Because today's adolescents are attaining puberty earlier and marrying later, they are more likely to engage in premarital sex than members of their parents' generation were.1 Adolescents who have premarital sex often fail to use contraceptives, thus exposing themselves to risks of unintended pregnancy and of sexually transmitted infections (STIs), including HIV.2 Globally, more than 13 million adolescents younger than 20 give birth each year, contributing roughly 10% of the total annual number of births.3 Moreover, about one-half of all HIV-infected individuals are younger than 25, and the majority of these young people are women.4
Unprotected premarital sex is especially prevalent in Sub-Saharan Africa. A recent study of female senior high school students in Nigeria, for example, found that their mean age at first sex was 15 and that 23% of those who were sexually experienced had ever been pregnant; the vast majority of these pregnancies (88%) ended in abortion.5 Another recent analysis conducted in Cameroon demonstrated that by age 18, the majority of adolescents, regardless of their marital status, are sexually experienced and have been exposed to risky sexual practices, including exchanging sex for money, having multiple partners and failing to use condoms.6
To reduce the incidence of unprotected premarital sexual activity, adults need to understand adolescent reproductive decision-making. Clearly, the social environment plays an important role in affecting adolescent behaviors.7 The social environment for adolescents includes their friends, sexual partners and family members, as well as their community, school and other institutions that work with youth.
Background
In this article, we focus on one aspect of the social environment—the family. The family obviously has a strong influence on young people's aspirations and values from an early age.8 For example, adolescents who learn about their parents' and elders' values regarding premarital sexual activity or contraceptive use are less likely to engage in sexual risk-taking than their peers who are not exposed to their elders' values.9 Family influence comes from parents and from other family members who interact with youth, such as aunts, uncles, older siblings and grandparents.10
The transition from traditional societies to modern societies that is occurring throughout the world is generating a radically different culture for reproductive and sexual decision-making among today's adolescents. When most of today's older generation were adolescents themselves, social roles and expectations were better defined. Individuals appointed by the community (familial or nonfamilial) taught adolescents a set of clear and unambiguous rules that governed sexual conduct. With increased urbanization, however, the role originally played by community-appointed teachers must now be assumed, in part, by an adult family member, and by parents in particular.11
Parents and other family members are often reluctant to talk to young people about reproductive health issues. This reluctance may result from adults' lack of reproductive health knowledge or from a concern that adolescents will interpret such communication as affirming the acceptability of premarital sexual activity.12Adults may also think that their information is outdated. When families do not provide reproductive health information, however, adolescents will seek that information from other sources, including their peers. Nonfamilial sources of information may provide incomplete or inaccurate messages about sexuality, which can inadvertently increase adolescents' participation in behaviors that expose them to the risk of unintended pregnancy and STIs, including HIV.13
Given the increasing importance of family influences on adolescent sexual risk-taking, it is crucial to examine adult perspectives on these behaviors to assess which adult family members act as facilitators and which act as barriers to adolescents adopting healthy behaviors. The role played by gender must also be understood: Men and women may have different attitudes, and these attitudes may depend on the gender of the adolescent.14
In this article, we report on data from Lomé, Togo, that were collected to address these issues. The findings can be used to design appropriate reproductive health programs not only for adolescents but also for adult family members (both male and female) whose attitudes may prevent adolescents from adopting protective reproductive health behaviors.
Data and Methods
Data
Lomé, the site for this study, is Togo's capital, with about one million inhabitants. This coastal city is the center of commercial and international activities in the country. Like the rest of Togo, Lomé has a heterogeneous population made up of many ethnic groups and languages; the majority of the ethnic groups in Lomé come from the south of the country.
This analysis uses baseline data from a study undertaken by the FOCUS on Young Adults Project and Tulane University, in collaboration with the Unité de Recherche Démographique of Lomé, to evaluate the youth center established in Lomé by the Association Togolaise pour le Bien-Etre Familiale (ATBEF, the International Planned Parenthood Federation affiliate of Togo). The center opened in March 1998, and these baseline data were collected roughly eight months later, in November 1998.
The sampling plan for the ATBEF study followed the design of the Togo Demographic and Health Survey (TDHS), which had been undertaken earlier that year (February-March 1998). For the TDHS, a random sample of 62 clusters from the total number of 866 clusters in Lomé was selected. The TDHS identified and listed all households in these 62 clusters, and randomly selected households and individuals for interview.
