Background

In 2019, more than 38 million people were living with HIV/AIDS worldwide [1]. Scaling up HIV antiretroviral therapy (ART) has since led to unprecedented achievements in improving the life quality and longevity of people living with HIV (PLHIV) and in reducing HIV transmission. PLHIV now have a life expectancy close to the general population when ART is initiated soon after diagnosis [2, 3]. The HIV Prevention Trials Network (HPTN 052), a multi-country study that aimed to determine the effects of ART on the transmission of HIV-1 from HIV-positive persons to serodiscordant partners, found that ART provides durable and reliable protection against HIV transmission [4]. These findings highlight the critical importance of “treatment as prevention” as a public health strategy to combat the HIV epidemic.

Retention in HIV care, defined as a person’s continued engagement in health services from enrollment to discharge/death [5], is critical to ensure access to ART, prevent HIV-associated complications, reduce mortality, and decrease community transmission of HIV. Individuals with interrupted HIV care are more at risk for opportunistic illnesses, such as Pneumocystis jirovecii pneumonia, than individuals who remained on ART [6]. Across different settings, PLHIV who experience a gap in their HIV care experience higher mortality rates than individuals who remain in HIV care [7, 8]. Evidence suggests that people diagnosed with HIV and who are not retained in medical care accounted for more than 60% of new HIV transmissions In contrast, individuals who are retained in care accounted for < 10% of new transmissions [9].

Despite the widespread availability of ART, critical implementation gaps persist in sustaining continuous engagement in HIV care. According to the latest Centers for Disease Control and Prevention (CDC) data, 42% of people diagnosed with HIV in the United States (US) are not retained in care, and 43% of new HIV transmissions are from individuals who are aware of their HIV infection but are not in care [10]. In a European cohort study, 50% of the patients had gone one year without a follow‐up visit [11]. A systematic review estimated 36-month retention at 65% in Africa [12]. Philips et al. found that irrespective of the data source or definition used for retention, the numbers of individuals retained at 24 months on ART varied between 41 and 72% globally [13]. These findings across settings highlight the critical relevance of evidence-based interventions to improve and sustain re-engagement into HIV care, as a large proportion of PLHIV will experience discontinuity of HIV care following HIV diagnosis.

Systematic reviews have been published on linkage to treatment, retention in care, and adherence [14, 15]. However, knowledge gaps persist for interventions targeting re-engagement into care for patients who had previously accessed services but later disengaged [15, 16] as published systematic reviews on re-engagement date from several years [16] or have been restricted to a specific geographical area (e.g., US) [15]. Identifying evidence-based interventions that can optimize re-engagement into care for PLHIV is necessary to achieve the second and third 95–95 UNAIDS targets by 2025 [17], which state that 95% of all people diagnosed with HIV infection will receive sustained ART and that 95% of people receiving ART will be virally suppressed.

We conducted a systematic review of interventions for re-engagement into HIV care to examine accumulated published evidence, identify similarities and differences across studies, provide an overview of the studies’ quality, and identify research gaps to inform future research.

Methods

Searches

The MEDLINE (PubMed), Embase, CINAHL, and PsycINFO databases were searched between January and March 2020. Keywords and database-specific terminology (e.g., MeSH) were used to capture the concepts of HIV and re-engagement (File S1). The search was restricted to papers published since 1996, and no restrictions were placed on the language of publication or publication status. We reviewed the reference lists of eligible studies to identify additional relevant studies.

Inclusion Criteria

Studies were eligible for inclusion if they evaluated an intervention aiming to improve re-engagement into HIV care, ART re-initiation, or viral suppression. We included randomized controlled trials (RCTs), cluster RCTs, interrupted time series, and other intervention studies that included a comparison group (historical or concurrent). We included studies with a study population of who had already been initiated on ART and were subsequently identified as out-of-care, lost to follow-up, or experienced interruption in their treatment. The definition of HIV interruption in care, out-of-care, and loss to follow-up varies widely in the scientific literature therefore we did not include one unique definition [7]. We excluded studies designed to improve linkage to HIV care or ART initiation for treatment naïve individuals. We also excluded studies that included HIV retesting as an intervention.

Study Selection and Data Abstraction

The final list of studies was de-duplicated electronically using covidence [18]. Two review authors (MCL and NB) completed the study selection; the appraisal process included reading the title and abstract to determine if the references needed further review (denoted by “yes,” “no,” or “unclear”). Two reviewers (MCL and NB) read the full text of studies meeting the screening criteria and classified each as “yes,” “no,” or “unclear” using a standardized template. References classified as “no” were excluded. For full texts categorized as “unclear” by at least one of the two reviewers, the reviewers reached a consensus to determine if the study should be included.

