Study details
The literature search yielded 4002 citations, of which 287 were duplicates, resulting in 3715 unique records. After screening
titles and abstracts, 192 potentially relevant studies were identified. Nine additional articles were found from hand searching of bibliographies. Following a careful examination of the 201 full-text articles and clinicaltrials.gov records, a total of 63 studies (34 RCT, 24 observational studies (OS) and five economic studies) met the inclusion and exclusion criteria and were included for qualitative evidence synthesis. The PRISMA flow of the review process is shown in Fig. 1. Nearly half (n = 30) of the studies were conducted in treatment naïve PLWH, 22 studies included treatment experienced PLWH and 11 included both treatment experienced and naïve PLWH.
Twenty-eight of the 63 studies were not included in the previous meta-analysis [11]. Most of the studies not included in the previous review reported efficacy (73%) and safety/tolerability (59%). Adherence was reported in 48% of the added studies. Fewer reported (or assessed) humanistic measures (patient-reported outcomes [PRO]) (14%) and economic endpoints (21%).
Key characteristics of the 63 studies are summarized in Additional file 1: Table S1 Of these 63 studies, only 14 studies reported outcome measures in evaluable format (n/N, mean, standard deviation, N, or median and inter-quartile range) and/or at consistent evaluation time points, with a clear comparison between STR and MTR, and were included for meta-analysis. Baseline demographics of the population in the 14 studies are shown in Additional file 2: Table S2.
The majority of the included RCTs had low risk of bias, with potentially high risk for blinding in treatment allocations. The overall quality of all observational studies was determined as medium and satisfactory. The detailed assessments are presented in Additional file 3: Table S3.
Adherence outcomes
While 30 of the 63 studies reported patient adherence outcomes, only eight studies reported quantifiable data and were included in the meta-analysis [17,18,19,20,21,22,23,24]. Seven of these eight studies [17,18,19,20,21,22,23] reported patient adherence as a dichotomous outcome (threshold defined per study protocol), two reported mean difference in medication adherence calculated using pill count and one reported using both formats [17, 24].
In the dichotomous adherence outcome analysis, adherence was significantly better in patients receiving STR than patients receiving once or twice daily MTR: 55.4% (range: 25.7% to 91.6%) versus 42.0% (range: 15.1% to 85.3%), odds ratio (OR) of adherence: 1.96, 95% confidence interval (CI) 1.66–2.33, p < 0.001. No heterogeneity was observed (Chi2 = 4.91; i2 = 0.0%) (Fig. 2a).
In the sub-analysis of STR versus once daily MTR, adherence was significantly better in patients receiving STR compared with those receiving once daily MTR: 80.5% (range: 76.5% to 85.4%) versus 71.8% (range: 68.0% to 75.0%), OR: 1.66, 95% CI 1.21–2.27, p = 0.002; Fig. 2a).
In the sub-analysis of STR versus twice daily MTR, adherence was numerically better in patients receiving STR compared with those receiving twice daily MTR: 84.1% (range: 82.6% to 85.4%) versus 66.8% of patients (range: 60.7% to 73.9%), OR: 2.53, 95% CI 1.13–5.65, p = 0.02; Fig. 2a).
Medication adherence based on ‘‘pill count’’ (two studies) was higher in the STR group (92.1% [range: 86.0% to 98.3%]) compared with 84.8% (range: 73.6% to 95.9%) in the collective MTR groups. The standardized mean difference (SMD) comparing medication adherence was also statistically significantly in favor of the STR group (SMD: 0.68, 95% CI 0.40–0.97, p < 0.001) in the two studies (Fig. 2b).
Efficacy outcomes
Twenty-four of the 63 studies reported efficacy data for viral load suppression and CD4 count. Eighteen studies were excluded from the meta-analysis since they reported time points other than 48 weeks or parameters not in a quantifiable format. Six studies [22, 25,26,27,28,29] provided analyzable data for viral load suppression (defined per protocol < 50 copies/ml) at 48 weeks and four studies [25,26,27, 30, 31] reported change in CD4 cell count at 48 weeks. There was a statistically significant difference in viral load suppression at 48 weeks between the STR and MTR groups (relative risk (RR): 1.05, 95% CI 1.02–1.09, p = 0.002) and low heterogeneity between the studies was observed (Chi2 = 7.54; i2 = 33.7%) (Fig. 3a). The difference (SMD) in CD4 cell count between STR and MTR was not statistically significant at 48 weeks (SMD: 0.029, 95% CI: − 0.06–0.12, p = 0.51), and no heterogeneity between the studies was observed (Chi2 = 1.76; i2 = 0.0%, Fig. 3b).
