A systematic review of sexual and reproductive health interventions for young people in humanitarian and lower-and-middle-income country settings

SRH search terms were selected based on the standard definition from the International Conference on Population Development (1994) and the World Health Organization’s (WHO) SRH strategies and guidance (2010). Search terms for SRH were generated by the authors in consultation with a WHO librarian and encompassed the following: general sexual and reproductive health, pregnancy, family planning, contraception, abortion, prenatal healthcare, antenatal health care, HIV/AIDS, STIs, prevention of mother-to-child transmission (PMTCT), maternal and newborn health, gender-based violence, and adolescent sexual health. We included studies conducted in the Middle East, low-income, lower-middle income, and middle-income countries as defined by the current World Bank country classifications using individual country names in the search. In addition to general humanitarian-crisis related terms, the WHO designations for current humanitarian emergencies (at levels 1, 2, and 3) were also used to guide identification of humanitarian settings. These included any type of humanitarian crises whether, man-made or natural disasters. We focused on LMICs because the resources available to address humanitarian crises in these settings are much different than those available in high-income countries, and a significant proportion of displaced persons seeking refuge from humanitarian emergencies are hosted by developing countries [2].

Our search strategy used the following format: (SRH related terms) and (intervention or education related terms) and (country/setting related terms). We restricted the search to randomized controlled trials, adolescent and young adult (ages 13–29) populations, and publications in the English language. We included studies only in peer-reviewed journals published between 2000 and 2018 and searched the following four databases: Pubmed, Psycinfo, Medline, and Embase. We then conducted a second search of upper-middle income countries using the same search strategy and hand-selected studies for title/abstract screening that were randomized controlled trials of SRH interventions among adolescents and/or young adults ages 13–29. We conducted this additional search in upper-middle income countries in order to more thoroughly review the literature on effective SRH interventions for young people. We also searched the grey literature by targeting international organizations involved in humanitarian work and reviewing any listed publications that included our SRH search terms. Finally, reference lists of prior systematic reviews on SRH domains in humanitarian settings were reviewed for potentially relevant studies. These multiple complementary search strategies helped ensure that we were as exhaustive as possible in our search strategy.

For studies in LMICs, we included only randomized controlled trials among adolescents and/or young adults ages 13–29. For Middle Eastern countries and humanitarian settings, we loosened our inclusion/exclusion criteria such that pilot randomized controlled trials and quasi-experimental designs among participants ages 10–49 (those of reproductive age) were considered for inclusion because we anticipated that there might be few RCTs on SRH interventions among adolescents and young adults in these settings. We also included studies that focused only on mental health outcomes if they met all of our search criteria because identifying effective interventions for improving mental health in these settings could help identify effective approaches for addressing mental health concerns among those who also have unmet SHR needs as well as inform the development of psychosocial components of SRH intervention packages in these contexts. Relevant interventions that targeted our specified outcomes included: educational, psychosocial, prevention, community-based, psychoeducational, empowerment, mental health, psychological, counseling, family-based, and training programs.

Exclusion criteria were as follows: studies conducted in high-income countries, secondary analysis of primary data, and evaluations of non-relevant interventions (medical, pharmacological, dietary, exercise or cash-transfer interventions).

All citations returned from the search databases and were downloaded and entered into a Microsoft excel database for screening of titles and abstracts based on our inclusion criteria. Data was then extracted from those studies selected for inclusion into another Microsoft excel database and subject to a secondary full text screening.

Extracted data included: first author, year of publication, title, trial duration, participant characteristics (age, gender), setting characteristics, country in which study took place, study design, type of intervention, intervention components, session topics (if available), recruitment incentives, follow-up incentives, intervention length, session format, follow-up duration, control condition, sexual and reproductive health outcomes, mental health outcomes and key findings.

For primary screening of titles and abstracts and secondary full text screening, two authors independently conducted screening and then cross-checked for consistency. Discrepancies were resolved via discussion between the two authors. If no agreement could be made, a designated third reviewer was consulted (see Fig. 1).

