Disease intervention specialists (DIS) deliver partner services programs encompassing a wide range of activities for HIV and sexually transmitted infections (STIs) including field testing and treatment, expedited partner therapy, case treatment follow-up, partner notification, counseling, and referral services.^1,2 They collect crucial data for disease surveillance. Additionally, they played a pivotal role in the COVID-19 response.³ Enhancing the DIS workforce is part of the federal strategy to strengthen outbreak response capacity.⁴

Past reviews have examined different aspects of DIS-delivered interventions for HIV and STIs.^5–9 Other reviews focused on new technologies in HIV/STI partner services.^10,11 Overall, there is evidence for a positive impact of DIS-delivered interventions. An updated systematic review is warranted considering evolutions in partner services, the DIS workforce, syndemic approaches, and epidemiological characteristics that influence outcomes such as test positivity among partners.

This review addressed the following questions: What is the program impact of partner services and other DIS-delivered interventions for HIV and STIs on clinical and epidemiological outcomes among index patients and their partners in the U.S.? How do outcomes vary across infections? What are the critical evidence gaps? For consistency with impact evaluations employing experimental or non-experimental designs such as pre-post or post-test comparisons between groups to assess program effectiveness,^12,13 this review considered comparisons to no intervention, alternative interventions not delivered by DIS, or different modes of delivery.

This systematic literature review used a narrative synthesis of findings following National Academy of Medicine and Patient Centered Outcomes Research Institute (PCORI) methods^14,15 and the Preferred Reporting Systems for Systematic Reviews and Meta-analyses (PRISMA) reporting guidelines.¹⁶ The protocol was registered with PROSPERO (CRD42022375961).¹⁷ The study did not require University at Albany institutional review board review because it was a literature review. Although the protocol encompassed DIS-delivered services for HIV, STIs, and tuberculosis, tuberculosis studies were excluded because analyses of the effectiveness of tuberculosis contact tracing were too dissimilar from HIV and STIs to allow for meaningful comparisons, and only 4 tuberculosis studies were found.^18–21 A second protocol deviation was excluding conference abstracts because they provided insufficient information to extract study characteristics and assess the strength of evidence.

The keyword search was implemented in PubMed/MEDLINE, CINAHL, ProQuest Dissertations & Theses Global, and Web of Science. The medRxiv preprint server was not included because it was infeasible to implement a similar keyword search with terms and Boolean operators; additionally, studies were not peer-reviewed. The keywords were developed based on the study team’s subject matter expertise, pilot searching, and consultation with 2 reference librarians. After the initial search, a second search was conducted with additional keywords for “contact investigation” (a synonym for contact tracing) and “venue-based testing” (a DIS-delivered intervention). The keyword search was completed in December 2022 and supplemented by a hand search of references in included studies and relevant review articles. Analyses including the hand search were conducted during 2023−2024. The Appendix contains additional details.

Population, intervention, comparator, outcomes, timing, and setting (PICOTS) screening criteria were used by authors EM, AM, HK, PS, and SJ. Additionally, included studies met a criterion of primary analysis of empirical data, excluding literature reviews and modeling studies that did not present new empirical data to derive model parameters. The population criterion included persons recently diagnosed with or at risk for HIV or non-viral STIs and their partners. Medical and community-based providers and program staff were excluded. The intervention criterion covered partner services and other DIS-delivered interventions; specific activities included interviewing index patients to elicit partner information, partner notification, screening, vaccination against related infections, prevention counseling, presumptive treatment, expedited partner therapy, linkage or referral to medical care and other services, direct delivery or referrals for HIV pre-exposure prophylaxis, and venue-based screening. Interventions not delivered by DIS were excluded; for example, expedited partner therapy was only included if DIS-delivered. Data-to-care interventions for persons with HIV presumed to be out of care and social network interventions were excluded because they target individuals previously in care. The comparator criterion required a comparison of outcomes between populations that did and did not receive partner services or a pre-post comparison of outcomes before and after receiving partner services, including studies that compared different program delivery modalities. Studies with a comparator but no statistical test of significance were excluded. In terms of outcomes, all outcomes for index patients, partners, and the community along the continuum of clinical evaluation, counseling, diagnostic testing, and treatment were considered. Receipt of partner services was excluded as an outcome. Regarding timing, all studies were considered. The setting criterion was the United States due to variation in interventions and terminology in non-U.S. settings.

