Risk of thromboembolic events in patients with polycythaemia vera: a study based on german health claims data Authors: Karina C Manz, Bashar Morouj, Marc Feuerbach, Rainer Ehneß, Pamela Lenti, Isabel Barnstorf, Ariane Höer, Haifa Kathrin Al-Ali Published: 2026 ABSTRACT AbstractThis retrospective cohort study evaluated thrombotic outcomes and treatment patterns in patients with polycythemia vera (PV) in German health claims data. Prevalent and incident patient populations were identified, and longitudinal cohorts were formed to assess thromboembolic event (TE) rates, phlebotomy frequency, and the development of hydroxyurea (HU) intolerance or resistance. Patients were stratified into four thrombotic risk categories based on prior TE, age ≥ 60 years, or both, with low risk defined as the absence of both risk factors. Between 2021 and 2023, approximately one-third of prevalent patients underwent phlebotomy annually, whereas incident patients required more frequent procedures. TE risk differed substantially by risk category: patients with event- and age-based high-risk had a 6.2-fold higher rate than low-risk patients (82.4 vs. 13.4 per 1,000 person-years). Both prior TE history and advanced age contribute to TE-risk. While age-based high-risk alone showed a slightly stronger effect than event-based high-risk (11.3% vs. 9.0%), patients with both risk factors had the highest TE incidence (21.1%), highlighting the clinical relevance of thromboembolic history beyond age alone. Among incident patients, TE-free survival declined from 91.9% at one year to 89.3% at three years post-diagnosis. During follow-up, 22% of HU-treated patients met proxy measures for intolerance or resistance to HU after a median of 337 days. These patients experienced higher TE rates compared to patients treated with ruxolitinib (72.9 vs. 46.7 per 1,000 person-years), although differences were not significant. This real-world analysis suggests a persistent TE risk in PV, particularly among TE-based high-risk patients, and supports the importance of risk-adapted monitoring and timely therapy optimization.Supplementary InformationThe online version contains supplementary material available at 10.1007/s00277-026-07119-5.Keywords: Polycythemia vera, Thromboembolic events, Claims data analysis, Germany Introduction Polycythemia vera (PV) is a Philadelphia chromosome-negative myeloproliferative neoplasm primarily driven by mutations in the JAK2 gene [1]. It is characterized by excessive erythrocyte production, often accompanied by leukocytosis and thrombocytosis. Management of PV focuses on reducing thromboembolic and cardiovascular risk by maintaining hematocrit below 45%, thereby reducing the incidence of major cardiovascular events and improving long-term outcomes [2]. Equally important is rapid and durable symptom control, as PV-related symptoms such as fatigue, pruritus, and splenomegaly significantly impair quality of life. Treatment strategies seek to delay disease progression, including transformation to myelofibrosis or acute myeloid leukemia, and to modify the disease course to reduce overall burden [3]. The most frequent complications are arterial and venous thromboses which represent the leading cause of morbidity and mortality in patients with PV [2, 4, 5]. Consequently, therapeutic strategies aim to reduce thromboembolic risk, with treatment decisions guided by individual risk profiles. Current guidelines classify patients into two risk categories: high-risk (age ≥ 60 years and/or prior TE event) and low-risk (all others) [4, 6]. Initial therapy typically combines phlebotomy with low-dose acetylsalicylic acid [4]. High-risk patients are recommended to receive cytoreductive therapy, most commonly hydroxyurea (HU) or ropeginterferon-alpha (ropeg INF-α) [4]. However, a substantial proportion of patients develop intolerance or resistance to HU, manifesting as inadequate disease control, severe cytopenias, or other adverse effects necessitating treatment discontinuation [4, 7, 8]. Similarly, tolerability issues with INF-α may also necessitate treatment changes [4]. For these patients, second-line options include ruxolitinib (RUX) or ropeg INF-α [4]. Despite these treatment algorithms, studies have shown that TE risk remains high [5, 9, 10]. A recent U.S. claims analysis of over 20,000 patients with PV reported that 25.1% of patients experienced at least one thromboembolic event (TE) during follow-up, with rates substantially higher than matched controls and particularly elevated in patients with prior TE [10]. TE-free survival probability declined progressively across all