Using estimated MSM populations as denominator we observe much narrower ranges of HIV incidence rates (and incidence of other STI) between cities and federal states than suggested by incidence rates based on general population denominators. This indicates the advantage of this approach, because comparability of epidemiological data between different regions is improved.
An explanation for the moderate differences in incidence rate estimates for metropolitan and non-metropolitan MSM may be the high partner seeking mobility in MSM populations. MSM from rural areas and smaller cities often seek sexual partners outside of their place of residence, often in the next gay centre. Thus, the likelihood to meet sexual partners infected with HIV and/or syphilis may not be too different – at least in a country like Germany with a very well developed traffic and public transport infrastructure, which allows MSM in many non-metropolitan and rural areas to reach the next gay centre within less than two hours. In recent years, the use of the internet for finding sex partners has become a highly plausible additional factor for reducing differences between metropolitan and rural areas.
If we compare the estimated incidence of newly diagnosed HIV infections among German MSM (mean 0.3%, peak values in metropolitan areas around 0.5–0.7%) with HIV incidence among MSM in other types of studies in the Netherlands, we find rates in the same order of magnitude, especially if a participation bias towards sexually more active MSM outside of regular relationships in those studies is considered. In prospective cohort studies in MSM in Amsterdam and Rotterdam incidence rates between 1.1 and 1.5/100 person-years were observed between 1999 and 2005 .
In most, but not all cities and regions, incidence of newly diagnosed Syphilis in 2007 was higher than incidence of newly diagnosed HIV infection. Outbreak-like incidence peaks even higher than in metropolitan areas can occasionally be observed in peripheral regions like Trier (border region in the Western part of Germany; see Figure 2/3, Additional file 1).
We also analyzed whether it is feasible to estimate regional prevalence of HIV infections among MSM. Direct regional HIV prevalence estimates for MSM populations in all 95 postal code regions of Germany based on participation rates of self-reported HIV positive MSM in the Internet surveys are however not reliable due to the relatively small sample sizes of approximately 403 HIV positive participants in the KABaSTI and 545 in the GMA-2007 survey, which are distributed across 95 postal code areas. To minimize inaccuracies and biases due to low numbers per area, we evaluated to which extent the relative proportion of HIV positive MSM residing in the gay centres and in the 16 federal states may be reflected by the regional distribution of KABaSTI and GMA-2007 survey participants.
When the regional prevalent HIV cases in MSM are estimated according to the cumulative incidence of HIV reports in MSM adjusted for the estimated number of deaths between 1980 and 2008 (RKI prevalence model), the estimated HIV prevalence compared to the prevalence estimate based on the distribution of HIV-positive survey participants is higher in all western German federal states, and considerably lower in Berlin, Saxony, and the other federal states in Eastern Germany.
The main reason for this difference is the change of the epidemiological dynamics during the German reunification in 1990. While the cumulative distribution of HIV from the RKI model also reflects the regional distribution of HIV infections among MSM during the first wave of HIV infections in the 1980s, prevalence estimates based on survey participant distribution rather mirrors a distribution of currently sexually active MSM, and thus neglects infections which occurred many years ago. These infected persons may still be alive, but meanwhile sexually less active or not using the internet to find partners. This is especially relevant for the discrepancies observed between Western and Eastern Germany. The German Democratic Republic (Eastern Germany) had not experienced the first wave of HIV infections in the 1980s. Thus, after the German reunification in 1990, MSM in the eastern part of Germany and East-Berlin had a much lower HIV prevalence than MSM in Western Germany and West-Berlin in the early 1990s.
On the other hand, the surveillance data based method probably underestimates the prevalent cases in eastern German MSM, because for Eastern Germany the estimates are predominantly based on the number of already diagnosed infections, while in Western Germany a proportion of as yet undiagnosed infections is included in the model by using the back-calculation method for the early period of 1980 until 1990 (back-calculation based on AIDS cases accounts also for undiagnosed HIV cases). Thus, the real prevalence in MSM in Eastern Germany may lie somewhere in between the two estimates.
