Health facility-based interventions and the uptake of contraception among people living with HIV: A systematic review & meta-analysis

Abstract

Background & objectives:

The prong 2 of 4 prong strategy introduced by the World Health Organization aims at averting unintended pregnancies among people living with HIV (PLHIV). This systematic review aimed to generate evidence on the effectuality of facility-based interventions in improving uptake of modern and dual contraception, for reducing unmet family planning (FP) needs and unintended pregnancies among PLHIV.

Methods:

Articles evaluating facility-based interventions to integrate human immunodeficiency virus (HIV) and FP published in English language were included. Eligible studies were identified from electronic and lateral search from three databases (PubMed, Cochrane Library and Web of Science) and grey literature. HIV care with no/minimal focus on FP was considered a comparator. Quality was assessed using design-appropriate tools. Descriptive analysis was presented in tables. Uptake of dual methods, unmet FP needs and unintended pregnancies were included in the meta-analysis to estimate pooled odds ratio (OR) with random effect model, P and I² values.

Results:

The search yielded 2112 results. After excluding duplicates and unfit articles, 17 were found eligible for review and nine for meta-analysis. The pooled OR for uptake of dual contraception was 1.69 (1.14, 2.5) (P=0.008; I²=90%), for unmet FP needs was 0.58 (0487, 0.69) (P<0.00001; I²=0%) and for unintended pregnancies was 0.6 (0.32, 1.1) (P=0.1, I²=38%).

Interpretation & conclusions:

The results of this meta-analysis suggest that health facility-based interventions to integrate HIV and FP services do result in improved uptake of dual methods and reduce unmet need for contraception along with a protective trend on incidence of unintended pregnancies. Such facility-based integration would ensure universal access to effective contraception and facilitate in achieving Sustainable Development Goals that aim at ending epidemics like HIV.

According to the World Health Organization (WHO), 38.4 million people were living with human immunodeficiency virus (PLHIV) in 2021, and 6,50,000 people died due to HIV-related reasons1. New HIV infections were acquired by 1.5 million, among which 1,60,000 were children (<15 yr of age)1. Across the world, around 0.7 per cent of adults are reportedly living with HIV1 as of 2022 year. Since the third target of the Sustainable Development Goals (SDG) is to end epidemics like AIDS2, it is essential to control the transmission of HIV between partners and from mother to child to achieve this target.

A four-prong strategy was recommended by the WHO to prevent parent-to-child transmission (PPTCT) of HIV. The first prong deals with prevention of the infection; the second focuses on prevention of unintended pregnancies; providing specific interventions to inhibit transmission of HIV from pregnant mother to infant is prong 3; care, support and treatment for mothers and children is prong 43. HIV testing and care with comprehensive family planning (FP) services would ensure a people-centred approach that would enable women living with HIV (WLHIV) to make informed decisions regarding their reproductive health and fertility desires3.

The efforts for PPTCT by the national and global organizations have been largely focussed on prong 3 through highly active antiretroviral therapy (HAART) rather than on preventing unintended pregnancies4,5. HAART has assisted in decreasing parent-to-child transmission to <2 per cent6. However, to end epidemics like acquired immune deficiency syndrome (AIDS) by 20302, strengthening all prongs of PPTCT is essential. The Glion Call to Action7 in 2004 for linking FP and HIV services had led to the initiation of several policy ideas and implementations that were targeted at achieving the same. Post this event, several studies and programme evaluations had demonstrated that the integration of HIV care to FP and vice versa could improve the uptake of FP services among PLHIV8,9,10,11,12,13,14. This would reduce the stigma related to it, thereby improving the quality of life and service coverage across this vulnerable population15.

Integration/linkage of HIV and FP services can be rendered by providing FP services within the HIV care facilities. Referral of PLHIV for FP services, training of service providers, Information, Education and Counselling (IEC) activities, data management, monitoring and evaluation are various components of service integration. These components of integration/linkage can empower PLHIV to achieve their fertility desires. The WHO prong 2 and 1 can be addressed by implementing these enlisted interventions.

The evidence pool on this subject continues to grow as the HIV/AIDS control programme across the globe improves as suggested by the six systematic reviews conducted from 2009 to 20178,9,10,11,12,13. Each of these reviews had expanded the scope of desired outcomes with the addition of newer evidence. The first review in 20098 concluded that more rigorously designed studies that compare the intervention with a vertical programme are required to obtain optimal results on the uptake of FP services while review in 20139 mentions a lack of studies with better rigor.

The last available review from 201710 had concluded that generalized outcomes like the uptake of modern contraception could be biased due to social desirability, and a better alternative is to use a more specific measure as dual methods. It also highlighted the lack of adequate evidence on reduction in unintended pregnancies.

