Data source

Data came from the international, web-based, cross-sectional study called ‘Positive Perspectives Wave 2’ that was conducted among adult PLHIV aged ≥18 years during 2019 (n=2389)^3,12-16. Participants were recruited from Argentina, Australia, Austria, Belgium, Brazil, Canada, Chile, China, France, Germany, Ireland, Italy, Japan, Mexico, Netherlands, Poland, Portugal, Russia, South Africa, South Korea, Spain, Switzerland, Taiwan, UK, and the USA. PLHIV were sampled non-probabilistically from existing panels of confirmed HIV sero-positive individuals, from national, regional, and local charities/support groups; and online support groups/communities and social media platforms. Surveys were translated into local languages. In the Asian region, 230 participants were recruited: China (n=50), Taiwan (n=55), Japan (n=75), and South Korea (n=50).

Measures

Ever switching of ART because of DDIs

Participants were asked whether they had ever changed their ART ‘to a different medication or combination of medications’, as well as the reasons for the change. Individuals who reported ever changing ART because ‘My HIV medicines did not work well with other medicines/drugs/pills I was taking’ were classified as reporting a past ‘major DDI’ with ART (i.e. experienced DDIs that were severe enough to warrant a complete switch of their ART). This is a very conservative construct as it likely captures only the manifestations of DDI that were severe enough to cause switching or modification of ART but not milder manifestations.

Current concerns, perceptions and treatment priorities

The survey assessed whether participants were currently concerned about DDIs, how their attitudes about DDIs had changed from when they first started HIV treatment versus now, the extent to which they prioritized issues related to DDI in their current treatment, and whether they were comfortable discussing DDI-related concerns with their healthcare providers.

Participants were classified as being currently concerned about the risk of DDIs if they responded ‘agree’ or ‘strongly agree’ (vs ‘strongly disagree’, ‘disagree’, or ‘neither agree nor disagree’) to the statement ‘I worry how my HIV medicines will affect other medications/drugs/pills I take’. To evaluate how attitudes towards ART may have changed over time among treatment-experienced individuals (those diagnosed for ≥1 year), eligible participants responded to two separate items in the survey that asked them what they prioritized the most when they first started ART, and what their priorities were now at the time of the survey. Both items were completed within the same survey but based on the respondent’s recall of the different time periods in question. Based on these responses, reducing DDIs was considered a new treatment priority if it was not deemed important at ART initiation (based on the respondent’s recall) but was deemed important now. Conversely, reducing DDIs was considered a current treatment priority if it was deemed important now, regardless of participants’ recall of its perceived importance at ART initiation. Among all participants, perceived comfort discussing DDI-related concerns with their healthcare providers was assessed by asking participants to rate to what extent they would feel comfortable discussing with their main HIV care provider ‘concerns about how your HIV medication affects other medications/drugs/pills you take’. Those answering ‘comfortable’ or ‘very comfortable’ (vs ‘very uncomfortable’, ‘uncomfortable’, or ‘neither comfortable nor uncomfortable’) were classified as being comfortable discussing DDI-related issues with their healthcare providers.

Discrete choice experiment to rank PLHIV’s current treatment priorities

To better understand the perceived relative importance (from the perspective of PLHIV) of the need to reduce the number of medicines taken as well as DDIs within the context of other improvements to HIV medicines, a discrete choice experiment¹⁷ was performed as part of the survey. Participants ranked the following ART attributes in order of perceived importance: reduced DDIs, smaller pills, fewer side effects, reduced long-term negative impacts, no food requirements, less medicines each day but just as effective, and non-daily regimens. More details about this question item have been described elsewhere¹⁸.

Clinical parameters

From a list of 21 health conditions (including a catch-all ‘other’ category), participants were asked to select which conditions they had ever been diagnosed with, and which ones they were currently taking medicines for at the time of the survey. Assessed conditions were mental illness, hypertension, hypercholesterolemia, insomnia, gastrointestinal disease (e.g. gastric ulcer), anemia, liver disease, arthritis, lung disease, substance use disorder, lipodystrophy, bone disease, diabetes, heart disease, neurological conditions, kidney disease, tuberculosis, malabsorption, and dementia. Each condition was recoded as a ‘yes’ or ‘no’ variable.

