Substance Use by Adolescents in Cape Town - Journal of Adolescent ...

JOURNAL OF ADOLESCENT HEALTH 2003;32:58 – 65

INTERNATIONAL ARTICLE

Substance Use by Adolescents in Cape Town: Prevalence and Correlates ALAN J. FLISHER, Ph.D., F.C.Psych. (S.A.), CHARLES D.H. PARRY, Ph.D., JANET EVANS, M.Sc., MARTIE MULLER, B.Sc. (Hons.), AND CARL LOMBARD, Ph.D.

Purpose: To document the prevalence rates for use of cigarettes, alcohol, and cannabis among high school students in Cape Town, and to investigate whether use of these substances is associated with a set of hypothesized psychosocial correlates. Methods: A multistage sampling procedure produced a sample of 2930 students in grades 8 and 11 at 39 high schools in Cape Town, who completed a self-administered questionnaire. The questionnaire contained items about whether the students had used various substances and that addressed the potential correlates of interest. We calculated prevalence rates with 95% confidence intervals and constructed a set of generalized estimating equations of use in the past month of cigarettes, alcohol, or cannabis on the hypothesized correlates. Results: The prevalence rates for previous month (recent) use of cigarettes, alcohol, and cannabis were 27%, 31%, and 7%, respectively. Rates were low for black females. Recent use of each of the substances was significantly associated with the number of days absent and the number of years lived in a city. Repeating a grade was significantly associated with previous month use of cigarettes and alcohol by colored (derived from Asian, European, and African ancestry) students and alcohol use by black grade 8 students (race classifications “colored” and “black” are as defined by the repealed population

From the Department of Psychiatry and Mental Health of the University of Cape Town, the Alcohol and Drug Abuse Research Group, and the Biostatistics Unit, Medical Research Council, Cape Town, Republic of South Africa. Address correspondence to: Alan J. Flisher, Ph.D., F.C.Psych. (S.A.), Department of Psychiatry, Groote Schuur Hospital, Observatory 7925, Cape Town, Republic of South Africa; E-mail: [email protected]. The work reported in this paper was supported by grants from the World Health Organisation Programme on Substance Abuse, the United Nations Development Programme, the South African Medical Research Council, and the Medical Faculty Research Committee of the University of Cape Town. Manuscript accepted June 20, 2002. 1054-139X/03/$–see front matter PII S1054-139X(02)00445-7

Registration Act of 1950). Not being raised by both parents was significantly associated with cigarette smoking by black and colored students, alcohol use by colored students, and cannabis use by female students. It was inversely associated with cigarette use by black students. Conclusions: It is necessary to identify the factors that protect black female adolescents from substance use. It is important to address demographic factors such as race classification and gender analytically if one is to avoid obscuring differences among groups. © Society for Adolescent Medicine, 2003 KEY WORDS:

Adolescence Alcohol Cannabis Correlates Gender differences Racial differences Substances Tobacco South Africa

There have been few methodologically sound studies that have aimed to document prevalence rates for substance use by South African adolescents. RochaSilva et al. (1) conducted a study with a representative sample of black youth aged 10 –21 years (n ⫽ 1378). Among their findings were that 34% had used alcohol in the previous 12 months and that cannabis use was confined to urban males, among whom 6% had used cannabis. However, this study was confined to black youth. Flisher et al. (2) had a sample of 7340 students that was representative of the population of high school students of all race classifications in the City of Cape Town. They found that 18% of

© Society for Adolescent Medicine, 2003 Published by Elsevier Science Inc., 360 Park Avenue South, New York, NY 10010

