Describing Engagement in the HIV Care Cascade: A Methodological Study

Background

To curb the HIV epidemic, UNAIDS committed to helping countries and regions meet the 90–90-90 targets.¹ By 2020, UNAIDS aimed for 90% of people living with HIV (PLWH) to know their HIV status, for 90% of those diagnosed HIV-positive to receive antiretroviral therapy (ART), and for 90% of patients receiving ART to be in viral suppression.¹ However, by 2020, only 73% of the 37.7 million people globally living with HIV have access to ART, only 79% of PLWH know their HIV status, and 20% of PLWH have yet to receive ART.²

Barriers to testing, staying in care, and starting or adhering to antiretroviral treatment exist and contribute to the loss of individuals at different stages of the HIV care cascade, preventing the fulfillment of the 90–90-90 targets.¹ As such, only a proportion of people living with HIV are engaged with all the steps required to achieve an undetectable viral load.³ Furthermore, according to UNAIDS, despite 79% of PLWH knowing their status, more than 20% of people diagnosed had not yet initiated treatment in 2018.² This is a concerning proportion because it shows that there are barriers to engagement in care that prevent PLWH from achieving optimal outcomes.²

A significant challenge in assessing the efficacy of public health programs and achieving the now expanded 95–95-95 targets is the lack of a standardized definition for engagement in care among PLWH.⁴ Varying definitions of engagement exist, which contributes to heterogeneity in research regarding how many individuals are actively participating and benefitting from HIV care.³ The overestimation of disengagement or the oversimplification of the complex cycle of entry and re-entry into care contribute to the misclassifications of engagement.^4–9 The implications of misclassifying engagement include failing to detect immunosuppression, inaccurately prioritizing interventions targeting viral suppression and retention in care, and failing to recognize disengagement from individuals who remain at a care facility.¹⁰ Furthermore, engagement often implies a moral framework of what constitutes a “good” patient compared to a “bad” patient.¹¹ Misunderstandings regarding engagement and a lack of agreement on what qualifies a patient as “engaged” in HIV care can contribute to bias in providers, which can be an additional barrier to care.

A lack of a standard definition of engagement precludes comparison of findings across studies and compromises researchers’ ability to pool data. Thus, defining engagement in care has the potential to reduce barriers to engagement, including inaccurate assessments of engagement, incorrect prioritization of interventions, and bias associated with “good” and “bad” patients. Furthermore, engagement requires more than a single measure, making it essential to evaluate each element included in the definitions of engagement found in the literature.^12–15 The objective of this methodological study was to identify and describe the various definitions for engagement in care used in HIV clinical trials.

Methods

We conducted a methodological study of randomized controlled trials (RCTs) with engagement as one of their outcomes. Studies were drawn from the database CASCADE (https://hivcarecascade.com/). CASCADE is a repository of clinical trials of interventions to improve the HIV care cascade, which includes trials aimed at investigating diagnosis and initiation of treatment, adherence to antiretroviral therapy, and retention in care.^1–4,16 These clinical trials have been curated by income level, vulnerable population, who delivered the intervention, the setting in which it was delivered, the intervention type, and the level of pragmatism of the intervention. The search strategies and data management used to build the database are reported elsewhere but include searches of multiple electronic databases and duplicate data extraction.¹⁷ CASCADE, as a result of a comprehensive search of PubMed, Excerpta Medica dataBASE, Cumulative Index to Nursing and Allied Health Literature, PsycINFO, Web of Science and the Cochrane Library, includes 298 trials published between 1995 and 2021.

Using the advanced search filter, we identified trials in which engagement was listed as an outcome. The identified trials were further categorized into those that explicitly defined engagement and those that did not. Explicit definitions of engagement were identified through specifications such as, “Engagement in care was defined as…” Those that did mention an explicit definition were further sorted into 6 categories (appointment adherence, laboratory testing, ART adherence, time period specification, intervention adherence, and quality of interaction) by two authors based on the content of the definitions. Trials that had multiple definitions of engagement were also included in these 6 categories.