Using a standardized data extraction form, two review authors (MCL and NB) extracted the following data elements from the included studies: authors, year of publication, type of publication (abstract or manuscript), the country in which the study was conducted, aims, study design, study period, setting (clinic, emergency department, community, health department), study population, intervention group, control group, outcomes, sample size, follow up time, statistical analysis, data source(s), and study results. These two review authors resolved differences through discussion.

Outcomes

Re-engagement into care was the primary outcome, defined either as an encounter with a healthcare worker (phone call or in-person), ART re-initiation, or viral suppression.

Analysis

We did not perform a meta-analysis due to the methodological heterogeneity in the study designs, study populations, outcomes, and statistical analysis. We extracted the results and reported measures of effect, p-value, and 95% confidence intervals (CIs) when available.

Quality Assessment

We used the Newcastle–Ottawa scale [19], which is designed for non-randomized studies, to assess the quality of the studies because all included studies were non-randomized [20]. The scale consists of three areas: (1) selection of the study groups, (2) the comparability of the groups, and (3) the study outcomes.

Results

Literature Search

We identified 765 references for screening, including five identified by reviewing the included studies’ list of references. We removed 348 duplicates. We reviewed 417 titles and abstracts and selected 125 for full-text review. Among them, we found nine publications (manuscripts or abstracts) that met our inclusion criteria (Fig. 1). Publications were excluded due to study design (n = 47), study aims (n = 45), study outcomes (n = 10), additional duplicates (n = 9), study population (n = 4), or type of paper (commentary) (n = 1).

Fig. 1
figure 1

Study selection flow chart

Full size image

Study Characteristics

Table 1 summarizes the study characteristics, and the results and methods are presented in Tables 2 and 3. Study designs used for the included studies included six retrospective studies [21,22,23,24,25,26], one case-cohort [27] and one prospective cohort [28], and one before and after study design [29]. Seven studies were conducted in the US [21,22,23,24,25, 27, 28], one in Canada [29], and one in Ethiopia [26]. Except for one study [21] that did not report the study period, all studies were conducted from 2001 to 2017.

Table 1 Summary of characteristics of the included studies
Full size table
Table 2 Study outcomes and results of included studies
Full size table
Table 3 Quality scoring of included studies
Full size table

Participants

Studies used different definitions to describe the populations eligible for the intervention. Studies included PLHIV who had no contact with a clinical provider or no HIV-related lab results varying from two months to twenty-four months. Four studies included different definitions of out-of-care for the intervention and control groups [23, 24, 28, 29]. The number of participants in the intervention group varied from 20 to 4,212 and from 33 to 3,343 in the control group.

Settings

Participants included PLHIV who sought HIV services at a health center or clinic [21, 22, 24, 26], part of a provincial HIV program [29], or seeking care at the emergency department or ambulatory clinic, inpatient clinic [25, 28]. Five studies included a partnership between health centers/clinics and a health department or community health centers [22,23,24, 28, 29].

Intervention

Interventions could be divided into three main categories: (1) the integration of clinic and HIV surveillance data to identify, trace, and contact individuals; (2) the engagement of health workers (clinic and/or public health) to provide an additional or different level of support from the standard of care to assist the person in re-engaging into HIV care; or (3) multi-component intervention. We categorized studies based on the intervention elements, which differed between the intervention and control groups.

Integration of Clinic and HIV Surveillance Data Versus Using One Data Source (Clinic or HIV Surveillance)

Sachdev et al. reported using an HIV surveillance registry to identify sexual partners of PLHIV who were not in care. HIV-positive partners who did not have viral load results in the prior 12 months or who had a detectable viral load more than 6 months ago were referred for navigation services to encourage re-engagement into care [24].

Engagement of Health Workers (Additional or Different Level of Support) Versus Standard of Care

Abrams et al. reported an intervention that included a primary care provider scheduling and confirming an appointment with the disengaged patient [21]. Limited information is included for the standard of care received by the control group. For the study conducted in Ethiopia, the intervention included a national program of trained adherence supporters who traced and counseled PLHIV [25]. In Sharp et al. the intervention group was supported by a in-person social worker who assessed barriers to care and facilitated ART re-linkage to care with the clinic [25]. The control group received support from a health educator (not in-person). In that study, an automated alert was sent to the clinicians when a match occurred between a patient registration and the out-of-care list (surveillance data). This alert feature was used for both the control and intervention groups therefore the unique element of the intervention is the engagement of the social workers to support re-engagement into HIV care. In the Canadian study, a province-wide prescriber alert was generated for ART interruption with direct referral to public health offices rather than for the providers only (control group) [29].

