Inclusion criteria

• Study design: any interventional study, including randomised controlled trials (RCTs), cohort studies, case-control studies and before-and-after studies.
• Population: females aged over 15 years living in LMICs (defined by the World Bank Group’s classification [16]).
• Intervention: any approach the author described as a "rights-based approach", or involving sexual and reproductive health interventions that authors explicitly related to "rights" (for example, an intervention that promoted the awareness of rights). We also included studies based on an intervention referred to as rights-based in another included paper. We only included interventions directed towards women (other than healthcare professionals) or their families. Studies using educational interventions were included in addition to non-educational interventions, as long as they were explicitly described by study authors as related to “rights”.
• Comparator: Acceptable comparators included clusters in which no intervention or fewer components of an intervention (for example without the rights-based components) were in place, or individuals not exposed to interventions, or exposed to fewer intervention components.
• Primary outcomes, measured through surveys or routine data (health or crime):
  • Gender-based violence (prevalence/incidence)
  • STIs (testing/prevalence/incidence)
  • Pregnancy and delivery-related outcomes (including antenatal, perinatal and post-partum healthcare access, pregnancy/birth complications)
  • Perinatal mental health (incidence/prevalence)
  • Unintended pregnancy (incidence)
  • Female genital mutilation (incidence/prevalence)
  • Contraceptive/condom use
  • Sexual and reproductive healthcare attendance (number of visits)
  • HIV-related outcomes (including viral load, CD4 count, medication adherence)
  • Mental wellbeing as measured by any wellbeing scale
  • Awareness of rights
• Secondary outcomes, measured through surveys or routine data (health or crime):
  • Healthcare attendance unrelated to sexual and reproductive health (number of visits)
  • Healthcare workers’ understanding of rights (difference between before and after intervention)

Exclusion criteria

• Study design: Qualitative studies
• Population: Children under 15 years. People who are male or transgender.

Following database searches, titles and abstracts were screened by two reviewers independently. Each included full-text article was then reviewed by two reviewers independently. At each state, if there were discrepancies regarding inclusion/exclusion decisions, they were addressed initially via discussion, and if necessary via a third author.

Rayyan Systematic Review Web App [17] and Microsoft Excel were used to record decisions.

Data were extracted by two reviewers independently onto a data extraction tool developed by the authors which included basic study details (journal, dates, authors, and location, sample size, population, study design, study duration, interventions, statistical analysis, results, ethics and funding). Only relevant results pertaining to populations eligible for inclusion in this systematic review were extracted- for example, where data was collected for males and females, only data presented on females were extracted.

Analysis

A narrative synthesis of included studies was undertaken. Due to study heterogeneity, meta-analysis was not undertaken. Outcomes were mapped to identify evidence gaps. Effectiveness of identified interventions was examined based on the author’s choice of outcomes.

Quality and risk of bias was assessed by two reviewers independently using the RoB-2 tool for cluster-randomised studies [18], and the ROBINS-I tool for non-randomised studies [19]. Uncontrolled before-and-after studies based on two cross-sectional surveys were given an overall assessment of serious or critical risk of bias [20]. Disagreements were addressed via discussion and if necessary via a third author.

Risk of bias in included studies

All studies were classed as high, serious or critical risk of bias, meaning that due to issues with the quality of the studies, there was a substantial risk that included studies overestimated or underestimated the true effect of the interventions. The risk of bias in cluster RCTs and non-randomised interventional studies is displayed in S1 and S2 Tables respectively. Uncontrolled cross-sectional before-and-after studies were all classified as serious or critical risk of bias (S3 Table).

Population and setting

As shown in Table 1, thirteen studies were undertaken among female sex workers (FSWs) in India: eight in south India as part of the evaluation of the India AIDS initiative, Avahan [23,31–33,35–38], and five in West Bengal, India, as part of the Sonagachi project [24,39–41,44].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of included studies.

https://doi.org/10.1371/journal.pone.0250976.t001

The remaining eleven studies were undertaken in Bangladesh (married women living in slums) [21], Philippines (FSWs) [22], Tanzania (women in the third trimester of pregnancy) [25], East Nepal (women who had delivered a baby within three years) [26], India (pregnant women [28] and women who had delivered a baby within twelve months [30]), Uganda (women living in rural areas) [29], Kenya (women who had given birth in last 24–48 hours) [42], the Amazon region of Brazil (FSWs- two papers using the same dataset) [34,43], and Egypt (women who were newly married, pregnant or post-partum) [27].