risk categories, underscoring persistent risk throughout the disease course [10]. In a previous claims-based study, we estimated PV prevalence at 28.6 per 100,000 and incidence at 3.3 per 100,000 adults (≥ 18 years) in 2021 [11]. Building on this work, the present study examines real-world thrombotic risk and treatment patterns in German PV patients. Specifically, major TE rates were analyzed across four risk categories defined by thrombotic history and age: low-risk (no prior TE and/or age < 60 years), event-based high-risk (prior TE), age-based high-risk (age ≥ 60) and event- and age-based high-risk (prior TE and age ≥ 60). Furthermore, TE-free survival from diagnosis, phlebotomy utilization, HU intolerance/resistance development, and thrombotic outcomes across therapies were evaluated. Understanding these treatment patterns is critical for optimizing management strategies to reduce complications and improve long-term outcomes. Methods Study design and data source This study was designed as a descriptive claims-based analysis to characterize epidemiology, treatment patterns, and associated outcomes in routine clinical practice. Anonymized health claims data from the sample database of the Institute for Applied Health Research Berlin (InGef) were used for this retrospective, non-interventional cohort study. The database comprises anonymized claims data from approximately four million individuals insured by the German statutory health insurance (SHI), representing about 5% of the German population [12]. It is considered representative of the German population by age and sex and was found to closely match national data in terms of morbidity, mortality, and drug utilization [12]. This database provides individual-level, longitudinal data across multiple healthcare sectors, including information on hospitalizations (admission and discharge dates, diagnostic and therapeutic procedures, main and secondary diagnoses), outpatient services (diagnoses, diagnostic and therapeutic procedures), and outpatient drug prescriptions (dates of physician prescription and pharmacy dispensation). Diagnoses are coded using the International Classification of Diseases, 10th revision in the German Modification (ICD-10-GM) [13]. For outpatient diagnoses, only the quarter of diagnosis is recorded rather than the exact date. Outpatient drug prescriptions are coded using Anatomical Therapeutic Chemical (ATC) codes [14], outpatient diagnostic and therapeutic procedures are coded according to the physician’s fee scale (EBM) [15], and inpatient procedures (including some inpatient drug administrations) are coded using German procedure classification (OPS) codes [16]. Data from January 1, 2016 to December 31, 2023 were analyzed with results reported for the period from January 1, 2018 to December 31, 2023. Study populations and cohorts First, for each reference year (2018–2023) adults aged ≥ 18 years with continuous insurance coverage during the two preceding calendar years and throughout the reference year (or until death) were identified. From this base population, multiple study populations and longitudinal cohorts were subsequently derived: Prevalent patient populations: Inclusion criteria required at least one PV diagnosis (ICD-10-GM code: D45; as a confirmed outpatient or main or secondary inpatient diagnosis) during the respective reference year (2018–2023). Exclusion criteria eliminated individuals with diagnoses of osteomyelofibrosis (D47.4), essential thrombocythemia (D47.3), other myeloproliferative neoplasms (C92.1, D47.0, D47.1, D47.5), or other erythrocytosis (D75.0, D75.1) in the respective reference year or the two years prior. Additional validation criteria required at least one of the following in the reference year or the two preceding years: treatment with hydroxyurea (HU) or interferon‑alpha (IFN‑α), a confirmed outpatient PV diagnosis by a hematologist, an inpatient PV diagnosis, bone marrow puncture, or phlebotomy. Prevalent patient populations: Inclusion criteria required at least one PV diagnosis (ICD-10-GM code: D45; as a confirmed outpatient or main or secondary inpatient diagnosis) during the respective reference year (2018–2023). Exclusion criteria eliminated individuals with diagnoses of osteomyelofibrosis (D47.4), essential thrombocythemia (D47.3), other myeloproliferative neoplasms (C92.1, D47.0, D47.1, D47.5), or other erythrocytosis (D75.0, D75.1) in the respective reference year or the two years prior. Additional validation criteria required at least one of the following in the reference year or the two preceding years: treatment with