Article Text

Download PDFPDF

Rising alcohol consumption and a high prevalence of problem drinking in black men and women in Cape Town: the CRIBSA study
  1. Nasheeta Peer1,
  2. Carl Lombard2,
  3. Krisela Steyn3,
  4. Naomi Levitt3,4
  1. 1Non-communicable Disease Research Unit, Medical Research Council (MRC), Durban, South Africa
  2. 2Biostatistics Unit, MRC, Cape Town, South Africa
  3. 3Chronic Disease Initiative for Africa, Department of Medicine, University of Cape Town (UCT), Cape Town, South Africa
  4. 4Division of Endocrinology and Diabetes, Department of Medicine, UCT, Cape Town, South Africa
  1. Correspondence to Dr Nasheeta Peer, Non-communicable Disease Research Unit, Medical Research Council, 491 Ridge Road, Overport, Durban, 4001, South Africa; nasheeta.peer{at}


Background To determine the prevalence and determinants of problematic alcohol use (CAGE ≥2) in 25–74-year-old black population in Cape Town in 2008/2009 and examine the changes in self-reported alcohol consumption between 1990 and 2008/2009 in 25–64-year-olds.

Methods In 2008/2009, a representative cross-sectional sample, stratified for age and sex, was randomly selected from the same townships sampled in 1990. Socio-demographic characteristics, the ability to cope with psychosocial stress (sense of coherence) and adverse life events were determined. Ordinal logistic regression analysis assessed the determinants of problem drinking.

Results There were 1099 participants, 392 men and 707 women, in 2008/2009. Prevalence of alcohol consumption in 2008/2009 (men: 68.5%, 95% CI 62.7 to 73.7; women: 27.4%, 95% CI 23.7 to 31.5) was higher than in 1990 (men: 56.7%, women: 15.1%). Prevalence of problem drinking was significantly higher in men (49.7%, 95% CI 44.6 to 54.9) than in women (18.1% 95% CI 15.3 to 21.2) (p<0.001). In men, greater alcohol use was associated with >7 years of education (p=0.012), being unemployed compared with employed (p=0.008) and coping poorly with stress (OR 1.02, 95% CI 1.01 to 1.05, p=0.042), and in women with spending more than half their life in the city (p<0.001) and coping poorly with stress (OR 1.02, 95% CI 1.01 to 1.04, p=0.039). The odds for greater alcohol use with increasing number of adverse life events, after adjusting for the other factors, was significant in men (OR 1.11, 95% CI 1.02 to 1.19, p=0.010) and women (OR 1.09, 95% CI 1.03 to 1.16, p=0.005).

Conclusions Problem drinking is a major problem in this population and requires urgent interventions to curtail the misuse.

Statistics from


The growing threat posed by non-communicable diseases (NCDs) has been recognised by the United Nations with the convening of the high-level meeting on NCDs in 2011. Notably, alcohol consumption was identified as one of the four major determinants of NCDs, along with tobacco use, unhealthy diets and lack of exercise.1 In addition to injuries and violence, it is strongly related to many cardiovascular diseases (CVDs), including hypertensive disease, haemorrhagic stroke and atrial fibrillation1–3 and has been linked to eight different cancers, various liver diseases and pancreatitis.1–3

In South Africa, alcohol misuse is a major social and health problem with approximately one in four adult men and one in 10 adult women experiencing symptoms of alcohol-related problems.4 The health burden is considerable with 7.1% of mortality attributable to alcohol misuse in 2000, more than double the global mortality average of 3.2%.2 ,5 Of the deaths attributable to alcohol misuse, injuries were the most frequent followed by CVDs.5 The leading contributors to the overall burden (morbidity and mortality) attributable to alcohol were interpersonal violence (39.0%), neuropsychiatric conditions (18.4%) and road traffic injuries (14.3%).

