Skip to main content

Early mortality in a cohort of people living with HIV in Rio de Janeiro, Brazil, 2004–2015: a persisting problem



Global mortality from AIDS-related diseases has been declining since 2005, resulting primarily from the widespread use and early initiation of combination antiretroviral therapy. Despite the significant improvements, high rates of early mortality, usually defined as that occurring within the 1st year of entry to care, have been observed, especially in resource-limited settings. This analysis draws upon data from an observational cohort of people with HIV (PWH) followed at a reference center for HIV/AIDS care and research in the city of Rio de Janeiro, Brazil, to identify the pattern and factors associated with early mortality.


The study population includes PWH aged 18 or older followed at the National Institute of Infectious Diseases Evandro Chagas who were enrolled between 2004 and 2015. The primary outcome was early mortality, defined as deaths occurring within 1 year of inclusion in the cohort, considering two follow-up periods: 0 to 90 days (very early mortality) and 91 to 365 days (early mortality). Cox proportional hazards models were used to identify the variables associated with the hazard of very early and early mortality.


Overall, 3879 participants contributed with 3616.4 person-years of follow-up. Of 220 deaths, 132 happened in the first 90 days and 88 between 91 and 365 days. Very early mortality rate ratios (MRR) show no statistically significant temporal differences between the periods 2004–2006 to 2013–2015. In contrast, for early mortality, a statistically significant decreasing trend was observed: mortality rates in the periods 2004–2006 (MR = 5.5; 95% CI 3.9–7.8) and 2007–2009 (MR = 3.9; 95% CI 2.7–5.7) were approximately four and three-fold higher when compared to 2013–2015 (MR = 1.4; 95% CI 0.7–2.7). Low CD4 count and prior AIDS-defining illness were strongly associated with higher hazard ratios of death, especially when considering very early mortality.


The present study shows an excess of mortality in the 1st year of follow-up with no changes in the mortality rates within 90 days among PWH from Rio de Janeiro. We note the significant impact of initiating treatment with immunosuppression, as evidenced by the increased risk of death among those with low CD4 cell count and with AIDS-defining illnesses.

Peer Review reports


Brazil, a middle-income country with an existing Unified Health System, preceded many other middle and low-income countries in providing treatment and support for people living with HIV (PWH). In 1996, combination antiretroviral therapy (ART) was provided free of charge to PWH with a CD4 cell count < 200 cells/μL or an AIDS-defining illness according to CDC 1993 criteria [1]. Over the years, Brazil’s HIV treatment efforts have changed to include newer drugs and an increasing number of salvage regimens. Furthermore, ART initiation criteria for PWH changed from 200 to 350 cells/μL in 2004 [2], from 350 to 500 cells/μL in 2008 [3], and then was made available regardless of CD4 count in 2014 [4]. The impact of treatment on overall as well as AIDS-specific mortality and morbidity has been shown in local studies from Brazil [5,6,7] and survival benefits have been calculated [8,9,10] and estimated using modeling strategies [11].

Beyond the provision of HIV care and treatment, described above, the Unified Health System (SUS, Sistema Único de Saúde), implemented in Brazil in 1988, played an important role in the success of the HIV program [12]. SUS provides a universal care through a hierarchical and regionalized health service in which the primary healthcare is conceived, for most individuals, as the first point of contact with healthcare. Relating to HIV care, SUS also encompasses voluntarily Testing and Counseling services through a network of facilities as well as laboratory monitoring. In the past, HIV care had been provided mainly by the specialized centers but now are also available within the primary healthcare [13].

However, despite the significant improvements, high rates of early mortality have been observed, especially in resource-limited settings [14, 15]. Early mortality has been defined as death occurring during the 1st year after initiating treatment, where the risk of mortality is greatest since the full effect of therapy has not yet been obtained. Mortality soon after starting ART is influenced by late presentation or severe immunodeficiency, which demonstrates weakness in healthcare programs targeting PWH [16,17,18]. In a 2016 national study of the continuum of care outcomes, Black ethnicity/skin color, lower education, residing in a less developed region (North and Northeast), and high levels of social vulnerability were independently associated with a higher likelihood of presenting to care with advanced disease and of not using ART [19]. In another analysis on the barriers to HIV testing, fear of stigma and discrimination stood out as important reasons for not testing [20]. Estimates of life expectancy of PWH who survive the 1st year of ART are much higher than at initiation, reflecting the importance of early treatment [21, 22].

The present study aimed to identify the pattern and factors associated with early mortality in PWH from a clinical cohort followed at a reference center for HIV/AIDS care and research in the city of Rio de Janeiro, Brazil.


We used information from the HIV Clinical Cohort followed at the STD and AIDS Clinical Research Laboratory of the National Institute of Infectious Diseases Evandro Chagas (INI/Fiocruz), in Rio de Janeiro. The HIV Clinical Cohort is an open cohort of individuals cared for at the institution for whom a database containing sociodemographic, laboratorial, clinical, outpatient, and hospital information has been routinely collected by trained abstractors. Information is updated regularly since 1998, using medical charts (electronic since 2004) of outpatient visits, hospitalizations, laboratory test results, and pharmacy records. Information on deaths is derived from medical records and the state mortality registry. Details of the methodology were previously described [23, 24].

Study population

For this analysis, we considered adult (18 years of age or older) PWH who started treatment in our institution between January 1, 2004, and December 31, 2015. The lower limit of the period was defined based on the 2004 update of the Brazilian guidelines recommending ART initiation with a CD4 count < 350 cells/µL or an AIDS-defining illness [2]. The upper limit references the last year in which information on dates and causes of death was systematically checked with the death registry of the State of Rio de Janeiro, thus minimizing the underreporting of deaths. Almost all patients enrolled in the cohort are from Rio de Janeiro state, 99.5%, of which 97.8% live in the metropolitan area.

