Mortality from AIDS and tuberculosis-HIV coinfection in the Chilean AIDS Cohort of 2000-2017

This article aims to assess the sociodemographic and epidemiological factors associated with AIDS and tuberculosis-HIV coinfection mortality in the Chilean adult population between 2000 and 2017. This is a retrospective observational study, evaluating the incidence density of TB-HIV coinfection mortality in the population over 14 years of age. We used data from the Chilean AIDS Cohort database, 17,512 people enrolled in highly active antiretroviral therapy in the public health system in Chile. The Kaplan-Meier survival function and Cox regression were applied. Incidence density of 0.05 for 39,283 person-years for mortality with TB-HIV coinfection was recorded, with an increase in new cases in people living with AIDS among Aymara and Mapuche indigenous populations. Risk factors included CD4 < 500 cells/mm 3 (HR = 3.2; 95%CI: 2.2-4.9), viral load at the start of treatment > 10,000 cop-ies/uL (HR = 1.3; 95%CI: 1.2-1.6). Having high school or higher education (HR = 0.76; 95%CI: 0.6-0.9) is a protective factor for mortality for coinfection. Mortality was concentrated in TB-HIV coinfected people with increasing mortality among women and indigenous populations. The paper contributes to the growing recognition of the role of social determinants in disease out-comes, and the requirement to improve community-focused and community-based testing, sex education in schools, and structural interventions to reduce the adult mortality in Chilean population. in the Chilean AIDS Cohort. We estimate an average survival of 11 years after initiating ART, which


Introduction
The HIV/AIDS prevalence rate was adjusted for age, being a weighted average of age-specific rates (Equation 2).
Psi = The population in the Ri in the standard population. The standard weights are then given by: In which 0 < Wi < 1. The age-adjusted HIV/AIDS prevalence rate (AAPR given by) (Equation 3) is:

Statistical analyses
Descriptive statistics for the sociodemographic and epidemiological profile of the population were developed. Hypothesis tests were applied to the categorical variables using Fisher's exact test and Pearson's chi-square test. Measures of central tendency (average, median and mod) and dispersion measures, standard deviation, and their respective confidence intervals for continuous measurements were estimated, once the normality of the distribution of the response variables was assessed using the Shapiro-Wilk test. Global survival functions were estimated using the Kaplan-Meier curve method for the crude analysis of sociodemographic and epidemiological factors associated with the survival of patients coinfected with TB-HIV by the log-rank test, accepting a probability of error of less than 5%, with their respective confidence intervals. Cox's semi-parametric model of proportional hazards ratios (HR) was applied 27,28,29 . Besides, the proportionality assumption of the errors were tested for the Cad. Saúde Pública 2021; 37 (6):e00212920 diagnosis of the Cox model and the subsequent evaluation of the proportionality of its residues, using the Schoenfeld test. The organization, cleaning and analysis of the data were performed in Stata, version 16 (https://www.stata.com).
The adjusted prevalence rates of PLWHA, and people that indicated ART and entered the Chilean AIDS Cohort were estimated using intervals with similar time gaps for the years 2000-2005, 2006-2011 and 2012-2017 to know and describe how the prevalence rates were distributed in the country, and to compare them between regions following the reports and current estimates on HIV and AIDS infections conducted by the Chilean Ministry of Health in the Nacional Prevention and Control Plan of HIV/AIDS and its reports 30 . Descriptive maps of spatial distribution of HIV/AIDS prevalence rates were constructed in QGIS, version 3.12.1 (https://qgis.org/en/site/).

Ethical aspects
The data used corresponds to data that use publicly available information under the Law n. 12,527 (November 18, 2011), provided for in Resolution n. 510 (April 7, 2016). We declare that the project was adhered to the regulatory frameworks of Chile, as established in Law n. 20,285 ("Transparency and access to public information"), and in Law n. 19,628 ("Protection of private life").

