Determining HIV incidence and prevalence in Turkey with mathematical modeling

Infectious diseases pose a significant threat to public health causing outbreaks, human casualties and economic burden. They also affect tourism, agriculture, trade, animal and human health. In recent years, diseases such as swine flu, avian flu, Ebola, Crimean-Congo Hemorrhagic fever have resulted in multi-country outbreaks and have resulted in mortality and major economic losses. Sexually transmitted diseases (STDs) such as HIV, gonorrhea, syphilis, Hepatitis B and Hepatitis C often considered and modeled differently than other infectious diseases because there are often new cases occurring although they may not cause large and global outbreaks. HIV is one of the most common STDs and there has not been a cure or vaccine to eradicate and prevent this disease completely. Therefore, significance and prevention of HIV increase in global scale and it is considered as a major threat for public health. Since it was first identified in 1980s, 35 million people died from HIV-related illnesses and HIV is still among the top ten causes of death in the developing countries. Moreover, there has been approximately 1,5 million people became infected with HIV every year. In this study, first we summarize and prevent fundamentals of HIV regarding its epidemiology, transmission and prevention. HIV weakens human immune system gradually by attacking a certain type of cells called CD4 T-cells in the immune system. Thus, infected individuals whose immune system is weakened are vulnerable to any diseases which are called opportunistic infections. Stage 3 HIV, also known as AIDS, is categorized when patients' health get worse significantly and CD4 T-cells drops below 200 cells/m3. Without any treatment, patients with Stage 3 HIV may die. However, if the patient is diagnosed and initiates treatment, the life expectancy of a HIV-infected person is "near-normal" that means it is close to life-expectancy of a healthy person. Recommended treatment for HIV is known as antiretroviral treatment (ART) and improves the life expectancy and quality of life in HIV-infected persons. HIV affects society, economy and public health by reducing the quality of life for infected persons, increasing healthcare costs with lifetime treatment of infected persons and increasing health parities due to stigma. HIV can stay dormant years without any symptoms. Without the treatment, it may result with mortality and treatment also significantly reduces HIV transmission which is termed as treatment as prevention (TaSP). Therefore, early diagnosis and early initiation of treatment for HIV infected persons is considered as the cornerstone of HIV prevention. In order to decrease HIV incidence and its effect on public health, new prevention programs and treatment guidelines are provided. UNAIDS, founded by United Nations to fight against HIV, and created 90-90-90 strategy. This strategy calls on countries to reach the following goals: 90% of people living with HIV diagnosed by 2020; 90% of diagnosed people on ART by 2020 and 90% of people in treatment with fully suppressed viral load by 2020. With the help of 90-90-90, UNAIDS monitors the dynamics of HIV in the world to achieve firm actions fighting against HIV. Turkey is located close to high HIV prevalence in addition to growing tourism opportunities, refugee and migration issues, increasing population, developing economic and commercial sectors, a large number of people working abroad and increasing number of drug users. Combined with geographic proximity to high HIV areas, these issues contribute to Turkey being one of the HIV risk areas. In Turkey, there are 17884 patients living with HIV in 2017. Newly infected people per year has tripled in the last five years compared to previous years. Due to this sudden increase in HIV diagnoses and lack of surveillance to provide a comprehensive picture of HIV epidemic in Turkey, we conducted this study and developed models to understand current situation. Because the number of diagnosed and reported patients are significantly less than estimated number of people living with HIV, it is suggested that there is a substantial amount of people who are undiagnosed and not aware of their infection. In addition, diagnosed patients may have some issues regarding the medication such as resistance and lack of adherence, the number of people on treatment drops below than recommended level. Although there are some literature on HIV epidemic in Turkey, there is a need for a comprehensive model that summarize the current situation in Turkey regarding incidence, prevalence and continuum of care. In this study, HIV incidence and prevalence in Turkey are determined with the help of two different mathematical modeling methods. Since it is not always possible to perform trial and error experiments in real life, modeling methods are very common to analyze various infectious diseases. Firstly, we develop a Bernoulli process model and estimate the annual number of people living with HIV. Bernoulli process model is based on risky behaviors and how often and how much these risky behaviors are repeated. As a second model, we develop a deterministic dynamic compartmental model which focuses on transmission and progression of HIV while considering diagnosis and treatment rates of HIV infected persons. Population is divided into several subpopulations that indicated disease status and severity and each group is considered as a compartment. Disease status is defined as susceptible and HIV positive persons. HIV positive persons further divided into three separate compartments based on their diagnosis and treatment status: undiagnosed patients, diagnosed patients without treatment, patients on ART. Furthermore, disease severity is included in the model by separating HIV positive compartments to two groups based on their CD4 levels: CD4 level is above 200 cells/m3 and CD4 level is below 200 cells/m3 or diagnosed as AIDS. In total, our model has 8 different compartments including death compartments. This model takes into account heterosexual patients as well as two HIV risk groups such as men who have sex with men (MSM) and people who inject drugs (PWID). The model estimates total number of infections, the number of diagnosed patients and the number of persons on treatment each year, in addition to estimate same metrics for each risk group. We also develop the HIV continuum of care in the national level and evaluate whether Turkey can reach UNAIDS' 90-90-90 targets. Our model suggests that the number of new infections will continue to rise if there is no change in the diagnosis and treatment levels in Turkey. Our main outcomes in this study are the number of newly infected patients each year and the total number of patients living with HIV in Turkey. Our contribution is twofold in this study. First, as an application this is one of few modeling studies that predict HIV incidence and prevalence by each risk group in Turkey and determine the HIV continuum of care by estimating diagnosis and treatment rates. These results could impact public health and policy by improving HIV prevention if these results inform decision making in this field. Second, methodologically we develop two mathematical models that are often employed in infectious disease modeling and compare their advantages and disadvantages. We also applied a two-step calibration method where we decompose the calibration parameters based on their interaction. In summary, we determine annual number of HIV infections by risk populations in 2016 using the Bernoulli process model and determine the annual number of infections by risk population between 2005-2030 and HIV continuum of care. We conducted one-way sensitivity analysis in order to determine parameters that affect the model results significantly. In the second model, a two-step calibration method is employed using the number of HIV and AIDS diagnosis as well as AIDS deaths in 2014 and 2015. The results of our model suggest that diagnosis and treatment rates should be improved to reach 90-90-90 targets and to prevent steady increase in HIV incidence in Turkey. By 2030, the number of HIV infected persons may triple and AIDS deaths may also increase compared to current values if there are no changes in prevention of HIV to increase diagnosis and treatment levels. In the final sections of this study, we compare the results obtained by two models and explain their advantages and disadvantages. Lastly, we conclude this study by presenting our key findings, limitations and future work.