Trends of HIV Mortality between 2001 and 2018: An Observational Analysis

Since the beginning of the epidemic in the early 1980s, HIV-related illnesses have led to the deaths of over 32.7 million individuals. The objective of this study was to describe current mortality rates for HIV through an observational analysis of HIV mortality data from 2001 to 2018 from the World Health Organization (WHO) Mortality Database. We computed age-standardized death rates (ASDRs) per 100,000 people using the World Standard Population. We plotted trends using locally weighted scatterplot smoothing (LOWESS). Data for females were available for 42 countries. In total, 31/48 (64.60%) and 25/42 (59.52%) countries showed decreases in mortality in males and females, respectively. South Africa had the highest ASDRs for both males (467.7/100,000) and females (391.1/100,000). The lowest mortalities were noted in Egypt for males (0.2/100,000) and in Japan for females (0.01/100,000). Kyrgyzstan had the greatest increase in mortality for males (+6998.6%). Estonia had the greatest increase in mortality for females (+5877.56%). The disparity between Egypt (the lowest) and South Africa (the highest) was 3042-fold for males. Between Japan and South Africa, the disparity was 43,454-fold for females. Although there was a decrease in mortality attributed to HIV among most of the countries studied, a rising trend remained in a number of developing countries.


Introduction
In 2019, around 38 million individuals were living with human immunodeficiency virus (HIV), and 690,000 died due to HIV-related illness. Since the beginning of the epidemic in the early 1980s, HIV-related illnesses have led to the deaths of over 32.7 million individuals [1]. However, there were approximately 39% fewer deaths due to HIV in 2019 than in 2010 [2][3][4].
Overall decreases in HIV-related mortality have mostly been attributed to the significant improvements in HIV/AIDS management, including highly active antiretroviral combination therapies [5,6]. Newer medications have better side effect profiles as well, which has improved adherence [7,8]. Highly Active Antiretroviral Therapy (HAART) has also contributed to the increased life expectancy of patients with HIV [7]. Along with the Joint United Nations Program on HIV/AIDS (UNAIDS), the WHO has framed various policies and guidelines to decrease the global burden of morbidity and mortality due to HIV [9]. Strategies were developed throughout the world to address the needs of high-risk populations and increase their awareness [10,11]. These multifactorial efforts have led to an overall decrease in HIV mortality worldwide; however, certain areas have continued to see rising mortality rates. Active efforts are needed in those areas [2].
In 2014, UNAIDS and partners set goals they named the 90-90-90 targets. These goals were to diagnose 90% of all HIV-positive people, to provide antiretroviral therapy (ART) for 90% of those diagnosed, and to achieve viral suppression for 90% of those treated by 2020 [12,13]. This led to more widespread and earlier administration of ART. It was thus important to examine how the setting of these targets-and the subsequent efforts to meet them-impacted global HIV mortality trends.
The primary objective of this study was to describe current ASDRs for HIV. The secondary objectives were to compute the absolute and percentage changes in ASDRs, categorized by country and gender, through an observational analysis of HIV mortality data from 2001 to 2018 from the World Health Organization (WHO) Mortality Database. We also compared HIV mortality trends in 48 nations between 2001 and 2018.

