Estimates of the prevalence of anxiolytic and sedative use are based on national and cross-national surveys of use in the general population. Multi-national surveys have shown a wide variation in past-year prevalence of use, ranging from 7.4% in The Netherlands to 17.6% in Belgium and 12.9% in the United States. [1,2] In most studies, the elderly and women are more likely to use anxiolytics, [1-5] but once men become users, they are as likely as women to become long-term users. [6,7] Predictably, rates of anxiolytic and sedative use are higher among psychiatric outpatients (ranging from 17 to 37%). [8,9] Although past-year prevalence of benzodiazepine use is high, most studies show that these drugs are generally being used appropriately,  and some authors suggest that they are used conservatively for the appropriate treatment of the approved indications of these medications, such as anxiety and insomnia. [11,12] The abuse of high doses of benzodiazepines for recreational purposes and without a therapeutic indication is rare and almost invariably occurs in the context of a polydrug substance use disorder. [10,13]
Physical dependence is manifested by a withdrawal syndrome that describes biochemical, physiologic, and/or behavioral changes that occur upon discontinuation of chronic drug therapy. However, it is only one of the nine DSM-III-R criteria used for the diagnosis of psychoactive substance use disorder. At least three of nine criteria must be met to classify a subject as dependent, and the duration of the drug problems must exceed 1 month. Physical dependence has been shown conclusively to occur after chronic (greater or equal to 6 months) administration of both high (daily dosage of > 40 mg of diazepam equivalents) and therapeutic doses (daily dosage of less or equal to 40 mg of diazepam equivalents) of benzodiazepines. [10,14,15] Therapeutic-dose benzodiazepine dependence is of considerable interest to the public and the scientific community because of the potentially large population at risk (i.e., 1 to 3% of the adult population of the Western World use these drugs regularly for more than 1 year) and because of the personal and social consequences of addictive disorders. Preoccupation with the health and social consequences of benzodiazepine misuse and dependence has led to uncertainty among physicians and patients about the rational use and the potential of abuse and dependence with this drug class. There are two important factors contributing to the confusion. One problem is the complexity and lack of precision of the terminology used in many studies (use, abuse, addiction, and psychological and physical dependence). [16,17] Some studies report long-term therapeutic use of benzodiazepines as benzodiazepine dependence or abuse when in fact their patients may not meet DSM-III-R or DSM-IV criteria for dependence. [18,19] The second factor is the differing opinions among health professionals and the public regarding social values associated with the chronic use of medications for certain indications (e.g., stress, panic).
A consequence of the concerns about the use of benzodiazepines has led to restrictions in their use in some countries with controversial results. A study conducted to determine the effects of the New York State triplicate prescription intervention showed that, although prescriptions of benzodiazepines in New York State decreased by 44 to 57% after the regulation, the use of older and less safe sedative-hypnotics increased with rates varying from 15 to 136%.  For example, use of intermediate-acting barbiturates increased by 27%, non-benzodiazepine hypnotic use increased by 87%, and meprobamate prescription use increased by 129%.  Under that program, except in patients with panic or convulsive disorders, benzodiazepine prescriptions could not be written for more than a 1-month supply.  The rationale for this regulatory intervention resulted from the relationship between the availability of some drugs and the problems associated with their use. This relationship between availability and problems associated with drug use, such as adverse effects, overdoses, and, with certain drugs, abuse and dependence, has also been shown for benzodiazepines. [23-26]
In most Western and Northern Hemisphere countries, benzodiazepines can only be obtained by a prescription order from a physician. In most developing countries, however, although legislation has been passed to exert the same control over drug sales, there are obstacles in implementing such legislation as well as economic difficulties that the public often faces when accessing the health care system. Thus, a substantial proportion of benzodiazepines are sold over-the-counter (OTC) in pharmacies, at the discretion of pharmacists and/or other health care workers.  In Chile, for example, a study showed that approximately 50% of the benzodiazepines sold in pharmacies were purchased OTC.  This setting provides a unique opportunity to study the influence of availability of benzodiazepines on patterns of use, abuse, and dependence. In Latin American and most Asian countries, benzodiazepines can be obtained without a prescription; however, at the present time, there are no studies available that examine these issues. Therefore, we tested the hypothesis that the patterns of benzodiazepine use are different depending upon benzodiazepine availability (prescription vs. OTC). We hypothesized that, if patterns of benzodiazepine use increase with OTC use, rates of benzodiazepine abuse and dependence would be significantly higher among OTC benzodiazepine users than among prescription users. We tested this hypothesis by conducting a household survey in January 1991 in Santiago, Chile, which, based on previous studies,  we expected an approximately equal chance of OTC and prescription use of these drugs. The survey also provided demographic characteristics of benzodiazepine users from Santiago, as well as other information regarding types of benzodiazepines used, indications for use, and DSM-III-R criteria for dependence met by this population.
