Forced sexual initiation among adolescent girls aged 15–19 years: evidence from four African countries using the performance monitoring for action survey

Abstract

Objectives

To assess the prevalence of forced sexual initiation and its associated factors among adolescent girls in four sub-Saharan African countries.

Design

Cross-sectional study design.

Setting

Four sub-Saharan African countries: Burkina Faso, Côte d’Ivoire, Kenya and Uganda included in this study.

Data source

Performance monitoring for action survey data collected in 2020.

Participants

A total of 2307 weighted samples of adolescent girls aged 15–19 years were included in this study.

Outcome measure

Forced sexual initiation among adolescent girls.

Results

The pooled prevalence of forced sexual initiation was 24% (95% CI 17% to 31%) among adolescent girls aged 15–19 years in four sub-Saharan African countries. Age at sexual debut, relationship with the household head, educational status, wealth status, marital status and country of residence were statistically significant factors associated with forced sexual initiation among adolescent girls.

Conclusions

Compared with that of the global target, the burden of forced sexual initiation among adolescent girls in the four sub-Saharan Africa was greater. This indicates that action is needed in this area to reach the global Sustainable Development Goals by 2030. Individual-level and community-level factors were associated with forced sexual initiation.

Strengths and limitations of this study.

• Nationally representative multicountry data provided generalisable evidence on forced sexual initiation in an understudied research area.

• A multilevel modelling approach was employed to effectively account for the hierarchical structure of the data.

• Inclusion of adolescent girls minimised recall bias related to forced sexual initiation compared with older age groups.

• The cross-sectional nature of the study limits the ability to establish temporal or causal relationships between variables.

• The use of secondary data limited the inclusion of important explanatory variables, such as the perpetrator’s age, relationship to the perpetrator and substance use factors.

Introduction

Forced sexual initiation refers to a woman’s first sexual intercourse that occurs against her will through pressure, coercion or physical force.^{1 2} It is recognised as a major public health problem with both immediate and long-term impacts on girls’ lives.³ Moreover, it constitutes a violation of women’s reproductive and sexual health rights and is considered a form of gender-based violence and human rights abuse.⁴ More broadly, sexual violence remains a significant public health, human rights and development challenge affecting societies worldwide. Evidence also indicates that being female is associated with an increased likelihood of early sexual victimisation.⁵

It is estimated that one in three women will experience sexual or physical violence in their lifetime by an intimate partner or sexual violence from a non-partner that has remained largely unchanged over the past decade.⁶ In particular, sub-Saharan Africa (SSA) is one of the regions with the highest global burden of sexual violence against girls, alongside South Asia. Within SSA, the burden is greater in the southern, eastern and western regions bear a greater burden than in the central regions.⁷

Adolescent girls bear a disproportionately higher burden of sexual violence. According to a 2024 UNICEF report, one in eight girls experienced rape or sexual assault before the age of 18.⁸ Similarly, the WHO 2024 report indicated that nearly a quarter of adolescent girls who had been in relationships experienced physical or sexual violence.⁹ In addition, a systematic review revealed that one out of four female children experienced sexual violence.¹⁰

Likewise, evidence from a 2015 study conducted among young adult women in low- and middle-income countries estimated the prevalence of forced sexual debut at 12%, with a greater burden observed in East and South Africa.¹¹ Furthermore, a study conducted in four sub-Saharan African countries among girls aged 12–19 years reported that 15% in Burkina Faso, 23% in Uganda, 30% in Ghana and 38% in Malawi experienced forced sexual initiation.¹² Consistently, a study conducted in South Africa revealed that nearly 46% of all sexually experienced young women reported that their first sexual encounter had been coerced.¹³ Similarly, studies from Malawi and Ghana reported that the prevalence of forced sexual initiation was 38.9% in Malawi,¹ while the prevalence of sexual violence at sexual debut in Ghana was 25.2%.¹⁴

Forced sexual practices against girls are strongly influenced by a society’s culture, norms and beliefs, particularly in African societies where patriarchal structures favour men over women, placing women in a culturally subordinate position with limited control over decision-making. As a result, women’s needs and desires are frequently disregarded, making them vulnerable to sexual violence.¹⁵ In addition, gender roles and expectations, religious beliefs, rape myths and bride prices all contribute to sexual coercion.¹⁶ Furthermore, in African traditions, issues concerning sex are rarely discussed openly and women are often not allowed to openly express sexuality.¹⁷ Consequently, the sociocultural environment shapes the lives of girls and often leads to their tolerance of sexual coercion.¹⁸