Multi-Component Intervention

Four studies reported combining clinic and surveillance data to identify out-of-care patients in addition to the engagement of health providers assisted in tracing and contacting PLHIV [22, 23, 27, 28]. Bove et al. reported integrating clinic and surveillance data and a  linkage specialist to trace, contact and re-engage individuals into HIV care [22]. Udeagu et al. also integrated clinic and public health HIV surveillance data to identify out-of-care individuals, and a patient navigator contacted those individuals to schedule appointments [27]. In the study by Hart-Malloy et al. the intervention included a collaboration with community health providers in addition to integrating clinic and HIV surveillance data [23]. Magnus et al. reported the effects of a public health information exchange (public health and clinic data), which provides real-time alerts to providers when a person living with HIV out of care seeks non-HIV care [28].

Effects of Interventions

Studies reported on re-engagement into care or ART re-initiation [21,22,23,24,25,26,27,28,29], time to re-engagement into care [21, 22, 26, 27], and viral suppression [22, 25, 28]. For studies defining re-engagement into care or ART re-initiation (yes/no), follow-up time varied from 90 days to 22 months (Table 2). Table S2 provides data analysis details used on each included publication to evaluate these outcomes.

Integration of Clinic and HIV Surveillance Data Versus Using One Data Source (Clinic or HIV Surveillance)

Outcome: Re-Engagement into Care

Sachdev et al. compared re-linkage to care (no specific definition provided) between two identification mechanisms (1. municipal STD clinic and 2. syphilis and HIV partner services) [24]. The control group had a higher proportion of individuals who re-engaged into care; however, the authors did not specify if the results were statistically significant [24].

Engagement of Health Workers (Additional or Different Level of Support) Versus Standard of Care

Outcome: Re-Engagement into Care

Abrams et al. reported a significantly higher proportion of individuals who re-engage in the intervention group (80%) than the control group (52%) (p < 0.01) and had a shorter time to re-engagement (no values reported) [21]. In contrast, Kremer et al. reported no statistically significant differences in the proportion of individuals who re-linked to care between the control and comparison groups (no values provided) [29]. For Sharp et al., the risk of being re-linked to care did not differ significantly between the intervention and control groups (adjusted RR 1.63, 95% CI 0.99–2.68).

Outcome: ART re-initiation

Kremer et al. found different results based on the definition used for ART re-initiation [29]. They found no statistical differences between intervention and control groups for ART re-initiation within four months (values not reported). However, when using time-to-ART re-initiation among persons who re-initiated ART after four months, the authors reported a significant difference for time-to-ART initiation (p = 0.004) between the intervention and control groups. Teklu et al. reported that the intervention group (which received navigation services) was significantly more likely to re-initiate ART than the control group (adjusted HR = 2.84, 95% CI not reported) and had a shorter time to re-start their treatment (p < 0.001) [26].

Outcome: Viral Suppression

Both Kremer et al. and Sharp et al. reported no or minimal effect of the intervention [25, 29].

Multi-Component Intervention

Outcome: Re-Engagement into Care

Bove et al. showed a statistically significant effect (HR 1.7, 95% CI 1.2–2.3) for time to re-linkage at 12 months and re-linkage at any time of the study (22 months) (RR 1.6, 95% CI 1.2–2.1) [22]. Udeagu et al. showed a higher odd of being re-engaged into care for the intervention group compared to the control group (OR 2.3, 95% CI 1.23–4.36) and days from case assignment to first contact (p < 0.001) [27]. Hart-Malloy et al. combined surveillance with clinical data to identify out-of-care patients and then connected them to field workers for re-engagement [23]. The proportion of individuals who re-linked to HIV care did not significantly differ between the intervention and control groups (p = 0.13). Magnus et al. evaluated public health information to identify out-of-care patients and generate an alert for providers [28]. Magnus et al. did not report if the findings differed significantly between intervention and control groups [28].

Outcome: ART Initiation

Magnus et al. reported that the odds of being prescribed ART did not significantly differ between the intervention and control group [28]. For Udeagu et al., no values were reported for ART re-initiation [27].

Outcome: Viral Suppression

Among the four studies with multi-component intervention, three reported viral suppression as the primary outcome, and all reported no or minimal effect of the intervention [22, 27, 28].

Quality of Included Studies

The quality of the included studies was assessed using the Newcastle–Ottawa Quality Assessment Scales [19]. All included studies had a score ranging from 4 to 7 out of 8 points (Table 3).