January 2003

males were regular smokers, 27% had engaged in binge drinking in the previous 2 weeks, and 8% had ever used cannabis. The remaining studies are outdated or suffer from methodological limitations such as small or unrepresentative samples or poor specification of variables (3,4). Rates of substance use may have changed in the light of demographic shifts; political and economic changes, such as the end of apartheid and the implementation of a democratic system of government; the AIDS epidemic, which accounts for about 40% of all deaths in the age group 15 to 49 years (5); and changes in the global production, distribution, and marketing of drugs. It is important to have contemporaneous data to inform intervention efforts such as school-based drug prevention programs and health policy (4). A further limitation of existing studies is that few have addressed correlates of substance use. Again, such data are necessary to suggest demographic subgroups that are at high risk and to point to possible risk factors that may be modifiable. Exceptions include studies addressing the associations between substance use and other risk behaviors (6,7), urbanization (8,9), high school dropout (10), and domains of the theory of planned behaviour (11). However, these studies were confined to one or two race groups, had small or unrepresentative samples, were confined to binge drinking, or were carried out some time ago. In addition, they included a limited range of potential correlates. Research conducted elsewhere has examined the associations between substance use and variables such as absenteeism (12), economic disadvantage (13,19), poor scholastic progress (14 –17), and family structure, such as not being raised by both parents (18 –21). One cannot extrapolate the results from other countries to South Africa. Not only may the results differ owing to the different circumstances in different countries, but the international studies are not necessarily consistent with each other. Amey and Albrecht (18), for example, found that the singleparent African-American family provides a greater protection against drug use than the two-biologicalparent family. Also using a national database, Hoffmann and Johnson (20) arrived at the opposite conclusion. The present study has two aims. First, to document the prevalence rates for substance use among a representative sample of high school students in Cape Town, South Africa. Second, to investigate whether use of selected substances is associated with: (a) absenteeism from school; (b) duration of residence in an urban area; (c) socio-economic status,

SUBSTANCE USE BY ADOLESCENTS IN CAPE TOWN

59

assessed by the number of people sharing a room with the student at night; (d) poor scholastic progress, assessed by whether the student had repeated a grade; and (e) not being raised by both parents.

Methods Sample The study population was all students in Grades 8 and 11 attending public schools in Cape Town, South Africa. Schools were stratified by postal code groupings, and 39 schools were selected such that the proportion of the selected schools in a selected stratum was directly proportional to the number of students in that stratum. Within each stratum, the selection probability of a school was proportional to the number of students in that school. Forty students were randomly selected from the combined class list of two randomly selected classes from each participating grade. An additional five students were selected as replacements for absentees. A maximum of five absent students were replaced.

Instrument The instrument was a self-completed questionnaire. Besides items about demographic characteristics (gender, grade, and race classification), the questionnaire contained items about whether the students had used various substances and that addressed the potential correlates of interest. The questions involving substances referred to smoking a whole cigarette; using alcohol (including beer and wine), other than a few sips; and smoking cannabis. For each substance, we asked whether the students had ever used the substance and, if they had, whether they had done so in the previous year and on how many days they had used the substance in the previous month. If a student had used a substance on one or more occasions in the previous month, we coded that student as having used the substance in the previous month. To assess the potential correlates of interest, we asked the students the number of days they had been absent from school in the first term (quarter) of that school year; the number of years they had lived in a city since birth; the number of people (besides themselves) that slept in the room with them at night, when they were at home; and whether they had ever repeated a grade at school. We also asked them to select who had raised them or brought them up. If they indicated that they had not been raised by both

60

FLISHER ET AL

their biological mother and their biological father, we coded them as not being raised by both parents. The instrument was translated from English into the other main languages spoken in Cape Town (Afrikaans and Xhosa) and then back-translated by other people who had these languages as home language. It has been used in a number of previous studies (2,6,8), and has been subjected to extensive pilot studies in small groups and classrooms. The test-retest reliability of the items involving lifetime use of substances has been investigated among students attending independent (private) high schools in the Cape Peninsula (22). The kappas (95% confidence intervals [CIs]) were as follows: .85 (.80 –.91) for ever smoked a whole cigarette; .78 (.71–.85) for ever used alcohol, other than a few sips; and .80 (.72–.88) for ever smoked cannabis. Overall, one can conclude that the test-retest reliability for the items involving substance use was substantial. Psychometric data are not available for the items addressing the potential correlates of interest. Students were asked if they had used a fictitious substance (Derbisol), and the five students that responded positively to this item were excluded.

