INTRODUCTION
Adolescent pregnancy is the leading cause of mortality in women aged 15–19 years in low- and middle- income countries. According to the World Health Organization (WHO), the global adolescent pregnancy rate is estimated at 46 births per 1000 girls, and Latin America and the Caribbean have the second highest rate of adolescent pregnancies in the world, estimated at 66.5 births per 1000 women aged 15–19 years1. Mexico has the highest adolescent pregnancy rate among the Organization for Economic Co-operation and Development (OECD) countries2. In 2017, two out of ten mothers who gave birth in Mexico were aged <20 years3.
Multiple factors contribute to adolescent pregnancies, including individual, family and environmental variables. Adolescent pregnancy is linked to poverty, malnutrition, drug use, not using contraceptive methods, and a lack of knowledge about sex. Most pregnancies among adolescent girls are unplanned, especially among the lowest quintiles of poverty. Previous reports showed that when pregnancy occurs in a socioeconomically disadvantaged adolescent, a greater risk of maternal and newborn morbidity and mortality has been associated with adolescent (15–19 years) pregnancy4,5.
The neighborhood environment can include both opportunities and barriers in the prevention of adolescent pregnancy. Exposure to psychoactive substance use is one of the disadvantageous neighborhood influences on adolescent pregnancy. In Mexico, the use of psychoactive substances increased between 2011 and 2016, mainly among those aged <25 years6. The abuse of these substances during adolescence is associated with risk behaviors among pregnant women and the children born from these pregnancies7.
Despite the high rates of adolescent pregnancy observed in Mexico, few studies have examined how various factors of the federal entities are related to adolescent pregnancy. This study aims to explore how the environmental and individual factors and psychoactive substance use are associated with adolescent pregnancy in the poorest social groups. We hypothesized that girls aged 15–19 years living in areas with higher prevalence of use of psychoactive substances, more crime, and greater marginalization, would have higher odds of pregnancy.
METHODS
This study involved a secondary analysis of cross-sectional data collected in the National Health and Nutrition Survey 2018 (Spanish acronym: ENSANUT 2018), which is representative at national and state with probabilistic, multi-stage, stratified and cluster sampling. The methodological details of the survey have been described previously8.
We used data from adolescent girls aged 15–19 years with complete information about reproductive health information. The total sample for this study included 4364 girls aged 10–19 years, but 1101 were eliminated because the respondents belonged to the richest social groups (1051 girls) or had incomplete information (50 girls). The final sample included 3263 girls aged 15–19 years, based on the weighting factor.
Outcomes
The dependent variable was adolescent pregnant among girls aged 15–19 years, which was assessed using two survey questions (did not have an adolescent pregnancy=0; had an adolescent pregnancy=1). A woman was considered to have had an adolescent pregnancy if she answered positively that she had been pregnant or was currently pregnant.
Covariates
We selected covariates at the individual level that had a theoretical association with adolescent pregnancy. However, information about sexual and contraceptive measures was not collected. At the individual level, the following social and demographic measures were studied: age, ethnicity (indigenous/not indigenous), school attendance, education level, marital status (unmarried or married/in union), health insurance (private vs public health insurance), depressive symptoms [the prevalence of depressive symptomatology was measured with the Depression Scale of the Center for Epidemiologic Studies, Brief Version (CESD-7)9], use of computers (whether used a computer during the previous 12 months at least once a week), use of cell phones (whether used the cell phone during the previous 12 months at least once a week), use of internet (whether used the internet during the previous 12 months at least once a week), if the participants were beneficiaries of the conditional cash transfer program ‘Progresa/Oportunidades/Prospera’ (CCT-POP)10, region (categorized as North, Center, Mexico City, and South), and residence (rural or urban).
Exposure variables and other environment variables
The influence of prevalence of psychoactive substance use on adolescent pregnancy in Mexican was assessed using data from the 2016 National Survey of Drug, Alcohol, and Tobacco Use (ENCODAT, for its Spanish acronym)6. We determined the prevalence of drug, alcohol, and tobacco use (Table 1). Furthermore, other environment variables, such as population density11, marginalization index12, and number of reported homicides13, were considered in the study.
