Objective: To investigate whether space-time clustering at diagnosis of type 1 diabetes mellitus (DM1) is related to clinical manifestations and to see whether there is a time-space clustering at birth of children who later develop DM1.
Patients: Using the method by Knox, clustering was analysed in all 1,144 children diagnosed between 1977 and 1994 in south-east Sweden.
Results: The strongest significance was obtained for the cut-off value of 5 km and 7 months (p<0.01). Using this cut-off, children with a short duration of symptoms before diagnosis had the same degree of clustering as children with a longer duration. Children diagnosed during autumn and winter tended to have a higher degree of clustering than children diagnosed during spring and summer. We found no significant clustering regarding birthplace and birth month.
Conclusion: This study is consistent with the existence of space-time clustering at diagnosis. The most plausible explanation is that infections elicit several cases of DM1 in children in whom the disease process has already begun.