Forced sex has long-term negative effects on the gynaecological and sexual health of affected girls. These effects include urinary tract infections, dyspareunia, pelvic pain, irregular menstrual cycles, unintended pregnancy, attempted abortion and adverse childbirth outcomes.^19–21 In addition, forced sexual initiation can lead to adverse sexual outcomes such as impaired sexual function and desire, hypoactive sexual desire, sexual aversion, arousal disorders, fear of sex and difficulties with touch.^22–25

Moreover, the literature indicates the effect of forced sexual initiation is associated with subsequent chronic diseases such as HIV, violence perpetration and subsequent sexual victimisation. Girls who experience forced sexual debut at an early age often suffer from severe mental health consequences, especially during stages of development when they are not yet capable of making rational decisions or controlling emotional impulses. Furthermore, poor mental health outcomes, including post-traumatic stress disorder and substance abuse, have also been linked with forced sexual debut.^{2 26–31} Similarly, women who experience coerced sexual debut are more likely to suffer from depression and other psychological problems.³² Consistent with these findings, a study conducted in China reported that forced early sexual initiation with a greater risk of suicide attempt.³³

According to previous studies, the age of the girl at sexual debut, place of residence, level of education, marital status, wealth index, relationship with the household head, country and region of SSA are associated with forced sexual initiation.^{1 12 13 34–39}

The world is striving to achieve gender equality and empower all women and girls. In this regard, one of the key focuses of Sustainable Development Goal (SDG) 5 is eliminating sexual violence against women and girls by 2030.³ Providing up-to-date evidence is important to present current evidence on the magnitude and contributing factors to forced sexual initiation among adolescent girls. However, to the best of our knowledge, there is limited and outdated evidence on the prevalence of forced sexual initiation and its associated factors among adolescent girls. Therefore, our study assessed the magnitude of forced sexual initiation and its associated factors among adolescent girls via performance monitoring for action (PMA) data from four sub-Saharan African countries.

Furthermore, the findings from this study will be insightful for policymakers, women and non-governmental organisations working on women’s sexual rights and equality in designing policies, targeted programmes, and interventions aimed at the prevention and reduction of forced sexual initiation in the four Sub-Saharan African countries and the broader SSA region.

Method

Data source and study setting

The study was conducted in four sub-Saharan African countries: Burkina Faso, Côte d’Ivoire, Kenya and Uganda. These countries are part of the PMA survey conducted in the region. The Bill & Melinda Gates Institute for Population and Reproductive Health at the Johns Hopkins Bloomberg School of Public Health funded the PMA surveys. We used the PMA data which are freely available at https://www.pmadata.org/countries.⁴⁰ PMA data are collected through a partnership between the Johns Hopkins Bloomberg School of Public Health (USA) and the respective countries’ Ministries of Health, public health institutes and statistical agencies.⁴¹ The PMA collects nationally representative cross-sectional and longitudinal data on key maternal and newborn health indicators, index pregnancy care, family planning, women’s empowerment and norms and other relevant national reproductive health priorities. The surveys are conducted annually and include follow-up with a selected cohort of women each year.

The PMA initially provided training for field staff, including resident enumerators (REs), quality control supervisors and supervisors, on the survey protocol and data collection tools, including consent administration and research ethics. Data were collected by female REs via a pretested, standardised questionnaire on mobile phones equipped with Open Data Kit (ODK) software, with close supervision of fieldwork. The ODK application enables REs and supervisors to collect and transfer survey data to a central ODK aggregate cloud server.

The study setting included four sub-Saharan African countries, namely, Burkina Faso, Côte d’Ivoire, Kenya and Uganda. According to a World Bank report in 2023, Burkina Faso is classified as a low-income country with a total population of approximately 23 million, of which 12 million (50.2%) are female and 73% reside in rural areas. The literacy rate among female youth (aged 15−24) is 52%. Côte d'Ivoire has a population of approximately 31 million, with 15.2 million (49.1%) females and 47% living in rural areas. It is classified as a lower-middle-income country, with a female youth literacy rate of 58%. Kenya’s population is approximately 55.3 million, with 27.8 million (50.3%) females and 70% residing in rural areas. It is also categorised as a lower-middle-income country, with a female youth literacy rate of 95%. Uganda has a population of approximately 48.6 million, with 24.4 million (50.2%) females and 73% living in rural areas. It is considered a low-income country, with a female youth literacy rate of 93%.⁴²

On the basis of the World Bank’s Country Policy and Institutional Assessment, countries are evaluated for their policies and institutions promoting gender equality on a scale from one to six, with the highest score indicating better gender equality. Kenya, Côte d'Ivoire and Burkina Faso each scored 3.5, whereas Uganda scored 3.0.⁴²

Study design and study period

A community-based cross-sectional study was conducted among women in 2020 in four sub-Saharan African countries.

