Health-seeking behaviours by gender among adolescents in Soweto, South Africa

Background Adolescents are an important age-group for preventing disease and supporting health yet little is known about their health-seeking behaviours. Objective We describe socio-demographic characteristics and health-seeking behaviours of adolescents in Soweto, South Africa, in order to broaden our understanding of their health needs. Design The Botsha Bophelo Adolescent Health Study was an interviewer-administered cross-sectional survey of 830 adolescents (14–19 years) conducted in Soweto from 2010 to 2012. Health-seeking behaviours were defined as accessing medical services and/or being hospitalised in the 6 months prior to the survey. Chi-square analysis tested for associations between gender, other socio-demographic and behavioural characteristics, and health-seeking behaviours. Results Of 830 adolescents, 57% were female, 50% were aged 17–19 years, 85% were enrolled in school, and 78% reported experiencing medium or high food insecurity. Males were more likely than females to report sexual debut (64% vs. 49%; p<0.0001) and illicit drug use (11% vs. 3%; p<0.0001). Approximately 27% (n=224) and 8% (n=65) reported seeking healthcare or being hospitalised respectively in the previous 6 months, with no significant differences by gender. Services were most commonly sought at medical clinics (75%), predominantly because of flu-like symptoms (32%), followed by concerns about HIV (10%). Compared to females, males were more likely to seek healthcare for condom breakage (8% vs. 2%; p=0.02). Relative to males, a significantly higher proportion of females desired general healthcare services (85% vs. 78%; p=0.0091), counselling (82% vs. 70%; p<0.0001), and reproductive health services (64% vs. 56%; p=0.02). Conclusions A quarter of male and female adolescents accessed health services in the 6 months prior to the interview. Adolescents reported a gap between the availability and the need for general, reproductive, and counselling services. Integrated adolescent-friendly, school-based health services are recommended to bridge this gap.

HIV/AIDS was the second leading cause of death globally in 2012, with adolescents in sub-Saharan Africa disproportionately affected (4).
Prior studies from Soweto in South Africa have identified high rates of unintended pregnancies among adolescents (5,6), leading to sexual and reproductive health needs like medical abortions (5), antenatal care (7), and prevention and treatment for sexually transmitted infections (STIs) including HIV/AIDS (8). In 2005, approximately 35% of adolescents in South Africa reported unintended pregnancies (5), indicating the need for accessible adolescent-friendly sexual and health reproductive services. Previous studies have reported that other healthcare services accessed by adolescents in South Africa include oncology care (9) and counselling for violence and psychological abuse (10).
The National Adolescent-Friendly Clinic Initiative (NAFCI) has attempted to improve the response of the South African public sector to the special needs of adolescent health by introducing 'adolescent-friendly' clinic accreditation. NAFCI has worked with clinics to build capacity and improve acceptability and quality of services for adolescents, especially with regard to sexual and reproductive health (11). In 2006, NAFCI was adopted by the Department of Health in South Africa transforming it to Youth Friendly Services (YFS) which trains healthcare providers and accredits facilities for adolescent health services. In the new YFS services, responsibility for media campaigns was not retained but is being championed by other non-governmental organisations like LoveLife that runs multimedia campaigns, peer outreach, youth centres, and peer education programmes as part of their HIV prevention campaign (12).
Research on health-seeking behaviours among adolescents in South Africa is limited. To help address this gap, we describe here the health-seeking behaviours of adolescents, by gender, within the peri-urban community of Soweto, South Africa.

Study design and setting
The Botsha Bophelo Adolescent Health Study (BBAHS) was an interviewer-administered cross-sectional survey of 830 adolescents conducted at the Kganya Motsha Adolescent Centre (KMAC) and the Perinatal HIV Research Unit (PHRU) in Soweto, South Africa, between June 2010 and June 2012. The BBAHS was a collaboration between Simon Fraser University in British Columbia, Canada, and the PHRU located in Soweto, South Africa.
KMAC was founded in 2008 targeting in and out-ofschool adolescents and was among the first health clinics in South Africa mandated solely for adolescents and designed to provide adolescent-friendly healthcare services. It offered HIV voluntary counselling and testing (VCT) and sexual and reproductive healthcare services for adolescents aged between 14 and 19 years at no cost.
Soweto is a township in the south of Johannesburg with a population of approximately 1.3 million inhabitants (13). In the most recent census, there were 315,116 adolescents aged between 15 and 19 years in the City of Johannesburg metropolitan municipality (14).

