HIV Diagnosis Time Differentiates ART Initiation: Implications from a Prospective Cohort Study in China

Background: We estimated the predictive effects of ART-related perceptions on the actual ART uptake behavior among ART naïve PLWH stratied by different time of HIV diagnosis under the new strategy. Methods: A prospective cohort study was conducted among ART naïve PLWH in Guangzhou, China from June 2016 to June 2017. Cox regression model was used to evaluate the predictive effects of ART-related perceptions on ART initiation among PLWH stratied by different timepoint of HIV diagnosis (i.e., before or after the update of the new treatment policy). Results: Among 411 participants, 150 and 261 were diagnosed before (pre-scaleup group) and after (post-scaleup group) the implementation of the new strategy, respectively. The ART initiation rate in the post-scaleup group (88.9%) was higher than that in the pre-scaleup group (73.3%) (p<0.001). A signicant difference of mean score was detected in each HBM construct between pre- and post- scaleup groups (p<0.05). After adjusting for signicant background variables, among all participants, only the self-ecacy (adjusted HR (HRa) =1.23, 95% C.I.: 1.06 to 1.43, p=0.006), has a predictive effect on ART initiation; in pre-scaleup group, all constructs of HBM-related ART perceptions were predictors of ART initiation (HRa=0.71 to 1.83, p<0.05), while in post-scaleup group, no signicant difference was found in each construct (p>0.05). Conclusions: The ART initiation rate was high particularly among participants who diagnosed after the new treatment strategy. The important role of the time of HIV diagnosis on ART initiation identied in this study suggested that future implementation interventions may consider to modify the ART-related perceptions for HIV patients who diagnosed before the implementation of the new ART strategy, while expand the accessibility of ART service for those who diagnosed after the implementation of the new strategy. Univariate cox regression models tted to assess the associations between the background variables and ART initiation in all participants. Variables with a p < 0.1 were included as candidates in the multivariate stepwise cox regression models, and variables with a p < 0.05 were dened as potential confounders. The associations between HBM-related ART perceptions and ART initiation was calculated by both univariate and multivariate analysis (adjusting for potential confounders). The same method was used for subgroup analysis. The results were presented as hazard ratios (HR) with 95% condence intervals (C.I.). The signicance of interaction effects of time of HIV diagnosis variable on the associations between the HBM-related ART perceptions and ART initiation was tested using multiple cox regression models. A total of 5 interaction models were tted (each construct of HBM-related ART perceptions × time of HIV diagnosis). SPSS for Windows (version 25.0; SPSS, Chicago, Illinois) was used for statistical analyses. A p value < 0.05 was considered statistically signicant.


