Prevalence of delayed antiretroviral therapy initiation among people living with HIV: A systematic review and meta-analysis

Objective HIV continues to be a global challenge. Key recommendations for HIV prevention and treatment are presented on rapid antiretroviral therapy (ART) initiation. However, several studies showed a high prevalence of delayed ART initiation. The aim of this systematic review and meta-analysis was to assess the prevalence of delayed ART initiation among HIV-infected patients globally. Methods This review summarised eligible studies conducted between January 2015 and August 2022 on the prevalence of delayed ART initiation in HIV-infected adults (age ≥ 15). Relevant studies were systematic searched through PubMed/Medline, EMBASE, Web of Science, China National Knowledge Infrastructure, Wanfang, and Chongqing VIP databases. Random-effects models were used to calculate pooled prevalence estimates. The heterogeneity was evaluated using Cochran’s Q test and I2 statistics. Moreover, potential sources of heterogeneity were explored using univariate subgroup analysis. Results Data on the prevalence of delayed ART initiation was pooled across 29 studies involving 34,937 participants from 15 countries. The overall pooled prevalence of delayed ART initiation was 36.1% [95% confidence interval (CI), 29.7–42.5%]. In subgroup analysis, the estimated pooled prevalence decreased with age. By sex, the prevalence was higher among male patients (39.3%, 95% CI: 32.2–46.4%) than female (36.5%, 95% CI: 26.9–50.7%). Patients with high CD4 cell count were more likely to delay ART initiation than those with low CD4 cell count (>500cells/mm3: 40.3%; 201-500cells/mm3: 33.4%; and ≤200cells/mm3: 25.3%). Conclusions Our systematic review and meta-analysis identified a high prevalence of delayed ART initiation. The prolonged time interval between diagnosis and treatment is a prevalent and unaddressed problem that should spur initiatives from countries globally. Further research is urgently needed to identify effective strategies for promoting the early ART initiation.


Introduction
HIV remains the leading cause of death and disability worldwide, endangering the health of the population and increasing the economic cost to health systems.By the end of 2021, 38.4 million people globally were infected with HIV, and 40.1 million people have died from AIDSrelated diseases since the beginning of the epidemic [1].Antiretroviral therapy (ART) is highly effective in slowing the progression of HIV to AIDS, reducing AIDS-related mortality and morbidity [2,3], and preventing further HIV transmission, as achieving undetectable virus means nontransmissible virus [4,5]; thus, timely initiation and full coverage of ART is necessary.However, only 87% of those who know their status have initiated ART, which is far from achieving the second '95' of the UNAIDS '95-95-95' target by 2025 [1,6].
Early initiation of treatment after diagnosis maximizes the benefits of ART at the individual and population levels.Randomized controlled trials in Uganda and South Africa demonstrated the feasibility and improvement in terms of treatment coverage and retention rates over three months with same-day initiation of ART, compared to the standard initiation group [7,8].The HPTN 052 trial in nine African countries clearly showed that early initiation of ART with CD4 counts between 350 and 550 cells /mm 3 facilitated the faster achievement of viral load suppression and reduced HIV transmission by approximately 96% compared to late ART initiation with CD4 count < 250 cells /mm 3 [9].Observational studies conducted in South Africa [10], Eswatini [11] and Ethiopia [12], which compared early treatment with standard care, reported similar results.Moreover, if ART is initiated immediately, the cost of hospitalization may be minimized in resource-poor and HIV-prone prevalence countries [13,14].As a result of this evidence, the World Health Organization (WHO) recommended in 2015 that all HIVinfected patients should start ART once they are diagnosed with HIV, regardless of stage of disease and CD4 cell count (known as "treat all").There is no reason to delay treatment decisions based on CD4 cell threshold [15,16].
However, despite knowing the benefits of early treatment and the expanding ART eligibility criteria, delayed initiation of ART persists and some countries face challenges in reaching the '95-95-95' targets [1].A retrospective study reviewing the rate of ART initiation in HIVinfected ART-naive women in Uganda, revealed delayed treatment rates of up to 72% [17].Wei et al. analyzed the timeliness of ART initiation in newly reported HIV-infected patients in Hunan Province, China, from 2014 to 2018, suggesting that 58% of patients did not initiate ART within one month of diagnosis [18].A pilot randomized controlled trial from 2015 to 2016 indicated that 67% of HIV-infected adults were linked to care and 42% initiated ART within three months of referral for treatment [19].A systematic review of HIV care retention has shown that the longer lag time between testing and treatment, the higher the rate of patients lost to ART care [20].Taking these studies together, reliable estimates of delayed ART initiation after the introduction of the "treat all" strategy and understanding various potential related factors are essential for informing ongoing and future intervention strategies to effectively improve coverage of ART.
In general, there is no unified, clear, evidence-based definition of delayed initiation of treatment.Various delayed ART initiation for time to ART initiation have been used in the studies, such as time from diagnosis to ART initiation, time from first clinical visit to ART initiation [21], time from eligibility to ART initiation [22], and time from enrolment in HIV care to ART initiation [23].Time from HIV diagnosis to commencement of ART was proposed as a more feasible and sensitive indicator of ART uptake [24].In our study, delayed ART initiation was defined as failure to initiate ART within 30 days of confirmation of HIV diagnosis, as previous studies have found that one month was sufficient for most patients to complete the steps needed to start treatment under routine care [25,26].
To the best of our knowledge, there are no current or ongoing systematic reviews of the prevalence of delayed ART initiation, although sufficient data exist to make pooled estimates and to explore the source of heterogeneity between studies through meta-analysis.To fill this critical gap, we conducted this study to estimate the global prevalence of delayed ART initiation and identify social and epidemiological factors related to delayed ART initiation in published observational studies in HIV-infected adults, by a systematic review and meta-analysis.

