Correlates of adolescent pregnancy in La Paz, Bolivia: findings from a quantitative-qualitative study.

Lipovsek, Vaija ; Karim, Ali Mebryar ; Gutierrez, Emily Zielinski 等

Adolescence is a transitional period characterized by opportunities for healthful development as well as risks for adverse health and social outcomes. It is now widely recognized that an adolescent's prospects for healthful development are influenced by myriad social, psychological, economic, and cultural factors (Kirby, 1999; Leffert et al., 1998; Resnick et al., 1997). Some of these (e.g., poverty, affiliation with peers who engage in high-risk behaviors, low self-esteem) appear to be risk factors for adverse health and social outcomes, while others (e.g., high parental educational expectations, "connectedness" with family and school, high self-efficacy) appear to be protective in that they tend to be associated with more favorable outcomes. Recent adolescent health research has assigned high priority to better understanding the relative importance of such factors so that youth programs might be developed to better reinforce the protective factors and mediate the risk factors that may be present in a given setting (Leffert et al., 1998; Resnick et al., 1997).

The present study was undertaken to address the question of why some female adolescents in La Paz, Bolivia, become pregnant while others in similar circumstances avoid early pregnancy. To investigate this question, the present study combined a nearest-neighbor case-control design with qualitative research methods (focus-group discussions). Study subjects for the case-control design consisted of matched pairs of female adolescents of the same age and residing in the same neighborhood, one of whom had experienced a pregnancy. Study subjects for the focus-group discussions were adolescent girls from the same population. The study focused on the following factors as possible explanations of the observed discordant outcomes: family structure, parental relationships, partner relationships, knowledge of pregnancy risk behaviors, self-esteem, and locus of control (internal versus external). The last three of these factors fall under the heading of what Kirby (1999) has termed "sexual psychosocial antecedents." Altho ugh our primary interest is in causal factors/antecedents of adolescent pregnancy, in many cases it is not possible to establish the direction of causality between factors and outcomes from the available data. For this reason, we use the term "correlates" instead of "antecedents."

Acquiring a better understanding of the correlates of pregnancy among Bolivian adolescents is important for several reasons. Bolivia, similar to many developing countries, has a very young population. Estimates from the 1998 Demographic and Health Survey indicate that 51% of the Bolivian population was 19 years of age or younger, including slightly over 10% between ages 15 and 19 (Republica de Bolivia, 1998). Most (63%) Bolivian females reach the secondary level of education, although by age twenty roughly 30% become pregnant (Republica de Bolivia, 1998). Recent estimates indicate a teen birth rate in Bolivia of 79 births per 1,000 females aged 15-19, well above the regional average (including North America) of 68 births (Republica de Bolivia, 1998; UNICEF, 1998). WHO/PAHO (1998) projections suggest that in the year 2000, 13% of all births in Bolivia will be to adolescents. Abortion in Bolivia is legal only if the health of the mother is in danger or in the case of rape, and thus voluntary termination of pre gnancies is not well documented and occurs under nonregulated conditions (Population Reference Bureau, 1994).

METHOD

The target population for the study was females 13-19 years of age residing in a peri-urban area of La Paz--the First Health District, known as La Garita de Lima. The total population of this area is estimated to be 250,000. There are 14 primary health care centers in the First Health District, all of which were sites for data collection.

A combination of quantitative and qualitative methods was used in gathering data for the study. The quantitative study design was a nearest-neighbor case-control design frequently used in epidemiologic research. Adolescents presenting at a primary health care center during the designated study period who were either currently pregnant or had been pregnant in the previous 12 months constituted the "cases." Adolescents of the same age (plus/minus one year) who resided in the same neighborhood as the cases, but who had never been pregnant, served as "controls." Study investigators familiar with the area, using the case's address as provided during her interview, visited residences nearest to the case's home until finding a suitable control (without revealing the identity of the control). Due to the intensity of the effort, one-to-one matching was used. Matching on neighborhood of residence has been shown in other studies to be effective in controlling for differences in socioeconomic status (Mock et al., 1993). The qualitative design consisted of focus-group interviews with adolescents from the same population. Girls who agreed to participate in the quantitative aspect (i.e., consented to a questionnaire interview) were also asked to participate in the focus-group discussions.

A structured questionnaire was administered to adolescents who agreed to be interviewed. Data were gathered in May-June 1999 by research assistants trained in interviewing techniques and provided written instructions. By the end of a six-week data collection period, 190 completed questionnaires had been obtained--95 cases and 95 controls. As adolescent pregnancy is a sensitive topic in Bolivia, the refusal rate among potential cases was high. The extent to which this may have biased study findings is unknown.