Results

Of the 298 studies in the database, 13 studies were eligible, of which five did not provide any explicit definition for engagement. The remaining 8 defined engagement using: appointment adherence (n=6), laboratory testing (n=2), ART adherence (n=2), time period specification (n=5), intervention adherence (n=5), and quality of interaction (n=1). A flow chart of the included studies is shown in Figure 1 and the characteristics of the study are shown in Table 1.

Table 1 Characteristics of the Studies Found from the CASCADE Database

Figure 1 Flowchart detailing the data screening and extraction process from the CASCADE database.

Of the 8 studies with explicit definitions of engagement, six defined engagement in care as appointment adherence, which was the most common definition.^18–23 Five studies specified periods during which the appointments must occur for patients to be considered engaged. However, the frequency of visits and time period in the patient’s medical history that were examined differed greatly. One study examined appointment adherence over a period of 3 months;²⁰ two examined adherence over a period of 6 months;^18,21 and another study examined adherence over a period of 12 months.¹⁹ One study examined appointment adherence by identifying if appointments were scheduled and attended within 60 days of each other.²² One study did not specify a time frame.²³

Two studies included laboratory testing in their definitions of engagement.^24,25 These studies either measured the frequency of viral load tests or CD4 counts during clinic visits or whether ≥ 1 laboratory test results were recorded within the past 12 months.^24,25

Two studies used ART adherence either as the whole definition of engagement or as a component that was included in the definition.^22,24 Another study identified engagement based on how often participants included the intervention being tested in their daily lives.²⁶ The remaining studies had no explicit definition for engagement in HIV care.^27–30

The most comprehensive definitions of engagement in the literature were studies that included four of the elements described in Figure 2. One study considered engagement to be a combination of appointment adherence, lab testing, ART adherence, and self-reported engagement measures.²⁴ A second study considered engagement to be a combination of appointment adherence, ART adherence, within a specific period, and self-reported engagement measures (ART adherence).²²

Figure 2 Definitions of engagement in each clinical trial. This bar graph summarizes definitions engagement versus the number of studies in which those definitions were used. *Self-reported measures included ART adherence, clinic attendance, viral testing frequency. **Quality of interaction with provider was examined using a using a 13-item questionnaire to rate patient interactions with their healthcare provider on a 4-point Likert scale.

Discussion

We identified 8 randomized trials using diverse definitions of engagement including appointment adherence, laboratory testing, ART adherence, time period specification, intervention adherence, and quality of interaction with the health care provider.

At first glance, appointment adherence seems to be an ideal measure of engagement; it is easy to measure, offers a more quantitative measure of engagement, and from a clinician and research standpoint, it offers little room for bias since it is not self-reported.¹⁴ There has also been evidence of a strong dose-response relationship between clinic visits and reduced mortality, suggesting that identifying and measuring clinic visits as a form of engagement in care may be valuable.³¹ However, there are other important considerations.

Appointment adherence is a complex measure as it can also be further categorized according to the number of missed visits or the length of time between visits. Missed visits are easy to measure, but require that visits be scheduled.¹⁴ The length of time between visits is also easy to measure and can be used to accurately estimate and determine loss to follow-up.¹⁴ However, the number of visits or length of time that is considered as disengagement is highly variable.¹⁴ When estimating appointment adherence, the number of visits required may vary based on patient needs and acuity,¹⁴ suggesting that the number of visits alone may be insufficient in capturing engagement. A more nuanced definition incorporating barriers to care, including stigma, accessibility concerns, distance to providers, and health literacy and knowledge is warranted.^18,19 Comparability between studies using appointment adherence is also limited due to variations in study protocols, methodologies, and appointment lengths and frequency.

The specification of a time frame in which to examine engagement standardizes how long patients have until they are considered disengaged. Despite this being the second most frequently reported definition, there is a lack of literature to support a suggested time period in which to measure whether a patient is engaged or not after diagnosis. In the studies examined in this paper, the time period specified ranged from the past 3 months to 12 months. Therefore, it seems that measuring engagement within the past year is an appropriate time period. However, it is important to note that people living with HIV may have different needs that are relevant at various time points. For example, the clinical visits may not overlap with the lab visits or visits for psychosocial support. Further, within the context of research, for reasons of feasibility or availability of resources, engagement may only be captured for brief periods.