While FP services are considered an essential component of HIV services, these are rarely addressed in the programme documents and guidelines. The programme focus is generally skewed towards diagnosing pregnant WLHIV and treating them and their infants with HAART. The concept of ‘dual methods’ is neither used nor emphasized in the programme documents16,17. The rationale behind this study was to underscore that despite existing evidence indicating a high unmet need for FP among WLHIV18,19,20,21, there is no clear guidance or an existing programme that focuses on addressing the second prong of prevention. This systematic review represents the first meta-analysis on this subject, emphasizing how specific interventions could promote the adoption of dual methods, thereby reducing unmet FP needs and unintended pregnancies. The aim was to generate evidence on the effectiveness of health facility-based interventions in improving uptake of modern contraception and dual methods among PLHIV in the reproductive age group, reducing unmet FP needs and incidence of unintended pregnancy among WLHIV.

Material & Methods

The protocol for this systematic review and meta-analysis was registered in PROSPERO (CRD42021242892)22. The manuscript was drafted adhering to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines. Published studies were searched in three electronic databases: PubMed, Cochrane Library and Web of Science. We looked for grey literature and also conducted lateral search.

Inclusion criteria: English language articles were included that measured one/more of the desired outcomes and considered only those that had a comparison group. Studies published from January 1, 2010 to November 15, 2022 were included. The population of interest was PLHIV in the reproductive age group. Health facility-based intervention studies at any level of care were included.

Exclusion criteria: Care was taken to avoid studies with titles/abstracts mentioning the study population to be exclusively high-risk groups like sex workers because these individuals may have different sexual practices and their contraceptive use cannot be generalized to the common public23. Individuals who suffered from WHO stage IV of HIV infection were also excluded as some contraceptive uses are contraindicated for them24, and community-based studies were excluded from the review.

Outcomes: The outcomes of interest were: (i) Uptake of modern contraception (when the participant used methods such as sterilization, intrauterine device (IUD), implants, injectable, oral pills and condoms25); (ii) Uptake of dual methods (use of two effective contraception methods, usually condom along with another modern/effective method26); (iii) Incidence of unmet FP needs (participants who had no intention to have children but were currently not using any method of contraception27); (iv) Incidence of unintended pregnancy (WLHIV reported her pregnancy to be undesired, i.e. either mistimed or unwanted28).

Data collection: Two reviewers assessed the titles and abstracts based on the inclusion and exclusion criteria. Rejected articles were documented with the reason for exclusion. If the reviewers had uncertainty regarding the inclusion/exclusion of any study, a third reviewer was involved in decision-making. Duplicates were removed and those studies selected for full-body review were recorded with title, authors, year, location, design, sample size, methodology, results and limitations.

Quality assessment: The quality of the studies was assessed with appropriate risk of bias tools. The Cochrane risk of bias tool was used for randomized control trial (RCT)29, Risk of Bias in Non-Randomized Studies of intervention for non-randomized trial30, Evidence Project Risk of bias tool for pre-post-intervention studies31, Critical Appraisal Skills Programme (CASP) for cohort studies32 and Appraisal of cross-sectional studies using AXIS tool33.

Analysis: The studies were assessed and summarized as text and tables. The narrative results focussed on elaborating the various components of the intervention and the desired outcomes reported by the studies. The output indicator-uptake of dual contraception and the outcome indicators-prevalence of unmet FP needs and unintended pregnancies were measured uniformly by the included studies and thus considered ideal for inclusion in the meta-analysis34. The analysis was conducted using RevMan 5.4.1 (https://revman.cochrane.org) and Comprehensive Meta-analysis 4 (https://meta-analysis.com) software. The desired outcomes were noted as dichotomous results. The pooled Odds ratio (OR) in the random effect model (REM) summarized the outcomes. The I² values identified heterogeneity, and P values measured significance. Funnel and forest plots identified publication bias and dispersion of effect sizes among the individual studies, respectively. Publication bias was further explored using statistical tests (Egger’s regression and Begg and Mazumdar rank correlation). In addition, a meta-analysis was conducted after allocating varying weights to the included studies based on the study design.

Results

The PRISMA diagram summarizes the identification, screening and inclusion of studies (Fig. 1). A total of 2112 studies were identified: 1258 from PubMed, 171 from Cochrane library, 567 from Web of Science and 116 from grey literature. Identified 2058 studies (after exclusion of duplicates) went through abstract/title screening. 1992 studies that did not fit into the inclusion criteria were excluded, and 66 were selected for full article review. In addition, 10 more studies were identified from lateral search. Therefore, a total of 76 articles were reviewed. Out of this, 59 studies were excluded for various design and outcome related issues (Fig. 1). Seventeen eligible articles were selected for the systematic review and nine for meta-analysis.