Data Sources
We extracted HIV mortality data from the WHO Mortality Database from 2001 to 2018 for the member nations with available data in October 2020. Countries were divided based on their WHO specified regions. Data were extracted based upon the International Classification of Diseases-10th Edition (ICD-10) diagnosis codes B20, B21, B22, B23, B24, and R75 for HIV. The WHO evaluated the quality of these data to ensure comparability and reliability, without adjustment for underreporting. The proportion of all deaths registered in the population covered by the vital registration system for a given country (referred to as completeness) were estimated by the WHO for the latest available year. For member states with incomplete vital registration systems, demographic techniques were used by the WHO to estimate the level of completeness of death recording for the specified population, in order to allow for the calculation of the death rate [14]. For inclusion criteria, we first evaluated the database to determine the countries with available data. Out of 194 member countries of the WHO, data on HIV mortality were available for 118 countries. Based on the WHO mortality database 2009-2017 completeness data, 107 countries with >20% completeness were included in our data review. Out of these 107 countries, we further excluded 59 countries that either did not have data for five years or had significant data breaks, defined as greater than three consecutive years.
Crude mortality rates were dichotomized by gender and reported by year. We computed age-standardized death rates (ASDRs) per 100,000 people using the world standard population. The ASDR was defined as mortality, weighted to the distribution of mortality per 5-year age group, according to the world standard population [15]. This removed the effects of historical events on age structure and controlled for differences in age structure in populations, producing age-specific mortality rates and more representative data. The estimated level of coverage for deaths with a recorded cause for death was calculated by dividing the total deaths reported for a country in a given year from the vital registration system by the total estimated deaths for that year for its national population. The national population estimates used were those of the UN Population Division. Best estimates of death rates by age and sex (adjusted for incompleteness and incomplete coverage) were applied to the national population data to obtain total estimated deaths. The WHO estimated coverage for a member state may have been less than 100% due to incompleteness of registration, coverage of only some parts of the national population, or differences between the vital registration the UN estimated de facto population. Population and birth recording in all countries are specified in the data as per the WHO standard for inclusion in the database [14].

Statistical Analyses
We computed male and female mortality rates and used a locally weighted scatterplot smoothing (LOWESS) plot to fit the rates of male and female mortality using SAS v9.4 (SAS, Cary, NC, USA) ( Figure 1). LOWESS was only used to make plots to visualize countryspecific trends and not to model changes across the countries. Mortality data were missing in a small subset of countries in the WHO Mortality Database for one or more calendar years. We excluded countries with missing data of more than three consecutive years during the observation period. There were no other modifications to the data. Changes in ASDR over the observation period are calculated as crude absolute differences between first and last data points for the earliest and most recent years available. Similar to our previous studies, percentage change was calculated as [(End ASDR − Start ASDR)/Start ASDR] * 100 for each gender and country [16,17]. We have also provided absolute changes in the ASDR. This can serve as a comparator to the percentage changes, especially for the countries with low baseline mortality.

Results
We analyzed data from 48 countries from the following regions (as classified by the  Figure 2 show mortality data for the most recent calendar year. South Africa had the highest ASDRs for both males (467.7/100,000) and females (391.1/100,000), whereas Egypt had the lowest ASDR for males (0.2/100,000) and Japan for females (0.01/100,000).
In the region of the Americas, the USA had the highest ASDR in 2007 for both males (47.04/100,000) and females (19.02/100,000), whereas Canada had the lowest ASDR in 2005 for both males (18.93/100,000) and females (5.38/100,000).
In the European region, Latvia had the highest ASDR in 2015 for males (57.91/100,000). For females, this was Estonia in 2016 (27.40/100,000), whereas Slovakia had the lowest ASDR in 2014 for males (0.29/100,000), followed by Hungary in 2017 (0.51/100,000). Turkey had the lowest ASDR in 2016 for females (0.24/100,000), followed by the Czech Republic in 2017 (0.31/100,000) and Finland in 2017 (0.41/100,000). Sufficient data were not available for women from Cyprus, Slovenia, Slovakia, and Malta. Out of the 34 European countries, half (17/34) had an ASDR of more than 5.26/100,000. Nine countries had an ASDR of more than 8.8/100,000 in men at the beginning of their respective study periods, ranging from 2001 to 2009, including Austria, France, Italy, Malta, Netherland, Romania, Spain, Serbia, and Switzerland. For the end of the study period, in years ranging from 2013 to 2018, 17 out of the 34 countries had an ASDR value of more than 3.4/100,000, and nine countries (26%) had an ASDR greater than 12/100,000. For females, we had data from 30 countries showing that half (15/30) had an ASDR greater than 1.7/100,000, and around a quarter (8/30) had an ASDR greater than 3/100,000, including Belgium, France, Italy, Luxembourg, Romania, Spain, Switzerland, and Serbia for years between 2001 and 2009 at the beginning of the study periods. For the end of the study periods, at years ranging from 2013 to 2018, half of the 30 countries had an ASDR value of more than 1.3/100,000, and eight countries (27%) had an ASDR with a value of more than 3.5/100,000, including Armenia, Estonia, Latvia, Romania, Georgia, Kyrgyzstan, Republic of Moldova, and Lithuania.
In the Western Pacific region, Malaysia had the highest ASDR in 2014 for males (24.54/100,000) as well as for females (3.79/100,000), followed by Singapore with an ASDR of 11.49/100,000 for males and 1.03/100,000 for females in the year 2017. In comparison, Japan had the lowest ASDR in 2017 for males (0.34/100,000) and females (0.01/100,000). Australia and New Zealand also had lower ASDRs.
In the Eastern Mediterranean region, Bahrain had the highest ASDR in 2014 for males (11.12/100,000) and females (1.6/100,000). In contrast, Egypt had the lowest mortality in 2015 for males (0.15/100,000) and females (0.12/100,000). Jordan and Kuwait showed an ASDR of 1.52/100,000 and 0.48/100,000 for males in 2015 and 2016, respectively. We did not have data for females in the other Eastern Mediterranean countries.
In Africa, out of the two countries included, South Africa had a higher ASDR in 2015 for males (467.27/100,000) as well as for females (391.09/100,000), whereas it was lower for Mauritius in 2018 for males (134.84/100,000) and females (31.22/100,000).
In the Southeast Asia region, Thailand showed an ASDR of 82.44/100,000 and 41.58/100,000 for males and females, respectively, in 2017, which was high compared to many other included countries.  Table 1 show HIV mortality at the beginning and the end of the study period. We have reported percentage change (PC) and absolute change (AC) in ASDR. A total of 31/48 countries (64.6%) showed a decrease in male mortality, whereas