Results
The highest HIV/AIDS prevalence rates were registered in the northern regions of the country, Arica y Parinacota, with a rate of 7.5 per 100,000 inhabitants, followed by Valparaiso region, with 5.0 per 100,000 inhabitants in the center and in the metropolitan region of Santiago with 4.2 per 100,000 inhabitants. These rates were greater than those estimated for the whole country, 3.0 per 100,000 inhabitants, and showed a significant increase between each observation period ( Figure 1). We emphasize the region of Arica y Parinacota and Tarapaca, where prevalence increased 3.3 times from the first observation period (Figure 1).
The distribution of cases of PLWHA enrolled in ART increased in the second period. It was higher in the cities of the north of the country in the Arica y Parinacota region, with a prevalence between 16.6 and 21.5 per 100,000 inhabitants, followed by the regions of Copiapo, Araucanía and Los Lagos with prevalence between 9.7 and 13.6 per 100,000 inhabitants ( Figure 1).
Regarding the specific prevalence rates by sex, the region that registered the highest rates per period correspond to the region of Arica y Parinacota ( Figure 1), with this region having the highest prevalence, showing a temporal pattern of admitted cases.
We identified 17,512 individuals with both HIV and TB, of whom 87.6% were men. Over time, women represented a larger portion of the sample. This reinforces the hypothesis of HIV feminization in Chile (Table 1).
The main route of transmission was the sexual, with injection drug use reported at low rates: 0.3 (2012-2017). Regarding age group, there is a significant increase among 15-29 years-old, which represented 42.7% of the population in the third period, compared to 34.2% in the period 2000-2005. This increase was slightly smaller in the 30 to 49 years-old age group, being 1.4 times greater than the period from 2000-2005.
During the period analyzed, we found 2,013 death recorded, representing 11% of the total, with a higher proportion of deaths, 31.8%, in the first period (2000)(2001)(2002)(2003)(2004)(2005). These decreased to 5.2% in the last period, which is consistent with a higher probability of survival due to the effects of ART and improved care ( Table 2).
Regarding CD4 lymphocytes, 72.5% of patients in 2000-2005 were enrolled with counts below 350 cells/mm 3 . This patient percentage decreased in the last period to 66.1%. However, when analyzing frank disease, there was a greater increase over time in people that entered in better health. For example, those in stage A in the first period were 43.9%, rising to 45.6% in the second period, and finally, 69.3% in the final period.
We found an incidence density of 0.05 per 39,283 person-years for TB-HIV coinfection mortality in the Chilean AIDS Cohort. We estimate an average survival of 11 years after initiating ART, which    * The differences in the "total n" are due to "missing values"; ** It represents total proportions and significant differences for Fisher's exact test with a probability of error < 0.001; *** Aymara indigenous people represent 32% and Mapuche indigenous people represent 60.3% of indigenous population.
means that people initiating ART in the country's public health system have a 53% probability of survival of 6 to 8 years. We found that men have a lower survival rate when compared with women (p < 0.001) ( Figure 2). With respect age groups, people over 40 years of age survived slightly less than the group between 30 to 39 and 15 to 29 years of age, with statistically significant differences. A lower survival was observed in people with elementary education or less compared those with middle school or higher. These differences are statistically significant (p < 0.001) (Figure 2).
We adjusted a multivariate model for people coinfected with TB-HIV, to evaluate the factors of risk associated with mortality in PLWHA (Equations 4-5). Our findings show that individuals enrolled in ART with TB have a much higher risk of dying (HR = 1.9; 95%CI: 1.2-1.7) (Figure 2). Besides, people coinfected with TB-HIV that entered with a CD4 count below 500 cells/mm 3      When assessing the risk of death due to coinfection with TB-HIV after the start of ART, we observe a higher chance of death, and a reduction in life expectancy associated with a number of different modifiable factors, implying serious questions about prevention, control and treatment. Since TB is a treatable disease, this represents a strong indicator of inequality in programming and delivery.