Study design and interventions

Four studies were cluster RCTs [21,28,29,40], two non-randomised interventional studies [24,39], one a time-series study design [23], and seventeen were retrospective observational studies based on one or more cross-sectional surveys [22,23,25–27,30–38,41–44].

All studies involved multiple components, including various elements alongside (or part of) a rights-based approach. These included education/workshops on rights and sexual health, sexual health service provision, capacity building, advocacy, community mobilisation, peer support and outreach, condom access and promotion, leaflets and posters, and community meetings (Table 1).

Outcomes and results

Table 2 shows the results of each study. Both Benzaken et al. 2003 and 2007 [34,43], and Jana et al. 1995 and 1998 [41,44] contained data from the same study so are reported together.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Outcomes identified in included studies.

https://doi.org/10.1371/journal.pone.0250976.t002

Findings

We found that rights-based interventions appear to be effective for certain outcomes within sexual and reproductive health, but evidence is of poor quality. Given the considerable risk of bias across all studies, the results must be interpreted with caution.

Two studies identified increased STI symptoms/diagnosis following intervention, but it is unclear whether this reflects a true increase. For instance, there was an increase in Herpes Simplex Virus-2 diagnosis in Reza-Paul et al.’s study post-intervention but the prevalence of other STIs decreased [36]. Halli et al. found increased STI symptoms among those with higher self-reported intervention exposure, although this could relate to a greater awareness of symptoms among those exposed to the intervention [37].

In keeping with a previous systematic review (mentioned above), which looked at interventions aiming to promote awareness of rights in order to increase use of maternity services [12], our study found mixed results in terms of the effect of interventions on healthcare access during pregnancy and delivery; of the four studies in our systematic review reporting this outcome, three were included in both reviews. The additional study reported no significant effect of intervention on whether women delivered in a healthcare setting (versus at home), whether a skilled birth attendant was present or whether a clean home delivery kit was used [26]. This relative lack of evidence for rights-based approaches in maternity settings is disappointing, and further studies are urgently needed, particularly given the recommendation for rights-based approaches in the OHCHR Technical Guidance [11].

In contrast to Stangl et al.’s systematic review of the impact of human rights programmes on HIV-related outcomes which found them to be largely effective [13], our systematic review showed mixed results: one study found a reduction in HIV prevalence [35], one reported no change [36] and one reported a significant increase [44] (although all were at critical risk of bias). Interestingly, only one study overlapped between the two systematic reviews, largely due to a different interpretation of a rights-based approach [23]. The overlapping study was limited to outcomes related to news reporting and intervention exposure/service use (not HIV-specific).

Population

Over half of the twenty-four studies included were undertaken among FSWs in India [23,24,31–33,35–41,44]. These studies were evaluating aspects of either the India AIDS initiative [23,31–33,35–38] or the Sonagachi Project [24,39–41,44]. The remaining eleven studies were undertaken across a fairly wide geographical footprint (India, Bangladesh, Philippines, Tanzania, India, Nepal, Kenya, Brazil and Egypt). Only seven studies were undertaken among a population that was not made up of FSWs [21,25,27–30,42], most of which were among women who were pregnant or had recently given birth. There remains a paucity of evidence on rights-based approaches among those who do not engage in sex work, and in non-maternity settings.

Study identification

Many studies identified in this systematic review were not identified from database searching. This is in keeping with a systematic review mentioned above where one of the four studies included was identified from searches and the other three from discussions with experts [12]. This may be because the word ‘rights’ is not often mentioned in key words, abstracts or titles [12]. We found that even if an intervention was described as related to ‘rights’ in one paper, another paper evaluating the same intervention (but evaluating different outcomes) did not necessarily refer to the same intervention as related to ‘rights’. For example, Reza-Paul et al. described a community mobilisation and outreach intervention as using a “rights-based approach” [36], whilst Ramesh et al.’s study did not mention ‘rights’ despite evaluating the same intervention [35].