We used principles derived from a narrative approach for data synthesis and conducted a thematic analysis of included studies due to the diverse range of study outcomes and intervention approaches. The narrative synthesis approach is an iterative process that involves developing a theory about how and why the intervention works and developing a synthesis of findings for included studies [24]. Findings were analyzed by SRH outcomes, intervention components (SRH and psychosocial components), intervention delivery information, and key findings (see Table 1). Themes for SRH intervention components were developed iteratively after manually coding the papers and extracting relevant data. Coding for SRH intervention components was conducted by two coders and cross-checked for consistency. Psychosocial intervention components were coded using the Practice Wise Clinical Coding system [80], which is an empirically derived and tested system for identifying common practice elements across different interventions. There were four discrepancies in coding of psychosocial components. Discrepancies in coding were resolved by discussion until consensus was obtained. Intervention components that were not captured by the existing codes were recorded as free text and reviewed to generate additional practice codes, which were cross-checked by the second coder. We identified seven additional intervention practice codes: (a) emotion regulation, (b) resilience building, (c) coping skills, (d) creative expression, (e) self-efficacy, (f) behavioral activation, and (g) interpersonal skills.

For those studies that reported significant effects of the intervention on SRH or mental health outcomes, summary charts illustrating common SRH and psychosocial components across studies were generated (see Figs. 2 and 3). Interventions were classified as effective by a significant between group intervention effect (p < .05) in which the intervention group demonstrated significant improvements in one or more SRH outcomes at the final follow-up assessment in comparison to the control group or to another intervention group. For studies reporting moderator effects where significant effects on given SRH outcomes were reported only for specific sub-groups, these effects are described in Table 1. In cases where moderation was reported, those SRH interventions that reported overall treatment effects or significant effects by gender were considered effective interventions; other moderation effects were not sufficient for an intervention to be considered effective [81]. In our presentation of findings, we focused primarily on those studies demonstrating effectiveness for SRH outcomes and not those that focused exclusively on mental health intervention components and outcomes because a systematic review focusing exclusively on mental health and psychosocial interventions for improving SRH among youth in LMIC and humanitarian settings was conducted previously (Turner et al., 2018; Prospero #: CRD42018081410).

In order to further extract data on common intervention components for those interventions designated as effective, written requests were sent to the authors of papers describing effective SRH interventions to obtain intervention manuals and protocols. Follow-up requests were sent to authors who did not respond (n = 7). An online literature search for intervention manual materials was also conducted when authors failed to respond to follow-up requests. Common SRH and psychosocial components were identified through extracting data from intervention manuals, when available, or through the description of intervention components provided in the text of the journal article in cases where treatment manuals were not available (n = 8). Two independent raters reviewed intervention manuals and journal articles to extract core psychosocial intervention components. In cases where there was discrepancy (n = 2), a third rater reviewed intervention material and consensus was arrived at through discussion.

Methodological reporting quality of included studies was assessed using the Effective Public Health Practice Project (EPPHP) Quality Assessment Tool for Quantitative Studies [82]. Several modifications similar to those described by Brown et al., 2016 were made to better characterize aspects of methodological quality in light of the limited information provided in some studies. Two raters assessed methodological questions independently, and discrepancies were resolved through discussion.

We identified eight studies that reported improvement in outcomes related to effective contraception or condom use, intentions to use contraception or condoms, or attitudes about contraception or condom use. All studies were randomized controlled trials except for the Mercy Corps (2015) report [31], which used a “post-test with control” design. Villaruel et al. (2010) evaluated the effects of a school-based sexual risk reduction intervention, “the Cuidate Program”, on youth in Mexico and found that those who received the intervention were significantly more likely to use a condom (OR, 1.75; 95% CI, 1.14–2.69; P < 0.05) or other contraception (OR, 1.53; 95% CI, 1.00–2.33; P < 0.05) during their first sexual encounter than those in the control group at 48-month follow-up [62]. However no effect was found for consistent condom use over time. Kaljee et al. (2005) examined the effects of an HIV/STI prevention intervention delivered in schools among Vietnamese youth and found that those who received the intervention reported significantly greater intentions to use condoms (p < .05) and perceived efficacy in condom use skills (p < .05) compared with those in the control condition at post-intervention and 6-month follow-up [63]. The authors were unable to measure condom use behavior due to low reported sexual intercourse. In a study of integrated family planning counseling services into HIV clinics among young women in Kenya, those who received integrated services reported significant increases in use of effective contraception (from 31.7 to 44.2% of encounters) at 24-month follow-up compared with controls and decreases in pregnancy rates at 48-month follow-up compared with controls at year one (rate ratio: 0.72; 95% CI 0.60–0.87) [48]. Atwood et al. (2012), examined an adapted version of “Making Proud Choices”, an HIV/STI and pregnancy prevention curriculum, delivered in schools for 6th grade youth in Liberia and found that youth who received the intervention reported significant improvements in positive attitudes towards condoms at 3-month and 6-month follow-up (β = .12, p < .05; β = .08, p < .05) [59]. Self-reported condom use self-efficacy was also significantly higher among those who received the intervention at 3-months and 9-months follow-up (β = .08, p < .05; β = .07, p < .10) compared with those in the control condition. Among youth who were sexually active, there was a significant effect on frequency of consistent condom use among those who received the intervention compared with controls at 9-month follow-up (β = .34, p < .05). Taylor et al. (2014) investigated a pregnancy prevention intervention delivered in schools for adolescents in South Africa and found that those who received the intervention reported significantly higher condom use than those in the control (54.2% vs. 36.7%, p < .01) [68].

