Correspondence (reply): In Reply

We would like to thank all authors of the correspondences for their interest in, and feedback on our article (1).

In response to Prof. Abholz: As shown in Figure 2, 63.3% of the pregnant women exclusively received the pre-test. For 12.7%, the diagnostic test was also performed. Therefore, 76% of the pregnant women were tested as defined in the Federal Joint Committee (G-BA) Maternity Guidelines (2); in our opinion, this does not reflect random implementation. Screening is an optional offer, according to the maternity guidelines, which is accompanied by a clear presentation of its advantages and disadvantages (see Annex 6 of the maternity guidelines [2]). It cannot be argued that pregnant women who forego the offered screening have received “substandard care”. Only 4.8% of the cohort received a diagnostic test without a documented pretest; the reasons for this approach cannot be derived from the billing data.

The reason why screening was introduced by the Federal Joint Committee (G-BA) was not the topic of the present analysis; as mentioned in our introduction, the decision was based on the expert report from the Institute for Quality and Efficiency in Health Care (IQWiG) (3).

In response to Prof. Stang: We had numerous discussions about how to present the relative frequencies in a meaningful and understandable manner. The representation of a circle diagram with an outer ring (applied test method) and an inner ring (the resulting diagnosis of GDM or not) seemed to us to be logical. The reference size for both rings is the entire study cohort (N = 567 191). If one sets the number of pregnant women with GDM and „only pre-test“ (N = 25 019) in relation to the total study cohort, 4.4% is obtained. If the inner ring were to represent only the pregnant women with GDM, the proportion would be 33% as described by the author. We deliberately chose to show the two subpopulations with respect to the same population (that is, the entire study cohort), since this seemed to us to be the most comprehensible representation (1). We will, however, take into account the valuable suggestion for future work and think critically about representations using circle diagrams.

In response to Dr. Reeske and Prof. Spallek: We share the opinion of the authors. Further studies on how to treat GDM, the treatment outcome, and the predictive factors for development of GDM are necessary, especially in light of indications of an increased prevalence, as can be derived from our analyses.

In response to Dr. Kleinwechter et al.: As described in the text and in Figure 1, women with pre-existing diabetes were not included in the prevalence calculation but rather were excluded from the population as a whole (N = 5363). Manifest diabetes was coded for the first time during the studied pregnancy for 1% of women (N = 5956). If these cases are subtracted from the total prevalence, the GDM prevalence is 12.2%, as indicated in the text. We chose not to provide detailed explanations about why GDM diagnoses were made without testing (1.8%) or were made based only on the pre-test (4.4%). These may be women who were diagnosed based on tests other than those prescribed by the maternity guidelines or who were tested but not billed within the nationwide panel doctor billing set, for instance, if testing occurred outside the recommended test period (4). However, based on the data available to us, we can only speculate about these reasons.

Using insulin treatments as an additional control, as requested by the authors, is not possible based on the available data set; to our knowledge, however, such a validation would also likely be distorted due to the known heterogeneous procedures of insulin treatment (4).

Since the selected population equals approximately 80% of the current births over one year, we assume that the cohort is representative. The limitations of the billing data analyses that should be considered are given in detail in the text. The resulting GDM prevalence corresponds to the current estimates for Europe.