hydroxyurea (HU) or interferon‑alpha (IFN‑α), a confirmed outpatient PV diagnosis by a hematologist, an inpatient PV diagnosis, bone marrow puncture, or phlebotomy. Incident patient populations: Inclusion criteria required a first PV diagnosis (confirmed outpatient diagnosis by a hematologist or main or secondary inpatient diagnosis) during the respective reference year (2018–2022; the 2023 data year was required to confirm incident case identification for 2022), while exclusion criteria eliminated individuals with diagnoses of osteomyelofibrosis (D47.4), essential thrombocythemia (D47.3), other myeloproliferative neoplasms (C92.1, D47.0, D47.1, D47.5), or other erythrocytosis (D75.0, D75.1) in the two years prior to the first observed PV diagnosis. Validation for incident cases further required presence of at least one of the following: bone marrow puncture within 90 days before or after the first PV diagnosis, HU or IFN‑α treatment within the first year from the first PV diagnosis, hospitalization due to TE with PV listed as a secondary diagnosis within the first year from the first PV diagnosis, or a phlebotomy within the first year from the first PV diagnosis. Incident patient populations: Inclusion criteria required a first PV diagnosis (confirmed outpatient diagnosis by a hematologist or main or secondary inpatient diagnosis) during the respective reference year (2018–2022; the 2023 data year was required to confirm incident case identification for 2022), while exclusion criteria eliminated individuals with diagnoses of osteomyelofibrosis (D47.4), essential thrombocythemia (D47.3), other myeloproliferative neoplasms (C92.1, D47.0, D47.1, D47.5), or other erythrocytosis (D75.0, D75.1) in the two years prior to the first observed PV diagnosis. Validation for incident cases further required presence of at least one of the following: bone marrow puncture within 90 days before or after the first PV diagnosis, HU or IFN‑α treatment within the first year from the first PV diagnosis, hospitalization due to TE with PV listed as a secondary diagnosis within the first year from the first PV diagnosis, or a phlebotomy within the first year from the first PV diagnosis. Longitudinal cohorts: Cohorts based on incident (incident patient cohort) or prevalent (prevalent patient cohort) status as well as treatment-specific cohorts were formed. Treatment-specific cohorts included prevalent patients who received HU treatment (HU-cohort), patients who received HU treatment and for whom proxy measures for HU intolerance/resistance (HUIT/R-cohort) were observed, or prevalent patients who received RUX treatment (RUX-cohort). HU intolerance/resistance was defined using proxy measures (see Table S1 in the supplement) as consensus criteria indicating HU intolerance/resistance rely on clinical parameters [4, 17] that are not available in the underlying data source. The prevalent patient cohort was stratified according to risk status at index (described below). For each cohort, an index date was defined (first diagnosis, first observable proxy measure for HU intolerance/resistance, or treatment initiation, respectively), a two‑year baseline period before index, and follow‑up through December 31, 2023. Censoring rules included death, loss to follow‑up, treatment switch to a different cytoreductive drug, treatment discontinuation, and progression to myelofibrosis (ICD-10-GM: D47.1, D47.4 as hospital main or secondary diagnosis or confirmed outpatient diagnosis), applied as appropriate for each cohort. Longitudinal cohorts: Cohorts based on incident (incident patient cohort) or prevalent (prevalent patient cohort) status as well as treatment-specific cohorts were formed. Treatment-specific cohorts included prevalent patients who received HU treatment (HU-cohort), patients who received HU treatment and for whom proxy measures for HU intolerance/resistance (HUIT/R-cohort) were observed, or prevalent patients who received RUX treatment (RUX-cohort). HU intolerance/resistance was defined using proxy measures (see Table S1 in the supplement) as consensus criteria indicating HU intolerance/resistance rely on clinical parameters [4, 17] that are not available in the underlying data source. The prevalent patient cohort was stratified according to risk status at index (described below). For each cohort, an index date was defined (first diagnosis, first observable proxy measure for HU intolerance/resistance, or treatment initiation, respectively), a two‑year baseline period before index, and follow‑up through December 31, 2023. Censoring rules included death, loss to follow‑up, treatment