Study setting and population
This was an analysis of prospectively collected data from a multicenter observational cohort of PLHIV on ART (SATREPS-HIV). This cohort was established under a Japanese government program, the “Science and Technology Research Partnership for Sustainable Development (SATREPS),” with the primary objective of assessing changes in treatment outcomes with the transition to SHI-based HIV services in Vietnam. One national-level hospital (National Hospital for Tropical Diseases, NHTD) and ten local ART facilities (seven provincial/city-level and three district-level facilities) in North Vietnam were included in the SATREPS-HIV (Supplementary Table S1). These ten local facilities comprised Dong Da General Hospital (DDGH), 09 Hospital, Nam Tu Liem Health Center (NTL), Quang Ninh General Hospital (QNGH), Hung Yen Hospital of Tropical Diseases (HYTD), Hai Duong Hospital for Tropical Diseases (HDHTD), Thanh Son District Medical Center (TSMC), Yen Binh District Medical Center (YBMC), Nghe An General Hospital (NAGH), and Ha Tinh Center for Disease Control and Prevention (HTCDC). These facilities were selected in consultation with the Vietnamese Ministry of Health from multiple perspectives, including region, facility level, HIV prevalence, and support from overseas donors. Additionally, because SATREPS was implemented within the framework of the Japanese government’s official development assistance, some facilities were selected with an intention to provide technical assistance to facilities with insufficient access to HIV services, such as HIV viral load (HIV-VL) testing.
The inclusion criteria for this study were as follows: written consent given to participate in the study, age 16 years or older, Vietnamese nationality, confirmed HIV infection, receiving ART for more than 6 months, holding an SHI card, and under the lower limit (20 or 50 copies/mL) for the most recent available HIV-VL. Owing to budget constraints, we principally randomly selected 100 participants from each local facility (total of 1000 participants from ten local facilities) for enrollment. However, a few facilities (TSMC and YBMC) only had approximately 100 HIV-infected outpatients. In these cases, all patients who met the inclusion criteria were enrolled. Regarding patient recruitment at NHTD, prior to SATREPS-HIV, the National Center for Global Health and Medicine established a hospital-based cohort of PLHIV taking ART to monitor clinical outcomes at NHTD. In this cohort, 1,287 PLHIV were under followed-up as of October 2019. Therefore, for convenience, we included all patients who were participating in this cohort and met the above inclusion criteria.
Patient enrollment was conducted during two periods: from December 2019 to March 2020 at NHTD and four local facilities, and from April to September 2021 at six other local facilities (Supplementary Table S1). This was because at the beginning of the SATREPS-HIV, it was unclear how the transition to SHI-based HIV services would proceed in many facilities. Therefore, six facilities were selected while monitoring the transition progress. If the selected patients did not appear at study sites for enrollment or did not give written informed consent, study participants were randomly selected until 100 participants were enrolled.
Participants were followed up at their registered facilities every 6 months to have a blood sample collected for HIV-VL and HIV-DR testing during their regular consultations. HIV-DR testing was performed for those with an HIV-VL ≥ 1,000 copies/mL. Each facility followed two study visit cycles during the Japanese fiscal year (FY), beginning in April and ending in March of the following year. The first visit cycle fell between April to September, and the second cycle fell between October to March. However, the duration of each cycle was extended during the period when transportation was restricted owing to the COVID-19 pandemic. Nevertheless, study visits were missed by those who were unable to visit the study site during the study period or who received ART by mail or at other facilities. All data collected in SATREPS-HIV, including demographics (e.g., sex, age) and HIV-related information (e.g., route of HIV transmission, ART history), clinical information (e.g., ART history, weight, HIV-VL and HIV-DR results), and follow-up status were reported by physicians or nurses at each study site using a cloud database. As for weight, only data from NHTD were collected because a weight scale was unavailable at many other hospitals.
Outcomes
The virological outcomes of ART were assessed using HIV-VL and acquired drug resistance detected during the follow-up period. Plasma samples were collected at each study site and were transported to NHTD for HIV-VL and HIV-DR testing. The quantitative measurement of HIV-VL was carried out using two automated systems, COBAS AmpliPrep/COBAS TaqMan and cobas 6800 (Roche Diagnostics, Basel, Switzerland) in the NHTD laboratory. The results for HIV-VL were divided into four categories: <50 copies/mL, 50–199 copies/mL, 200–999 copies/mL, and ≥ 1,000 copies/mL. We examined the risk of viremia, defined as HIV-VL ≥ 200 copies/mL27,28, and its associated factors. HIV-DR genotyping was performed for study participants with treatment failure (HIV-VL ≥ 1,000 copies/mL). Sanger sequencing was conducted using the 3500 Genetic Analyzer (Thermo Fisher Scientific, Waltham, MA, USA) for the regions HIV-1 protease (amino acids 1–99), reverse transcriptase (1–560), and integrase (1–289) using the in-house method29. The Stanford HIV Drug Resistance Database (https://hivdb.stanford.edu/) was used to evaluate drug resistance mutations (DRMs) to nucleos(t)ide reverse transcriptase inhibitors (NRTIs), NNRTIs, protease inhibitors (PIs), and INSTIs.
