Abstract
Objective:
We aimed to investigate potential causes of higher risk of treatment interruptions within the multi-country START trial in 2015.
Methods
We defined baseline as the date of starting ART and a treatment interruption as discontinuing ART for ≥2 weeks. Participants were stratified according to randomisation arm and followed from baseline to the earliest of end date of the initial phase of START, death, date of consent withdrawn, or the date of first treatment interruption. Cox regression was used to calculate hazard ratios (HR) and 95% confidence intervals for factors that may predict treatment interruptions in each arm.
Results
Of the 3,438 participants who started ART, 2286 were in the immediate arm and 1152 in the deferred arm. 12.9% of people in the immediate arm and 10.5% of people in the deferred arm experienced at least one treatment interruption by 3 years after starting ART. In adjusted analyses, age (HR for 35–50 years: 0.75 (95% CI: 0.59, 0.97) and >50 years: 0.53 (0.33, 0.80) vs. <35 years), education status (HR for post-graduate education vs. less than high school education (0.23 (0.10, 0.50)) and region (HR for United States vs. Europe/Israel (3.16 (2.09, 4.77)) were significantly associated with treatment interruptions in the immediate arm. In the deferred arm, only age and education status were significantly associated with treatment interruptions.
Conclusion
Within START, we identified younger age and lower educational attainment as potential causes of ART interruption. There is a need to strengthen adherence advice and wider social support in younger people and those of lower education status.
Keywords: HIV, Interruption, ART, Treatment, Discontinuation
Introduction
In 2015, results published from the large-scale, multi-country Strategic Timing of AntiRetroviral Treatment (START) trial showed that amongst people with CD4 count above 500/mm3 starting treatment straight away, rather than delaying until the CD4 count was lower than 350 cells/mm3 or an AIDS disease had occurred, reduced the risk of developing serious AIDS and non-AIDS diseases [1].
People who start ART early will tend to be on treatment for a longer period than those who defer treatment, and whilst current drugs are substantially more tolerable now than they once were, most are still not without side effects [2]. This, along with other reasons such as travel, important distracting life events, perceived and possibly real impact of treatment on quality of life, may result in treatment interruptions. This is despite the evidence from the earlier Strategies for Management of Antiretroviral Therapy (SMART) trial showing the detrimental effects of treatment interruption [3].
It has been shown that certain groups are more likely to interrupt treatment than others. Whilst studies are differentiated by both definitions of interruption and inclusion criteria, recent studies suggest men are at higher risk of interruption than women [4–6], although some earlier studies have shown a higher risk in women [7, 8]. People who inject drugs and those of younger age have also been shown to be at an increased risk of treatment interruption [7–11], as have those with low educational attainment [10].
Whilst risk factors associated with treatment interruptions have previously been studied, we place focus on early ART initiators in a multi-national trial. Using data from the START trial, we aimed to gain insights into what factors potentially cause a raised risk of treatment interruption.
Methods
The START trial was designed and conducted by the International Network for Strategic Initiatives in Global HIV Trials (INSIGHT). People were eligible for the study if they were HIV-positive, aged >18 years of age, had not yet initiated ART and had two CD4 counts of >500 cells at least 2 weeks apart within 60 days before enrollment. People were randomised to either immediate initiation of ART or deferred initiation until CD4 cell count dropped to 350 cells/mm3 or occurrence of an AIDS defining disease or another condition that dictated the use of antiretroviral therapy (e.g. pregnancy). (For full details regarding the original study design, see the study protocol, available at NEJM.org.)
For these analyses, we defined baseline as the date of starting ART. A treatment interruption was defined as stopping all antiretroviral treatment for at least two weeks, regardless of the cause of interruption – i.e. both clinician-directed and self-reported interruptions (>75% of interruptions were self-reported) were included. Self- reported interruptions were ascertained using information from the ‘Adherence to ART’ questionnaires (1–203 and 1–203A) and amalgamated with the prescribed ART information. Participants were followed from baseline (the start of ART, the earliest of which was May 2009) to whichever occurred first of the following: end date of the initial phase of START (26 May 2015, based on an interim review of the study data by the study’s Independent Data and Safety Monitoring Board), death, date of consent withdrawn, last known date of being alive, or the date of first treatment interruption.