Participants
Participants were recruited through a stratified sampling approach across all regions (locally known as townships) that comprise Soweto. The sampling strategy was developed to target more females than males, those of older age (17Á19 years) to include a higher proportion of sexually active adolescents and to ensure representation from all geographic locations within Soweto including those from formal and informal housing. Recruitment was conducted in 44 identified townships and at KMAC. A minimum of 15 adolescents were recruited from each township. Only those between 14 and 19 years and living in Soweto were eligible for enrolment. To ensure representation and inclusion of the hard-to-reach and out-of-school population within townships, recruitment occurred in a variety of locations such as community centres, shopping centres, and areas around schools. Participants were approached and informed about the study. Those willing to participate provided contact details for scheduling interview appointments. Interviews were completed at KMAC or PHRU, whichever was more convenient for individual participants.
To avoid enrolling participants more than once, identification was required for enrolment and recorded in a separate file. Each participant was then given a unique identifier code containing no personal information in order to maintain confidentiality.

Data collection
Interviewers administered an online survey with participants. It assessed socio-demographics, sexual and reproductive health, HIV testing history, substance use (drugs and alcohol), and depressive symptoms. Trained, multilingual interviewers administered the survey in English or Isizulu (a local language in Soweto) via an online internet platform, Survey Monkey, using iPads or computers (Survey Monkey, 2012). The survey took an average of 60 min to complete.

Measures
Socio-demographic measures included gender, age (515, 15Á17 and 17Á19 years), type of housing (living in a shack or other [house or apartment]), employment (yes vs. no), current enrolment in school (yes vs. no), whether parents were alive (yes vs. no), level of education (up to primary vs. high school or higher), and food insecurity.
The nine-item Household Food Insecurity Access Scale was used to measure food access within a 4-week time period (Cronbach alpha 00.82). Possible responses were 'never', 'rarely', 'sometimes,' and 'often'. A score of 3 or more was considered a reflection of medium or high food insecurity with scores of 0Á9, and higher scores indicating greater food insecurity (15).
Risk behaviours were assessed by asking whether the participant had ever had a STI, alcohol in the previous 6 months, and ever used drugs. Adolescents were asked if they had ever had sex (yes vs. no), had ever tested for HIV (yes vs. no), and their sexual orientation (heterosexual vs. non-heterosexual).
Depression was measured with the 20-point Center for Epidemiologic Studies Depression (CES-D) 9.3 scale (Cronbach alpha 00.81). It is used for the assessment of depressive symptomatology in the general population (16). Individual items are scored between 0 and 3, with the scale score ranging from 0 to 60. Based on previous evaluations of the CES-D among adolescent populations, we used a score ]24 to determine the presence of depressive symptoms (17).
The primary outcome variable, health-seeking behaviour, was defined as having responded 'Yes' to the question, 'Have you gone for medical/healthcare in the past 6 months?'. To assess where adolescents sought healthcare services and for what reasons, we asked; 'Where did you go for medical/healthcare in the past 6 months?' and 'What was the reason you required medical/healthcare in the past 6 months?'. To determine hospital service use we asked; 'In the past 6 months, have you been hospitalised for any reason?', and 'What was the reason for hospitalisation?'. Additional questions about health-seeking behaviours were, 'Have you ever tested for HIV?'; and to determine the potential for gaps in service needs, we asked; 'What social and health services would you or people you know like to have that you cannot currently find?'.

Ethical considerations
Ethical approval to conduct the study was given by the ethics committees of the University of the Witwatersrand, Johannesburg, South Africa, and Simon Fraser University, Burnaby, Canada. The study was conducted in accordance with the Helsinki Declaration of 1975 as revised in 2008. Participants received 50 ZAR (Â$7) for travel reimbursement. Written informed consent was obtained for all participants. Participants younger than 18 years required written parental consent with participant assent.