Fig. 1.

PRISMA flowchart.

Description of studies: Seventeen studies were included in the systematic review: four RCTs and one two-year follow up analysis of an RCT18,35,36,37,38, two non-RCTs39,40, six pre-post-intervention studies41,42,43,44,45,46, one cohort study47 and three cross-sectional comparison studies48,49,50. Among these studies, 15 belonged to African countries: four from Kenya36,38,47,50 and two from South Africa35,42, Uganda37,48 and Malawi44,49. One study each belonged to Zimbabwe39, Nigeria40, Tanzania41, Botswana45 and Zambia46. Apart from the African countries, one study each belonged to India18 and the United Kingdom43.

The components of interventions implemented in these studies focussed on clients (counselling, referral and IEC activities), providers (training on systematic counselling and FP) or the facility-based strengthening of integration (availability of barrier methods and sterilization within the HIV facility). These interventions were, however, not mutually exclusive. Implementation of one or more of these interventions was considered integration of HIV-FP services. Four studies implemented client counselling36,38,39,48; four studies had counselling along with referral of clients for FP services18,35,41,49; two studies organized IEC activities for the clients18,39; one study practiced client referral and IEC activities together42; ten studies practiced provider training on FP services and/or method of counselling18,35,36,40,41,42,43,45,46; FP commodities were made available within the HIV clinic36,38,43,46,47,48,50. One study implemented44 an electronic medical record, a software-based prompt to the provider to initiate FP questionnaire and services. A FP voucher valid for three months postpartum along with SMS reminders and educational counselling were given in one study38.

Among the studies, uptake of modern contraception was assessed by 16 studies35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50 and dual methods by seven18,35,36,37,41,46,50. Four studies evaluated the prevalence of unmet FP needs41,46,48,50. Unintended pregnancies were reported by only four18,40,43,50. The Table presents the study designs, quality assessment and components of interventions. All four RCTs18,35,36,37 had a low risk of bias (Table). The sample size of the RCTs ranged from 12135 to 358437. Furthermore of these four studies, three18,35,36 were used for meta-analysis. These RCTs contributed 3988 participants to the meta-analysis, thereby reducing the bias within the analysis. The Supplementary Table ^{(566.7KB, pdf)} summarizes the features of the studies, interventions implemented and results with the measures of significance.

Table.

Study design, quality and components of intervention of studies included in the systematic review

Study short tile, year and country	Components of intervention	Study design	Risk/quality*
Mantell et al35, 2017, South Africa	Counselling and referral of clients Training of providers	RCT	Low risk of bias
Joshi et al18, 2016, India	Counselling, IEC activities and referral of clients Training of providers		Low risk of bias
Grossman et al36, 2013, Kenya	Counselling of clients Training of providers		Low risk of bias
Cohen et al37, 2017, Kenya	Availability of FP services within HIV clinic		Low risk of bias
Atukunda et al38, 2019, Uganda	Counselling, family planning voucher postpartum SMS reminders to utilize the voucher within three months postpartum		Low risk of bias
Sarnquist et al39, 2014, Zimbabwe	Counselling, IEC activities amongst clients	Non-RCT	Low risk of bias
McCarraher et al40, 2011, Nigeria	Provider training Facility assessment and monitoring Community mobilization		Low risk of bias
Baumgartner et al41, 2014, Tanzania	Counselling, referral and follow up of clients Training of providers	Pre-post interventional studies	Good
Hoke et al42, 2014, South Africa	Referral and IEC activities for clients Training of providers		Good
Weilding et al43, 2016, UK	Training of providers Availability of FP services within HIV clinic		Fair
Tweya et al44, 2017, Malawi	EMR system to prompt providers		Good
Hawkins et al45, 2021, Botswana	Training of providers		Fair
Medley et al46, 2022, Zambia	Training on providers Availability of FP services within HIV clinic		Fair
Kosegi et al47, 2011, Kenya	Availability of FP services within HIV clinic	Cohort study	Good
Wanyenze et al48, 2015, Uganda	Counselling Availability of FP services within HIV clinic	Cross-sectional comparative study	Fair
Phiri et al49, 2016, Malawi	Counselling and referral of clients		Poor
Chen et al50, 2020, Kenya	Availability of FP services within the HIV clinic		Fair