Discussion
We analyzed mortality due to HIV/AIDS in 48 countries for males and 42 for females between 2001 and 2018 data from the WHO Mortality Database. At the end of the study period, the five leading countries with the highest ASDRs in males were South Africa, Mauritius, Thailand, Latvia, and Estonia. The five countries with the highest ASDRs in females were South Africa, Thailand, Mauritius, Estonia, and Latvia. Egypt, Japan, Kuwait, Hungary, and Sweden had the lowest ASDR in males. In females, Japan, Australia, Egypt, Turkey, and the Czech Republic had the lowest ASDRs. In males, 17 countries had percentage increases in ASDRs over the study period, with the highest increases in Kyrgyzstan, Georgia, Latvia, the Republic of Moldova, and Estonia (Range: +6998.62%-+436.49%). The same countries had an absolute increase in ASDR, with South Africa and Mauritius being the top two (+168.95 and +109.68, respectively). However, some of the countries with massively large percentage changes (ex Kyrgyzstan: +6998.62%) had absolute changes that were comparable to other countries (AC of +29.29). This seems to be likely owing to the exceedingly small proportion of HIV-related deaths during the beginning year (e.g., Kyrgyzstan ASDR: 0.42/100,000 in 2002). In females, 13 countries had a percentage change and absolute change increase, with the highest PC increase in Kyrgyzstan, Georgia, Latvia, Republic of Moldova, and Estonia (Range: +6998.62%-+676.97%) and the highest AC increase in the same countries as well as South Africa and Mauritius. This multifold mortality increase in some countries is worrisome. This indicates a significant risk that the progress made in slowing the HIV epidemic could be reversed without continued efforts.
Thirty-one countries had both percentage change and absolute change decreases in males. Twenty-five countries had percentage change and absolute change decreases in females, which is encouraging. In 2018, the disparity in ASDR rates in males, measured using rate ratios, between the lowest rates observed in Egypt and highest rates observed in South Africa was 3034. In females, rate ratios between the lowest rates observed in Japan and highest rates observed in South Africa were 41,211. These large differences help demonstrate that countries are on remarkably diverse trajectories with regard to the burden of HIV/AIDS. Trends from the Global Burden of Disease (GBD) database have shown high mortality in Sub-Saharan Africa and South Asia, as well as increased mortality in South East Asia. Apart from these regions, in our study, we have also observed increasing ASDRs in Eastern Europe and the Central Asia region (Latvia, Lithuania, Estonia, Kyrgyzstan, Georgia, and Armenia) as well as in the Central Europe region (the Czech Republic and Croatia) [18]. This finding highlights the importance of the harmonization of 'real world' data and the increasing need for independent multinational registries to better understand health-system-level differences in policy, clinical practice, and outcomes. Seventy-six million people have tested positive for HIV since the start of the epidemic. Thirty-eight million people globally were living with HIV in 2019, of whom 81% knew their status, 67% were accessing treatment, and 59% were virally suppressed. From 2010 to 2019, new HIV infections declined by 23%, from 2.1 million to 1.7 million. The global community has made considerable progress in addressing this unprecedented epidemic [1]. In our study, high HIV mortality areas were concentrated in resource-limited countries such as the African (South Africa, Mauritius) and the South East Asian regions (Malaysia). The WHO Mortality Database does not have data of prevalence or incidence. Studies in the past using the GBD database have shown that in 2015, 75% of new cases were in Sub-Saharan Africa, followed by South Asia (8.5%) and Southeast Asia (4.7%). The overall high prevalence and incidence could be affecting this high mortality substantially [18]. Antiretroviral therapy (ART) has extended the life expectancy for most people living with HIV. The percentage of people receiving ART has changed over time, overall and by region. A Study by Lau et al. showed that in 2014-2015, 22% of HIV-positive patients in Europe did not receive ART [19]. In low-and middle-income countries, two-thirds of HIV/AIDS-related deaths occur in individuals not on ART [20] 20 . The number of individuals who lost their lives to HIV/AIDS in 2015 is 45% lower than the peak of the epidemic in 1998 [13,21]. HIV/AIDS remains a significant cause of death despite these gains and global trends mask persistent regional and subregional variation [22].
