The impact of specific HIV treatment-related adverse events on adherence to antiretroviral therapy: a systematic review and meta-analysis

Abstract

Poor adherence to antiretroviral therapies (ARTs) in human immunodeficiency virus (HIV)-infected patients increases the risk of incomplete viral suppression, development of viral resistance, progression to acquired immune deficiency syndrome and death. This study assesses the impact of specific treatment-related adverse events (AEs) on adherence to ART in the adult HIV patient population. A systematic review of studies involving adult HIV-infected patients aged ≥ 16 years that reported an odds ratio (OR) for factors affecting adherence to ART was conducted through a search of the EMBASE(®) and Medline(®) databases. Database searches were complemented with a search of titles in the bibliographies of review papers. Studies conducted in populations limited to a particular demographic characteristic or behavioural risk were excluded. To qualify for inclusion into a meta-analysis, treatment-related AEs had to be defined similarly across studies. Also, multiple ORs from the same study were included where study sub-groups were distinct. Random effects models were used to pool ORs. In total, 19 studies and 18 ART-related AEs were included in meta-analyses. Adherence to ART was significantly lower in patients with non-specific AEs than in patients who did not experience AEs [OR = 0.623; 95% confidence interval (CI): 0.465-0.834]. Patients with specific AEs such as fatigue (OR = 0.631; 95% CI: 0.433-0.918), confusion (OR = 0.349; 95% CI: 0.184-0.661), taste disturbances (OR = 0.485; 95% CI: 0.303-0.775) and nausea (OR = 0.574; 95% CI: 0.427-0.772) were significantly less likely to adhere to ART compared to patients without these AEs. Knowledge of specific treatment-related AEs may allow for targeted management of these events and a careful consideration of well-tolerated treatment regimens to improve ART adherence and clinical outcomes.

Figure 1.

Flow of citations through the screening process. ∗The population of interest was originally patients with chronic diseases; this was later restricted to patients with HIV.

Figure 2.

Forest plot of meta-analyses of ORs pertaining to the effects of treatment-related general AEs on adherence. Each OR represents a pooled estimate for the corresponding adverse health outcome. All meta-analyses were conducted using a random effects model. ∗This OR was calculated using multiple studies and data from mutually-exclusive sub-groups within the same study. CI, confidence interval; I², heterogeneity index; OR, odds ratio.

Figure 3.

Forest plot of meta-analyses of ORs pertaining to the effects of treatment-related mental health AEs on adherence. Each OR represents a pooled estimate for the corresponding adverse health outcome. All meta-analyses were conducted using a random effects model. CI, confidence interval; I², heterogeneity index; OR, odds ratio.

Figure 4.

Forest plot of meta-analyses of ORs pertaining to the effects of treatment-related sensory AEs on adherence. Each OR represents a pooled estimate for the corresponding adverse health outcome. All meta-analyses were conducted using a random effects model. ∗This OR was calculated using multiple studies and data from mutually-exclusive sub-groups within the same study. ‡This OR was derived from mutually-exclusive sub-groups in a single study. CI, confidence interval; I², heterogeneity index; OR, odds ratio.

Figure 5.

Forest plot of meta-analyses of ORs pertaining to the effects of treatment-related gastrointestinal AEs on adherence. Each OR represents a pooled estimate for the corresponding adverse health outcome. All meta-analyses were conducted using a random effects model. ∗This OR was calculated using multiple studies and data from mutually-exclusive sub-groups within the same study. †This OR was derived from mutually-exclusive sub-groups in a single study. CI, confidence interval; I², heterogeneity index; OR, odds ratio.