This study is a collaborative research project involving the Addiction Research Foundation, the Department of Psychiatry and the Faculty of Pharmacy of the University of Toronto, the Faculty of Chemical and Pharmaceutical Sciences of the University of Chile, and the Corporation for Health and Social Policy (CORSAPS), a private health policy research institution in Santiago, Chile. CORSAPS is a nonprofit organization affiliated with the University "Academy of Christian Humanism," and has a close link with the Ministry of Health and the Parliament.
This study included all inhabitants of Santiago, 15 years and older as stratified by data of the Instituto Nacional de Estadisticas (INE) (Chilean National Institute of Statistics).  Santiago is the capital city of Chile with approximately 5,200,000 inhabitants in 1990 (39.7% of the Chilean population). The Chilean national product value (U$ 1,809 per capita) is approximately half of the world average (U$ 3,470 per capita).  Health expenditures constitute approximately 2.5% of the gross national product. Forty-four percent of the Chilean population is considered to live at poverty levels and 18% at extreme poverty levels.  Our sample for the study was constituted by 1,500 persons. Age and gender were the variables for proportional stratification and were based on the demographic data published by INE for the year 1990. The population was separated into four age groups: 15 to 24 years, 25 to 39 years, 40 to 59 years, and 60 years and older.
Sample size and selection
Sample size was calculated with an estimated error of 2% and a confidence level of 88%. The sample size calculated was 1,482, which was then approximated to 1,500 individuals. 
The 1,500 individuals were selected in three stages:
1. A random computer selection of the blocks included in the city.
2. Selection of a household to contact was done systematically. The interviewer was positioned in the northwest corner of the assigned block and walked clockwise counting the houses of the assigned block and then divided by 4. The result yielded the interval for which to contact households until completing four households in the block. Provisions for buildings were also included. In the event that persons selected refused to participate in the survey, they were replaced systematically, and the interviewer contacted the household immediately before and, if necessary, the one after. If the original block did not yield the four households needed, the same procedure was applied in the north, west, south, and east blocks, in that order, if necessary.
3. One person who matched a predetermined gender and age range was selected from each household.
The Chilean Health Care System
A brief description of Chilean Health Care System organization is relevant because it impacts the way that benzodiazepines are prescribed, sold, and used. The Chilean Health Care System has both private and public sector components. Approximately 30% of the population has some form of private insurance, whereas the rest has access to a minimal care plan that does not cover medication costs.  The public sector access to the Health Care System is characterized by lack of resources and long delays in its delivery. Thus, most patients go directly to a pharmacy to obtain their medications without getting a physician's prescription. Pharmacists are pressured by patients and thus are caught in the dilemma between dispensing the medications or complying with loose regulations. The latter often results in the denial of pharmacologic treatment to those patients who have difficulty accessing the Health Care System. Therefore, more than 50% of medications in Chile are dispensed OTC. 
The following definitions were used for this study:
Lifetime prevalence of benzodiazepine use: An individual who had used a benzodiazepine at least once in his/her lifetime.
Past-year prevalence of use: An individual who had used a benzodiazepine at least once during the past year.
Frequent long-term user: Anyone who used a benzodiazepine every other day for longer than 1 year (past-year prevalence).
Daily long-term user: Anyone who used a benzodiazepine daily for more than 1 year (past-year prevalence).
Benzodiazepine abuser: Anyone meeting the following criteria: (a) using the benzodiazepine in doses higher than the recommended therapeutic doses for the major indication of the drug irrespective of duration of use and (b) having substantial problems associated with drug use (e.g., family or work problems). Problems such as dizziness or vague complaints of "malaise" or "nerves" were not considered sufficient to score criterion (b) as positive.
Benzodiazepine-dependent patient: Anyone who met the following criteria: (a) daily use of a benzodiazepine for 1 year or more and (b) meeting DSM-III-R criteria for sedative-hypnotic or anxiolytic dependence.
Interviews and training
Sixty interviewers were volunteers from either 6th-year students of the Faculty of Chemistry and Pharmacy, University of Chile, or 6th-year students of the Psychology Department of the University Diego Portales in Santiago.