Introduction
The HIV epidemic remained a global challenge [1]. Based on recent evidence that early antiretroviral therapy (ART) initiation was effective in preventing HIV transmission and slowing the progression of HIV-related diseases as compared to delayed treatment [2][3][4], WHO proposed the 'Treatment as Prevention' strategy in 2013, and recommended immediate ART for all people living with HIV (PLWH) irrespective of their CD4 counts (CD4) in 2015 [5]. Globally the ART coverage was 59% in 2017 [1], and the rate varies across countries (37-72%) [6]. In China, by the end of September 2018, the reported number of PLWH was 850,000, and about 80,000 people are newly diagnosed every year [7]. In June 2016, China launched the new ART strategy which provides free ART irrespective of clinical stage and CD4 count and immediate initiation after diagnosis to all PLWH (hereinafter referred to as 'the New Strategy') [8].
In addition, the evolution of ART guidelines from advocating a cautious application to a near universal approach may also have crucial improvement on ART initiation. A study in Ethiopia found that PLWH who were diagnosed after year 2008 (WHO guideline increased the threshold of CD4 to initiate ART from 200 to 350 cells/ul in 2008) were more likely to initiate ART than those diagnosed before 2008 (18). Another study in South Africa found that among PLWH with CD4 between 201 to 350 cells/ul, those diagnosed after 2011 (South Africa guideline increased the threshold of CD4 to start ART from 200 to 350 cells/ul in 2011) had higher ART initiation rate than those previously diagnosed [19]. Research in China among HIV positive adolescents [20] and MSM [21] showed similar results. These cohort studies mentioned above con rmed the positive effects of broadened spectrum of eligible individuals on improving timely treatment and ART coverage.
However, there is still a signi cant number of PLWH who had not initiated ART despite the scale-up strategy, leading to a gap to the UNAIDS 90% ART coverage goal [22]. How these potential ART users (e.g., past and recent diagnoses) perceive the treatment is essential to the implementation of the new strategy, as perceptions are known to be strong predictors of actual behaviors [14]. According to literature review, most existing research among ART naïve PLWH focused on the intention of early ART. For example, research in UK found that 43-47% ART naïve PLWH were willing to start ART [23]; research in Australia found that 72% of HIV positive men who have sex with men (MSM) believed the necessity to start ART as early as possible [24]. Studies in China also reported that early ART was acceptable for over 60% of HIV positive MSM [21,25,26].
However, the above studies focused on ART intention and perceptions through cross-sectional designs. So far there is a lack of studies investigating how ART-related perceptions predict actual ART initiation behaviors prospectively after the implementation of new ART strategy.
Behavioral theories help people understand, interpret and improve health-related behaviors [27,28]. There is a number of behavioral theories being successfully applied in HIV/AIDS area, among which the Health Belief Model (HBM) is one of the most widely used. The HBM has been applied to study late ART initiation [29], ART adherence [30,31], HIV testing [32], and was also successfully used to guide behavioral interventions [33,34]. According to HBM, perceived severity, perceived susceptibility, perceived barriers, perceived bene ts and self-e cacy are the key constructs, which can be used as potential predictors of ART initiation [27,28]. How ART naïve PLWH perceive the treatment, whether different subgroups (e.g., PLHW diagnosed before and after the implementation of new strategy) have different perceptions, and how such differences affect ART initiation behavior? Answers to these questions are warranted when implementing the new ART strategy.
This cohort study was conducted right after the New ART strategy (universal ART) being implemented. We aimed to investigate ART-related perceptions, ART initiation behaviors and the predictive effects of perceptions on ART initiation among ART naïve PLWH in China. As the time of HIV diagnosis (before or after the implementation of the new strategy) may affect PLWH's perceptions toward ART and may lead to different behaviors, we also conducted the subgroup analyses and examined the interactions between perceptions and the time of diagnosis on ART initiation.

Study design and population
This cohort study was conducted from June 2016 to June 2018 in Guangzhou, the capital city of Guangdong Province (China started the new ART strategy since June 1st, 2016) (8). By the end of 2015, 7,311 HIV cases had been reported in Guangzhou and 5,056 PLWH (69.2%) were receiving ART (Guangzhou Yearbook, 2016). Six out of 11 districts with the highest HIV prevalence in Guangzhou were selected as the study sites, among which 3 were urban and 3 were suburb districts. All participants were recruited during the rst year of the study period and followed up for at least 12 months to assess ART initiation behaviors.
The study population comprised of individuals who were aged ≥ 18 years, con rmed as HIV positive, clinically eligible for ART, never experienced ART before, and could provide informed consent. In total, 439 eligible participants were recruited, of which 411 (93.6%) nished the baseline survey. Ethical approval was obtained from the Ethics Committee of the School of Public Health, Sun Yat-sen University.

Data collection
According to National Practical Guideline for the Follow-up and Management of PLWH, district-level Centers for Disease Control and Prevention (CDC) or Community Healthcare Centers (CHCs) (sub-district level) are responsible for providing routine care of PLWH in community, including regular follow up and health education [35]. In this study, the trained health care workers at CDC or CHCs contacted all PLWH in the jurisdiction who met the inclusion criteria and invited them to participate in the study. Potential participants were briefed about the study design, ensured the con dentiality issues, and emphasized that refusal to participate would not affect their access to any regular health care services. After informed consent, face-to-face interviews were administered in separate private rooms using structured questionnaires.

Background information
Background information included socio-demographic characteristics (e.g., age, gender, income) and HIV-related health status (e.g., time of HIV diagnosis, latest CD4 count) ( Table 1). Participants were dichotomized by different time of HIV diagnosis into pre-scaleup group (before June 1st, 2016: the implementation of the new strategy) and post-scaleup group (after June 1st, 2016).

