Advertisement

Impact of risk factors on the occurrence of arterial thrombosis and venous thromboembolism in adults with primary immune thrombocytopenia – Results from two nationwide cohorts

Open AccessPublished:April 17, 2019DOI:https://doi.org/10.1016/j.thromres.2019.04.016

      Highlights

      • 7225 patients from France and 2490 from Sweden with primary ITP have similar incidence rates of AT and VTE.
      • Comorbidities present at ITP baseline increase the risk of both AT and VTE.
      • Age, male sex and a previous AT are the most important risk factors for AT in this patient group.
      • Age and cancer are the strongest risk factors for VTE.

      Abstract

      Background

      Previous studies have found that patients with Immune thrombocytopenia (ITP) have an increased risk of arterial thrombosis (AT) and venous thromboembolism (VTE). However, risk factors for thrombosis in adults with primary ITP remain unassessed in large cohorts.
      Aim
      To assess the occurrence and impact of risk factors for AT and VTE in patients with primary ITP in France and Sweden.

      Methods

      Both countries have national health databases, including hospital diagnoses and drug dispensing data. Adults with incident primary ITP identified using algorithms between the years 2009–2015 in France, and 2009–2016 in Sweden were included. Cumulative incidence rates (IR) of AT and VTE were calculated by risk factors and multivariable Cox models were used to estimate associations.

      Results

      The study included 7225 patients from France and 2490 from Sweden. The IR of AT were 15.0 (95% CI: 13.4–16.7) and 14.7 (95% CI: 12.4–17.5) per 1000 person-years, respectively. The incidences of VTE were 6.9 (95% CI: 5.9–8.1) and 6.5 (95% CI: 5.1–8.4), respectively. Increasing age, male sex and a previous AT were associated with AT in both countries and so were exposure to antiplatelet drugs in France and a history of VTE and chronic kidney disease in Sweden. Increasing age and a history of VTE were associated with VTE in both countries, in France also cancer.

      Conclusion

      The IR of AT and VTE were similar in France. Age and male sex remained the most important risk factors for AT, age for VTE.

