Delays between chikungunya case diagnosis and entry into the disease surveillance system.

There is often a long delay of weeks or months between initial diagnosis of a chikungunya case and entry of that case into the monitoring system. Varied delays make nowcasting on a weekly basis more difficult, as we lack complete data on both the most recent week and the weeks shortly preceding it. The first row describes the distribution of delays across the sample as a whole. Only 26% of cases are reported after 1 week, and 74% after 4 weeks. We further find evidence of a tail of cases with very long delays, with 13% of cases still not entered 8 weeks after notification. The second row describes the distribution of delays during epidemic periods, which are weeks in which the case count exceeds 104 cases. This is very similar to the sample as a whole. Similarly, we find that the distribution of delays in non-epidemic periods (third row) and the distributions of delays by year (fourth to seventh rows) do not differ greatly to the distribution of delays for the sample as a whole. This suggests that any differences in the performance of the model in these different time periods are unlikely to be due to differences in the structure of the delays.