Patient preferences and health state utilities associated with the treatment process of antiretroviral therapy for people living with HIV

Like most studies designed to estimate treatment process utilities [29, 34], this study was conducted using vignette-based methods. Generic preference-based measures such as the EQ-5D were designed to assess overall health status and are unlikely to be sensitive to differences in treatment process. In contrast, the vignette-based approach is well suited for isolating the impact of treatment process on utility.

Four health state vignettes were developed and refined based on published literature, clinician interviews, and a pilot study. All health states included the same description of a person living with HIV, but differed in the description of the treatment process. The four treatment processes were selected to represent two common oral ART regimens and two LAI regimens [10, 22, 24, 25, 35]. The health states were valued in a time trade-off (TTO) utility elicitation study with a sample of PLHIV in the UK. One-on-one TTO interviews were conducted by videoconference from November 2020 to January 2021. Because the four health states varied only in the treatment process, all resulting differences in preference and utility can be attributed entirely to these treatment process differences.

Informed consent was obtained prior to each interview, and the study protocol was approved by an institutional review board (Ethical and Independent Review Services; Study 20173–01).

A targeted literature review was conducted to inform the initial draft of the health states. This review focused on the experience of living with HIV [4, 10, 17, 18, 36], ART [10, 12, 36, 37], and the ART treatment process [12, 15, 16, 38‐41]. Based on this literature, initial drafts of four health states were developed, along with a detailed background information document providing details on the four ART options. Then, interviews were conducted with four clinicians (three HIV specialists and one infectious disease specialist; two MD, one MBBS, one MB BCh; three from London, UK, and one from Omaha, Nebraska) to refine the health states and background information document. These clinicians reported treating between 10 and 200 PLHIV per month, including 95% taking ART and over 90% at undetectable viral loads. Clinicians averaged over 22 years of experience working with HIV.

These interviews were conducted using a semi-structured interview guide that included questions on the clinicians’ professional background, description of HIV with an undetectable viral load, and description of daily oral and LAI ART. Health states were developed in an iterative process. Clinicians were interviewed multiple times, and interviews continued until clinicians agreed that the health states and background information document accurately described typical patient experiences with ART treatment regimens.

All four health states began with the same description of a person living with HIV, treated with ART and having an “undetectable viral load.” After the description of HIV, each health state described one of four treatment processes. Health state A described a single tablet taken daily, and health state B described two tablets taken together daily. Both descriptions also stated that the tablets should be taken at the same time every day, and that the tablets need to be carried while traveling to ensure that treatment is not skipped on those days.

Health states C and D described injections administered monthly and once every two months, respectively. These descriptions stated that the injections must be administered at a clinic or hospital, must occur around the same day each month, are administered in the buttocks, and only take a few minutes. The health states also specified that if an individual is receiving injections as scheduled, there is no need for additional oral medication. Complete health state text is presented in the online supplementary material.

A background information document (included in supplementary material) was developed to inform respondents about potential advantages and disadvantages of oral and LAI routes of administration for ART. This background information page was developed to ensure that participants were aware of the potential implications of the two routes of administration before they were asked to report their health state preferences. To ensure that respondent preferences were based on the treatment process and not on assumptions of drug efficacy or adverse event profiles, the background information document stated that the treatments were equally effective with similar rates of side effects, but that “research has shown that some patients prefer the daily oral treatment, while others prefer the long-acting injectable treatment.” A series of bullet points then described some of the primary reasons that PLHIV have reported for liking/disliking the two routes of administration (three positive points and three negative points for each route of administration). Participants were instructed to consider these advantages and disadvantages of the two routes of administration during the health state ranking and utility elicitation. All treatment attributes in this background information document were selected based on published literature [12, 15, 16, 37, 38] and clinician input. The clinicians helped to refine this document so that it was consistent with their observations of typical patient experiences.

A pilot study was conducted with 16 PLHIV (75.0% male; mean age = 39.5 years) in the UK. Participants completed a TTO valuation and provided feedback on the health states and procedures. Based on this feedback, health states were edited for clarity and ease of understanding. The participants often reported that the description of living with HIV was similar to their own experience. All participants said the health states and background information document were clear and easy to understand. Data from the pilot study were not included in the main valuation analysis sample.

