Medical and nonmedical stimulant use among adolescents: from sanctioned to unsanctioned use
===========================================================================================

* Christiane Poulin

## Abstract

**Background:** The past decade has seen a generalized upward trend in the prevalence of adolescent use of substances, including stimulants. The purpose of this article was to determine the prevalence of and risk factors for the medical and nonmedical use of stimulants, and the diversion of prescribed stimulants among adolescent students, and to demonstrate links between medical use, nonmedical use and the diversion of stimulants.

**Methods:** A self-reported anonymous questionnaire was administered in 1998 to a random sample of students in grades 7, 9, 10 and 12 in New Brunswick, Nova Scotia, Prince Edward Island, and Newfoundland and Labrador about their medical and nonmedical use of stimulants (Benzedrine, Dexedrine, Ritalin, Cylert, diet pills, “speed,” “uppers,” “bennies” and “pep pills”). A total of 13 549 students completed the questionnaire, representing a 99% participation rate among the students present at school on the day of the survey.

**Results:** Of the 5.3% of students who reported medical use of stimulants in the 12 months before the survey, 14.7% reported having given some of their medication, 7.3% having sold some of their medication, 4.3% having experienced theft and 3.0% having been forced to give up some of their medication. Nonmedical stimulant use by students who did not have a prescription for stimulants was significantly related to increased numbers of students who gave or sold some of their prescribed stimulants, at both the school class and individual student levels (*p* < 0.001).

**Interpretation:** Although the vast majority of adolescent students taking prescribed stimulants appeared to be using their medication as sanctioned, a link was found between medical and nonmedical stimulant use and the diversion of medication from sanctioned to unsanctioned use.

The past decade has seen a generalized upward trend in the prevalence of substance use among adolescents, with Canada and the United States reporting marked increases in the use of many substances, including stimulants.1,2,3,4,5 The annual prevalence of nonmedical stimulant use among adolescents in Nova Scotia increased from 5% in 1991 to 11% in 1998.2,3,4 In the United States, from 1992 to 1997 the annual prevalence of stimulant use increased from 8.2% to 12.1% among students in grade 10 and from 7.1% to 10.1% among students in grade 12.1

Concurrently, major increases in the prescribing of methylphenidate have been observed in Canada and the United States.6,7,8 In Canada, the amount of methylphenidate prescribed increased about five-fold from the early to the mid-1990s.6 In the United States the number of prescriptions given to youths increased about three-fold from 1990 to 1995, the largest increase (311%) occurring among high school students 15–19 years of age.7,8 About 90% of all methylphenidate in the United States is thought to be prescribed to children and adolescents with attention deficit/hyperactivity disorder.7,8 Considerable variation in methylphenidate use has been reported according to sex, age, geographic region and health care system.6,7,8,9

Increased prescribing of stimulant medication is of concern, in part because of the increased potential for the diversion of the drug from the licit to the illicit market.10,11 Little is known about actual diversion of stimulant medication. In the peer-reviewed literature, there is a report on attempted prescription fraud, where large quantities of amphetamines were to be procured illegally from legitimate manufacturers and physicians.12 In the case of youths, however, diversion activity may take place on a smaller scale and on a more personal level. For example, a representative survey in Wisconsin enquired whether schoolchildren prescribed methylphenidate had been approached to sell, give or trade their medication.13 There is also anecdotal evidence of youths being forced to give up their stimulant medication, experiencing theft, or willingly giving or selling some of their pills.14,15

The present research provides population-based data about the actual outflow of prescribed stimulants from sanctioned to unsanctioned use among adolescents. The objectives of the research were (a) to determine the prevalence of and risk factors for the medical and nonmedical use of stimulants, and the diversion of prescribed stimulants among adolescent students, and (b) to demonstrate links among medical use, nonmedical use and the diversion of prescribed medication to unsanctioned use in the general adolescent population.