Search strategy
We searched three English language electronic databases, PubMed/Medline, Web of Science, and Embase, and three Chinese language Databases, CNKI (China National Knowledge Infrastructure), Wanfang, and Chongqing VIP databases.Our comprehensive search strategies combined terms and variants of HIV ("human immunodeficiency virus", "acquired immunodeficiency syndrome"), treatment (therapeutics, "antiretroviral therapy, highly active"), and delay (Time-to-Treatment, defer, same-day, early, instant, immediate, rapid, quick) using both controlled vocabularies and free text words.Search queries were optimized to fit the specific feature of each database (S1 File).The electronic database search was supplemented by scanning the reference list of included articles, previous reviews and published meta-analysis.All study steps of this systematic review and meta-analysis were conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines (S1 Table) [27].

Inclusion and exclusion criteria
Studies complying with the following inclusion criteria were included: (1) Studies published in English or Chinese between January 2015 and August 2022; (2) HIV-infected adults (�15Y); (3) Studies reported time to ART initiation (namely, time from diagnosis to ART initiation) and delayed initiation prevalence or gave sufficient information to calculate this estimate.We excluded case reports, study protocols, commentaries, correspondence, letters, reviews or systematic reviews and studies carried out with pregnant women or people with serious, opportunistic infections due to the clinical management of these patients being complex.

Selection of studies
All searched articles from different databases were exported to the Endnote reference manager, and duplicates were removed.Then, two authors (Tao and Xie) independently evaluated the potentially relevant articles by checking the titles and abstracts.Those articles that did not meet the inclusion criteria were removed, and the full texts of the remaining articles were retrieved based on the inclusion and exclusion criteria.Any discrepancies that arose during the selection process were resolved by the adjudicator (Xiao).

Data extraction and quality assessment
The following elements was extracted from included articles, with use of a preconceived data abstraction form: (1) Characteristics of the study: author name, publication year, study year, study design, country, country income and continent; (2) Characteristics of the sample: sample size, age, gender; (3) HIV-related data: CD4 cell count, the time between HIV diagnosis and ART initiation, the number of participants who delayed ART initiation.
We assessed the methodological quality of included studies using the Newcastle-Ottawa Scale [28].Though this tool, risk of bias was assessed across the domains of selection, comparability and outcome for the risk of bias revision, and high-quality articles were determined when the score of the scale is five and above out of nine scales.Two authors (Tao and Xie) independently assessed study quality, and disagreements were discussed until a consensus was reached.