Procedure The selected students completed the questionnaire during a normal school period. The seating was arranged such that confidentiality was preserved. Members of the research team administered the questionnaire, with no school staff being present. After completing the questionnaire, the students placed it in an envelope, which they sealed before handing it in. The level of student participation was satisfactory and no student refused to participate. The study was approved by the Research Ethics Committee, Faculty of Health Sciences, University of Cape Town.

Analysis Prevalence rates and means with 95% CIs were calculated, taking the multistage stratified design into account. The Survey Data Analysis Program (SUDAAN) (23) was used for the analysis. The “without replacement design” option was used and the design stages included were postal code area, school, and grade. Sampling weights were computed using the number of students in the school, the

JOURNAL OF ADOLESCENT HEALTH Vol. 32, No. 1

number of students in the grade (in a specific school), and the number of students sampled from the grade. Prevalence rates were provided for lifetime, previous year, and previous month use of tobacco, alcohol, and cannabis, stratified by grade, race classification, and gender. We also provided the lifetime, previous year, and previous month prevalence rates for binge drinking. The race classifications are as defined by the repealed population Registration Act of 1950. There are dangers of analyzing the data by race classification because the groups do not have anthropological or scientific validity. However, there are differences among the groups for many indicators of health, mediated by political and economic factors (24). For those who had used, and not used, each of cigarettes, alcohol, and cannabis in the previous month, we provided the percentages of students who were male, were in grade 11, and were of each race classification. For the hypothesized correlates, we provided the relevant percentages or means. To further investigate the association between use in the previous month of each substance and the demographic variables and each potential correlate, we constructed a series of generalized estimating equations (GEEs). We used this approach to take the clustering of the schools into account. The dependent variable for each model was whether the substance had been used in the previous month, which was modeled as a binary outcome using the logit link function. The independent variable for each model was a demographic characteristic or a hypothesized correlate. Each model was adjusted for grade and the clustering according to school. Finally, we constructed a further set of GEEs. For each model the dependent variable was whether a specific substance had been used in the previous month (as in the previous models) and the independent variable was a hypothesized correlate. However, we did an analysis of variance to test for interactions among the demographic variables of gender, grade, and race classification (and their interactions) and the dependent variables. Where such significant interactions were found, we included the relevant terms in all the models for that substance. We also tested the interaction between each of the demographic variables (and their interactions) and the independent variables. Where appropriate, we presented the results taking such interactions into account. Each of the models was also adjusted for the clustering according to school. The results of all models are presented as adjusted odds ratios with their 95% CIs.

January 2003


61

Table 1. Prevalence Rates for Substance Use (N ⫽ 2779) Grade 8

Boys: Cigarettes Lifetime Past year Past month Alcohol Lifetime Past year Past month Cannabis Lifetime Past year Past month Girls: Cigarettes Lifetime Past year Past month Alcohol Lifetime Past year Past month Cannabis Lifetime Past year Past month a

Grade 11 a

Blacks

Coloreds

Whites

Blacks

Coloredsa

Whites

21.9 (11.5–32.3) 8.9 (2.2–15.7) 7.7 (3.7–11.6)

43.1 (36.7– 49.5) 27.0 (21.6 –32.2) 27.3 (21.3–33.3)

47.0 (39.3–54.8) 33.7 (24.5– 43.0) 28.4 (19.0 –37.7)

50.7 (41.7–59.7) 29.5 (22.1–36.8) 33.3 (26.0 – 40.5)

63.5 (56.6 –70.3) 50.6 (42.8 –58.3) 46.5 (38.4 –54.6)

56.8 (46.3– 67.3) 40.4 (28.3–52.5) 35.1 (26.7– 43.5)

34.7 (26.5– 43.0) 16.1 (10.9 –21.2) 17.7 (11.8 –23.5)

39.7 (33.3– 46.1) 22.6 (16.9 –28.3) 21.9 (17.2–26.5)

49.8 (39.4 – 60.3) 33.9 (23.0 – 66.4) 22.5 (13.8 –33.2)

55.4 (46.5– 64.4) 35.6 (27.4 – 43.8) 37.6 (29.5– 45.7)

69.0 (61.7–76.4) 54.3 (46.1– 62.5) 47.7 (39.4 –56.0)