Six training sessions were done with interviewers divided into four groups of 15 each and with one of the principal investigators (M. Busto, I. Ruiz) as trainers. The six sessions covered the following areas:
Session 1: General introduction, objectives, and relevance of the study. General aspects related to the methodology.
Session 2: Instrument presentation. Instructions for questionnaire completion. Importance of adequate presentation to the subject. Discussion and clarification of the definitions and terms. A final draft of the instrument was given to each trainee, with the task of piloting the questionnaire in one benzodiazepine user and using it for role-playing exercises.
Session 3: Collection of piloted instruments. Discussion of difficulties and incorporation of suggestions arising from the exercise (e.g., changes in language, clarification of terms).
Session 4: Instructions for instrument use were repeated, and each trainee was informed of any mistakes made when completing the pilot questionnaire. Appropriate corrections were discussed and implemented when necessary.
Session 5: Explanation of the system for the selection of the household and subject as explained above.
Session 6: Final review of the entire interview and sampling process. Appropriate material, interviewer code, and dates of completion given. Telephone numbers of the investigators were provided so that they could be contacted for questions or problems.
Interviewers were instructed to identify themselves to the targeted subject as a university student (showing identification), assess the appropriateness of the subject, explain that the objective of the survey was to "study the use of some medications commonly used by people," reassure the subject of the anonymity and confidentiality of the data (subjects will not be identified by name; data will be kept in locked cabinets and not be reported individually but by group), and then after obtaining consent from the subject, proceed with the interview.
The instrument in this study was a structured questionnaire used to collect information in several areas.
1. Demographic data: Age, gender, occupation, marital status, and education. Similar demographic data were collected about the head of the household (if different than the person interviewed). Data on the quality of the household (e.g., good quality = 100-140 m2, one to two bathrooms; poor quality = needs construction, self-construction, inadequate bathroom or sanitary condition), water supply, number of persons living in household, and address (30 questions).
2. Drug use data: The initial question was designed to ensure a relaxed atmosphere. The subject was asked about analgesic use: "The following is a list of drugs commonly used to treat pain, colds and fever. Can you tell me which of the following you have used at least once over the last year (check all that apply)." A list including acetylsalicylic acid, dipyrone, acetaminophen, and other common analgesics was presented and scored accordingly. Then, after a reference to hectic lifestyles, stress, and tension of everyday life, 21 questions (27 extra optional questions available depending on the answers provided) were asked about his/her benzodiazepine use. Each individual was shown a list of benzodiazepines and then were told: "The following is a list of tranquilizers, sedatives and hypnotics (i.e., 'nerve pills,' 'sleeping pills' or 'tension pills'). Please circle the ones that you have ever used." If the person still had difficulty remembering the names of drugs used, the interviewers were trained to provide oral descriptions of commonly used benzodiazepines, and if further clarification was required, the interviewers asked to see the medication to confirm the drug name. The following information was then obtained:
- Lifetime and past-year use: Dose, time, situation, and frequency of use.
- Indication for use: A list of the most frequent reasons for use was provided (e.g., tension, anxiety, insomnia).
- Initial use: Age of first use, information received on how medication should be used, source of benzodiazepine (e.g., "Who recommended that you use benzodiazepines initially?").
- Source: For example, physician prescription, pharmacy, family member. If the answer was "pharmacy," the interviewer asked a specific question about the method used to obtain the benzodiazepine ("always on prescription," "sometimes on prescription," "never on prescription").
- Questions related to all nine DSM-III-R criteria for abuse and dependence, including (1) use for longer or at higher doses than planned; (2) persistent desire or one or more unsuccessful attempts to control use; (3) great deal of time spent in getting the drug; (4) recurrent use when substance use is hazardous or frequent withdrawal or intoxication; (5) important social, occupational, and recreational activities given up because of substance use; (6) continued use despite significant problems; (7) marked tolerance (i.e., increase of 50%); (8) characteristic withdrawal symptoms when attempting to stop; and (9) substance used to relieve or avoid withdrawal.
3. Psychosocial variables: The instrument also contained questions on the extent of social participation (e.g., involvement in sports, hobbies, etc.), perception of risk with drug use, evaluation of nondrug strategies to cope with stress, anxiety, and insomnia, perceived ability to control events (external vs. internal locus of control on drug use), and degree of satisfaction with social interactions.
The questionnaire was revised, corrected and approved by a panel of anthropologists, pharmacists, physicians, psychologists, and psychiatrists to ensure appropriateness of the questions, language, and definitions. Interviews ranged from 10 to 60 minutes to complete, depending on the extent of substance use of the individual.
