Gender,Power,and Health: Measuring and Assessing Sexual Relationship Power Equity Among Young Sub-Saharan African Women and Men,a Systematic Review

Selection Criteria

Our review follows Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guidelines for systematic reviews (Liberati et al., 2009). In order to meet the objectives of this review, studies were included (1) if they were conducted within sub-Saharan Africa, (2) if the study was specific to youth and young people (eligible age ranged from 10 to 30), (3) if they were related to SRP, and (4) if they included a quantitative SRP equity measure or a measure of relationship control. The age range of 10–30 was used in order to capture individuals that met the United Nations, WHO, and Commonwealth definitions of youth (The Commonwealth, 2018; United Nations Population Fund, 2014; WHO, 2014).

Search Methods

University library electronic searches were conducted in March 2019 using the following databases: Pubmed, Ovid Med, Psych info, and Web of Science for manuscripts meeting the selection criteria. We further reviewed the gray literature using Google Scholar, reviewed the reference lists of manuscripts in the topic area, consulted with topic experts, and reviewed databases of a topic-specific forum provided by the Sexual Violence Research Initiative (SVRI; 2019). The search was updated to include any additional papers in November 2019. The search was open-ended; however, since the original SRPS was developed in 2000 (Pulerwitz et al., 2000), publication dates for papers ranged from 2004 to 2019.

SRP search strategies conceptualized SRP by including search strategies including power [psychological], controlling behavior, and relationship control. Examples of search strategies used within this review can be found in Supplementary Material S1.

Study Selection

All titles and abstracts found through the search strategy were inputted into EndNote (2017), and any duplicated titles or abstracts were removed. Abstracts were reviewed by KC and JN independently for eligibility, if there was a discrepancy, the full text article was reviewed, and additional co-authors were consulted on eligibility. Following abstract selection, the same authors independently reviewed full-text articles detailing any further reasons for exclusion. Abstracts and full-text articles not meeting the eligibility were excluded, and the reason for exclusion was noted. Studies were excluded if they were qualitative, they did not include a measure of SRP, were not focused specifically on youth, or were conducted outside of sub-Saharan Africa.

Data Extraction

Detailed notes were made by KC and JN regarding each manuscript including author, year of publication, country, the aim of each study, study design, analysis, details on the SRP equity measure, and relevant findings in relation to the SRP equity measure. Where available, notes were made on the underlying theoretical framework, any validation process for the SRP equity scale measure, and the original source of the scale. After notes were made, we summarized study characteristics, definitions, and descriptions of the SRP equity measures, scale scoring, justifications for changes to the scale, demographic differences in SRP equity, and associations between measures of SRP equity and HIV risk factors among young men and women reported in the reviewed manuscripts. KC and JN further independently used 15 questions to rate the quality of each of the studies based on an adapted Critical Appraisal Skills Program (CASP) quality assessment score sheet (Hatcher et al., 2015). Quality assessment responses included 0 (nonresponsive), 1 (partially responsive), or 2 (fully responsive). The quality assessment of each manuscript included in our review can be found in Supplementary Material S2. The CASP checklist was based on the 1994 JAMA user’s guides to the medical literature and refined through piloting work with health care professionals.

Study Characteristics

Two thirds of manuscripts (n = 21, 72%) used cross-sectional data, four were cohort studies utilizing multiple time periods of data, one was a mixed-method study using cross-sectional quantitative data, one manuscript assessed the results of a trial using quasi-experimental methods examining time series analyses pre- and post-intervention, and two were quasi-experimental mixed-method evaluation studies (see Supplementary Material S3, S4, and S5). Manuscripts were published between 2004 and 2019, the majority of papers were from South Africa (n = 24), with one study taking place in each of the following countries: Kenya, Ghana, Malawi, and Zimbabwe. One study covered three sub-Saharan African countries: Namibia, Ethiopia, and Tanzania. Of the 29 papers, 10 included only young women (S4), five included only young men (S5), and 11 included both (S3). Nearly half (14/29) were published either from the Stepping Stones (Jewkes et al., 2006) or Stepping Stones and Creating Futures trials (Gibbs et al., 2017).