Statistical analysis
Statistical analysis was performed using Stata 15.The prevalence estimates of delayed initiation of ART and their 95% confidence interval (CI) were calculated and pooled by time elapsed from HIV diagnosis to ART initiation using a random effect analysis model, accounting for the expected between-study significant heterogeneity [29].Q-tests and I 2 statistics were used to assess the level of heterogeneity among studies [30].The I 2 heterogeneity was categorized as follows: low (25-50%), moderate (50-75%) and high (>75%).Subgroup analysis was undertaken to investigate source of between-study heterogeneity, in terms of country, continent, study year, study design, sample size, age group, gender, CD4 cell count.Sensitivity analysis was carried out by serially omitting each study to assess the influence of individual studies on the robustness of the pooled estimate.Furthermore, publication bias was evaluated though visual inspection of the funnel plot and Egger's tests [31].

Study results and characteristics
The PRISMA flow diagram outlines the search strategy to identify the articles (Fig 1).In total, 34937 publications were identified, of which 34533 were excluded as duplicated or irrelevant according to title and abstract.Full-text assessment was undertaken for leaving 404 potentially articles.After a detailed full-text review, 29 eligible articles  were included in the analysis.

Prevalence of delayed initiation of ART
Analyzing the global prevalence of delayed initiation of ART among patients using the random-effects model, we found the pooled prevalence of delayed ART initiation was 36.1% (95% CI: 29.7-42.5%;I 2 = 99.8,p<0.001).

Subgroup analysis
Subgroup analyses were conducted to compare the prevalence of delayed ART initiation by regions, research design, age, gender, and CD 4 cell count (Table 2).

Sensitivity analysis
We performed a sensitivity analysis using the random-effects model to identify the effect of individual studies on the overall meta-analysis.No significant changes in the overall pooled prevalence for delated ART initiation were found on the removal of a single study (S1 Fig).

Publication bias
Visual inspection of the funnel plot of studies reporting on the delayed treatment revealed significant asymmetry (Fig 3).The egger's test showed evidence of significant publish bias (P = 0.027) which further support the finding of the funnel plot.