The structured questionnaire covered the following topics: marital status and living arrangements, family background, sexual experience, parental and partner relationships, knowledge of risk of pregnancy, self-esteem, and locus of control. The questionnaire was pretested at a maternity hospital in La Paz among adolescents presenting for a prenatal check-up and among adolescent girls of similar age and socio-demographic characteristics at a nearby secondary school.

After preliminary analysis of the data from the structured questionnaires, cases and controls associated with four of the sites (Chamoco Chico, El Tejar, Cuidadela Ferroviaria, and La Portada) were recruited to participate in focus-group discussions. However, due to the sensitivity of the subject matter, none of the adolescents who were, or had been, pregnant agreed to participate in the focus groups. Three focus-group discussions were conducted with girls who had never been pregnant, stratified by age (13-14 years, 15-16 years, and 17-18 years). Each focus group had 6-8 participants and lasted approximately 90 minutes.

The focus-group discussions centered around perceived risks and benefits of early pregnancy, its prevention, and its consequences. As the focus groups took place after the completion of the questionnaire data collection and preliminary analysis, they were intended to validate and further inform the quantitative results.

Of the initial 95 case-control pairs, four pairs were subsequently dropped due to age discrepancies of more than one year, as were three pairs that did not match sufficiently well with regard to place of residence. This left 88 matched pairs available for analysis which, although limited, is supported by other research as sufficient to employ the case-control design (Caton et al., 2000).

All variables measured were discrete, and most were analyzed as such: variables pertaining to marital status, living arrangements, family background, sexual experience, parental and partner relationships, and locus of control. In addition, continuous scales were created to measure two constructs, self-esteem and knowledge of risk of pregnancy. A scale with 12 items pertaining to perceptions and feelings about relationships with peers and partners was created to measure self-esteem (Cronbach's [alpha] .73), and a scale with 9 items was created to measure knowledge regarding pregnancy (Cronbach's [alpha] = .43). The alpha coefficient of the knowledge scale should be interpreted with some caution, as inconsistent responses may be indicative of inconsistent knowledge rather than low scale reliability (Zimet, 1992). All the items used for the two scales had binary responses. Accordingly, scales were developed using latent trait analysis (McDonald, 1999), which assumes that there is a single latent attribute per s cale and that all items in the scale measure that attribute. Classical item analysis (item-total score correlation and item-rest correlation) was used to select items for each scale (McDonald, 1999). Items for locus of control did not scale well using the classical item analysis method and therefore a scale was not constructed to measure this attribute. The items composing each scale and percentage of positive responses (i.e., positive selfesteem, internal locus of control, and correct knowledge of risk of pregnancy) are presented in Table 1.

Conditional logistic regression was used to test bivariate associations between the self-esteem index and pair wise-matched cases and controls. Odds-ratios for all other factors were estimated using the Mantel-Haenszel method. Matching was accounted for using the nonparametric McNemar chi-square statistic. Multivariate analyses were not conducted due to the differing missing values in the independent variables, which caused the number of case-control pairs to drop significantly (from the initial 88 pairs). Therefore, the low power of the sample size precluded meaningful multivariate results.

The focus-group interviews were audio recorded and transcribed by the Bolivian researcher. They were reviewed and synthesized around the main themes of interest: perceived risks and benefits of early pregnancy, its prevention, and its consequences.

RESULTS

Before turning to the quantitative findings of the study, it is informative to consider some of the consequences of early pregnancy perceived by focus-group participants, none of whom had been pregnant. These were overwhelmingly negative. Younger focus-group participants expected punishment from family ("they would hit you," "father would reject you because you made him look bad"), from peers ("friends would not spend time with you anymore"), and from the community ("you would be kicked out of school; people would talk badly about you"). A girl who gets pregnant, it was stated, would "not want to leave the house" or "would want to kill herself."

The perceptions of the older focus-group participants (ages 15-18) were comparable--they expected to be punished by their family and spoken of badly by the community. They stated that the family might support them, but that more often a young woman would face serious pressure from her family to marry the father of the child. They expected the community to look badly upon them ("people don't want their daughters to be friends with you"), but they also stated that their friends would stand by them and help them. As with the younger girls, these young women felt that getting pregnant would be a terrible outcome for them: "girls first think about abortion or suicide or of going away." One girl said, "if you can't talk to your parents, then you look for marriage to get out of the house."