Sample size and sampling procedure

The PMA surveys are nationally representative and use a multistage stratified cluster design with urban and rural strata. The results are representative at both the national and urban/rural levels. Census-based enumeration areas (EAs) were randomly selected, followed by a random selection of households within each EA, and data were collected by interviewing eligible participants.

At baseline, 35 households were randomly selected within each EA for interviews. The residential enumerator (RE) administered a household questionnaire, including completing a census of household members and guests who slept there the previous night for all selected households who consented to participate. Eligible females aged 15−49 years, who were usual members, consented to participate in the female panel survey. Guests that slept at the households were included in the female cross-sectional survey.

All female household members aged 15–49 years were eligible to participate in the PMA female survey. The four countries included in this study collected data in phases starting in 2020 via an open panel design. Women who participate in early phases are followed annually unless they are lost to follow-up, in which case new participants are enrolled to replace them. For this analysis, we used data from the first-phase survey of each country’s household and female questionnaires due to the nature of the data, since each country’s baseline (phase 1) survey collected baseline information related to our outcome variable, whereas in other phases, baseline information was collected only for those enrolled to replace the lost-to-follow-up samples.

Phase 1 of the Household and Female Survey includes 167 EAs in Burkina Faso, 122 in Côte d’Ivoire, 308 in Kenya and 122 selected EAs in Uganda. Further details on the PMA data collection methods are explained elsewhere.⁴¹

For this study, we used female and household records from the phase 1 PMA surveys of all four countries. The study population consisted of adolescent girls aged 15–19 years, as the outcome variable data were collected for this age group. Due to the sensitivity of the study question in this setting, there was a higher missing rate to the outcome variable among adolescent girls. Samples with missing data on the outcome variable were excluded from the analysis, in accordance with the PMA Guideline (figure 1). After exclusion of 76 EAs due to missing outcome data, a total of 643 EAs (160 in Burkina Faso, 119 in Côte d’Ivoire, 254 in Kenya and 110 in Uganda) were included in the final analysis.

Figure 1. Flow diagram of study sample selection from four sub-Saharan African countries PMA survey. PMA, performance monitoring for action.

Variables

Outcome variables

Forced sexual initiation was assessed by asking respondents about their first sexual experience via the following question: ‘Were you forced against your will?’ If the respondent answered ‘yes’, she was considered to have experienced forced sexual initiation.

Explanatory variables

On the basis of the literature review, multiple explanatory variables were included in this study. These variables were classified into individual-level and community-level categories. Individual-level variables included age at sexual debut (<15, 15−17 and ≤18), relationship to the head of household (part of the nuclear family (includes husband, wife and children); extended family (other relatives not part of the nuclear family); house help; other), marital status (single, married/living with partner, widowed/separated), educational status (no education, primary education, secondary education/post primary vocational, higher education) and wealth status (poor, middle, rich). Race and ethnicity variables were not available in the PMA dataset for the countries included in this study and therefore could not be included in the analysis.

The community-level variables included place of residence (rural, urban) and country of residence (Kenya, Burkina Faso, Côte d'Ivoire and Uganda). Community-level variables were not self-reported. Urban/rural residence was based on census classifications provided by the respective National Statistical Offices during survey design and assigned at the EA level. Thus, these classifications were prespecified before data collection and derived from EA-level information rather than respondent self-report. No rurality index score was used.