All studies identified in the keyword search were reviewed manually for eligibility in 2 phases. First, the titles and abstracts were screened based on the primary research and PICOTS criteria using hierarchical criteria (ordering: setting, primary research, intervention, comparator, population, and outcomes). At this stage, screeners (AM, HK, and PS) were flexible in progressing studies to the second stage if exclusion was unclear. Second, full-text articles were screened. Following the PCORI guidance that relaxes standards related to dual screening and data abstraction if fact-checking and quality control procedures are in place,¹⁵ the first stage screening was conducted by 1 author. Screeners were trained by double-coding at least 30 studies with the first author and flagged articles with unclear determinations for review by a second screener. In the first stage, 534 (33.3%) of the 1,605 journal articles were reviewed by a second screener. The lead author reviewed all ProQuest studies that were not removed by screeners for non-US setting or lacked relevance. In the second stage full-text review, all 138 studies were screened by 2 authors. Where decisions were unclear, one or more additional authors reviewed to resolve via consensus. For additional studies identified in the hand search of references of included articles from the keyword search, 2 screeners (AM, HK) reviewed references independently and verified each other’s decisions.

Reviewers (AM, HK, PS, and SJ) manually extracted the following into a standardized, pre-piloted form (see the Appendix): stated aims, intervention description, infections examined, study design, statistical analysis, comparator, data source and coverage, location, population, study sample, outcome measures, key findings, and reported limitations. One reviewer extracted information, with a second verifying the extraction. Questions were resolved through consensus and the lead author when needed. Additionally, all summaries were reviewed by the senior author. Only findings reporting a statistical test of significance (p-values or confidence intervals) were extracted. Anticipating heterogeneity across studies, no restrictions were placed on effect measures (e.g., risk ratios or mean differences). If there were multiple analyses, the highest-order analysis was extracted (e.g., multivariable regression results were extracted rather than bivariate results). Missing or unclear information was not confirmed with study investigators because it was infeasible to contact all authors.

Findings were summarized by distilling study characteristics and outcomes from the data extraction sheets into summary tables. A narrative synthesis summarized findings by infection, specific interventions, and components of the disease intervention cascade (interview, testing, treatment, and risk behaviors for index patients; partner identification, notification, testing, and treatment; and population-level epidemiological outcomes). Due to the diverse outcome measures, interventions, and infections, it was infeasible to perform a meta-analysis or conduct other quantitative syntheses.

For each study, the strength of evidence on program effectiveness was determined by considering holistically the JBI critical appraisal checklists,²² study design, and key limitations. The appraisal was specific to studies with a research design suitable for an impact evaluation. One reviewer completed the checklists, writing detailed comments assigning an overall strength of evidence score of “high,” “medium,” or “low” based on the checklist scores, study design, and study limitations (e.g., small sample sizes or nonequivalent groups with insufficient adjustment in multivariable analysis). A second reviewer verified scores and wrote additional comments regarding their own interpretations. All disagreements were resolved through consensus and discussion with the lead author (EM) when needed. See the Appendix for additional details.

It was infeasible to conduct a formal assessment of the certainty in the body of evidence for the effectiveness of partner services and other DIS-delivered interventions due to the variability of studies with respect to specific interventions, infections included, and study design. Instead, the key limitations and strength of evidence were described narratively.

The keyword search yielded 1,915 unique records after removing duplicates, of which 1,777 were removed after screening titles and abstracts, 106 were removed after full-text review, and 4 were removed due to a focus on tuberculosis.^18–21 The 28 included studies were supplemented with 2 studies based on hand searching references, yielding 30 studies (see Figure 1). A common study exclusion was an insufficient comparator. In the first stage (title and abstract review), among the 423 studies that met the first 3 hierarchical criteria of primary research, setting, and intervention, 201 (47.5%) were removed due to an unsuitable comparator. Among the 94 studies assessed in the full-text review that met the first 3 hierarchical criteria, 61 (64.9%) were removed due to an unsuitable comparator.