Some of the consequences anticipated by focus-group participants, none of whom had been pregnant, are borne out by the data obtained from the structured questionnaires (Table 2). Two-thirds of cases were either married or cohabiting, compared with only 8% of controls. Early pregnancy and entry into marriage/union thus appear to be inextricably bound in this area of La Paz. Adolescent pregnancy also has significant consequences with regard to school attendance and employment. Less than one-third of cases attended school full time as compared with 76% of controls, and over three times as many cases as controls were employed. Nearly one-third of cases were neither attending school nor working (they likely were caring for infants/children), compared with only 3% of the controls.

It should be noted, however, that the direction of causation between adolescent pregnancy and marriage, school matriculation, and employment status is uncertain. It is possible that adolescents in La Paz who get pregnant early are self-selected with regard to their aspirations toward education, marriage, and children. However, these were not mentioned in the focus-group discussions as a rationale for/benefit of early pregnancy.

It may also be observed in Table 2 that there were no differences between cases and controls in age at first boyfriend and number of lifetime boyfriends. However, it is not possible to discern the differences in these relationships that contributed to diverging pregnancy outcomes, since the controls were not asked if they had ever engaged in sexual intercourse.

Findings regarding the association between adolescent pregnancy and family background factors (place of birth, parents' place of birth, and number of male or female siblings) are presented in Table 3. As may be observed, cases and controls did not differ significantly with regard to any of the family characteristics considered.

Associations between parental relationship factors and adolescent pregnancy are displayed in Table 4. Significant differences are observed between cases and controls on three of the six factors considered. Cases were 2.44 times as likely as controls to have reported substantial fighting with their parents, 0.48 times as likely to have said that their parents were affectionate, and 0.39 times as likely to have said that their parents were supportive (p < .05 for all three factors). As adolescent pregnancy in a conservative society such as Bolivia may well affect the quality of child-parent relationships, at least in some cases a pregnancy may have been the cause instead of the outcome of adverse relationships. However, a number of prior studies in which the causal sequencing is clear have shown positive parental relationships to be important protective factors for adolescent pregnancy (Kirby, 1999; Leffert et al., 1998; Resnick et al., 1997).

The combination of these findings and the results from the focus groups suggest that Bolivian adolescents do not tend to view parents as a key source of information or support for matters concerning sexuality, contraception, and pregnancy. When asked where they would turn for support on such matters, focus-group participants named friends as their first choice. Some also suggested going to the health center, although some acknowledged that contact with health facilities carried the risk of stigmatization--"if you go there, everyone will think you are pregnant."

With regard to relationships with partners, cases and controls differed significantly on only one of the seven factors considered- whether respondents had talked with their partner about getting pregnant (see Table 5). Cases were 3.5 times more likely to have talked with their partner about pregnancy than were controls (p <.01). However, as such discussions may well have occurred after they became pregnant, this factor cannot necessarily be viewed as an antecedent of adolescent pregnancy.

Only two items measured the association between peer influence and adolescent pregnancy. One of them--whether the respondent saw her friends often--was found to be protective regarding early pregnancy (p <.01). However, it is possible that, as with parental-relationship variables, relationship with peers was substantially altered as a result of an early pregnancy.

It is noteworthy that there were no significant differences on the other partner dimensions considered, particularly given the emphasis on partner communication and negotiation in many adolescent reproductive health programs. The focus-group discussions shed some light on this topic, highlighting the importance of partner (and peer) pressure in decisions to have sex. Regarding partner pressure, older focus-group participants stated: "if you do it, he won't leave you," "if you refuse, he may look for another girl," and "if you refuse, other girls will make fun of you." On the other hand, one participant observed that "if a girl refuses to have sex, her partner may leave her, but he may also lose interest and leave her after he's had sex with her." In the event of pregnancy, they saw two possibilities: "either he leaves you, or he's made to marry you."

In addition, while most girls believed that it was the responsibility of both partners to think about protection, they felt that it was really up to them to take care of themselves. However, they identified embarrassment and the fear of being perceived as promiscuous by their partners if they demanded the use of contraception. They stated that most girls use rhythm as a method to avoid pregnancy, and although most knew of other methods, they did not use them.

Findings regarding associations between adolescent pregnancy and two sets of factors falling under the general heading of "sexual psychosocial" factors are presented in Table 6. This table shows the association between self-esteem and knowledge regarding pregnancy indices and the outcome of interest (pregnancy). Although the alpha score for the pregnancy knowledge index is low, this is not seen as problematic in the case of knowledge because, as mentioned previously, inconsistent responses may be indicative of inconsistent knowledge as opposed to low scale reliability (Zimet, 1992).