Reliability testing of the instrument and interviewer was performed by repeating 5% of randomly selected interviews. Address and confirmation of the interview was obtained in 100% of the cases. Subject(s) were re-interviewed (selected questions only) in 81.3% of the cases. Reliability was high, ranging from 88 to 100%, depending on the question.
Quality control of data collection
One thousand five hundred sixty interviews were performed by the 60 interviewers, 1,500 for the sample plus one extra interview per interviewer for replacement, if needed. All questionnaires were thoroughly reviewed by the investigators to control for completeness and for consensus on whether responses were consistent with a positive diagnosis based on DSM-III-R criteria. Incomplete questionnaires, wrong selection of subject (e.g., less than 15 years of age), and subjects with inconsistent answers were eliminated (N = 24, 1.5% of sample). These interviews were replaced from the extra 60 interviews performed, matched by age and gender. Data were coded and entered into an IBM-compatible computer and analyzed with the statistics of proportion, chi squared with Yates' correction if appropriate.  Contingency tables were prepared, and the appropriate level of significance was set at 0.05.
The 1,500 interviewed individuals were stratified by age and gender, and the sample was not found to be significantly different from the Santiago population as obtained from the INE for the study (chi squared = 3, df = 4, p < 0.17).  Figure 1 shows the population breakdown according to benzodiazepine use; as shown in the figure, the lifetime prevalence of benzodiazepine use was 42.9% (643 respondents reported use of this drug at least once in their life). Past-year prevalence of use was 31.4%: 471 individuals reported using a benzodiazepine at least once in the past year. Of these, 87 (18.5%) only used a benzodiazepine a few times per year, 76 (16.1%) used a benzodiazepine a few times in the month, 69 (14.6%) used a benzodiazepine every other day for less than 1 year, 54 (11.5%) used a benzodiazepine every other day for longer than 1 year (frequent long-term user), 93 (19.7%) used a benzodiazepine daily for less than 12 months, and 88 (18.7%) of users and 5.9% of the total sample met criteria for daily long-term use, that is, reported having used the medication daily for longer than 12 months. The remaining 4 (1%) did not remember the frequency of use.
(Table 1) provides the demographic characteristics of the 471 past-year benzodiazepine users. Use was significantly higher among women (women:men ratio 2:1), middle-aged, and elderly subjects as well as those of lower education (p < 0.01). Socioeconomic and marital status were not related to rates of use.
Type of benzodiazepine used
Diazepam was the most commonly used benzodiazepine in the past year (43%). Alprazolam, bromazepam, lorazepam, and chlordiazepoxide were the next most commonly used benzodiazepines (17%, 13%, 11%, and 10%, respectively). Benzodiazepines were most frequently used for their accepted indications: anxiety (33.3%) and insomnia (28.9%).
Seventy-one percent (335) of the subjects started using benzodiazepines on the recommendation of a physician. General practitioners were most frequently mentioned as the initiator of benzodiazepine use (185 cases [55.2%]). Psychiatrists and neurologists were the next most frequent initiators at 10.7% each, followed by cardiologists at 6.9%. Family or acquaintances were cited as initially recommending benzodiazepine use in 21.2% of cases and only 1.3% (6 cases) by pharmacists.
Two hundred thirty-eight (71%) subjects reported that their family doctor provided information regarding indications for use and possible side effects, whereas 90.1% of those patients whose benzodiazepine use was initiated by a cardiologist received this information. A substantial proportion of the subjects (21.5%) were never instructed about the length of time they should be using the drug (no significant differences among specialties in this respect).
Source of benzodiazepine
The source of medication was as follows: pharmacy: 362 (69.1%), community clinic: 80 (15.3%), family: 74 (14.1%), and others: 8 (1.5%). These numbers total more than the number of past-year benzodiazepine users because several people obtained their medication from more than one source. Of those obtaining the benzodiazepine from a pharmacy (N = 362), 162 (44.8%) never used a prescription, 77 (21.3%) reported using a prescription "sometimes," whereas the remaining 123 (33.9%) always used a prescription (Figure 1).