The ATBEF youth center study aimed to collect data from a representative sample of both adolescents and adults to permit an evaluation of adult perspectives on adolescent reproductive health issues. To that end, the households from the 62 clusters were revisited and relisted. A random sample of young people was selected for interview from the new household listing, with the goal of obtaining a representative sample of 2,000 youth (1,000 males and 1,000 females), and a total of 2,083 10-24-year-olds were interviewed.
Likewise, a random sample of adults aged 30 and older was selected for interview, with the goal of obtaining a representative sample of 1,100 individuals. A total of 1,027 interviews (with 585 women and 442 men) were successfully completed; all interviews were conducted in the home of the respondent, and interviewers were matched to the sex of the respondent.
Of the interviewed adults, only 3% (N=35) were not parents. While we lack information on the specific age-range of respondents' children, we could determine through the screening questionnaire that 88% of the entire sample were parents or guardians of a child who was older than age eight. By targeting for interview adults aged 30 and older, we expected that the majority of their children would be at least adolescents or older. However, since many adults in our sample might not have had adolescent-age children, the data reflect adults' perspectives on adolescent behavior in general rather than parents' perspectives on their own adolescent's behavior. We hypothesized that adults' attitudes may differ by the age of their children, and we control for this possibility by controlling for the age of the adult (i.e., older adults are more likely to have adolescent-age children than are younger adults). In this article, we focus on data collected from the adult sample only.
Methods
While most of the survey items in the ATBEF questionnaire were directed to the total sample of adults, several communication questions were asked only of parents or guardians of either a son or a daughter aged nine and older. The questionnaire contained specific questions on topics that surveys rarely ask about—that is, how adults perceive adolescent sexual and contraceptive behaviors. The questionnaire also asked about adults' own knowledge, attitudes and practices regarding reproductive health, but those data, which were also collected in the 1998 TDHS, are not emphasized here.
We used simple chi-square tests to determine whether differences were significant by gender. Multivariate regression techniques (logistic regression and multinomial logistic regression) were then used to analyze factors associated with holding positive attitudes toward premarital sex and contraceptive use. We present the results of multivariate models separately for females and males, but we also ran a model that combined men and women to determine whether differences were significant by gender.
For the multivariate analyses, the three dependent variables are adults' approval of premarital sex for adolescent females; adults' approval of contraceptive use outside of marriage; and adults' approval of adolescents' use of contraceptives. This last variable originally had seven response options (approve, approve with reservations, disapprove, tolerate, leads to sterility, leads to loose values and "other"), which we collapsed into three—approve; somewhat approve (approve with reservations and tolerate); and disapprove (disapprove, leads to sterility, leads to loose values and "other").
The recategorization scheme may bias the results somewhat toward the "approve" category, or it may mean that "somewhat approve" includes some persons who actually "somewhat disapprove." Because of these potential biases, we emphasize in the multivariate results the distinction between "approve" and "disapprove" (in a multinomial logistic model); there are few differences between the "somewhat approve" group and the two other groups (results not shown).
Responses to the item on attitudes toward contraceptive use outside of marriage had three response options—approve, disapprove or indifferent. As only 3% of respondents said they were indifferent, their responses were dropped from the analysis. Finally, we assessed with a yes-no question whether each adult thought a male or female youth should have sex before marriage.
Although additional items on adults' attitudes were contained in the questionnaire, they were not included in the multivariate analyses. These questions, for which bivariate data only are presented, are gender-specific and were asked only of parents with a child of that gender. The questions gauging attitudes toward either a son or daughter practicing contraception and using a condom had four response options (approve, approve with reservations, tolerate and disapprove), which were regrouped into three—approve, somewhat approve (approve with reservations and tolerate) and disapprove. Questions on adult attitudes toward young people (of either gender) having premarital sex had the same initial four options and were similarly recategorized into three. Given the potential biases toward approval such a recategorization creates, we present multivariate results for the attitude variable on whether an adolescent male or female should have sex before marriage using a dichotomous response option (yes or no).
Age was included in all multivariate models (30-34, 35-39, 40-44, 45-49 and 50 and older) to control somewhat for the age that the children of respondents were likely to be at the time of the interview. We also hypothesized that younger adults are likely to have less conservative attitudes* toward adolescent sexuality compared with older adults.