Measures and definitions of SRP varied widely within manuscripts. Although the original SRPS had 23 items and two sub-scales, the reported number of items included in the scales in reviewed manuscripts ranged from four to 15, with a median of nine items and a mode of 13 items (of 24 papers that reported the number of scale items). None of the papers used both of the original subscales in their analyses, and few reported the full list of scale items in the published manuscript. All scales that were listed in the reviewed manuscripts can be found in Supplementary Material S6.

In Pulerwitz et al. (2000) original study, the scale is named “the Sexual Relationship Power Scale”; however, within the 29 reviewed papers, scale names were varied. The scale names included “controlling behavior” (Closson, Hatcher, et al., 2019; Gibbs, Govender, & Jewkes, 2018; Gibbs, Jewkes, et al., 2018; Hatcher et al., 2014; Naidoo et al., 2015), “relationship control” (Christofides et al., 2014a; Jewkes et al., 2014; Nduna et al., 2013), “relationship power” (Teitelman et al., 2016) “power in relationship” (Harrison et al., 2006), “Power Disparity Scale” (Bingenheimer & Stoebenau, 2016), “Relationship Power Inequity Scale” (Closson, Dietrich, et al., 2019; Jewkes et al., 2010; Zembe et al., 2015), “Gender Attitudes and Control Scale” (Shai et al., 2012; R. Jewkes et al., 2016), and “unity in power processes” (Conroy, 2014). For the purposes of comparison, the term SRP equity will be used throughout our review to refer to all measures of relationship control and power dynamics in relationships.

All papers with the exception of two, which conducted study-specific scale development (Conroy et al., 2016; Harrison et al., 2006), stated that the scale they used was modified from the original (Pulerwitz et al., 2000). Only three papers described a process for piloting and removing scale items using factor analyses (Conroy, 2014; Harrison et al., 2006; Pulerwitz et al., 2018). Of the 29 papers, five papers did not report the Cronbach’s α of the scale used. Of the remaining papers, Cronbach’s αs ranged from .54 to .90. Where authors reported the scale Cronbach’s αs by gender (n = 7), all were higher for women compared to men, with the exception of two papers (Closson, Dietrich, et al., 2019; Harrison et al., 2006).

Across items in the scale, there were a number of similarities with the adapted scales and the original scale which are included in S6. Themes across scales included unequal decision-making power (e.g., “My partner does what he wants even if I don’t want him to”), male controlling behavior (e.g., “my partner tells me who I can spend time with”), jealousy (e.g., “My partner gets jealous when I wear things that make me look too beautiful”), male partner surveillance (e.g., “My partner always wants to know where I am”), and male partner expectations (e.g., “Because my partner buys me things, he expects me to please him”). Wording of items differed across papers, depending on the aim. For example, Zembe et al. (2015) were interested in examining differences in SRP and experiences between different types of partnerships and thus specified which partner (main, casual) items were referring to (Zembe et al., 2015), while the remaining papers only asked about primary partners. All papers that included measures of controlling behaviors for men asked men to think about their own controlling behaviors, except for five which asked men about their female partner’s controlling behaviors (Belus et al., 2019; Sayles et al., 2006; Steffenson et al., 2011; Teitelman et al., 2016).

Two papers described the development of their own scales related to power dynamics in relationships. Some of the items in Harrison et al.’s (2006) and Conroy’s (2014) papers were similar to the original SRPS, including items around decision-making power (e.g., “We come to agreements together about what we want to do,” Harrison and “My partner and I discuss important matters together,” Conroy). However, there were differences in the items, including more emphasis on violence (e.g., “My partner sometimes gets angry in such a way that he hits me”) and dominant scripts around gender roles (e.g., “If a woman refuses sex, she usually doesn’t mean it”) in Harrison’s paper and an emphasis on “unity” in Conroy’s paper (e.g., “My partner shows they care about me”).