(Table 2) shows the impact of OTC benzodiazepine availability on the prevalence of benzodiazepine use, long-term use, and dependence for the 362 subjects who reported obtaining the benzodiazepine in a pharmacy. Most subjects (66%) had obtained the benzodiazepines without always using a prescription. Daily long-term users (N = 77) and subjects who met DSM-III-R criteria for dependence as discussed in the Methods section (N = 44) who obtained their benzodiazepine from a pharmacy were not more likely than other users to obtain benzodiazepines OTC (66% and 70% vs. 66%, chi squared = 0.77, not significant), thus rejecting our hypothesis. Of the 88 daily long-term users, 50 subjects (57%) (3.3% of the total sample) met three or more DSM-III-R criteria, thus meeting our criteria for dependent subjects. Only six subjects (0.4% of the total sample) met DSM-III-R criteria for severe benzodiazepine dependency, meeting seven or more criteria for dependence. All of these six subjects met criteria for daily long-term use, tolerance, withdrawal symptoms, and use of the benzodiazepine to reduce withdrawal symptomatology. No subject reported recreational use or met DSM-III-R criteria for benzodiazepine abuse as defined in the Methods section.
Those who used benzodiazepines in the past year either less often than daily or for less than 1 year (383 patients) reported significantly fewer DSM-III-R criteria for psychoactive substance use disorders (number of criteria met; x +/- SD = 1.9 +/- 1.7) than benzodiazepine-dependent subjects (x +/- SD = 4.5 +/- 1.3) (t-test = 12.87, p < 0.05). No significant differences were found between this same group of 383 nondaily/non-long-term users and 38 long-term nondependent users with respect to the number of DSM-III-R criteria met. However, some of the 471 subjects classified as users (N = 64) answered "yes" to more than three DSM-III-R benzodiazepine-dependence criteria. In 28 (43.8%) of these patients, the patterns of benzodiazepine use supported the view that these patients may have been benzodiazepine dependent and that our criteria for scoring a subject as "dependent" may have been too stringent (for example, they may have been taking the drug daily for 11 months only or two to three times a week for more than 1 year). However, for the remaining 36 users, their pattern of benzodiazepine use (i.e., used the drug only "occasionally" in the past year) was unlikely to lead to benzodiazepine dependence and these subjects could have been incorrectly classified as dependent if pattern of use was not examined, and thus could be considered false positives.
Also, the type of DSM-III-R criterion scored differed among the three categories (Figure 2). Nondaily or non-long-term users frequently reported difficulties in stopping use (DSM-III-R criterion No. 2) (60%) but only 12% reported tolerance (DSM-III-R criterion No. 7), and 18% described withdrawal symptoms (DSM-III-R criterion No. 8). In contrast, 92% of subjects classified as dependent reported difficulties trying to diminish or stop use (DSM-III-R criterion No. 2), 52% reported tolerance (DSM-III-R criterion No. 7), and 64% described withdrawal symptoms (DSM-III-R criterion No. 8). The differences were statistically significant (p < 0.01). Nondependent daily long-term users were similar to users with respect to the DSM-III-R criterion reported.
These data indicate the prevalence of benzodiazepine use, long-term therapeutic use, and dependence in Santiago, Chile. In addition, this article provides the opportunity to determine whether easy availability of benzodiazepines is a risk factor that is associated with increased use, frequent long-term use, daily long-term use, and dependence on these medications. Our results show that increased availability of benzodiazepines led to higher rates of use and daily long-term benzodiazepine use than those described for other countries. [1-5] However, daily long-term benzodiazepine users and dependent subjects were not more likely than other users to obtain the benzodiazepine OTC, suggesting that although availability increases benzodiazepine use, it is not a major risk factor contributing to the abuse and dependence of these drugs.
Despite recent decreases in anxiolytic use reported in the literature, [2,10] the high prevalence of benzodiazepine use found in this study is striking. It is almost three times higher than the average use in cross-national surveys conducted in 1991, [33,34] although prevalence varied widely among countries. Consistent with most studies on sedative-hypnotic use, benzodiazepine use in Chile was higher among women and increased with age. [1-5,10] In contrast, the prevalence of daily use for greater or equal to 12 months (daily long-term users) is not significantly higher than in many studies where the rates of daily use among some users (e.g., elderly subjects) have been reported to be as high as 25% of users.  Thus, our study, as in most other studies of benzodiazepine use, indicates that the most common pattern of benzodiazepine use was occasional use. 
