Epidemiology of substance use among secondary school learners in Atteridgeville , Gauteng

Adolescent substance use is a serious concern owing to potential shortand long-term outcomes for the health and safety of the individual as well as the number of associated adverse social consequences.1-6 The association between heavy episodic adolescent alcohol use and road traffic accidents, suicide, violence and high-risk sexual behaviour is well documented.2,3 There is evidence that excessive alcohol use in adolescence leads to alcohol dependence, illicit substance use and criminal behaviour in adulthood.2,4 Short-term consequences of cannabis use include high-risk sexual behaviour and an increase in sexually transmitted infections.5 Frequent use of cannabis can result in dependence and psychosis.6

Introduction.Adolescent substance use has a number of adverse consequences for both the individual and society.Anecdotal evidence suggested the existence of a serious substance use problem among learners in Atteridgeville, part of the City of Tshwane Metropolitan Municipality in Gauteng, South Africa.
Objectives.The objectives of the study were to determine the prevalence rates and age of initiation of substance use among learners attending secondary schools in Atteridgeville, and the factors, if any, associated with cannabis use.
Methods.This was a cross-sectional study.A cluster sampling technique was employed.Twenty-two of the total of 191 grade 8 -11 classes in the nine Atteridgeville secondary schools were randomly selected, with all 895 learners in the selected classes being invited to participate.Self-administered questionnaires were used to collect data.Statistical analysis was conducted using the survey estimation commands in STATA 10.0..

Results
. The lifetime prevalence rates for the three most commonly used substances were 51.4% (95% confidence interval (CI) 41.5 -61.5%) for alcohol, 25.2% (95% CI 17.1 -33.3%) for cigarettes and 13.2% (95% CI 8.3 -18.2%) for cannabis.Alcohol was also found to have the lowest mean age of initiation at 14.6 years (standard deviation 2.0).Based on CRAFFT screening scores, 30.3% (95% CI 24.5 -36.1%) of learners met the criteria for possible problem substance use.Following multivariate analysis, the factors found to be significantly associated with lifetime cannabis use were age, gender, lifetime cigarette use, lifetime alcohol use, the number of illicit drug users among the learners' five closest friends, and an older sibling with a history of illicit drug use.

Conclusion.
The results of the study suggest that substance use among learners in Atteridgeville is widespread and that a comprehensive intervention strategy is required.increases the likelihood of adolescent substance use. 18Simons-Morton 19 found that alcohol use by an adolescent was related to the number of drinkers among the individual's five closest friends.Peers were found to have had a greater influence on learners initiating substance use than the family, school and media among learners in the Free State. 11e present study was conducted in Atteridgeville, South Africa.Atteridgeville is a township located south-west of the Pretoria central business district and forms part of the City of Tshwane Metropolitan Municipality. 20Established in 1939 by the then government as a township for black Africans, it is today home to approximately 250 000 people. 20There are nine high schools in Atteridgeville.A number of media reports since 2006 have highlighted the existence of substance use, particularly the use of a combination of cannabis, heroin and cutting agents, a combination referred to locally as nyaope, by learners in Atteridgeville.The objectives of this study, conducted among learners attending high schools in Atteridgeville, were to determine the prevalence of use of various substances, the age of initiation of substance use, the proportion of learners who may require further intervention for their substance use, and the factors associated with cannabis use.

Design and sampling
This was a cross-sectional study of learners attending the nine high schools in Atteridgeville.Because grade 12 learners were preparing for national examinations they were excluded from the survey, as were learners who were absent on the day of the survey.A singlestage cluster sampling technique was employed with each class of learners being regarded as a primary sampling unit.A single sequential list was compiled of all 191 grade 8 -11 classes in the nine secondary schools.Twenty-two classrooms were then selected from this list by simple random sampling.All the learners in the selected 22 classrooms were invited to participate in the survey.

Measurement tools
The measurement tool for learners was a self-administered questionnaire, available in English and Northern Sotho (Sepedi).It included questions related to socio-demographic characteristics, socio-economic characteristics, scholastic characteristics, use of alcohol, cigarettes and illicit substances, sites where substances were obtained and family and peer alcohol and drug use.It also comprised of the CRAFFT screening instrument.The CRAFFT score is derived from six questions. 21Based on the findings and recommendations of Knight et al., 21 we regarded a CRAFFT score of 2 or more as indicating possible problem substance use requiring further assessment. 21The CRAFFT screening tool was preferred to other instruments because it was designed specifically for adolescents and screens for multiple substances, and the questions are easy to understand. 21,22The validity and reliability of the CRAFFT tool have been tested in various settings, 21,23,24 but it should be noted that no validation studies on the CRAFFT tool have been conducted in South Africa.Knight et al. 21found that validity of the instrument was not significantly influenced by gender, race or age.In a recent review by Dhalla et al. 24 of the CRAFFT's psychometric properties, the instrument's internal consistency was found to be between 0.65 and 0.86, sensitivity between 0.61 and 1.00 and specificity between 0.33 and 0.97.The values for validity were obtained at optimal cutpoints. 24

Ethical considerations and procedures
The study received ethics approval from the Research and Ethics Committee of the University of Pretoria.Authorisation to conduct the study was granted by the Gauteng Department of Education.A 'passive' parental consent procedure was employed.A leaflet was sent to the learners' guardians, which gave them information about the nature of the research and their right to refuse their child's participation.A refusal form was attached to the leaflet.If this form was not returned, it was assumed that consent had been given by a learner's guardian.Informed assent was then sought from those learners whose parents did not refuse to allow them to participate.Those learners whose parents did not want them to participate were not given a questionnaire, and those who themselves wished not to participate were asked to leave the questionnaire blank.