^*Method of scoring the studies.The Evidence project tool, CASP and AXIS checklists did not have a method to score the studies. A score of >4 suggested that the study is of good quality; 3 was fair and <3 as poor quality while using the evidence project tool. For CASP, among the 11 questions if the study gave ≤2 ‘no/can’t-tell’ it indicated good quality; 3-5 ‘no/can’t tell’ meant fair quality and ≥6 indicated poor quality. For AXIS, among the 20 questions, <3 negative answers meant good quality; 4-8 indicated fair quality and >9 was of poor quality. EMR, electronic medical record; HIV, human immunodeficiency virus; CASP, critical appraisal skills programme; FP, family planning; IEC, information, education and counselling; RCT, randomized control trial; SMS, short message services

Uptake of modern contraception: Out of the16 studies that assessed the uptake of modern contraception, the major components of interventions implemented in these studies were counselling, referral, IEC activities for clients and training of service providers. Among the various modern contraception options, IUD and implant use, followed by sterilization, injectable, oral pills and condoms, showed a commendable uptake. Utilization of modern contraception improved in 14 studies. Four studies40,41,42,48 reported the improvement in the uptake of modern contraception, in general, to be significant. A Zambian study46 reported a significant uptake of injectables along with a significant reduction in the use of oral pills. A study from Kenya50 reported a significant improvement in the uptake of long-acting reversible contraception methods in specific, along with modern methods in general. One study47 reported a significant improvement in the uptake of modern contraception only when condom was included as a method and a reduction otherwise. A two-year later, prospective analysis37 of an RCT36 showed a significant reduction in use of modern contraception, when condom was included as a method.

Uptake of dual methods: Seven studies assessed dual methods. These studies implemented client counselling, referral and IEC activities with the training of service providers and the availability of FP commodities from the facility. The uptake improved in six studies18,35,36,41,46,50 with significance reported by two18,46. A two-year later analysis of an RCT37 reported a significant reduction in the uptake of dual methods in both the control and intervention groups.

Unmet family planning (FP) needs & unintended pregnancies: Four studies analyzed unmet FP needs. All studies implemented client counselling and provider training as a part of their intervention and reported a reduction in unmet FP needs among the intervention group, of which two studies46,50 reported significance. Three studies reported the incidence of unintended pregnancies. Two studies18,43 reported a reduction in the incidence of unintended pregnancies among the intervention group.

Meta-analysis:

Uptake of dual methods: This outcome was measured by seven studies, four RCTs and three comparative studies. The P values from both Egger’s test (P=0.056) and Begg’s test (P=0.548) were greater than 0.05, suggesting the absence of publication bias. While considering all studies, the pooled OR in REM with 95 per cent confidence interval (CI) (Fig. 2) for uptake of dual contraception was 1.69 (1.14, 2.5) (P=0.008; I²=90%). The analysis was repeated by considering all studies but allocating different weights based on the study designs. The pooled OR was 1.57 (0.95, 2.61; P=0.07).

Fig. 2.

(A) Forest plot. (B) funnel plot for the uptake of dual methods among people living with HIV and publication bias. SE, standard error; OR, odds ratio.

Unmet family planning (FP) needs & unintended pregnancies: Since unintended pregnancies are a subset of the unmet FP needs, these indicators were considered together for this analysis. The P values from both Egger’s test (P=0.19) and Begg’s test (P=0.45) were greater than 0.05, suggesting the absence of publication bias. Four studies41,46,48,50 directly measured unmet FP needs and three others18,43,50 measured unintended pregnancies. The pooled OR in REM (Fig. 3) with 95 per cent CI was estimated to be 0.57 (0.47, 0.67) (P<0.0001; I²=0%). The analysis was repeated after allocating different weights based on the study designs. The pooled OR was 0.604 (0.33, 1.09; P=0.09).

Fig. 3.

(A) Forest plot. (B) funnel plot after considering unintended pregnancies as a subset of unmet FP needs, among people living with HIV and publication bias.

Four studies measured only unmet FP needs41,46,48,50. The pooled OR in REM (Fig. 4) with 95 per cent CI was estimated to be 0.58 (0.48, 0.69) (P<0.00001; I²=0%). The analysis was repeated by considering all studies but allocating different weights based on the study designs. The pooled OR was 0.65 (0.29, 1.46; P=0.305).

Fig. 4.

Forest plot for unmet FP needs among people living with HIV.

Three studies measured only unintended pregnancies18,43,50. The pooled OR in REM (Fig. 5) with 95 per cent CI was estimated to be 0.6 (0.32, 1.1) (P=0.1; I²=38%). The analysis was repeated by considering all studies but allocating different weights based on the study designs. The pooled OR was 0.605 (0.27, 1.34; P=0.219).

Fig. 5.