Also, our data support the view that very few patients abuse and/or become severely dependent on benzodiazepines even when these medications can be obtained OTC. Dependence on benzodiazepines, however, is not the only factor to consider when the risks associated with benzodiazepine availability are examined. Benzodiazepine overdose, although rarely fatal, is both common and costly and so are possible acute interactions with alcohol leading to psychomotor impairment and traffic accidents. [26,35,36] A cautionary word is necessary, however, because, contrary to practice in developed countries, benzodiazepine use, frequent long-term use, and daily long-term use in this study was found to occur mostly (76%) unsupervised by a physician. Also, therapeutic indication(s) for the drug use was often unclear, drug doses and effectiveness of therapy frequently were not monitored, subjects may not have received information about the recommended duration of use, and withdrawal symptoms may have been self-treated.
Another interesting observation of this study is the high prevalence of diazepam use in Chile (43% mentioned diazepam as the benzodiazepine used). This is in contrast with the patterns of use in developed countries where data indicate dramatic shifts in benzodiazepine use from slowly eliminated (e.g., diazepam) to rapidly eliminated benzodiazepines (e.g., lorazepam and triazolam).  Explanations for the popularity of diazepam in Chile and other developing countries include the fact that it is an effective and fast-acting anxiolytic medication. Also, the trade name is well-known because it was one of the first benzodiazepines introduced in the world market. The low cost of the product (in Chile, approximately U$ 0.25 for 20 tablets and at least one third of the cost of the other benzodiazepine compounds available) in a population where medication costs are an important factor leading to drug use may be another reason for its popularity. Finally, it is possible that diazepam is chosen over other benzodiazepines because of its rapid absorption and fast onset of effects. 
The findings of this study suggest that OTC availability of benzodiazepines does increase occasional benzodiazepine use but is only one of the many risk factors affecting long-term use, abuse, and dependence on these medications. Intrinsic pharmacologic effects of drugs, individual attitudes, psychiatric comorbidity, and genetic factors are among the complex determinants of long-term use, drug abuse, and dependence. 
Our data were based on strict operational definitions of abuse and dependence. An evaluation of the literature shows that lack of clear definition is a well-known problem of drug utilization studies. In many studies, the terms employed in association with drug use are applied inconsistently and/or improperly. In a recent review of the use of benzodiazepines in anxiety disorders, Shader and Greenblatt  proposed some operational definitions for tolerance, abuse, and dependence. Our definition of abuse required the use of high doses of benzodiazepines, which is often, but not always, an indication of abuse. Our survey did not directly address the question of whether the benzodiazepines were specifically used for inducing euphoria as suggested in Shader and Greenblatt's definitions, thus we may have missed a small number of subjects who took therapeutic doses of benzodiazepine for the purpose of deliberate intoxication or pleasure. Also, even though our operational definition of dependence was strict, it is likely that most of the subjects identified as dependent were physically dependent on the drug, that is, an objective withdrawal syndrome would occur upon drug discontinuation. Physical dependence only describes a phenomenon that has a well-defined scientific meaning (neuroadaptation), but as Shader and Greenblatt adequately point out, it may not necessarily be linked to tolerance, abuse, or psychologic dependence. Given the limitations of our survey, it is difficult to determine how many of the 50 patients classified as "dependent" fall into the category of "physical dependence" only as opposed to patients who have developed a range of problems (i.e., social, medical, or psychiatric) in association with drug use in addition to physiologic dependence.
The findings of this study also raise important issues with respect to the most appropriate interventions to encourage rational drug use in general and of the benzodiazepines in particular. Although easy (OTC) availability of benzodiazepines is associated with higher occasional and frequent use of these medications, OTC availability is not associated with indiscriminate abuse and dependence of these drugs. It can be argued that underreporting may account for the finding that no subject met DSM-III-R criteria for abuse or recreational use of these drugs. This is certainly one of the well-recognized methodologic limitations of surveys. However, our finding that benzodiazepine abuse is rare is consistent with other surveys in the general population where benzodiazepine abuse rates are usually less than 1%  and with other experimental studies of benzodiazepine abuse liability. 
Several limitations of our methodology, which are also limitations of most surveys, must be acknowledged. For example, the interviewers in this survey were university students and thus not capable of distinguishing between difficulties with discontinuing or withdrawing from benzodiazepines from recurrence of anxiety or rebound insomnia. A qualified professional such as a psychiatrist would have been needed to make this interpretation. Also, the survey method may not have provided a truly random sample of the Santiago population. If selected individuals refused to participate (< 3%) or if they were not available, they were replaced by other persons who matched the demographic quotas. In addition, this study also emphasizes the need for careful interpretation of data obtained from surveys: 2.4% of the population could have been erroneously classified as dependent on benzodiazepines because their survey response score was positive for three or more DSM-III-R criteria. However, a close and careful look at their pattern of benzodiazepine use showed that this was unlikely from a pharmacologic point of view.