Data analysis
All data were analysed using the statistical package STATA version 10.0.As cluster sampling was employed in the study, statistical analysis needed to take intra-cluster correlation into account.This was accomplished by using the survey estimation commands in STATA.Prevalence rates with 95% confidence intervals (CIs) were calculated for lifetime and past month use of various substances.Frequencies and summary statistics were obtained for age of initiation of substance use.A total CRAFFT score was calculated for each participant, and the proportion of learners with a CRAFFT score of two or more was determined.Multiple logistic regression was used to determine the factors associated with lifetime cannabis use.The independent variables that were significantly associated with lifetime cannabis use at a significance level of 90% in bivariate analysis were entered into the initial full regression model.Backwards stepwise regression was then carried out.

Results
A total of 22 classes comprising 895 learners were invited to participate in the study.Eight hundred and nine learners completed questionnaires, which equates to a response rate of 90.4%.A total of 81 (9.1%) learners who were invited to participate were absent on the days of data collection.Excluding those who were absent, the response rate would be 99.4% (809/814).
As the sample of classes was randomly selected, the participants were not evenly distributed by grade.Of the 809 participants, 232 (28.7%) were in grade 8, 101 (12.5%) in grade 9, 194 (24.0%) in grade 10 and 282 (34.9%) in grade 11.The mean age of learners participating in the study was 16.2 years (standard deviation (SD) 1.8).The sample consisted of marginally more females (50.3%) than males (49.8%).All the learners were black Africans.

Prevalence rates of substance use
Alcohol was the substance most commonly used by the learners, followed by cigarettes and cannabis.Over half of the study participants (51.4%) had consumed alcohol at some point in their lives.The prevalence of alcohol use for the past month was 18.1%.Of the 146 learners who had reported alcohol use in the past month, 75 (51.4%)exhibited a binge drinking pattern (5 or more drinks on a single day) on at least one day in the past month.Table 1 details the lifetime prevalence rates of various substances (with 95% CIs) as well the pastmonth prevalence rates of the most commonly used substances.
When substance use was stratified by grade and gender, the highest prevalence rates among the participants for lifetime and pastmonth alcohol, cigarette and cannabis use was found among grade 11 males.For all grades, males had higher lifetime alcohol, cigarettes and cannabis prevalence rates than females.With the exception of grade 8, males had higher rates of past-month alcohol use, pastmonth binge drinking and past-month cigarette use than females.For males there was a trend for past-month alcohol use, past-month binge drinking and past-month cannabis use to increase with grade, but with most of the 95% CIs overlapping.There were no consistent grade trends for females for past-month use of any of the substances.The lifetime and past-month prevalence rates for alcohol, cigarettes and cannabis stratified by grade and gender are set out in Table 2.

Age of initiation of substance use
The mean age in years of first use of alcohol, cigarettes, cannabis and nyaope among the participants was 14.6 (SD 2.0), 14.8 (SD 2.3), 15

Factors associated with lifetime use of cannabis
The variables that maintained a significant association with lifetime cannabis use following multiple logistic regression analysis are displayed in Table 4. Male gender, age 17 years and older, lifetime cigarette use, lifetime alcohol use and older sibling illicit drug use were associated with significantly higher odds of lifetime cannabis use.An increase in the number of illicit drug users among five closest friends also significantly increased the odds of lifetime cannabis use.

