Background During the first few months of HIV infection (primary infection) viral loads, and hence infectivity, are high. Estimating the proportion of onward transmissions occurring during primary infection is crucial in predicting the population-level impact of post-primary interventions, such as behavioural changes following diagnosis, mass treatment and the transmission of drug resistant strains after treatment failure.

A recent study of discordant couples shows that the transmission rate in primary infection may be 10 times that of asymptomatic infection. The short duration of primary infection means that, despite high infectivity, transmission opportunities are limited and are highly sensitive to the sexual behaviour of the population. Previous analyses of the rapidly expanding San Francisco epidemic of the 1980s attributed a large proportion of transmissions to primary infection. The role of primary infection in a generalised epidemic in lower risk populations has yet to be  investigated.

Methods Re-analysis of published data on transmission in discordant partnerships, dependent on the stage of infection, allows estimation of the basic reproductive number, R₀, in this cohort. This study is also used to parameterise an HIV transmission model to investigate how the role of primary infection depends on the stage of the epidemic and the sexual behaviour of the population.

Results In the discordant couples cohort the first 5 months of infection contribute 0.84 new cases to the total R₀, suggesting that transmissions during primary infection alone cannot cause an epidemic. In an expanding epidemic doubling the risk behaviour of the population approximately doubles the proportion of cases due to primary infection and the exponential growth rate. Once an epidemic has reached equilibrium the proportion of new cases due to primary infection does not depend on the level of risk behaviour in the population.

Conclusion Primary infection was responsible for a significant proportion of transmissions in the rapidly expanding epidemic amongst high risk individuals in San Francisco. In lower risk populations with more developed epidemics, primary infection is responsible for only a small proportion of new infections. Identifying and advising those with asymptomatic infections to reduce their risk behavior can affect the epidemic. The opportunities for transmission of treatment-induced drug resistant strains may be more frequent than previously thought.