Table 1 summarizes study characteristics. Eleven studies focused on HIV exclusively,^23–33 15 focused on non-viral STIs exclusively,^34–48 and 4 included both HIV and bacterial STIs.^49–52 These studies examined contact tracing for HIV and syphilis,⁴⁹ increasing the intensity of partner notification and cluster investigation for syphilis,^35,37 partner notification or partner services,^{23–25,27–29,36,40,48} DIS-delivered expedited partner therapy for non-viral STIs,^38,44 field-delivered therapy for gonorrhea and chlamydia,⁴⁵ on-site placement of DIS for HIV³¹ or STIs,^46,47 and modifications to the delivery of HIV/STI partner services such as telehealth or different interview techniques.^{26,30,32–34,39,41–43,50–52} The most common study design was an external control group,^{23–26,29,33,36,40,42,49,50,52} followed by pre-post designs without external control groups,^{30,35,37,45–47,51} RCTs,^{28,34,38,41,44} pre-post designs with external control groups,^27,31,32,39 and external control group via pre-post designs.^43,48 Studies were distributed regionally across the West,^{24,26,27,34,38,41,42,45,46,48,50,51} Northeast,^{23,25,29–33,36,39,47} and South;^{28,35,37,43,44,49,52} only 1 study was in the Midwest.⁴⁰ Most studies were in urban locations, with the exception of 2 statewide studies^49,51 and 3 studies assessing outcomes in multiple local health departments or counties.^28,36,38 Two studies were linked, providing an analysis of the same intervention (intensified partner notification and cluster investigation for syphilis) using an identical study design and data source.^35,37

Health outcomes among index patients were considerably more favorable with HIV/STI partner services than without. Persons with HIV reached by DIS had higher linkage to care and established care,²³ HIV viral suppression,⁴⁹ and condom use with notified partners.²⁷ Compared to a reminder system, field follow-up yielded higher return rates for chlamydia treatment.⁴⁰ Expanding STI partner services to all men who have sex with men (MSM) with bacterial STIs and testing their sex partners for HIV prior to case closure yielded increased HIV testing and new HIV diagnoses.⁵¹

Partner outcomes were also more favorable under HIV/STI partner services. Patients receiving HIV partner services were more likely to be linked to HIV care²⁶ and notify partners,²⁴ DIS elicited more partners per HIV index patient compared to community clinicians,²⁹ and HIV partner notification was higher when completed by public health counselors versus by patients.²⁸ Incorporating field testing into HIV partner services resulted in more notified partners testing for HIV.³⁰ One state health department’s campaign to reduce syphilis yielded more persons treated prophylactically per index patient.^35,37 One study had mixed results: among men with nongonococcal urethritis who received nurse referral, field follow-up, or DIS interviews without field follow-up, nurse referral counseling elicited the highest number of partners per index patient but field follow-up yielded more treated female sex partners per index patient.⁴⁰

Studies examining population-level epidemiological outcomes found reductions in reported rates of gonorrhea. There were fewer repeat gonorrhea cases and incident diagnoses among civilians when gonorrhea patients treated at a military base received interviews and contact tracing under the direction of a health department representative,⁴⁸ counties with a higher percent of partners brought to preventive treatment had lower reported gonorrhea rates,³⁶ and reported gonorrhea rates were lower when health department staff reprioritized patients to prioritize high-morbidity geographical areas.³⁹

There was mixed evidence for DIS-delivered expedited partner therapy and field-delivered therapy. There was an increase in chlamydia and gonorrhea index patients treated⁴⁵ but no differences in their reported condom usage.⁴⁴ The 2 studies examining partner outcomes had divergent findings: one reported increased patient-delivered therapy and decreased chlamydia test positivity and gonorrhea diagnoses,³⁸ whereas the other reported no difference in time between female index patients’ initial visits until treatment of their male partners, no reduction in repeat trichomoniasis rates, and a lower percentage of male partners with verified treatment in the DIS group compared to the patient-delivered partner therapy group.⁴⁴ The divergent findings are potentially attributable to differences in the strength of evidence (with improvements reported by the study with more suitable evidence for causal inference⁴⁴) or different infections.

Most partner services outcomes improved after placing DIS in HIV clinics,⁴⁶ an STI clinic,⁴⁷ and healthcare facilities.³¹ However, the specific benefits were mixed. Although index patients were more likely to be interviewed^46,47 and diagnosed earlier,⁴⁶ there were no observed differences in index patient treatment.⁴⁶ Two studies found mostly favorable outcomes among partners (such as partners elicited, notified, and treated)^31,46 while another study reported no differences in partner-related outcomes.⁴⁷ These divergent outcomes are not readily attributable to differences in suitability of causal evidence, infection, geography, or time period.

