[Comparison of sentinel surveillance of influenza like infections in Martinique and in Guadeloupe during influenza A(H1N) 2009 pandemic: use of a phone survey and hospital surveillance]

Abstract

Background: Sentinel general practitioner networks monitor influenza-like infections (ILI) in Martinique and in Guadeloupe (French West Indies). During the A(H1N1)2009 pandemic, they gave an ILI incidence estimation higher in Martinique than in Guadeloupe. In October 2009, a telephonic survey was launched in both islands to assess the number of ILI diagnoses performed by general practitioners since the beginning of the pandemic. This paper compares the results of sentinel surveillance, of telephonic survey and of hospital surveillance in Guadeloupe and in Martinique.

Methods: On each island, the sentinel network gathers a representative sample of voluntary general practitioners. Each week, they report the number of ILI they diagnosed the past week. Times series of these weekly numbers were modelized using the Serfling method with the upper limit of the confidence interval of the expected value representing the epidemic threshold. The telephone survey was conducted from October 2, 2009 to October 12, 2009 in Martinique and from October 13, 2009 to October 21, 2009 in Guadeloupe. The quota method was used for sampling individuals older than 14 years, leading to 507 interviews in Guadeloupe and 508 in Martinique.

Results: The epidemic lasted 12 weeks in both islands, from August 3 to October 25 in Martinique and from August 17 to November 8 in Guadeloupe. During August and September, estimated attack rate in Martinique was 5.52% (CI95: 5.23-5.83) from the sentinel network versus 8.3% (CI95: 6.0-11.0) from the telephone survey. In Guadeloupe, it was 2.13% (CI95: 1.97-2.24) from the sentinel network versus 6.9% (CI95: 4.8-9.5) from the telephone survey. An equivalent number of confirmed hospitalized cases was observed in the two islands.

Conclusion: These results suggest that the sentinel network underestimates ILI incidence in Guadeloupe. According to Emergency Room activity for ILI, it seems possible that ILI incidence was actually higher in Martinique. A lower proportion of swab sampling among ILI hospitalized people could partly explain the observed differences in hospitalization ratio, in severity and in lethality between the two islands.