HBM-related ART perceptions
The HBM-based items of ART perceptions were developed through literature review, in-depth interviews with 15 PLWH, and focused group discussions. The response categories of the questions in the rst four constructs were rated on a ve-point Likert scale (1='extremely disagree', 5='extremely agree.').
Perceived severity was measured by four items, referring to the individual's recognition of the adverse consequences of not being treated in time (e.g., 'If you do not get treatment in time, the chance of virological failure will increase in the future').
Higher scores represented stronger belief in the serious consequences if not starting ART.
Perceived susceptibility in this study was the likelihood of transmitting HIV to one's HIV-negative sexual partner(s) and was measured by 'Non-timely participation in treatment will increase the chances of HIV transmission to sexual partners.' A higher score indicated stronger belief in the chances of HIV transmission if not starting ART.
Perceived bene ts were measured by six items, referring to the individual's recognition of the bene ts of timely participation in treatment (e.g., 'Starting ART may increase my CD4 counts effectively'). Higher scores indicated higher perceived bene ts of starting ART.
Perceived barriers were measured by six items, referring to obstacles that an individual think they may encounter if they initiate ART (e.g., 'Early ART may lead to early side effects'). Higher scores indicated stronger belief in barriers if starting ART.
Self-e cacy was measured by ve items, referring to the individual's con dence in their ability to participate in treatment successfully (e.g., 'How con dent are you that you will schedule your personal and work life in order to start ART?').
Responses were measured on a ve-point scale ('not', 'a little', 'not sure', 'some' and 'very'). Higher scores represented a higher level of self-e cacy to start ART immediately.
The above scales showed good internal consistency with Cronbach's alpha coe cients ranging from 0.79 to 0.92 (Table 2). For item analyses, the responses were dichotomized (e.g., agree/extremely agree, neutral/disagree/extremely disagree).

Statistical analysis
In between-group comparison of the background information (participants diagnosed before or after the new strategy), t test, Chi-square or rank sum test was performed accordingly. Kaplan-Meier method was used to calculate the treatment rate of the two subgroups and Log-rank test was used to test the difference.
Univariate cox regression models were tted to assess the associations between the background variables and ART initiation in all participants. Variables with a p < 0.1 were included as candidates in the multivariate stepwise cox regression models, and variables with a p < 0.05 were de ned as potential confounders. The associations between HBM-related ART perceptions and ART initiation was calculated by both univariate and multivariate analysis (adjusting for potential confounders). The same method was used for subgroup analysis. The results were presented as hazard ratios (HR) with 95% con dence intervals (C.I.

ART-related perception
The difference of mean scores is signi cant in each construct of HBM-related ART perception between pre-and post-scaleup groups (p < 0.05). Speci cally, the participants in the post-scaleup group perceived higher severity (p < 0.001), higher susceptibility (p = 0.001), higher bene ts (p = 0.012), while lower perceived barriers (p = 0.028) and higher self-e cacy (p < 0.001) than those in the pre-scaleup group. Regarding speci c items of HBM-related ART perception scale, 69.

The predictive effect of background variables on ART initiation
In univariate analysis of all participants, predictors of ART initiation include: older age, married/cohabited with others, living in the suburb area, higher CD4 levels, self-reported as poor health status, and diagnosed HIV after the new strategy (Table 3).
In multivariate analysis (data not tabulated), participants in the post-scaleup group were more likely to initiate ART than those in the pre-scaleup group (multivariate HR (HR m ) = 2.08, 95% CI: 1.65 to 2.64,p < 0.001).    Table 3.
In univariates and adjusted analyses of post-scaleup group, none of the constructs of HBM-related ART perceptions has predictive effect on ART initiation (p > 0.05) ( Table 4).