Kaplan-Meier analyses, stratified by randomisation arm were used to estimate the proportion of people who experienced a treatment interruption. Univariable and multivariable Cox proportional-hazards models, stratified by randomisation arm were used to calculate hazard ratios and 95% confidence intervals for characteristics measured at time of start of ART that may predict treatment interruptions: age, gender/mode of acquisition, education status, region, CD4 and viral load. We also calculated hazard ratios for specific antiretroviral drugs prescribed as the initial regimen; the NRTI backbone was categorised as TDF + 3TC/FTC, ABC+ 3TC/FTC, ZDV+3TC or ‘other combination’ and the ‘third’ drug in the regimen was categorised as EFV, other NNRTI, boosted ATV, boosted DRV, other boosted PI, any integrase inhibitor or ‘other’.
Results
Of the 4,684 HIV positive participants who were randomised 3,438 participants started ART for the first time prior to May 2015 (median time from start of ART to study end/withdrawal/death = 31.0 ((Inter-quartile range (IQR): 21.8, 43.1) months); 2286 of these participants were in the immediate arm and 1152 in the deferred arm. The median time from randomisation to starting ART in the immediate arm was 0.2 (0.1, 0.5) months and in the deferred arm was 18.7 (10.3, 29.0) months. In total, 380 (11.1%) participants experienced at least one treatment interruption and of these, 40 participants (10.5%) did not restart treatment before the study end date. By 3 years after starting ART, 12.9% (Kaplan-Meier estimate) of people in the immediate arm and 10.5% of people in the deferred arm experienced at least one treatment interruption.
The median duration of interruption amongst those who restarted ART was 43 (22, 93) days and 59 (30, 164) days in the immediate and deferred arms, respectively. Baseline characteristics stratified by randomisation arm are shown in Table 1.
Table 1:
Baseline characteristics stratified by treatment interruption and randomisation arm
Treatment interruption | |||||
---|---|---|---|---|---|
Immediate (n=2286) | Deferred (n=1152) | ||||
No | Yes | No | Yes | ||
1993 (87.2) | 293 (12.8) | 1065 (92.4) | 87 (7.6) | ||
Age | <35 years | 869 (84.7) | 157 (15.3) | 427 (89.5) | 50 (10.5) |
35–50 years | 874 (88.6) | 112 (11.4) | 464 (94.3) | 28 (5.7) | |
>50 years | 250 (91.2) | 24 (8.8) | 174 (95.1) | 9 (4.9) | |
Gender/mode of | IDU | 28 (75.7) | 9 (24.3) | 12 (80.0) | 3 (20.0) |
acquisition | MSM | 1138 (88.9) | 142 (11.1) | 733 (94.5) | 43 (5.5) |
Heterosexual female | 462 (83.2) | 93 (16.8) | 173 (86.9) | 26 (13.1) | |
Heterosexual male | 265 (88.0) | 36 (12.0) | 101 (91.0) | 10 (9.0) | |
Other | 100 (88.5) | 13 (11.5) | 46 (90.2) | 5 (9.8) | |
Education | Less than high school | 544 (80.1) | 135 (19.9) | 203 (86.0) | 33 (14.0) |