Outcome definition

The primary outcome assessed was early mortality, defined as deaths occurring within 1 year from cohort enrollment. As such, the start of follow-up was defined based on the date of the first visit at our clinic when cohort enrollment was initiated, and the end of follow-up was determined by death or censoring. We considered two follow-up periods to better characterize early mortality: 0 to 90 days (very early mortality), and 91 to 365 days (early mortality). Accordingly, for any participant, study follow-up is at most 1 year of the first clinic visit. Among non-deceased individuals, censoring was defined according to the available follow-up information. Participants with follow-up information beyond 1 year from the first clinic visit were censored at 365 days. Participants with follow-up information less than 1 year from the first clinic visit were censored on the last known contact date with the clinic, given the possibility that participants may have moved to another clinic or out of state.

Independent covariates

We considered the following variables as potentially associated with early mortality: year of first visit (2004–2006, 2007–2009, 2010–2012, 2013–2015), age, self-reported ethnicity/skin color (White, Black, Brown), years of education (≤ 5, 6–9, ≥ 10), ART naïve before first visit (yes/no), CD4 cell count at first visit (≤ 50, 51–200, 201–350, > 350 cells/µL), viral load count at first visit (< 400, 400–999, ≥ 1000 copies/mL) and presence of prior AIDS-defining illness (yes/no) [25]. CD4 cell counts and viral load measurements were assessed using a window of 180 days before and after first visit, and if more than one measurement was available, the closest value to the first visit was assigned. Prior AIDS-defining illness was based on the 1993 Centers for Disease Control and Prevention (CDC) case definition with a window of 180 days before to 30 days after first visit.

We grouped participants into the following categories based on their gender and, among men, their sexual orientation (heterosexual, homosexual or bisexual): Cis-W: cisgender women; Trans-W: transgender women (i.e. assigned male sex at birth and who identified as a woman according to information on the medical chart); MSM: gay and bisexual men who have sex with men; and MSW: men who have sex with women. In Rio de Janeiro, as in most regions of Brazil except the South, most HIV transmissions result from sexual intercourse. Accordingly, we excluded from the analyses a small number of participants who acquired HIV through other modes of HIV acquisition.

Statistical analysis

Participants’ characteristics are described according to the year of first visit with medians (interquartile ranges) for continuous variables and absolute and relative frequency for categorical variables. We used Chi-squared tests to evaluate differences in the study population by year of first visit. Mortality rates and 95% confidence intervals (CI) were estimated, according to the year of first visit for each participant, using Poisson regression models. We evaluated the temporal trends in the very early and early mortality rates by estimating mortality rate ratios (and 95% CI) for the periods of study assuming the last period (2013–2015) as the reference category. Cox proportional hazards models were used to identify the variables associated with the hazard of early death. Covariates found to be significantly associated with the outcome (p < 0.25) in the unadjusted models were then analyzed together in a multivariate model. All variables selected from the unadjusted model were kept in the adjusted model independently of being statistically significant or not.

The results of the Cox models were interpreted as hazard ratios [26]. To overcome the problem resulting from missing data on some variables, we used bootstrapping multiple imputation methods combined with nonparametric regression models [27, 28]. To increase the statistical efficiency of the imputation process, five independent repetitions of the imputed values database were performed [29, 30].

All analyses were performed using R software [31]. Multiple imputation was done using the Hmisc library [32]. Poisson model parameters were estimated using the basic library of R, stats [31]. The parameters of Cox proportional hazards models were obtained based on the Hmisc and rms libraries [32, 33], and the risk proportionality assumption was evaluated based on the scaled Schoenfeld residuals, available in the survival library [34, 35].


From January 2004 to December 2015, 3991 individuals diagnosed with HIV aged 18 or older at the time of the first visit were included in the HIV Clinical Cohort of INI/FIOCRUZ. We excluded from the analysis six observations reffering to Indigenous (N = 2) and Asian descent (N = 4) participants. As participants with non-sexual routes of HIV acquisition do not constitute a homogeneous group, making it necessary to create several categories with a small number of observations, we also excluded them from the analysis (injection drug users [N = 33], vertical transmission [N = 31], work accident [N = 9], blood transfusion [N = 32] and haemophilia [N = 1]). At the end of this process the, the study population was composed of 3879 individuals whose HIV transmission occurred through sexual contact.

Table 1 presents participants’ characteristics by year of first visit. Among demographic characteristics, there was an increase in the proportion of young people (18–24 years) and a decrease in the proportion of participants aged 35–49 years old over time, reflecting the significant drop in median age from 37 (IQR 29–43) to 33 years (IQR 27–42). Over the periods, the proportion of participants of White ethnicity/skin color decreased from 61.9 to 36.5% between 2004–2006 and 2013–2015 while the proportion of participants with 10 or more years of education increased from 49.3 to 62.0%, in the same periods. The gender/sexual preference categories remained relatively stable except for the significant increase in the number of Trans-W (0.6% in 2004–2006 to 10.8% in 2013–2015).