72.9 (62.1– 83.7) 62.3 (47.3–77.4) 54.7 (40.2– 69.2)

10.5 (6.0 –15.0) 5.0 (1.5– 8.5) 3.3 (0.9 –5.7)

5.5 (3.3–7.7) 2.9 (1.2– 4.5) 3.0 (1.0 –5.1)

11.6 (7.0 –16.1) 7.1 (3.2–11.1) 3.6 (0.0 –7.6)

23.0 (12.3–33.8) 12.1 (6.3–18.1) 6.0 (2.3–9.8)

37.5 (30.8 – 44.1) 24.4 (17.1–31.7) 18.7 (12.6 –24.7)

31.2 (23.9 –38.6) 23.2 (17.5–28.9) 8.9 (4.6 –13.3)

5.9 (2.3–9.5) 3.9 (1.0 – 6.8) 3.5 (0.6 – 6.4)

45.5 (38.2–52.9) 30.3 (24.1–36.5) 31.3 (24.2–38.4)

47.1 (32.5– 61.7) 41.7 (26.2–57.3) 25.8 (9.9 – 41.7)

5.9 (3.1– 8.8) 4.3 (1.7– 6.9) 3.0 (0.5–5.6)

62.9 (57.9 – 67.8) 46.4 (41.5–51.3) 43.2 (37.6 – 48.8)

58.7 (52.4 – 65.1) 46.8 (41.2–52.5) 42.4 (36.9 – 48.0)

16.2 (8.8 –23.6) 6.7 (3.8 –9.6) 7.7 (4.1–11.2)

32.5 (27.2–37.8) 18.5 (13.2–23.8) 20.5 (15.4 –25.6)

52.8 (37.9 – 67.6) 39.8 (23.5–56.2) 25.0 (11.1–38.9)

18.3 (14.1–22.4) 9.6 (6.2–13.1) 8.4 (5.0 –11.8)

55.6 (49.3– 62.0) 41.3 (35.6 – 47.0) 33.2 (27.0 –39.4)

75.7 (67.1– 84.4) 64.1 (53.4 –74.7) 56.8 (45.8 – 67.7)

1.2 (0.0 –2.9) 0.7 (0.0 –2.0) 0.7 (0.0 –2.0)

4.7 (2.3–7.1) 1.8 (0.4 –3.1) 2.7 (1.1– 4.3)

7.1 (1.7–12.5) 6.2 (0.6 –11.8) 2.4 (0.0 –5.4)

2.7 (0.4 –5.0) 1.0 (0.0 –2.5) 0.7 (0.0 –2.0)

14.1 (10.5–17.7) 8.7 (5.9 –11.5) 5.0 (2.8 –7.2)

27.1 (19.4 –34.8) 18.8 (12.0 –25.7) 6.6 (2.8 –10.4)

Coloreds refers to Asian, European, and African descent.

Results The total sample size was 2946. Of these, 16 were Asian and were excluded owing to their small number, and 151 did not report their gender, grade, or race classification. The analysis was confined to the remaining 2779 students. Of these, 1617 (58.2%) were female; 1413 (50.8%) were in grade 11; and 787 (28.3%) were black, 1455 (52.4%) colored (derived from Asian, European, and African ancestry) and 537 (19.3%) white. The prevalence rates for previous month (recent) use of tobacco, alcohol, and cannabis were 27%, 31%, and 7% respectively. Table 1 displays the prevalence rates (with their 95% CIs) for lifetime, previous year, and previous month use of cigarettes, alcohol, and cannabis, stratified by gender, grade, and race classification. For black students in each grade, the rates were higher for boys. For black students in grade 8, the CIs for boys and girls did not overlap in the majority of cases, whereas in grade 11 they did not overlap in any cases. Conversely, for colored and white students, there were no clear trends in the gender differentials, and the CIs overlapped in all cases except for cannabis smoking by colored students in grade 11.