      Keywords

      1. Introduction

      Immune thrombocytopenia is an autoimmune hematologic disorder characterized by a platelet count below 100 ∗ 109/L, which is classically associated with an increased risk of bleeding [
      • Rodeghiero F.
      • Stasi R.
      • Gernsheimer T.
      • Michel M.
      • Provan D.
      • Arnold D.M.
      • Bussel J.B.
      • Cines D.B.
      • Chong B.H.
      • Cooper N.
      • Godeau B.
      • Lechner K.
      • Mazzucconi M.G.
      • McMillan R.
      • Sanz M.A.
      • Imbach P.
      • Blanchette V.
      • Kuhne T.
      • Ruggeri M.
      • George J.N.
      Standardization of terminology, definitions and outcome criteria in immune thrombocytopenic purpura of adults and children: report from an international working group.
      ]. ITP can be either primary, where the underlying cause is unknown, or secondary to an underlying disorder e.g. a systemic auto-immune disease [
      • Cines D.B.
      • Liebman H.A.
      The immune thrombocytopenia syndrome: a disorder of diverse pathogenesis and clinical presentation.
      ]. Some 60–70% of adults with newly diagnosed primary ITP develop a chronic condition of the disease, defined as ITP lasting >12 months [
      • Rodeghiero F.
      • Stasi R.
      • Gernsheimer T.
      • Michel M.
      • Provan D.
      • Arnold D.M.
      • Bussel J.B.
      • Cines D.B.
      • Chong B.H.
      • Cooper N.
      • Godeau B.
      • Lechner K.
      • Mazzucconi M.G.
      • McMillan R.
      • Sanz M.A.
      • Imbach P.
      • Blanchette V.
      • Kuhne T.
      • Ruggeri M.
      • George J.N.
      Standardization of terminology, definitions and outcome criteria in immune thrombocytopenic purpura of adults and children: report from an international working group.
      ,
      • George J.N.
      • Woolf S.H.
      • Raskob G.E.
      • Wasser J.S.
      • Aledort L.M.
      • Ballem P.J.
      • Blanchette V.S.
      • Bussel J.B.
      • Cines D.B.
      • Kelton J.G.
      • Lichtin A.E.
      • McMillan R.
      • Okerbloom J.A.
      • Regan D.H.
      • Warrier I.
      Idiopathic thrombocytopenic purpura: a practice guideline developed by explicit methods for the American Society of Hematology.
      ,
      • Moulis G.
      • Germain J.
      • Comont T.
      • Brun N.
      • Dingremont C.
      • Castel B.
      • Arista S.
      • Sailler L.
      • Lapeyre-Mestre M.
      • Beyne-Rauzy O.
      • Godeau B.
      • Adoue D.
      • Group C.I.
      Newly diagnosed immune thrombocytopenia adults: clinical epidemiology, exposure to treatments, and evolution. Results of the CARMEN multicenter prospective cohort.
      ]. Chandan et al. found an increased risk of CVD in patients with ITP compared with the general population [
      • Chandan J.S.
      • Thomas T.
      • Lee S.
      • Marshall T.
      • Willis B.
      • Nirantharakumar K.
      • Gill P.
      The association between idiopathic thrombocytopenic purpura and cardiovascular disease: a retrospective cohort study.
      ] and a Danish study reported an increased mortality risk of 50% in patients with ITP due to cardiovascular diseases (CVD) compared with the general population [
      • Frederiksen H.
      • Maegbaek M.L.
      • Norgaard M.
      Twenty-year mortality of adult patients with primary immune thrombocytopenia: a Danish population-based cohort study.
      ]. The mechanisms behind the increased risk of thromboembolic events in patients with ITP is to a large extent unknown but could partly be linked to a pathological complement activation on/by platelets contributing to vascular inflammation and thrombosis [
      • Peerschke E.I.
      • Yin W.
      • Ghebrehiwet B.
      Complement activation on platelets: implications for vascular inflammation and thrombosis.
      ,
      • Peerschke E.I.
      • Andemariam B.
      • Yin W.
      • Bussel J.B.
      Complement activation on platelets correlates with a decrease in circulating immature platelets in patients with immune thrombocytopenic purpura.
      ]. An increased risk of thrombosis after splenectomy has also been reported in other studies [
      • Ruggeri M.
      • Tosetto A.
      • Palandri F.
      • Polverelli N.
      • Mazzucconi M.G.
      • Santoro C.
      • Gaidano G.
      • Lunghi M.
      • Zaja F.
      • De Stefano V.
      • Sartori R.
      • Fazi P.
      • Rodeghiero F.
      • Gruppo Italiano Malattie E.d.A.A.
      • G.S.I.T.P. Thrombocytopenias Working Party
      Thrombotic risk in patients with primary immune thrombocytopenia is only mildly increased and explained by personal and treatment-related risk factors.
      ]. Boyle et al. reported an increased cumulative incidence of venous thromboembolism (VTE) in patients with ITP after splenectomy [
      • Boyle S.
      • White R.H.
      • Brunson A.
      • Wun T.
      Splenectomy and the incidence of venous thromboembolism and sepsis in patients with immune thrombocytopenia.
      ], and an almost 3-fold increased remaining risk after splenectomy compared with the general population (adjusted for age) was reported by Thomsen et al. [
      • Thomsen R.W.
      • Schoonen W.M.
      • Farkas D.K.
      • Riis A.
      • Fryzek J.P.
      • Sorensen H.T.
      Risk of venous thromboembolism in splenectomized patients compared with the general population and appendectomized patients: a 10-year nationwide cohort study.
      ]. In the Nordic countries, Norgaard et al. reported an increased rate of stroke in Danish patients who had a splenectomy for various indications [
      • Rorholt M.
      • Ghanima W.
      • Farkas D.K.
      • Norgaard M.
      Risk of cardiovascular events and pulmonary hypertension following splenectomy - a Danish population-based cohort study from 1996-2012.
      ]. The knowledge about the risk for VTE in non-splenectomized patients is limited and the prevalence and impact of baseline risk factors for arterial thrombosis (AT) and VTE in patients with ITP have, to our knowledge, not yet been assessed in large cohorts. Including data from almost 10,000 patients with primary ITP in France and Sweden made it possible to assess incidence rates (IR) of AT and VTE, as well as the prevalence and the impact of baseline risk factors in adults with primary ITP.

