A review of HIV-specific patient-reported measures of perceived barriers to antiretroviral therapy adherence: what themes are they covering?

In 2017, there were 2.2 million people living with human immunodeficiency virus (PLHIV) in western and central Europe and North America, with approximately 77% accessing antiretroviral therapy (ART) [ 1 ]. However, only 63% of PLHIV on ART are estimated to attain the ideal of at least 95% adherence [ 2 ]. Indefinitely maintaining ART adherence may be extremely difficult, given the numerous factors that can impede it [ 3 ]. While newer, more potent ART regimens may make perfect adherence less necessary [ 4 ], adherence difficulties are tied to a range of medically relevant psychosocial and structural issues. These include depression, alcohol/substance misuse, and health service-related barriers [ 5 ]. Indeed, regularly identifying a patient’s potential barriers to ART adherence is explicitly recommended in some HIV treatment guidelines [ 6 ]. Doing so could help address previously undetected problems and prevent virologic failures. Nevertheless, how best to do this remains less clear. Given the many recognized barriers to ART adherence, such an assessment could prove time-consuming [ 7 ]. Patient-reported outcome measures (PROM) could offer a solution and their use is growing in healthcare [ 8 ]. While published initiatives of their implementation in HIV care are few (e.g., [ 9 , 10 ]), using them to screen for barriers prior to the clinic visit could offer a quick and affordable solution and lead to more patient-centered counseling and intervention [ 7 ]. Yet there may be few comprehensive HIV-specific self-report measures for capturing and succinctly scoring patient perceived barriers to properly taking ART in developed countries [ 11 ]. It is also unclear to what extent PLHIV participated in their creation, considering that patient involvement is deemed essential to a PROM’s content validity [ 12 ]. In a previous research phase, our team generated a conceptual framework of ART adherence barriers based on a synthesis of qualitative studies with PLHIV in developed countries, to design a new PROM for use in routine HIV care in Canada and France [ 13 ]. With this review, we seek to: 1) identify existing patient-reported measures of barriers to ART adherence used in developed countries, and 2) examine their coverage of this patient-informed conceptual framework. Forty-one qualitative studies with adult PLHIV on barriers to ART adherence in developed countries were synthesized with thematic analysis to create our framework. It defines 6 broad interrelated themes under which are grouped 20 subthemes of barriers. Details on this framework are published elsewhere [ 13 ]. On July 4, 2018, four databases were searched for patient-reported measures of barriers to ART adherence: EMBASE, MEDLINE, PsychINFO, and Health and Psychological Instruments. Searches were adapted to each database and targeted words in the abstract referring to: 1) HIV; 2) adherence; 3) barriers; and 4) antiretroviral therapy. The searches were limited to English-language publications from 1996 and human adults (18 or 19 years and older). The precise search strings used are available upon request. Duplicates of all identified records were eliminated. Then, the title and abstract of each record were screened and the full-texts of all potentially relevant records were examined. Records of conference abstracts and opinion articles were excluded. A tenth of deduplicated records and 15% of full-texts were reviewed by IT to calculate interrater reliability with Cohen’s kappa [ 14 ] and percent agreement. The references of retained full-texts were also searched. Instruments (e.g., questionnaires, checklists, subscales) were included if they served to quantify perceived barriers to ART adherence. Specifically, eligible instruments allowed respondents to indicate factors that prevented them from taking the medication, as prescribed. Instruments also needed to be HIV-specific (i.e. designed or adapted for PLHIV), used in developed countries [ 15 ], based on patient report, and published in English no earlier than 1996, when combination ART became the new standard of care. If several versions of an instrument were found, only the most complete version was retained, unless item content differed meaningfully between them, in which case all were retained. Instruments with fewer than 3 items were excluded. They were also excluded if all relevant instrument items were not obtained, after contacting the author(s). We extracted the following information for each retained measure: instrument and/or study name, if appropriate; instrument items; publication or version year of the document from which the instrument items were extracted; number of items; author description of what the instrument measures; mention and form of patient involvement in its development; and first author and year of the