The secular trend in CD4+ T cell count at initiation of antiretroviral therapy in China CURRENT

Background: We aimed to look at the trends in CD4 cells counts (at initiation of ART) over calendar years and to investigate its influential factors in Chinese populations. Methods: We searched PubMed, Wanfang, Chinese Biomedical Literature Database, and China National Knowledge Infrastructure to acquire relevant papers published from January 2002 to Twelve 2015. We abstracted the mean or median CD4+ T cell counts from the included studies. Random-effect meta-regressions were used to estimate the mean CD4+ T cell counts across studies and corresponding 95% CIs. Results: A total of 220 eligible articles, which included 426283 HIV-positive individuals, were identified in this meta-analysis. The mean CD4+ T cell counts at ART initiation increased from 136.63 (95%CI=120.63–152.63) cells/µL in 2003 to 199 (95%CI=188.38– 209.62) cells/µL in 2014. In the univariate meta-regression model, we found that the estimated change in CD4+ T cell counts at ART initiation was +8.73 cells/μL per year (95% CI= 4.57 to 12.90 cells/μL per year). The highest CD4+ T cell counts at ART initiation (441.00 cells/μL, 95% CI=438.96-186.33 cells/µL) was reported in serodiscordant couples, followed by pregnant women (385.82 cells/μL, 95% CI=327.19-444.44 cells/µL), MSM (293.78, 95% CI=126.59-460.96 cells/µL), and children267.36 cells/μL, 95% CI=126.65-408.07 cells/µL. Conclusions: In China, the CD4+ T cell counts at ART initiation have dramatically increased during the past decade. However, there remains a substantial gap between CD4+ T cell counts at ART initiation and the WHO updated HIV treatment guidelines in some groups, highlighting the need for strategies to improve earlier ART initiation.

been reported in China, and the reported patients continued to increase [1]. Although the HIV/AIDS epidemic remains a low-prevalence trend as a whole, the people living with HIV/AIDS have concentrated in some high-risk groups in China. In 2010 an estimated 9.08% of injecting drug users and 5.98% of men who have sex with men(MSM) were infected with HIV [2]. Antiretroviral therapy (ART) has greatly reduced HIV-attributable mortality and morbidity and improved the quality of life of HIV-infected patients. Evidence also indicates that ART is the most effective prevention strategy, which can effectively reduce the likelihood of transmitting HIV to others [3], 'Treatment as prevention' has been proposed because recent evidence has demonstrated that expanded ART coverage among HIV-infected individuals was associated with lower transmission rates at a population level [4].
Early initiation of ART with a higher CD4 cell counts (> 500 cells/µL) is a good predictor of successful ART. In 2015, the Strategic Timing of AntiRetroviral Treatment (START) trial was stopped early because the results showed that risk of developing serious illness or death was reduced by 53% due to the early initiation of ART [5]. Another randomized clinical trial had also reported a reduction in the hazard of death, AIDS, or a serious clinical condition of 40% if they start taking antiretroviral drugs earlier [6]. On the basis of these evidences, World Health Organization (WHO) has recently issued new ART guidance which recommends starting ART in HIV-positive adults with a CD4 + T cell counts ≤ 500 cells/µL [7].
In 2014, the Chinese government also adjusted the criteria for initiation ART of adults and children living HIV/AIDS, so ART can now be initiated at CD4 cell counts of 500 cells per µL or less [1]. However, despite these changes in worldwide clinical guidance, CD4 + T cell counts at ART initiation in sub-Saharan Africa and developed countries have not appreciably increased during the past decade [8,9]. Although ART has been rapidly scaled up in Asia, to our knowledge no systematic review has assessed temporal trends of the CD4 + T cell counts at ART initiation in Asian country. We aimed to look at the trends in CD4 cells counts (at initiation of ART) over calendar years in China.

Search Strategy and Selection Criteria
We searched PubMed, Wanfang, Chinese Biomedical Literature Database, and China National Knowledge Infrastructure to acquire relevant papers published from January 2002 to December 2015. The following detail search strategy was used: (HIV, "human immune*", "acquired immunodeficiency syndrome", AIDS) and (China, Chinese) and (ART, HAART, "antiretroviral*'' , CD4 , "CD4 cell count" , "T cell "). In addition to database searches, we also screened the references of the selected paper to identify any additional articles. No language restrictions were the whole process of database searches. We chose 2002 as the beginning year for database searches because it was the first year of published WHO guidelines for ART.
The studies that 1 focused on Chinese people living with HIV/AIDS; (2) reported a summary measure of CD4+ T cell counts at ART initiation; and (3) published during 2002-2015,were eligible. We excluded reviews, meeting reports and news articles. We excluded studies that did not report original CD4+ T cell counts data, had fewer than 30 participants, or with a pre-specified CD4+ T cell counts range for inclusion. We also excluded the studies that only reported the number or proportion of patients with CD4 cell counts in particular ranges. When one study presented results of multiple years, we treated them as different computed yeas.

Study Selection and Data Extraction
The titles and abstracts of the retrieved papers were carefully screened by two of us (W.M,, Z.H.. ). One author searched the database and deleted the unavailable papers, and then another author checked the accuracy. When disagreements occurred between these two authors, a third author (H.Z.H,) was invited and a seminar was hold to discuss and resolve the disagreements. All data were extracted and filled using a data abstraction form that developed by us. Characteristics of studies (authors, years, design, regions, population), data type, number of intervals, average interval length (years), and CD4+ T cell counts at initiation (cells/µL) were extracted.
Data type was divided according to the sources of data, and the sources of data included the data from national free antiretroviral treatment program (NFATP), the data from prevention of mother-to-child transmission (PMTCT) program, the data from clinical trial, and the data from descriptive study.
The study region was categorized to eastern China, north China, northeast China, northwest China, south-western China, and south-central China. The income level was divided into low income, low-middle income, high-middle income, and high income based on the gross domestic product per individual in 2013 in China [10,11].
We abstracted the mean CD4 cell counts from every include paper. When the mean value was unavailable, we would reconstruct mean value or use median value. We abstracted the imputed study years using the method described previously by Siedner MJ et al.. We treated the first years of data collection as the imputed year if the last years was unavailable, and we treated 2 years before the publication year as the imputed year when the first and the last years both were unavailable [8].