switch to a different cytoreductive drug, treatment discontinuation, and progression to myelofibrosis (ICD-10-GM: D47.1, D47.4 as hospital main or secondary diagnosis or confirmed outpatient diagnosis), applied as appropriate for each cohort. Operationalization of relevant procedures and treatments are shown in Table S2 in the supplement. An overview of the study populations and longitudinal cohorts is shown in Fig. 1. Overview of the study populations and study cohorts. Patients were included in their respective longitudinal cohorts up to and including 31 December 2022, in order to allow for a follow-up of at least one year (or until death) Risk stratification Patients included in the prevalent patient cohort were stratified according to four categories based on risk status at index: Low-risk: <60 years old and no diagnosis for a TE during baseline. Low-risk: <60 years old and no diagnosis for a TE during baseline. Event-based high-risk: <60 years old and at least one diagnosis for a TE during baseline. Event-based high-risk: <60 years old and at least one diagnosis for a TE during baseline. Age-based high-risk: ≥60 years old and no diagnosis for a TE during baseline. Age-based high-risk: ≥60 years old and no diagnosis for a TE during baseline. Event- and age-based high-risk: ≥60 years old and at least one diagnosis for a TE during baseline. Event- and age-based high-risk: ≥60 years old and at least one diagnosis for a TE during baseline. TEs were identified by documented ICD-10-GM codes considering hospital main diagnoses (see Table S3 in the supplement for operationalization); thus, only considering major events requiring hospitalization. Statistical analysis The execution of phlebotomies was cross-sectionally analyzed in terms of treatment frequency in the prevalent and incident patient populations per year. The frequency of phlebotomy was assessed only from 2021 onward, since the procedure (EBM number 13505) has been billable, in addition to the basic fee, only since April 2020 by specialists in internal medicine focusing on hematology and internal oncology. Cohorts were described in terms of baseline characteristics. In longitudinal analyses, TE rates and risk were analyzed over time across HUIT/R and RUX-cohorts, with comparison between these two cohorts. TE risk was also examined by patient risk status within the prevalent patient cohort and over time from diagnosis within the incident patient cohort. Post-index TEs were identified by the first TE diagnosis after index. TE rates are reported per 1,000 person-years (PY). Post-index TE-free survival was evaluated using Kaplan-Meier analysis. Time from initiation of HU treatment to occurrence of HU intolerance or resistance was analyzed using Kaplan-Meier analysis within the HU-cohort. As an exploratory extension, outcomes following censoring due to progression to myelofibrosis were examined, including therapy switch to RUX, therapy switch to IFN‑α, therapy switch to busulfan, death, or loss to follow-up before any of these events occurred. Additionally, an exploratory analysis within the HUIT/R-cohort descriptively assessed the first observable event after the index date, considering therapy switch to RUX, therapy switch to IFN‑α or busulfan, progression to myelofibrosis, death, and loss to follow-up before any of these events occurred. Results Prevalence and incidence Between 2018 and 2023, the number of patients with prevalent PV identified in the database ranged from 822 in 2018 to 966 in 2023, while the number of patients with incident PV varied between 29 in 2018 and 91 in 2020 (Fig. 1). These counts correspond to a one-year prevalence of 31.5 per 100,000 in 2023 and a one-year incidence of 2.9 per 100,000 in 2022. Use of phlebotomies Between 2021 and 2023, about one-third of prevalent patients underwent at least one phlebotomy during the respective year (Fig. 2). Among these patients, the average number of phlebotomies was around three per year (mean ± SD by year: 2021, 3.18 ± 2.17; 2022, 3.12 ± 2.22; 2023, 3.07 ± 2.29). Phlebotomies were more frequent among incident patients, with about one in four patients receiving six or more phlebotomies within the first year after diagnosis (Fig. 2). Newly diagnosed patients who underwent at least one phlebotomy received on average between four and five phlebotomies in that first year (mean ± SD by year: 2021, 5.17 ± 3.22; 2022, 4.35 ± 2.62). The proportion of patients without phlebotomies in the first year declined from 34.8% (2021) to 21.6% (2022). Number of observed phlebotomies in the prevalent patient populations (in the respective