research article publication affiliated with the measure. Based on Weiring et al. [ 16 ], patient involvement was defined as explicit mention of patient participation in either determining the outcome measured (e.g., in developing its framework or domains); generating items; and/or verifying content validity, including comprehensibility (e.g., through interviews). Our methods draw on the approach taken by O’Brien et al. [ 17 ]. To compare instrument items against our conceptual framework, we used content analysis [ 18 ], allowing for the creation of new themes to accommodate the items. We sought to map each item to the framework, using the qualitative analysis software, Atlas.ti (v8). Items could be coded for several subthemes. KE mapped all instrument items. IT mapped 10% of the items ( n = 43) to calculate percent agreement on each item’s main subtheme. To assess coverage of the concept of barriers to ART adherence, instrument breadth (representation of all original framework themes) and depth (representation of all original subthemes) were evaluated. Coverage was expressed with means (i.e. average instrument breadth and depth) and proportions (e.g., percentage of (sub)themes represented). We did not consider the number of items representing each (sub)theme. We reviewed a total of 1540 records, removing 730 duplicates (see Fig. 1 ). Following deduplication and exclusion of irrelevant records, based on title/abstract screening, the full-texts of 59 records were examined. Percent agreement was 90.1% for the deduplicated records and Cohen’s Kappa was 0.62, indicating substantial agreement [ 14 ]. Percent agreement for the full text articles was 88.9% and Cohen’s kappa was also 0.62. Relevant records and their references yielded 31 instruments for inclusion in the review. Two instruments were excluded [ 19 , 20 ], given incomplete access to their items. Search flow diagram Table 1 provides details on the instruments. Descriptions of an instrument could vary. All but one were described as measures of “reasons” (for “missing a dose”, “taking treatment breaks”, “nonadherence”, etc.) ( n = 21) or “barriers” (to “adherence”, “taking antiretrovirals”, etc.) ( n = 4) or both ( n = 5). They originated from the Unites States ( n = 20); Western Europe: Denmark, Germany, United Kingdom, and Sweden ( n = 4); Australia ( n = 3); Canada ( n = 3); and Romania ( n = 1). On average, they contained 13.5 items ( SD = 5.8), with a range of 3 to 23. For 9 measures, patient involvement was reported. Its specified forms included interviews ( n = 5), consultation ( n = 3), and piloting/pretesting/pre-experimentation ( n = 3). The version or publication year of the included instruments ranged from 1999 to 2017. An indication of their influence, authors reported adapting the Adult AIDS Clinical Trials Group (AACTG) adherence instruments [ 21 ] for 8 measures. Two original AACTG instruments were also included. Percent agreement for the item mapping was 88.4%. Thirty-five items were not mapped to the framework. Twenty-three of these, from 5 instruments, concerned “Likely clinically justified reasons” for not taking a specific antiretroviral agent or treatment (e.g., “Recommended by doctor”, “Changing regimens”). These items did not qualify as barriers, as they concerned situations in which the medication no longer seemed clinically indicated. Similarly, 4 other items related to “ How a person was non-adherent” (e.g., “Doubled up on a dose because you missed a dose”), falling beyond the framework’s scope. Finally, 8 items (/408, 2%) could not be confidently mapped, for lack of clarity (e.g., “You had a bad event happen that you felt was related to taking the pills”). Table 2 reports the findings on instrument breadth and depth. On average, breadth was 4.4/6 themes ( SD = 1.2). The majority of instruments covered the broad themes of “Lifestyle factors” (94%), the “Characteristics of antiretroviral therapy” (90%), “Cognitive and emotional aspects” (84%), the “Social and material context” (84%) and the “Health experience and state” (61%). Less than a quarter (23%) covered the “Healthcare services and system” theme. As to depth, it was, on average, 7.0/20 subthemes ( SD = 3.0). Individual subthemes were addressed in between 3% and 88% of instruments. A majority of instruments contained at least one item on the subthemes of “Demands and organization of daily life” (88%) (e.g., change/break in daily routine, away from home, forgot, fell asleep/overslept, ran out of pills); “Side effects” (81%); “Affect” (71%), especially, feeling depressed/overwhelmed; “Beliefs” about adherence, ART or HIV (63%) (e.g., felt like drug was toxic/harmful); “Instructions” for ART (61%) (e.g., too many pills, problems taking pills at specific times); “HIV stigma and privacy” (61%) (e.g., did not want others to notice); and “Bodily signals” (52%), particularly, feeling sick or ill.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.