Interhospital Transfers in ICU- and Non-ICU-Treated Sepsis Patients in Germany Between 2016 and 2019

Interhospital transfers (IHTs) are an important component of sepsis care, especially when hospitals lack resources or specialized expertise. For Germany, data on IHT practices and patient-level predictors for transfers to higher-volume hospitals in ICU vs. non-ICU-patients are missing.

To assess the frequency and timing of IHT among sepsis patients in Germany and identify factors associated with IHT in higher-volume hospitals.

Retrospective cohort study.

We used de-identified health claims data from Germany's largest statutory health insurer (AOK-Die Gesundheitskasse, approximately 32% population coverage). Patients 16 years or older hospitalized with sepsis (2016-2019) were identified by International Classification of Diseases, 10th Revision, German Modification and procedural codes.

We measured the frequency of IHT. IHT were defined as admissions to a different hospital within 1 day of discharge from the index hospitalization. We used competing risk models to estimate time-to-transfer and generalized additive models stratified by ICU treatment to identify predictors of transfers to higher-volume hospitals. We hypothesize that the state of health before sepsis, characteristics of sepsis and its treatment during the index stay, are associated with the probability of an IHT.

Of 347,826 patients with sepsis, 14.4% experienced greater than or equal to 1 IHT. Transition probabilities were highest for IHT from lower to higher-volume hospitals (e.g., 47.0% from the lowest to highest case volume stratum). Transitions were more likely within the same regional category than between different types. ICU-treated patients had a delayed but prolonged probability for IHT (peak: days 30-60); non-ICU patients showed earlier peaks (days 5-30). In multivariate models, increasing age decreased the IHT probability to a higher-volume hospital. Nosocomial infection was also associated with a significantly lower probability of such transfers (non-ICU: -5.6%; ICU: -2.9% probability for IHT to higher-case-volume hospitals). ICU patients in rural/suburban hospitals had reduced transfer likelihood compared with their non-ICU counterparts, whereas sex was not significantly associated.

Around one in seven sepsis patients in Germany had at least one IHT, often from lower-volume or rural hospitals to higher-volume centers. Transfers occurred earlier in non-ICU patients and later and more prolonged in ICU patients, and increasing age and nosocomial infection were associated with a lower probability of transfer to higher-volume hospitals. Our findings highlight that both the timing and determinants of IHTs differ across patient subgroups, underlining the need for further research on the factors driving transfer decisions and their impact on patient outcomes.