Table 1
Exposure variables and other environment variables
We used a harmonized dataset of individual and state-level data for 32 states (Mexico is a federal republic composed of 32 states). Therefore, we aggregated state-level data to individual data for girls aged 15–19 years based on the girl’s place of residence.
Statistical analysis
The descriptive analysis was done according to whether or not women had an adolescent pregnancy, using means and standard deviations for continuous variables, and percentages and 95% confidence intervals for categorical variables. Means were compared with Student’s t-test, and the statistical significance of observed differences between groups was determined with the chi-squared test. Finally, multilevel binary logistic regression models were performed with individuals nested within states, to test the association between adolescent pregnancy and psychoactive substance use within the context of an ecological analysis.
Models were adjusted for individual and environmental variables whose bivariate tests had a p<0.25. A model was fitted separately for each of psychoactive substance (prevalence of illegal drug use, non-prescription use of medical drugs, alcohol abuse and daily tobacco use). Also, excluded variables were reintroduced in the model to assess their association in the multivariable environment, variables that became significant or that changed the coefficient of state variables by more than 10% were maintained in the model. All models were adjusted by the same set of individual-level covariates: age, school attendance, education level, use of computers, use of cell phones, use of internet, and being a CCT-POP beneficiary.
Statistical analyses were performed using STATA version 16.0 (StataCorp, Stata Statistical Software, 2019). Significance was set at alpha=0.05 with a 95% CI. All analyses considered the original multistage stratified sampling design of the surveys using the SVY module for complex samples.
RESULTS
Table 2 provides the individual characteristics of the weighted sample by outcome; 15.9% were girls who had a pregnancy, and the average age was significantly higher in the group with adolescent pregnancy (17.8 ± 1.2 years) than in the group without adolescent pregnancy (16.7 ± 1.4 years). Girls who quit school had a higher proportion of pregnancies than those who attended school. The proportion of adolescent pregnancy was significantly higher for girls with primary school or less (40.5%) and middle school (21.1%) than for those with a high school (8.7%) or college (8.2%). Most married or in union girls reported having been pregnant (79.3%). Among those using computers, cell phones and internet, the proportion of women reporting ever having been pregnant was lower than among those who did not use Information and Communication Technology (ICT). The prevalence of depressive symptomatology and indigenous background were not significantly associated with adolescent pregnancy.
Table 2
Baseline characteristics of girls according to adolescent pregnancy outcome
| Individual characteristics | Without adolescent pregnancy | With adolescent pregnancy | p | Total |
|---|---|---|---|---|
| Total, n | 2695 | 568 | 3263 | |
| Total (weighted), n (%) | 3356268 (84.14) | 632734 (15.86) | 3989002 | |
| Age (years), mean ± SD | 16.69 ± 1.35 | 17.76 ± 1.24 | <0.001* | 16.97 ± 1.48 |
| Ethnicity | % (95% CI) | % (95% CI) | % (95% CI) | |
| Indigenous | 87.4 (81.69–91.51) | 12.6 (8.49–18.31) | 0.227 | 6.52 (5.18–8.19) |