High school graduate | 447 (88.6) | 69 (13.4) | 216 (91.9) | 19 (8.1) | |
Completed vocational training | 189 (90.4) | 20 (9.6) | 121 (93.8) | 8 (6.2) | |
Some college/university | 344 (83.4) | 41 (10.7) | 197 (92.5) | 16 (7.5) | |
Bachelor’s degree | 354 (94.4) | 21 (5.6) | 250 (97.3) | 7 (2.7) | |
Any post-grad education | 115 (94.3) | 7 (5.7) | 78 (95.1) | 4 (4.9) | |
Region | Africa | 430 (87.8) | 60 (12.2) | 117 (86.0) | 19 (14.0) |
Latin America | 483 (82.8) | 98 (17.2) | 253 (93.7) | 17 (6.3) | |
Europe and Israel | 701 (93.3) | 50 (6.7) | 465 (94.5) | 27 (5.5) | |
United States | 196 (80.3) | 48 (19.7) | 134 (87.6) | 19 (12.4) | |
Australia | 49 (87.5) | 7 (12.5) | 34 (94.4) | 2 (5.6) | |
Asia | 144 (82.8) | 30 (17.2) | 62 (95.4) | 3 (4.6) | |
NRTI backbone | TDF + 3TC/FTC | 1797 (88.7) | 228 (11.3) | 958 (93.0) | 72 (7.0) |
ABC + 3TC/FTC | 61 (85.9) | 10 (14.1) | 65 (92.9) | 5 (7.1) | |
ZDV + 3TC | 133 (70.7) | 55 (29.3) | 37 (80.4) | 9 (19.6) | |
Other combination | 2 (100.0) | 0 (0.0) | 5 (83.3) | 1 (16.7) | |
Third drug | EFV | 1445 (87.1) | 215 (12.9) | 539 (92.9) | 41 (7.1) |
Other NNRTI | 92 (94.9) | 5 (5.1) | 144 (92.3) | 12 (7.7) | |
ATV | 190 (83.0) | 39 (17.0) | 85 (87.6) | 12 (12.4) | |
DRV | 147 (90.2) | 16 (9.8) | 116 (92.1) | 10 (7.9) | |
Other PI | 25 (80.7) | 6 (19.3) | 24 (85.7) | 4 (14.3) | |
INSTI | 91 (89.2) | 11 (10.8) | 153 (95.0) | 8 (5.0) | |
Other | 3 (75.0) | 1 (25.0) | 5 (100.0) | 0 (0.0) | |
CD4 at start of ART | cells/mm3 | 647 (579, 757) | 666 (596, 788) | 405 (319, 563) | 463 (329, 593) |
Viral load at start of ART | Copies/mL | 14328 (3413, 46000) | 9453 (2609, 35880) | 41700 (12700, 120000) | 35100 (12761, 97957) |
In the immediate arm, lower age, IDU/heterosexual female risk groups, lower education status, higher CD4 count, lower viral load and receiving ZDV and 3TC as the NRTI backbone were significantly associated with treatment interruptions in univariable Cox proportional hazards model (Table 2, immediate arm, univariable analyses). Region was also significantly associated with treatment interruptions; those from Africa, Latin America, United States and Asia were at an increased risk of treatment interruption compared to those from Europe and Israel. In multivariable analyses, only age, education status, region and NRTI backbone remained significantly associated with treatment interruptions (Table 2, immediate arm, multivariable results). Similar univariable results were seen in the deferred arm, albeit with stronger associations between the factor of interest and the risk of treatment interruption (Table 2, deferred arm, univariable results). In multivariable analyses, only age and education status were significantly associated with treatment interruptions (Table 2, deferred arm, multivariable results).