These data underscore the need for regular monitoring of alcohol consumption and misuse in South Africa. Although the most recent national surveillance data on alcohol use was obtained in 2008,6 on a regional level these have not been examined in the black population of Cape Town in almost two decades. Regular monitoring of alcohol use is essential to raise awareness among policy makers and the public of the patterns of alcohol consumption and the magnitude of problematic alcohol use so as to place greater emphasis on these conditions in the health policy agenda.

The Cardiovascular Risk in Black South Africans (CRIBSA) study aimed to determine the prevalence of alcohol use in black men and women in Cape Town in 2008/2009 and to compare these findings with a similar study conducted in 1990. The time span between the two studies was marked by dramatic political transformation with the first national democratic elections held in 1994. Accompanying this were major economic and social changes which may have influenced the pattern of alcohol consumption during this time period. This population and city were selected because black Africans comprise 80% of the nation's population7 and the Western Cape Province, of which Cape Town is the capital, has among the highest prevalence of alcohol intake and problem drinking in the country.6 Furthermore, this study examined the prevalence and determinants (socio-demographic and psychosocial stress) of problem drinking using the CAGE questionnaire in 2008/2009. These associations were examined because they have been found to be related to alcohol use and misuse.8–10


Study population and sampling procedure

In 2008/2009, a sample of 25–74-year-old men and women in the predominantly black residential areas of Langa, Guguletu, Crossroads, Nyanga and Khayelitsha in Cape Town was selected to participate in this cross-sectional study. The residents of these townships, historically earmarked solely for the black African population, today remain over 90% black. These areas were selected to ensure comparability with a 1990 study, the methodology of which has been described previously.11 The sampling procedure for the current study included a 3-stage cluster-sampling and has been described in detail elsewhere.12 The prespecified age and gender quotas included disproportionate sampling across age groups to ensure at least 50 men and women in each sex category. Among individuals excluded were those on tuberculosis or antiretroviral therapy or those who had received cancer treatment within the previous year.

Data collection

Socio-demographic data, including education level, employment status, housing type and assets defining wealth, were collected by questionnaires administered by trained fieldworkers. The assets recorded included ownership of consumer items (durable goods), dwelling characteristics in terms of wall and flooring materials, the source of drinking water and toilet facilities.

Psychosocial stress was examined using three tools. These included the Brugha Life Events questionnaire comprising of 12 questions related to negative life events such as illness, death, financial or marital difficulties and their impact.13 The impact was categorised and scored as no impact (0), some impact (1) or significant impact (2), separately, for an adverse event that occurred within the previous 6 months or more than 6 months ago. Another psychosocial tool used was Antonovsky's sense of coherence (SOC) scale, previously validated in South Africa,14 that consisted of 13 items measuring comprehensibility (cognitive), manageability (instrumental/behavioural) and meaningfulness (motivational),15 with a low SOC inferring a poor ability to cope with stressors.16 Psychosocial stress was also evaluated using the locus of control (LOC) set of six questions, which determined the individual's perceived sense of control over his/her environment and life. A low score construed poor perceived control and a high score good perceived control.17

Problematic alcohol use was assessed using the CAGE set of four questions, a brief and commonly used screening instrument including in South Africa,18 ,19 that was developed by Ewing for detecting alcohol abuse and dependence.20 With an average sensitivity of 0.71 and specificity of 0.90, high test–retest reliability (0.80–0.95) and adequate correlations (0.48–0.70) with other screening measures, the CAGE questionnaire was found to be a valid and reliable tool.21 Nonetheless, a positive screen needs to be followed by a clinical evaluation using standard criteria for a proper diagnosis. The acronym stands for four yes/no questions: Have you ever felt that you should Cut down on your drinking? Have people Annoyed you by criticising your drinking? Have you ever felt bad or Guilty about your drinking? Have you ever had a drink first thing in the morning to steady your nerves or get rid of a hangover (Eye-opener)? Problematic alcohol use was deemed present if two or more CAGE questions were answered affirmatively.