Table 1 Demographic and clinical characteristics of cohort participants according to year of first visit

Overall, most participants (77.9%) were ART-naïve before their first visit to INI. The proportion of participants with a CD4 cell count > 350 cells/µL at first visit increased slightly from 40.7% in 2004–2006 to 45.6% in 2013–2015. The proportion of participants with a low viral load (< 400 copies/mL) increased almost 8 percentage points over time. Finally, we observed an increase in the proportion of individuals with prior AIDS-defining illnesses from 20.5% in 2004–2006 to 29.1% in 2010–2012, followed by a reduction, reaching 24.4% in 2013–2015. Overall, 23% of participants had an AIDS-defining illness at first visit, out of which, the great majority (81%) were ART-naïve. The most prevalent types of AIDS-defining illnesses were extrapulmonary tuberculosis (28%), pulmonary tuberculosis (26%), and Pneumocystis jirovecci pneumonia (24%).

Table 2 shows the number of deaths, loss to follow-up (LTFU), and person-years of follow-up, as well as the respective mortality rates and rate ratios, according to the year of first visit and follow-up period. There were 220 deaths within the 1st year after cohort inclusion, and the vast majority occurred in the first 90 days (60%). The proportion of AIDS-related deaths was higher in the first 90 days when compared to 91 to 365 days. Approximately 87% of deaths that occurred in the first 90 days were AIDS-related, while in the 91 to 365 days follow-up period they represented 66%. As the number of deaths in the first 90 days of follow-up was higher than that observed between 91 and 365 days in all years of first visit, with exception to 2004–2006, and given that the exposure time in the first follow-up period is much shorter, mortality rates in the first 90 days were much higher regardless of the year of first visit (overall 14.2 in the first 90 days versus 3.3 between 91and 365 days) (Table 2).

Table 2 Number of deaths, loss to follow up (LTFU), person-years, mortality rates (MR) per 100 person years and mortality rate ratios (RR) of cohort participants, according to year of first visit and follow-up period. INI cohort, 2004–2015 (N = 3879)

Considering very early mortality rate ratios, there were no statistically significant temporal differences between the periods 2004–2006 to 2013–2015. This can also be seen at the Kaplan–Meier curves in Fig. 1. In contrast, for early mortality, a statistically significant decreasing trend was observed: mortality rates in the periods 2004–06 (RR = 3.93; 95% CI 1.95–8.78) and 2007–09 (RR = 2.82; 95% CI 1.39–6.32) were approximately three and two-fold higher than that observed in the 2013–2015 period.

Fig. 1
figure 1

Kaplan–Meier curves, stratified by year of first visit and follow-up period of cohort participants. INI cohort, 2004–2015 (N = 3879)

Regarding LTFU, 4.0% of the participants who entered the cohort between 2004 and 2015 were LTFU, with the highest LFTU in the 2013–2015 period. This represents an LTFU rate of 4.34 per 100 person-years.

The results of Cox proportional hazards models for very early and early mortality are presented in Tables 3 and 4. The proportionality assumption was not verified for prior AIDS-defining illness in the very early mortality model nor the variables age, period of first visit, and prior AIDS-defining illness for the outcome early mortality. However, analysis of the Schoenfeld residuals plots shows no major deviations from the proportional hazards assumption (Additional file 1: Figs. S1 and S2). Besides, we can interpret the effects of these variables as an average risk ratio [36].

Table 3 Hazard ratios (HR) of dying in the follow-up period of 0–90 days after first visit, their respective 95% confidence intervals (CI), and p-values for Cox proportional hazard models
Table 4 Hazard ratios (HR) of dying in the follow-up period of 91–365 days after first visit, their respective 95% confidence intervals (CI), and p-values for Cox proportional hazard models

Variables associated with the hazard of very early mortality included all clinical variables (except for viral load), age, and education. A 10-year increase in age increased the mortality hazard by 25% (HR = 1.25; 95% CI 1.05–1.47, Table 3). Participants with 5 years or less of education had a risk of mortality 56% higher than those with 10 or more. The risk of mortality among patients who were not ART-naïve was 2.52 times that observed among those who were ART-naïve before first visit (HR = 2.52; 95% CI 1.37–4.62). Having a CD4 count lower than 50 cells/μL or between 51 and 200 cells/μL increased the hazard of mortality by approximately nine and seven-fold (HR = 9.98; 95% CI 3.32–29.94 and HR = 8.05; 95% CI 2.98–21.77, respectively) compared to those with a CD4 count above 350 cells/μL. Moreover, the hazard of mortality among participants with prior AIDS-defining illness was seven times that of participants without diagnosis (HR = 7.13; 95% CI 4.11–12.37).

All demographic variables were associated with early mortality outcome, with exception of ethnicity/skin color (Table 4). Age showed a larger effect than that observed for the very early mortality model (HR = 1.57 CI 1.25–1.97), < 5 years of education double the mortality hazard as compared to individuals with 10 years or more of education (HR = 1.91; 95% CI 1.03–3.53). MSM had a higher hazard of death when compared to cis women (HR = 2.07; 95% CI 1.01–4.24), as did MSW (HR = 1.92; 95% CI 0.96–3.82), although borderline significant. Trans-W (HR = 2.84; 95% CI 0.63–12.82) also had a higher hazard of death compared to cis-women, but not statistically significant, with a p-value of 0.17. Additionally, the period of first visit was statistically significant (reflecting the relationship found between this variable and the mortality hazard, Table 2). When considering the clinical variables, participants who had their first visit in 2004–2006 and 2007–2009 showed mortality hazard 3.5-fold (HR = 4.50; 95% CI 1.80–11.23) and 2.5-fold (HR = 2.59; CI 1.03–6.51) higher, respectively, when compared to the period 2013–2015. Participants with CD4 count ≤ 50 cells/μL had a risk of death of approximately 2.5 times (HR = 2.45; 95% CI 1.09–5.49) than of those who had a CD4 count higher than 350 cells/μL. Finally, participants with prior AIDS-defining illness at cohort entry had a three-fold increase (HR = 4.32; CI 2.33–8.00) on the hazard of mortality compared to those who did not.