| Not indigenous | 83.91 (81.95–85.7) | 16.09 (14.3–18.05) | 93.48 (91.81–94.82) | |
| School attendance | ||||
| Yes | 97.33 (96.4–98.03) | 2.67 (1.97–3.6) | <0.001* | 61.39 (58.88–63.84) |
| No | 63.16 (59.26–66.89) | 36.84 (33.11–40.74) | 38.61 (36.16–41.12) | |
| Education level | ||||
| Primary school or less | 59.49 (50.19–68.15) | 40.51 (31.85–49.81) | <0.001* | 6.41 (5.35–7.67) |
| Middle school | 78.94 (75.51–82.01) | 21.06 (17.99–24.49) | 41.62 (39.15–44.12) | |
| High school | 91.3 (89.34–92.94) | 8.7 (7.06–10.66) | 48.01 (45.52–50.51) | |
| College | 91.78 (81.86–96.5) | 8.22 (3.5–18.14) | 3.96 (3.11–5.03) | |
| Marital status | ||||
| Unmarried | 96.74 (95.83–94.76) | 3.26 (2.54–4.17) | <0.001* | 83.42 (81.49–85.18) |
| Married/in union | 20.74 (16.51–25.71) | 79.26 (74.29–83.49) | 16.58 (14.82–18.51) | |
| Health insurance | ||||
| Private | 89.32 (85.56–92.18) | 10.68 (7.82–14.44) | 0.002* | 26.64 (24.34–29.07) |
| Public | 82.26 (80.04–84.28) | 17.74 (15.72–19.96) | 73.36 (70.93–75.66) | |
| Depressive symptoms | ||||
| No | 84.22 (82.28–85.99) | 15.78 (14.01–17.72) | 0.783 | 88.48 (86.69–90.05) |
| Yes | 83.49 (77.76–87.97) | 16.51 (12.03–22.24) | 11.52 (9.95–13.31) | |
| Use of computers | ||||
| Yes | 89.44 (85.86–92.19) | 10.56 (7.81–14.14) | <0.001* | 33.54 (31.2–35.98) |
| No | 81.46 (79.19–83.54) | 18.54 (16.46–20.81) | 66.46 (64.02–68.8) | |
| Use of cell phones | ||||
| Yes | 85.04 (83.14–86.77) | 14.96 (13.23–16.86) | <0.001* | 89.63 (88–91.06) |
| No | 76.33 (69.27–82.19) | 23.67 (17.81–30.73) | 10.37 (8.94–12) | |
| Use of internet | ||||
| Yes | 89.45 (86.38–91.89) | 10.55 (8.11–13.62) | <0.001* | 35.24 (32.9–37.64) |
| No | 81.25 (78.83–83.44) | 18.75 (16.56–21.17) | 64.76 (62.36–67.1) | |
| CCT-POP | ||||
| Non-beneficiary | 81.9 (79.22–84.3) | 18.1 (15.7–20.78) | 0.004* | 43.66 (41.19–46.16) |
| Beneficiary | 87.02 (84.5–89.19) | 12.98 (10.81–15.5) | 56.34 (53.84–58.81) | |
| Region | ||||
| North | 80.28 (75.97–83.98) | 19.72 (16.02–24.03) | 0.17 | 14.05 (13.09–15.06) |
| Center | 84.73 (81.06–87.8) | 15.27 (12.2–18.94) | 37.59 (35.34–39.9) | |
| Mexico City | 89.74 (80.75–94.8) | 10.26 (5.2–19.25) | 7.74 (6.44–9.27) | |
| South | 83.86 (81.29–86.13) | 16.14 (13.87–18.71) | 40.62 (38.79–42.48) | |
| Residence | ||||
| Urban | 84.72 (82.48–86.72) | 15.28 (13.28–17.52) | 0.351 | 68.24 (66.36–70.07) |
| Rural | 82.89 (79.35–85.93) | 17.11 (14.07–20.65) | 31.76 (29.93–33.64) |
Table 3 presents the environmental characteristics where adolescent women resided at the time of survey by outcome. Although none of the environmental variables showed a significant association, we observed that adolescent women who had a pregnancy lived in areas with a higher prevalence of psychoactive substance use than those who did not have a pregnancy.
Table 3
Characteristics for the states of girls without adolescent pregnancy (N=2695) and with adolescent pregnancy (N=568)
Based on multilevel logistic models, girls who quit school were more likely to have had a pregnancy compared to those who attended school (AOR=9.60, 95% CI: 6.64–14.00). Similarly, girls with college (AOR=0.21, 95% CI: 0.09–0.53) or high school education (AOR=0.26, 95% CI: 0.16–0.44) were less likely to have has a pregnancy that those with primary education or less. Girls who did not use internet had a greater risk of pregnancy than those who used internet (AOR=1.50, 95% CI: 1.10–2.01), and being CCT-POP beneficiary decreased a woman’s risk of adolescent pregnancy (AOR=0.50, 95% CI: 0.37–0.67) (Table 4).