Statistical analysis

Data analyses were done using STATA ver.11. Descriptive statistics, including crude prevalence, were calculated using the weights based on the sample design and adjusted for the realised sample. A principal component analysis of the pooled data, based on the assets that defined wealth, was used to develop an asset index22 and categories of relative wealth were created using tertiles.

Univariate analyses are presented as mean values and SDs for the continuous data (psychosocial scores), and as percentages and SEs for the prevalence data. Survey-based OR and 95% CI for the associations of the socio-demographic variables with alcohol use were calculated (unadjusted). The mean psychosocial scores across alcohol use categories were compared using an analysis of variance test adjusted for the survey design.

Survey multiple ordinal logistic regression analyses determined the independent associations of the socio-demographic and psychosocial variables with alcohol use. Alcohol use was categorised as non-drinkers, non-problematic drinkers and problematic drinkers. A model with the socio-demographic factors only showed a significant interaction between gender and education as well as percentage of life spent in a city (p=0.030). Therefore, gender-specific models were fitted when incorporating the psychosocial variables.

The psychosocial measures of SOC, LOC, adverse life events, impact of events that occurred within the previous 6 months and impact of events that occurred more than 6 months ago were modelled independently as continuous variables. Adverse life events and their impact were entered in separate models to avoid the problem of colinearity. The correlation was 0.827 for life events and impact >6 months, and 0.673 for life events and impact <6 months. The possible non-linear effects of the psychosocial variables were investigated by graphical presentations using lowess plots of alcohol versus a particular score by gender. Quadratic terms of the psychosocial variables were added to the models where indicated. These quadratic terms were not significant in any of the extended models. Consequently, gender-specific main effects models are reported in the table with the ordinal logistic regressions; the data for the other psychosocial variables when they independently replaced SOC in the same models are presented in the same table and changes in the direction or significance of the other variables are noted. The proportional odds assumption of the ordinal logistic models was investigated using a likelihood ratio test. In men, this assumption was violated in the model with impact of events that occurred >6 months ago. Multinomial regression estimates are reported for this specific model.

The common age range of 25–64 years between the two studies was used for the comparison between the 1990 and 2008/2009 data. A direct comparison of the 2008/2009 and 1990 datasets could not be conducted because of the geographic and demographic changes that occurred in some residential areas during this period; the populations increased markedly with a concomitant expansion in area size. Consequently, the reported estimate of the 1990 study was used as the population reference value. Further, problematic alcohol use could not be compared because the CAGE instrument was used only in 2008/2009.

The University of Cape Town's Research and Ethics Committee approved the study. All participants signed informed consent.


The realised study sample comprised of 1099 participants, 392 men and 707 women (64% and 108% of the planned sample, respectively), with 17 excluded because they did not meet the inclusion criteria. The response rate was 86%; of the 187 non-responders (ie, the selected people who the study team were unsuccessful in contacting), 79 (42%) were men.

Among 25–74-year-olds, self-reported alcohol consumption was 67.5% (95% CI 62.0 to 72.5) in men and 26.9% (95% CI 23.4 to 30.8) in women in this study. The highest prevalence was in 25–34-year-old men (72.6%) and women (31.8%). The lowest rates of alcohol consumption were in 65–74-year-olds (men 50.0%, women 18.4%).

The prevalence of self-reported alcohol consumption among 25–64-year-old adults in 2008/2009 was 46.9% (95% CI 43.1 to 50.7) which differed from the 1990 reference value of 35.5%. The rates in 2008/2009 were higher than the 1990 reference values in both men (2008/2009: 68.5%, 95% CI 62.7 to 73.7; 1990: 56.7%) and women (2008/2009: 27.4%, 95% CI 23.7 to 31.5; 1990: 15.1%). In contrast to the pattern by age category in 2008/2009 where alcohol use was most frequent in 25–34-year-old men and women, in 1990, the highest prevalence was in 45–54-year-old men (62.0%) and women (19.0%) (figure 1).