reference year) and in the incident patient populations (in the first year after first diagnosis) Risk of thromboembolic events Baseline characteristics of the longitudinal cohorts are shown in Table 1. Table 1Baseline characteristics of longitudinal cohortsVariableIncident patient cohort (n = 341)Prevalent patient cohort (n = 1,397)HU-cohort (n = 312)HUIT/R-cohort (n = 120)RUX-cohort (n = 85)Age (mean, SD)67.0913.5667.0914.3273.9810.2774.4810.0072.4510.04Female (n, %)14341.963145.217455.86957.54249.4Cardiovascular risk factors1 ≥ 1 CVRF (n, %)19858.180457.618860.37360.84350.6 Arterial hypertension (n, %)24371.3103273.926584.910285.07284.7 Hypercholesterolemia/dyslipidemia (n, %)14843.461444.015148.45747.53338.8 Obesity8926.133724.15116.32420.01315.3 Diabetes (n, %)8825.834524.76520.81915.878.2 Tobacco use (n, %)4814.119614.0216.7108.3< 5-Risk status Low-risk (n, %)8023.532523.3216.786.7910.6 Event-based high-risk (n, %)113.2674.8113.5< 5-< 5- Age-based high-risk (n, %)15445.262144.513743.96453.33945.9 Event- and age-based high-risk (n, %)9628.238427.514345.84436.73440.0CVRF Cardiovascular Risk Factor, HU hydroxyurea, IT/R intolerance/resistance, RUX ruxolitinib1Claims-based risk factors may be underestimated due to incomplete documentation, particularly concerning life-style related factors such as smoking and obesityNote: values are not shown if the sample size was < 5 to ensure data protection Baseline characteristics of longitudinal cohorts CVRF Cardiovascular Risk Factor, HU hydroxyurea, IT/R intolerance/resistance, RUX ruxolitinib 1Claims-based risk factors may be underestimated due to incomplete documentation, particularly concerning life-style related factors such as smoking and obesity Note: values are not shown if the sample size was < 5 to ensure data protection The incident patient cohort included 341 patients (mean age: 67.1 years; 41.9% female). A post-index TE was documented in 27 patients (7.9%), with 10 events diagnosed at the time of initial PV diagnosis. The Kaplan-Meier analysis showed TE-free survival probabilities of 91.9% at one year (365 days) post-index, declining to 89.3% at 2 years (730 days) and 83.2% at 3 years (1,095 days) post-index (Fig. 3). Kaplan-Meier analysis evaluating the probability of thromboembolic event-free survival in the incident patient cohort. Note: values are not shown, if the sample size was < 5 to ensure data protection. Censoring: progression to myelofibrosis, death, loss to follow-up The HUIT/R-cohort included 120 patients and the RUX-cohort included 85 patients. Patients in the RUX cohort were younger on average (72.5 vs. 74.5 years) and had a lower proportion of women (49.4% vs. 57.5%) compared with the HUIT/R-cohort (Table 1). Post-index TE incidence was higher in the HUIT/R-cohort (9.2% vs. 7.1%). The rate of post-index TEs was 1.57-fold (95% CI: 0.58–4.58) higher in the HUIT/R-cohort than the RUX-cohort (72.9 vs. 46.7 per 1,000 person-years). The Kaplan-Meier analysis indicated TE-free survival probabilities at one year (365 days) post-index of 90.5% for the HUIT/R-cohort and 94.2% for the RUX-cohort (Fig. 4). However, differences between the curves were not significant (log-rank χ² = 0.43, p-value = 0.51). Kaplan-Meier analysis evaluating the probability of thromboembolic event-free survival in the RUX- and HUIT/R-cohorts. HU – hydroxyurea, IT/R – intolerance/resistance, RUX – ruxolitinib. Censoring: treatment switch to a different cytroreductive drug, treatment discontinuation, progression to myelofibrosis, death, loss to follow-up. Note: values are not shown, if the sample size was < 5 to ensure data protection The prevalent patient cohort included 1,397 patients (mean age: 67.1 years, 45.2% female; Fig. 1). Patients were classified by risk as 23.3% low-risk, 44.5% age-based high-risk, 27.5% event- and age-based high-risk, and 4.8% event-based high-risk (Table 1). Post-index TE incidence was highest among event- and age-based high-risk patients (21.1%), followed by age-based high-risk patients (11.3%), event-based high-risk patients (9.0%), and low-risk patients (4.3%). Compared with low-risk patients, the post-index TE rate was 6.2-fold higher in event- and age-based high-risk patients (82.4 vs. 13.4 per 1,000 PY), 2.9-fold higher in age-based high-risk patients (39.1 vs. 13.4 per 1,000 PY), and 2.2-fold higher in event-based high-risk patients (29.7 vs. 13.4 per 1,000 PY). The Kaplan-Meier analysis confirmed the lowest TE-free survival in event- and age-based high-risk patients, with similar TE-free survival in age-based high-risk patients and event-based high-risk patients (Fig. 