Two studies examined alternative interview techniques, finding that adding cues to aid recall yielded more partners elicited and located³⁴ but that social network analysis interviews did not yield different partner elicitation or other partner outreach outcomes compared to traditional partner notification interviews.⁴³ Studies comparing different delivery modes found that telephone partner services for HIV and syphilis had worse partner outcomes along the continuum compared to inperson partner services.^32,50 Evidence on internet and text message partner services was mixed. One study compared both modes to standard HIV partner service, finding that success along the partner services continuum varied by notification type; the likelihood of contacting partners was highest for text messaging partner services, notifying contacted partners was highest for internet partner services, and testing notified partners for HIV was highest for traditional partner services.³³ In an RCT, MSM receiving web-based partner notification services had fewer partners tested, although there was low power to detect differences due to the small sample.⁴¹ MSM enrolled in text message reminder services did not have different subsequent asymptomatic STI diagnosis rates compared to MSM without reminders.⁴² Compared to traditional HIV/STI partner services, email-based partner notification for pseudonymous sex partners for which email was the only available contact information resulted in fewer sex partners notified and tested.⁵² However, it is difficult to conclude definitely that text message reminders or email-based partner notification are ineffective because the text message reminder groups opted into the intervention and many “control” participants already used reminders,⁴² and the study of pseudonymous sex partners had a low sample size and did not adjust for differences between groups.⁵²

Overall, the strength of the body of evidence on program effectiveness was low with respect to the study designs’ suitability for assessing the causal impact of partner services and other DIS-delivered interventions on clinical and epidemiological outcomes. One study had a strength of evidence rating of “high”; most studies were rated as “low” (n=18, 60%) or “medium” (n=11, 36.7%). (See Table 2 for the strength of evidence ratings and the Appendix for the JBI critical appraisal checklists.)

In addition to specific issues affecting the strength of evidence described in the preceding narrative description of study findings, there were few RCTs, which provide the highest level of causal evidence. Specific weaknesses in the RCTs were suspected errors in the reported results,^34,44 a low sample size that led to a study’s early termination,⁴¹ and insufficient information for a complete assessment.²⁸ Many studies using external control groups had non-equivalence of treatment groups and potential selection bias, such as comparing outcomes among those who accepted the intervention to those who did not, without appropriate statistical adjustment for individual-level characteristics or other confounders. Most studies with external control groups only examined “post” data, rather than collecting before-and-after data in both groups for a differences-in-differences design. In many instances, there was missing or unclear information about the study designs such as descriptions of the interventions delivered, potential spillover effects, participant attrition, selection biases, or sufficient details about the statistical analyses performed to allow for a comprehensive understanding of the approach and replication.

Overall, this systematic review identified evidence across multiple studies and settings for the favorable impact of HIV/STI partner services on the continuum of index patient and partner outcomes, and on reported gonorrhea incidence. DIS-delivered expedited partner therapy and field-delivered therapy improved treatment of STI index patients but there was mixed evidence for partner-related outcomes. Most partner services outcomes improved after placing DIS in clinics. Compared to in-person delivery, telephone partner services yielded worse partner outcomes; however, there is insufficient evidence regarding internet-based or text message partner services.

The first evidence gap is the suitability of the evidence for assessing the causal impact of DIS-delivered interventions. Although there is a large literature on these interventions, many studies were excluded from this review because their designs were not suitable for impact evaluation, such as descriptions of program participants and program outcomes,^53–55 comparisons of partner services outcomes between participants with different demographic characteristics,^56–58 or not having a statistical test to assess the significance of comparison.^59–61 Based on a holistic assessment of the study design, the critical appraisal checklists, and key limitations, almost all studies were rated as providing low or medium strength evidence. In the full-text review, approximately two-thirds of articles meeting the primary research, setting, and intervention criteria were excluded due to the lack of a suitable comparator for an impact evaluation. Many articles described the outcomes of specific outbreaks (e.g., number of index patients treated; number of contacts/partners identified, tested, and treated; and test positivity among contacts/partners) without a comparison group to assess how results would have differed in the absence of DIS-delivered interventions. Few studies used an RCT design; as noted by Kerani et al.,⁴¹ the difficulty of recruiting subjects limits the feasibility of implementing this study design. Another common limitation was not adjusting for group differences to ensure suitability for comparing treatment and control groups; for example, several studies provided unadjusted bivariate comparisons between groups despite nonequivalence of treatment arms. Similarly, studies with pre-post designs and external control groups did not employ methods that allow for strong causal inference.^27,31,32,39 Another common methodological issue was the primary focus on program process measures such as testing and partners elicited; although these are important for understanding the indirect impact of partner services, only 5 studies examined outcomes such as future diagnoses and repeat infections.^{36,38,39,44,48}