Retention in care was assessed according to the follow-up status at each study visit. The status was coded using the following five categories: under follow-up, transfer, death, skip, and loss to follow-up. Skip was defined as one missed appointment, and loss to follow-up was defined as two consecutive missed appointments.
To assess the effectiveness of switching to DTG-containing regimens, we assessed DTG use, viremia, and their associations. Furthermore, to evaluate the tolerability of DTG-containing regimens, the reasons for discontinuation of the regimens were identified and classified into the following categories: lack of effectiveness (either virological, immunological or clinical), toxicity (i.e., renal, allergic, gastrointestinal, neuropsychiatric, hepatic, and osteoarticular dysfunction or reactions potentially owing to the prescribed drug), stockout (defined as a situation in which physicians are unable to prescribe the drug because it is out of stock), and other reasons (e.g., drug–drug interactions, pregnancy or desire to become pregnant, and patient’s or physician’s choice).
To examine weight gain after switching to a DTG-containing regimen, we assessed the average weight per visit among patients from NHTD taking the regimen.
Statistical analyses
The primary outcomes of this study were virological outcomes (HIV-VL level and acquired DRMs) and treatment retention during the follow-up period. We descriptively analyzed the prevalence of each HIV-VL level and follow-up status according to study visit, acquired DRMs (any DRM and DRMs by drug class) in patients with HIV-VL ≥ 1,000 copies/mL. Additionally, we calculated the incidence of viremia (HIV-VL ≥ 200 copies/mL) during the follow-up period in all study participants. We counted viremia developed at any time after enrollment. The follow-up time was calculated from the first day of clinical follow-up in this study up to the last clinical follow-up or the date of first viremia. Cox proportional hazard models were used to identify factors independently associated with the risk of viremia during the follow-up period. The following covariates were included in the models: facility level, sex, age, injection drug use (IDU), men who have sex with men (MSM), time from HIV diagnosis to registration (< 10 or ≥ 10 years), time from HIV diagnosis to ART initiation (< 1 month, ≥ 1 and < 6 months, ≥ 6 and < 12 months, or ≥ 12 months), the most recent CD4 count before the first visit (< 200/µL, 200–349/µL, or ≥ 350/µL), and HIV-VL at the first visit (< 20/mL, 20–199/mL, 200–999/mL, or ≥ 1000/mL). These covariates were collected at enrollment. Factors that were associated with viremia with a p-value < 0.1 in the univariable model were included in the multivariable model.
The secondary outcomes were effectiveness and tolerability of switching to DTG-containing regimens. We descriptively assessed the reasons for discontinuation of DTG-containing regimens. The incidence of viremia after DTG initiation among study participants who used a DTG-containing regimen was also calculated. In this analysis, we only included participants who were on a DTG-containing regimen for at least 12 weeks to ensure sufficient and stable therapeutic effects of DTG, and those who had an HIV-VL test result ≥ 12 weeks after starting DTG. We counted viremia developed after DTG initiation. However, because only participants who had taken DTG-containing regimens for 12 weeks or more were included in this analysis, viremia that developed within 12 weeks after DTG initiation was not counted. The follow-up time was calculated from the first day of DTG initiation to the last clinical follow-up or the date of first viremia. For study participants who discontinued DTG-containing regimens, only the period during which they were taking DTG was included in the follow-up period, and only viremia while taking a DTG-containing regimen was counted. As for examination of weight gain after switching to a DTG-containing regimen, we calculated the difference in mean weights among four visits by sex, from October 2019 to February 2023, and a paired t-test was conducted. Only patients from NHTD who had data for all visits were included in the examination. The analyses were performed using SAS 9.4 software (SAS Institute Inc., Cary, NC, USA) or Stata 16 (StataCorp, College Station, TX, USA). All tests were two-sided, with the significance level set at 5%. Missing data were excluded from the analyses.