5). Kaplan-Meier analysis evaluating the probability of thromboembolic event-free survival in the prevalent patient cohort stratified by risk status. Note: values are not shown, if the sample size was < 5 to ensure data protection. Censoring: progression to myelofibrosis, death, loss to follow-up HU intolerance/resistance The HU-cohort included 312 patients (mean age: 74.0 years, 55.8% female; Table 1). During follow-up, proxy measures of HU intolerance or resistance were observed in 22.1% of patients (n = 69), occurring on average after 451 days (median 337 days). The Kaplan-Meier analysis showed a probability of remaining free from HU intolerance/resistance of 85.5% at one year (365 days), 76.2% at 2 years (730 days), and 68.3% at three years (1,095 days) following HU initiation (Fig. 6). A total of 93 patients (29.8%) were censored due to progression to myelofibrosis. Among these, most did not switch to an alternative cytoreductive drug; 11 patients (11.8%) switched to RUX, and 5 patients (5.4%) switched to IFN‑α. Twenty-six patients (28.0%) died, while 51 patients (54.8%) were lost to follow-up before any of the predefined events were observed. Kaplan-Meier analysis evaluating the probability of HU intolerance/resistance-free survival in the HU-cohort. Censoring: treatment switch to a different cytroreductive drug, progression to myelofibrosis, death, loss to follow-up In the HUIT/R cohort, the first observed event after index was progression to myelofibrosis in 33 patients (27.5%), death in 19 patients (15.8%), and switch to RUX in 16 patients (13.3%). Fewer than five patients switched to IFN-α or busulfan (n < 5). Overall, 50 patients (41.7%) were lost to follow-up before any event was recorded. The mean observation period for these patients was 1,259 days (median: 1,281 days), corresponding to approximately 3.5 years. Discussion This retrospective study based on German health claims data provides insights into the epidemiology and management of PV in routine clinical practice. The observed one-year prevalence of 31.5 per 100,000 in 2023 and one-year incidence of 2.9 per 100,000 in 2022 are consistent with previous estimates based on the same database [11]. This study further identified substantial variation in TE risk across patient subgroups. Between 2021 and 2023, approximately one-third of patients with prevalent PV underwent phlebotomy at least once per calendar year, with an average of three procedures annually. This aligns with current guideline recommendations for hematocrit control [4]. However, frequent phlebotomies may signal inadequate disease control and have been associated with increased TE risk in patients receiving HU [18]. Moreover, repeated phlebotomy procedures impose a considerable burden on patients, including time commitment and potential quality-of-life impairment [19, 20]. While phlebotomy remains essential for reducing thrombotic risk, the resulting iron deficiency may paradoxically promote thrombosis through upregulation of prothrombotic genes [21]. The higher frequency of phlebotomies among incident patients, with about one in four receiving six or more procedures in their first year after diagnosis, likely reflects the initial management phase aimed at achieving hematocrit control. The decline in the proportion of incident patients without phlebotomy during the first year after diagnosis (from 34.8% in 2021 to 21.6% in 2022) may partly result from improved documentation practice since phlebotomy became billable in 2020. The numerically lower TE rate observed in patients treated with RUX (RUX-cohort) compared to HU-treated patients who met our proxy measures for intolerance/resistance (HUIT/R-cohort) (72.9 vs. 46.7 per 1,000 person-years) is directionally consistent with findings from clinical trials which demonstrated TE risk reduction with RUX [22, 23]. However, this descriptive analysis cannot be interpreted as evidence of treatment effectiveness due to the non-randomized, unadjusted nature of the comparison and the small sample sizes. Further studies are needed to better characterize the relationship between treatment patterns and thromboembolic risk in routine clinical practice. The analysis of HU intolerance/resistance development showed that in approximately 22% of patients, proxy measures for HU intolerance/resistance were detected during follow-up, occurring on average after 451 days. The Kaplan-Meier analysis indicated that about one-third of patients may develop signs of HU intolerance/resistance within three years of treatment initiation. This aligns with earlier studies showing that HU effectiveness