In this study, we investigated risk factors associated with very early and early mortality in a cohort of PWH cared for at the Evandro Chagas National Institute of Infectious Diseases (INI/Fiocruz). As in previous analyses, we focused on deaths occurring within 1 year after first visit but further divided this period into two categories of follow-up: 0 to 90 days and 91 to 365 days. This analysis strategy allowed us to identify distinct trends in mortality rates per period and to assess whether different risk factors are associated with early mortality in each follow-up period. In the 12-year study period (2004–2015), there was a large expansion in the use of ART, as well as advancements in the treatment of HIV/AIDS. Globally, in this period, there was an increase in the survival of PWH though lower than that of the general population [21, 37]. This gap is even greater in the context of developing countries, where high rates of early mortality persist [14, 15, 38, 39].

Our results on the mortality rates during the 1st year highlight important excess mortality when compared with results from developed countries [17, 40, 41], but are on the same level as those reported in other developing countries [17, 42]. Regarding very early mortality, meta-analysis estimates show that the estimated mortality rate of 11.5% is much higher than the estimated for low- and middle-income countries, 6.0% [38]. However, attention should be drawn to the great heterogeneity among the populations analyzed, as mortality rates range from 1% (0.3–2.5%) in a Chinese cohort of patients recruited in 2003–2004 and followed every 6 months until May 2010 [43] to 18.4% (14.5–23.0%) in a cohort of patients in a rural hospital in Tanzania who started ART between October 2003 and November 2006 [44].

Importantly, our results show that over time very early mortality rates have remained high while there was a decreasing trend for mortality rates during 91 to 365 days of follow-up, with the periods 2010–2012 and 2013–2015 having similar mortality rates. The high very early mortality rate in all periods (between 10 to 12 per 100 person-years) may be a result of changes in the composition of the study cohort. In the last decade, HIV infection has increasingly affected people of lower socioeconomic status [45]. Among our study participants, we observed a decrease in the proportion of White participants and an increase in the proportion of Black and Brown participants. There is an accumulation of evidence about the social and economic disadvantages of the Black and Brown population compared to the White population in Brazil [46,47,48]. Notably, during the past decade, studies have shown an increase in the affirmation of Black and Brown identity in Brazil, with a consequent increase in the self-report as Black or Brown in National studies [49]. Nevertheless, this fact could not explain the changes observed in the composition of the cohort as they are much more pronounced than those observed in the national study cited previously. The excess of very early mortality over time may also result from health system failures in identifying PWH at an early stage of the disease [50]. Despite the advances made in terms of HIV treatment, efforts still need to be made to expand HIV surveillance system.

Regarding the survival analysis, our findings corroborate those of other studies, although some variables were evaluated differently. This includes the gender/sexual preference variable, in which MSM and MSW were at a higher risk of dying when compared to the category of ciswomen. Studies have shown that women are more likely to look for health services, which has been suggested as the main explanation for the lower mortality rates found among them in the general population [51], a fact also evidenced when it comes to early mortality among PWH [52,53,54]. Sexual and reproductive health needs may also play a role. Prenatal and childbirth coverage in Brazil is close to 100% and the coverage of testing for HIV infection during pregnancy is over 80% [55,56,57]. Men tend to be diagnosed later than women with AIDS-related illnesses, thus increasing the risk of mortality, particularly early mortality [10, 58,59,60]. Here, it is important to highlight the significant burden of tuberculosis in our setting, as PWH with TB tend to seek care later than expected, with low CD4 count, greatly contributing to early mortality [61, 62]. Our results have equally evidenced the burden of TB, with extrapulmonary and pulmonary TB being the two most frequent AIDS-related illnesses.

It is worth mentioning the changes in the effects of the variable gender/sexual preference after adjustment. While in the unadjusted model the MSW category is statistically significant (HR = 2.26; 95% CI 1.17–4.38) and has a higher hazard ratio than MSM (HR = 1.42; 95% CI 0.74–2.72), in the adjusted model we can observe that MSM is the category statistically significant and with higher HR (HR = 2.07; 95% CI 1.01–4.24) when compared to MSW (HR = 1.92; 95% CI 0.96–3.82). These findings are similar to those presented by Coelho and colleagues [58] while analyzing competing risks in this same cohort with participants enrolled between 2000 and 2011. Transwomen also had a substantial difference between models, with a much higher hazard ratio (HR = 2.84; 95% CI 0.63–12.82), although not statistically significant since the number of transwomen in the cohort is small (N = 170). Albeit the 2.8-fold higher hazard of mortality when compared to ciswomen, transwomen have a higher CD4 count at first visit and entered more recently in the cohort, two factors that are protective. Therefore, this can possibly be explained by non-AIDS mortality [63].

Even though the percentage of participants with CD4 count > 350 cells/µL at first visit increased from 2004–2006 to 2013–2015, it consistently remained below 50% among the periods analyzed. This is a global trend, even in high-income countries [64]. CD4 count has a substantial impact on early mortality, and, even in recent periods with better treatments and earlier initiation, deaths occurring very early after reaching care were not affected. The results obtained for the clinical indicators at first visit are consistent with other studies, revealing that participants with low CD4 cell counts and ART naïve present higher risks of early mortality [65,66,67]. Indeed, we show that low CD4 counts dramatically increased the hazard of mortality in the first 90 days of follow-up. In this regard, though the Brazilian recommendations for ART initiation have been for all PWH irrespective of CD4 cell count, those with low CD4 cell counts should be prioritized since they may need to initiate, in addition to ART, prophylaxis or treatment for opportunistic infections to reduce their high mortality [4].