The multivariate models also included educational level (none, primary, and secondary or higher), since more-educated adults are likely to have a more modern perspective on adolescent behavior, and the number of consumer goods or household amenities, since persons of higher socioeconomic status are likely to hold more modern views of adolescent sexual behavior.
The models additionally controlled for religion (Catholic, Protestant and other Christian, Muslim, and traditional or none), because some religions are more conservative toward adolescent premarital sexual activity and contraceptive use than others. Marital status (married or unmarried) was included to assess whether adults' perspectives vary by their own living arrangement. We hypothesized that married adults would be more conservative than unmarried adults.
Finally, we entered employment status into the models as an additional control for modern lifestyles, which are probably associated with less-conservative views. For the multivariate model of women only, three employment categories were used (self-employed, housework or "other"), while men's employment was assessed with four categories (public-sector job, private-sector job, self-employed and "other"). For men, we hypothesized that those who are self-employed are likely to have less-modern views than those who work in the public or private sector. The expected direction of the employment effect is not completely clear for women, however, because staying at home rather than working is in some cases considered to be a modern lifestyle, while in others housework is perceived to be a more traditional one. Additionally, homemakers may have greater contact with their children than working women, and thus may have more realistic perspectives on adolescent reproductive behaviors.
Sample Characteristics
The women in the sample were slightly younger than the men (e.g., 57% were aged 30-39, compared with 45%, see Table 1). Women were also much more poorly educated than men (e.g., 38% of women had had no schooling, compared with only 10% of men). The sample was primarily Christian (71-72%), with higher proportions being Catholic as opposed to belonging to Protestant and other Christian denominations (42-44% vs. 27-30%). The sample had reasonable access to durable goods, with 89% of women and 95% of men owning at least one of the items measured (i.e., a radio, television, refrigerator, motorcycle or car, and electricity in the home). Also, while the same proportion of men and women had never married (5%), women were more than twice as likely as men to be divorced or widowed (15% vs. 6%).
To determine the representativeness of the ATBEF adult sample from Lomé, we compared the characteristics of 30-49-year-old women and 30-59-year-old men with similar-aged women and men surveyed in Lomé for the 1998 TDHS. Among the women, the two samples differed significantly by marital status, by whether the woman was in a monogamous or polygynous union and by the woman's relationship to the head of the household. For example, women in the ATBEF sample were significantly more likely (p<.001) than those interviewed for the TDHS to be currently married (86% vs. 81%) and to be the spouse of the household head (68% vs. 59%). Among married women, those in the ATBEF sample were significantly more likely to be in a monogamous union than those in the TDHS (75% vs. 63%).
Men in the ATBEF sample were significantly less likely than those in the TDHS sample to be Muslim, but significantly more likely to report either no religion or a traditional religion. Moreover, men interviewed for the ATBEF youth center evaluation were significantly more likely to be single (5%) and to be the head of the household (95%) than were those interviewed for the TDHS (0% and 88%, respectively).
Finally, among respondents of both genders, those in the ATBEF sample were more likely to be current contraceptive users than those in the TDHS sample. This significant difference in overall contraceptive use reflects both a greater reliance on traditional methods among women in the ATBEF sample compared with those in the TDHS sample (30% vs. 15%), and a greater reliance on modern methods reported by men in the ATBEF sample compared with men in the TDHS sample (43% vs. 23%). These findings may represent true differences in contraceptive use across the samples, or they might instead reflect differences in the wording of the questions. (For example, the ATBEF item was worded: "What contraceptive method do you often use when you have sexual relations?" while the TDHS item was worded, "Right now, do you use a method to avoid a pregnancy?")
The significant differences between the two samples—especially the greater likelihood among ATBEF respondents to be married and to be either the head of the household or married to one—potentially limit the generalizability of our study. Thus, the results presented here need to be considered in terms of how selected married adults in Lomé perceive adolescent behaviors, and are not necessarily representative of all adults. Moreover, the male ATBEF sample is less likely to be Muslim than the general male population of Lomé.