Discussion
The current study is the first meta-analysis using a comprehensive search strategy to explore the global prevalence of delayed ART initiation among HIV-infected patients.The studies included in this review showed a high prevalence of delayed ART initiation.Our study revealed that the pooled prevalence was 38.0%, lower than the pre-2015 delayed initiation rate ranging from 41%-85% [61][62][63][64].A previous systematic review of six studies conducted in sub-Saharan Africa in 2012 by Mugglin et al. found that the median time of the whole period from HIV diagnosis to ART initiation ranged from a few weeks to more than six months [65].In our study, the median time from HIV diagnosis to ART initiation varied from five days to 40 days, which is still a gap from the WHO recommendation of starting treatment within seven days or on the same day of confirming HIV diagnosis [66].
Although the WHO guidelines encourage early initiation of ART and HIV treatment has proven to be effective, the decision to start treatment takes into account various factors.It is increasingly recognized that, earlier ART initiation is associated with a low incidence of severe HIV morbidity, disease progression, better immune recovery and longer survival time [67].Despite this, some patients may opt to delay initiation of ART due to concerns about side effects, pill interaction and financial constraints.This is particularly true for patients who have non-AIDS-defining conditions and serious, opportunistic infections that fall outside the scope of this study [68].In addition, different countries have recommended ART initiation at varying levels of CD4 count because of medical resources and economic concerns [67,69].
In this systematic review, we identified studies from 15 countries.Surprisingly, patients in high-income countries delayed treatment at a much higher rate than patients in upper-and lower-middle income countries.This might be explained by the fact that implementing "treat all" is rapidly expanding in upper-and lower-middle income countries, 82 of which have officially adopted the recommendation to provide universal HIV treatment to HIV-infected individuals [6].In addition, many effective interventions, including simplified ART initiation procedures [8,70], short message service (SMS) health messages [71], and grassroots prevention and control service model which integrated the resources of the health and medical service network and accelerated HIV treatment services [52], have been implemented in upperand lower-middle income countries, which accelerated the initiation of treatment.The relationship between national income levels and late treatment initiation needs to be further explored.
Our systematic review found that delayed initiation of ART was more common in male patients than in females, a finding that is in line with the previous study conducted in 6 sub-Saharan African countries showing that men had a significantly higher risk of delaying treatment after enrollment in HIV care than women [72].Hibbard et al. also reported that females use HIV health services more often than males [73].However, gender inequalities and gender-based violence persist, increasing women's blocks access to services.Previous studies revealed a positive association between use of ART among women and sexual and physical partner violence [74,75].In 2021, UNAIDS Board adopted the new Global AIDS Strategy 2021-2026, End Inequalities, End AIDS, which propels the process of ending AIDS [76].Strategies and attention on tackling delayed initiation of ART should thus be focused equally on both genders.We discovered that patients younger than 30 years of age were more likely to delay initiation of treatment than patients older than 50 (47.3%VS 43.3%), which is consistent with other studies conducted among general HIV-infected patients [77,78].A possible reason for the age difference is that since HIV requires lifetime treatment, patients who start treatment at a younger age will get a longer duration of therapy, and long-term medicine would be a serious burden.Financial stress has been identified as a main barrier to initiating and adhering to HIV treatment among young adults [79].Substantial subsidies for ART drugs, as well as more social benefits for young adults, are expected to increase their affordability, especially in terms of paying out-of-pocket for prescription ART drugs, so that they can rapidly start HIV treatment and achieve higher ART coverage [80].
Our study also found that a higher CD4 cell count was related to the late onset of ART.Although the implementation of the new WHO guidelines on HIV treatment, regardless of CD4 count, was effective for timely initiation of ART, especially for patients who are unavailable to services and have a high probability of delayed or non-initiation of treatment after diagnosis [81,82], a high CD4 cell count remains a risk factor for timing initiation of treatment.This may be because many asymptomatic HIV-infected patients consider themselves to be healthy in the presence of high CD4 counts.As a result, they thought they could delay treatment until they develop HIV-related symptoms [83].Previous studies have also confirmed that feeling healthy and the late appearance of HIV-related symptoms are independent factors for the late initiation of ART [84].ART initiation counseling should highlight the benefit of early treatment and motivate them to participate in the care system.Additional support and other intervention measures to facilitate the care cascade over time were needed for those patients.
Given the current prevalence of delayed ART initiation, effective measures are needed to increase patient awareness of early treatment.Failure to initiate treatment rapidly after diagnosis hinders the effectiveness of ART as a prevention method for HIV mortality and morbidity, suggesting a great need to explore patient concerns about starting treatment and develop strategies to overcome them.Several studies have identified factors associated with delayed ART initiation, including structural and individual-level barriers [23,85,86].The inconvenience of treatment as structural barriers, side effects, stigma, and lack of support as individual-level barriers were the most common concerns about ART initiation.Increased social support from close friends and family has been shown to be effective in facilitating ART initiation [87].A systematic review suggested that home-based ART services as an alternative to a centralized medical care system were another effective way to deliver ART and increase linkage to care [88].
Despite the robust findings, our study has some limitations.First, some relevant literature may have been missed, since only English or Chinese articles were included, and grey literature was not searched.Additionally, most of data were derived from studies in China and South Africa, and the pooled estimates could not be well representative of the world.Further, some of the studies in our analysis were conducted in patients who initiated ART, and those not receiving treatment were not included in the study, which may lead to an underestimation of the rate of delayed ART initiation.Lastly, the observed heterogeneity was high in the overall analysis.Although subgroup analysis found some factors that affect the result, the extent to which we can explain observed heterogeneity is limited.

Conclusions
Our systematic review revealed a high prevalence of delayed ART initiation.Comprehensive strategies and collaboration between countries and multilateral institutions to accelerate ART initiation have become increasingly critical, and the WHO recommendation supporting rapid initiation should be implemented.The prevalence of delayed treatment was higher in younger patients and those with high CD4 counts, suggesting more effective intervention strategies to promote early treatment should be provided to those high-risk patients.
Fig 2 shows the forest plot of the 29 studies reporting the prevalence of delayed treatment.