Participants were recruited from a patient database that was populated as follows: collaborating with patient support organizations and charities, digital marketing to targeted audiences (e.g., via Facebook, Twitter, Google), and patient referrals. Recruitment messages for the current study were sent via email to PLHIV in the database. If participants responded with interest, they were screened by phone for eligibility. To be eligible for this study, participants were required to be over 18, a UK resident, diagnosed with HIV, virologically suppressed (i.e., viral load under 50 copies per milliliter) as indicated by their most recent test, and currently treated with ART. Because interviews were conducted by videoconference, participants were required to have a desktop computer, laptop, or tablet with video capabilities, and have a physical mailing address where they could receive a packet of study materials. Respondents were required to provide proof of receiving ART for HIV in one of three ways: a photo of the medication or medication packaging, a prescription note containing their name and the medication name, or a letter from their hospital/clinic/doctor/nurse.

The four health states finalized in the pilot study were used to assess preference and elicit health state utilities in the larger utility elicitation. Six trained interviewers conducted one-on-one interviews, following a semi-structured interview guide. Each interviewer was observed by the principal investigator (PI) at least once (by joining the Microsoft Teams meeting) to ensure consistency in interview procedures.

First, participants were introduced to the differences between oral and LAI treatment process, using the background information document. The interviewer reviewed each bullet point in this document to ensure that the respondent understood these details. Then, the four health states were presented in random order and reviewed at the same level of detail. The participant was then given an opportunity to read the materials independently and ask questions about the health state content. After confirming that the participant understood the health states and had no further questions, participants were asked to rank the health states from most preferable to least preferable and explain their preferences. Direct quotes were recorded by the interviewer.

After completing the ranking, participants valued the health states in a TTO task with a 20-year time horizon. TTO methods have been described extensively in previous publications [28]. For each health state, participants were offered a series of choices between a life of 20 remaining years in the health state being rated or a shorter period of time in full health, followed by dead. Choices were presented in a way that alternated between longer and shorter periods of time in full health (e.g., 20 years, 0, 1, 19, 2, 18, 3…). Each health state received a utility value on a scale with the anchors of dead (0) to full health (1) based on the choice in which the respondent was indifferent between 20 years in the health state being evaluated and x years in full health. The resulting utility estimate (u) is calculated as u = x/20.

Different TTO procedures were planned for health states perceived to be worse than dead. However, no participants perceived a health state to be worse than dead, so these procedures were never implemented.

To maintain data quality and ensure the respondents fully understand the health state vignettes and TTO task, the optimal approach for vignette-based utility elicitation involves one-on-one interviews in which the interviewer and respondent can share the study materials. Because of the COVID-19 pandemic, however, interviews could not be conducted in person. Therefore, a three-step approach was used to simulate the experience of an in-person interview. First, a package was sent to the participant with paper copies of all materials that would be necessary for the interview, including the health states, the background information document, and questionnaires. Participants were instructed not to review any of the materials prior to the interview.

Second, the one-on-one interviews were conducted by videoconference via Microsoft Teams. At the beginning of the interview, participants were instructed to open the package that had been sent to them. They were also asked to work at a table or desk large enough for the four health states to be spread out and placed in order of preference. With the videoconferencing software, the interviewer and participant could see each other and point to parts of the health states and background information document as if they were working together in the same room. The TTO choices (e.g., 20 years in full health vs. a shorter amount of time in the health state being rated) were presented in a PowerPoint slide deck with the screen sharing feature of Microsoft Teams.

Third, the PI and/or project manager (PM) were available for questions during every interview. For example, if an interviewer was unsure about how to answer a participant’s question or how to clarify an aspect of the TTO task, the interviewer would send a text message to the PI or PM for assistance. If a brief response would suffice, the PI or PM could respond by text message. If a more thorough response was necessary, the PI or PM could join the Microsoft Teams meeting to help clarify the issue. With this approach to supervision, it was possible to simulate the experience of collecting data as a team in a single location, thus maximizing data quality and consistency across interviewers.