Discussion
The finding that alcohol is the substance most commonly used by learners in Atteridgeville is consistent with studies conducted among learners in other parts of South Africa. 10,11,25hen considering substance use by gender, males in this study generally had higher prevalence rates than females.A striking finding of the Atteridgeville study is the high lifetime prevalence of alcohol use in female learners when compared with black female learners in previous South African studies. 10,17The higher rates of lifetime alcohol use in female learners in Atteridgeville may be part of a national trend of increased use of alcohol among black female learners in the period since those studies were conducted.The wide gender difference in cannabis use in this study has been demonstrated previously among black learners in other studies. 9,10,17isser and Routledge 26 suggest that differing male and female social roles could be one of the explanations for gender differences in the prevalence of substance use.According to Masitsa, 11 cannabis use by females has historically been considered taboo amongst black people in South Africa.
In this study alcohol was found to have the lowest mean age of initiation, followed by cigarette smoking and cannabis.This study confirms the findings of the first South African National Youth Risk Behaviour Survey 7 and a study conducted among primary school learners in the Tshwane metropolitan area, both of which reported that a significant number of learners use alcohol and cigarettes for the first time while still at primary school. 7,12In addition, the Atteridgeville study found that some learners had used cannabis while still at primary school, which is consistent with the findings of Visser and Moleko. 12This finding is particularly concerning as a younger age of initiation of illicit substance use has been found to increase the risk of illicit substance abuse at a later stage. 27is study did not specifically address the frequency and quantity of each substance being used, in order to keep the questionnaire as concise as possible.Instead the CRAFFT screening tool was used to differentiate between respondents who may have just experimented with a substance and those who may have a substance use problem.
Because the CRAFFT is a composite score based on 6 questions, it was only possible to calculate a CRAFFT score for 697 learners as a number of learners had not answered one or more of the 6 questions.The CRAFFT screening test results for Atteridgeville does suggest that that a number of learners may meet the criteria for problem use, abuse and dependence.The CRAFFT results, however, may not have been representative of the learners' recent substance use, as the questions refer to lifetime behaviour.
Multivariate analysis was used to identify the factors that were significantly associated with the lifetime use of cannabis.The finding of a significant association between learners' cannabis use and older sibling illicit drug use is consistent with the findings of a crossnational European study. 28The finding of a significant association between cannabis use and number of the five closest friends who use cannabis is consistent with the findings of Simons-Morton 19 regarding alcohol use.It is also in keeping with the results of a study by Chabrol et al. 29 that the number of peers using cannabis was significantly associated with cannabis use among French secondary school learners.There may be a number of explanations for this finding in the Atteridgeville study.The learner may be selecting friends with similar risk-taking behaviours, conforming to subcultural norms or being actively influenced by peers. 30,31Lifetime cannabis use was found to be significantly associated with lifetime cigarette use and lifetime alcohol use in this study.The association between cannabis and alcohol use is consistent with the findings of Taylor et al. 10 in KwaZulu-Natal.These findings are also in keeping with those of Faeh et al. 15 that alcohol, tobacco and cannabis use cluster in adolescents, i.e. a high proportion of adolescents use two or more of these substances.
This study had several limitations.It relied on self-reporting by learners, and it has been suggested that in survey research on substance use there may be a tendency towards under-reporting. 32n order to maximise reporting in this study, guarantees of confidentiality were given to the participants and an anonymous self-administered questionnaire was used.A number of learners failed to answer questions related to the use of 'hard' drugs.It is not clear whether this was deliberate under-reporting.The exclusion of learners absent on the day of the survey may have resulted in an underestimate of the prevalence of substance use, as a higher rate of absenteeism has been found in those who use substances. 17The cross-sectional design of the study limits the inferences that can be drawn from some of the associations found in the study, owing to uncertainty about the temporal relationship.The generalisability of the findings of this study are limited.As grade 12 learners were not included in the study, the findings only apply to grade 8 -11 learners in Atteridgeville.Generalisation of these findings to learner populations in other communities should be done with caution, as there is likely to be variation in substance use patterns between communities.

Conclusion
This study has confirmed some important findings from previous South African and international substance use research among adolescents, including the important role of peers in an adolescent's life.In addition the results have confirmed that the concerns regarding substance use among learners in Atteridgeville are not unfounded.The high prevalence rates, combined with medical and psychosocial consequences of using these substances, highlight the public health importance of the problem and the need for urgent intervention.School-based interventions to prevent substance use should start at primary school level and incorporate methods that have been shown to be effective in countering the effect of peer pressure.It is important that national and provincial initiatives to reduce substance use among learners be strengthened, including full implementation of South Africa's National Drug Master Plan. 33urther research should be conducted, including similar studies in other communities, longitudinal studies to establish temporality, in-depth qualitative studies to understand the contemporary social determinants, and validation studies to establish the most appropriate screening tool for adolescent substance use in South Africa.

Fig. 1 .
Fig. 1.Distribution of age of initiation of alcohol, cigarettes, cannabis andnyaope use among study participants.

Table 2 . Prevalence rates (% (95% CI)) of alcohol, cigarette and cannabis use, by grade and gender
SD 2.2) and 15.8 (SD 2.6), respectively.The distribution of age of first use of alcohol, cigarettes, cannabis and nyaope is displayed in Fig.1.Some of the participants had first started using alcohol, cigarettes and cannabis while still at primary school.A significant number of participants who had ever used alcohol had had their first alcoholic drink by the age of 14, with the vast majority of participants having had their first drink by the age of 16.First use of cigarettes and cannabis among the participants occurred mainly between the ages of 14 and 17 years and 15 and 17 years, respectively.First use of nyaope most commonly occurred at 17 years of age.

Table 4 . Multivariate analysis of factors associated with lifetime use of cannabis*
Multiple logistic regression.Variables originally entered in model: grade categories, age categories, gender, number of days absent categories, repeated a grade, lifetime alcohol use, lifetime cigarette use, parental illicit drug history, sibling illicit drug history and number of illicit drug users among five closest friends. *