Data management and statistical analysis

Stata V.17 was used for data management and analysis. Data completeness was checked through exploratory data analysis, and any item non-response was excluded from the analysis. Frequency distributions were run to exclude responses marked as ‘Do not Know’ (DK) or ‘No Response’ (NR). The sample was weighted prior to statistical analysis via the weighting variables ‘FQweight’ and ‘svyset’ to account for unequal selection probabilities, as the unit of analysis was female respondents, in accordance with the PMA Analysis Manual (V.5).⁴³

In accordance with PMA guidelines, a complete-case analysis was used as the primary analytical approach; however, potential bias due to missing outcome data was assessed using sensitivity analyses. To evaluate the pattern of missingness, we examined predictors of outcome non-response using multivariable logistic regression. In addition, a sensitivity analysis was conducted using inverse probability weighting (IPW). IPW was derived from a logistic regression model predicting the probability of having complete outcome data based on observed covariates. The inverse of these predicted probabilities was used as analytical weights in the regression models. The IPW approach assumes that missingness is conditionally independent of the outcome given observed covariates.

The PMA data have a clustering nature; women are nested within an EA, so women in the same EA are more likely to share similar characteristics. This violates the classical logistic regression assumption of the independence of observations. Hence, a multilevel binary logistic regression model with a fixed and random effect component was used to handle the clustering nature of the observations. We fitted four models for the analysis: the null model, model one (individual-level factors), model two (community-level factors) and model three (both individual-level and community-level factors). For each model, we reported the log likelihood and deviance. We then selected the best fitted model with a higher log likelihood and lower deviance.

In the random effect analysis, we computed the intraclass correlation coefficient (ICC), proportional change in variance (PCV) and median OR (MOR). The ICC measures the proportion of total variance in forced sexual initiation attributed to a cluster. It is calculated via the following formula: ICC=σ²/(σ² + π²/3), where σ² is the variance between clusters. An ICC greater than 0.10 in the null model indicated that there was clustering, so we used a mixed effect model rather than a classical model for the analysis.⁴⁴

The PCV assesses the variation in the model’s variance when different factors are included to explain the outcome. It is calculated via the following formula: PCV=[(V₀−V₁)/V₀]×100; V₀ is the variance of the null model, and V₁ is the variance of the other models. The MOR indicates the central value of the OR between the highest-risk and lowest-risk regions when two EAs are chosen at random. MOR=exp (sqrt (2×Va)×0.6745); Va is the variance of the random effect model.⁴⁵

In the fixed effect analysis, first, we selected potential factors in the bivariable analysis on the basis of a p<0.20; then, we performed multivariable analysis. Finally, variables with p values <0.05 and adjusted OR CIs did not include one declared significant factor. During the multilevel analysis, the multicollinearity of independent variables was checked by the variance inflation factor (VIF), and those variables with a VIF >10 were considered to have high multicollinearity and were excluded from the analysis to avoid variable redundancy.⁴⁶

Patient and public involvement

None.

Results

Background characteristics of the study participants

We included 2307 (weighted) samples from adolescent girls aged 15–19 years. One-fourth (24.7%) of the study participants’ sexual debut was less than 15 years. More than one-third (32.9%) of the study participants were from Kenya. More than two-thirds (72.1%) of the study participants lived in rural areas, and more than half (60.1%) of the study participants were single (table 1).

Table 1. Background characteristics of study participants.

Variable	Frequency	Proportion (%)
Age at sexual debut (in years)
Less than 15	570	24.7
15–17	1475	64.0
18 and above	262	11.3
Educational status
No education	403	17.5
Primary education	809	35.1
Secondary education	1052	45.6
Higher education	44	1.9
Relation to household head
Part of nuclear family	1673	72.5
Part of extended family	489	21.2
House help	29	1.2
Other	116	5.0
Marital status
Single	1403	60.8
Married/living with partner	615	26.7
Divorced/separated	289	12.5
Wealth status
Lowest tertile	895.1	38.8
Middle tertile	748	32.4
Highest tertile	662	28.7
Residence
Urban	643	22.9
Rural	1664	72.1

The distribution of the sample before weighting, as well as background characteristics stratified by country for both unweighted and weighted samples, is presented in online supplemental informations 1 and 2.

Forced sexual initiation among adolescent girls

The prevalence of forced sexual initiation among adolescent girls in four SSA countries overall was 23.1% (95% CI 21.4% to 24.8%) (figure 2).

Figure 2. Forced sexual initiation among adolescent girls in four sub-Saharan Africa countries.

In terms of country, Uganda (30.5%) had the highest magnitude of forced sexual initiation, whereas Kenya (16.3%) had the lowest proportion of forced sexual initiation among adolescent girls (figure 3).

Figure 3. Forced sexual initiation among adolescent girls across countries of sub-Saharan Africa.