Results
Adult Attitudes on Adolescent Behaviors
•Bivariate analyses. In the ATBEF adult sample, only 21% of women and 25% of men said they approved of young people having sex before marriage, and an additional 22% of women and 27% of men somewhat approved (Table 2). Thus, women were significantly more likely than men to hold disapproving attitudes toward adolescents engaging in premarital sex (58% vs. 48%).
A specific question on whether an adolescent female or male should have sex before marriage (yes or no) assessed whether adults' attitudes varied by the adolescent's gender. Although slightly smaller proportions of adults stated that sex before marriage would be acceptable for an adolescent female (27-32%) than for an adolescent male (32-35%), the difference is not significant.†
Overall, just 27% of women disapproved of family planning use among married couples, while the proportion disapproving of use among unmarried couples was far higher (40%, Table 2). Adult women's rates of disapproval were highest for young people using contraceptives (48%), and for one's own son or daughter using contraceptives (46% and 48%, respectively). Women's relatively low approval rates of young people's contraceptive use likely reflects their reluctance to approve of young people having sex before marriage at all. Among men, roughly 23-25% disapproved of contraceptive use among both married and unmarried couples, and a slightly higher proportion disapproved of young people practicing contraception (31%).
Men were significantly more likely than women to approve of contraceptive use in general. It is interesting to note that the level of approval of condom use by people of all ages for STI and HIV prevention is high among both men and women, and even exceeds that of approval for married couples using family planning. The high level of approval for condom use in this population (79% of women and 85% of men) suggests that HIV and AIDS messages have been well understood in Lomé. These condom use messages, however, appear to not always be linked to family planning messages.
•Multivariate analyses. Net of all social and demographic variables, older women (those aged 45 and older) were significantly less likely than 30-34-year-old women to approve of adolescent females having premarital sex, as were Muslim women compared with women who reported no religion or a traditional one (Table 3, page 182). Married women were similarly less likely to approve of young women having premarital sex than were unmarried women. The converse of this association—the greater likelihood of approval among unmarried women—may stem from these women's tendency to rationalize their own sexual activity.
As with women, the factors among men that significantly reduced their likelihood of approving of adolescent females having premarital sex included older age and being Muslim. Additionally, educated men were significantly more likely to approve than were uneducated men. (Although we do not present results of the multivariate model of attitudes toward young men having premarital sex, that model showed the same significant associations as the model for approval of adolescent women having premarital sex.)
Lomé women aged 45 and older were significantly less likely than those aged 30-34 to approve of unmarried couples' contraceptive use. (While married women were similarly less likely than unmarried women to approve, that association was only marginally significant, at p=.054.) Additional factors associated with women's approval of unmarried couples' contraceptive use included socioeconomic status and religion. For example, women living in households with 3-6 of the items measured were more likely to approve than were those living in households that had none, and Muslim women were significantly more likely than those of either no religion or a traditional religion to indicate that contraceptive use would be permissible among unmarried couples. (Muslim women were also significantly more likely to approve than were Protestant women, results not shown.)
Similar factors were associated with women's approval of young people practicing contraception. The data from the multinomial logistic regression comparing approval with disapproval indicate that age and number of household items significantly affected the likelihood that women would approve of young people's contraceptive practice. However, in this model, women's schooling also emerged as a significant predictor, with better-educated women being significantly more likely to approve than women who had received no schooling.
Among men, the only factors associated with their attitudes toward unmarried couples adopting contraception were older age and having a secondary education. Once all variables were controlled for, men older than 50 were less likely to approve than were those aged 30-34, while men with a secondary or higher education were significantly more likely to approve than were less-educated men.
These variables were also significant, and in the same direction, in the model assessing men's attitudes toward young people's contraceptive use, while a third variable emerged as significant in this model as well—socioeconomic status. That is, men with 5-6 household items were significantly more likely than those with none to approve of contraceptive use among young people. (This association was not significant when the reference group was 0-2 items.)
The multivariate model that combined data from both men and women (not shown) revealed no significant differences by the gender of the adult respondent in whether they approved of adolescent females having premarital sex. However, in the joint model, which assessed men's and women's attitudes toward contraceptive use, men were significantly more likely than women to approve of contraceptive use by both unmarried couples and by young people.