Figure 1

The prevalence of alcohol use in 25–64-year-old men and women in 1990 and 2008/2009 (±SE).

The prevalence of problem drinking, as measured by the CAGE questionnaire, was higher in men (49.7%, 95% CI 44.6 to 54.9) than in women (18.1% 95% CI 15.3 to 21.2) (p<0.001) (table 1). However, among those who drank alcohol, symptoms of problem drinking were high in both men (73.4%) and women (67.1%) (p=0.107). Problem drinking was highest in 35–44-year-old men (56.7%) and in 25–34-year-old women (21.0%) and generally decreased with age.

Table 1

Prevalence and associations of socio-demographic factors with alcohol use in men and women*

Men who were problematic alcohol users were likely to be more educated and less likely to be employed or receiving a pension compared with being unemployed. Problematic drinking in women was more frequent with urbanisation and less likely in pensioners compared with the unemployed (table 1).

In both men (p=0.035) and women (p=0.010), there were significant differences in the mean SOC scores across the three categories of alcohol use. Table 2 shows that participants with problematic alcohol use had the lowest mean SOC scores and had experienced a greater number of adverse life events. The mean impact of adverse events that had occurred more than 6 months previously was significantly different across the three categories of alcohol use in women (p=0.024) but not in men (p=0.591).

Table 2

Mean±SD of psychosocial variables presented by categories of alcohol use in men and women

In the multiple ordinal logistic model with SOC for men, alcohol use was associated with higher education level (p=0.012), work status (p=0.026) and a lower ability to cope with stress as indicated by a lower SOC (OR 1.02, 95% CI 1.01 to 1.05, p=0.042) (table 3). In the separate model with SOC for women, spending more than half their life in the city (p<0.001) and coping poorly with stress (OR 1.02, 95% CI 1.01 to 1.04, p=0.039) were relevant. When adverse life events replaced SOC in the same models, the odds for alcohol use with increasing number of adverse events, after adjusting for the other factors, were significant in men (OR 1.11, 95% CI 1.02 to 1.19, p=0.010) and women (OR 1.09, 95% CI 1.03 to 1.16, p<0.005). In these models with increasing adverse life events, work was no longer significant in men (p=0.095). In women, being older compared with 25–34-years-olds was less likely to be associated with alcohol use (p=0.049). In men, the impact of lifetime events that occurred >6 months ago was inversely associated with non-problem drinkers (OR 0.91, 95% CI 0.84 to 0.98, p=0.010) and not significant in problem drinkers (OR 1.01 95% CI 0.96 to 1.07, p=0.655).

Table 3

Multiple ordinal logistic regression models for alcohol use categorised as non-drinkers, drinkers with CAGE <2 and drinkers with CAGE ≥2


Alcohol use

The higher prevalence of alcohol use found in men than in women in this study is a universal finding but the magnitude of the difference varies among populations and cultures.8 Although the prevalence of self-reported alcohol consumption in 2008/2009 was higher than in 1990 and previous national studies6 ,18 ,19 in both genders, of concern is the steep rise in consumption among women who traditionally had low rates. However, the prevalence in this study was comparable with women in Western Cape (24.1%–33.1%).6 ,18 ,19 Higher alcohol consumption among white and coloured compared with black women in this region in the past may partly account for these comparable rates.