As may be observed, there was a strong inverse linear relationship between level of self-esteem and likelihood of adolescent pregnancy (p < .01). That is, as the score on the self-esteem index increased, the likelihood of pregnancy decreased. As with other correlates, it is unclear whether diminished self-esteem is a cause or an effect of unwanted pregnancy. In the focus-group discussions, the participants spoke at length of negative consequences of unwanted pregnancy, including rejection by the partner, peers, and even the broader community, as well as possible punishment from parents. In light of this, it would seem that pregnant girls in La Paz are very likely to have lowered self-esteem as a consequence of the response of their immediate environment to their pregnancy.

The index measuring knowledge regarding pregnancy was not found to be associated with the outcome. As cases and controls did not differ significantly on any of the items, knowledge alone does not seem to be an explanatory factor for the disparities in pregnancy outcomes. However, the focus-group discussions yielded important insight into the difficulties adolescents in La Paz have in obtaining accurate information on reproduction and contraception. A number of the younger girls reported that they did not know anything about sexual relations, expressing that they were too young to talk about it, and that they had not yet been taught about it in school. They had, however, heard of condoms, and gave various sources of information for questions pertaining to sex and contraception, naming (in order of preference): friends, older sister or cousin, mothers, and the physical education teacher at school.

Older girls said they were taught the biological aspects of sex in school. When asked how to prevent pregnancy, they mentioned using condoms, going to the pharmacy, going to the health clinic, and waiting until marriage. When asked about where information about sex or contraceptives could be obtained, the girls listed (in order of preference): friends (a comfortable and easily accessible source of information, although some girls expressed concern about the accuracy of the information obtained); media; school (although they reported that the information given was often in unfamiliar terms, was confusing and limited, and was not in tune with their reality); and family (information obtained from family was also seen as limited and not always relevant, and girls reported being uncomfortable asking parents about these topics).

In addition to self-esteem and knowledge regarding pregnancy, there were also three items that attempted to measure locus of control. Little correlation was found between these three items (through item-total score correlation and item-rest correlation). For this reason, a scale was not formed, but each of the items was tested individually, and no association was found between these items and the outcome (the individual items are listed in Table 1).

While the focus-group discussions did not address locus of control specifically, the (perceived or real) inability of girls to demand the use of contraceptives from their partners, or obtain them from pharmacies or health centers, points to the low level of control Bolivian girls exert upon such important decisions. This should not, however, be mistaken for lack of awareness or desire to act, as the focus-group participants clearly noted they knew that contraceptive methods existed and where they could be obtained.

DISCUSSION

The present study sought to identify some of the factors responsible for discordant pregnancy outcomes among matched pairs of female adolescents in La Paz, Bolivia. One important factor was parent relationships, which were found to be strongly correlated with early pregnancy: never-pregnant girls were more likely than girls who were or had been pregnant to have reported being shown affection and to feel supported by their parents. Conversely, ever-pregnant girls reported higher levels of fighting in their parents' home than did controls. These findings are consistent with recent literature indicating that "connectedness" with parents is a powerful protective factor against a broad range of negative health and development outcomes among adolescents, and that strained parent-child relations and a coercive parenting style are often antecedents of adolescent risk behaviors and negative outcomes (Caton et al., 2000; Bearman et al., 1999; Cole, 1998).

Interpreting this result in the light of focus-group discussions suggests that a serious strain on the parent-child relationship is nearly unavoidable following an unintended pregnancy. Of equal importance, however, were the focus-group findings that communication with parents regarding sexuality is generally difficult for all Bolivian girls. Focus-group participants stated that it was difficult to discuss issues of sexuality and contraceptives with parents due to embarrassment, and that information received from them was seen as limited and not relevant to their lives. In addition, fear of punishment and rejection was expressed very strongly regarding parental relationships, particularly among the younger girls. Rejection could take the form of physical punishment or disappointment of parents or, in a more direct way, the girl could be rejected by the family by being forced to marry the father of the child. These results suggest that adolescents do not see their family as a source of support in a situation o f need. Improved parent-child communication relating to matters of sexuality and contraception would thus seem an important focus of future adolescent health initiatives in Bolivia.