Quality Assessment
Because of the majority of included study were descriptive epidemiologic studies, we used the Agency of Healthcare Research and Quality (AHRQ) assessment to assess the study quality. The total score of the AHRQ is 11, and the score from 0 to 3 was assessed as low quality, the score from 4 to 7 was assessed as moderate quality, and the score from 8 to

Statistical Analysis
Random-effect model was used to estimate the mean CD4+ T cell counts across studies and its 95% CIs because that it is more conservative and provide better estimates with wider confidence [13]. The I 2 (and its 95% CIs) and tau-square were used to assess the between-study heterogeneity, and the H 2 was used to assess the within-study heterogeneity [14,15]. The following subgroup analyses were performed: economic level (high income, high-middle income, low-middle income, low income), study regions (eastern Some sensitivity analyses were performed to assess the robustness of the outcome, for example re-performed all primary analyses using fixed effect model, and re-performed all primary analyses after excluding studies with fewer than 50 participants. We also reperformed all primary analyses including studies reporting data that spanned ≤5 years only, because study with long-term data might easily produce bias. We performed all analyses using Stata software (version 13.0).

Characteristics of included studies
We identified 8292 articles from database searches and 68 additional articles through screening the references of the retrieved articles. During the titles and abstracts screening, 5550 articles were excluded, leaving 382 full-text articles to be assessed for eligibility. Finally, we included 220 articles in this meta-analysis (Supplementary Appendix   Table 1). Figure 1 notes the process of selecting articles.  (Table 1).

Meta-regression analyses of temporal trend in CD4+ T cell counts at ART initiation
In the univariate meta-regression model, we found that the estimated change in CD4 cell counts at ART initiation was +8.73 cells/μL per year (95% CI= 4.57 to 12.90 cells/μL per year). Adjusting for study regions, study populations, data type, and average interval length did not significantly modify the estimated annual change in CD4 cell counts (+7.46 cells/μL, 95% CI=3.64 to 11.29 cells/μL) ( Table 2).
The subgroup analyses were noted in

Sensitivity Analyses
Sensitivity analyses did not result in substantive changes to our findings ( Table 2).
Exclusion of studies that with fewer than 50 participants yielded a similar estimate of the trend (+8.47 cells/μL, 95% CI= 4.04-12.89 cells/μL). Exclusion of studies that reporting >5 years of data also yielded a similar estimate of the trend (+8.23 cells/μL, 95% CI= 3.24-

Evaluation of Publication Bias
We generated funnel plots and visually examined the asymmetry of these funnel plots. For total sample, the asymmetry observed in the funnel plots was minimal. We also assessed the asymmetry funnel plot using the Egger's linear regression test and Begg's test, and the results noted that there were no significant publication bias across all outcomes.

Discussion
Our analyses presented that the mean CD4+ T cell counts at ART initiation was increased from 136.63 cells/µL in 2003 to 199 cells/µL in 2014, and the estimated change in CD4+ T cell counts at ART initiation was +8.73 cells/μL per year. However, CD4+ T cell counts at ART initiation remained below present treatment guidelines in general population and elder patients. Moreover, remarkable variation between regions was found in our analyses.
We therefore argue that greater resources should be focused on earlier ART initiation, especially in the groups that with low CD4+ T cell counts at ART initiation. The results of this meta-analysis were robust due to a large number of include papers and numerous sensitivity analyses.

Recent published meta-analysis in developed country and sub-Saharan Africa both
reported that CD4+ T cell counts at ART initiation have not increased over the past decade [8,9]. On the contrary, our study showed substantial improvements in the mean A very interesting finding in our study was that higher CD4 cell counts at ART initiation were reported in serodiscordant couples and pregnant women. According to the recent updated Chinese guidelines, ART should be initiated for all HIV-positive pregnant women and serodiscordant couples irrespective of CD4 cell count, which may be the main reason that CD4 cell counts at ART initiation were significant higher in serodiscordant couples and pregnant women than that in other groups [18]. In China, nationwide PMTCT of HIV program was incorporated into child health-care system or antenatal care (ANC) services, which could provide the effective test-and-treat strategies for all HIV-infected pregnant women and partially promote early ART initiation [19]. Recent meta-analyses observed an estimate of nearly 80% of Chinese pregnant women who had 100% ART adherence and HIV vertical transmission rate substantially decreased from 31.8% prior to the program to Our study had some limitations. First, all included articles in this meta-analysis were published articles. Published articles may be more likely to be based on data from centers partnering with academic or research institutions, which often provide additional financial resources and/or health infrastructure for study sites. Second, a high heterogeneity was observed between studies. Although we performed subgroup analyses by income, regions, years, years spanned by data collection, data type, populations, and sample size, and these factors may be the sources of between-study heterogeneity. However, other unmeasured factors likely affected the detected heterogeneity; unfortunately, we did not obtain sufficient information about these aspects for further analysis. The last, the number of studies in pregnant women and serodiscordant couples was too small, so the results would be interpreted with caution.

Conclusion
In summary, due to the new WHO guidelines recommendation that increasing eligibility to  Abbreviations: CI, confidence interval; PMTCT, prevention of mother-to-child transmission;NFATP,national free antiretroviral treatment program. # adjusted by regions,populations, data type, and average interval length