Using the frequency of lab testing as a measure for engagement is objective and does not require patient interviews. One prospective, observational, cohort study (n=2909) examined HIV laboratory testing events as a proxy for engagement in care by analyzing laboratory tests associated with medical encounters.³² They concluded that laboratory testing had a high positive predictive value for actual engagement, which was defined as having at least 2 medical encounters within 12 months that were at least 90 days apart.³² This measure is not without its limitations. It requires that testing be available and accessible. However, because lab testing is associated with active engagement in care, it can be prioritized as an essential part of engagement.

ART is rarely considered the sole aspect of the definition for engagement, as there are several barriers to ART adherence, including patient-provider trust, systemic stigma and discrimination, and a lack of social support.³³ Furthermore, using ART adherence as a definition for engagement blurs the lines between engagement in care and adherence to treatment. Yet, adherence to medication is a critical part of engagement and is highly correlated with viral suppression.³⁴ Another study found that participants that were prescribed ART medications were more likely to attend scheduled visits, suggesting a relationship between ART adherence and other forms of engagement in care.³⁵

Only one study recommended adherence to the RCT intervention as a measure for engagement. Adherence to a new or foreign intervention in care may not be an accurate reflection of an individual’s engagement in care. This is because implementing a new intervention in their care involves new factors and different barriers; for instance, if the intervention was a novel engagement strategy through telephone reminders, and a patient rarely uses their cell phone or has limited use due to work, failure to engage with this intervention may not mean that the patient is not engaged in care. This definition may only have value in research settings or for evaluating the feasibility of a program.

Self-reported engagement has several advantages, as it encompasses all the components to consider regarding measures of engagement, including stigma, access to clinics or labs, health literacy, etc., because it can individualize engagement through self-reflection. However, in theory, self-reported engagement allows for recall bias, social desirability bias and is difficult to standardize. There can be efforts to mitigate this bias. One study in the UK found that there was a high level of accuracy of self-reported lab testing, as only 16% of PLWHIV did not correctly self-report whether their viral load was undetectable and 25% did not correctly self-report their CD4 count.³⁶ Another study found that asking participants on what days they had taken their ART, compared to what days they missed their ART, was more accurate.³⁷ Furthermore, in examining recall bias, one-month recall was comparable with 3-day recall, suggesting that one month can be a valid recall timeframe.³⁷ However, inaccurate self-reporting was associated with sexual orientation, non-white ethnicity, lower socioeconomic status, poor English fluency, lack of a support network, and nondisclosure of HIV status.³⁶ As such, self-reporting is highly susceptible to being impacted by systemic and individual barriers.

Literature supports that the quality of patient-provider interactions affects patient engagement in care, especially regarding provider qualities such as empathy and effective communication.³⁸ This measure of engagement acknowledges that frequency may not be superior to quality. In one qualitative study, provider competence was a barrier to engagement, especially for men as many male providers assumed that their male clients had sex with women only.³³ Additionally, this measure can act as a predictor of other measures of engagement. One study found that patients who felt that their providers always listened carefully to them kept 7% more of their appointments than patients who felt that their providers usually, sometimes, or never listened to them carefully.³⁸ While there is no standard way of measuring the quality of interaction with a provider, the quality of patient-provider interactions may lead to increased appointment adherence, lab testing, and ART adherence.

Strengths and Limitations

This study was done to provide an overview of definitions of engagement in HIV intervention research. To our knowledge, this is the first study summarizing the nuances of engagement in HIV care. Additionally, this study highlights the benefits of using CASCADE, a database of curated clinical trials investigating the HIV care cascade. Using such a selective database allowed this study to exclusively examine high-quality evidence-based research. One key limitation of this work is that we focused solely on randomized trials. Other study designs may have useful definitions of engagement that were not considered here.

Conclusion

Randomized trials assessing the safety and efficacy of HIV interventions define engagement in care in various ways, making it challenging to compare findings across studies. We recommend future work aimed at establishing a unified definition for engagement in HIV care.