This study has strengths and limitations that should be acknowledged. A strength is the thoroughness with which mortality information was sought after, including medical chart review and record linkage of mortality information using the State of Rio de Janeiro mortality database. Brazil instituted national reporting of mortality in 1977 and has a very robust system in place with high completeness of data regarding the cause of death, particularly in Rio de Janeiro [65]. Additionally, our institute provides primary, secondary, and tertiary care such that most deaths occurred within our institute and were thus captured locally. Nevertheless, it is important to highlight a limitation of our population with regards to how representative they are of PWH cared for at other clinics. INI is the largest HIV care provider in Rio de Janeiro state and the level and quality of care equal that of university hospitals and/or other research-based HIV services. Specifically, INI provides specialty and tertiary care, services not provided at the many primary care facilities. Given these peculiarities, PWH cared for at INI are not representative of PWH overall but can be taken as representative of those cared for at reference services of Brazil. Also noteworthy is the use of the multiple imputation techniques, thus avoiding the exclusion of participants with missing values and increasing the power of the statistical tests used in the models. Limitations include the maintenance of variables with evidence that the Cox proportional hazards assumption had not been verified [34, 68]. However, since there are limitations to proportional hazards hypothesis testing, we interpreted the effects as an average of risk ratios over the follow-up period [36, 69, 70].


Based on information from a cohort of PWH followed at a reference institution in Rio de Janeiro, early mortality rates were much higher than those observed in developed countries. This excess mortality is strongly associated with advanced disease and, therefore, with the patients’ level of immunodeficiency at the time of first visit. Late presentation of PWH to care is, therefore, one of the most significant problems for care providers and PWH. Among developing countries, Brazil was the first country to make ART available to all PWH. Thus, it is of fundamental importance that public policies are designed to promote early diagnosis and linkage to care, so patients can benefit from early ART initiation, preventing the evolution of the disease, thus increasing survival and improving the quality of life of PWH.

Availability of data and materials

The datasets generated and/or analyzed during the current study are not publicly available due Institutional Review Board (IRB) restrictions but are available from the corresponding author on reasonable request.


  1. Centers for Disease Control and Prevention. 1993 Revised Classification System for HIV infection and expanded surveillance case definition for AIDS among adolescents and adults. 1992. Accessed 9 Feb 2020.

  2. Brasil. Recomendações para Terapia Anti-Retroviral em Adultos e Adolescentes Infectados pelo HIV 2004. Brasília: Ministério da Saúde. Secretaria de Vigilância em Saúde. Programa Nacional de DST e Aids; 2004.

  3. Programa Nacional de DST e Aids. Recomendações para terapia anti-retroviral em adultos infectados pelo HIV: 2008. Brasília: Programa Nacional de DST e Aids; 2008.

    Google Scholar 

  4. Brasil. Protocolo Clínico e Diretrizes Clínicas para Manejo da Infecção pelo HIV em Adultos. Brasília, DF: Ministério da Saúde. Secretaria de Vigilância em Saúde. Departamento de Vigilância, Prevenção e Controle das Infecções Sexualmente Transmissíveis, do HIV/Aids e das Hepatites Virais; 2018.

  5. Grinsztejn B, Luz PM, Pacheco AG, Santos DVG, Velasque L, Moreira RI, et al. Changing mortality profile among HIV-infected patients in Rio de Janeiro, Brazil: shifting from AIDS to non-AIDS related conditions in the HAART Era. PLoS ONE. 2013;8:e59768.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  6. Ribeiro SR, Luz PM, Campos DP, Moreira RI, Coelho L, Japiassu A, et al. Incidence and determinants of severe morbidity among HIV-infected patients from Rio de Janeiro, Brazil, 2000–2010. Antivir Ther (Lond). 2014;19:387–97.

    Article  Google Scholar 

  7. Raboni SM, Ribeiro CE, Almeida SM, Telles JPM, Azevedo M, Schaitza GA. Impact of public health strategies on reducing AIDS mortality in southern Brazil. Int J STD AIDS. 2017;28:54–62.

    Article  PubMed  Google Scholar 

  8. Lima TdA, Beyrer C, Golub JE, Mota JCd, Malta MS, Silva CMFPd, et al. Inequalities in HAART uptake and differential survival according to exposure category in Rio de Janeiro, Brazil. Cad Saude Publica. 2018;34:e00009617.

    PubMed  Google Scholar 

  9. Rossi SMGd, Maluf ECP, Carvalho DS, Ribeiro CEL, Battaglin CRP. Impact of antiretroviral therapy under different treatment regimens. Rev Panam Salud Publica. 2012;32:117–23.

    Article  PubMed  Google Scholar 

  10. Mangal TD, Meireles MV, Pascom ARP, de Almeida CR, Benzaken AS, Hallett TB. Determinants of survival of people living with HIV/AIDS on antiretroviral therapy in Brazil 2006–2015. BMC Infect Dis. 2019;19:206.

    Article  PubMed  PubMed Central  Google Scholar 

  11. Luz PM, Girouard MP, Grinsztejn B, Freedberg KA, Veloso VG, Losina E, et al. Survival benefits of antiretroviral therapy in Brazil: a model-based analysis. J Int AIDS Soc. 2016;19:20623.

    Article  PubMed  PubMed Central  Google Scholar 

  12. Greco DB, Simão M. Brazilian policy of universal access to AIDS treatment: sustainability challenges and perspectives. AIDS. 2007;21(Suppl 4):S37-45.