Scoring the Scale

In the original 23-item SRPS that is comprised of two subscales, Pulerwitz combined the score from the two subscales and trichotomized the score into “high,” “medium,” and “low” levels of power by dividing the scale into three equal parts (Pulerwitz et al., 2000). Pulerwitz recommends that the scale be scored by calculating the mean score for each respondent—by summing the scores for each respondent and dividing the sum by the number of nonmissing items—and then depending on the distribution of the data, researchers can create a tercile with higher scores indicating higher relationship power. Respondents with more than a third of missing items should be dropped from the analysis. Despite the original paper using and recommending the scale be used as a tercile, only 7/29 (24%) of the papers included in this review scored the SRPS into terciles, three of which combined two terciles into one group. Half of papers used a continuous score, five dichotomized the scale into high or low SRP, one study created quartiles (four), and one study did not report on details of the scoring. Only two studies (Closson, Dietrich, et al., 2019; Pulerwitz et al., 2018) used the original scoring of the scale mentioned above.

Justification for Changes to the Scale

Relationship dynamics and the ways in which control functions in a relationship varies widely across contexts. Contextual differences across populations, countries, and age groups provide justification for scale adaptation and modifications. However, despite a wide range of scale items and scale score included in the review, only nine (31%) papers reported any information regarding the justification for scale development and/or adaptation. Three of the papers came from a national sample of South African youth and indicated that they reduced the scale down to four items through piloting work in order to address difficulties in translation, lack of comprehension, and space constraint in the survey (Pettifor et al., 2004; Sayles et al., 2006; Steffenson et al., 2011).

One paper from South African data (Harrison et al., 2006) and one from Malawi (Conroy, 2014) created their own scales to measure power inequities as well as other relationship dynamics through multiple qualitative and quantitative scale development and validation processes. Both papers describe conducting qualitative in-depth interviews in the local language to elicit a pool of items that could be used to develop scales related to SRP and gender relations. Qualitative data collection and analysis were followed by quantitative validation through factor analyses and test–retest reliability.

Pulerwitz et al. (2018) describe the adaptation process of the original 15-item Relationship Control SRP subscale by removing one item due to difficulty in translation, and the addition of two items said to be important for young women in this context. A confirmatory factor analysis was then performed on a 16-item Relationship Control subscale, in which one item was dropped with an inter-item correlation <.30, resulting in a final 15-item scale.

One paper reported removing one item from the eight included in the Stepping Stones and Creating Futures midline survey due to low factor loading (<.4; Closson, Hatcher, et al., 2019) and another created two gender-specific modified scales using an exploratory factor analysis (Closson, Dietrich, et al., 2019). Finally, Teitelman et al. (2016) indicated they removed four items from the original 23-item scale in order to not conflate the relationship between SRP and the outcomes of interest, and an additional four items that the study team felt were not clear indicators of relationship control for adolescents in South Africa.

Demographic Differences and Potential Drivers of SRPS

Six of the papers included in the review used SRPS as a main outcome or exposure of interest, while the remaining manuscripts included the scale as a covariate or secondary objective in their study. In Pulerwitz and colleagues’ (2018) study among young Kenyan women, the youngest aged participants (aged 15–17) had higher SRP than young women in the older age-group (21–24 years; Pulerwitz et al., 2018). Closson and colleagues (2019) also found that younger men had higher SRP equity.

There were mixed findings around differences between women and men. Two studies reported no significant differences in SRP equity between men and women (Conroy, 2014; Harrison et al., 2006), one study found that young men had lower SRP equity than young women (Teitelman et al., 2016), and one study found women had lower SRP equity than men (Naidoo et al., 2015). Closson and colleagues found that men from Soweto (vs. Durban) had higher SRP equity and women in age-similar relationships and relationships less than 2 years in length had greater SRP equity (Closson, Dietrich, et al., 2019). One study with young men found that those with heightened stresses related to their gender roles including success, power, and competition, as well as restrictive emotionality had lower SRP equity (Closson, Hatcher, et al., 2019).

Across papers, SRP equity was tested against four main themes including (1) violence, (2) mental health, (3) sexual behavior, and (4) sexual and reproductive health outcomes (Table 1).

OPEN IN VIEWER

Table 1.

Key Findings From the 29 Manuscripts Reviewed.