Mercy Corps (2015) evaluated a Safe Spaces program compared with Safe Spaces in combination with a livelihood education program and a control condition delivered in schools and homes for girls (ages 10–18) in Niger [31]. The Safe Spaces program incorporates education on life skills, risk associated with early marriage and pregnancy, and reproductive health knowledge, whereas livelihood education focuses on savings and loan information, livestock management, and gardening. The authors reported significant improvements in attitudes towards contraception use in both those who received Safe Spaces (p < .01) and those who received Safe Spaces plus Livelihood (p < .05) compared with controls. Matthews et al. (2016) evaluated a school-based HIV and intimate partner violence (IPV) prevention program among youth in South Africa and found that youth who received the intervention reported significantly greater condom knowledge than youth in the control condition at 12-month follow-up (p < .001) [69]. The program involved an educational program during school delivered by study research staff, a youth friendly health service in school, and a peer-outreach component to spread information on violence prevention in school. Raj et al., (2016) evaluated a couples-based gender equity and family planning intervention among young married couples in India and found that women in the intervention group were significantly more likely to use effective contraception at 18-month follow-up (AOR = 1.57–1.58, p = 0.05) [66]. In a couples-based HIV risk-reduction and PMTCT intervention delivered to young couples in ANC clinics in South Africa, participants in the intervention group had greater odds of increased condom use at post-intervention compared with those in the control group (OR = 5.1, 95% CI (OR) = (2.0, 13.3) [70].

We identified six studies that reported significant improvements in HIV/STI knowledge, reductions in self-reported STI symptoms or increases in STI service usage. In a study investigating a multi-pronged HIV prevention intervention that included a youth, parent, and community component in Zimbabwe, youth (ages 18–22) who received the intervention reported significantly higher STI knowledge compared (AOR = 1.32; 95% CI: 1.08–1.61) and pregnancy prevention (AOR = 1.59; 95% CI:1.27–1.99), but no differences were found on HIV knowledge [25]. Aninyana et al. (2015) evaluated an SRH education intervention that involved a school-based curriculum for youth, peer-outreach activities, and youth friendly services provider training and found that youth who received the intervention demonstrated higher odds of STI service usage (AOR 2.47; 95%CI: 1.78–3.42), but no significant differences on HIV testing service usage [74]. Another study that investigated an intervention involving school-based educational sessions delivered by teachers, community activities (i.e., community-based condom promotion by youth), and provision of youth friendly services found that youth who received the intervention reported significantly greater HIV transmission knowledge (male rate ratio: 1.44 (1.25, 1.67); female rate ratio: 1.41 (1.14, 1.75)), and greater STI transmission knowledge (male rate ratio: 1.28 (1.07, 0.54); female rate ratio: 1.41 (1.06, 1.88)) [54]. Okonofua et al. (2003) evaluated an intervention to improve STI treatment seeking that incorporated education on STI prevention and referrals to STI services in a school-based curriculum as well as series of community educational lectures and found that youth who received the intervention reported significantly increases in STI knowledge (mean increase in STI’s named was .63 compared with the control group, 95% CI: 0.39–0.86) and STI treatment-seeking (from 17.5 to 40.7%; OR = 3.24, 95% CI: 1.84–5.74) [30]. The authors also reported significantly reduced STI prevalence (OR = 0.68,95% CI = 0.48–0.95) compared with youth in control schools at post-intervention. In the school-based HIV and IPV prevention program reported by Matthews et al. (2016), youth who received the intervention reported significantly greater HIV knowledge than those in the control arm at 12-month follow-up (p < .01) [69]. Finally, the HIV/STI prevention intervention for youth in Vietnam also reported significantly greater HIV/STI knowledge among youth who received the intervention (p < .05) compared with youth in the control condition at post-intervention and 6-month follow-up [63]. Jones et al. (2013) also found increased HIV knowledge at post-intervention among young couples completing the HIV-risk reduction intervention (F(1, 476) = 13.9, p < .001) [70]. All studies were randomized controlled trials.

