HIV/AIDS AND SEXUAL NETWORS Dimitri Fazito (CEDEPLAR/UFMG) International Workshop on Demography of Lusophane African Countries 22nd - 24th of May, 2007 : HIV/AIDS AND SEXUAL NETWORS Dimitri Fazito (CEDEPLAR/UFMG) International Workshop on Demography of Lusophane African Countries 22nd - 24th of May, 2007
The global AIDS epidemic in 2006 : The global AIDS epidemic in 2006 An estimated 39.5 million people are living with HIV/AIDS. The vast majority are aged 15-49 years.
4.3 million people were newly infected with the virus in 2006.
2.9 million people died of AIDS.
There are 11,000 new infections and nearly 8,000 deaths daily.
2.3 million children (under 15 years) are living with HIV.
Nearly one-third of the world’s HIV-infected people – or 13 million – lives in countries classified by the World Bank as heavily burdened by debt. Of the 41 poorest and most indebted countries, 34 are in sub-Saharan Africa.
Vulnerable Groups : Vulnerable Groups Children: Globally, 2.3 million children are living with HIV;
Women: 2.5 times more vulnerable to HIV infection than men. UNAIDS estimates that 60% of all people living with HIV in sub-Saharan Africa were women;
Young People: More than one-third of all people living with HIV/AIDS are under the age of 25, accounting for 2 million infections each year. In sub-Saharan Africa, more than half of all new infections are among young people, with girls being particularly affected;
Sex Workers: High rates of HIV infection have been found among sex workers. Higher proportion in Asia, especially among women;
Injecting Drug Users: UNAIDS estimates that injecting drug use accounts for one-third of new infections outside sub-Saharan Africa, especially in Europe, North and Latin America and Asia;
Prisioners: The prevalence of HIV infection in prisons is higher than that in the general population. In South Africa it is estimated that 41% of prisioners are HIV positive.
Estimated HIV/AIDS, 2003 : Estimated HIV/AIDS, 2003
Slide5 : Table 1: Maplecroft’s HIV/AIDS Index (HAI) Worldwide (2006) HIV/AIDS Index (HAI): level of prevalence in adults (%) + total number of infected adults (year) + country’s capacity of disease contention
Countries Studied: 148
Category Risk: extreme (0–2.5), high (2.5–5.0), medium (5.0–7.5) and low (7.5–10) Source: Maplecroft & UNAIDS, 2007
Why Networks Matter : Why Networks Matter Sexual behaviors are socially sanctioned in groups (eg. dyads, personal networks, cliques and cores) within the context of social norms (cultural values and social interactions);
Why Networks Matter : Why Networks Matter How social structure influences sexual behavior?
Slide8 : Local network involvement
The strength and qualities of particular network ties (“direct embeddedness”)
Degree, tie strength, condom use, etc
One’s position in the overall network (“structural embeddedness”)
Centrality, local-network density, transitivity, membership.
Global network structure
The global structure of the network affects how goods can travel throughout the population.
Distance distribution
Connectivity structure
Among the most challenging tasks for modeling networks is building a robust link from the first to the second. Why Networks Matter
Why Networks matter : Why Networks matter Disease transmission occurs through diffusion networks ( “one-by-one” personal contacts);
Sexual risk is a function of relational and structural composition of networks (dyads and cliques);
Network ties established within structuring environments do not occur at random – the network “clustering” effect;
A simplified multi-layered framework : A simplified multi-layered framework Social units (y)
individuals
...
Ties among social units (x)
person-to-person
...
Settings (s)
geographical
sociocultural
... For example:
Interactions between tie variables depend on node attributes
social selection effects
Interactions between ties depend on proximity through settings
context effects
The Network “Clustering” Effect : The Network “Clustering” Effect When different processes can lead to similar macro signatures:
For example: “clustering” typically observed in social nets
Sociality – highly active persons create clusters (eg. Leaders, drug-dealers, brokers)
Homophily – assortative mixing by attribute creates clusters (eg. Ethinic cliques, religious communities)
Triad closure – triangles create clusters (eg. Work and schoolmates)
Slide12 : Friend of a friend, or birds of a feather?
Homophily:: People tend to chose friends who are like them, in grade, race, etc. (“birds of a feather”), triad closure is a by-product
Transitivity:: People who have friends in common tend to become friends (“friend of a friend”), closure is the key process
Slide13 : Why do Networks Matter? Local vision
Slide14 : Why do Networks Matter? Global vision
Slide15 : Networks are structurally cohesive if they remain connected even when nodes are removed Node Connectivity 0 1 2 3 Disease Transmission and the Network Density
Variation in the Timing and Intensity of HIV Epidemic : Variation in the Timing and Intensity of HIV Epidemic The rate of sexual partner acquisition
The impact of “core groups” activities
The presence of different sexually transmitted diseases (infection amplification)
Higher mobility (migration)
The rate of concurrent (simultaneous) sexual partnerships and duration
The rate of partnership stability
Definition of Concurrency : Definition of Concurrency
Why concurrency matters : Why concurrency matters Less protection afforded by sequence 2. virus-eye view: Less time lost locked in partnership 3. Larger “connected
component” in the
network
Connectivity in sparse networks : Connectivity in sparse networks High degree hubs
Low degree linking
Both have mean degree = 1.9
Connectivity in sparse networks and Concurrency : Connectivity in sparse networks and Concurrency “Low degree” “High degree” -Some individuals are highly connected (core transmitters) -Perceived as “high risk” -Potentially more likely to motivate prevention behavior -Most individuals are less connected -Perceived as “lower risk” -Potentially less likely to motivate prevention behavior
Slide21 : Structural degree and cohesion gives rise automatically to a clear notion of embeddedness, since cohesive sets nest inside of each other supporting concurrency partnership and contagion 17 18 19 20 2 22 23 8 11 10 14 12 9 15 16 13 4 1 7 5 6 3 2 Structural Properties: Concurrency and Speed of HIV/AIDS Transmission
Degree Networks, Cohesion, Concurrency and transmission core : Degree Networks, Cohesion, Concurrency and transmission core Bicomponents
in red Source: Martina Morris, Univ. of Washingtion, used with permission from a presentation given at a
meeting on concurrent sexual parnerships and sexually transmitted infections at Princeton University, 6 May 2006.
Slide23 : Worldwide, almost all studies show increased risks
with increased sexual partners Partner reduction has been associated with declines in HIV at the population level in both concentrated and generalized epidemic settings Multiple sexual partnerships
Uganda vs. US and Thailand : Morris et al. (2006) Uganda vs. US and Thailand
Empirical Findings: Rate of Concurrency and Duration : Thailand’s population has many more partners, but the network connections are extremely short duration.
Despite much higher contact rates, transmission dynamics are dampened, and prevalence will remain low
Uganda’s population has fewer partners, but the network is more continuously connected over time.
This long term concurrency amplifies transmission dynamics, allowing prevalence to rise much higher.
Empirical Findings: Rate of Concurrency and Duration
Concluding Remarks: the importance of networks : Concluding Remarks: the importance of networks Large populations exhibit network structure
Social, sexual, infrastructure, transportation
Large epidemics need to be understood as many small epidemics linked by networks (clustering and overlapping effects)
Incorporating “multi-scale” structure of the world in epidemic models can explain multi-modality and resurgence of HIV/AIDS
“Rare events” (e.g. one person getting on a plane) can have big consequences. Such events can be modeled by Network Models (eg. Small World, Random Graphs, Free Scale Networks)
Population structure itself can be used as control measure (e.g. intermediate connections)