SourceStudy TypeCountrySample SizeAge Range or Mean	SRPS Reference	Included Details on Adaptations or Modifications?	Included Scale Score?	Number of Items, Included List of Items in Scale (Yes, No)	Cronbach Reported?	Determinants of SRP	Theme 1	Theme 2	Theme 3	Theme 4	Intervention Evaluation
Belus et al. (2019) Cross-sectional South Africa 286 heterosexual couples Men’s mean age = 26 and women’s mean age = 24	Pulerwitz et al. (2000)	Used the eight-item Decision-Making Dominance Scale from original scale		8 range 0–8	.84 for women and .81 for men		X		X
Closson (2019)Cross-sectionalSouth Africa164 young women and 87 young men16–25 years	Pulerwitz et al. (2000, (2002)Jewkes et al., 2002	Conducted gender-specific exploratory factor analyses to produce two modified scales	Women mean = 35.2 (3.8) Men mean = 35.1 (4.6)	Full = 13Modified = 9 for men and 8 for women	Full women = .63modified women = .67full men = .64modified men = .70	X	X		X
Closson (2019)Cross-sectionalSouth Africa449 young men18–30 years	Jewkes et al. (2006)	Conducted factor analysis on eight-item scales used in Stepping Stones Creating Futures (SSCF)Higher score = more controlling behaviors	Median = 9 (Q1–Q3 = 6–11)	7, no	.83	X
Kilburn et al. (2018)Cash transfer Randomized Control Trial (RCT)South Africa2,448 young women13–20 years	Pulerwitz et al. (2000) Dunkle (2004)	NR	Median = 28 (23 to 32, range 1–36)	NR	NR						X
Pulerwitz et al. (2018)KenyaCross-sectional1,101 young women15–24 years	Pulerwitz et al. (2000)	Factor loading by age-group, with one question added		15, yes	15–17 years = .7918–20 years = .7921–24 years = .82	X	X		X
Gibbs (2018a)Cross-sectionalSouth Africa677 men and 680 women18–30 years	Jewkes et al. (2006)	No; higher score = more controlling behaviors	Mean men no IPV = 10.1 (9.6–10.5)Mean men IPV perpetration 11.5 (11.2–11.9)Mean women no IPV experience 8.4 (7.9–8.9) IPV experience 11.1 10.8–11.3	8, no	.88 women and .88 men	X	X
Gibbs (2018b)Cross-sectionalSouth Africa232 (122 women and 110 men)18–30 years	Jewkes et al. (2006)	Larger scores = less controlling behavior	Cut scores into terciles, does not provide scoring detail	8, no	Men = .67women = .71			X
Teitelman et al. (2016)Cross-sectionalSouth Africa786 (409 men and 377 women)14–18 years	Pulerwitz et al. (2000)	Used all eight items in making (DMD) Scale and three condom-related questions and one on multiple partners were removed(four additional items not listed removed)	Females 45.13 (SD = .48)Males 43.22 (SD = .45)Overall (44.15, SD = .33)	15, no	.90		X		X
Jewkes et al. (2016)Cross-sectionalSouth Africa1,370 young men15–26 years	No reference “Gender and Relationship Control Scale”	No information on development and no information on scoring	NR	13, no	.69		X
Bingenheimer & Stoebenau (2016)Cross-sectionalGhana605 young women13–19 years	Pulerwitz et al. (2000)	No, also scored with three response options very true, somewhat true, and not true at all	NR		.76					X
Zembe et al. (2015)Mixed-methodsSouth Africa259 young women16–24 years	Pulerwitz et al. (2000) and Pettifor et al. (2004)	No,Dichotomized to answering yes/no and dichotomized into high/low relationship power	NR	15 but presents 11 (See Supplementary Materials S6)	No		X		X
Stockl et al. (2015)Cross-sectional399 ever-partnered women from Ethiopia, 320 women from Namibia, 789 ever-partnered women from Tanzania15–24 years	No reference	No control (ref), medium control, and high control	NR	Unknown	No		X
Naidoo et al. (2015)Cross-sectionalSouth Africa1,612 young men and 1,409 young women18–24 years	No reference	Higher scores = lack of relationship control (only used the Relationship Control subscale)	No	4?	.81					X