As of June 2020, access to ART for HIV-positive patients was 72% in South Africa, 82% in Mauritius, and 80% in Thailand, compared to the global average of 67% [21]. Out of the 48 countries studied, ASDRs were highest in these countries, with 467.27/100,000, 134.84/100,000, and 82.44/100,000 in males, respectively, and 391.10/100,000, 41.58/100,000, and 31.22/100,000 in females, respectively [12,23]. The disease continues to claim lives in regions with high treatment coverage levels, this suppors a complex nature of the care model. After more than a decade of ART scale-up and more than 19.5 million people on treatment, an increase in drug resistance is inevitable. Data from several national surveys from low-income and middle-income countries have shown an increased prevalence of pretreatment HIV drug resistance to non-nucleoside reverse-transcriptase inhibitors. Drug resistance testing is largely not available in low-and middle-income countries and might be contributing to the high mortality despite good ART coverage [24]. Non-HIV-related conditions are also emerging as prominent health concerns in settings where ART is widely available. A prospective cohort study of PLHIV patients in Spain found that the most common non-AIDS events developing in these patients were psychiatric, hepatic, renal, cardiovascular, and malignant diseases [25]. Interestingly, the leading causes of non-AIDSassociated mortality in HIV-positive individuals were malignancy, most commonly: lung, followed by liver, with which hepatitis C virus was frequently found associated [21]. Additionally, hepatitis B virus coinfections are high in endemic regions (East and Southeast Asia) [26]. A large body of evidence indicates that HIV-positive individuals are at aboveaverage risk for cardiovascular disease [27,28]. Coinfection with tuberculosis remains a major public health concern and is the main cause of high mortality in HIV globally, particularly in resource-limited settings [29]. Older age, low CD4 count, and cigarette use are independent predictors of mortality in HIV-positive individuals [30]. Additional risk factors for non-AIDS events include a late stage at initial treatment, coinfection with hepatitis C virus, and injecting drug abuse [31]. This situation raises the question of how the health needs of HIV patients should be addressed beyond access to ART. Given these trends, the management of HIV patients in low-and middle-income countries should focus on reasons of virologic failure in patients receiving ART, providing universal access to better quality health services, and timely identification and management of potential non-AIDS complications.
The rapid expansion of ART began in 2004 [32]. One of the areas most influenced by the implementation and availability of ART was Eastern and Southern Africa. HIV and population mortality rates saw a steep decline [22]. Initially, HIV-positive individuals would only qualify for ART if they were WHO stage 3 or 4 or had a CD4 count less than 200 [26]. These requirements have been evolving ever since. A recent study in Zimbabwe demonstrated the effectiveness of "Test and Treat" protocol; of 972 people newly diagnosed HIV-positive, 94% were enrolled in HIV care, 79% were initiated on ART, and 71% were retained alive on ART at three months [33]. In 2014, UNAIDS also created the "90-90-90" approach in a further attempt to eradicate the virus [34]. Treatment, as well as pre-exposure prophylaxis (PrEP), has seen constant advancements influencing the mortality trends [29]. The Centers for Disease Control and Prevention recommends high-risk individuals take PrEP to limit their chances of contracting disease [35]. ART initiation is also one of the most effective ways to prevent and help resolve opportunistic infections. At the time of diagnosis of HIV and before initiation of ART, all individuals must be screened for latent tuberculosis (TB) infection [36], given the potential for reactivation of latent TB and the need for a mycobacterial treatment regimen before the initiation of ART [36].