can vary, and some patients may develop intolerance or resistance [8, 24]. Intolerance and/or resistance to HU has been associated with inadequate disease control, increased risk of thrombotic complications, as well as progression to myelofibrosis or acute leukemia [7, 24]. Thus this finding supports the importance for regular monitoring of patients for signs of intolerance and resistance and the prompt initiation of alternative therapies (RUX or INF-α) when needed, as recommended by guidelines [4]. We observed a high rate of progression to myelofibrosis in the HU-cohort, with 29.8% of patients censored for this reason. Progression to myelofibrosis is a major late complication associated with poorer prognosis [25]. Notably, most patients who progressed did not receive alternative cytoreductive therapy, despite German guidelines recommending JAK inhibitor therapy for myelofibrosis [26]. This finding could indicate gaps in optimal disease management in Germany and highlights the need for improved adherence to guideline-based care. Thromboembolic risk varied markedly across patient subgroups, supporting the relevance of prior TE history in addition to age in current risk stratification. Patients classified as event- and age-based high-risk exhibited the highest TE rate (82.4 per 1,000 PY), a 6.2-fold increase compared with low-risk patients (13.4 per 1,000 PY). Age-based high-risk patients had a 2.9-fold higher rate (39.1 per 1,000 PY), and event-based high-risk patients a 2.2-fold higher rate (29.7 per 1,000 PY). These differences align with existing evidence identifying advanced age and prior thrombosis as the strongest predictors of vascular complications in PV [27]. Kaplan-Meier analysis of TE-free survival in incident patients revealed an initial steep decline, followed by a gradual but persistent decrease from 91.9% at one year to 89.3% at three years post-diagnosis. This pattern may reflect that PV diagnosis is sometimes triggered by a TE. The subsequent steady decline suggests an ongoing risk of thrombosis throughout the disease course, consistent with findings from long-term follow-up studies [27, 28]. Strengths and limitations This study has several strengths, including the use of a large, representative German health claims database with longitudinal follow-up, enabling comprehensive capture of healthcare utilization across multiple sectors. The administrative nature of the data ensures complete ascertainment of coded diagnoses and procedures. Furthermore, the approach builds on previous methodological work with this database, allowing for consistent case identification and outcome assessment [11]. However, certain limitations must be acknowledged. Reliance on administrative data for case identification and outcome ascertainment may introduce misclassification bias, particularly for HU intolerance/resistance, as clinical parameters and symptom burden are not available in claims data. The proxy measures for HU intolerance/resistance, although developed in consultation with a clinical expert, have not been validated against gold-standard clinical assessments. Furthermore, thromboembolic events (TE) were defined to include hemorrhagic events, which may have led to a slight overestimation of TE incidence. Additionally, key clinical parameters such as hematocrit levels, molecular markers, cardiovascular risk factors and symptom burden are not or only partially captured in administrative databases, limiting our ability to assess treatment adequacy, disease control or comparative treatment effectiveness. As this was a descriptive claims-based analysis, the observed differences between treatment groups should be interpreted with caution, particularly given the absence of adjustment for potential confounders. In addition, potential regional or center-level variation in treatment patterns and outcomes within Germany was not assessed in the present analysis. Future studies may further explore regional differences in care delivery and guideline implementation and integrate claims data with clinical registries or prospective data to better characterize predictors of outcomes and treatment strategies in PV. The relatively short observation period of up to six years and the small sample sizes in the treatment-specific cohorts may have further limited the statistical power to detect clinically meaningful differences. Conclusions Leveraging a large, representative German health claims database, this study provides robust and recent estimates of the prevalence and incidence of PV in Germany, alongside real-world insights into treatment patterns and