From these data, we created a tally of non-HIV conditions participants had ever been diagnosed with (range: from 0 for those living only with HIV, to ≥7 conditions), as well as conditions for which they were currently taking medicines at the time of the survey, including HIV (range: from 1 for those taking medicines only for HIV, to ≥7 conditions). Counts of comorbidities and concurrent treatments 7 or greater were collapsed as ‘≥7’ because of sample size considerations.

Self-reported virologic control was defined as a response of ‘undetectable’ or ‘suppressed’ versus ‘detectable,’ ‘unsuppressed,’ ‘I don't know,’ or ‘prefer not to say’ to the question: ‘What is your most recent viral load?’. Polypharmacy was defined as taking ≥5 pills per day for HIV or non-HIV conditions, or currently taking medicines for ≥5 conditions, including HIV³. Suboptimal adherence was defined as a report of one or more reasons for missing ART for ≥5 times within the past month¹². Self-ratings of physical, mental, sexual, or overall health as ‘good/very good’ were classified as optimal health (vs ‘very poor/poor’/‘neither good nor poor’).

Statistical analysis

Prevalence estimates for DDI-related endpoints were stratified for participants in the Asian region versus other non-Asian geographical regions. All multivariable analyses were however performed on pooled data using the full dataset (n=2389) to increase sample size. Adjusted prevalence ratios were calculated in a multivariable Poisson regression model¹⁹ to explore factors associated with a past major DDI (coded as 1 if they ever switched ART because of DDIs, coded as 0 if they never switched ART, or switched for reasons other than DDIs). Secondary outcomes explored in the multivariable analysis were a report of DDIs as a current treatment priority, as well as a report of DDIs as a new treatment priority, as defined earlier. The independent variables assessed in the exploratory regression models were age, gender, geographical region, education level, domicile, HIV duration, number of concomitant medications, and ART formulation (i.e. single, or multi-tablet regimen). To examine how the types of current medical treatments for other non-HIV conditions were associated with participants’ concerns about their risk of ART-related DDIs, adjusted prevalence ratios were calculated for each co-treatment assessed, with perceived concern over DDI risk (yes or no) as the dependent variable, and condition-specific treatment status as the independent variable (coded as 1 if taking medication for the condition now, 0 if otherwise), adjusting for age, gender, and geographical region. Similarly, to examine how the type of non-HIV comorbidity ever diagnosed of was associated with participants’ past switching of HIV medication because of ART-related DDIs, adjusted prevalence ratios were calculated with DDI-induced past ART switching as the outcome (ever vs never), and ever diagnosis of the specified condition as the predictor (coded as 1 if ever diagnosed of the condition, 0 otherwise), adjusting for age, gender, and geographical region.

To measure the relationship between DDI experience/concerns (exposure variable) and various health-related outcomes, we classified the exposure variable into three categories: 1) reported neither a history of DDI-induced ART switching nor any concerns about DDI risks; 2) reported concerns about DDI risks but no history of DDI-induced ART switching; and 3) reported history of DDI-induced ART switching (i.e. past major DDI). Estimation of prevalence ratios within a Poisson regression model adjusted for age, gender, geographical region, and presence of non-HIV comorbidities; the last was included as a control variable to assess the independent effect of number of medications being administered (modifiable factor) separate from conditions ever diagnosed of (non-modifiable factor). Statistical significance was set at p<0.05. All analyses were conducted using SAS, Cary, NC, v9.4 and R v3.4.

DDI-related experiences and perceptions

Of all participants from the Asian region, 59.1% (136/230) reported they were concerned about the risk of DDIs (China 56.0%, Japan 57.3%, South Korea 58.0%, Taiwan 65.4%), 58.3% (134/230) were concerned about having to take more medicines as they grew older (China 72.0%, Japan 42.7%, South Korea 64.0%, Taiwan 61.8%), and 25.9% (44/170) of those who had ever switched ART in the Asian region attributed it to DDIs (China 43.4%, Japan 23.4%, South Korea 20.7%, Taiwan 15.8%) (Table 1). Switching ART because of DDIs within the Asian sample was around two-fold higher among men who have sex with women [37.0% (20/54)] than men who have sex with men [15.8% (9/57), p=0.016]; those living in non-metropolitan [36.1% (26/72)] than in metropolitan areas [18.4% (18/98), p=0.011], and those on multi-tablet ART regimens [32.2% (38/118)] than on single-tablet regimens [12.0% (6/50), p=0.006]. When compared to several other geographical regions, participants in the Asian region were more likely to switch ART on account of DDIs, as shown in Figure 1. Only 55.7% (128/230) of Asian participants felt comfortable discussing DDIs with their healthcare providers (China 50.0%, Japan 72.0%, South Korea 26.0%, Taiwan 65.4%) (Figure 1). Furthermore, participants in the Asian region reported the lowest percentage (57.4%) of those who perceived that their healthcare provider met their needs and took into consideration the things important to them (Figure 1).