Similarly, another systematic review found that despite the studies included using what authors defined as human rights-based approaches, many studies did not explicitly refer to the protection and promotion of human rights in reference to the intervention being evaluated [13]. Thus, intervention descriptions and whether interventions are ‘rights-based’ is subjective, and some relevant studies may have been missed in this systematic review. Given the importance of women achieving their sexual and reproductive health rights [6,11], future studies should ensure they are explicit in their use of a rights-based approach or at least refer to ‘rights’ in relation to the intervention if they are promoted so that the available evidence can be appropriately evaluated.

Outcomes not identified from included studies

Not all important sexual and reproductive health outcomes were identified in the papers included. No results were identified regarding legal action or convictions related to gender-based violence; this is consistent with a previous systematic review of systematic reviews of gender-based violence prevention among adolescent girls in low-income countries which found a paucity of evidence [45]. Gurnani et al. described 4,600 rights violations towards FSWs reported to the police during the study period, and crisis management teams supported 92% of these, but it was unclear whether this was a change from previously [23]. Ratcliffe et al. found that 10% of women who attended their Open Birth Days intervention filed a complaint to the hospital regarding their treatment following the intervention, compared to no women before the intervention, but they did not refer to legal actions or convictions [25]. Additionally, outcomes related to perinatal mental health, unintended pregnancy or abortion, delivery outcomes, female genital mutilation or mental wellbeing were not identified in any included papers.

Limitations

All included studies used complex interventions, usually involving multiple components with rights being some part of it. There was no consistent approach to incorporating rights into programmes. Even in situations where studies found the interventions effective, it was unclear whether or not this was due to the rights-based aspect. Detail regarding interventions in many papers was limited and it was often unclear how the rights-based element was incorporated, exacerbating the challenge in identifying whether a rights-based approach is effective.

Our inclusion criteria were based on the author recognising that the approach was rights-based and making that explicit. Not having our own definition of what we would consider a rights-based approach means that there may have been studies that were not included, but which examined interventions that did have a rights-based approach. Note that several studies we included did not mention that the interventions studied were rights-based, but because they examined the same intervention that had been described that way in other publications, we included them. This suggests there may be studies our systematic review has missed. Similarly, it is possible that some of the studies included would not have met a strict definition of what a rights-based approach should entail. Reassuringly, none of the included studies were obviously in this category.

Since this systematic review was limited to females, some studies were excluded due to outcomes not being separated by gender, and some outcomes within included studies were also excluded for this reason. For example, Rana et al. included relevant outcomes that were not separated by gender (even if it may be assumed that most people answering the question were women): data were available on knowledge of circumstances when abortion is legal or illegal, HIV and STIs, and contraceptive methods but as these data were not available for women only [26], they were not included in this systematic review.

As discussed, the quality of studies overall was poor. Most studies used an uncontrolled before-and-after study design. These are by nature prone to bias as there is no way of knowing what might have happened without the intervention and what else occurred in the time period which might actually have been responsible for any findings: they are therefore at high risk of confounding [46]. Moreover, some studies did not undertake any statistical analysis of results nor explain why analysis was not undertaken [23,25,41,44]. We did not undertake a meta-analysis due to the heterogeneity of populations, interventions and outcomes. This limited us in undertaking a rigorous estimate of effectiveness.

Conclusions

This is the first systematic review to evaluate the evidence on rights-based approaches to sexual and reproductive health including maternity, gender-based violence, sexual health and HIV. We undertook a comprehensive search of the literature from three relevant leading electronic databases, as well as reference searching of included studies and relevant systematic reviews. By synthesising the evidence base, we have illuminated the evidence gaps and what should be done in future research to improve the quality of the evidence base. Future research should be more explicit about the use of rights-based approaches and specify more precisely what distinguishes a rights-based approach from others, in a way that would allow evidence to be gathered. Moreover, evidence is needed across a diverse range of populations (not limited to FSWs and maternity settings). Given that rights-based approaches are recommended widely [6,47], high quality (ideally cluster randomised) controlled studies need to be undertaken urgently to determine whether rights-based approaches to sexual and reproductive health are effective in LMICs.