These multivariate results consistently demonstrate that Lomé women hold more conservative attitudes toward contraceptive use than Lomé men; that is, they are less likely to approve of contraceptive use either by young people or unmarried people. Also, older women and men tend to have more conservative attitudes toward adolescent sexual activity and contraceptive use. We observed less-conservative attitudes among more-educated men and women and among women of higher socioeconomic status (i.e., those who had more household amenities and durable goods). Finally, while Muslim women (and men) were less likely to approve of premarital sex, women were also more likely to approve of contraceptive use by unmarried couples; this finding suggests that Muslim women tend to have protective attitudes, not necessarily conservative attitudes.
Attitudes on Adolescent Knowledge
To supplement our findings on adults' attitudes toward adolescent behaviors, we also examined adults' perspectives on adolescents' access to reproductive health information. Roughly 70% of the Lomé adults interviewed perceived children to be well-informed on reproductive health issues, with no differences by the gender of the adult (Table 4).
Multivariate analyses (not shown) demonstrated that age was the main factor associated with believing that children are informed; we hypothesized that compared with younger adults, older adults may perceive children to be informed on these issues simply because older adults are more likely to be exposed to older children. On the other hand, older adults themselves may be uninformed, and thus be more likely to assume that children would know more about this subject than they do. The reasons behind this age distinction remain to be explored with future data that would control for the ages of respondents' actual children.
Overall, men were again more liberal than women in that they were significantly more likely to report being open to sexuality questions from children (81% vs. 77%). Moreover, more-educated men and women were more open to such discussions than their less-educated counterparts (not shown).
Finally, both women and men most commonly mentioned STIs and HIV (75% and 80%, respectively) when asked which subjects they would like children to be taught about. This finding may be related to the increasing attention being paid to HIV in Togo and to the fact that HIV is fatal and as yet has no cure. The next most common topic indicated by adults (about half as often) was contraception (36-40%).
The remaining responses revealed some distinctions by sex of the adult. Men were significantly more likely than women to say that they would like children to receive information on anatomy and puberty, while women were significantly more likely than men to think that children needed information on sexual activity. It remains to be explored in further analyses whether women and men would want different types of information to be imparted to females than to males.
Only the adults who were parents or guardians of a child aged nine and older were asked whether they had ever discussed a reproductive health topic with their child. Cross-tabulations demonstrate no significant difference between women and men in whether they held such a discussion with a child (Table 5). However, when the gender of the child is considered, women were significantly less likely than men to report having discussed a reproductive health topic with their son (27% vs. 34%, p<.05), but there was no significant difference at the bivariate level between men and women in their likelihood of having a discussion with a daughter.
When women only were considered, however, they were significantly more likely to have had such discussions with their daughters than with their sons (41% vs. 27%, p<.001); among men, there was no significant difference in having had such discussions by the child's gender. This finding may indicate that women are better able to communicate with a daughter than with a son, or that women feel that girls have a greater need for reproductive health information at an earlier age than do boys.
Multivariate analyses that included parents or guardians of a son only found no significant differences between women and men in the likelihood of a reproductive health discussion with a son (not shown). These models controlled for age, education, employment, number of household items, marital status and religion. Conversely, when we considered in a multivariate model only parents or guardians of a daughter, women were significantly more likely than men to report such communication with a daughter (not shown). Furthermore, multivariate models of communication with either a daughter or a son showed that parents aged 35 or older were significantly more likely to have discussed any reproductive health topic with their child than those aged 30-34. Again, this age difference may reflect the fact that older parents would be more likely to have older children, whom they perceive to have a relatively greater need for reproductive health information.
We also examined how parents of a daughter (of any age) would react if they learned that their unmarried daughter were pregnant. Their reaction, which provides a perspective of the family context within which adolescents might become pregnant unintentionally, was most commonly acceptance of the pregnancy (67% of mothers and 59% of fathers). The second most common response (mentioned half as often), however, was anger, which did not differ by the parent's gender. Other common responses were that the parent would feel deceived (both mothers and fathers), and that they would eject their daughter from the house (mentioned by a significantly higher proportion of fathers than of mothers). Finally, fathers were significantly more likely than mothers to report that they would feel dishonored if their daughter had an out-of-wedlock pregnancy.