Table 2:
Univariable and multivariable hazard ratios for treatment interruptions stratified by randomisation arm
Immediate arm | Deferred arm | ||||||||
---|---|---|---|---|---|---|---|---|---|
Univariable | Multivariable | Univariable | Multivariable | ||||||
Hazard ratios (95% CI) | P-value | Hazard ratios (95% CI) | P-value | Hazard ratios (95% CI) | P-value | Hazard ratios (95% CI) | P-value | ||
Age1 | <35 years | 1 | 1 | 0.01 | 1 | 0.01 | 1 | 0.01 | |
35–50 years | 0.74 (0.58, 0.95) | 0.01 | 0.75 (0.59, 0.97) | 0.53 (0.33, 0.84) | 0.54 (0.34, 0.88) | ||||
>50 years | 0.58 (0.38, 0.90) | 0.52 (0.33, 0.80) | 0.49 (0.24, 0.99) | 0.43 (0.21, 0.90) | |||||
Gender/mode | IDU | 2.34 (1.19, 4.59) | 0.001 | 1.63 (0.80, 3.30) | 0.46 | 3.40 (1.06, 10.95) | 0.001 | 3.30 (0.95, 11.50) | 0.17 |
of acquisition1 | MSM | 1 | 1 | 1 | 1 | ||||
Heterosexual female | 1.71 (1.31, 2.22) | 1.24 (0.88, 1.74) | 2.86 (1.75, 4.66) | 1.85 (0.98, 3.49) | |||||
Heterosexual male | 1.18 (0.82, 1.70) | 0.98 (0.66, 1.47) | 1.78 (0.90, 3.55) | 1.64 (0.78, 3.44) | |||||
Other | 1.09 (0.62, 1.92) | 0.97 (0.54, 1.76) | 1.87 (0.74, 4.73) | 1.57 (0.61, 4.09) | |||||
Education1 | Less than high school | 1 | <0.0001 | 1 | <0.0001 | 1 | 0.0001 | 1 | 0.03 |
High school graduate | 0.63 (0.47, 0.84) | 0.48 (0.35, 0.66) | 0.52 (0.30, 0.92) | 0.58 (0.31, 1.07) | |||||
Completed vocational training | 0.42 (0.27, 0.68) | 0.45 (0.27, 0.76) | 0.35 (0.16, 0.77) | 0.51 (0.22, 1.16) | |||||
Some college/university | 0.49 (0.34, 0.69) | 0.35 (0.24, 0.52) | 0.49 (0.27, 0.88) | 0.50 (0.25, 0.98) | |||||
Bachelor’s degree | 0.25 (0.16, 0.39) | 0.22 (0.14, 0.36) | 0.16 (0.07, 0.37) | 0.22 (0.09, 0.53) | |||||
Any post-grad education | 0.26 (0.12, 0.55) | 0.23 (0.10, 0.50) | 0.28 (0.10, 0.78) | 0.42 (0.14, 1.28) | |||||
Region1 | Africa | 2.16 (1.48, 3.15) | <0.0001 | 1.04 (0.67, 1.63) | <0.0001 | 3.35 (1.86, 6.05) | 0.001 | 1.26 (0.60, 2.66) | 0.39 |
Latin America | 2.97 (2.11, 4.18) | 2.70 (1.90, 3.84) | 1.38 (0.75, 2.53) | 1.23 (0.66, 2.31) | |||||
Europe and Israel | 1 | 1 | 1 | 1 | |||||
United States | 3.14 (2.11, 4.67) | 3.16 (2.09, 4.77) | 2.29 (1.27, 4.12) | 1.99 (1.06, 3.73) | |||||
Australia | 1.86 (0.84, 4.10) | 2.03 (0.91, 4.52) | 0.98 (0.23, 4.12) | 1.15 (0.27, 4.86) | |||||
Asia | 2.97 (1.89, 4.68) | 2.23 (1.37, 3.65) | 0.90 (0.27, 2.97) | 0.81 (0.24, 2.75) | |||||
CD4 at start of ART1 | Per 50 cells higher | 1.04 (1.01, 1.07) | 0.01 | 1.02 (0.99, 1.05) | 0.21 | 1.04 (0.99, 1.09) | 0.13 | 1.02 (0.97, 1.08) | 0.35 |
Viral load at start of ART1 | Per 1 log higher | 0.83 (0.74, 0.94) | 0.004 | 0.95 (0.82, 1.09) | 0.45 | 0.87 (0.68, 1.11) | 0.25 | 1.08 (0.83, 1.40) | 0.57 |
NRTI2 | TDF + 3TC/FTC | 1.24 (0.66, 2.33) | <0.0001 | 1.54 (0.79, 3.01) | <0.0001 | 0.97 (0.39, 2.39) | 0.01 | 1.01 (0.39, 2.64) | 0.35 |
Backbone | ABC + 3TC/FTC | 1 | 1 | 1 | 1 | ||||
ZDV + 3TC | 2.77 (2.06, 3.71) | 2.14 (1.53, 3.01) | 2.78 (1.39, 5.56) | 1.90 (0.80, 4.51) | |||||
Third drug2 | EFV | 1 | 0.12 | 1 | 0.30 | 1 | 0.46 | 1 | 0.36 |
Other NNRTI | 0.44 (0.18, 1.06) | 0.52 (0.21, 1.29) | 1.19 (0.62, 2.26) | 1.39 (0.66, 2.93) | |||||
ATV | 1.28 (0.91, 1.80) | 1.37 (0.95, 1.97) | 1.66 (0.87, 3.17) | 2.10 (1.04, 4.27) | |||||
DRV | 0.71 (0.43, 1.18) | 1.06 (0.61, 1.83) | 0.99 (0.49, 1.97) | 1.29 (0.58, 2.88) | |||||
Other PI | 1.40 (0.62, 3.15) | 1.35 (0.60, 3.06) | 1.90 (0.68, 5.29) | 0.73 (0.22, 2.44) | |||||
INSTI | 0.84 (0.46, 1.55) | 1.19 (0.63, 2.24) | 0.79 (0.37, 1.69) | 1.01 (0.43, 2.36) |
Multivariable model mutually adjusted for age, gender/mode of acquisition, education, region, CD4 and viral load
Multivariable model adjusted for age, gender/mode of acquisition, education, region, CD4 and viral load
Discussion