Discussion
This study was one of the rst studies to investigate the ART initiation in a cohort among ART naïve PLWH. The results revealed that ART initiation rate was differentiated by the HIV diagnosis period and a great gap of the median time interval from enrollment to ART initiation was also identi ed between the two subgroups. In comparison, the ART initiation rate in the post-scaleup group is close to the 90% target, but the pre-scaleup group still has a large gap. Developing targeted interventions for ART naïve PLWH who were diagnosed before the new strategy might be bene cial for achieving the second 90 target-90% ART coverage.
The high levels of HBM-related ART perceptions in this study may possibly be explained by the increased evidence of less side effect and more health bene ts of ART [3,4]. We found that the participants in the post-scaleup group have more favorable ART-related perceptions than those in the pre-scaleup group in all constructs. Several reasons might account for such difference. First, the post-scaleup group received the latest and consistent ART knowledge, while the pre-scaleup group experienced multiple changes in treatment strategies, and ART knowledge obtained may be diverse and inconsistent. Second, the participants in the pre-scaleup group might still under the impression that early treatment was harmful to health and may lead to early drug resistance even if the treatment strategy changed. It may be hard to change the established perception.
Third, due to the possible inadequate health promotion of the new strategies, the previously diagnosed PLWH might not comprehensively and accurately understand the new strategies. Therefore, it is necessary to conduct in-depth qualitative and quantitative research to understand the perceptions of HIV patients who diagnosed before the new strategy and to guide the future intervention designs to promote treatment.
Previous studies suggested that ART-related perceptions can be modi ed to promote the actual ART initiation behavior [13,14,29,[36][37][38][39][40]. The different perception levels between the two subgroups might primarily contribute to different ART initiations. Therefore, forming favorable ART-related perceptions for the pre-scaleup group may have a positive effect on ART initiation. Future research may consider using HBM for guiding intervention designs [31], such as delivering the information based on the mobile platform [41], designing short and user-friendly articles and vivid videos [42], developing a website and telephone service platform to provide more convenient consultation and support for PLWH [43], promoting advocacy in speci c groups through peer support [44,45], promoting ART initiation by supporting medical staff with knowledge, training and management skills and integrating such skills into PLWH's routine care service [46].
In addition, we found that HBM-related ART perceptions had different predictive effects on the ART initiation in different subgroups. In the pre-scaleup group, HBM-related ART perceptions have a predictive effect on the ART initiation, but in the post-scaleup group, there was no signi cant predictive effect. The reason may be that post-scaleup group has reached a higher level of HBM-related ART perceptions and the variation is not enough to explain the treatment behavior, and the prescaleup group experienced multiple changes in treatment strategies and the perception variation was large [47,48]. Therefore, under the new strategy it may be critical to promoting the proper perceptions for the patients who diagnosed before the new ART strategy. For the post-scaleup group, implementing the immediate ART initiation program [49], increasing the accessibility of treatment could promote the ART coverage in suburban area, given that PLWH who were diagnosed with HIV within15 days (91.5% vs 77.1%) and those living in urban (97.2% vs 80.5%) have higher treatment rate than others.
We found an interaction effect of the time of HIV diagnosis and HBM-related ART perceptions on ART initiation. According to the interaction diagram, no interaction effect of HBM-related ART perceptions and time of HIV diagnosis was found in postscaleup group, but such interaction effect was suggested in the pre-scaleup group, which is consistent with the analyses of the effects of HBM-related ART perceptions on treatment behavior in subgroups.
In this study, we didn't nd the predictive effect of CD4 levels on ART initiation among all participants and subgroup analysis. Although we observed that the higher latest CD4 levels had the lowest treatment rate of all PLWH, such association was not signi cant after adjusting for the time of HIV diagnosis. This result is inconsistent with previous studies which showed that higher CD4 levels was negatively associated with ART initiation [12]. The unobserved treatment disparity of different CD4 groups in initiating ART is quite promising.
This study has several limitations. First, we only analyzed the predictive effects of background variables and HBM-related ART perceptions on ART initiation and did not involve other factors such as psychosocial variables. Future research may consider these aspects. Second, the study de ned the time to ART initiation as the interval between baseline enrollment and actual ART uptake, instead of the time interval from HIV diagnosis to ART initiation. It may not be able to directly examine the impact of new strategy on ART uptake behaviors. Third, we analyzed the ART initiation and its predictors in the two subgroups, but the imbalance in the number of cases in two subgroups may cause insu cient e cacy. Finally, the data collected by the baseline survey was self-reported data and there may be reporting bias.

Conclusions
In summary, ART initiation, HBM-related ART perceptions and predictive effect of perceptions on ART initiation were different among PLWH diagnosed at different time periods. ART coverage was high under the new strategy, especially among participants who diagnosed after the new treatment strategy. However, a gap still exists referring to the 95% treatment rate target in 2030 [50], especially in the pre-scaleup group. The important role of the time of HIV diagnosis on ART initiation identi ed in this study suggested that future implementation interventions may consider to modify the ART-related perceptions for HIV patients who diagnosed before the implementation of the new ART strategy, while expand the accessibility of ART service for those who diagnosed after the implementation of the new strategy.

Consent for publication
Not applicable.
Availability of data and materials The datasets used and/or analysed during the current study is available from the corresponding author on reasonable request.

Competing interests
YTH is a member of the editorial board of BMC Infectious Diseases and others authors have no competing interests to declare.

Funding
The study was supported by the National Natural Science Foundation Project (