Jewkes et al. (2014)Pre- and posttests and intention to treatSouth Africa232 young men and women18–30 years	Pulerwitz et al. (2000)	High = more equitable	Men baseline = 19.4Female baseline = 22.2Men postintervention = 21.7Women post intervention = 22.8	8?	No						X
Conroy (2014)Cross-sectionalMalawi3,932 men and women16–29 years	Connell (1987) Cromwell & Olsen (1975)	Yes, items were developed using qualitative interviews	NR	NR	Total = .65women = .74men = .53		X
Christofides et al. (2014a)Cohort dataSouth Africa983 young men15–26 years	Pulerwitz et al. (2000)	NoHigher scores = more equity	Mean score = 21 for men fathering an incident pregnancyMean score = 20 for men not fathering an incident pregnancy	8	.73					X
Christofides et al. (2014b)Cohort dataSouth Africa174 young women15–18 years	Pulerwitz et al. (2000)	NoSplits into tertiles and does not report the mean or median score	NR	10	.73					X
Nduna et al. (2013)Cross-sectionalSouth Africa1,415 young women and 1,368 young men aged 15–26	Pulerwitz et al. (2000)	Higher score meant more liberal power dynamics	NR	10				X
Shai et al. (2012)Cross-sectionalSouth Africa1,219 young men15–26 years	No reference	No, created tertiles	NR	13	.69				X
Steffenson et al. (2011)Cross-sectionalSouth Africa7,409 young men and women15–24 years	Pettifor et al. (2004)	No, only references Pettifor et al. (2004)	NR	4	No				X
Shai et al. (2010)Cross-sectionalSouth Africa1,204 young women15–26 years	No reference and combined SRPS with Gender Attitudes Scale	NR	Min = 13 and Max = 47 (no mean or SD reported)	13	.69				X
Jewkes (2010)CohortSouth Africa1,099 young women15–24 years	Dunkle (2004) Pulerwitz et al. (2000)	NR	No compared the lowest tercile was compared to the highest two tercile	10	.73					X
Nduna et al. (2010)Cohort dataSouth Africa1,002 young women and 976 young men15–26 years	Pulerwitz et al. (2000)	NR	No compared “more liberal power dynamics in the relationship versus less liberal”	13	Women = .68men = .54			X
Dunbar et al. (2010)Mixed-method evaluationZimbabwe50 young women16–19 years	Dunkle (2004) Pulerwitz et al. (2000)	NR	Scores divided into terciles and categorized as high, medium, and low	7	No						X
Dunkle et al. (2007)Cross-sectionalSouth Africa1,288 young men15–26 years	Dunkle (2004, 2006)Pulerwitz et al. (2000)	NR	Four score levels representing quartiles of the scale and one level of men not reporting a main girlfriend	13	.69					X
Sayles et al. (2006)Cross-sectionalSouth Africa3,890 young women and 3,519 young men15–24 years	Pulerwitz et al. (2000)Pettifor et al. (2004)	Only cites original study Pettifor et al. (2004)		4	No				X
Jewkes et al. (2006)Cross-sectionalSouth Africa1,295 young women15–26 years	Dunkle (2004) Pulerwitz et al. (2000)			10	.73					X
Harrison et al. (2006)Cross-sectionalSouth Africa101 young men and 199 young women18–24 years	Developed own scales based on literature and gender script theory	Extensive ground-up work using daily diaries, in-depth qualitative interviews, survey instrument development, factor analyses, and test–retest reliability		Power in relationships = 4Egalitarian sexual scripts = 4	Power in relationships = .76 men and .70 womenEgalitarian sexual scripts = .69 men and .68 women				X
Pettifor et al. (2004)Cross-sectionalSouth Africa4,066 young women15–24 years	Pulerwitz et al. (2000)	Pilot test was conducted with the full scale. Items were removed due to lack in translating concepts, lack of comprehension, and space constraints. Factor analysis was conducted among five items, with four remaining	Score dichotomized and scored 0–4 with score of 0–2 indicating high relationship control and 3–4 indicating low relationship control	4	.69				X	X

