Article Text

Download PDFPDF
HIV seroconversion interval and demographic characteristics: no evidence of selection bias
  1. Philippe Vanhems1,2,
  2. Robert Allard3,
  3. Muriel Dhénain1,
  4. Christian Chidiac4,
  5. Dominique Peyramond4,
  6. Jean-Louis Touraine5,
  7. Christian Trépo6,
  8. Jacques Ritter2,
  9. Jacques Fabry1,2,
  10. The Collaborators Of The Lyon Hiv Hospital Database*
  1. 1Hygiène Hospitalière et Epidémiologie, Hôpital Edouard Herriot, Lyon, France
  2. 2Laboratoire d'Epidémiologie et de Santé Publique, INSERM U271, Université Claude Bernard, Lyon, France
  3. 3Department of Public Health, and McGill University, Montreal, Quebec, Canada
  4. 4Service de Maladies Infectieuses et Tropicales, Hôpital de la Croix Rousse, Lyon, France
  5. 5Service d'Immunologie Clinique, Hôpital Edouard Herriot, Lyon, France
  6. 6Service d'Hépatologie, Hôtel-Dieu, INSERM U271, Lyon, France
  1. Dr P Vanhems, Laboratoire d'Epidémiologie et de Santé Publique, INSERM U271, 8 av, Rockefeller, 69373 Lyon Cedex 08, France philipva{at}


Objectives: To determine if the interval between the last negative and the first positive HIV test is associated with demographic characteristics of HIV seroconverters.

Methods: A prospective cohort of patients with HIV seroconversion enrolled in the Lyons HIV hospital database was analysed. Comparisons of demographic characteristics were performed after stratification on the duration of the interval between the last HIV negative screening test and the first HIV positive screening test, which ranged from 1 day to 24 months. Linear regression methods were used to identify the covariates associated with a negative HIV antibody test followed by a positive test.

Results: Age (p=0.54), sex (p=0.78), heterosexual route of infection (p=0.78), other route (p=0.40) compared with homosexual route, and estimated year of HIV infection (p value ranged from 0.84 to 0.95) were not associated with a shorter seroconversion interval after multivariate analyses. The presence of an acute HIV illness was the only predictor of a short seroconversion interval (p=0.006) with a reduction of 84 days of the interval when it was reported.

Conclusions: No selection bias for demographic characteristics of HIV seroconverters seems associated with the length of the seroconversion interval, at least for intervals ≤24 months.

  • selection bias
  • HIV seroconversion
  • study design
  • incidence cohort

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • * The list of collaborators is given at the end of the paper.