thromboembolic risk. The findings suggest a persistent burden of thrombotic events across all patient subgroups, with markedly higher rates observed in high-risk categories, supporting the importance of current risk stratification approaches. The findings also suggest potential gaps in disease control in routine clinical practice. The observed frequent phlebotomy use, development of HU intolerance or resistance, and the sustained thromboembolic risk over time indicate that unmet treatment needs persist across risk groups despite current management strategies. Addressing established treatment goals, including hematocrit control, symptom reduction, and prevention of disease progression, remains essential to mitigate complications and improve long-term outcomes in PV. This population-level description of healthcare utilization and disease burden and may help inform future research on disease management and outcomes in PV. Further studies are needed to better characterize predictors of outcomes, treatment response, and patient management in routine clinical practice. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements We thank Julia Lovera for her editorial assistance with this manuscript. Author contributions KCM, MF, RE, PL, AH, and HKA were involved in designing the study. BM was involved in carrying out the analyses. All authors were involved in interpreting the results, contributed to the manuscript, and approved the submitted version. Funding Funding was provided by Novartis Pharma GmbH. Data availability The data analyzed in this study was retrieved from the Institute for Applied Health Research Berlin (InGef) Research Database and cannot be made available in the manuscript, the supplementary files, or in a public repository due to German data protection laws (Bundesdatenschutzgesetz). To facilitate the replication of results, anonymized data used for this study are stored on a secure drive at the InGef GmbH. Access to the data used in this study can only be provided to external parties under the conditions of the cooperation contract of this research project and can be assessed upon request, after written approval (contact: info@ingef.de), if required. Declarations Ethics approval Since all patient-level data in the database are anonymized to comply with German data protections regulations and German federal law, approval of an Ethics Committee was not required. The research was conducted in accordance with the Declaration of Helsinki. Competing interests The study was funded by Novartis Pharma GmbH. The authors had complete autonomy for the process of designing the study, carrying out the analyses, interpreting the results, and preparing the manuscript. This also includes the full right to publish the results without limitation.AH is an employee of the IGES Institut GmbH, which is a paid consultant to Novartis Pharma GmbH for conducting the study and preparing the manuscript. At the time of this study, KCM was an employee of the IGES Institut GmbH. BM is an employees of the InGef – Institute for Applied Health Research Berlin GmbH, which was contracted and reimbursed by IGES Institut GmbH for providing the data and carrying out the analyses.MF, RE, IB, and PL are employees of Novartis Pharma GmbH. RE and IB hold stock and/or stock options of Novartis Pharma AG. HKA received research support from Bristol-Myers Squibb, Incyte, and Novartis, consulting fees from AbbVie, AOP Health, Bristol-Myers Squibb, GSK, Novartis, and Otsuka, advisory board fees from AbbVie, AOP Health, Blueprint, Bristol-Myers Squibb, GSK, and Novartis, and a travel grant from Alexion. Patient consent statement Since anonymized administrative claims data were used and presented only in aggregate form, informed consent of patients was not required. The pre-existing database did not allow direct identification of patients, and no contact was made with patients. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement The data analyzed in this study was retrieved from the Institute for Applied Health Research Berlin (InGef) Research Database and cannot be made available in the manuscript, the supplementary files, or in a public repository due to German data protection laws (Bundesdatenschutzgesetz). To facilitate the replication of results, anonymized data used for this study are stored on a secure drive at the InGef GmbH. Access to the data used in this study can only be provided to external parties under the conditions of the cooperation contract of this research project and can be assessed upon request, after written approval (contact: info@ingef.de), if required.