Safety and tolerability outcomes
Of the 63 studies, 37 reported safety outcomes with data relevant to adverse events (AE), laboratory abnormalities, mortality, and tolerability (treatment discontinuation). Six studies [22, 25,26,27,28, 32] reported analyzable data for the safety and tolerability outcome parameters. All six studies [22, 25,26,27,28, 32] reported rates of discontinuation due to any reason, three reported protocol-defined SAE [22, 25, 26], five reported Grade 3 to 4 AE [22, 25, 26, 28, 32], two reported Grade 3 to 4 laboratory abnormalities [25, 32] and two reported mortality [25, 26].
Meta-analyses of SAE, grade 3 to 4 AE and mortality revealed no statistically significant differences between STR and MTR groups (Fig. 4a) with RR of any SAE (RR: 0.96, 95% CI 0.64–1.45, p = 0.86), Grade 3 to 4 AE (RR: 0.83, 95% CI 0.59–1.17, p = 0.29) and mortality (RR: 0.49, 95% CI 0.05–4.65, p = 0.53) and minimal heterogeneity among the studies (Chi2 = 0.17–4.55; i2 = 0.00–12.17%). Grade 3 to 4 laboratory abnormalities were significantly less likely for the STR group versus the collective MTR groups, RR: 0.68, 95% CI 0.49–0.94, p = 0.02, with no heterogeneity in the studies.
The risk of discontinuation due to any reason was lower in the STR group versus the MTR group, RR: 0.69, 95% CI 0.47–1.00, p = 0.05, —. 4B. Moderate heterogeneity was observed in the tolerability studies (Chi2 = 9.71, i2 = 48.49%), potentially due to variation in study design and/or population.
Humanistic or patient-reported outcomes (PRO)
Seven [19, 20, 25, 27, 33,34,35] of the 63 studies reported evaluable results of PRO associated with ART. The most commonly used instruments in the studies included the HIV Treatment Satisfaction Score (HIV-TSQ), HIV Symptom Index (HIV-SI) and the 36-Item Short Form Health Survey (SF-36). Variability among the instruments used meant that the meta-analysis could not include PRO data. Five studies [25, 27, 33,34,35] reported positive impact with results favoring STR in patient satisfaction, symptom control and overall health status. Among these five studies, the impact of STR or MTR on mental health was rated as neutral. Among the five, only one directly compared STR and MTR [25]. In this study, STR was associated with a higher HIV-TSQ score as compared to MTR at weeks 4 (21.5 versus 13.3, p < 0.0001) through week 24 (23.1 versus 14.5, p < 0.0001) [25]. Further, STR resulted in lower rates of patient-reported AE such as diarrhea (30% versus 46%, p < 0.001) and bloating (33% versus 41%, p = 0.039) compared with MTR (p < 0.05 for both) [25]. Similar improvements in the STR and MTR arms in the other two studies, regardless of pill burden [19, 20].
Economic summary
Thirteen [23, 36,37,38,39,40,41,42,43,44,45,46,47] of 63 studies reported economic outcomes. Ten studies were summarized in the previous review [11]. Three additional studies were identified and are reported in this publication [23, 44, 45]. One study, evaluating overall ART changes in 3850 PLWH, found that modifying therapy resulted in a mean additional cost of €14 (SD €216; range −€528 to +€831) per month per patient [44]. Toxicity and therapy simplifications were cited as the leading causes for regimen changes [44]. Economic outcomes have been modeled using simulations and insurance claims data including comprehensive computer-based microsimulation to compare the cost-effectiveness of STR to MTR for initial treatment and concluded that the ICER of STR to MTR is $26,383 per quality-adjusted life year [45]. A multivariate regression model study using Medicaid medical and pharmacy claims data was completed to determine the impact of ART pill burden in 2174 PLWH [23]. Patients taking STR had a lower risk of hospitalization (HR: 0.71, 95% CI 0.59–0.86, p = NS) and extended time to hospitalization (median: 1508 versus 1032 days, p = 0.0042) [23].