Volume 14, Number 6—June 2008
Dispatch
Underreporting of Human Alveolar Echinococcosis, Germany
Table 2
Log linear estimates of the total number of alveolar echinococcosis cases, Germany, 2003–2005*
Models† | df | AIC | x | N | 95% CI for N |
---|---|---|---|---|---|
Saturated: Interaction (RKI, EER) and (RKI, PAS) and (EER, PAS) | 0 | 0.00 | 174 | 288 | 129–2020 |
Interaction (RKI, EER) and (RKI, PAS) | 1 | –1.49 | 83 | 197 | 143–358 |
Interaction (RKI, PAS) and (EER, PAS) | 1 | –1.33 | 70 | 184 | 148–253 |
Interaction (RKI, EER) and (EER, PAS) | 1 | –1.66 | 285 | 399 | 189–1961 |
Interaction (RKI, PAS) | 2 | –3.33 | 70 | 184 | 150–242 |
Interaction (RKI, EER) | 2 | –1.57 | 134 | 248 | 171–430 |
Interaction (EER, PAS) | 2 | –0.08 | 93 | 207 | 163–287 |
Independent (no interactions) | 3 | –1.94 | 89 | 203 | 163–268 |
*RKI, Robert Koch Institute; EER, European Echinococcosis Registry; PAS, pathologists’ survey; df, degrees of freedom; AIC, Akaike Information Criterion (measures how well the model fits the data [small values indicate a better fit]); x, estimate of unascertained cases; N, estimate of total number of cases (total number of observed cases + x); CI, goodness-of-fit–based confidence interval.
†Each model includes all first-order terms. The first model (saturated) adjusts for dependencies between all 3 source pairs; the second model adjusts for possible dependencies between RKI and EER, and between RKI and PAS, etc.