This study also found that never-pregnant girls were more likely to have had high self-esteem than those who were or had been pregnant. Self-esteem was measured through a series of statements concerning perceptions and feelings about relationships with peers and partners. The statements revolved around sexuality and sexual decision-making in which the fear of rejection by both peers and partner figured prominently. The finding of a protective effect of self-esteem is consistent with much of the research literature (Kirby, 1999; Modrcin-Talbott et al., 1998; Connelly, 1997; Cole, 1997; Mullis & Mullis, 1997; Morgan, Chapar, & Fisher, 1995), although it should be noted that the literature is not entirely consistent on the relationship between self-esteem and sexual risk-taking. Some studies have found no relationship between self-esteem and pregnancy status after controlling for key background factors (e.g., age and socioeconomic status) and others have observed higher levels of self-esteem among adolescents wh o practiced risky sexual behaviors and had more sexual partners. A recent study of secondary school students in urban Peru found that the effects of self-esteem on sexual risk-taking behavior varied by gender--among boys, high self-esteem was associated with higher likelihood of sexual activity, but with a lower likelihood of sexual activity among girls (Magnani, Zielinski Gutierrez, Seiber, & Vereau, 2000). While the issue of causality must be approached with some caution, the statements that measured self-esteem in the present study generally revolved around perceptions and beliefs that would not necessarily be fundamentally changed with an unintended pregnancy. A plausible explanation is that low self-esteem among girls is in fact both an antecedent and an outcome of unwanted pregnancy; that is, girls with low self-esteem may be predisposed to sexual risk-taking, but girls who get pregnant in an environment that subsequently ostracizes them might also exhibit lowered self-esteem.

The finding that girls who spend more time with their friends were less likely to be pregnant is of some interest. While the negative influence of peers has been explored in the literature (Bearman et al., 1999), less is known about the detrimental effects of peer isolation. One possible interpretation of these findings is that girls with some form of peer network are more likely to be protected against adverse health outcomes, including unplanned pregnancy. Such an interpretation is supported by the focus-group results, where friends were cited as an important, albeit not entirely reliable, source of reproductive health information. The most reliable sources of such information cited in the focus groups were female siblings. Based on these data, it is possible to hypothesize that girls who were more isolated (that is, had no female siblings and did not spend much time with friends) were more likely to have become pregnant.

No clear effects of variations in partner relationships on the likelihood of adolescent pregnancy were observed in the quantitative data gathered for the study. This might be attributable to the limited power Bolivian adolescent females appear to have in such relationships. The information extracted from the focus groups indicates that there is little communication between adolescent partners in Bolivia about pregnancy and contraception. Focus-group participants reported an unwillingness to talk about contraception with their partners or request that condoms be used due to embarrassment and fear of appearing promiscuous. Nonetheless, they stated it was up to them to protect themselves against pregnancy, and most used the rhythm method to do so. However, given the relatively low use/effectiveness of the rhythm method and its inability to protect girls against sexually transmitted diseases, this is a less than optimal safeguard.

The fear of being rejected by partners also figured prominently in the focus-group discussions. Participants reported fearing that if they did not consent to sexual relations, their partners would leave them. Paradoxically, they also stated that a partner might leave a girl after he has had sex with her (he might lose interest in her), and that a partner often leaves a girl after she has become pregnant. This leaves a girl in a situation of having to balance her own interests, her family's expectations of her, and her desire for an intimate relationship.

Unlike in other recent studies, no significant differences were observed between cases and controls with regard to locus of control. It is clear from the focus-group discussions, however, that control issues figured prominently in decision-making concerning sex and contraception. Focus-group participants identified pressure from peers and partners as powerful factors; for example, that a partner would pressure a girl into having sex to show love for him and that even other girls would look down upon a girl in their group who did not have sexual relations.

Overall, the combination of these different findings make it unclear whether controls had avoided pregnancy by luck, by protecting themselves with contraceptives, or by a strong sense of self that allowed them to not be influenced by peer and/or partner pressure to engage in sexual intercourse. Nevertheless, the findings do suggest that adolescent females in La Paz lack trustworthy support networks-with their families, peer groups, and partners-that would enable them to seek information regarding sex and contraception and act upon that information. The findings further suggest that while individual factors such as high self-esteem are associated with being able to avoid unwanted pregnancy, structural factors also contribute to the health and development status of adolescent females. The focus-group discussions revealed a complex web of such factors: girls are encouraged (sometimes pressured) to have sexual relations by partners and peers, but strict social norms preclude open discussion about sexuality and c ontraception between parents and youth, as well as between adolescent partners. At the same time, transgression of norms (getting pregnant) elicits punishment and ostracism from family, peers, partners, and the broader community.

