Discussion
The COVID-19 school closures and containment measures notably impacted the daily life of vulnerable adolescent schoolgirls in western Kenya. This study follows girls from the start of their third year of secondary school through to the conclusion of their final school examinations to measure COVID-19’s impact on their sexual health and schooling. Girls, whose schooling was disrupted due to COVID-19, experienced a threefold risk of dropping out of school and 3.4 times the risk of changing schools relative to their peers. Girls who experienced school closures also had twice the risk of falling pregnant prior to completing school, with 1 in 10 COVID-19 cohort girls becoming pregnant prior to sitting their examinations. This group also saw increased sexual activity, with nearly one in two COVID-19 cohort girls becoming sexually active during follow-up. These girls were twice as likely to report their first sex was not desired when compared with girls who did not experience COVID-19 during secondary school. Girls in the COVID-19 cohort also reported increased work hours and four out of every five girls in this group reported their household income decreased during the pandemic.
Our results show that adolescent girls in this study area, which typifies many rural areas of sub-Saharan Africa, are particularly vulnerable to sexual and reproductive harms during emergencies. These findings are in line with other studies that have shown heightened rates of adolescent pregnancies after population-level crises. For example, during the west-African Ebola outbreak, in Sierra Leone teenage pregnancies increased by 25% and were hypothesised to represent underlying vulnerabilities to transactional sex and sexual exploitation.15 While reported transactional sex was too low to detect in our study, our findings showed that girls experiencing COVID-19 restrictions were twice as likely to report that their first sex was undesired, indicating a possible rise in sexual coercion. Studies have hypothesised that, during emergencies, increased economic strain may push girls to engage in sex in exchange for money or favours.30 Wider studies have also shown that girls’ facing sexual coercion or abuse obtain lower schooling and lifetime earnings,31 with adolescent pregnancy itself being a major driver of school dropout.11 These outcomes underscore the longevity of effect resulting from harms experienced during adolescence.
Our study highlighted the detrimental and potentially long-term effects COVID-19 containment measures had on girls schooling. Girls enrolled in secondary school during COVID-19 were 3.4 times more likely to change schools relative to their non-COVID-19 peers. In addition to being disruptive to education, among adolescents, domestic migration has been associated with unstable living arrangements and early engagement in high-risk sexual activity and informal work.32 Girls in the COVID-19 cohort were also three times as likely to drop out of school. During Ebola many countries employed school closures as a policy response to contain the outbreak; however, long after schools reopened, school enrolment of young girls remained significantly lower than pre-Ebola.33 34 These impacts disproportionately affect lower income and more marginalised students and particularly older adolescents who take on additional paid or unpaid work during crises.35 36 Studies have shown that once engaged in economic activities, older adolescents may have to retain these responsibilities even after schools reopen to help sustain their families.30 35 36 Our results echo these findings, showing how girls engaged in non-school-related work had increased work hours in the aftermath of the COVID-19 pandemic.
Schools act as a social vaccine in preventing adolescent SRH harms, including pregnancy.11–13 In late secondary school, many students transition into full-time boarders in Kenya; thus, the closure of schools may have affected girls’ living environment in addition to their learning environment. This was not unique to Kenya: school closures affected an estimated 80% of young learners globally, most acutely affecting students in low-income countries that had the fewest resources to support remote learning. However, in higher-income countries, where much of schooling was moved online and learning was expected to continue, inequalities were also found to widen, leaving behind students with limited access to internet, no suitable place to do homework and those from households with unstable housing. In the Netherlands, for example, one study found that students from disadvantaged households experienced 60% more learning loss than those in the general population;37 another study in Canada is projecting that the student skills gap due to socioeconomic factors in adolescents will widen by more than 30%.38 In the USA, where school closures were not uniform across states, one study found that school closures were higher among lower-income schools with higher numbers of minority students and higher numbers of students who were eligible for free lunches.38 Evidence is growing that food instability among youth is rising due to school closures. In India, for example, where over half of all children are undernourished, the Mid-day Meal, a school-based nutrition programme provided minimum calorie and protein requirements to 80% of primary school students nationwide prior to school closures.39 Moreover, schools serve as the primary link between adolescents and child protective services. Concern is rising that school closures are leaving adolescents to suffer without support at a time when mental health outcomes have deteriorated in adolescents.40 A recent study in the USA identified that child maltreatment reports dropped during COVID-19, possibly highlighting a break in the referral chains for vulnerable children and youth.41 Our findings add to this mounting body of evidence, which warns of rising inequalities due to COVID-19-related school closures, and emphasise the multifaceted role schools play in adolescent health.
In Kenya, increased domestic abuse within households has been reported, partly due to pandemic-related stress and income loss.42 In our study, girls voiced decreasing household incomes, feeling less safe at home and increasing crime in their communities in the aftermath of the pandemic. Girls also reported heightened pandemic-related stress. This is in line with what has been documented worldwide: studies have evidenced higher levels of anxiety among females, younger individuals, those of lower incomes and those who experienced loss of income.43–46
Some limitations of this study should be considered. First, girls in the COVID-19 cohort were followed up for an additional four months due to their delayed KCSE examination. This unequal follow-up may have given them additional time to fall pregnant relative to their pre-COVID-19 peers. In response we employed person–time analysis to adjust for this unequal follow-up. To note, because this study was completed immediately after the COVID-19 cohort girls sat their final examinations, girls who were in the early months of their pregnancy and sat their examination may have been missed. Consequently, the estimated pregnancy prevalence in the COVID-19 cohort may be an underestimate of all girls who became pregnant while still enrolled in secondary school. Because the trial was still active in the study area for a year after pre-COVID-19 girls completed their school, all pregnancy outcomes occurring during school were documented in full. Second, behavioural data were self-reported and certain sensitive indicators, such as sexual activity and sexual behaviours, may be under-reported. We also resensitised girls at every survey round on the importance of truthful reporting which may have improved response bias but led to differential reporting between survey rounds. In response, we took a cumulative measure of girls reporting sexual activity at any survey. Third, this study only measured effects in a secondary school-going population and may not reflect vulnerability to harms of other populations such as out-of-school girls. Lastly, this study was embedded within a cRCT that was testing different interventions to improve girls schooling and SRH harms, with girls in the COVID-19 cohort experiencing interventions for an additional year relative to their pre-COVID-19 comparators. We note that any beneficial impact from being in the trial longer would push the measure of effect in the COVID-19 cohort towards the null, suggesting the results presented here may be conservative estimates of the pandemic’s effects on girls’ health and schooling relative to non-trial schoolgirl populations.