## Methods

The present study is based on the 1998 Student Drug Use Survey in the Atlantic Provinces, a self-reported anonymous survey of students in grades 7, 9, 10 and 12 of the public school systems in New Brunswick, Nova Scotia, Prince Edward Island, and Newfoundland and Labrador.4,16,17,18 The 1998 survey was the second application of a protocol developed and pilot tested in 1994/95.19 Ethics approval was obtained from the Dalhousie University Faculty of Medicine Ethics Committee. The sample design was a cluster sample of randomly selected classes stratified by grade and by either health region (in Nova Scotia and Newfoundland) or school district or board (in New Brunswick and PEI).

The survey used a computer-scannable instrument of 94 items. The methods to assess validity and reliability of the survey are reported elsewhere.20 A test-retest took place in Nova Scotia in March 1998, with an initial sample of 240 students in grades 7, 9, 10 and 12, yielding 225 completed test-retest surveys paired on encoded unique identifiers. The test-retest revealed fair to good agreement for medical stimulant use in the 30 days before the survey and for medical and nonmedical stimulant use in the 12 months before the survey (kappa values 0.68, 0.60 and 0.62 respectively).21

Nonmedical stimulant use in the 12 months before the survey was defined as an affirmative response to the question “In the past 12 months, have you taken stimulants (Benzedrine, Dexedrine, speed, uppers, bennies, pep pills) without a prescription or without a doctor telling you to do so?” Medical stimulant use in the 12 months before the survey was defined as an affirmative response to the question “In the past 12 months, have you taken stimulants (Benzedrine, Dexedrine, Ritalin, Cylert, diet pills) as prescribed for you by your doctor?” Medical stimulant use in the month before the survey was defined as an affirmative response to the question “In the past 30 days, how often did you usually take stimulants (Benzedrine, Dexedrine, Ritalin, Cylert, diet pills) as prescribed for you by your doctor?”

The diversion of prescribed stimulants was defined as an affirmative response to any of the following questions: “Have you ever given any of your stimulant pills to friends? Have you ever sold any of your stimulant pills? Have any of your stimulant pills ever been taken away from you against your will (by force or threats)? and Have any of your stimulant pills ever been stolen from you?”

The statistical analysis took into account the stratified cluster sample design. The prevalence estimates were weighted according to the number of students responding in each stratum and the total number of students enrolled in each stratum. The extent of ever giving, selling, or experiencing coercion or theft was expressed as proportions of students who reported medical stimulant use in the year before the survey. The standard errors used to compute the 95% confidence intervals (CIs) were adjusted for the intracluster correlation by means of the Kish design effect.22

Considered as potential risk factors for stimulant use were sex, grade, province, any alcohol use, any cigarette smoking and any cannabis use. The proportions of students reporting use and non-use of stimulants were compared using logistic regression analysis, initially with unadjusted odds ratios (ORs) and subsequently with adjusted ORs in multivariate models. Statistical significance was set at *p* < 0.05 with Bonferroni adjustment for the number of variables in the multivariate models, which resulted in a *p* value of 0.008 or 0.01 depending on the model.

Finally, the influence that the number of students who gave some of their prescribed stimulant medication had on nonmedical stimulant use was determined at the school class and individual student levels, using multivariate logistic regression analysis. Giving and selling prescribed medication were assessed in separate multivariate models because of collinearity. The samples of students in the 2 mainland provinces, and of those in the 2 island provinces, were combined in order to preserve an adequate sample size according to place.

## Results

A total of 13 549 students in grades 7, 9, 10 and 12 completed the questionnaire, representing a 99% participation rate among the students present at school on the day of the survey. About 13% of the students in the participating classes were absent on the day of the survey. The mean age of the respondents was 15.2 years. The median age was 13 years in grade 7, 15 years in grade 9, 16 years in grade 10 and 18 years in grade 12. The samples in the 4 provinces did not differ significantly in terms of age, sex and absenteeism.

### Stimulant use

Overall, 2.6% of the students reported medical stimulant use in the 30 days before the survey. Significantly more male than female students reported such use (3.2% v. 2.1%, *p* < 0.001). The prevalence was significantly lower among grade 12 students than among students in grades 7 to 10 (1.6% v. 2.8% to 3.2%) and among students in Newfoundland and Labrador than among those in the other 3 provinces (1.6% v. 2.7% to 3.3%).