Discussion

Re-engagement into HIV care is a key pillar of successful, longitudinal HIV care and treatment programs at the local and national levels. This systematic review highlights a lack of high-quality, standardized definitions of re-engagement and should galvanize researchers in the field to improve research methods and strategies for this critical component of HIV care. Nine studies met the inclusion criteria for this systematic review on interventions to improve re-engagement into care among out-of-care PLHIV. For the included studies, the main intervention centered on either (1) integration of clinic and HIV surveillance data; (2) additional or different level of support provided by healthcare workers (clinic and/or public health staff); or 3) combination of interventions. For re-engagement into care, three studies reported statistically significant positive findings [21, 22, 27] three studies found no difference [23, 25, 29], and two studies did not specify if results differed between the intervention and control groups [24, 28].For ART initiation, one study found a significant difference between intervention and control groups [26], one study reported mixed results depending on the definition of ART initiation [29], one study reported no difference between the intervention and control groups [28], and one study did not specify if results differed [27]. None of the five studies reported a significant difference between the intervention and control groups for viral suppression [22,23,24,25, 28, 29].

The ultimate goals of HIV treatment are to achieve viral suppression, prevent adverse health outcomes for individual patients, and prevent HIV transmission in the community. None of the studies that included viral suppression as an outcome showed a significant difference between the intervention and control groups. This finding suggests that additional strategies are needed to optimize and sustain re-engagement into care. Client-centered strategies to sustain care continuity and achieve viral suppression are warranted. Barriers interfering with their continuity of care, such as access and availability of quality HIV services, social support, social and cultural norms, substance abuse, and stigma, among others, should also be addressed to improve continuous engagement [30,31,32].

We assessed the quality of the included studies using the Newcastle–Ottawa Quality Assessment Scale, used for non-randomized studies [19]. All studies scored between four and seven out of eight. In many studies, the choice of control groups was a concern as they often differed from the intervention groups in terms of characteristics. For example, Magnus et al. used in-care patients who had experienced discontinuity in care at some point along with their HIV care as their control group [28]. One of their outcomes of interest included viral suppression. In the control group, which was comprised of patients in care, patients were more likely to be virally suppressed than out-of-care patients, limiting the ability to compare both groups accurately.

Also, studies varied greatly in statistical analysis types, and potential confounders were not systematically controlled across the different studies (Table S2). Despite this low-to-moderate-quality score, other quality issues were identified, which were not captured in the Newcastle–Ottawa Scale. Seven studies did not fully report all outcomes, and three did not specify the statistical analysis used [21, 23, 24, 26,27,28,29]. We also found two abstracts that contained limited information on the study population, outcomes, statistical analysis, and results [21, 24]. In addition, similarities between the intervention and standard of care may have limited the effects of the intervention. In some studies, the standard of care or services provided for the control group was either not clearly presented or differed slightly from the intervention group [21,22,23, 25]. For example, in one study, the intervention group received an intervention by a social worker while some individuals in the control group received an intervention by a health educator [25]. Both groups may have received comparable services, but these services were offered via different healthcare cadres, which may have contributed to the minimal difference between both groups.

Out-of-care individuals and re-engagement in care were defined differently across all included studies, which complicates the ability to compare the interventions’ effectiveness. This heterogeneity highlights the need to use or develop standard definitions when possible to avoid misclassification of individuals and outcomes. For retention and loss to follow-up, different definitions have been used or proposed [5, 33,34,35]. However, the included studies did not use available standard definitions. For re-engagement, gaps persist in the literature to define metrics for re-engagement into care. Half of the studies included viral suppression as an outcome, which is the ultimate goal from a clinical and public health perspective.

Strengths and limitations

We conducted a broad search of the literature, irrespective of language, and included manuscripts, conference proceedings, and grey literature. We reviewed over 417 titles; most of the studies focused on factors associated with re-engagement into care rather than the evaluation of interventions. We were unable to conduct a meta-analysis due to methodological differences across studies. Eight out of the nine studies were conducted in North America. This finding points to a major knowledge gap as nearly 70% of PLHIV live in sub-Saharan Africa (SSA) [36]. All studies using electronic medical records (EMRs) or surveillance data were conducted in the United States, even though EMR systems are increasingly being implemented to assist with delivering quality health care in SSA [37]. Future studies are warranted to examine the effectiveness of EMR systems and applications in optimizing continuity of care in different settings in combination with other strategies. Furthermore, to identify evidence-based interventions for re-engagement into HIV care, rigorously designed studies are needed, such as cluster randomized control trials, step-wedge studies, or before-and-after studies with a concurrent control group.

In conclusion, this review highlights significant gaps in the current understanding of the evidence-based interventions to re-engage patients living with HIV who are lost to care. Better and consistent definitions of control and intervention groups, interventions used, and outcomes analyzed are clearly needed. More work is urgently required in SSA, which carries a disproportionate burden of HIV, and where the success of local and national HIV programs clearly hinges. Finally, novel strategies and interventions for re-engaging PLHIV into care will also be required to raise the quality and ultimate success of HIV care and treatment programs.