were important because the study was designed specifically to eliminate threats to the validity of the findings that were present in the earlier secondary analyses. |
The Essex County data were also used to investigate the questions mentioned above about the role of the physiological criteria in the diagnosis of alcohol dependence. These include tolerance to alcohol and evidence of alcohol withdrawal symptoms. To investigate the importance of tolerance and withdrawal, the likelihood of remission was compared between alcohol-dependent subjects with tolerance and/or withdrawal and alcohol-dependent subjects without these physiological reactions. The results showed that increased tolerance to alcohol was not associated with chronicity of alcohol dependence. However, having withdrawal as part of the baseline alcohol dependence syndrome did predict chronicity. Further, when tremors were present as part of the alcohol withdrawal syndrome, chronicity was predicted even more strongly. |
The findings suggested that withdrawal symptoms constituted an important component of the DSM-IV alcohol dependence criteria but that the role of tolerance was unclear. These findings are consistent with the results of other researchers. This type of data is part of what the American Psychiatric Association will review when the process of preparing DSM-V is begun. |
Replication of Essex County results
Replication of scientific research results is important to ensure that they are valid and not an epiphenomena (secondary condition) of a specific research situation. It became important to confirm the Essex County findings in national and international data to ensure that the results were generalizable to subjects outside one specific area. We also needed to ensure that the process of selecting subjects by screening for heavier-than-average drinking had not somehow introduced bias or error in the results. For this purpose, my group analyzed data from the 1992 NLAES national survey mentioned above. We also analyzed data from a World Health Organization (WHO) study that I had participated in from the planning stages to publication of the results. |
A replication of the external criterion variables analysis in the NLAES data again indicated that the subjects with DSM-IV alcohol dependence were more likely to have been in treatment than those with an abuse diagnosis or no diagnosis. Further, they were more likely to have been suicidal, to have had blackouts and to have a positive family history of alcohol problems. Thus, the national, unscreened subjects confirmed our findings from the screened Essex County sample. |
Using NLAES data, the differentiation between abuse and dependence was again addressed. The results indicated that most subjects with a current diagnosis of dependence only were unlikely to have previously met criteria for abuse only. Subjects who currently meet the criteria for both dependence and abuse were not likely to have previously met criteria for abuse only. Those subjects likely to have met abuse criteria formerly were the ones who still met it currently. Generally, subjects stayed relatively true to type in their present and past diagnoses. These cross-sectional results replicated the longitudinal findings from the Essex County community study. |
Data from the WHO study was used to replicate the convergent case identification study described above. These data were obtained from 11 research centers around the world. Each center contributed about 150 subjects, including both patients and untreated community residents. The results confirmed the findings from the Essex County study. Cross-system agreement on diagnoses of alcohol dependence was generally excellent. Cross-system agreement on cases of alcohol abuse was generally only fair or poor. The results supported the validity of the DSM-IV and International Classification of Diseases, Tenth Revision (ICD-10) diagnosis of alcohol dependence but not the validity of diagnoses of alcohol abuse. |
Reliability as a precondition for validity
In psychometric terms, reliability is considered a precondition for validity. If a condition cannot be recognized and diagnosed consistently from rater to rater (e.g., from physician to physician, or from research interviewer to research interviewer), then all further research on the condition is impaired. My group has conducted test-retest studies of the reliability of the alcohol dependence and alcohol abuse diagnoses in several different samples, including patient and general population samples. I have also worked with others, including the WHO study, on this same issue. |
Generally, data from these studies show that the reliability of dependence is very high. However, the test-retest reliability for the alcohol abuse diagnosis when diagnosed according to DSM-IV rules (e.g., the diagnosis is not made when dependence is diagnosed) is usually only fair to poor. This finding suggests a problem with the abuse category. However, when abuse is diagnosed independently of dependence (e.g., regardless of whether dependence is present or not), its reliability improves considerably. Therefore, the poor reliability of the alcohol abuse diagnosis appears to result at least in part from the DSM-IV rule that alcohol abuse is diagnosed conditionally on alcohol dependence--preventing a diagnosis of DSM-IV alcohol abuse if a person has ever had dependence, for example. |
Reconsidering abuse
Should alcohol dependence and alcohol abuse be considered valid medical/psychiatric diagnoses when diagnosed in the general population? Taken in sum, I believe that the diagnosis of alcohol dependence as currently found in DSM-IV and the international ICD-10 system is valid when made in community or general population subjects. However, the results indicate that continued thought needs to be given to the diagnosis of alcohol abuse and the relationship of abuse to dependence in the DSM system. |
Closer inspection of the grounds for a diagnosis of alcohol abuse both in the Essex County and NLAES samples showed that about half the subjects received the abuse diagnosis from meeting one criterion only: "hazardous use." In almost half of these cases with only hazardous use, the criterion was met based on a few instances of driving after drinking too much. The lack of wisdom in such behavior is not in question. However, the grounds for making a psychiatric diagnosis based only on this behavior can be questioned. Perhaps raising the threshold for an abuse diagnosis so that such cases will be excluded will improve its reliability and validity. It is also possible that removing the DSM-IV and ICD-10 hierarchical condition for diagnosing abuse only in the absence of dependence will improve its reliability and validity. Answering these questions will take further research in my group as well as collaboration with other researchers and independent work from other groups. |
|
| |