Region-specific increases in HIV mortality are well known. In our data, South Africa, Thailand, and Mauritius were all found to have increased ASDRs. Several explanations for this pattern may exist. Despite the expansion and the availability of ART, resource-limited areas such as those in South Africa have associated delays in diagnosis and initiation of treatment, which ultimately led to increased mortality [17]. In addition, high-risk activities such as increased IVDA and sex trafficking are more common in certain areas of Asia and Africa and likely contribute to the increased HIV prevalence and mortality [37]. Particular areas where TB is more prevalent have an increased rate of reactivation which increases mortality risk in the HIV-positive population [36].
The strengths of this investigation include the use of annual mortality data collected from national surveillance statistics from the WHO. These data have made it possible to assess population-level trends over an extended observation period, allowing comparisons in directions rather than total annual mortality rates. Despite this study's strengths, there are several limitations which should be considered when interpreting the results. Out of 194 countries, data of only 48 countries have been analyzed due to the non-availability of data. Additionally, the start and end years of countries varied based on the data availability. In 2015, 75.4% of new cases were in Sub-Saharan Africa [18]. We understand that South Africa and Mauritius do not represent the population of Sub-Saharan Africa. However, due to the lack of availability and completeness of the data, we could not include these countries in our study. We did not attempt to assess the prevalence of morbidity associated with HIV, as the WHO Mortality Database does not report this. ICD coding for HIV mortality presents many challenges and is one of the limitations of all database studies. To achieve comparability over time, the WHO has tried to standardize ICD-10 coding for HIV with detailed guidelines. However, issues still exist when medical certification is not possible, especially in Sub-Saharan Africa, as they occur at home without a medical doctor. The WHO has developed computerized coding of verbal autopsy to address this issue using algorithmic methods. Validation studies showed that based on the symptoms indicative of AIDS-defining illnesses from verbal autopsy, 89% of deaths among HIVpositive individuals are attributable to HIV [38]. Finally, as with any observational study, causal statements regarding the observed trends cannot be made. However, one notable strength of using longitudinal data is the ability to comment on overall trends within individual countries after standardization and to report these differences between health systems. The discussion is provided to assist future researchers, policymakers, and public health experts in focusing their efforts.

Conclusions
Although there has been a decrease in mortality attributed to HIV among most of the countries studied, a rising trend remains in a number of developing countries. A renewed and heightened commitment to address this ongoing epidemic by these countries, healthcare agencies, and the global community is called for.
Author Contributions: Each of the authors significantly contributed to this manuscript. C.J., K.P., H.S., J.D.S. and J.S. contributed to this paper's concept and design. J.D.S., C.C., D.C.M. and C.J. were significantly involved in the statistical analysis. D.C.M., K.P., C.J., A.W., O.A.O., H.S., A.R., R.G., J.S. and J.D.S. were extensively involved in drafting the manuscript. J.S. and J.D.S. were engaged in the document's critical revision and contributed important intellectual content. All authors are responsible for the contents and have read and approved the manuscript for submission. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no specific grant from any funding agency in public, commercial, or not-for-profit sectors.
Institutional Review Board Statement: Not applicable for studies not involving humans or animals. The data was retrieved from open source deidentified data available online.

Informed Consent Statement: Not applicable.
Data Availability Statement: We extracted HIV mortality data from the WHO Mortality Database from 2001 to 2018 for the member nations with available data.

Conflicts of Interest:
The authors do not have any conflict of interest or any disclosures.