Basis for public health surveillance of Guillain-Barre syndrome in Sweden

We explored the basis for public health surveillance (PHS) of Guillain-Barre syndrome (GBS) in Sweden. Time-series analyses were performed by using autoregressive integrated moving average (ARIMA) models on CBS-related hospital-discharge data, namely monthly incidence rates in Sweden during the period 1978-1992 and bimonthly numbers of new cases in Stockholm County during the period 1973-1992. In general, the predicted values for 1993 fitted well with the observed figures. However, in line with prior analyses, the observed monthly incidence rates for April and May 1993 among the under 40 years age group in Sweden were below the lower limit of the 95% confidence intervals for the corresponding predicted values. Furthermore, a seasonal pattern was detected. When the 1993 forecasts were used for retrospective examination of the data, significant variations in CBS incidence rates were found for certain time periods. Some such signals corresponded to known high-risk periods in 1978 and 1983. The results suggest that monthly or bimonthly forecasts of GBS incidence can be obtained from ARIMA models applied to registered CBS diagnostic data. PHS of GBS could have detected the zimeldine-induced CBS epidemic in 1983 as well as the increases in CBS incidence with unknown cause in 1978 and 1983.