Older adults at high risk of HIV infection in China: a systematic review and meta-analysis of observational studies

There is an increasing prevalence of human immunodeficiency virus (HIV) infection in older adults in China, but the findings across prevalence studies have been mixed. This is the first meta-analysis of the prevalence of HIV infection and its moderating factors in older adults in China. Two investigators systematically and independently searched both international (PubMed, PsycINFO, Web of Sciences and EMBASE) and Chinese (WanFang, CNKI, and CQVIP) databases. HIV infection rates in older adults were analyzed using the random-effects model. Altogether 46 studies were included in the analysis. The pooled prevalence of HIV infection in older adults was 2.1% (95% CI [1.9%–2.3%], I2 = 99.4%). Subgroup analyses revealed that men who have sex with men (MSM), hospital population samples, publications after 2014, studies conducted in the western region of China, and higher study quality were significantly associated with higher HIV infection rate. This meta-analysis found that the HIV infection prevalence in older adults is significantly higher than the general population in China. Attention should be given to this urgent public health issue, and effective HIV/AIDS preventive, screening and treatment measures are warranted in this population. PROSPERO: CRD42019124286.

The inclusion criteria were studies that: (1) reported prevalence of HIV infection in older adults or provided information to calculate the prevalence. The diagnosis of HIV infection was based on study-defined criteria; (2) were cross-sectional or cohort studies (only baseline data were included) with meta-analyzable data; (3) were conducted in older adults (i.e., aged 50 years and older) in China. Case series, reviews, and meta-analyses were excluded.

Outcome measures
The outcome measure of the meta-analysis was the prevalence of HIV infection in older adults. For each study, the prevalence was calculated by the number of HIV-infected older adults divided by the total number of older adults.

Data extraction
Systematic literature search and data extraction were independently conducted by the same two investigators. The titles and abstracts of potential publications were screened separately by the same two investigators before the full texts were read for eligibility. Any inconsistencies in the process were discussed and resolved by a third reviewer (Yuan-Yuan Wang). The following information was extracted: year of publication, survey period, study site, sampling method, sample size and response rate, mean age, sex, education, occupation, province, rural or urban area, definition of older adults (e.g., above 50 years), and transmission route (e.g., commercial sex).

Quality assessment
The same two investigators independently evaluated the methodological assessment, using the critical appraisal for epidemiological studies (Loney et al., 1998) that contains 8 items covering three aspects: sampling, measurement and analysis. The total score of this instrument is 8; the total score of 7-8 was considered as 'high quality', 4-6 as 'moderate quality', and 0-3 as 'low quality ' (Loney et al., 1998). Any inconsistencies were resolved by a discussion with a third investigator.

Data analysis
The Comprehensive Meta-Analysis software, Version 2.0 (http://www.meta-analysis.com/) and Open Meta-Analyst (http://www.cebm.brown.edu/openmeta/) were used to synthesize data (DerSimonian & Laird, 1986) in all the meta-analytic outcomes. Due to different sampling methods, study designs and demographic and clinical characteristics between studies, the random-effects model was used (DerSimonian & Laird, 1986). The heterogeneity of outcomes were assessed using I 2 , with I 2 > 50% as significant heterogeneity (Higgins et al., 2003). Following the recommendation of the Cochrane handbook (Higgins et al., 2019) and other studies (Li et al., 2020;Xu et al., 2020;Yang et al., 2020), publication bias was assessed using the funnel plots and Begg's test and Tweedie's trim-and-fill analysis. All data analyses were 2 tailed, and the significant level was set at 0.05.
Subgroup and meta-regression analyses were performed to examine the moderating factors of HIV infection prevalence. Subgroup analyses were conducted according to the following categorical variables: (1) sexual orientation: MSM vs. not specified; (2) sex: male predominance (male percentage ≥ 60%) vs. no-predominance; (3) study site: hospital vs. community; (4) route of transmission: commercial sex vs. non-specified; (5) education: equal to and below primary school vs. above primary school education; (6) occupation: predominance of farmers (i.e., farmer percentage ≥ 60%) vs. no such predominance; (7) economic region: western vs. middle of vs. eastern region of China; (8) area: rural vs. urban; and (9) cut-off age for older adults: ≥50 years vs. ≥60 years, and (10) publication year: during and before 2014 vs. after 2014 (using median splitting method). Meta-regression analyses were conducted to examine the moderating effects of continuous variables (such as quality assessment score and study period) on the results when the number of included studies was at least 10.

Literature search and study characteristics
A total of 3,641 publications were identified in the initial literature search. Finally, 46 studies with 363,399 subjects met the entry criteria and were included in the meta-analysis (Fig. 1); four studies were published in English Feng et al., 2009;Ning et al., 2018;Xie et al., 2014), and the remaining were published in Chinese language. Table 1 shows the study characteristics. All studies were published between 2004 and 2018. Most studies (89%) used ≥ 50 years as cut-off age for older adults. Of the 46 studies, only one reported treatment information in HIV infected older adults, i.e., 53.75% of whom had HAART treatment (Ning et al., 2018). Nine studies reported the HIV infection rate in all age groups (Chen et al., 2017;Chen et al., 2013;Du & Meng, 2016;Feng et al., 2009;Wang et al., 2018a;Zhao, Su & Jiang, 2015;Zhong et al., 2017;Zhu, Miu & Zhang, 2014), therefore only data in older adults were extracted for analyses; while five studies focused on older MSM adults. Thirteen studies reported transmission routes, while the remaining (71.7%) did not.