In terms of residence, the prevalence of forced sexual initiation among adolescent girls was greater among urban residents (24.7%; 95% CI 21.5% to 27.9%) (figure 4A). In terms of educational status, the highest prevalence of forced sexual initiation was observed among girls with no formal education, whereas it was lowest among girls with higher education (8.7%; 95% CI 3.4% to 17.2%) (figure 4B). In terms of wealth status, girls from middle-wealth tertile households experienced a greater magnitude of forced sexual initiation (24.2%; 95% CI 21.1% to 27.3%) (figure 4C).

Figure 4. Forced sexual initiation among adolescent girls by residence (A), educational status (B), and wealth status (C) in sub-Saharan Africa.

Factors associated with forced sexual initiation

With respect to the models and random effects, the null model indicated that 24.6% (ICC=0.246) of the proportion of forced sexual initiation was due to between-cluster variation. On the basis of model comparison, the final best-fit model is model 3 (both individual and community variables), since it has a lower deviance. In this model, the MOR of 2.60 indicates the central value of the OR between the highest-risk and lowest-risk regions when two EAs are randomly selected (table 2).

Table 2. Random effect parameters of the models and model comparison.

Parameter	Null model	Model 1	Model 2	Model 3
Variance (95% CI)	1.074 (0.74 to 1.56)	1.128 (0.77 to 1.64)	0.941 (0.63 to 1.41)	1.005 (0.68 to 1.50)
ICC	0.246	0.258	0.222	0.234
PCV	ref	−5.02	12.4	6.4
MOR	2.68 (2.27 to 3.29)	2.75 (2.31 to 3.39)	2.52 (2.13 to 3.10)	2.60 (2.20 to 3.22)
LL	−1244	−1209	−1225	−1191
Deviance	2488	2418	2450	2382

ICC, intraclass correlation coefficient; LL, Log-Likelihood; MOR, median OR; PCV, proportional change in variance.

Age at sexual debut, educational status, relationship with the household head, wealth status, marital status and country of residence had statistically significant associations with forced sexual initiation among female adolescent girls in four SSA countries.

Compared with girls aged 18 years and above, girls aged 15–17 years were 2.75 times (AOR=2.75, 95% CI 1.78 to 4.26) and 1.49 times (AOR=1.49, 95% CI 1.04 to 2.29) more likely to experience forced sexual initiation, respectively. Compared with those with higher education, girls who attended only primary education had 3.61 times greater odds of experiencing forced sexual initiation (AOR=3.61, 95% CI 1.16 to 11.30). Compared with girls who are part of nuclear families, such as daughters, house-helper girls had 2.72 times (AOR=2.72, 95% CI 1.22 to 6.04) greater odds of experiencing forced sexual initiation (table 3).

Table 3. Factors associated with forced sexual initiation among adolescent girls (model 3 represents the final adjusted model).

Variables	Null model	Model 1 AOR (95% CI)	Model 2 AOR (95% CI)	Model 3 AOR (95% CI)
Age at sexual debut (in years)
Less than 15		2.74 (1.772 to 4.24)*		2.78 (1.80 to 4.31)*
17		1.55 (1.04 to 2.29)*		1.52 (1.02 to 2.25)*
18 and above		1		1
Educational status
No education		3.13 (0.96 to 10.1)		2.66 (0.81 to 8.73)
Primary education		3.74 (1.19 to 11.7)*		3.60 (1.15 to 11.3)*
Secondary education/post primary vocational		2.93 (0.96 to 9.09)		2.91 (0.93 to 9.08)
Higher education		1		1
Relation to house head
Part of nuclear family		1		1
Part of extended family		0.62 (0.36 to 1.08)		0.61 (0.35 to 1.06)
House help		2.74 (1.23 to 6.10)*		2.68 (1.22 to 5.90)*
Other		1.06 (0.66 to 1.73)		1.00 (0.61 to 1.61)
Wealth tertiles
Lowest tertile		1		1
Middle tertile		1.29 (0.97 to 1.72)		1.36 (1.01 to 1.82)*
Highest tertile		1.15 (0.84 to 1.59)		1.14 (0.78 to 1.63)
Marital status
Married/living with partner		1		1
Divorced/separated		2.21 (1.04 to 4.68)*		1.55 (1.03 to 2.34)*
Single		1.04 (0.80 to 1.37)		1.18 (0.85 to 1.63)
Residence
Urban			1	1
Rural			1.26 (0.94 to 1.68)	1.26 (0.90 to 1.76)
Country of residence
Kenya			1	1
Burkina Faso			2.02 (1.39 to 2.095)*	2.32 (1.53 to 3.53)*
Côte d'Ivoire			2.56 (1.73 to 3.82)*	2.80 (1.83 to 4.27)*
Uganda			2.87 (1.93 to 4.28)*	2.84 (1.85 to 4.35)*

^*Statistically significant (p value<0.05).