Figure 1

Percentage of participants who reported various perceptions and beliefs regarding drug-drug interactions and other aspects of care, by geographical region within pooled analysis of people living with HIV in all participating locations (N=2389)

Among Asians who participated in the discrete choice experiment, 4.6% ranked the need to reduce DDIs in first place of importance relative to the other ART attributes assessed, 10.3% ranked it in second place, while 20.0% ranked it in third place (Figure 2). First, second, and third place rankings for ART regimens with less medicines each day but just as effective, were 9.7%, 16.4%, and 16.9%, respectively (Figure 2). Ranking of the seven ART attributes perceived as most important based on their cumulative shares of first, second, and third place positions were as follows among Asian participants: reduced long-term impact on my body (80.1%), fewer side effects (71.7%), ‘longer-lasting medicine so I don't have to take it every day’ (47.1%), less HIV medicine each day but just as effective (43.0%), less chance of affecting other medicines/drugs/pills I take (34.9%), no food restrictions or requirements (15.9%), and smaller pills (7.2%).

Figure 2

Rankings of the perceived relative importance of various attributes of HIV medications among people living with HIV from the Asian region (N=230)

Factors associated with DDI-related experiences and perceptions

Following adjustment for region, age, and gender within a pooled analysis, current medical treatment for certain conditions increased the likelihood of being worried about the risk of DDIs (Figure 3). Compared to those not currently on medication for the specified conditions below, those currently on medication reported greater concern about the risk of DDIs with their ART: hypercholesterolemia (adjusted prevalence ratio, APR=1.17; 95% CI: 1.03–1.33), mental health disorders (APR=1.19; 95% CI: 1.06–1.33), insomnia (APR=1.19; 95% CI: 1.04–1.36), heart disease (APR=1.21; 95% CI: 1.01–1.46), arthritis (APR=1.22; 95% CI: 1.01–1.47), bone disease (APR=1.28; 95% CI: 1.06–1.54), gastrointestinal disease (APR=1.29; 95% CI: 1.13–1.48), kidney disease (APR=1.30; 95% CI: 1.01–1.66), and lipodystrophy (APR=1.40; 95% CI: 1.07–1.83) (all p<0.05) (Figure 3).

Figure 3

Adjusted prevalence ratios with 95% confidence intervals for self-reported concern over risk of drug-drug interactions, comparing individuals currently taking medications for each specified condition vs those not taking medications for that condition currently within pooled analyses of people living with HIV in all participating locations, 2019 (N=2389)

There was also a strong association between certain conditions ever been diagnosed (regardless of whether currently being treated), and a report of past DDI-induced ART switching (Figure 4). The strongest associations in this regard were noted for bone disease (APR=1.95; 95% CI: 1.35–2.82), heart disease (APR=2.08; 95% CI: 1.45–2.98), diabetes (APR=2.08; 95% CI: 1.50–2.88), gastrointestinal disease (APR=2.16; 95% CI: 1.66–2.82), malabsorption (APR=2.60; 95% CI: 1.61–4.21), kidney disease (APR=2.73; 95% CI: 1.90–3.94), and dementia/Alzheimer’s disease (APR=3.63; 95% CI: 2.19–6.01) (all p<0.05).

Figure 4

Adjusted prevalence ratios with 95% confidence intervals for report of having ever switched HIV medications on account of drug-drug interactions, comparing individuals ever diagnosed of each specified condition vs those never diagnosed of that condition within pooled analyses of people living with HIV in all participating locations, 2019 (N=2389)

When expressing the outcome past DDI-induced ART switching as a percentage of all participants (not just those who switched), the probability of reporting this outcome increased with increasing number of comorbidities as well as with increasing number of concurrent treatments (Figure 5). The percentage of the entire pooled sample that ever switched ART because of DDIs ranged from 10.6% (148/1402) among those currently on ART only, to 26.4% (14/53) among those receiving medications for ≥7 conditions (including HIV). Consistent trends were seen when examining past DDI-induced ART switching in the entire pooled sample as a function of comorbidities ever diagnosed; this percentage ranged from 9.4% (44/470) among those with exactly one non-HIV comorbidity ever diagnosed, to 22.7% (32/141) among those with ≥7 non-HIV comorbidities ever diagnosed. Notably, even among those reporting they had never been diagnosed with any non-HIV comorbidity, 11.7% (116/993) still reported past DDI-induced ART switching.