Four studies reported significant effects for sexual self-efficacy, including condom negotiation self-efficacy, self-efficacy for safe sex, and sexual refusal self-efficacy. In the study of an HIV prevention intervention among youth in Zimbabwe mentioned above, Cowan et al. (2010) found that female youth who participated in the intervention reported significant increases in condom negotiation self-efficacy compared with those in the control condition (AOR:1.17; 95% CI: 0.95–1.43), but no significant differences were found for male youth [25]. Similarly, Baioochi et al. (2017) investigated a school-based program to increase girls’ empowerment and boys’ education about gender equity among youth (ages 10–16) in Kenya, and those who received the intervention reported significant increases in general self-efficacy, described as perceived ability to cope with stress and manage difficulties (mean score increase = .19, 95% CI: 0.08–0.39) [46]. Finally, in a study on a church-based HIV prevention intervention for youth and families in Kenya, youth who received the intervention reported increases in self-efficacy for safe sex compared with controls at 1-month follow-up (b = .41, p < .01), but effects were not sustained at 3-month follow-up [47]. In the study on the HIV prevention intervention among youth in Vietnam, youth who received the intervention reported increases in condom use self-efficacy and condom negotiation self-efficacy compared with those in the control condition (p < .001) at post-intervention and 6-months follow-up. All studies were randomized controlled trials [63].

We identified three studies that found significant intervention effects for improving accurate SRH knowledge. Leventhal et al. (2016) examined a psychosocial curriculum plus a health curriculum compared with the psychosocial curriculum alone and a control condition delivered in schools among girls in India (average age = 13) [64]. The authors found that those who received both the health and psychosocial curriculum combined and the health curriculum alone reported significantly greater reproductive health knowledge (i.e., menstrual hygiene) than girls in the control condition (p < .05) at post-intervention. The study by Ross et al. (2007) among youth in Tanzania reported that those who received the SRH intervention reported significant increases in pregnancy prevention knowledge (male rate ratio: 1.66; 95% CI: 1.55–1.78; female rate ratio: 1.58; 95% CI: 1.26–1.99) compared with those in the control condition [54]. In the Mercy Corps (2015) study evaluating Safe Spaces alone and Safe Spaces plus a livelihood-training component among girls in Niger, both groups demonstrated significantly greater SRH knowledge compared with controls [31]. Girls in Safe Spaces were 69.7% more likely to know one benefit of delaying pregnancy to 18 or older (p < .01), whereas girls in the livelihood program were 27.7% more likely (p < .05).

Five studies were identified that reported improvements in gender-based violence. Matthews et al. (2016) found that youth who received the HIV and intimate partner violence (IPV) prevention intervention were significantly less likely to report experiencing sexual violence than those in the control arm at 12-month follow-up (35.1 vs. 40.9%; OR 0.77, 95% CI 0.61–0.99; t [30] = 2.14) [69]. Devries et al. (2017) investigated the Good Schools Toolkit, which includes education on sexual and emotional violence and relationship power, among youth in Uganda (ages 11–14) [58]. The authors found that those who received the intervention reported significant reductions in levels of violence (male student AOR = 0.34, 95% CI: 0.22–0.53; female students AOR = 0.55, 95% CI: 0.36–0.84). In the school-based curriculum developed by Baiocchi et al., (2017), results also showed that following the girls’ empowerment and boys’ gender equality education intervention components, there was an estimated 3.7% decrease (p = 0.03; 95% CI: 0.4–8.0), risk of sexual assault in the intervention group [25]. Raj et al. (2016) found that of those participating in the gender equity intervention, women were less likely to report IPV (AOR = 0.48, p = 0.01), and men were less likely to report acceptance of IPV in comparison to those in the control condition at 18-month follow-up (AOR = 0.51, p = 0.004) [66]. In the couples-based HIV-risk reduction intervention in South Africa, there was a significant reduction in reported acts of violence and verbal aggression among those in the intervention group (McNemar’s test, p = 0.001 for intervention, p = 0.49 for control; McNemar’s test, p = 0.01 for intervention, p = 0.10 for control, respectively) [70].