Medical stimulant use in the year before the survey was reported by 5.3% of the students (Table 1). Simultaneous adjustment for all student characteristics revealed that alcohol use, cigarette smoking and cannabis use were all independent risk factors for medical stimulant use. The prevalence of medical use in the year before the survey was significantly lower among students in grade 12 than among students in grades 7 to 10 after adjustment for all variables (OR 0.4, *p* < 0.001).

View this table:
[Table1](http://www.cmaj.ca/content/165/8/1039/T1)

Table 1. 

Nonmedical stimulant use in the year before the survey was reported by 8.5% of the students (Table 1). Again, alcohol use, cigarette smoking and cannabis use were found to be independent risk factors. The strengths of those associations were considerably greater for nonmedical use (OR range 2.3 to 5.9) than for medical use (OR range 1.5 to 2.2). Once substance use was taken into account, the only demographic characteristic found to be an independent protective factor for nonmedical stimulant use was residence in Newfoundland and Labrador (OR 0.6, *p* < 0.001).

Medical and nonmedical stimulant use in the year before the survey were not mutually exclusive. About 3.5% of the students reported only medical use, 6.8% reported only nonmedical use, and 1.8% reported both medical and nonmedical use.

### Diversion of prescribed stimulants

Of the students who reported medical stimulant use in the year before the survey, 14.7% reported having given, and 7.3% having sold, some of their prescribed stimulants (Table 2). About 80% of the students who reported having sold some of their medication also reported having given some away. Grade and province were not associated with giving or selling stimulant medication. The only drug use pattern found to be predictive of giving or selling stimulant medication was nonmedical stimulant use. Compared with students who did not report nonmedical stimulant use, those who did report nonmedical use were about 3.3 and 4.6 times more likely to report having given or sold some of their medication, respectively. Male sex was found to be an independent risk factor for selling medication.

View this table:
[Table2](http://www.cmaj.ca/content/165/8/1039/T2)

Table 2. 

Regarding coercion and theft, 3.0% (95% CI 1.6%–4.4%) and 4.3% (95% CI 2.7%–5.9%) of the students taking prescribed stimulants in the 12 months before the survey reported that some of their pills had been taken from them against their will or had been stolen, respectively. Sex, grade and province were not found to be risk factors.

### Relation between stimulant use and diversion of prescribed stimulants

The relation between medical use, nonmedical use and diversion was examined on the basis of 2 assumptions. First, it was assumed that actively giving or selling stimulants and, reciprocally, taking or buying stimulants might happen among students in the same school class. Second, it was assumed that students who reported nonmedical stimulant use who did not have a prescription for stimulants might obtain the drugs from students with a ready supply, that is, from students reporting medical stimulant use.

Nonmedical stimulant use was found to be significantly associated with the giving of prescribed stimulant medication. With the school class as the unit of analysis, the proportion of classes experiencing nonmedical stimulant use was found to increase in relation to the number of students in a class who reported having given some of their medication away. (For this analysis, the responses of 181 students in 63 of the 719 participating classes were excluded because the class had fewer than 6 students.) For example, in Nova Scotia and New Brunswick, 66%, 78% and 89% of the classes with 0, 1, and 2 or 3 students who gave medication, respectively, had at least 1 student who reported nonmedical stimulant use in the year before the survey (*p* < 0.001) (Fig. 1). Nonmedical stimulant use was also found to be significantly associated with the giving of prescribed stimulant medication on the individual student level. Compared with students in classes in which no one reported having given some of his or her prescribed medication away, those in classes with 3 students who did so were twice as likely to report nonmedical stimulant use (OR 2.1, *p* < 0.001). The proportion of classes and individual students experiencing or reporting nonmedical stimulant use, in relation to the giving of medication, were significantly higher in Nova Scotia and New Brunswick than in PEI and Newfoundland and Labrador (*p* < 0.001) (Fig. 1). Similar findings were observed relative to the selling of stimulant medication.