Interventions seeking to reinforce protective factors for Bolivian adolescents might, therefore, focus on improving communication and dialogue between parents and children, between partners, as well as in the larger community about sexuality, sexual risks, and ways to protect oneself against unwanted pregnancy. Since sex education exists in Bolivian schools, this can be an opportunity to discuss not only biological aspects of reproduction, but also relationships, communication, and negotiation with partners. Myths regarding promiscuity and use of contraception could be dispelled, and joint responsibility for the use of contraception among sexually active adolescents could be encouraged. A successful school- and mass media-based peer promotion initiative recently implemented in Paraguay provides a model for such an intervention (Magnani, Robinson, Seiber, & Avila, 2000).

This study illustrates the pressing need for more in-depth research into the factors that may influence the likelihood of pregnancy among young women living under similar circumstances. The biases that affected this study are, in some ways, difficult to overcome in a study limited by data collection only after pregnancy occurs. Nevertheless, the study findings do highlight the combination of structural and personal factors that interact to produce health and development risks for adolescent females in Bolivia. The findings suggest that targeting only the behavior of adolescent girls is not likely to be sufficient. In order to be successful, it is likely that parents, partners, and the community at large have to be involved in the process of reducing risks and creating opportunities for healthful development of Bolivian adolescents. The potential benefits of being able to closely target programs to reach young women at risk, and to offer programs capable of mediating the risks, are of interest to all in the f ield of adolescent reproductive health.

Table 1

Percentage of Positive/Correct Responses to Items Constituting the
Self-Esteem, Locus of Control, and Knowledge of Pregnancy Indices by
Cases and Controls

Items Cases Controls

Self-esteem

 I feel very attractive when my 76.5 88.4
 partner asks me for sex, and that's
 why I accept *
 Girls who have sex are more mature 75.6 72.4
 It would be terrible for me if I 62.4 81.0
 could not satisfy my partner
 sexually **
 If I ask questions about sex, my 51.8 64.7
 friends and family will think badly of me
 If I don't want to lose my partner, 81.2 89.8
 I have to have sex
 Being capable of having sex makes me 88.0 94.2
 feel more important
 In this day and age, it's only the shy 70.6 77.9
 and insecure girls who don't dare to
 have sex
 When I'm with my group, I don't dare 50.6 64.0
 to say what I think about sex because
 they might criticize me
 I am afraid my group will make fun 86.9 94.2
 of/reject me, and that's why I have
 sex like the other girls in the group
 Girls who can't hold on to a boyfriend 84.7 90.7
 are losers
 I may be afraid of getting pregnant, 60.7 81.4
 but I'm more afraid of being rejected
 by my partner **
 To give in to the "proof of love" is 64.7 88.1
 the only way my partner can be sure
 of me **

Locus of control

 I follow my friends' advice regarding 90.7 86.2
 sex, because that way they respect me
 more
 Following the rules of my religion 60.0 49.4
 protects me from any danger
 My boyfriend knows how to 41.2 30.9
 direct/control our relationship, and
 that's why I trust him

Knowledge of pregnancy

 A woman can get pregnant...

 When the man puts the penis in the 91.9 90.7
 vagina and leaves his milk
 Only when the man puts his penis 43.9 54.9
 in the vagina
 When she has sex without protection 95.3 86.9
 When she's taking pills and she 69.1 64.1
 forgets a day
 When the man hugs and kisses her 98.9 95.2
 strongly
 The first time she has sex with 76.3 71.8
 a man without protection
 When she is menstruating 27.7 26.5
 When the penis enters only a little 73.4 61.3
 and leaves the milk
 When she protects herself with 3.5 10.7
 birth control

Significant difference (two-tailed) between cases (n = 88) and controls
(n = 88)

* p < .05

** p < .01
Table 2

Percentage Distribution of Cases and Controls by Marital Status,
Occupation, and Dating.

Factor Category Cases Controls

Marital status Married *** 18.8 3.4
 Cohabiting ** 47.1 4.5
 Single** 34.1 92.0
 Total respondents (n) 85 88
Occupation Full-time student ** 32.9 76.1
 Working ** 21.2 5.7
 Part-time student 14.1 14.8
 Neither ** 31.8 3.4
 Total respondents (n) 85 88
Age at first boyfriend 8-12 10.6 6.2
 13-14 34.1 42.0
 15-16 48.2 45.7
 17-20 7.1 6.2
 Total respondents (n) 85 81
Number of lifetime 1 21.3 15.8
 boyfriends 2 18.7 26.3
 3 25.3 17.1
 4 + 34.7 40.8
 Total respondents (n) 75 76