There was a trend for the rates to increase from grade 8 to grade 11, with the 95% CIs not overlapping in the majority of cases. The exception to this trend was black females, for whom there was no consistent trend. For black females, the CIs for the two grades overlapped in all cases. Finally, the only clear and consistent trend for race classification was that black females had lower rates for each substance than their colored and white counterparts. The CIs for black females did not overlap with those of colored or white females, except for cannabis use by grade 8 females. When adjusting for grade and the clustering effect owing to school, there was a significant association between recent use of cigarettes, alcohol, and cannabis and the number of days absent in the first quarter, the number of years lived in a city, and repeating a grade, (Table 2). Conversely, we did not detect a significant association between recent use of any of these substances and the number of people with whom the students shared a room at night and not being raised by both parents (Table 2). Three of these findings remained valid when adjusting for demographic variables and taking significant interactions among demographic variables

FLISHER ET AL

62


Table 2. Potential Correlates of Substance Use in the Previous Month: Descriptive Data and Results of Generalized Estimating Equations (N ⫽ 2779)a Cigarettes

Alcohol b

Number of days absent in first term (mean)c Number of years lived in a city (mean)d Number of people sharing room at night (mean) Repeated a grade (%) Not raised by both parentse (%)

Cannabis b

No

Yes

ORb (95% CI)

3.05

4.1

1.3 (1.3–1.3)*

No

Yes

No

Yes

Odds ratio (95% CI)

2.9

4.2

1.5 (1.5–1.6)*

3.0

3.8

1.4 (1.4 –1.4)*

12.5

13.9

1.2 (1.2–1.2)*

12.5

14.0

1.2 (1.2–1.3)*

12.7

14.8

1.6 (1.5–1.6)*

1.7

1.5

1.0 (0.9 –1.1)

1.7

1.4

1.0 (0.8 –1.1)

27.1 43.3

36.7 40.3

1.6 (1.3–1.9)* 1.1 (0.9 –1.3)

28.6 43.3

44.5 38.1

2.3 (1.5–3.5)* 1.1 (0.8 –1.5)

1.7 26.4 44.1

1.39 39.5 40.0

OR (95% CI)

0.9 (0.9 – 0.98)* 1.7 (1.4 –2.0)* 1.0 (0.9 –1.2)

* p ⱕ .05. Percentages, means and odds ratios are adjusted for grade and clustering according to school. Percentages are column percentages. b Each row refers to a separate model. c Odds ratios refer to an increase of 10 days. d Odds ratios refer to an increase of 5 years. e Refers to biological parents. a

and the independent variables into account (Table 3). Specifically, the number of days absent in the first term and the number of years lived in a city remained significantly associated with recent use of all three substances for all subgroups of students. Also, the absence of a significant association between the number of people sharing a room with the student at night and recent use of any of the substances and was confirmed for all subgroups of students. However, the remaining findings were more complex. Repeating a grade, for example, was significantly associated with substance use for subsets of demographic subgroups and substances. Specifically, it was significantly associated with recent use of cigarettes and alcohol by colored students, and with alcohol use by black grade 8 students. Finally, not being raised by both parents was significantly associated with recent cigarette smoking by black and colored students, alcohol use by colored students, and cannabis use by female students (Table 3).

Discussion This study has contributed to knowledge by producing contemporaneous prevalence data for substance use by high school students in Cape Town and investigating the associations among recent use of cigarettes, alcohol, and cannabis, and key psychosocial correlates. Furthermore, there were important differences in the results when adjusting for significant confounders and taking into account significant interactions between demographic variables and potential psychosocial correlates. This emphasizes the

importance of addressing demographic variables analytically if one is to avoid obscuring important differences among groups (25). It is, however, important to acknowledge the limitations of the study. First, the sample was limited to students who were present at school on the day the study was undertaken. Dropouts and absentees were not included. There is evidence from previous studies in Cape Town (10) and elsewhere (26) that rates of substance use are higher for these subgroups. Indeed, the present study found that recent use of tobacco, alcohol, and cannabis was associated with the number of days absent in the first quarter of the school year. However, data for those who actually attend school are relevant for school-based intervention programs. Second, the study relied on selfreport data. A proportion of those who tend to exaggerate their substance use may have been excluded from the study by the omission of those who answered affirmatively to use of a fictitious drug. Although we went to great lengths to ensure confidentiality and anonymity, we had no way of assessing underreporting of substance use. The prevalence rates reported above may thus represent low estimates of the actual prevalence rates. Finally, the study was cross-sectional, which limits the extent to which conclusions can be drawn about the causal nature of the associations between the correlates and substance use. Comparisons of the prevalence rates reported here with international studies are hampered by differences in sampling strategies, the ages of students in a grade, and question formats. However, it would