![Figure1](http://www.cmaj.ca/https://www.cmaj.ca/content/cmaj/165/8/1039/F1.medium.gif)

[Figure1](http://www.cmaj.ca/content/165/8/1039/F1)

**Fig. 1: Nonmedical stimulant use reported by adolescent students in Atlantic Canada who did not have a prescription for stimulant medication (shown as proportions of school classes and individual students, stratified by province), according to the number of students in their class who gave away some of their prescribed stimulant medication.** 

## Interpretation

The present study provides a population-based estimate of the proportion of students with prescriptions for stimulant medication who experienced a diversion of their medication to nonmedical use. The vast majority of students taking stimulant medications appeared to be using them as sanctioned. A relatively small proportion who were prescribed stimulants had their medication stolen or were forced to give it up. Musser and colleagues13 found that 16% of Wisconsin school children had been approached to give, sell or trade their prescribed methylphenidate; unfortunately, the authors did not ask whether the children had actually done so. In the Atlantic provinces about 15% and 7% of the students who reported medical stimulant use in the 12 months before the survey reported actually having given or sold, respectively, some of their medication. Thus, a subset of students apparently misused or abused their stimulant medication for its recreational potential or currency as a street drug.

In addition, the present survey provides empirical evidence about the relation between medical and nonmedical stimulant use and the diversion of prescribed stimulants among students. The most direct evidence was the reporting of giving or selling of stimulants by some students with prescriptions. Students most likely to have given or sold some of their medication were those who also reported nonmedical stimulant use. Medical and nonmedical stimulant use were also associated with substance use. Compared with students who reported medical stimulant use, students who reported nonmedical use appeared to be much more committed users of alcohol, cigarettes and cannabis. At the school class level, an increasing proportion of students per class reporting the giving or selling of medication was found to be an independent predictor of at least 1 student per class reporting not having a prescription but nonetheless using stimulants. Thus, it appears that the school class may constitute not only an administrative collective but also a natural social group or informal economic market in which drug transactions can occur.

The main limitation of the study was that it relied on self-reported information. School-based surveys are thought to systematically underestimate the prevalence of risk behaviours in the larger adolescent population, because youths not in school, through either absenteeism or school dropout, are more likely to engage in such behaviours than are adolescents in school.23 Thus, the present study may provide an underestimate of the prevalence and diversion of stimulants in the adolescent population of Atlantic Canada. As well, the study is cross-sectional and thus provides correlational rather than causal evidence about the relation between diversion and nonmedical stimulant use. Finally, the questionnaire asked about various stimulants in a single question rather than in separate questions, which may have led to an obscuring of the ratios of interest. For example, several studies have reported a preponderance of females taking diet pills1 and males taking methylphenidate.7,8,9

Ultimately, public health and clinical practice must inform each other. From a clinical perspective, physicians prescribing stimulant medication should be vigilant concerning potential abuse, particularly among adolescent patients known or thought to be using other substances. Physicians and parents should keep track of stimulant medication, especially when several months' supply is prescribed.24 The present study should not detract from the a priori value of methylphenidate and dextroamphetamine, which are well-established and safe medications of proven efficacy for attention deficit/hyperactivity disorder.25 However, this study does underscore the need for a broad understanding of benefit versus risk, one that recognizes the impact of prescribing practices at the societal as well as the individual level.

## Acknowledgments

Thanks are extended to Linda Van Til, Department of Health and Social Services, Prince Edward Island; Jim Baker, Department of Health, Nova Scotia; Ron Tizzard, Department of Health and Community Services, Newfoundland and Labrador; and Bob Jones, Department of Health and Community Services, New Brunswick, for reviewing an earlier draft of the document.

This research was supported by Medical Research Council of Canada grant MA-14706 and a Clinical Scholar Award from the Faculty of Medicine, Dalhousie University. Funding for data collection was provided in part by the Departments of Health in Nova Scotia, Newfoundland and Labrador, Prince Edward Island and New Brunswick.

## Footnotes

*   *This article has been peer reviewed.* 
    
    *Competing interests:* None declared.

## References

1.  1. Johnson LD, O'Malley PM, Bachman JG. *Secondary school students*. Vol 1 of *National survey results on drug use from the Monitoring the Future Study, 1975–1998*. Washington: US Department of Health and Human Services; 1999. NIH pub no 99–4660.
    