Also, our findings that benzodiazepine availability does not significantly increase rates of abuse and dependence of these drugs may be interpreted as a cautionary note against strict regulatory interventions on these medications. Although it is difficult to draw definite public health conclusions based on self-reported survey information, a large body of literature shows that restricting prescribing choices (e.g., formularies) is extremely successful in decreasing drug use. [20,42,43] However, as the implementation of the triplicate prescription in New York State has shown, restrictions on benzodiazepine use may set in motion compensatory changes in prescribing patterns that sometimes lead toward the use of less efficacious or other more costly medications.  Studies with other drug classes also support these observations.  Interventions designed to improve the quality of drug therapy should be done in the context of all other drugs available as alternatives. Restrictive measures may create considerable problems and may prove to be of little consequence from a quality-of-care point of view.  Continuing education of physicians, health professionals, and patients on the appropriate use of benzodiazepines seems to be a better alternative to stringent regulations on these medications.
We wish to thank Mr. Cristian Rocco and Mr. Francisco Espinosa for their invaluable help in the conduct of this study. We also thank Mr. Luis Saavedra (CORSAPS) for his help with the data analysis, Dr. C.A. Naranjo for his invaluable suggestions, and Ms. Tina Stuart for the typing of the manuscript.
1. Balter MB, Manheimer DI, Mellinger GD, Uhlenhuth EH. A cross-national comparison of anti-anxiety/sedative drug use. Curr Med Res Opin 1984;8[suppl]:5-20.
2. Balter MB. Benzodiazepine use/abuse: an epidemiologic appraisal. Presented at symposium: Triplicate prescription: issues and answers. New York, February 28, 1991.
3. Greenblatt DJ, Shader RI, Koch-Weser J. Psychotropic drug use in the Boston area: a report from the Boston Collaborative Drug Surveillance Program. Arch Gen Psychiatry 1975;32:518-25.
4. Wysowski DK, Baum C. Outpatient use of prescription sedative-hypnotic drugs in the United States, 1970 through 1989. Arch Intern Med 1991;151:1779-83.
5. Parry HJ, Balter MB, Mellinger GD, Cisin IH, Manheimer DL. National patterns of psychotherapeutic drug use. Arch Gen Psychiatry 1973;28:769-83.
6. Romach MK, Busto U, Sobell LC, Sobell MB, Somer GR, Sellers EM. Long-term alprazolam use: abuse, dependence or treatment? Psychopharmacol Bull 1991;27:391-5.
7. Mellinger GD, Balter MB, Uhlenhuth EH. Prevalence and correlates of the long-term regular use of anxiolytics. JAMA 1984;251:375-9.
8. Gottschalk LA, Bates DE, Fox RA, James JM. Psychoactive drug use: patterns found in samples from a mental health clinic and a general medical clinic. Arch Gen Psychiatry 1971;25:395-7.
9. Samarasinghe DS, Tilley S, Marks IM. Alcohol and sedative drug use in neurotic outpatients. Br J Psychiatry 1984;145:45-8.
10. American Psychiatric Association. Task force report on benzodiazepine dependence, toxicity, and abuse. Washington, DC: American Psychiatric Association, 1990.
11. Marks J. The benzodiazepines: use, overuse, misuse, abuse? Lancaster, England: MTP Press, 1985.
12. Nagy A. Long-term treatment with benzodiazepines: theoretical, ideological and practical aspects. Acta Psychiatr Scand Suppl 1987;335:47-55.
13. Woods JH, Katz JL, Winger G. Benzodiazepines: use, abuse, and consequences. Pharmacol Rev 1992;44:151-347.
14. Busto U, Sellers EM, Naranjo CA, Cappell H, Sanchez-Craig M, Sykora K. Withdrawal reaction after long-term therapeutic use of benzodiazepines. N Engl J Med 1986;315:854-9.
15. Hollister LE, Motzenbecker FP, Degan RO. Withdrawal reactions from chlordiazepoxide ("Librium"). Psychopharmacologia 1961;2:63-8.
16. Lader MH. Benzodiazepines-the opium of the masses? Neuroscience 1978;3:159-65.
17. Carranza J. Long-term use and abuse of benzodiazepines. Pharmacopsychiatry 1980;13:254-8.
18. Noyes R, Garvey MJ, Cook BL, Perry PJ. Benzodiazepine withdrawal: a review of the evidence. J Clin Psychiatry 1988;49:382-9.