January 2003


63

Table 3. Odds Ratios (95% Confidence Intervals) for Use of Each Substance in the Previous Month for Those Variables Qualifying for Inclusion in the Generalized Estimating Equations by Selected Demographic Variables (N ⫽ 2779) Cigarettes

Alcohol b

a

Demographic Number of days absent in first terme

Number of years lived in a cityf Number of people sharing room at night Repeated a grade

Not raised by both parentsg

Black Colored White — — Black Colored White

Black Colored White

OR (95% CI’s) 1.1 (1.1–1.1)* 1.9 (1.8 –2.0)* 2.2 (2.1–2.3)* 1.2 (1.1–1.2)* — 1.1 (0.7–1.7) 2.4 (2.0 –2.9)* 1.3 (0.9 –1.8)

Cannabis c

a

Demographic —

— — Grade 8 Black Colored White Grade 11 Black Colored White 0.5 (0.3– 0.9)* Black 1.4 (1.1–1.8)* Colored 1.1 (0.7–1.8) White

OR (95% CI’s)

a

Demographic

ORd (95% CI’s)

1.4 (1.3–1.4)*

—

1.2 (1.2–1.2)*

1.2 (1.2–1.3)* —

—

1.5 (1.4 –1.6)*

Male Female

0.9 (0.6 –1.4) 2.4 (1.5–3.9)*

2.3 (1.5–3.6)* 3.1 (2.3– 4.1)* 1.3 (0.7–2.4) 1.1 (0.7–1.7)* 1.5 (1.2–2.0)* 0.7 (0.3–1.2) 0.9 (0.6 –1.4) 1.6 (1.3–2.0)* 1.0 (0.7–1.5)

* p ⱕ 0.05. a Refers to levels of the demographic variables of gender, grade, or race classification for which there were interactions with the independent variable. “Coloreds” refers to Asian, European, and African descent. b Adjusted for gender, grade, race classification, gender ⫻ grade; gender ⫻ race classification; grade ⫻ race classification; and gender ⫻ grade ⫻ race classification. c Adjusted for gender, grade, race classification, gender ⫻ grade; and gender ⫻ race classification. d Adjusted for gender, grade, race classification, and gender ⫻ race classification. e Odds ratios refer to an increase of 10 days. f Odds ratios refer to an increase of 5 years. g Refers to biological parents.

appear that the rates are comparable with those observed in comparable studies conducted elsewhere (14,27,28). The only South African study with which the results are comparable was conducted in 1990 (2). Although the sampling strategies differed between the two studies, both aimed to produce samples that were representative of all students in Cape Town in the selected grades. It was found that rates of cigarette and alcohol use had remained fairly constant, whereas rates of cannabis use had almost doubled. This may be owing to (a) more vigorous enforcement of laws aiming to reduce drug use in the United States and other Western countries, resulting in the exploitation of new markets such as those in South Africa; and (b) increased access to cannabis because the borders have become more open with the demise of apartheid. A striking finding was the low prevalence rates for black females. This is reflected in the low prevalence rates for black females compared with black males, and compared with colored and white females. Furthermore, the rate did not increase between grade 8 and 11, which indicates that whatever protective factors were operating continue to exert

their effect until grade 11. This low prevalence for black females is consistent with previous findings in Cape Town (2) and South Africa as a whole (1). It is necessary to identify the protective factors that prevent black female adolescents from acquiring the higher rates of substance use demonstrated by black males and colored and white students of both genders. One may be able to take steps to ensure the continued operation of the protective factors for black females and their enhancement for other students. Although qualitative studies involving drinking among adolescent boys in Cape Town have been conducted (29,30), there are no studies involving drinking among adolescent girls or use of other substances among boys or girls. There is an urgent need to fill this gap. The association between school absenteeism and substance use has been documented in previous studies internationally (12), but has not been studied in South African populations. There could be a direct causal relationship between absenteeism and substance use in either or both directions, or both the absenteeism and substance use could reflect a personality trait such as unconventionality (12). What-