    

2.  2. Poulin C, Elliott D. The prevalence of alcohol, tobacco and cannabis use and of harmful consequences among Nova Scotia adolescent students: implications for prevention and harm reduction. CMAJ 1997;156(10):1387-93. Available: [www.cma.ca/cmaj/vol-156/issue-10/1387.htm](http://www.cma.ca/cmaj/vol-156/issue-10/1387.htm)
    
    [Abstract/FREE Full Text](http://www.cmaj.ca/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NDoiY21haiI7czo1OiJyZXNpZCI7czoxMToiMTU2LzEwLzEzODciO3M6NDoiYXRvbSI7czoyMToiL2NtYWovMTY1LzgvMTAzOS5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=) 

3.  3. Poulin C, Van Til L, Wilbur B, Clarke B, MacDonald CA, Barcelo A, et al. Alcohol and other drug use among adolescent students in the Atlantic provinces. Can J Public Health 1999;90:27-9.
    
    [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=10910561&link_type=MED&atom=%2Fcmaj%2F165%2F8%2F1039.atom) 

4.  4. Poulin C, Baker J. *Nova Scotia student drug use 1998* [technical report]. Halifax: Nova Scotia Department of Health and Dalhousie University; 1998.
    
    

5.  5. Adlaf EM, Paglia A, Ivis FJ, Ialomiteanu A. Nonmedical drug use among adolescent students: highlights from the 1999 Ontario Student Drug Use Survey. *CMAJ* 2000;162(12);1677-80. Available: [www.cma.ca/cmaj/vol-162/issue-12/1677.htm](http://www.cma.ca/cmaj/vol-162/issue-12/1677.htm)
    
    

6.  6. Miller A, Lee SK, Raina P, Klassen A, Zupancic J, Olsen L. *A review of therapies for attention-deficit/hyperactivity disorder*. Ottawa: Canadian Coordinating Office for Health Technology Assessment; 1998. Available: www .ccohta .ca/newweb/pubapp/pdf/adhd\_tech\_rep.pdf (accessed 2001 Sept 13).
    
    

7.  7. Safer DJ, Zito JM, Fine EM. Increased methylphenidate usage for attention deficit disorder in the 1990s. Pediatrics 1996;98:1084-8.
    
    [Abstract/FREE Full Text](http://www.cmaj.ca/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6MTA6InBlZGlhdHJpY3MiO3M6NToicmVzaWQiO3M6OToiOTgvNi8xMDg0IjtzOjQ6ImF0b20iO3M6MjE6Ii9jbWFqLzE2NS84LzEwMzkuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 

8.  8. Zito JM, Safer DJ, dosReis S, Gardner JF, Boles M, Lynch F. Trends in the prescribing of psychotropic medications to preschoolers. *JAMA* 2000; 283; 1025-30.
    
    

9.  9. LeFever GB, Dawson KV, Morrow AD. The extent of drug therapy for attention deficit-hyperactivity disorder among children in public schools. Am J Public Health 1999;89:1359-4.
    
    [CrossRef](http://www.cmaj.ca/lookup/external-ref?access_num=10.2105/AJPH.89.9.1359&link_type=DOI) 
    
    [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=10474553&link_type=MED&atom=%2Fcmaj%2F165%2F8%2F1039.atom) 
    
    [Web of Science](http://www.cmaj.ca/lookup/external-ref?access_num=000082214600011&link_type=ISI) 

10. 10. Haislip GR. Drug diversion control systems, medical practice, and patient care. NIDA Res Monogr 1993;131:120-31.
    
    [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=8413522&link_type=MED&atom=%2Fcmaj%2F165%2F8%2F1039.atom) 

11. 11. Morrow RC, Morrow AL, Haislip G. Methylphenidate in the United States, 1990 through 1995 [letter]. Am J Public Health 1998;88:1121.
    
    [CrossRef](http://www.cmaj.ca/lookup/external-ref?access_num=10.2105/AJPH.88.7.1121&link_type=DOI) 
    
    [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=9663168&link_type=MED&atom=%2Fcmaj%2F165%2F8%2F1039.atom) 
    
    [Web of Science](http://www.cmaj.ca/lookup/external-ref?access_num=000074610600029&link_type=ISI) 

12. 12. Whalen RP, Silverman R. Diversion of narcotic and stimulant drugs. N Y State J Med 1975;75(5):735-7.
    
    [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=1054792&link_type=MED&atom=%2Fcmaj%2F165%2F8%2F1039.atom) 

13. 13. Musser CJ, Ahmann PA, Theye FW, Mundt P, Brosete SK, Mueller-Rizner N. Stimulant use and the potential for abuse in Wisconsin as reported by school administrators and longitudinally followed children. J Dev Behav Pediatr 1998;19:187-92.
    
    [CrossRef](http://www.cmaj.ca/lookup/external-ref?access_num=10.1097/00004703-199806000-00006&link_type=DOI) 
    
    [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=9648044&link_type=MED&atom=%2Fcmaj%2F165%2F8%2F1039.atom) 
    
    [Web of Science](http://www.cmaj.ca/lookup/external-ref?access_num=000074270600006&link_type=ISI) 

14. 14. Frei R. Ritalin becoming school drug problem. Students robbing classmates of their prescription drugs. *Med Post* 1998;34(26). Available: [www.medicalpost.com/mdlink/english/members/medpost/data/3426/04A.htm](http://www.medicalpost.com/mdlink/english/members/medpost/data/3426/04A.htm) (accessed 2001 Sept 13).
    
    

15. 15. Chisholm P. The ADD dilemma. *Maclean's* 1996 Mar 11:42-4.
    
    

16. 16. Barcelo A, Jones B, Grobe C. *Provincial Student Drug Use Survey: highlights 1998*. Fredericton: New Brunswick Department of Health and Community Services, Provincial Epidemiology Service; 1998.
    
    

17. 17. MacDonald CA, Holmes PR. *Newfoundland and Labrador student drug use 1998* [technical report]. St. John's: Government of Newfoundland and Labrador, Department of Health, Addictions Services; 1998.
    
    

18. 18. Van Til L, MacMillan H, Poulin C. *Prince Edward Island student drug use 1998* [technical report]. Charlottetown: Department of Health and Social Services; 1998.
    
    

19. 19. Poulin C, Clarke B, Balram C, Wilbur B, Bryant E. *Student drug use surveys in the Atlantic provinces: a standardized approach*. Halifax: Dalhousie University; 1996.
    
    

20. 20. Poulin C, MacNeil P, Mitic W. The validity of a province-wide student drug use survey: lessons in design. Can J Public Health 1993;84:259-4.
    
    [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=8221500&link_type=MED&atom=%2Fcmaj%2F165%2F8%2F1039.atom) 
    
    [Web of Science](http://www.cmaj.ca/lookup/external-ref?access_num=A1993LW65700011&link_type=ISI) 

21. 21. Fleiss JL. *Statistical methods for rates and proportions*. 2nd ed. New York: John Wiley & Sons; 1981.
    
    

22. 22. Kish L. *Survey sampling*. New York: John Wiley & Sons; 1965.
    
    

23. 23. Johnston LD, O'Malley PM. Issues of validity and population coverage in student surveys of drug use. NIDA Res Monogr 1985;57:31-54.
    
    [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=3929114&link_type=MED&atom=%2Fcmaj%2F165%2F8%2F1039.atom) 

24. 24. Garland EJ. Intranasal abuse of prescribed methylphenidate. J Am Acad Child Adolesc Psychiatry 1998;37:1212-3.
    
    

25. 25. Elia J, Ambrosini PJ, Rapoport JL. Treatment of attention-deficit-hyperactivity disorder. N Engl J Med 1999;340:780-8.
    
    [CrossRef](http://www.cmaj.ca/lookup/external-ref?access_num=10.1056/NEJM199903113401007&link_type=DOI) 
    
    [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=10072414&link_type=MED&atom=%2Fcmaj%2F165%2F8%2F1039.atom) 
    
    [Web of Science](http://www.cmaj.ca/lookup/external-ref?access_num=000079034100007&link_type=ISI)