Significant difference (two-tailed) between cases and controls

** p < .01

*** p < .001
Table 3

Associations Between Family Background Factors and Adolescent Pregnancy

 95%
 OR confidence McNemar's
Factors Categories (adjusted) interval [chi square]

Birthplace Other 1.00
 This city 0.65 0.30 - 1.38 1.29
Parents' birthplace Other 1.00
 This city 0.88 0.49 - 1.57 0.20
Male siblings None 1.00
 At least one 0.83 0.36 - 1.92 0.18
Female siblings None 1.00
 At least one 0.47 0.21 - 1.05 3.57

 No. No. of
 of discordant
Factors pairs pairs

Birthplace
 77 28
Parents' birthplace
 82 45
Male siblings
 88 22
Female siblings
 88 28
Table 4

Associations Between Parental Relationship Factors and Adolescent
Pregnancy

 95%
 OR confidence McNemar's
Factors Categories (adjusted) interval [chi square]

Relationship with Regular/bad 1.00
parents is... Good 0.64 0.33 - 1.24 1.78

Is communication No 1.00
with parents good? Yes 1.07 0.52 - 2.22 0.03

Is there much No 1.00
fighting in your Yes 2.44 1.13 - 5.31 5.45 *
home?

Are your parents No 1.00
affectionate? Yes 0.48 0.23 - 0.98 4.24 *

Do you trust your No 1.00
parents? Yes 0.47 0.20 - 1.09 3.24

Are your parents No 1.00
very supportive? Yes 0.39 0.16 - 0.93 4.84 *

 No. of
 No. of discordant
Factors pairs pairs

Relationship with
parents is... 82 36

Is communication
with parents good? 77 29

Is there much
fighting in your 72 31
home?

Are your parents
affectionate? 75 34

Do you trust your
parents? 71 25

Are your parents
very supportive? 68 25

* p < .05 (two-tailed)
Table 5

Associations Between Partner Relationship and Peer Influence Factors and
Adolescent Pregnancy

 95%
 OR confidence McNermar's
Factors Categories (adjusted) interval [chi square]

Partner relationship

 Overall, Regular/bad 1.00
 Relationship with Good 0.73 0.34-1.60 0.62
 partner is...

 Have you ever No 1.00
 talked with Yes 3.50 1.41-8.67 8.33 **
 partner about
 pregnancy?

 Is communication No 1.00
 with partner Yes 1.00 0.47-2.10 <0.01
 good?

 Do you fight No 1.00
 often with your Yes 0.87 0.41-1.82 0.14
 partner?

 Is your partner No 1.00
 affectionate? Yes 0.60 0.18-1.82 1.00

 Do you trust your No 1.00
 partner? Yes 1.36 0.43-2.96 0.62

 Is your partner No 1.00
 very supportive? Yes 0.77 0.34-1.75 0.39

Peer Influences

 Any of your No 1.00
 friends drink Yes 0.59 0.27-1.23 2.31
 alcohol regularly?

 Do you get No 1.00
 together with Yes 0.29 0.12-0.62 12.10 **
 friends often?

 No. No. of
 of discordant
Factors pairs pairs

Partner relationship

 Overall,
 Relationship with 72 26
 partner is...

 Have you ever
 talked with 67 27
 partner about
 pregnancy?

 Is communication
 with partner 67 28
 good?

 Do you fight
 often with your 64 28
 partner?

 Is your partner
 affectionate? 66 16

 Do you trust your
 partner? 67 26

 Is your partner
 very supportive? 65 23

Peer Influences

 Any of your
 friends drink 82 35
 alcohol regularly?

 Do you get
 together with 83 40
 friends often?

** p < .01 (two-tailed p value)
Table 6

Associations Between Self-Esteem and Knowledge Regarding Pregnancy
Indices and Adolescent Pregnancy

 95% confidence
Factors OR (adjusted) interval p value

Self-esteem index 0.03 <0.01 - 0.27 .002
(range:0-1;alpha:0.73)

Knowledge regarding 0.87 0.24 - 3.12 .833
pregnancy index
(range:0-1;alpha:0.43)

REFERENCES

Bearman, P., Bruckner, H., Brown, B. B., Theobald, W., & Philliber, S. (1999). Peer potential: Making the most of how teens influence each other. Washington, DC: The National Campaign to Prevent Teen Pregnancy.

Caton, C. L., Hasin, D., Shrout, P. E., Opler, L. A., Hirshfield, S., Dominguez, B., & Felix, A. (2000). Risk factors for homelessness among indigent urban adults with no history of psychotic illness: A case-control study. American Journal of Public Health, 90(2), 258-263.