Statistical analysis
Overall frequencies and frequencies by gender were determined for categorical socio-demographic and behavioural measures such as age-group and ever testing for HIV. Reliability of the assessment scales CES-D and food insecurity were assessed using the Cronbach alpha statistic.
Frequencies and percentages overall and by gender were determined for the items 'Have you gone for medical/ healthcare in the past 6 months?', 'Where did you go for medical/healthcare in the past 6 months?', 'What was the reason you required medical/healthcare in the past 6 months?', 'What social and health services would you or people you know like to have that you cannot currently find?', 'In the past 6 months, have you been hospitalised for any reason?', and 'What was the reason for hospitalisation?'. Where applicable, all items with the response 'Other' and 'Please specify' were categorised to fit in with the pre-specified categories or plausible medical classifications. Chi-squared tests were used to investigate associations between variables. 'Other (please specify)' responses to the item 'What social and health services would you or people you know like to have that you cannot currently find?' were assessed descriptively.
Where missing values occurred, we performed a complete case analysis. All statistical tests were performed at a 5% level of significance in SAS Enterprise Guide 5.1 (18).
Almost half (47%, n0378) of the adolescents reported ever testing for HIV, and 2.4% (n020) reported ever having an STI. Though 55.5% (n 0461) of participants reported sexual debut, males were more likely to report this than females (64.2% vs. 49.1%; p B0.0001). Although the proportion of males and females reporting alcohol use was similar (66.9% vs. 62.0%; p 00.1557), more males than females reported drug use (10.7% vs. 3.2%; pB0.0001). A high score of depressive symptomatology was reported in 48.1% of adolescents and was similar between males and females.
Characteristics of health-seeking behaviours Figure 1 shows the most commonly desired services among adolescents. These were general health (85% among females and 78% among males), counselling (82% among females and 70% among males), reproductive health (64% among females and 56% among males), and addiction counselling (25% among females and 29% among males). Slightly more than one-quarter of participants had accessed medical care in the past 6 months with the majority seeking care through medical clinics (76% of females and 73% of males). The main reason for seeking medical care was flu-like symptoms (30% among females and 35% among males). Almost 8% of adolescents had been hospitalised in the previous 6 months, and the most common reason was injury (33% among females and 26% among males). All bold values refer to variables that are statistically significant.
In Table 2, the location and reasons for adolescent health-seeking behaviour in the past 6 months are shown. Overall, 27.0% (224/830) of the adolescents reported seeking medical care in the previous 6 months, and the proportions by gender were similar (25.6% vs. 28.0%; p00.4474). Healthcare was most commonly sought at local medical clinics (74.6%), followed by doctor's practice (8.5%) and hospitals (6.3%). The proportion of female adolescents seeking healthcare in the doctor's practice was significantly higher than for males (12.8% vs. 2.2%; p00.0052). Approximately 5.4% of adolescents sought healthcare through a school clinic, and only 3.1% sought care through a family planning clinic. Overall, approximately 7.6% sought care through a homeopath and/or a traditional healer.
The most common reason for seeking healthcare was flu-like symptoms (32.1%) followed by concerns about HIV (10.3%), injuries (8.5%), and dental problems (7.1%). Compared to females, males were more likely to seek care for condom breakage (7.7% vs. 1.5%; p00.0205). Overall 4.9% of adolescents sought healthcare for birth control and 3.6% for depression and/or suicidal thoughts. Approximately 8.8% of the young male participants sought healthcare for circumcision-related reasons.