Traditionally, alcohol consumption was the preserve of elders and other senior or authority figures, and less common among the youth and women of childbearing age.23 The weakening of social and cultural barriers that may have previously prevented women from drinking and increased female autonomy with greater spending power may contribute to a rise in the proportion of women, particularly younger women, who consume alcohol as seen in this study. Of interest, however, in the GENACIS study, alcohol use was generally more prevalent with older age.8

In keeping with the weakening of social and cultural traditions, the highest rate of alcohol consumption among men in 2008/2009 was in the younger and the lowest in the older population, unlike the findings in 1990. That almost three-quarters of the younger men drank alcohol may indicate the success of alcohol advertisements which are usually aimed at the younger market with advertisements placed for maximum exposure to young people.24 Drinking is linked to notions of financial and social success,25 and promoted as a lifestyle associated with fun, popularity, partying and other evening activities.3 ,26 Alcohol advertising is documented to influence young people's behaviour by normalising drinking in many different settings, bringing about positive beliefs about drinking, and encouraging young people to drink alcohol sooner and in greater quantities.25

Problematic alcohol use

The high prevalence of problem drinking symptoms among those who drank alcohol, including women, suggests that the majority of alcohol consumers were problem drinkers. The pervasive nature of this drinking pattern among alcohol consumers indicates a culture of problem drinking and the high rates observed are a manifestation of the prevailing societal norms and attitudes that favour erratic heavy drinking. Among South Africans in general there seems to be a high level of acceptance of heavy drinking, particularly among men.23 The major burden that alcohol misuse places on the health, economic and social well-being of the individual3 ,6 ,27 ,28 needs to be emphasised in order to change the South African attitude towards problem drinking.

In accordance with the findings in this study where unemployed men were more likely than employed men to consume alcohol and be problem drinkers (p=0.008), a review by Henkel (2011) found that risky alcohol consumption (associated with hazardous, binge and heavy drinking) was more prevalent among the unemployed.29 The association of poverty with problematic alcohol intake speaks of the relatively low real price of alcohol in South Africa with easy access to alcoholic beverages likely contributing to easier uptake and overindulgence.23 Furthermore, the high rates of unemployment and poverty encourage the production and sale of cheap home-brewed alcohol as a source of income, thereby also contributing to problem drinking.

Additionally, unemployment, poverty and urban living are likely to be associated with psychosocial stress, which may also predispose individuals to problematic alcohol intake. The associations of lower SOC, and a greater number of adverse life events and their impact with problematic alcohol use in this study underscore the relationship between psychosocial stress and problem drinking. These findings are in keeping with Conger's Tension Reduction Hypothesis which postulates the use of alcohol for its tension-reducing properties, that is, individuals consume alcohol in response to stressful situations to obtain relief from psychosocial distress.10

The unidimensional general LOC scale, which showed no relation with alcohol use in this and other studies, has been adapted for drinking behaviour with the developments of drink-related LOC and drug-related LOC.30 These drink-specific LOC instruments have been significantly associated with patterns of alcohol consumption unlike the unidimensional construct and highlight that control beliefs are multidimensional rather than unidimensional.30

The association of urbanisation with problem drinking in women may be related to the psychosocial stress that accompanies urbanisation; often unplanned urban growth is frequently found to be associated with poverty and alcoholism.31 Living in poor neighbourhoods in urban areas, where crime, unemployment and so on are common, was found to be related to heavy drinking.10 Additionally, this relationship may be a manifestation of weakening social and cultural traditions, particularly in urban settings. This is supported by higher alcohol consumption and problem drinking reported in South African urban compared with rural women.6

The association of higher education level with alcohol use among men accorded with a pattern of greater alcohol consumption and high-risk drinking behaviour prevalent among the better educated in a Brazilian study.32 On the other hand, in the USA, while alcohol use was positively related to education, various measures of heavy drinking were associated with lower education levels.9 Further research and analyses are likely required to untangle the multifaceted associations of higher education and unemployment with alcohol use and problem drinking in this community.

Addressing the problem

Despite a culture of problem drinking and the high disease burden attributable to alcohol in South Africa, the focus on alcohol control has been disconcertingly inadequate. The failure to acknowledge and address alcohol in South Africa as an increasingly important factor for CVD, other NCDs and the current extremely high burden of violence and injury is in stark contrast to the successful reduction in tobacco control.33 A comprehensive national alcohol strategy that cuts across different sectors including health, social welfare, trade and industry, and agriculture, among others, is urgently required to curtail the misuse of alcohol.4 ,28

Recently, there has been a renewed interest in addressing the problem of rising alcohol misuse in South Africa.34 The Minister of Health has suggested a ban on alcohol advertising and sponsorship, and to increase the age of legal alcohol consumption from the current 18 to 21 years.25 ,35 The Minister's stance is based on the decrease in smoking that followed the ban on tobacco advertising.