Figure 5

Dose-response relation between number of non-HIV comorbidities ever diagnosed of ^a as well as the total number of conditions currently receiving medications for ^b, and the outcome of having ever changed HIV medications because of drug-drug interactions ^c within pooled analyses of respondents in all participating locations, 2019 (N=2389)

Of all participants in the Asian region who had been living with HIV for at least one year, 34.6% (71/205) identified the need to reduce DDIs as a treatment priority when they first started ART (China 40.0%, Japan 28.2%, South Korea 25.0%, Taiwan 46.9%); the same percentage overall, 34.6%, also identified this as a treatment priority now (China 33.3%, Japan 33.8%, South Korea 25.0%, Taiwan 44.9%). More so, 14.6% (30/205) of those living with HIV for at least one year did not consider the need to reduce DDIs as a priority when they first started ART, but now identified it as a current priority. Within pooled multivariable analysis using the full dataset, the recognition of the need to reduce DDIs as a current priority (regardless of perception at ART initiation), generally increased with increasing number of concomitant medications when compared to those on ART only; prevalence ratios, by range of concurrent treatments, were 1.58 (95% CI: 1.39–1.80) among those receiving medications for two conditions (including HIV), to 2.05 (95% CI: 1.65–2.55) among those receiving medications for ≥7 conditions (including HIV). In contrast, when the outcome was new concerns about DDIs, the relationship with the number of concurrent treatments was significant only to a certain threshold of concurrent treatments (≥4) after which increasing number of concurrent treatments was no longer statistically significant (Table 2).

Older adults aged ≥50 years were more likely to perceive the need to reduce DDIs as a current treatment priority, than younger adults within pooled analysis (APR=1.14; 95% CI: 1.02–1.27). Compared to the Asian region, the adjusted likelihood of ever switching ART on account of DDIs did not differ significantly among participants in North America but was significantly lower among participants in Europe (APR=0.65; 95% CI: 0.46–0.90), Latin America (APR=0.41; 95% CI: 0.24–0.67), and other regions (APR=0.28; 95% CI: 0.17–0.47) (Table 2).

Associations between DDI experiences and health-related outcomes

When compared to those with no history/current concerns about DDIs within pooled multivariable analyses, those reporting past DDI-induced ART switching reported higher likelihood of suboptimal adherence (APR=1.91; 95% CI: 1.58–2.30) and lower likelihood of self-reported viral suppression (APR=0.73; 95% CI: 0.66–0.81). They also reported lower likelihood of optimal health on all assessed domains, including physical (APR=0.62; 95% CI: 0.54–0.72), mental (APR=0.71; 95% CI: 0.62–0.82), and overall (APR=0.65; 95% CI: 0.56–0.75), as well as lower likelihood of treatment satisfaction (APR=0.85; 95% CI: 0.77–0.94). Conversely, they were more likely to perceive room for improvement with their current HIV medication (APR=1.62; 95% CI: 1.40–1.87). Those with no report of a past DDI-induced ART switching but nonetheless concerned about the risk of DDIs also reported less favorable health-related outcomes when compared with those having no history/concerns about DDIs, as shown in Table 3.

Limitations

Some limitations exist to this study. First, these are cross-sectional, self-reported data and only associations can be drawn. Second, the data may not be fully representative of the respective geographical regions because of the non-probabilistic sampling. Furthermore, the use of pooled data to generate estimates means that the results may not be representative of any specific country or region (e.g. the Asian region). Finally, there were no data on which specific medicines participants were taking for the condition(s) assessed; data only existed for the conditions for which medicines were being taken. Thus, our indicator of ‘concurrent treatments’ merely reflects counts of comorbidities being treated, not actually co-medications. Also, the list of health conditions assessed was not exhaustive. Finally, this study did not consider interactions between ART and herbal treatments.