    Article  PubMed  Google Scholar 

  13. Melo EA, Maksud I, Agostini R. Cuidado, HIV/Aids e atenção primária no Brasil: desafio para a atenção no Sistema Único de Saúde? Rev Panam Salud Publica. Organización Panamericana de la Salud. 2018;42:e151.

    Google Scholar 

  14. Lawn SD, Harries AD, Anglaret X, Myer L, Wood R. Early mortality among adults accessing antiretroviral treatment programmes in sub-Saharan Africa. AIDS. 2008;22:1897–908.

    Article  PubMed  Google Scholar 

  15. Grinsztejn B, Veloso VG, Friedman RK, Moreira RI, Luz PM, Campos DP, et al. Early mortality and cause of deaths in patients using HAART in Brazil and the United States. AIDS. 2009;23:2107–14.

    Article  PubMed  Google Scholar 

  16. Trickey A, May MT, Vehreschild J-J, Obel N, Gill MJ, Crane HM, et al. Survival of HIV-positive patients starting antiretroviral therapy between 1996 and 2013: a collaborative analysis of cohort studies. Lancet HIV. 2017;4:e349–56.

    Article  Google Scholar 

  17. Braitstein P, Brinkhof MWG, Dabis F, Schechter M, Boulle A, Miotti P, et al. Mortality of HIV-1-infected patients in the first year of antiretroviral therapy: comparison between low-income and high-income countries. Lancet. 2006;367:817–24.

    Article  PubMed  Google Scholar 

  18. Gupta A, Nadkarni G, Yang W-T, Chandrasekhar A, Gupte N, Bisson GP, et al. Early mortality in adults initiating antiretroviral therapy (ART) in low- and middle-income countries (LMIC): a systematic review and meta-analysis. PLoS ONE. 2011;6:e28691.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  19. Pascom ARP, Meireles MV, Benzaken AS. Sociodemographic determinants of attrition in the HIV continuum of care in Brazil, in 2016. Medicine (Baltimore). 2018;97:S69–74.

    Article  Google Scholar 

  20. Kerr LRFS, Mota RS, Kendall C, Pinho AdA, Mello MB, Guimarães MDC, et al. HIV among MSM in a large middle-income country. AIDS. 2013;27:427–35.

    Article  PubMed  Google Scholar 

  21. Johnson LF, Mossong J, Dorrington RE, Schomaker M, Hoffmann CJ, Keiser O, et al. Life expectancies of South African adults starting antiretroviral treatment: collaborative analysis of cohort studies. PLoS Med. 2013;10:e1001418.

    Article  PubMed  PubMed Central  Google Scholar 

  22. May M, Gompels M, Delpech V, Porter K, Post F, Johnson M, et al. Impact of late diagnosis and treatment on life expectancy in people with HIV-1: UK Collaborative HIV Cohort (UK CHIC) Study. BMJ. 2011;343:d6016.

    Article  PubMed  PubMed Central  Google Scholar 

  23. Campos D, De C, Vieira Lisboa S, Luiz A, Matzenbacher, Grinsztejn B, et al. Banco de dados de indivíduos HIV positivos para fins de pesquisa clínica: elaboração e atualização. Florianópolis; 2006. p. 6.

  24. Pacheco AG, Saraceni V, Tuboi SH, Moulton LH, Chaisson RE, Cavalcante SC, et al. Validation of a hierarchical deterministic record-linkage algorithm using data from 2 different cohorts of human immunodeficiency virus-infected persons and mortality databases in Brazil. Am J Epidemiol. 2008;168:1326–32.

    Article  PubMed  PubMed Central  Google Scholar 

  25. CDC. AIDS-defining conditions. Atlanta: Centers for Disease Control and Prevention; 2008.

  26. Colosimo EA, Giolo SR. Análise de sobrevivência aplicada. São Paulo: Edgard Blücher; 2006.

    Google Scholar 

  27. Harrell Jr FE. Regression modeling strategies: with applications to linear models, logistic regression, and survival analysis. Switzerland: Springer; 2010. Accessed 7 Sept 2019.

  28. Van Buuren S, Brand JPL, Groothuis-Oudshoorn CGM, Rubin DB. Fully conditional specification in multivariate imputation. J Stat Comput Simul. 2006;76:1049–64.

    Article  Google Scholar 

  29. Allison PD. Missing data. Thousand Oaks: Sage Publications; 2002.

    Book  Google Scholar 

  30. Berglund P, Heeringa SG. Multiple Imputation of missing data using SAS. Cary: SAS Institute; 2014.

    Google Scholar 

  31. R Core Team. R: a language and environment for statistical computing. Vienna: R Foundation for Statistical Computing; 2021.

  32. Harrel Jr. FE. Hmisc: Harrell Miscellaneous. R package version 4.5-0. 2021. Accessed 5 Jan 2019.

  33. Harrel Jr. FE. rms: Regression Modelling Strategies. R package version 5.1-3.1. 2019.

  34. Grambsch PM, Therneau TM. Proportional hazards tests and diagnostics based on weighted residuals. Biometrika. 1994;81:515–26.

    Article  Google Scholar 

  35. Therneau TM. survival: Survival Analysis. R package version 3.2–11. 2021.

  36. Allison PD. Survival analysis using SAS: a practical guide. 2nd ed. Cary: SAS Institute; 2010.

    Google Scholar 

  37. Teeraananchai S, Kerr SJ, Amin J, Ruxrungtham K, Law MG. Life expectancy of HIV-positive people after starting combination antiretroviral therapy: a meta-analysis. HIV Med. 2017;18:256–66.