After the utility elicitation, participants completed two questionnaires. First, they completed a demographic and clinical form. Second, they reported health-related quality of life on the SF-12 [42]. The SF-12 has been found to be a valid and reliable measure in PLHIV [43‐46].

Statistical analyses were conducted with SAS version 9.4. Descriptive statistics were used to summarize demographic data, SF-12 physical/mental component summary scores, health state preferences, and utilities (frequencies and percentages for categorical variables; means and standard deviations for continuous variables). Paired t-tests were conducted to examine differences between utility means (e.g., utility of health state A vs. utility of health state B), and independent t-tests were used to test for subgroup differences in utilities by age [median split], gender, and employment status [employed vs. not employed]). Post hoc descriptive analyses were conducted to provide utilities for subgroups of patients categorized based on preference for oral or LAI treatment.

A total of 268 potential participants were eligible based on screening. Of those, 226 were scheduled, and 205 attended interviews. Four of these participants who were eligible at screening became ineligible by their interview date because of test results indicating that they were no longer virologically suppressed. Therefore, the analysis includes data from 201 interviews. See Table 1 for demographics.

Participants reported being diagnosed with HIV an average of 14.1 (SD = 8.7) years prior to their interview and beginning ART an average of 11.3 (SD = 6.8) years prior. Almost all participants were currently on a regimen of one pill per day (n = 92; 45.8%) or multiple pills taken at the same time daily (n = 99; 49.3%). Most participants (n = 187; 93.0%) reported missing their medication fewer than one day per week. Only one participant (0.5%) reported having used LAI ART. The most commonly reported comorbid health conditions were anxiety (n = 78; 38.8%), depression (n = 67; 33.3%), hypertension (n = 25; 12.4%), and arthritis (n = 19; 9.5%). Mean (SD) SF-12 Mental and Physical Component Scores were 50.38 (12.14) and 42.74 (10.25), respectively. These scores are somewhat lower (i.e., indicating worse health) than those found in other studies reporting SF-12 values of PLHIV [44], but this study occurred during the COVID-19 pandemic, which could have affected these scores.

Participants ranked the health states from most preferable to least preferable (Table 2). Almost all participants said the most preferable health state was either D representing injections once every two months (n = 119; 59.2%) or A representing a single daily tablet (n = 75; 37.3%). Preference orders (listed in order from most to least preferable) reported by at least 5% of the sample included DCAB (n = 70; 34.8%), ABDC (n = 58; 28.9%), DABC (n = 31; 15.4%), DACB (n = 14; 7.0%), and ADBC (n = 10; 5.0%).

After providing their rankings, participants explained their preferences. Common reasons for preferring daily oral treatment included the ease of oral administration, convenience of daily tablets, fear of needles, and the fact that oral medication is taken at home without requiring additional medical appointments. Common reasons for preferring LAI administration were the convenience of less frequent dosing, no risk of missing a daily dose, elimination of the daily reminder of HIV, and the convenience of traveling without bringing medication. Quotations from participants are presented in Table 3.

Health state D (injections every two months) had the highest mean utility score at 0.910, followed by A (single tablet daily) at 0.908, B (two tablets daily) at 0.905, and C (injections every month) at 0.900 (Fig. 1). Of the 201 respondents, 128 (63.7%) had the same utility score for all four health states, while 73 (36.3%) had differences in utility scores. Significant differences in mean utilities were found between pairs of health states differing in number of daily tablets (A vs. B) and frequency of injections (C vs. D) (Table 4). However, the utility difference between the two most preferred health states (A and D) was not significant. The majority of respondents (n = 152; 75.6%) had the same utility score for these two health states, while 23 participants (11.4%) had a higher utility for A, and 26 (12.9%) had a higher utility for D. There were no significant between-group differences in utilities by age, gender, or employment status.

Additional exploratory descriptive analyses were conducted so that utilities could be reported separately for those who preferred oral ART and those who preferred injectable ART as reported during the ranking task (Table 5; Results described in online supplementary material).