Spatial patterns of tuberculosis incidence in Cologne (Germany).

In many European countries, the decline of tuberculosis notification rates levelled off in the mid 1980s. Germany has been facing only a very modest resurgence of tuberculosis (TB) in the early 1990s, but until now, the notification rate does not yet fulfil the WHO definition of a low incidence country. Mainly immigration from high incidence countries is held responsible for the delayed decline and temporary increase of TB. This paper seeks to investigate the inner-urban strength of association between tuberculosis and several potential risk factors in the city of Cologne using geographical and statistical methods within a small-area division based, GIS-supported ecological study, which never before has been conducted for a German city. Standardised annual TB incidence rates for 78 urban subdistricts between 1986 and 1997, disaggregated by age, have been analysed. Twelve independent variables representing contemporary ethnic and socioeconomic conditions are used to calculate Poisson regression models. For the entire population and for the <60 age group, a strong positive association was found between tuberculosis incidence and share of immigrants. Interestingly enough the share of German-origin immigrants from eastern Europe was seen not to be instrumental in increasing TB rates. Reflecting strong ecological correlations between variables depicting economic conditions and TB as well as immigration variables, the deprivation of certain ethnicities rather than high prevalence immigration background is inferred to be significant to TB level. As for the 60 + age group regression analysis fails to model the TB incidence patterns sufficiently it seems reasonable to assume an autonomous, spatially disaggregated TB epidemic of the elderly, echoing a severe post-war epidemic by reactivation, only marginally associated with contemporary living conditions, but contributing substantially to the current TB incidence level. This assumption is reinforced by further results of the statistical analysis (deviance of regression null models, testing for Normal distribution, time trends). It could be shown that in Cologne the HIV/AIDS epidemic is not significantly influencing the TB incidence rate of men aged 30-49 being the most AIDS-afflicted population subgroup.