Mathematical models to quantify and optimize HIV intervention in British Columbia

CMS Meeting Hamilton

2014-12-07

UBC: Bernhard Konrad, Daniel Coombs

BCCDC: Warren Michelow, Mark Gilbert

General background

HIV in Canada

Canada

  • 70,000 HIV+ (2009)
  • 2,500 new diagnosis every year
  • one new infection every 3h

BC

  • 400 new diagnosis every year
  • 3500 unaware of infection (25%)

Vancouver

  • 6000 HIV+ (1.2%)
  • 250 new diagnosis every year      
  • ~50% MSM, ~25% IDU, ~25% HET

Goal: Understand the non-improvement for MSM

Modelling Epidemics

\frac{dS}{dt} = -\beta SI/N
​dt​​dS​​=−βSI/N
\frac{dI}{dt} = \beta SI/N - \gamma I
​dt​​dI​​=βSI/N−γI
\frac{dR}{dt} = \gamma I
​dt​​dR​​=γI

Find parameters for Vancouver MSM population

Infection rate = (Rate of contact) x (Per-contact risk)

Removal rate

  • Death and anti-retroviral treatment
  • Sexually active
  • Stable partnerships
  • Number of sexual contact
  • Network of contacts
  • Type of exposure
  • Serosorting

Study 1: TasP

Treatment as prevention:

  • Treatment for personal and population benefits
  • Infectiousness correlates with viral load
  • Treatment decreases VL => less infectious

Study 1: TasP

Study 1: TasP

  • BC Centre for Excellence in HIV/AIDS
  • Vancouver resource rich
    • Free treatment by universally accessible health care
    • centralized laboratory monitoring
  • For every 1% increase in number of individuals on suppressed HAART, HIV incidence rate decreased by 1%
  • Declining trend in new HIV diagnosis in BC unique in Canada
  • TasP favourable cost-benefit ratio

Study 1: TasP

Choose parameters to fit Vancouver MSM:

  • Infection rate    <--> Prevalence
  • Testing rate       <--> HIV awareness
  • Treatment rate <--> Treatment percentage

Study 2: Acute infection

  • Infectiousness correlates with viral load
  • Viral load very high shortly after exposure
  • How much of the epidemic is driven by acute infections?

Study 2: Acute infection

  • Pooled NAAT tests at HIV testing clinics
  • Recruitment between April 2009 and June 2012
  • Very difficult to catch early infections:
    • 13 participants with acute infection
    • 12 with early infection
  • Questionnaires on sexual activity, number of partners, use of condoms, etc.

Goal: Find acute/early infections in Vancouver

Study 2: Acute infection

Findings:

  • 64% were not expecting a positive result
  • Likelihood of "risky sex" decreased for first 3 months after diagnosis, remained low during follow-up, but increased over time
  • All participants eventually resumed sexual activity, 92% resumed anal sex
  • The 25 cases would not have been detected otherwise, hence prevented secondary cases.

Study 2: Acute infection

Study 2: Acute infection

Choose parameters to fit Vancouver MSM:

  • # acute infections found

Study 2: Acute infection

Choose parameters to fit Vancouver MSM:

  • Effect of recent diagnosis
  • Difficult to find early infections
  • Are there any differences between the infected and non-infected group?

Study 3: Negative cohort

Why study HIV negative cohort?

  • 166 HIV- MSM (tested at health clinic in Vancouver)
  • 1 year follow-up with 4 questionnaires
  • Questions similar to positive cohort
  • 33 men reporting risky sex in the past 6 months, those invited to interview.

Study design

Study 3: Negative cohort

  1. Have you had a HIV test in the LAST 5 months?
  2. How many men have you had anal sex with in the LAST 3 months?
  3. How often do you use condoms when you are the top when having anal sex?

Example questions:

Findings: ??? - in progress

  • Is the at-risk population homogeneous?
  • If not,
    • how much do they differ?
    • how stable are the compartments?

Study 3: Negative cohort

Distinguish between high-risk, low-risk

Different infection rate and testing behaviour

Study 3: Negative cohort

(How often) Do people switch between groups?

Current and Future Work

  • Learn from HIV negative cohort
    • Characterize, quantify risk groups
    • How stable are risk groups?
  • Combine study results into single model(s)
  • Are ODEs the best choice?
    • Exponential rates
    • Relatively small populations
  • Predict outcome of interventions

Thank you!

Made with Slides.com