On average (year 2014–2016) a total number of 172 aTTP episodes per year was projected (95% confidence interval [95%CI]: 132–212) in the German overall population (children and adults). The majority were newly diagnosed aTTP cases (n = 121; 95%CI: 105–129), and 51 were recurrent aTTP cases (95%CI: 27–84). The related average annual aTTP incidence was 1.47 per million inhabitants (95%CI: 1.28–1.57) and the related average annual incidence of aTTP episodes was 2.10 per million inhabitants (95%CI: 1.60–2.58). Looking at the German adult population the average annual aTTP incidence was 1.70 per million adults (95%CI: 1.48–1.81) and the related average annual incidence of aTTP episodes was 2.42 per million adults (95%CI: 1.86–3.00). As determined by our systematic literature search these findings represent the first national incidence estimates for aTTP in Germany.
The described projection is bound to the national estimates of TMA (ICD-M31.1) that are based on the national hospitalization statistics and on the G-DRG database [24,25,26,27,28,29], hence the available national data defined the maximum limit of the projection. Based on the hospital-level data we determined the proportion of confirmed aTTP diagnoses in primary and secondary TMA cases. To account for statistical uncertainty of the study sample, the data from the eight participating hospitals were utilized using logistic regression. This statistical uncertainty was expressed by the lower and upper 95%CI around the presented incidence estimates.
As TMA and HUS are both assigned mainly to the same DRG in Germany (L72Z named Thrombotic Microangiopathy and Hemolytic Uremic Syndrome) , we also looked for aTTP patients coded as HUS (ICD-10 D59.3). By these means we identified two aTTP cases that were primarily miscoded as HUS (please refer to Table 2). Therefore, for the projection we considered these cases as TMA. No further aTTP cases were coded as HUS, so for the projection the national data for TMA (ICD-10 M31.1) were used as basis.
In our study, patients were considered to have a confirmed diagnosis of aTTP if ADAMTS13 activity were < 10% and/or the medical records explicitly mentioned the diagnosis of aTTP. A potential alternative definition would have been ADAMTS13 activity < 10% (as only criterion) or ADAMTS13 activity < 10% combined with a positive autoantibody (AAB) test, with the disadvantage that non-tested readmissions and borderline results (rated as aTTP) would have been excluded, both of which might have been resulted in an underestimation of the aTTP incidence.
Hence the applied definition of a confirmed diagnosis of aTTP was rated as the most reliable definition as it may deliver the most realistic estimate and no systematic underestimation on the number of aTTP cases. In 17% of cases (n = 15 of n = 87 cases) there was no current ADAMTS13 activity measurement available. However, all those cases without ADAMTS13 activity measurement were classified as recurrent aTTP by the treating physician based on the clinical symptoms and by confirmed prior aTTP episodes. In further 11% patients (n = 10 of n = 87 cases) the ADAMTS13 activity measurements presented borderline values (slightly above 10%) but these were rated as aTTP by the treating physicians (e.g. as the patient had a previous aTTP episode or as ADAMTS13 activity was measured after first plasma exchange therapy). In this context it is also of interest, that only for 59% of all study cases (351 of 600 cases) ADAMTS13 measurements were available, which explains why the ADAMTS13 activity alone was not regarded as reliable definition for the diagnosis of aTTP in the presented study.
The described approach might also include congenital TTP cases as anti-ADAMTS13 antibodies measurement was only available in 69% of patients (n = 60 of n = 87 cases); hence a differentiation between congenital TTP and acquired TTP, on the basis of the absence of anti-ADAMTS13 antibodies, was not possible in all patients. However, due to the rare event of congenital TTP forms and thorough clinical evaluation the used approach most likely has only a mild impact for overestimation of the aTTP incidence.
For better comparison of different studies detailed analysis of the used methodologies is essential. In this context it is most important to assure that only aTTP was included, whether initial and/or recurrent aTTP episodes were considered, what kind of populations (e.g. adults/children/total) were analyzed and which approach was used for the definition / diagnosis of aTTP (e.g. ADAMTS 13 activity < 10%).
In other countries, the TTP incidence was estimated at 1.5 (France) , 3.1 (USA) , and 6.0 (UK) [16, 17] cases per million. These varying incidence rates are a result of varying definitions and of population-based differences.
The data from the French registry were calculated on the basis of a large cohort, which was enrolled in the registry over 15 years. They only included patients with a first TMA episode and an ADAMTS13 activity < 10% which was measured in one reference center with a highly standardized method. Therefore the quality of the results can be considered as very reliable. Looking at the applied methods the comparison of our approach to the French cohort seems to be reasonable with the difference that we also report the incidence of recurrent aTTP episodes, while the French group reports only the incidence of (initial) aTTP. Comparing the results we show a good accordance, as in our projection for Germany 1.47 per million inhabitants (1.5 per million in France) are estimated to have an initial aTTP manifestation. In adults the related incidence of aTTP is 1.70 per million.
In the UK TTP registry the incidence of TTP episodes was calculated on the basis of a clinical diagnosis according to national guidelines , excluding other conditions such as HUS and HELLP, but patients with secondary TTP (e.g. due to HIV infection or drug-induced TTP) were included [16, 17]. As this approach did not rely on ADAMTS13 measurement, the real incidence of aTTP might have been overestimated in the UK study.
In the United States the Oklahoma TTP-HUS registry provides estimates for the incidence of initial and recurrent aTTP in a mixed population (adult and children), allowing a good comparison to our assessment. Reese et al. reported a standardized aTTP incidence rate (newly diagnosed cases) of 2.17 (95%CI: 2.00–2.34) per million observed in the Oklahoma TTP-HUS registry , compared to 1.47 (95%CI: 1.28–1.57)) per million population in our study. Page et al. Identified a combined incidence of initial and recurrent aTTP episodes of 3.1 per million  as compared to 2.10 per million inhabitants (95%CI: 1.60–2.58) identified in the present study. The slightly lower incidence found in our study in our study could be explained with the predominant Caucasian origin of the German population in contrast to the Oklahoma registry. As the aTTP incidence was found to be higher in the black population compared to non-blacks (Incidence rate ratio 7.09),  this might explain the higher incidence observed in the Oklahoma registry.
Besides the aspects that were already discussed above, there are further limitations of our study. Our incidence estimate is reflecting a mainly Caucasian population and only eight of thirty invited hospitals provided retrospective data and were hence included into this study. Centers from the North, West and South of Germany were equally represented, with a lack of centers from the eastern parts of Germany. As these eight centers are major TMA centers the number of cases that were retrospectively recorded (n = 600) reflects approximately 10% of all German HUS / TMA cases (n = 5906) [24,25,26,27,28,29] in this period. With regard to the patient numbers (based on the primary TMA diagnosis) the participating centers reflect about 25% of patients in all top 30 TMA hospitals. However, the logistic regression analysis of the hospital-level data helps to describe potential inaccuracies with 95% confidence intervals.
A further limitation is that we have only selected major TMA centers as basis for our assessment. It is difficult to predict, but possible, that the inclusion of smaller centers (outside the top 30) might have altered the proportion of aTTP cases in relation to all TMA hospitalization cases. However, as our findings compare well to the findings of the French  and the Oklahoma  aTTP registry, the impact of this potential selection bias is rated minor.