A second evidence gap is limited geographic representation. New York City and Seattle/King County were over-represented, corresponding to well-established investigators and study teams that frequently publish in this subject area. Notably, there was only one study from Texas⁵² and no studies from California or Florida, despite those states having a disproportionately high number of incident diagnoses of HIV and STIs.^62,63 Evidence from additional locations, particularly from local health departments and rural settings that may have lower levels of staff and other resources, would allow for assessment of the extent to which findings are generalizable. It would improve the ability to assess whether the impact on clinical and epidemiological outcomes differs by setting.

As partner services programs evolve, there are several evidence gaps to prioritize. First, updating the scientific evidence with more recent data, research designs with higher internal validity, and improved adjustment for non-equivalent treatment groups and selection bias are important given changes in the epidemics that can influence outcomes such as reinfection and diagnosis rates and to support future modeling studies. Quantifying these outcomes is essential. Although RCTs can offer the highest level of evidence, they are often unrealistic in the context of limited funding for public health systems and services research⁶⁴ and STI research,⁶⁵ the length of time to execute studies, and pragmatic considerations. Alternative approaches include quasi-experimental designs such as synthetic matched controls or differences-in-differences using surveillance and programmatic data, or stepped-wedge designs whereby updated interventions are phased into different facilities or localities over time. Second, evidence on effectiveness could be supplemented with evidence on implementation strategies using research designs such as hybrid implementation-effectiveness trials. Delays in the uptake of health interventions are well-documented^66,67; additional contextual challenges for HIV/STI prevention include chronic public health workforce shortages,⁶⁸ critical gaps in public health infrastructure,⁶⁹ and the backlash against public health,⁷⁰ compounded by rising STI rates.⁵⁷ As such, implementation science is a critical addition to impact evaluation to ensure that effective DIS-delivered interventions are integrated into practice within this broader context. Third, the literature would benefit from cross-disease approaches to assessing impact following the Centers for Disease Control and Prevention’s national strategic plan that emphasizes syndemic approaches to increase collaboration and service integration across infectious diseases.⁷¹ Although DIS provide services for multiple infections including tuberculosis and COVID-19, all included studies focused exclusively on HIV and/or STIs; the effectiveness of DIS intervention on other infections remain separate literatures. Only 4 studies included both HIV and STI outcomes; this is a missed opportunity given efforts to integrate services. A fourth area for future research is evaluating new partner services strategies, such as the extent to which molecular epidemiology for HIV/STI partner services improved clinical and population outcomes (versus the current literature which primarily uses the molecular data for descriptive epidemiological analyses), the role of new data systems following enhancements under the Data Modernization Initiative,⁷² new innovations developed during the COVID-19 pandemic such as enhanced use of telehealth services, and new strategies implemented as a result of the DIS Workforce Development Funding supplement.⁷³

DIS are a critical workforce for a reimagined public health system that can respond quickly to emerging infections.⁷⁵ The findings on improved partner services outcomes after placing DIS in clinics and better outcomes with in-person versus telephone-based partner services suggests that although virtual program delivery can be beneficial particularly in rural areas, Ending the HIV Epidemic in the US jurisdictions, and local regions with provider deserts, technology may not always be an adequate substitute for in-person DIS involvement. As the role of DIS expands and partner services programs evolve to integrate services across disease siloes and incorporate new surveillance system technologies, program evaluations will be critical to demonstrate the value of these public health investments and identify areas for improvement. Opportunities to strengthen the evidence base include establishing a streamlined set of core metrics, implementing more rigorous causal inference methodologies to assess impact, linking data systems to integrate program and clinical outcomes data, and supplementing impact evaluations with evidence on implementation strategies.