Table 4
Adjusted associations between adolescent pregnancy and psychoactive substance use
| Predictor | Category | Adolescent pregnancy | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Model 1 | Model 2 | Model 3 | Model 4 | ||||||
| AOR | 95% CI | AOR | 95% CI | AOR | 95% CI | AOR | 95% CI | ||
| Age | 1.76* | 1.55–2.01 | 1.76* | 1.55–2.01 | 1.76* | 1.55–2.01 | 1.76* | 1.55–2.01 | |
| School attendance | Yes (Ref.) | 1 | 1 | 1 | 1 | ||||
| No | 9.64* | 6.64–14.02 | 9.62* | 6.64–13.94 | 9.63* | 6.63–13.99 | 9.61* | 6.64–13.90 | |
| Education level | Primary or less (Ref.) | 1 | 1 | 1 | 1 | ||||
| Middle school | 0.87 | 0.53–1.42 | 0.86 | 0.53–1.40 | 0.85 | 0.52–1.41 | 0.87 | 0.53–1.43 | |
| High school | 0.26* | 0.16–0.44 | 0.26* | 0.16–0.43 | 0.26* | 0.15–0.43 | 0.26* | 0.16–0.44 | |
| College | 0.21* | 0.09–0.53 | 0.21* | 0.09–0.53 | 0.21* | 0.08–0.52 | 0.21* | 0.09–0.53 | |
| Use of computers | Yes (Ref.) | 1 | 1 | 1 | 1 | ||||
| No | 0.91 | 0.60–1.38 | 0.91 | 0.59–1.38 | 0.91 | 0.60–1.38 | 0.91 | 0.60–1.39 | |
| Use of cell phones | Yes (Ref.) | 1 | 1 | 1 | 1 | ||||
| No | 1.14 | 0.71–1.84 | 1.15 | 0.72–1.83 | 1.13 | 0.70–1.84 | 1.15 | 0.72–1.85 | |
| Use of internet | Yes (Ref.) | 1 | 1 | 1 | 1 | ||||
| No | 1.49* | 1.10–2.01 | 1.49* | 1.10–2.02 | 1.50* | 1.11–2.02 | 1.48* | 1.09–2.02 | |
| CCT-POP | No (Ref.) | 1 | 1 | 1 | 1 | ||||
| Yes | 0.50* | 0.37–0.67 | 0.50* | 0.37–0.68 | 0.50* | 0.37–0.69 | 0.50* | 0.37–0.67 | |
| Population density by statea | 0.90* | 0.88–0.93 | 0.89* | 0.86–0.92 | 0.84* | 0.82–0.87 | 0.85* | 0.82–0.88 | |
| Number of homicides** | 1.08* | 1.03–1.13 | |||||||
| Use illegal drugs | 1.22* | 1.17–1.27 | |||||||
| Use medical drugs without prescription | 2.00* | 1.66–2.39 | |||||||
| Use tobacco daily | 1.05* | 1.03–1.08 | |||||||
| Abuse alcohol | 1.10* | 1.03–1.18 | |||||||
AOR: adjusted odds ratio; all models adjusted by the set of individual variables of each girl: age, school attendance, education level, use of computers, use of cell phones, use of internet and CCT-POP.
** Model 4 adjusted by number of homicides: re-scaled using z-scores, so that a one-unit change represents a one standard deviation change in homicides of states.
Regarding environmental factors, psychoactive substance use was directly associated with an increase in adolescent pregnancies. The increase in one unit of prevalence of illegal drug use, non-prescription use of medical drugs, alcohol abuse and daily tobacco use was associated with an increase in the odds of adolescent pregnancy (AOR=1.22, 95% CI: 1.17–1.27; AOR=2.00, 95% CI: 1.66–2.39; AOR=1.10, 95% CI: 1.03–1.13; and AOR=1.05, 95% CI: 1.03–1.18, respectively). In the four models, higher state population density was associated with higher odds of adolescent pregnancy. In addition, in Model 3 we observe that a one standard deviation increase in homicides was associated with a 8% increase in the odds of inadequate early childhood education (AOR=1.08, 95% CI: 1.03–1.13).