Forest plot for unintended pregnancies among people living with HIV.

Discussion

The objective of this systematic review was to collate evidence on the effectiveness of integrated HIV-FP interventions on the uptake of modern and dual methods, and the incidence of unmet FP needs and unintended pregnancies among PLHIV. This review presents the first meta-analysis on the topic to the best of our knowledge. This study demonstrates the odds of uptake of dual methods due to the intervention and its impact on the unmet FP needs and unintended pregnancies among WLHIV. The examination of the pooled OR using different approaches with weights determined by sample size and study design consistently yielded similar outcomes.

Fifteen out of 17 studies included in this review belonged to Africa. The interventions were either client, provider or facility oriented. The common interventions that gave significant results were training the service provider, along with client counselling and referral for FP services. Among the studies that assessed the uptake of modern contraception, 14 out of 16 studies reported an improvement. It was observed that uptake improved when condom was considered as a method of modern contraception.

Seven studies analyzed the uptake of dual methods, and six out of these reported an improvement with a pooled OR of 1.69 (P=0.008). Four studies assessed the unmet need for FP. Four studies assessed unintended pregnancies. The pooled OR for unmet needs was 0.58 (P<0.0001), suggestive of the protective effect of the intervention on reducing unmet FP needs. The pooled odds for unintended pregnancies also showed a protective trend due to the intervention. The Begg’s and Egger’s tests suggest the absence of publication bias.

Evidence in the literature shows six published systematic reviews that have a similar research question to ours8,9,10,11,12,13. A general conclusion from all these reviews was that integrating HIV-FP services was a feasible and cost-effective method to empower HIV clients in achieving their fertility desires. All these reviews highlighted a lack of evidence on dual methods and unintended pregnancies.

The systematic review published in 201710 mentions that as social desirability bias and inconsistent condom use could affect generalized outcomes like the uptake of modern contraception, a better alternative is to use a more specific measure as dual methods which have been particularly measured in this review. Various Indian studies have also identified an inadequate emphasis given to the promotion of dual methods among WLHIV18,51. The 2017 review also identified lacunae in evidence of unintended pregnancies and unmet FP needs. We have tried to address it by estimating the pooled odds of these outcome indicators when the intervention is implemented. We also assessed all studies using design-specific tools to improve the quality assessment, unlike a single tool used in the 2017 review. Furthermore evidence has been collated quantitatively using meta-analysis. Thorough literature searches ensured the inclusion of all relevant data. The inclusion of studies with lesser risk of bias in the meta-analysis has also helped in reducing the heterogeneity and publication bias, as reflected from the I² value, Begg’s and Egger’s tests and funnel plots.

Limitations: Several methodological and outcome parameters had limitations in the studies that were reviewed. The included studies differed in terms of design and on how frequently and detailed the interventions that were implemented. However, the outcome measurements were common in all the reviewed studies. These studies were from different countries, and hence, the sociodemographic characteristics of the participants were not uniform. This review encompassed 15 studies conducted in African countries. It is worth noting that this might constrain the extent to which the findings can be generalized. The high I² test value is reflective of this heterogeneity in the case of measuring the uptake of dual methods. Given this heterogeneity, it was difficult to conclude how dual versus single method use impacts unmet FP needs or unintended pregnancies.

Dependence of contraceptive use on women/couple behaviour and discrimination towards PLHIV in public health settings contribute to the subjectivity of these outcomes. The follow up period of such interventions always falls short as the pattern of contraception uptake changes throughout a person’s reproductive life span. Even though certain retrospective studies with long follow up were included, the general follow up period was not long enough to make any conclusive statements with certainty, especially on unintended pregnancies. We have presented the data using a difference of proportion. Although these measures do not adjust for confounders, they helped project a change in the pattern of the outcomes. Furthermore we were unable to measure a few other outcomes such as acquired infant HIV infection rate, consistent condom use, pregnancy termination rates and knowledge among WLHIV, as the search did not yield substantial results.

Overall this review demonstrates that integrating HIV-FP services is effective to improve the uptake of dual methods and reducing unmet FP needs. Preventing unintended pregnancies among WLHIV could avert the health burden on WLHIV and the presumed cost burden on the health system and families. Access to integrated services could facilitate in accomplishing the SDG target. Access to effective contraception is a matter of right for PLHIV as is for the general population. HIV programmes must invest in effective measures to prevent discrimination of this vulnerable population and the evidence from this review could help strengthen advocacy in this regard. However, it is equally essential to understand the cost-effectiveness of these interventions such that programmes can test their feasibility and adapt them.

Financial support and sponsorship

None.

Conflicts of interest

None.