Using data from a diverse multi-national cohort of participants in the START trial, we found that 12.9% of people who started ART in the immediate arm and 10.5% of those who started ART in the deferred arm had at least one treatment interruption after 3 years of starting ART. Treatment interruptions took place on average 5 months after starting ART and generally lasted from 3 to 15 weeks. We found that after adjusting for potential confounders, participants who started ART were more likely to experience treatment interruptions if they were of younger age and had lower education status. Region and the NRTI backbone used was significantly associated with treatment interruptions in the immediate arm but not in the deferred arm.
Our finding of a higher risk of treatment interruption (regardless of randomisation arm) amongst those of younger age is consistent with earlier studies [7–11]. It is possible that younger people are less stable both in terms of living conditions (less likely to be in permanent housing) and socially, and this may impact on taking treatment regularly.
Level of education was also independently associated with treatment interruptions in both the immediate and deferred arms. The association between education and treatment interruptions/low adherence has been shown in other studies [10, 12, 13]. Education attainment is also linked to a greater ability to deal with life events due to enhanced psychosocial skills [14]. Hence those experiencing difficult life events who have higher education attainment may be less likely to interrupt therapy than those with lower education attainment, and they may also have greater support available when such events occur. Higher education is generally seen, at least partially, as a proxy for socio-economic status and is likely to be linked to increased health literacy and in turn better adherence to medication. We note that socio-economic status is a potentially important unmeasured confounder in our study which prohibited exploration of the independent potentially causal roles of education and socio economic status.
Region was significantly associated with treatment interruption. People from the United States were over 3 times more likely to experience treatment interruptions than people from Europe and Israel, whilst those from Latin America and Asia had over double the risk of experiencing a treatment interruption. Regional differences could partially reflect differences in health care systems leading to different selection factors affecting who is enrolled. For example, within the United States, people from lower socio-economic status may have been more likely to enrol in the trial to gain access to free medication.
General adherence to ART may also explain the finding between region and interruption. In a systematic review of young adults, Kim et al found that people from North America had the lowest average ART adherence, though this was followed by those from Europe and South America, with higher levels seen in Africa and Asia [15]. A meta-analysis which included HIV-infected adults in North America and sub-Saharan Africa also showed low levels of adherence amongst people from North-America compared to sub-Saharan Africa [16]. This may be due to the epidemic being more generalised in sub-Saharan Africa and hence ART provision and adherence counselling being more widespread. In the United States, the epidemic is more focussed amongst certain communities, in which healthcare may not be fully utilised, despite the overall richer resource setting. However, it is likely that the epidemic is similarly focussed in certain communities in Europe and hence this explanation is unlikely to fully explain the association seen. Another possible explanation is that health systems within Europe place a greater emphasis on the importance of adherence and risks of interruption by delivering specific counselling to people which may not be available for a range of reasons, including financial constraints on healthcare budgets, outside of Europe. Other factors such as physician experience [17], relationship with healthcare provider [18] and mental health[19] have also been shown to be associated with the risk of treatment interruptions, whilst outside the setting of a randomised controlled trial, factors such as different health care policies and treatment options may also contribute to this risk.