Note. Theme 1: violence, Theme 2: mental health, Theme 3: sexual and reproductive health behaviors, Theme 4: sexual and reproductive health outcomes. NR = not reported; SRPS = Sexual Relationship Power Scale; SRP = sexual relationship power; IPV = intimate partner violence.

Associations Between Measures of SRP Equity and HIV Risk Factors Among Young Men and Women

Quality of Papers

Supplementary Material S2 presents combined results from the independent quality assessment conducted by KC and JN. Overall, the quality of papers was fair (Voss & Rehfuess, 2013). All papers presented clear objectives and accurately defined their study population. However, some papers lacked a clear exclusion criterion for their specific analysis and many had limitations of generalizability. The majority of papers adequately presented and described the measures used within their study. However, few studies provided detailed reporting of the SRPS, the validation process, and how scale items were worded (Conroy, 2014; Harrison et al., 2006; Pulerwitz et al., 2018). Many of the studies included in the review were sub-studies of larger interventions or research programs, as such some of the descriptions of the study as a whole were minimal, referencing the original protocol or trial paper. Also, many of the papers included did not have a strong theoretical framework and were interested in more exploratory analyses examining factors related to an outcome. As such, many of these papers did not meet the quality criteria for “key confounding variables measured and adjusted for” and made determining whether exposure and outcome measures were clearly defined difficult. Also, given that most of the studies included in the review were cross-sectional, it was challenging to determine the criteria for temporality (if the exposure was measured before the outcome).

Research

We found great variation in SRP measures and limited reporting of adaptations and modifications to the original SRPS. As such, it is difficult to determine and compare levels of SRP equity across study settings. Adaptations to quantitative scales in survey research are necessary for constructs to be contextual; however, these adaptations create challenges for the comparability of findings across studies. This is particularly true when there is a lack of information regarding how and why adaptations and modifications to existing measures were done.

Results from our review indicate few consistencies across papers in the number of items used. Reliability of measures varied widely and reviewed papers presented limited detail into the adaptation and/or modification processes of such scales. Given that the original Pulerwitz et al. (2000) scale was developed and validated among a sample of 833 mainly Latina women in the United States, there is a need for researchers to describe how they have adapted the SRPS to ensure that the items asked are accurately measuring the construct of SRP equity for young men and women in sub-Saharan African contexts. Also, detailed descriptions of the items used, removed, and modified should be reported in order to better understand which items function best within a diverse range of contexts. It is possible, that due to journal-specific word count requirement, authors may be limited in how much detail they can provide regarding the scales included in their analyses. Given the limited information regarding adaptation processes in the literature, our study team is currently investigating young men and women’s perceptions of items used within the South African adaptation of the SRPS, with South African youth participating in a youth-engaged cohort study (Closson, Dietrich, et al., 2019). This research will help to ensure that young people accurately understand the items asked, identify and address any potential issues in translation, and if there are missing aspects of SRP equity that are currently not being asked.

Only two of the studies in the review used the original scoring recommendation of dividing total scale score by the number of items (Closson, Dietrich, et al., 2019; Pulerwitz et al., 2018). The majority of papers in this review thus have differential scores that are not comparable across studies. Future research using the SRPS should consider scoring the SRPS using the original authors recommendations and depending on the distribution of the data, potentially creating terciles in order to compare levels of SRP equity across studies and populations.

Associations between SRP equity scores and sexual behavior and condom use is dependent on which items were included in the scale. Although Pulerwitz recommends that questions around condom use be removed when assessing the relationship between SRP equity and condom use, at least one study that found an association between SRP equity and condom use did not remove this item (Pettifor et al., 2004). Given that not all papers included the full list of items used, we are unable to determine whether some of the studies’ associations were conflated or not. Despite this potential bias in results, there were relatively consistent results showing associations between SRP equity and increased condom use among young women (Pettifor et al., 2004; Shai et al., 2010; Teitelman et al., 2016). For men, however, results were inconsistent, finding that SRP equity was associated with reduced likelihood of condom use among men in one study (Teitelman et al., 2016) and increased condom use in two others (Harrison et al., 2006; Shai et al., 2012). This may partially be explained by the way in which the scale assessed SRP equity for young men. For example, Teitelman’s (2016) findings highlights that men in relationships where they perceive their partner to be controlling are less likely to use condoms, and those reporting they themselves are controlling were also less likely to use condoms. These results may be due to multiple interacting factors beyond SRP equity influencing condom use for young men (Chirinda & Peltzer, 2014; Manyaapelo et al., 2017; Mehra et al., 2014). Associations between SRP equity and sexual partners were more consistent across genders, where studies found that young women and young men with higher SRP equity were more likely to report fewer partners (Harrison et al., 2006; Teitelman et al., 2016). All studies around sexual behavior were cross-sectional, highlighting a need to examine time-dependent associations between SRP equity and HIV-risk behavior.