19. Tyrer P. Dependence on benzodiazepines. Br J Psychiatry 1980;137:576-7.
20. Weintraub M, Singh S, Byrne L, Maharaj K, Guttmacher L. Consequences of the 1989 New York State triplicate benzodiazepine prescription regulations. JAMA 1991;266:2392-7.
21. Shader RI, Greenblatt DJ, Balter MB. Appropriate use and regulatory control of benzodiazepines. J Clin Pharmacol 1991;31:781-4.
22. New York State Department of Health. Benzodiazepines. Epidemiology Notes 1989;4 (No. 12).
23. de Lint J, Schmidt W. The epidemiology of alcoholism. In: Israel Y, Mardones J, editors. Biological basis of alcoholism. Toronto: Wiley-Interscience, 1971:423-42.
24. Ciraulo DA, Renner JA. Alcoholism. In: Ciraulo DA, Shader RI, editors. Clinical manual of chemical dependence. Washington, DC: American Psychiatric Press, Inc., 1991:1-94.
25. Fox JE. Outpatient alcoholism coverage to be tried. US Med 1983;19:24-5.
26. Busto U, Kaplan HL, Sellers EM. Benzodiazepine-associated emergencies in Toronto. Am J Psychiatry 1980;137:224-7.
27. Danhier A, Brieva J, Villegas G, Yates T, Perez H. Estudio de la utilizacion de psicofarmacos en la ciudad de Concepcion. Rev Chil Neuro-Psiquiat 1988;26:32-8.
28. Instituto Nacional de Estadisticas (INE). Compendio estadistico 1989. Santiago, Chile: Ministerio de Economia, Fomento y Reconstruccion, 1989.
29. World Bank Report on World Development 1990. In: Situacion y atencion de la salud en Chile. Santiago, Chile: Departamento de Comunicaciones, Ministerio de Salud, 1991:8.
30. Comision Economica para America Latina y el Caribe (CEPAL). Departamento de Estadisticas. In: Situacion atencion de la salud en Chile. Santiago, Chile: Departamento de Comunicaciones, Ministerio de Salud, 1991:8-10.
31. Cochran NG. Tecnicas de muestreo. Editorial continental. Mexico DF, 1971:109-11.
32. Departamento de Comunicaciones, Ministerio de Salud. Situacion de la salud en Chile. Santiago, Chile, 1991.
33. Famuyiwa OO. Psychotropic drug prescription in Nigeria. Acta Psychiatr Scand 1983;68:73-81.
34. Anonimo. Temas de actualidad. Informe epidemiologico sobre el uso y abuso de sustancias psicoactivas en 15 paises de America Latina y el Caribe. Bol Of Sanit Panam 1989;107:595-644.
35. Greenblatt DJ, Allen MD, Noel BJ, Shader RI. Acute overdosage with benzodiazepine derivatives. Clin Pharmacol Ther 1977;21:497-514.
36. Sellers EM, Busto U. Benzodiazepines and ethanol: assessment of the effects and consequences of psychotropic drug interactions. J Clin Psychopharmacol 1982;2:249-62.
37. Busto U, Lanctot KL, Isaac P, Adrian M. Benzodiazepine use and abuse in Canada. Can Med Assoc J 1989;141:917-21.
38. Busto U, Bendayan R, Sellers EM. Clinical pharmacokinetics of non-opiate abused drugs. Clin Pharmacokinet 1989;16:1-26.
39. Busto U, Sellers EM. Pharmacokinetic determinants of drug abuse and dependence. A conceptual perspective. Clin Pharmacokinet 1986;11:144-53.
40. Shader RI, Greenblatt DJ. Use of benzodiazepines in anxiety disorders. N Engl J Med 1993;328:1398-405.
41. Griffiths RR, Bigelow G, Liebson I. Human drug serf-administration. Double-blind comparison of pentobarbital, diazepam, chlorpromazine and placebo. J Pharmacol Exp Ther 1979;210:301-10.
42. Carruthers G, Goldberg T, Segal H, Sellers EM. Drug utilization: a comprehensive literature review. Department of Health Administration, Community Health Division, Faculty of Medicine, University of Toronto, 1987.
43. Brahams D. Benzodiazepine overprescribing: successful initiative in New York State. Lancet 1990;336:1372-3.
44. Shenfield GM, Jones AN, Paterson JW. Effect of restrictions on prescribing patterns of dextropropoxyphene. BMJ 1980;281:651-3.