Further, we cannot rule out the possibility of other unmeasured confounding; participants from countries in which a higher rate of treatment interruption was seen may have personal attributes (e.g. attitudes to health care and provision) which contribute to a higher risk of interruption but were not captured within the study.
People receiving ZDV+3TC in the immediate arm had a significantly increased risk of interruption compared with those receiving other NRTI backbone regimens. A considerable proportion of people in Latin America (20%) and Africa (8%) were receiving ZDV+3TC as their first regimen and one likely reason for discontinuation amongst these people is the poorer toxicity profile associated with ZDV compared to newer drugs[20]. Most of these people had enrolled onto the trial prior to the end of 2013 and the proportion being prescribed this backbone combination has since declined.
Whilst people interrupt treatment for a range of reasons, it is of concern that within our multi-national study in which we were able to separately analyse risk factors for treatment interruptions amongst early initiators, selected sub-groups are still more likely to interrupt treatment. Further work to confirm these findings is needed. Interrupting therapy is associated with long term poorer outcomes [3]. Improved approaches to support young people and those with lower educational attainment to sustain their taking of antiretrovirals are required.
Acknowledgements
The authors would like to thank all participants in the START trial. Please see N Engl J Med 2015; 373:795–807 for a complete list of START investigators.
All authors have read and approved the final manuscript.
LBM performed the analyses and wrote the paper.
SC, JL, CS and AP extensively reviewed the analyses and the paper
GRL, HM, IV and PP reviewed the paper
Funding
Supported by the National Institute of Allergy and Infectious Diseases (United States), Agence Nationale de Recherches sur le SIDA et les Hépatites Virales (France), National Health and Medical Research Council (Australia), National Research Foundation (Denmark), Bundes ministerium für Bildung und Forschung (Germany), European AIDS Treatment Network, Medical Research Council (United Kingdom), National Institute for Health Research, National Health Service (United Kingdom), and the University of Minnesota. Antiretroviral drugs were donated to the central drug repository by AbbVie, Bristol-Myers Squibb, Gilead Sciences, GlaxoSmithKline/ViiV Healthcare, Janssen Scientific Affairs, and Merck
Bibliography
- 1.Lundgren JD, Babiker AG, Gordin F, Emery S, Grund B, Sharma S, et al. Initiation of Antiretroviral Therapy in Early Asymptomatic HIV Infection. The New England journal of medicine 2015; 373(9):795–807. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Srinivasa S, Grinspoon SK. Metabolic and body composition effects of newer antiretrovirals in HIV-infected patients. European journal of endocrinology 2014; 170(5):R185–202. [DOI] [PubMed] [Google Scholar]
- 3.El-Sadr WM, Lundgren J, Neaton JD, Gordin F, Abrams D, Arduino RC, et al. CD4+ count-guided interruption of antiretroviral treatment. The New England journal of medicine 2006; 355(22):2283–2296. [DOI] [PubMed] [Google Scholar]
- 4.Giles ML, Achhra AC, Abraham AG, Haas AD, Gill MJ, Lee MP, et al. Sex-based differences in antiretroviral therapy initiation, switching and treatment interruptions: global overview from the International Epidemiologic Databases to Evaluate AIDS (IeDEA). Journal of the International AIDS Society 2018; 21(6):e25149. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Kranzer K, Ford N. Unstructured treatment interruption of antiretroviral therapy in clinical practice: a systematic review. Trop Med Int Health 2011; 16(10):1297–1313. [DOI] [PubMed] [Google Scholar]
- 6.UNAIDS. Blindspot: Reaching out to men and boys In; 2017.