AOR, Adjusted Odds Ratio.

Sensitivity analyses were conducted to assess the robustness of the findings to missing outcome data. Multivariable logistic regression indicated that outcome missingness was associated with several observed covariates, including age at sexual debut, marital status, country of residence and relationship to the household head. The IPW analysis produced results broadly consistent with the complete-case analysis, with no meaningful changes in the direction or magnitude of the main associations. Associations for age at sexual debut, marital status, country of residence and relationship to the household head remained stable. However, associations for educational status and wealth tertile were attenuated and no longer statistically significant. Overall, the findings suggest that the main conclusions are robust to potential bias arising from missing outcome data (online supplemental information 3).

Discussion

The current study aimed to assess the prevalence of forced sexual initiation and its associated factors among adolescent girls aged 15–19 years in four SSA countries. The prevalence of forced sexual initiation among adolescent girls was 23.1% (95% CI 21.4% to 24.8%). There was variation across residence, wealth status and educational status. The factors significantly associated with forced sexual initiation included age at sexual debut, relationship with the household head, educational status, wealth status, marital status and country of residence.

The prevalence of forced sexual initiation among adolescent girls in four SSA countries was higher than that reported in a previous study conducted in low- and middle-income countries.¹¹ One possible explanation is that the population in the previous study consisted of reproductive-aged women, whereas our study focused on adolescent girls, who are more vulnerable to sexual violence.⁴⁷ Another reason is the study setting: the previous study included middle-income countries in addition to lower-income countries, which may have lowered the pooled estimate of forced sexual initiation. In contrast, our study included lower-income and lower-middle-income countries from SSA, which have a greater burden of non-consensual sexual practice than women do.⁷ Additionally, differences in the study years may have contributed, as the previous study used data collected from 2004 to 2011 and was conducted in 2015, whereas our study used data collected in 2020.

The prevalence of forced sexual initiation in our study was lower than that reported in other studies.¹ The possible reason could be that their study included only girls who had initiated sex, whereas our study included all girls, regardless of sex status.

The prevalence of forced sexual initiation was greater in girls residing in urban areas than in those living in rural areas. This difference may be because girls in cities are more aware of their sexual rights and have better access to youth support services. In contrast, rural girls may hesitate to report sexual coercion due to feelings of embarrassment and fear of stigmatisation.⁴⁸ Additionally, the wider socioeconomic gap between the rich and the poor, along with higher rates of illicit drug use among both girls and offenders in urban areas, could contribute to this variation.^49–51 Additionally, the migration of adolescents from rural to urban areas may increase their vulnerability to sexual violence due to a lack of protection and support.⁵²

Girls who made their sexual debut before the ages of 15 and 18 were more likely to experience forced sexual initiation than those who made their sexual debut after the age of 18. This pattern is consistent with previous research indicating that early adolescence, particularly before age 17, is a period of heightened vulnerability to sexual violence.⁴⁷ It may also reflect the fact that girls under 15 or 18 are still legally and developmentally considered children, and therefore often lack the capacity, power or support to resist coercion. In some cases, perpetrators may also include individuals within trusted social networks, such as peers, family members or acquaintances.⁵³ Furthermore, younger adolescents are less likely to have comprehensive sexual education, more limited autonomy in sexual decision-making, and greater susceptibility to exploitation and abuse. Financial and social dependency on others may further increase their vulnerability.^{54 55} In addition, this observed association may partly reflect a compositional effect, whereby fewer girls engage in consensual sexual activity at younger ages. As a result, sexual debut occurring during early adolescence is more likely to involve coercion, which may increase the proportion of forced sexual initiation in this age group.

Compared with those with higher levels of education, girls who had only primary education were more likely to experience forced sexual initiation. This finding is supported by previous findings,⁵⁶ which suggest that higher educational attainment reduces the risk of sexual violence. This protective effect may be explained by the fact that girls with higher education are more likely to receive comprehensive sexual and reproductive health education and to be aware of their sexual and reproductive rights.⁵⁷ Furthermore, higher education may empower women by improving self-esteem, enhancing knowledge and skills, and increasing their employment opportunities, thereby promoting gender equality, strengthening socioeconomic status, and improving decision-making autonomy,^{58 59} all of which may help them to prevent sexual coercion.⁶⁰ This association may also partly reflect reporting differences, as more educated adolescents may have greater awareness and understanding of sexual coercion, making them more likely to recognise and report such experiences.

