Background: The incidence of extra-pulmonary tuberculosis (EPTB) in the Netherlands shows a seasonal trend, with a peak in spring and a trough in autumn. Possible causes of this peak are winter crowding and a seasonal decrease in immune competence in spring. A third explanation may be a reporting bias.
Objective: To investigate the role of winter crowding by a time-series analysis of notification data. DNA fingerprinting clustering status can differentiate between recent and remote infections. Seasonality in clustered cases would reflect enhanced transmission in winter and/or seasonally lowered immunity, while seasonality in unique cases would only reflect seasonally lowered immunity.
Methods: We fitted (seasonal) auto-regressive moving average models to culture-positive TB notifications in the Netherlands (1993-2008) to assess seasonality. We then used seasonal trend Loess decompositions to derive the seasonal pattern, and compared the heights of the seasonal peaks.
Results: Clustered and unique EPTB notifications showed a seasonal trend that was absent in clustered and unique PTB notifications. The seasonal peak in clustered EPTB cases was not significantly higher than in unique EPTB cases.
Conclusions: The similar timing and height of the seasonal peak of clustered and unique EPTB cases suggests that winter crowding is unlikely to cause the seasonal trend in notifications.