Effectiveness and Safety of Dolutegravir Two-Drug Regimens in Virologically Suppressed People Living With HIV

A Systematic Literature Review and Meta-Analysis of Real-World Evidence

Y S Punekar; D Parks; M Joshi; S Kaur; L Evitt; V Chounta; M Radford; D Jha; S Ferrante; S Sharma; J Van Wyk; A de Ruiter


HIV Medicine. 2021;22(6):423-433. 

In This Article


Study Identification

A systematic search of Embase, MEDLINE, MEDLINE In-Process and Cochrane databases was performed to identify real-world studies evaluating the effectiveness and/or safety of DTG in virologically suppressed PLHIV switching to DTG with 3TC or RPV (published in any language between 1 January 2013 and 4 April 2020, inclusive). Abstracts published in major HIV/AIDS conference proceedings between and including 1 January 2013 and 4 April 2020 were hand-searched to supplement the literature searches. Full details of the search strategy (including search terms and strings) are presented in Table S1. Conferences included in these searches are presented in Table S2.

Following the identification and removal of duplicate publications, a two-step screening process was undertaken to identify suitable studies: step 1 – the titles and abstracts of all publications identified by the literature searches were reviewed for eligibility; step 2 – full-text copies of all relevant publications identified during step 1 were obtained and reviewed against the same eligibility criteria. Eligible populations included adult PLHIV (studies including only children were excluded); no limits were applied based on gender or race. Eligible studies included observational cohort studies (both retrospective and prospective), case–control studies, claims-database studies and case series. Eligible interventions included DTG-based drug regimens. Case reports detailing information for only one patient and case series providing evidence for four patients or less were excluded.

Linked publications were identified based on population, sites and study period. Alongside linking publications, studies were also reviewed by the team to assess whether there was potential duplication in cohorts and populations for which results were being reported. Where duplication of cohort/population was suspected, only the publication reporting the highest number of people receiving DTG + 3TC or DTG + RPV, the overarching study, was included in the analysis.

The Downs and Black assessment tool was used to assess the methodological quality of the included studies.[28]

Data were extracted from selected publications by two independent reviewers, with any discrepancies resolved by a third reviewer. Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) data extraction and reporting guidelines were observed.[29–31]


Outcomes included in the meta-analysis were the proportion of patients with viral failure (VF), VS using a snapshot algorithm (VSS), VS on treatment (VSOT), and the proportion of patients discontinuing treatment at weeks 48 and 96. VSS was calculated as follows: intent-to-treat (ITT) population – (VF + discontinuations). This analysis was conducted to supplement the results reported in the identified publications, aiming to overcome potential biases of overestimating the proportion of virally suppressed population, as many of the studies did not account for participants who were lost to follow-up or discontinued. The VF was defined as plasma viral load ≥ 50 copies/mL obtained in two consecutive measurements and/or > 1000 copies/mL obtained in a single measurement. The VSOT was defined as the proportion of patients achieving a specific reduction in HIV RNA copies/mL (usually < 50 copies/mL in accordance with FDA guidance[32]), over the number patients who remain on DTG + 3TC or RPV group at weeks 48 and 96.

Data Analysis and Quality Assessments

Extracted data (including VS) were analysed using a fixed-effects model assuming asymptotically normally distributed variance and a random-effects model using restricted maximum likelihood or Bayesian methods to estimate variance. Root mean square error (RMSE) criteria were used to determine the best-fit model and guided the choice of the model used for the estimates. Statistical heterogeneity between studies was assessed using the following equation: I 2= (Q – df/Q) × 100%, where I is the level of inconsistency, Q is the χ 2 statistic and df the degrees of freedom. Publication bias for each outcome was analysed using a funnel plot (sample size vs. estimated effect size) and effect modification (i.e. between-study variations in treatment duration, dosage) analysed using meta-regression. Data are presented as the proportion of patients (percentage), standard error (SE; for study data), 95% CI (for meta-analysis data), weighting (for fixed and random effects), funnel plot asymmetry (P-value) and heterogeneity (I 2).