Discussion and Conclusions
Recent studies have demonstrated that because of delayed marriage and greater involvement in premarital sex, young Africans are increasingly exposed to the risks of a premarital birth and of contracting an STI, including HIV.15 Earlier generations generally initiated sex within marriage (especially women) or married if a premarital birth took place. In the current ATBEF adolescent sample, however, 44% of unmarried 15-19-year-old males and 48% of similar females had ever had sex; the percentage of unmarried 20-24-year-old men and women who had ever had sex rises to 85% and 90%, respectively.16 The 1998 Togo DHS found that similar proportions of 45-49-year-old women had been sexually active by either ages 15-19 or 20-24, although the majority of this sexual activity took place within marriage.17
Given the potential for differing generational perspectives on premarital sex and contraceptive use, programs that target adolescents may fail to meet their objectives if adults who interact with adolescents act as barriers to adolescent adoption of safe-sex practices. Information on how adults can constitute such barriers may be crucial for the design and implementation of effective reproductive health programs for youth.
Our study found that adult perspectives on many adolescent behaviors differed by the gender of the adult. For example, women hold more conservative attitudes than men toward many behaviors, and they are especially unlikely to approve of adolescents and unmarried couples using contraceptives. Moreover, men are more open than women to their children's request for sexuality information. We found no differences, however, between men and women in whether they thought that adolescent females (or males) should have premarital sex.
Although women were significantly less likely than men to have held a reproductive health discussion with a son at the bivariate level, that association lost significance once we controlled for age and for other social and demographic variables. Conversely, the lack of a significant difference at the bivariate level between women and men in whether they had spoken with a daughter gained significance in the multivariate analysis. That is, net of all factors, Lomé women are significantly more likely than Lomé men to have communicated with a daughter about reproductive health. Thus, adolescent females appear to have greater access to reproductive health information from women, who tend to hold more conservative attitudes than men.
Such conservative values among adult women may affect adolescent risk-taking, especially if women are the primary source of reproductive health information in the family. Adolescents (of both genders) who rely on their mothers or on another female adult for their reproductive health information may not receive enough information or may get more conservative information, which might be inconsistent with adolescents' own attitudes and behaviors or with those of their peers.
Our study demonstrated that both older and less-educated women and men hold more conservative attitudes toward young people's sexual and contraceptive behaviors. Community-based reproductive health programs need to acknowledge the existence of these more conservative attitudes among older women and men, and approach these influential adults to elicit their support, since without it such programs are likely to fail. Furthermore, more-educated women and men are often respected by their peers, and may be useful educators within a community-based program.
Finally, our study demonstrated that Muslim women have especially protective attitudes toward adolescent sexual risk-taking. These attitudes were apparent in Muslim women's relatively greater disapproval of premarital sex and greater approval of contraceptive use among unmarried couples. Muslim women's protective attitudes should be fostered by programs that operate in communities with large Muslim populations.
This study is a first step at assessing whether adults might constitute a barrier to adolescent contraceptive use by demonstrating gender differences in adult perspectives on adolescent reproductive behaviors. A recent qualitative study in Lusaka, Zambia, which examined this issue using focus-group discussions, revealed that parents would be upset if they found out that their child had received family planning services at a clinic.18
Thus, our findings and those of others suggest that future studies need to assess whether parents and other adult family members act as a direct barrier to adolescents engaging in healthy reproductive behaviors by examining the role of adults' perspectives on adolescent outcomes. This can best be done with a large matched sample of parents (and guardians) and their own adolescent children. Given the potentially important role of parents and other adults in the household as barriers to adolescent contraceptive use, adult women and men (specifically parents or other adult role models) should be considered in reproductive health strategies that target adolescent risk-taking.
This involves activities at multiple levels. For example, supportive policies are needed that permit young people to access reproductive health information and services without parental consent. Clinic-based activities that provide information and services to adolescents in a discrete environment are also needed to permit them to access these services without the risk of encountering a parent or other influential adult in the family or neighborhood. Finally, activities that target parents and adults specifically are needed to promote positive attitudes toward behaviors that enhance and maintain reproductive health.
These activities can be developed through the media, through work with parent groups affiliated with schools or religious groups, or through community-based distribution of brochures.19 The messages provided to adults and parents need to emphasize the potential negative consequences of unintended pregnancies and STIs among adolescents, the need to promote responsible adolescent sexual behaviors, and the appropriateness and availability of reproductive health services for young people in the community.