- 7.Holkmann Olsen C, Mocroft A, Kirk O, Vella S, Blaxhult A, Clumeck N, et al. Interruption of combination antiretroviral therapy and risk of clinical disease progression to AIDS or death. HIV medicine 2007; 8(2):96–104. [DOI] [PubMed] [Google Scholar]
- 8.Moore DM, Zhang W, Yip B, Genebat M, Lima VD, Montaner JS, et al. Non-medically supervised treatment interruptions among participants in a universally accessible antiretroviral therapy programme. HIV medicine 2010; 11(5):299–307. [DOI] [PubMed] [Google Scholar]
- 9.Kavasery R, Galai N, Astemborski J, Lucas GM, Celentano DD, Kirk GD, et al. Nonstructured treatment interruptions among injection drug users in Baltimore, MD. Journal of acquired immune deficiency syndromes (1999) 2009; 50(4):360–366. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Taffe P, Rickenbach M, Hirschel B, Opravil M, Furrer H, Janin P, et al. Impact of occasional short interruptions of HAART on the progression of HIV infection: results from a cohort study. Aids 2002; 16(5):747–755. [DOI] [PubMed] [Google Scholar]
- 11.Samji H, Taha TE, Moore D, Burchell AN, Cescon A, Cooper C, et al. Predictors of unstructured antiretroviral treatment interruption and resumption among HIV-positive individuals in Canada. HIV medicine 2015; 16(2):76–87. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Burch LS, Smith CJ, Anderson J, Sherr L, Rodger AJ, O’Connell R, et al. Socioeconomic status and treatment outcomes for individuals with HIV on antiretroviral treatment in the UK: cross-sectional and longitudinal analyses. The Lancet Public health 2016; 1(1):e26–e36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Kleeberger CA, Buechner J, Palella F, Detels R, Riddler S, Godfrey R, et al. Changes in adherence to highly active antiretroviral therapy medications in the Multicenter AIDS Cohort Study. AIDS 2004; 18(4):683–688. [DOI] [PubMed] [Google Scholar]
- 14.Skalicka V, van Lenthe F, Bambra C, Krokstad S, Mackenbach J. Material, psychosocial, behavioural and biomedical factors in the explanation of relative socio-economic inequalities in mortality: evidence from the HUNT study. International journal of epidemiology 2009; 38(5):1272–1284. [DOI] [PubMed] [Google Scholar]
- 15.Kim SH, Gerver SM, Fidler S, Ward H. Adherence to antiretroviral therapy in adolescents living with HIV: systematic review and meta-analysis. Aids 2014; 28(13):1945–1956. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Mills EJ, Nachega JB, Buchan I, Orbinski J, Attaran A, Singh S, et al. Adherence to antiretroviral therapy in sub-Saharan Africa and North America: a meta-analysis. JAMA 2006; 296(6):679–690. [DOI] [PubMed] [Google Scholar]
- 17.Sangsari S, Milloy MJ, Ibrahim A, Kerr T, Zhang R, Montaner J, et al. Physician experience and rates of plasma HIV-1 RNA suppression among illicit drug users: an observational study. BMC infectious diseases 2012; 12:22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Stone VE, Clarke J, Lovell J, Steger KA, Hirschhorn LR, Boswell S, et al. HIV/AIDS patients’ perspectives on adhering to regimens containing protease inhibitors. Journal of general internal medicine 1998; 13(9):586–593. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Carrico AW, Riley ED, Johnson MO, Charlebois ED, Neilands TB, Remien RH, et al. Psychiatric risk factors for HIV disease progression: the role of inconsistent patterns of antiretroviral therapy utilization. Journal of acquired immune deficiency syndromes (1999) 2011; 56(2):146–150. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Lewis W, Day BJ, Copeland WC. Mitochondrial toxicity of NRTI antiviral drugs: an integrated cellular perspective. Nature reviews Drug discovery 2003; 2(10):812–822. [DOI] [PubMed] [Google Scholar]