Compared with those from first-degree families, girls who worked as house help had a greater likelihood of experiencing forced sexual initiation. Previous studies have also highlighted that house workers are at greater risk of forced sexual initiation⁶¹ and sexual coercion.⁶² One reason is that female domestic workers usually live independently and without parents’ protection and care and thus become more vulnerable to sexual coercion. They also have less access to legal protection, thereby becoming the best targets for perpetrators, who take advantage of their vulnerability and need to avoid punishment they would have faced were the girls still at home. A third explanation is economic hardship; the lack of basic necessities such as food and shelter may compel these girls to experience sexual coercion even from employers, other male household members, or brokers who exploit their positions of power.⁶²

In terms of wealth status, girls from middle-wealth households were more likely to experience forced sexual initiation than were those from lower-wealth households. This finding contradicts other studies, which indicate that lower wealth status is associated with a greater risk of sexual coercion.⁶³ A possible explanation for the increased risk among girls from middle-wealth households could be that reduced concern about basic necessities allows for greater engagement in relationships. Additionally, their increased access to social environments such as schools and recreational events, along with a greater risk of substance use, may expose them to more peer pressure and increase their involvement in sexual activity, which may lead to sexual coercion.^64–66 This association may also be partly explained by socioeconomic differences, as adolescents from higher socioeconomic backgrounds may have greater awareness and understanding of sexual coercion, making them more likely to recognise and report such experiences.

Compared with currently married peers, divorced girls were more likely to experience forced sexual initiation. This could be explained by broader structural and relational factors. In some settings within SSA, where marriage by abduction or forced marriage has been documented,⁶⁷ the perpetrator of sexual violence may become the spouse; thus, forced sexual initiation can occur within the context of marriage itself. Furthermore, forced sexual initiation has been linked to subsequent exposure to abusive or violent marital relationships,⁶⁸ including ongoing intimate partner violence and revictimisation, even when the spouse was not the initial perpetrator.^{69 70} As a result, these sustained experiences of violence may contribute to marital instability and increase the likelihood of separation or divorce.

Adolescent girls from Burkina Faso were more likely to experience forced sexual initiation compared with those from Kenya. The possible explanation could be political instability and a lack of civil administration, as experienced by some Sahel countries such as Burkina Faso, which had a political stability index estimate of −2.0.⁷¹ Political unrest and armed conflict lead to poverty, displacement and disruptions in education, all of which increase a girl’s susceptibility to violence, including forced sexual initiation.⁷² Similarly, adolescent girls from Côte d’Ivoire were more likely to experience forced sexual initiation compared with girls from Kenya. The possible explanation is the greater prevalence of early marriage and early sexual debut among adolescents in Western Africa countries,⁷³ which contributes to an increased risk of forced sexual initiation.⁷⁴ Adolescent girls from Uganda were more likely to experience forced sexual initiation compared with girls from Kenya. Previous studies have also highlighted that the national estimate of sexual violence among women in Uganda is higher than in Kenya.^{75 76}

The strengths of the study include the use of nationally representative multicountry data to estimate forced sexual initiation, providing evidence in an understudied area. This study used an appropriate statistical approach, a multilevel model, to accommodate the hierarchical nature of the data. We included adolescent girls in our study, which reduces recall bias related to forced sexual initiation compared with studies involving older women. The limitations of the study include the cross-sectional nature of the data, which may limit the ability to establish temporal relationships. Other limitations include the absence of some explanatory variables, such as the perpetrator’s age, the relationship with the perpetrator, and the use of addictive substances due to the use of secondary data. In addition, the IPW sensitivity analysis showed that associations for educational status and wealth tertile were insignificant, suggesting caution in interpreting these findings.

Conclusions

The prevalence of forced sexual initiation among adolescent girls in four SSA countries was greater than the global target, indicating the need for intervention in this region to achieve the global SDGs by 2030. Age at sexual debut, relationship with the household head, educational status, wealth status, marital status, and country of residence were statistically significant factors associated with forced sexual initiation among adolescent girls.

Data availability statement

Data are available in a public, open access repository.