DISCUSSION
The most relevant findings from this work can be summarized as follows: First, we found that a higher prevalence of illegal drug use, non-prescription use of medical drugs, and tobacco and alcohol consumption in federal entities were associated with greater odds of adolescent pregnancy. Second, regarding individual characteristics, not attending school, low education level, not using the internet, and not being a CCT-POP beneficiary, were associated with an increased risk of adolescent pregnancy.
Adolescent substance use is an enduring problem in Mexico. One of the problems associated with the psychoactive substance use is risky sexual behavior among adolescents, since sexual activity and lack of contraceptive use are carried out under the effects of these substances. These behaviors are associated with unplanned pregnancies14–16. Our study adds evidence related to the presence of strong associations of higher prevalence of psychoactive substance use with greater risk of adolescent pregnancy in Mexico, after adjustment for individual characteristics. Studies have reported that the unintended pregnancy rate was highest among adolescent girls who used marijuana, cocaine, and opioid analgesics. Adolescents who used tobacco, alcohol, or other drugs, were more likely to be sexually active compared to adolescents who did not used psychoactive substances17.
Several studies reported that marijuana use among adolescents is associated with sexual risk behaviors18,19. The activation of cannabinoid receptors can increase sexual desire and sexual satisfaction in women18. These receptors affect brain regions that influence pleasure, memory, thought, concentration, sensory and time perception20. Sumnall et al.21 found that cannabis use improved the sexual experience and facilitated the sexual encounter.
Our findings also showed that living in more densely populated areas was associated with lower odds of having adolescent pregnancy. Large cities allow greater access to education, childcare facilities, and better employment opportunities for girls22. The highest rates of adolescent pregnancy were found in populations living in more marginalized areas where residents have a lack of information and restricted access to services23. Economic growth and development plays an important role to understand determinants associated with adolescent pregnancy rate24.
Recent studies found that adolescents living in neighborhood with higher homicide rates have a significant impact on health and adolescent well-being25–29. Our findings are similar to this previous evidence, which may be due to widespread violence and organized crime in Mexico. Crime can potentially exacerbate drug use that has unintended consequences such as adolescent pregnancy and sexually transmitted diseases, and becoming a victim of physical or sexual abuse26,30–32.
Additionally, our findings show that adolescent pregnancy was strongly associated with school attendance, use of Information and Communication Technology, education, and targeted programs. Prior research found that adolescents living in neighborhoods with concentrated poverty are associated with higher unemployment rates, lower education level, and higher adolescent pregnancy rates, initiating sexual intercourse at younger ages33–38. Furthermore, being a CCT-POP beneficiary in Mexico was associated with a lower prevalence of early unions and pregnancies, as well as a higher school prevalence, which could be due to the cash incentives provided by the program, so that adolescents stay in school39,40.
Strengths and limitations
To our knowledge, this is the first study in Mexico showing nationally representative information with strong association between adolescent pregnancy and psychoactive substance use. Nonetheless, some limitations must be considered. First, causality cannot be determined due to the cross-sectional nature of the study design. Second, it was not possible to analyze whether a girl/woman used illegal drugs or non-prescribed medical drugs. Third, the environmental factors were measured at the federal entity level, which was the geographical level that was available for ENCODAT. The lack of disaggregated information on key exposures might limit our ability observe associations. For example, we acknowledge that homicide rates might not capture all the violence occurring in an area, which might be underestimating the association between the true contextual violence and adolescent pregnancy. Finally, we do not have the information of new psychoactive substances (NPS) use. Previous studies showed that adolescents may be more vulnerable to the harmful effects of these drugs than adults41,42.
CONCLUSIONS
Our findings show that a mix of individual and environmental factors was associated with adolescent pregnancy. Individual factors such as school attendance, education level, and cash transfer programs, as well as environmental factors such as prevalence of psychoactive substance use and population density, influenced adolescent pregnancy. Our study provides insights that can be used to lead policies and plan actions to prevent adolescent pregnancy and reduce high rates in the Mexican population.