Prevalence of HIV infection in older adults and moderating factors
The pooled prevalence of HIV infection in older adults was 2.1% (95% CI [1.9%-2.3%], I 2 = 99.3%, Fig. 2). The results of the subgroup analyses are presented in Table 2. MSM population, hospital population samples, publications after 2014, and studies conducted in the western region were significantly associated with higher HIV infection rate. Metaregression analyses revealed that higher study quality was significantly associated with higher HIV prevalence (β = 0.84, p < 0.001, Fig. S1). Meta-regression analysis did not find any significant association between study periods and the prevalence of HIV infection (β = −0.03, p = 0.76).

Publication bias
Although funnel plot was visually asymmetrical (Fig. S2), Begg's test did not find statistically significant publication bias (p = 0.21). The Duval and Tweedie trim-and-fill analysis suggests that 8 studies would need to be imputed to achieve an approximate normal error distribution. Including these 8 studies could lead to a lower prevalence of 0.096 (95% CI [0.094-0.099]).

DISCUSSION
This was the first meta-analysis to examine the prevalence of HIV infection in older adults in China. The meta-analysis revealed that the pooled prevalence of HIV infection in older Chinese adults was 2.1%, which was substantially higher than the figure reported in the Chinese general population (0.05%) (National Bureau of Statistics of China, 2018). The high HIV infection rate could be due to several reasons. The life expectancy of HIV-infected adults has been significantly prolonged due to widespread use of HAART (Bhaskaran et al., 2008;Greenbaum et al., 2008;Xing et al., 2014); e.g., the National Free Antiretroviral Treatment Program (NFATP) has covered more than 97% of HIV-infected people in Table 1 Characteristics of studies included in this meta-analysis.  Xie, 2014Zhu, 2016Wang, 2018Hong, 2016Xu, 2017Liu, 2004Li, 2015Lin S, 2018Shao, 2018Zhou, 2013Li P, 2018Qin, 2015Liu M, 2016Yu, 2017Dou, 2014Huang, 2018Ning, 2018Zhao, 2015Chen, 2013Lu, 2014Lu, 2015Deng, 2017Pan, 2014Wang, 2014Wu, 2015Zhu J, 2014Mai, 2017Chen Z, 2016Min, 2014Ma, 2015Fu, 2013Shi, 2016Lin W, 2018Zhou, 2015Du, 2016Chen R, 2017Li, 2018Zhu Y, 2014Su, 2016Feng, 2009Zhong, 2017 Overall ( China (Zhang et al., 2009). The high HAART adherence rate in Chinese HIV patients, as confirmed by a recent meta-analysis (Wang et al., 2018b), would be expected to increase life expectancy and many patients are living into their older adulthood. Furthermore, many studies have indicated increasing transmission via commercial sexual activities among older Chinese men after retirement as a major reason for HIV infection  Zhou et al., 2014). It has been suggested that prevention of HIV transmission among older MSM should be an urgent priority in China's HIV/AIDS strategy (Ning et al., 2018). The high proportion of HIV infection in older adults is growing major public health challenge in China. Compared to younger adults, physical and psychiatric comorbidities, such as pneumonia, depression and insomnia, are usually more common in HIV-infected older adults (Ding et al., 2017), which could lead to heavy personal, family and economic burden. Therefore, appropriate allocation of healthcare resources and developing effective preventive strategies for HIV-infected older adults in China should be considered (Xing et al., 2014). Consistent with previous findings on HIV prevalence in China , we found in this meta-analysis that MSM was associated with a higher risk of HIV infection; the HIV prevalence in MSM older adults was 11.8%, which was the highest among all subgroups. Compared to community populations, hospital population samples were significantly associated with higher HIV infection rate, which is probably because older adults with HIV infection were more likely to receive HIV testing than those in the community. Compared to middle and eastern economic regions, HIV infection prevalence in older adults was significantly higher in the western region of China, being less developed than other parts of China. Therefore, the lack of access to HIV treatment and prevention measures in the western region could be associated with higher HIV infection rate. Due to degradation of traditional Chinese family structure and lack of family support, retired older men are also more likely to have engaged in commercial sex, particularly in under-developed western regions of China (Huang, Maman & Pan, 2012).
Unexpectedly, commercial sex as the transmission route was not significantly associated with higher HIV infection rate. This appears inconsistent with previous findings that commercial sex is a major route for HIV infection transmission among older Chinese men Zhou et al., 2014). Studies published after 2014 were significantly associated with higher HIV infection rate, which we were unable to explain adequately. However, we found that higher study quality was significantly associated with higher HIV prevalence. Due to sigma and discrimination associated with HIV/AIDS, many sufferers, particularly older adults in China usually deny or conceal their diagnosis in order to avoid ''loss of face''. High quality studies may identify patients more systematically, and obtain a more accurate and often higher rate HIV infection.
There were several limitations in this meta-analysis. First, similar to other meta-analyses of epidemiological studies Long et al., 2014;Wang et al., 2018b;Winsper et al., 2013), there was substantial heterogeneity, although subgroup analyses were performed. The heterogeneity may be associated with different sampling methods, study designs, diagnostic criteria of HIV infection and demographic and clinical characteristics between studies. Second, most studies did not report the transmission route, therefore further sophisticated analyses could not be conducted. Third, due to the cross-sectional design of included studies, the causal relationship between HIV infection and related variables could not be explored.
In conclusion, this meta-analysis showed that the prevalence of HIV infection in older adult population is significantly higher than the general population in China. Attention should be given to this urgent public health issue, and effective HIV/AIDS preventive, screening and treatment measures are warranted in this population.

Grant Disclosures
The following grant information was disclosed by the authors: University of Macau: MYRG2015-00230-FHS, MYRG2016-00005-FHS.