Cole, F. L. (1997). The role of self-esteem in safer sexual practices. Journal of the Association of Nurses in AIDS Care, 8(6), 64-70.

Connelly, C. D. (1998). Hopefulness, self-esteem, and perceived social support among pregnant and nonpregnant adolescents. Western Journal of Nursing Research, 20(2), 195-209.

Kirby, D. (1999). Antecedents of adolescent sexual risk-taking, pregnancy, and childbearing: Implications for research and programs. Washington, DC: National Campaign to Prevent Teen Pregnancy.

Leffert, N., Benson, P., Scales, P., Sharma, A., Drake, D., & Blythe, D. (1998). Developmental assets: Measurement and prediction of risk behaviors among adolescents. Applied Development Science, 2(4), 209-230.

Magnani, R. J., Robinson, A., Seiber, E., & Avila, G. (2000). Impact of a health communication program on adolescent reproductive health knowledge, attitudes, and behaviors in three cities in Paraguay. Washington, DC: FOCUS on Young Adults.

Magnani, R. J., Zielinski Gutierrez, E., Seiber, E., & Vereau, D. (2000). Antecedents of sexual activity and condom use among secondary school students in urban Peru. Washington, DC: FOCUS on Young Adults Program/Pathfinder International.

McDonald, R. P. (1999). Test theory: A unified treatment (pp. 231-249). Mahwah, NJ: Erlbaum.

Mock, N. B., Magnani, R. J., Dikassa, L., Rice, J. C., Abdoh, A., & Bertrand, W. E. (1993). The utility of case-control methods for health policy and planning analysis: An illustration from Kinshasa, Zaire. Evaluation and Program Planning, 16(2), 199-205.

Modrcin-Talbott, M. A., Pullen, L., Zandstra, K, Ehrenberger, H., & Muenchen, B. (1998). A study of self-esteem among well adolecents: Seeking new direction. Issues in Comprehensive Pediatric Nursing, 21(4), 229-241.

Morgan, C., Chapar, G. N., & Fisher, M. (1995). Psychosocial variables associated with teenage pregnancy. Adolescence, 30(118), 277-289.

Mullis, A. K., & Mullis, R. L. (1997). Vocational interest of adolescents: Relationships between self-esteem and locus of control. Psychological Reports, 81(3), 1363-1371.

Population Reference Bureau. (1994). The world's youth 1994: A special focus on reproductive health.

Republica de Bolivia, Ministerio de Hacienda e Instituto Nacional de Estadistica and Macro International. (1998). Encuesta Nacional de Demografia y Salud, 1998. Calverton, MD, USA. Demographic and Health Surveys.

Resnick, M. D., Bearman, P. S., Blum, R. W., Bauman, K E., Harris, K M., Jones, J., Tabor, J., Beuhring, T., Sieving, R. E., Shew, M., Ireland, M., Bearinger, L. H., & Udry, J. R. (1997). Protecting adolescents from harm: Findings from the National Longitudinal Study of Adolescent Health. Journal of the American Medical Association, 278(10), 823-832.

UNICEF. (1998). Progress of nations. New York: UNICEF. WHO/PAHO. (1998). Health in the Americas, 1998 Edition (Volume 2). Washington, DC: Pan American Health Organization.

Zimet, G. D. (1992). Reliability of AIDS Knowledge Scales: Conceptual issues. AIDS Education and Prevention, 4(4), 338-344.

This research was supported by funds provided by USAID/Bolivia to the FOCUS on Young Adults Program/Pathfinder International under Cooperative Agreement No. CCP-A-00-96-90002-00.

The coordination and logistical assistance of Gladys Pozo, Pathfinder/Bolivia, in carrying out the study is gratefully acknowledged, as are the helpful comments of FOCUS on Young Adults Program colleague Nancy Murray on an earlier version of this article.

Vaija Lipovsek, Ali Mehryar Karim, Emily Zielinski Gutierrez, Robert J. Magnani, FOCUS on Young Adults Program, Tulane University Medical Center, School of Public Health & Tropical Medicine, New Orleans, Louisiana; Maria del Carmen Castro Gomez, independent consultant, La Paz, Bolivia.

Reprint requests to Varja Lipovsek, Tulane University Medical Center, School of Public Health & Tropical Medicine, Department of International Health & Development, 14-40 Canal Street, Suite 2200, New Orleans, Louisiana 70112. Electronic mail may be sent to vlipovs@tulane.edu.