Power dynamics within youth’s sexual relationships vary widely depending on the type of relationship. One major limitation of the SRPS for use among youth is that it primarily only asks the participant to discuss the level of power inequity occurring within their primary relationships. Prior studies on relationships among young people in sub-Saharan Africa have demonstrated that it is common for both young men and women to have multiple concurrent partnerships (Crankshaw et al., 2016; Maughan-Brown et al., 2014; Steffenson et al., 2011). Moreover, many of the SRPS items are heteronormative in nature. Given the shifting relationship dynamics of young people, it is challenging to assess whether the partner that youth are referring to when answering the SRPS is the same partner perpetrating IPV or passing HIV. Also, the scale does not allow for use within nonheterosexual relationships. If many sexual acts are occurring outside of young men and women’s perceived primary partners, current measures of SRP equity will not be able to capture potential SRP inequities within these relationships. For example, Harrison et al. (2006) found that men who were more dominant were less likely to use condoms in their nonprimary relationships. Future research should consider asking questions around relationship control and decision-making power for the multiple relationships that young men and women may have, including within nonheterosexual relationships.

Practice

No study found any significant association between mental health and SRP equity among young women after adjusting for potential confounders, which requires further investigation due to the adverse traumatic effects associated with IPV (Cavanaugh et al., 2010; Fulu et al., 2013; Hansrod et al., 2015). It is likely that mental health outcomes may be more closely linked to experiences of IPV, which is a common consequence of low SRP equity for young women (Kiene et al., 2017; Tsai et al., 2016). Among young men, one study did find that young men with greater SRP equity had less depressive symptoms (Gibbs, Govender, & Jewkes, 2018). However, this analysis was cross-sectional, and given previous longitudinal studies that have found no association between SRP equity and depression among men (Nduna et al., 2010), future research is needed to establish the temporality and directionality of the association. High levels of unemployment and poverty among young men growing up in South Africa, and other sub-Saharan African nations impact young men’s mental health and contribute to gender role strain, which has been associated with increased controlling behaviors as well as perpetration of IPV (Closson, Hatcher, et al., 2019; Gottert et al., 2016, 2018). Despite the important role mental health plays in terms of SRP equity and perpetration of violence, in many sub-Saharan African settings, there exists few resources and support for mental health. With a growing body of literature on masculinities and how gender role strain may result in young men forming harmful masculinities resulting in behaviors that impact the health and well-being of their partners (Gibbs et al., 2014; Jewkes et al., 2011a, 2011b; Morrell et al., 2013), increased efforts are being rolled out to attempt to provide opportunities for young men to address gender role stressors, including restrictive emotionality, through men’s programs aimed at redressing harmful formations of masculinity (Peacock & Barker, 2014; Peacock & Levack, 2004; Seidler, 2007).

Although papers related to health impacts, including HIV acquisition and pregnancy incidence, were more likely to use cohort (time series) data, only a small number of papers included in the review looked at these outcomes (Christofides et al., 2014a, 2014b; Jewkes et al., 2010). Jewkes and colleagues’ (2010) seminal paper found that low SRP equity was associated with HIV incidence (Jewkes et al., 2010). Using data from the same intervention, Christofides and colleagues (2014) found no association between low SRP equity and subsequent incidence of unwanted pregnancy among young women; however, there was an association between experiencing physical IPV and incidence of unwanted pregnancy at 12-month follow up (Christofides et al., 2014a).