Discussion
Between 2000 to 2017, 157 deaths due to TB-HIV coinfection were registered, which represents 13.3% of the total deaths in PLWHA in the Chilean AIDS Cohort. This represents a failure of treatment in a marginalized population with less access to public health, as reported in other studies in the country 31,32 . Diagnosis of TB in HIV-positive persons can be limited, in part, to poor infrastructure and failure to address issues of culture, as demonstrated in studies conducted in Latin America countries, added to a delay in the start of ART and multidrug resistant TB 22,33 .
After analyzing overall mortality, which includes 17 years of observation of PLWHA in treatment in the public health system, we showed a higher mortality due to the effect of comorbidities, in which Cad. Saúde Pública 2021; 37(6):e00212920  This decrease has been accelerating in recent decades, which coincides with the findings from other studies 13,35,36 .
Estimates indicate that in the Latin American and Caribbean region there would be 2.1 million and 330,000 PLWHA, respectively, in 2019, and 132,000 new infections in the population over 15 years-old 37 . These numbers reflect the inadequacy of control and prevention programs, since these data show only a slight decrease since the last report in 2016, with 120,000 new HIV infections, far from the path to the 90% reduction target by 2030 1,3 .
Regarding key populations, men who have sex with men (MSM) constitute the largest group of new cases (60%) for the period 2014-2017. These findings are consistent with other studies conducted among MSM, using respondent-driven sampling (RDS) in the Santiago Metropolitan Region 38,39,40 , prevalence studies conducted in Latin American countries 41,42,43 and Angola in Sub-Saharan Africa 44 .
Selected areas of Chile showed a progressive and chronic increase in the prevalence of HIV/AIDS in the adult population, of 5.3 per 100,000 inhabitants, particularly in the northern regions of the country (Arica y Parinacota, Tarapacá, Antofagasta) and in the center of the country (Valparaíso and the Santiago Metropolitan Region). There are multiple reasons for this uneven pattern of AIDS and TB-HIV mortality. In urban areas, the availability of ART and test-and-treat have led to prevention and avoidance of services that aid transmission. In more rural and marginalized areas, especially indigenous areas, the lack of consistent public HIV programs, history of relatively low prevalence, and poor access to care exacerbate the conditions of marginality.
Studies in Brazil among MSM show that HIV rates are rising steeply among young populations, who report early sexual initiation, a high number of sexual partners and lack of condom use 35,37 .
Cad. Saúde Pública 2021; 37(6):e00212920 Not only for Brazil, but, for example, a study conducted in the province of Luanda (Angola) by our group 44 showed that the opening of borders and the pace of social change such as rapid urbanization, internal and external migration, low levels of HIV knowledge and perceived risk, diversity of sexual encounters and low condom use may contribute to increasing HIV infection among MSM. Given what we know about TB and HIV, and the late initiation of ART, programs addressing TB need to be reinforced.

Conclusion
Our findings showed that there is a lower probability of survival in TB-HIV coinfected people in the Chilean AIDS Cohort, especially for those with a delay in the initiation of ART. Furthermore, there is a concentration of the epidemic at younger ages. There is a process of accelerated feminization of the epidemic, and spread to vulnerable emerging groups such as the Aymaras indigenous population in northern Chile, indigenous Mapuche in the center and south of the country, and among migrants. We strongly advocate for the need to improve access to community-based TB and HIV testing, targeted sex and health education in schools and structural interventions to reduce stigma and discrimination, both for sexual and ethnic minorities. Social policies to protect the human rights of minorities and vulnerable populations are required. Thinking of the overall investments in health care and ART, the potential for ongoing transmission, and the emergence of MDR-TB, not investing in improved TB programs seems counterintuitive.

Study limitations
Due to limitations of our data, we could not explore more robustly coinfection with TB-HIV in indigenous groups. However, our results do suggest the direction of the rapidly evolving phenomenon of HIV/AIDS and coinfections of the TB-HIV epidemic in Chile.

Contributors
C. Sanhueza-Sanzana contributed to the study conception, data collection, analysis and interpretation, and manuscript writing. L. Kerr contributed to the study conception and design, methodology, software ownership, analysis and interpretation, and review and editing. C. Kendall contributed to the study conception and review and editing. All authors approved the final version of the manuscript. Cad