Multilevel interventions are also required to target high-risk drinkers. Screening initiatives to identify individuals at risk and brief intervention programmes for hazardous and harmful drinking are necessary.1 The inclusion of such programmes in the public healthcare sector, while increasing the burden on primary healthcare services, may be offset by fewer hospital admissions for alcohol intoxication and other alcohol-related conditions and should be considered in the rollout of the National Health Insurance scheme. The focus therefore needs to shift from the current provision of tertiary care for alcohol dependence, often associated with poorer outcomes, to cost-effective brief intervention strategies for early alcohol use disorders.6 Even simply screening for alcohol use and the provision of an alcohol health education leaflet may reduce drinking.36

The study limitations include the cross-sectional study design that precludes conclusions about causal associations between problematic alcohol use and its determinants. Furthermore, the low sample realisation in men (64%) that is characteristic of epidemiological studies in this country necessitated higher sampling weights and a loss of precision. Alcohol use, problem drinking and psychosocial stress were self-reported measures which may lead to same source bias. However, the use of the CAGE instrument aimed to overcome the reluctance and lack of social desirability among participants to discuss alcohol misuse.

What is already known on this subject

In South Africa, alcohol misuse is a major social and health problem with approximately one in four adult men and one in 10 adult women experiencing symptoms of alcohol-related problems. This underscores the need for regular monitoring of alcohol consumption and misuse. Although on a national level the most recent surveillance data on alcohol use were obtained in 2008, on a regional level these have not been examined in the urban black population of Cape Town in almost two decades.

What this study adds

Self-reported alcohol consumption increased 20.8% and 81.5% in men and women, respectively, between 1990 and 2008/2009. Among 25–74-year-old adults, in 2008/2009, the prevalence of alcohol consumption was much higher in men (67.5%) than in women (26.9%). However, among those who drank alcohol, problem drinking was high in both men (73.4%) and women (67.1%), and found to be significantly associated with psychosocial stress. This study highlights the urgent need for concerted prevention and control measures in the form of effective and comprehensive alcohol policies that cut across all related government departments.


The authors would like to thank the participants, fieldworkers, MRC research nurse fieldworkers Debbie Jonathan and Theresa Gogela, fieldwork coordinator Erica April, study manager Serena van Haght, statisticians Nomonde Gwebushe, Rebecca Shanmugam and Ria Laubscher, and Dr Kirsty Bobrow. We also thank the City of Cape Town.



  • Contributors NPwrote and researched the manuscript, and performed the statistical analyses. KS contributed to conception and design of the study, interpreted data and reviewed/edited the manuscript. CL contributed to the analysis and interpretation of the data, and reviewed/edited the manuscript. NL contributed to conception and design, interpreted data and reviewed/edited the manuscript. All authors read and approved the final manuscript.

  • Funding This work was supported by an unrestricted grant from Servier Laboratories (South Africa); the Medical Research Council of South Africa; the Initiative for Cardiovascular Health Research in Developing Countries (IC Health) Foundation Council; and Brigham and Women's Hospital, Harvard University.

  • Competing interests NL has received honoraria from Novartis for serving on the steering committee for the Navigator Trial and travel support from Novo Nordisk, Eli Lilly Laboratories and Sanofi Aventis.

  • Patient consent Obtained.

  • Ethics approval The University of Cape Town's Research and Ethics Committee approved the study.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Disclaimer This was a collaborative study between the University of Cape Town's Department of Medicine and the Medical Research Council's Non-communicable Disease Research and Biostatistics units.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.