Addressing gender inequities and experiences of violence toward women are global health priorities. In our review, a number of papers showed associations between having low SRP equity and experiencing IPV among young women (Belus et al., 2019; Closson, Dietrich, et al., 2019; Conroy, 2014; Pulerwitz et al., 2018; Stockl et al., 2015) and perpetrating IPV among young men (Gibbs, Jewkes, et al., 2018). However, SRP equity may not be directly associated with other health outcomes such as depression or pregnancy that have been associated with increased experiences of IPV (Christofides et al., 2014b, 2014; Nduna et al., 2013). Previous evaluations of gender transformative interventions, such as Stepping Stones, have found some evidence that educational training and workshops centered on shifting harmful gender norms had reduced the perpetration of IPV among young men; however, there was no documented improvement in SRP equity (Jewkes et al., 2008). Moreover, interventions providing conditional cash transfers may function to reduce IPV through mediating pathways of reduced sexual partners and delaying sexual debut, however, not through SRP equity (Kilburn et al., 2018). Given the strong association between SRP equity and IPV, these findings appear to be counterintuitive and may be the result of SRP equity measures not accurately capturing experiences of power imbalances in sub-Saharan African youth’s relationships. An increased understanding of how SRP manifests within the relationships of young people can inform violence and mental health practice, specifically treatment and prevention strategies.

Policy

In order for effective policies to be implemented, there is a need to improve the quality and standards of research in the area of gender equity and VAWG globally (Jewkes, 2010). At present, little is known regarding pathways from SRP equity to IPV and HIV. Gibbs et al. (2018) was the only reviewed paper that used structural equation modeling to examine the multiple intersecting pathways that are potentially driving low SRP equity in relationships and then in turn how SRP inequities result in negative outcomes, such as perpetrating IPV (Gibbs, Jewkes, et al., 2018). Identifying potential drivers of SRP inequities is key for informing interventions and policies which may help to mitigate risk of experiencing SRP inequities. Given the complexity in the pathways from SRP inequity to HIV and other sexual and mental health outcomes, future research is needed to assess gendered pathways between SRP equity and HIV. Moreover, there is a gap in understanding the temporal relationships between SRP equity and sexual and reproductive health outcomes, as well as mental health.

The majority of papers included in this review rely on the assumption that young women with more SRP equity will in turn have increased agency to negotiate safe sex, including condom use. The results presented herein should take into consideration that some young women with high decision-making power may choose to not negotiate condom use for multiple reasons beyond having limited agency in their relationships, such as simply not wanting to use a condom or that condomless sex is associated with increased feelings of emotional intimacy (Gebhardt et al., 2003). Complexities in intimacy, sexual pleasure, power, and control within young people’s relationships may indicate why one study found that increased SRP equity was associated with reduced condom use among young women (Harrison et al., 2006). Furthermore, the studies presented here assume that sex is mostly negotiated and consensual and yet evidence suggests otherwise (Nduna, 2020). Although HIV-prevention studies have not delineated transmissions that occur from consensual versus nonconsensual sex, the risk inherent in sexual violence should always be borne in mind when designing HIV-prevention interventions for sub-Saharan Africa. With the roll out of diverse options for HIV prevention including pre-exposure prophylaxis, future quantitative and qualitative studies should examine condom use preference that move beyond the assumption that all women with improved sexual agency will increase condom negotiation and use.

Limitations

Due to the varied ways in which SRP equity was measured, including discrepancies in the number of items used, we were unable to conduct a meta-analysis. Also, given that many of the papers included in the review were both exploratory and cross-sectional in nature, we were unable to determine some dimensions of the quality, direction of effect, and temporality in the results presented within the reviewed papers. This review specifically was interested in exploring the use of the SRPS within quantitative youth-specific studies in sub-Saharan Africa. There are a number of other studies which have used the SRPS with older cohorts that include young men and women, and studies that have qualitatively explored the role of SRP equity within the sexual lives of young men and women growing up in sub-Saharan Africa; however, due to our inclusion criteria, they were not included in this review. The exclusion of these studies may have resulted in missed opportunities for examining the use of the SRPS within sub-Saharan African contexts.