The SRH components of intervention packages demonstrating significant effects on SRH outcomes focused primarily on education/knowledge building and skills building. The most frequent educational components were: SRH education (i.e., puberty, reproductive anatomy, pregnancy; 7/17), HIV/STI prevention education (7/17), gender equity education (6/17), and education about effective contraceptive methods (including condom use) (6/17). Education on gender-based violence was the next most common component (4/17). Three studies also incorporated parent education sessions on youth SRH, and three included education on healthy relationships. For skill-building, sexual decision making skills components were incorporated in three studies, and building condom use skills through demonstration and practice was incorporated into those interventions focused on improving knowledge of effective contraceptive methods. Three studies incorporated youth friendly services as an intervention component to increase youth access to SRH services. For studies reporting pregnancy prevention as an SRH intervention component, we coded pregnancy prevention as SRH education.

Of the 17 studies that reported significant intervention effects on SRH outcomes, approximately 50% (9/17) incorporated psychosocial components into the intervention package. The most common psychosocial components included were assertiveness training (4/17) communication skills (4/17), and problem solving (4/17), (see Fig. 3). Other common psychosocial intervention components included emotion regulation, goal setting, resilience-building, cognitive, modeling, coping skills, and psychoeducation-staff.

The majority of the effective SRH interventions were delivered in school settings (13/17). Three of these interventions were delivered in multiple settings that included schools and either community centers or clinics. Other intervention delivery settings were clinics (3/17), churches (1/17) or village community settings (1/17). The most frequent delivery format for interventions was group-based (15/17), with two group-based interventions delivered to female youth only. Four interventions involved an individual counseling component, four involved parent education and discussion groups, and four involved a community component (i.e., community outreach activities, public lectures, etc.). Multi-component intervention approaches that included youth groups and either a parent group or community component (or both) were also reported in 8/17 studies. Three studies incorporated peer outreach training. Two studies involved couples-based interventions that included male partner involvement. Four studies incorporated service provider training for SRH services, and three of these implemented youth friendly health services as an intervention component. Only four studies described both facilitator training and supervision procedures; five studies provided some description of facilitator training procedures but failed to report on supervision. Interventions varied greatly in terms of session frequency, length, and overall duration of the intervention (see Table 1).

Results of the quality assessment rating in accordance with the EPHPP criteria are displayed in Fig. 4. Studies reported strong participant study completion rates, with 11 studies indicating completion rates of 80% or higher, and only two studies failed to report participant withdrawal and dropout information. There were also several methodological weaknesses of these studies. A major methodological weakness was the failure to blind both participants and assessors or the lack of reporting about whether blinding occurred. Only one study reported that assessors were blinded, and no studies reported that participants were blinded to study condition. Another weakness was reporting of whether contamination likely occurred; only one study reported that participants in the comparison condition likely had some exposure to the intervention program. Many studies also failed to report information on the reliability or validity of assessment tools, with only four studies including information on validity and eight on reliability. Finally, only six studies described processes through which fidelity to the intervention was assessed or monitored, and no studies reported results of fidelity assessments.

We restricted our search to those studies published in the English language, which may have limited the number of articles retrieved from our search. Qualitative studies were excluded from the search, which can often provide rich information regarding feasibility and acceptability of interventions as well as guide selection of culturally appropriate incentives to improve participation and retention during intervention implementation. We limited our search to randomized controlled trials (with the exception that quasi-experimental studies conducted in humanitarian settings or the Middle East were included) in order to gather information on those SRH interventions that were evaluated through rigorous scientific methods, which may have limited the studies retrieved in low resource settings that lack funding and expertise for more costly evaluations of interventions. Because we restricted our participant age criteria to exclude studies that included participants ages 10 and under, we may have excluded some effective SRH interventions for adolescent populations if the study included younger participants. It should be noted that our criteria for including an intervention as effective were that the intervention demonstrated significant improvements in one or more of the reported SRH outcomes over time. There were several instances in which an intervention showed significant effects on one SRH outcome, while failing to impact several others (or impacting outcomes in unexpected directions for particular subgroups). We retained these studies because they still met our inclusion criteria, but the inconsistency in improvement across SRH outcomes raises some questions as to thus overall effectiveness of those interventions. Inclusion of some studies conducted in upper-middle income countries could be less relevant for populations in the Middle East and humanitarian settings due to cultural, social, economic, and political differences among these contexts; thus their findings may not necessarily generalize to humanitarian contexts. Heterogeneity of measurement tools used to assess SRH outcomes across studies and lack of psychometric information on outcome tools obfuscated comparison of effects across studies.