    Article  CAS  PubMed  Google Scholar 

  38. Brennan AT, Long L, Useem J, Garrison L, Fox MP. Mortality in the first 3 months on antiretroviral therapy among HIV-positive adults in low- and middle-income countries: a meta-analysis. J Acquir Immune Defic Syndr. 2016;73:1–10.

    Article  CAS  PubMed  Google Scholar 

  39. Bisson GP, Ramchandani R, Miyahara S, Mngqibisa R, Matoga M, Ngongondo M, et al. Risk factors for early mortality on antiretroviral therapy in advanced HIV-infected adults. AIDS. 2017;31:2217–25.

    Article  PubMed  Google Scholar 

  40. Marazzi MC, Liotta G, Germano P, Guidotti G, Altan AD, Ceffa S, et al. Excessive early mortality in the first year of treatment in HIV type 1-infected patients initiating antiretroviral therapy in resource-limited settings. AIDS Res Hum Retrovir. 2008;24:555–60.

    Article  CAS  PubMed  Google Scholar 

  41. Tuboi S, Schechter M, McGowan C, Cesar C, Krolewiecki A, Cahn P, et al. Mortality during the first year of potent antiretroviral therapy in HIV-1-Infected patients in 7 sites throughout Latin America and the Caribbean. JAIDS J Acquir Immune Defic Syndr. 2009;51:615–23.

    Article  PubMed  Google Scholar 

  42. Gesesew HA, Ward P, Woldemichael K, Mwanri L. Early mortality among children and adults in antiretroviral therapy programs in Southwest Ethiopia, 2003–15. PLoS ONE. 2018;13:e0198815.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  43. Liao L, Xing H, Su B, Wang Z, Ruan Y, Wang X, et al. Impact of HIV drug resistance on virologic and immunologic failure and mortality in a cohort of patients on antiretroviral therapy in China. AIDS. 2013;27:1815–24.

    Article  PubMed  Google Scholar 

  44. Johannessen A, Naman E, Ngowi BJ, Sandvik L, Matee MI, Aglen HE, et al. Predictors of mortality in HIV-infected patients starting antiretroviral therapy in a rural hospital in Tanzania. BMC Infect Dis. 2008;8:52.

    Article  PubMed  PubMed Central  Google Scholar 

  45. Pellowski JA, Kalichman SC, Matthews KA, Adler N. A pandemic of the poor: social disadvantage and the U.S. HIV epidemic. Am Psychol. 2013;68:197–209.

    Article  PubMed  PubMed Central  Google Scholar 

  46. Marteleto LJ. Educational inequality by race in Brazil, 1982–2007: structural changes and shifts in racial classification. Demography. 2012;49:337–58.

    Article  PubMed  Google Scholar 

  47. Marteleto L, Dondero M. Racial inequality in education in Brazil: a twins fixed-effects approach. Demography. 2016;53:1185–205.

    Article  PubMed  Google Scholar 

  48. de Souza PF, Ribeiro CAC, Carvalhaes F. Desigualdade de oportunidades no Brasil: considerações sobre classe, educação e raça. Revista Brasileira de Ciências Sociais. 2010;25:77–100.

    Article  Google Scholar 

  49. IBGE. PNAD Contínua - Características gerais dos domicílios e dos moradores 2018. Brasília: IBGE, Coordenação de Trabalho e Rendimento; 2019.

    Google Scholar 

  50. Antinori A, Coenen T, Costagiola D, Dedes N, Ellefson M, Gatell J, et al. Late presentation of HIV infection: a consensus definition. HIV Med. 2011;12:61–4.

    Article  CAS  PubMed  Google Scholar 

  51. Wang Y, Hunt K, Nazareth I, Freemantle N, Petersen I. Do men consult less than women? An analysis of routinely collected UK general practice data. BMJ Open. 2013;3:e003320.

    Article  PubMed  PubMed Central  Google Scholar 

  52. Akinyemi JO, Adesina OA, Kuti MO, Ogunbosi BO, Irabor AE, Odaibo GN, et al. Temporal distribution of baseline characteristics and association with early mortality among HIV-positive patients at University College Hospital, Ibadan, Nigeria. Afr J AIDS Res. 2015;14:201–7.

    Article  PubMed  Google Scholar 

  53. Boulle A, Schomaker M, May MT, Hogg RS, Shepherd BE, Monge S, et al. Mortality in patients with HIV-1 infection starting antiretroviral therapy in South Africa, Europe, or North America: a collaborative analysis of prospective studies. PLoS Med. 2014;11:e1001718.

    Article  PubMed  PubMed Central  Google Scholar 

  54. Suligoi B, Zucchetto A, Grande E, Camoni L, Dal Maso L, Frova L, et al. Risk factors for early mortality after AIDS in the cART era: a population-based cohort study in Italy. BMC Infect Dis. 2015;15:229.

    Article  PubMed  PubMed Central  Google Scholar 

  55. UNAIDS. 2020 Global AIDS Update⁠—Seizing the moment⁠—tackling entrenched inequalities to end epidemics. 2020;384.

  56. Domingues RMSM, Szwarcwald CL, Souza PRB, Leal MdC. Prenatal testing and prevalence of HIV infection during pregnancy: data from the ‘Birth in Brazil’ study, a national hospital-based study. BMC Infect Dis. 2015;15:100.