Other factors that may explain some of the differences between the surveillance data and survey based estimates:
Selective migration of HIV-positive MSM after HIV diagnosis from rural areas to larger cities and between cities (net gains to be expected especially for Berlin, Frankfurt and Leipzig, net losses for rural areas in Baden-Wuerttemberg, Bavaria, Lower Saxony, Rhineland-Palatina, Schleswig-Holstein, Mecklenburg-Vorpommern, Saxony-Anhalt, Thuringia and the cities Stuttgart and Hanover).
Underrepresentation of (HIV positive) MSM among the survey participants from the respective state/city (especially for Bavaria, where a difference between MSM population estimates based on proportion of MSM website user profiles and the proportion of survey participants has been described )
Overestimation of the proportion of HIV positive men living in a city by geographical attribution based on the postal code of the health care provider in the surveillance data based model (may be relevant especially for cities with large catchment areas in densely populated areas, such as Hamburg, Munich, Frankfurt, Duesseldorf and Stuttgart). In our experience the distinction between health care provider and patient postal code is not always reliable, thus a larger proportion of allocations than currently acknowledged may be based on the health care provider postal code.
The discrepancy between the two estimates disappears for the eastern German states except Berlin and becomes smaller for the western German states if we make a tentative adjustment in the survey based model for the federal states in Eastern Germany and reduce the prevalence estimate by 50%. Due to a later starting point of the HIV epidemic in MSM who live in the region of the former German Democratic Republic an adjustment of the estimate is justified. A 50% reduction could be justified by the fact that at the time of German reunification approximately 50% of the total cumulative HIV cases in Germany had already occurred in the former Western part of the country, and the number of prevalent cases would have a considerable impact on the number of new infections occurring in the period after the reunification. For the united federal state of Berlin which is geographically located in the eastern part of the country but is composed of the former West-Berlin (2.1 Mio. inhabitants) and the former East-Berlin (1.3 Mio. inhabitants), such kind of adjustments are more difficult. While MSM in West-Berlin had a similar or even higher HIV prevalence as other large cities in Western Germany, HIV prevalence in MSM in East-Berlin was dramatically lower before reunification. How much the estimate for Berlin based on the observed prevalence among survey participants could be reduced to account for the "reunification effect" is unclear. However, it does not seem realistic to explain the large difference between survey and surveillance based estimates for Berlin by such a "reunification effect".
For the relative proportions of the federal states of Bavaria and North Rhine-Westphalia some adjustments might be reasonable as well because we observed a slightly skewed representation of survey participants from these two states compared with the MSM website profile data from the largest German MSM website (GayRomeo): from the KABaSTI participants who were recruited on GayRomeo 14.6% reported residence in Bavaria, 19.3% residence in North Rhine-Westphalia, compared with 10.6% and 25% of all survey participants. But again, even if we adjust the data according to these proportions, the surveillance based estimate for Bavaria will remain higher than the survey based estimate.
In Germany, a major challenge for regional prevalence estimates arises from temporal changes of the spread of HIV in MSM populations, mainly from the different epidemiological dynamics in the Western and Eastern part of the country before reunification. Another factor which is difficult to assess is selective migration of HIV positive men to metropolitan areas after HIV diagnosis. As life expectancy and quality of life of people living with HIV have improved during the last decade, such selective migration may have played an increasing role in recent years. Because of the existence of larger sexual networks of HIV-positive MSM in metropolitan areas and because of better access to quality medical HIV care, MSM diagnosed with HIV infection in non-metropolitan areas may see even larger benefits from moving to metropolitan areas than their non-infected peers. The pronounced differences between the surveillance and survey based HIV prevalence estimates for Berlin and Bavaria may be an indication for such selective migration processes. Questions about post-HIV diagnosis migration of HIV positive MSM in clinical surveillance studies could be used to verify this hypothesis.