64

FLISHER ET AL

ever the nature of the association, this finding suggests that interventions to improve school attendance should comprise a component of a comprehensive approach to the prevention of substance abuse. An association between duration of residence in an urban area and substance use has been documented in a study conducted in 1990 among black high-school students in Cape Town (8). The present finding adds to the robustness of the previous finding by replicating it in a contemporaneous sample of students of all race classifications. Rapid urbanization is frequently accompanied by housing difficulties, crime, poverty, unemployment, and separation from extended families, which may in turn lead to substance use (8). One might expect that substance use would be associated with the number of people with whom the student shares a night because this is associated with economic disadvantage, which has been associated with substance use (13,19). However, contradictory evidence about the relationship between substance use and economic disadvantage has appeared (31), with which the present findings are consistent. It would appear that the relationship between economic disadvantage and substance use is more complicated than previously thought. Repeating a grade was significantly associated with recent cigarette use by colored students and recent alcohol use by colored students and black grade 8 students. Although the association between poor scholastic performance and substance use has been repeatedly documented (14 –17), it is not clear why this applied only to subsets of students for recent cigarette and alcohol use only. Not being raised by both parents was associated with cigarette and alcohol use by colored students and with cannabis use by females. The finding that not being raised by both parents is associated with substance use is consistent with previous research (18 –21). Among the explanations that have been offered for this relationship are that families with one biological parent are more likely to be characterized by circumstances which in turn are associated with substance use, such as economic hardship, increased residential mobility, and psychosocial stress (20). However, as with the previous variable of repeating a grade, it is difficult to explain why not being raised by both biological parents is associated with substance use only for subgroups of students for specific substances. Further research is needed to address this issue.


For black students, there was an inverse association between not being raised by both parents and cigarette smoking. Although this finding certainly requires replication, it is consistent with an analysis of a national household survey in the United States (18). This study found that white and Latino adolescents from two-biological-parent families are least likely to use alcohol and marijuana, whereas black adolescents from such families are among the most likely users of these substances (18). We thank the principals, staff, and students at the schools where the research was undertaken, and the support of the Western Cape Education Department.

References 1. Rocha-Silva L, De Miranda S, Erasmus R. Alcohol, Tobacco and Other Drug Use among Black Youth. Pretoria: Human Sciences Research Council, 1996. 2. Flisher AJ, Ziervogel CF, Chalton DO, et al. Risk-taking behaviour among Cape Peninsula high-school students: Parts III to V. S Afr Med J 1993;83:477–85. 3. Flisher AJ, Cruz C, Eaton L, et al. Review of South African Research Involving Adolescent Health. Report Submitted to the Youth Development Trust. Cape Town: Department of Psychiatry, University of Cape Town, 1999. 4. Parry CDH, Bennetts AL. Alcohol Policy and Public Health in South Africa. Cape Town: Oxford University Press, 1998. 5. Dorrington R, Bourne D, Bradshaw D, et al. The Impact of HIV/AIDS on Adult Mortality in South Africa. Cape Town: Medical Research Council, 2001 (Technical Report). 6. Flisher AJ, Chalton DO. Adolescent contraceptive non-use and covariation among risk behaviors. J Adolesc Health 2001;28: 235–41. 7. Flisher AJ, Ziervogel CF, Chalton DO, et al. Risk-taking behaviour among Cape Peninsula high-school students: Parts IX and X. S Afr Med J 1996;86:1090 –8. 8. Flisher AJ, Chalton DO. Urbanisation and adolescent risk behaviour. S Afr Med J 2001;91:243–9. 9. Strebel P, Kuhn L, Yach D. Determinants of cigarette smoking in the black township population of Cape Town. J Epidemiol Community Health 1989;43:209 –13. 10. Flisher AJ, Chalton DO. High-school dropouts in a workingclass South African community: Selected characteristics and risk-taking behaviour. J Adolesc 1995;18:105–21. 11. Morojele NK, Parry CDH, Ziervogel CF, et al. Prediction of binge drinking intentions of female school-leavers in Cape Town, South Africa, using the theory of planned behaviour. J Subst Use 2000;5:240 –51. 12. Jones SP, Heaven CL. Psychosocial correlates of adolescent drug-taking behaviour. J Adolesc 1998;21:127–34. 13. Lowrie R, Kann L, Collins JL, Kolbe LJ. The effect of socioeconomic status on chronic disease risk behaviors among US adolescents. JAMA 1996;276:792–7. 14. Resnick MD, Bearman PS, Blum RW, et al. Protecting adolescents from harm. Findings from the National Longitudinal Study on Adolescent Health. JAMA 1997;278:823–32. 15. Jackson C, Hendriksen L, Dickinson D, Levine DW. The early use of alcohol and tobacco: Its relation to children’s compe-