    Article  PubMed  PubMed Central  Google Scholar 

  57. Tomasi E, Fernandes PAA, Fischer T, Siqueira FCV, Silveira DS da, Thumé E, et al. Quality of prenatal services in primary healthcare in Brazil: indicators and social inequalities. Cadernos de Saúde Pública. Escola Nacional de Saúde Pública, Fundação Oswaldo Cruz; 2017;33. Accessed 3 Apr 2021.

  58. Coelho L, Grinsztejn B, Castilho JL, De Boni R, Quintana MSB, Campos DP, et al. Mortality among HIV-infected women, heterosexual men, and men who have sex with men: insights from an observational cohort study. Lancet HIV. 2016;3:e490–8.

    Article  PubMed  PubMed Central  Google Scholar 

  59. Sobrino-Vegas P, Moreno S, Rubio R, Viciana P, Bernardino JI, Blanco JR, et al. Impact of late presentation of HIV infection on short-, mid- and long-term mortality and causes of death in a multicenter national cohort: 2004–2013. J Infect. 2016;72:587–96.

    Article  PubMed  Google Scholar 

  60. Santos ME, Protopopescu C, Ribeiro RA, Benzaken AS, Pereira GFM, Stevens A, et al. Population attributable fractions of mortality in people living with HIV: roles of delayed ART, hepatitis coinfections and social factors. AIDS. 2020;34:1843–54.

    Article  CAS  PubMed  Google Scholar 

  61. Saraceni V, Cohn S, Cavalcante SC, Pacheco AGF, Moulton LH, Chaisson RE, et al. Prevalent tuberculosis (TB) at HIV diagnosis in Rio de Janeiro, Brazil: the TB/HIV in Rio (THRio) Cohort. J Acquir Immune Defic Syndr. 2014;67:98–101.

    Article  PubMed  PubMed Central  Google Scholar 

  62. Saraceni V, Benzaken AS, Pereira GFM, Andrade KB, Oliveira PB, Arakaki-Sanchez D, et al. Tuberculosis burden on AIDS in Brazil: a study using linked databases. PLoS ONE. 2018;13:e0207859.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  63. Mendes WG, Silva CMFPd. Homicide of lesbians, gays, bisexuals, travestis, transexuals, and transgender people (LGBT) in Brazil: a spatial analysis. Cien Saude Colet. 2020;25:1709–22.

    Article  PubMed  Google Scholar 

  64. IeDEA and COHERE Cohort Collaborations. Global trends in CD4 cell count at the start of antiretroviral therapy: collaborative study of treatment programs. Clin Infect Dis. 2018;66:893–903.

    Article  Google Scholar 

  65. Mutasa-Apollo T, Shiraishi RW, Takarinda KC, Dzangare J, Mugurungi O, Murungu J, et al. Patient retention, clinical outcomes and attrition-associated factors of HIV-infected patients enrolled in Zimbabwe’s National Antiretroviral Therapy Programme, 2007–2010. PLoS ONE. 2014;9:e86305.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  66. Tachbele E, Ameni G. Survival and predictors of mortality among human immunodeficiency virus patients on anti-retroviral treatment at Jinka Hospital, South Omo, Ethiopia: a six years retrospective cohort study. Epidemiol Health. 2016;38:e2016049.

    PubMed  PubMed Central  Google Scholar 

  67. Tran DA, Ngo AD, Shakeshaft A, Wilson DP, Doran C, Zhang L. Trends in and determinants of loss to follow up and early mortality in a rapid expansion of the antiretroviral treatment program in Vietnam: findings from 13 outpatient clinics. PLoS ONE. 2013;8:e73181.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  68. Mills M. Introducing survival and event history analysis. 1st ed. Los Angeles: SAGE Publications Ltd; 2011.

    Book  Google Scholar 

  69. Wasserstein RL, Lazar NA. The ASA statement on p-values: context, process, and purpose. Am Stat. 2016;70:129–33.

    Article  Google Scholar 

  70. Greenland S, Senn SJ, Rothman KJ, Carlin JB, Poole C, Goodman SN, et al. Statistical tests, P values, confidence intervals, and power: a guide to misinterpretations. Eur J Epidemiol. 2016;31:337–50.

    Article  PubMed  PubMed Central  Google Scholar 

Download references


The corresponding author would like to thank the Escola Nacional de Saúde Pública Sergio Arouca (ENSP/FIOCRUZ) in providing a Ph.D. scholarship.


Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) and Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro (FAPERJ). This work was supported by the NIH-funded Caribbean, Central and South America network for HIV epidemiology (CCASAnet), a member cohort of the International Epidemiologic Databases to Evaluate AIDS (leDEA) (U01AI069923).

Author information

Authors and Affiliations



Conceived study: PHACL, LEC, VGV, BG, PML. Study implementation and procedures: PHACL, SWC, RIM, LEC, PML. Data analysis: PHACL, PML. Wrote paper: PHACL, LEC, SWC, RIM, VGV, BG, PML. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Pedro H. A. C. Leite.

Ethics declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of INI/Fiocruz (CAAE 0032.0.009.000-10) and was conducted according to the principles expressed in the Declaration of Helsinki. All patients signed a free and informed consent term before entering the study and all records/information were anonymized before analysis.

Consent for publication

Not applicable.

Competing interests

Authors declare no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1: Figure S1.

Plots of the Schoenfeld residuals against transformed time for the very early mortality model (0 to 90 days). Figure S2. Plots of the Schoenfeld residuals against transformed time for the early mortality model (91 to 365 days).

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Leite, P.H.A.C., Coelho, L.E., Cardoso, S.W. et al. Early mortality in a cohort of people living with HIV in Rio de Janeiro, Brazil, 2004–2015: a persisting problem. BMC Infect Dis 22, 475 (2022).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: