Correspondence (reply): In Reply

In our article (1) we reported two different response rates according to international standards (2), which referred to different baseline totals (response rate 3 according to AAPOE: all likely households; cooperation rate: all target persons who had been contacted successfully). For this reason, the suggested multiplication of the two rates is not admissible. As shown in the cited references, the achieved response rates meet the expectations for population surveys (3). However, in order to be able to generalize the results to the population, the design and adjustment weightings of the analyses are additionally crucial. Still, we did mention in the article the limitations that apply to each and every telephone health survey.

Collecting data on self-reported diseases diagnosed by a physician is the common method used in health surveys to estimate disease prevalence rates at the population level. We did not validate these self-reports—this would have been impossible in a nationwide telephone survey of this order of magnitude.

In Figure 2 we described which federal state had above-average or below-average values for both indicators. The confidence interval for Germany is 8.2% to 8.7%; this is significantly different to Baden-Württemberg (6.7% to 7.8%). It also means marginally overlapping confidence intervals for Rhineland-Palatinate (8.7% to 10.8%) and Saxony-Anhalt (8.5% to 11.2%).

Other authors have also reached the conclusion that differences in cardiovascular mortality between federal states cannot plausibly be explained with method-related biases of the cause of death statistic (4). Our summary of the most important cardiovascular disorders may have balanced the effects of coding difference at state level further. When all limitations are taken into account, using the official cause of death statistic is the only option for studying differences in regional death rates.