January 2003

16.

17.

18.

19.

20.

21.

22.

23.

24.

tence and parents’ behavior. Am J Public Health 1997;87:359 – 64. Singh H, Mustapha N. Some factors associated with substance abuse among secondary school students in Trinidad and Tobago. J Drug Educ 1994;24:83–93. Yamada T, Kendix M, Yamada T. The impact of alcohol consumption and marijuana use on high school graduation. Health Econ 1996;5:77–92. Amey CH, Albrecht SL. Race and ethnic differences in adolescent drug use: The impact of family structure and the quantity and quality of parental interaction. J Drug Issues 1998;28:283–98. Duncan TE, Duncan SC, Hops H. Latent variable modeling of longitudinal and multilevel alcohol use data. J Stud Alcohol 1998;59:399 –408. Hoffmann JP, Johnson RA. A national portrait of family structure and adolescent drug use. J Marriage Fam 1998;60: 633–45. Shucksmith J, Glendinning A, Hendry L. Adolescent drinking behaviour and the role of family life: A Scottish perspective. J Adolesc 1997;20:85–101. Flisher AJ, Evans J, Muller M, Lombard CL. Test-retest reliability of self-reported adolescent risk behaviour. J Adolesc 2003 (in press). Shah BV, Barnwell BG, Bieler GS. SUDAAN User’s Manual, Release 7.5. Research Triangle Park, NC: Research Triangle Institute, 1997. Ellison GTH, De Wet T, Ijsselmuiden CB, Richter L. Desegregating health statistics and health research in South Africa. S


65

Afr Med J 1996;86:1257–62. 25. Headen SW, Bauman KE, Deane GD, Koch GG. Are the correlates of cigarette smoking initiation different for black and white adolescents? Am J Public Health 1991;81:854 –8. 26. Eggert LL, Seyl CD, Nicholas LJ. Effects of a school-based prevention program for potential high school dropouts and drug abusers. Int J Addict 1990;25:773–801. 27. National Institutes of Health. Epidemiologic Trends in Drug Abuse. Volume II: Proceedings of the International Work Group on Drug Abuse. Bethesda, MD: National Institutes of Health, Division of Epidemiology, Services and Prevention Research, 1999. 28. Currie C, Hurrelmann K, Settertobuite W, et al. Health and Health Behaviour Among Young People. Health Behaviour in School-aged Children: A WHO Cross-National Study (HBSC) International Report. Geneva: World Health Organisation, 2000. 29. Ziervogel CF, Ahmed N, Flisher AJ, et al. Adolescent alcohol misuse: A qualitative investigation. Int Q Community Health Educ 1997–98;17:25–41. 30. Ziervogel CF, Morojele NK, Van der Riet J, et al. A qualitative investigation of alcohol drinking among male high school students from three communities in the Cape Peninsula, South Africa. Int Q Community Health Educ 1998;17:271–95. 31. Friedman AS, Ali A. The interaction of SES, race/ethnicity and family organisation (living arrangements) of adolescents, in relation to severity of use of drugs and alcohol. J Child Adolesc Subst Abuse 1997;7:65–74.