Discussion
This study adds to the literature on health-seeking behaviours of adolescents. The primary strength of the study is that, unlike many other adolescent studies, it did not restrict recruitment to a type of venue, such as schools or clinics, but rather obtained the sample from the general population, thereby maximising the potential of the study to include vulnerable adolescents who may be outside of formal systems.
Our findings show that about one-quarter of adolescents sought healthcare services in the 6 months prior to the interview. Uptake appeared much higher in a schoolbased study in the US, where in their senior year, 71% of participants had visited a healthcare provider, 44% visited a school health centre, 41% visited a private doctor's office, and 26% visited an emergency room (19). The differences in uptake may be explained by poorer economic circumstances in our participants who access the private sector infrequently, and the lack of school health centres in public sector South African schools. It is known that adolescents prefer to be separated from adult service provision (20), and the absence of school health services may indeed limit the ability to reach adolescents for health promotion.
In our study, the conventional public healthcare sector was the predominant entry point for adolescent care. In this sector, although general health, counselling, and reproductive health services do currently exist, adolescents perceived that these very services were unavailable to adequately meet their needs. This disconnection deserves further consideration. It is possible that adolescents were not aware of these services being present in their area. However, some adolescents noted barriers to care including perceived sub-optimal service quality. Adolescents' desire for existing services to be improved and their ideas for improvement are consistent with findings from a recent audit of healthcare facilities in South Africa commissioned by the Department of Health (21). In that audit, only 30% of facilities were found to be compliant with 2) 0 (0) Á Other: Not stated (n 05), concern about breast cancer (n 02), eye illness (n05), renal positive and caring attitudes, 50% with cleanliness and 54% with availability of medicines and supplies. Taken together, these realities may discourage adolescents from seeking healthcare. One study showed that if clinics adhere to NAFCI standards, the quality of adolescent-friendly services could be significantly improved. However, NAFCI adherence on its own is insufficient: the health system should include adolescent voices in determining their own health needs, should educate them about health rights and inform them about service availability (11). While the proportion of female adolescents seeking healthcare in our study was slightly higher than that of males, they were not significantly different, suggesting that health systems are a viable entry point for health promotion for both male and female adolescents. Other African studies have found gender disparity in adolescent health-seeking behaviours, with significantly more females accessing healthcare than males (22Á24). In our study, the differences in service utilisation were not by gender, but rather by sexual activity and by HIV testing uptake. This suggests that health service utilisation for both male and female adolescents may follow from the initiation of sexual activity. Again it highlights the opportunity that health facilities have to promote sexual and reproductive health to male and female adolescents.
In our setting, the most common reason for seeking healthcare was flu-like symptoms, and this did not differ by gender. This finding is different to that of a recent multi-country study across Africa, North America and Asia, which, unlike our study, had a qualitative design. It found that reasons for seeking healthcare differed by gender: female adolescents sought healthcare for sexual and reproductive health reasons while males sought healthcare for drug and alcohol related reasons (24).
Adolescents indicated the need for more accessible counselling services, especially related to violence, family, and drug addictions. This is underscored by the high observed prevalence of reported depression symptoms. Previous research from South Africa demonstrates the high prevalence of substance use and inter-personal violence  (10,25,26). While older reports from South Africa had found that epilepsy and pneumonia were the leading causes of adolescent hospitalisation (27), injuries were the leading cause for hospitalisation in our study. Although specific causes of injuries were not explored, it is likely that the injuries may be associated with the high levels of violence in South Africa (28).
There are gender implications that need to be borne in mind when providing adolescent-friendly health services. More males in our study reported drug use, earlier sexual debut, and consultations for condom breakages. Though much has been written about sexual health services for females, our work demonstrates the importance of including sexual health and drug counselling services for adolescent males. The association between male adolescents and drug use has been documented in other studies (29).
Our data reflect the vulnerable nature of adolescents in Soweto: just over three-quarters of 14Á19 year olds surveyed reported medium-to-high food insecurity. This reflects prevalent poverty, which is an important social determinant of health (3). Multi-sectoral interventions to address food insecurity are needed on a large scale. Indeed, in a few African studies, conditional cash-transfer systems have been noted to reduce risky behaviours in girls (30Á32). The South African government already provides child support grants to about 60% of children below the age of 18 years, and there is evidence to show this money is being used for essential needs (33). However, these grants are not conditional on anything except low parent or guardian income. We recommend removing barriers to accessing social grants for vulnerable youth who are both in and out-of-school, linking vulnerable schools with existing food and nutrition schemes, and re-establishing links between public education and health delivery systems.
Some limitations to this study should be noted. Measures were self-reported to an interviewer and therefore possibly influenced by recall and social desirability bias. We did not collect information on injury types. Neither did we explore what exactly participant concerns about HIV were as a reason for seeking healthcare. Voluntary response bias, defined as self-selected volunteers, may have occurred during the sampling process where adolescents were brought into a youth-friendly clinic for interviews from non-healthcare settings. The statistical analysis did not focus on causal inference as this was a cross-sectional non-random sample. Though not generalizable, our findings have relevance to other settings in sub-Saharan Africa where there is need for more research and programming in supporting health-seeking behaviours among adolescents.

Conclusions
Despite the existence of general, reproductive healthcare and counselling services within the South African healthcare system, adolescents reported a gap between their needs and the availability of these services; this may reflect unacceptability and perhaps inaccessibility of these services. One recommendation to address this is to revive schools-based health facilities in South Africa by incorporating gender-appropriate services to promote health and treat disease. Last, adolescent voices within health services planning are required to ensure adolescentfriendly, relevant, and appropriate health service delivery.