      2. Methods

      We included nationwide data from France and Sweden on patients with primary ITP and performed a cohort study to estimate occurrences of AT and VTE. We also undertook an assessment of risk factors.

      2.1 Data sources

      In France, the national health system database Système National des Données de Santé, (SNDS) records linkable health information with virtual coverage of the entire French population (66 million inhabitants). It links information from the hospital database (named Programme de médicalisation des systems des informations), that contains notably inpatient discharge diagnoses in all public and private hospitals in France; and from the inter-scheme consumption (Données de consommation inter-régimes - DCIR) data set, that contains all outpatient reimbursed health expenditures, including community dispensing of reimbursed drugs, as well as chronic diseases allowing full reimbursement that are notified by general practitioners. Diagnoses are coded using the International Classification of Diseases, version 10 (ICD-10). An anonymous identifier is derived from the social security number of insured individuals and can be used for linkage between the registers in the database [
      • Moulis G.
      • Lapeyre-Mestre M.
      • Palmaro A.
      • Pugnet G.
      • Montastruc J.L.
      • Sailler L.
      French health insurance databases: what interest for medical research?.
      ,
      • Bezin J.
      • Duong M.
      • Lassalle R.
      • Droz C.
      • Pariente A.
      • Blin P.
      • Moore N.
      The national healthcare system claims databases in France, SNIIRAM and EGB: powerful tools for pharmacoepidemiology.
      ].
      The Swedish National Patient Register (NPR) was founded in 1964, and from 1987 it includes nationwide data for all hospitalizations with recording covering details of hospitalizations such as date and duration of stays, hospital and department names, surgical procedures and discharge diagnoses with a main diagnosis and up to 30 contributory diagnoses (coded with the ICD, 7th–10th versions). The quality of the register is of high standard with almost complete coverage of inpatient care [

      Socialstyrelsen, (2018).

      ]. The coverage on information from outpatient care has increased over the years. The ICD-codes for several diseases have been validated using information from medical charts with a positive predictive value of 85–95% [
      • Ludvigsson J.F.
      • Andersson E.
      • Ekbom A.
      • Feychting M.
      • Kim J.L.
      • Reuterwall C.
      • Heurgren M.
      • Olausson P.O.
      External review and validation of the Swedish national inpatient register.
      ]. Information on dispensed drugs can be obtained from the Swedish Prescribed Drug Register (PDR). The register has been available and linkable to other registers in Sweden since July 2005. The PDR contains detailed information on filled prescriptions including product, quantity, dates of prescription and dispensing. The register has complete coverage of all prescribed and dispensed drugs [
      • Wettermark B.
      • Hammar N.
      • Fored C.M.
      • Leimanis A.
      • Otterblad Olausson P.
      • Bergman U.
      • Persson I.
      • Sundstrom A.
      • Westerholm B.
      • Rosen M.
      The new Swedish Prescribed Drug Register--opportunities for pharmacoepidemiological research and experience from the first six months.
      ,
      • Wallerstedt S.M.
      • Wettermark B.
      • Hoffmann M.
      The first decade with the Swedish prescribed drug register - a systematic review of the output in the scientific literature.
      ]. Drugs given in hospital are not covered in the register. The unique personal identification number given to each Swedish resident at birth or immigration is used to link the different registers.

      2.2 Identification of patients

      In France, patients with a diagnosis of ITP from 2009 to 2015 were identified with the ICD-10 code D69.3 (immune thrombocytopenia). Adult patients (≥18 years old) with incident primary ITP were identified using an algorithm combining diagnosis codes and drug exposures, which demonstrated high positive predictive values in identifying the date of ITP diagnosis, detailed elsewhere [
      • Mezaache S.
      • Derumeaux H.
      • Ferraro P.
      • Capdepon P.
      • Steinbach J.C.
      • Abballe X.
      • Palas D.
      • Saichi N.
      • Desboeuf K.
      • Lapeyre-Mestre M.
      • Sailler L.
      • Moulis G.
      Validation of an algorithm identifying incident primary immune thrombocytopenia in the French national health insurance database.
      ]. In Sweden, NPR was used to identify adult patients with a diagnosis of incident primary ITP from 2009 to 2016 in out- or inpatient care using the ICD-10 code D69.3 (immune thrombocytopenia).
      Patients with a D69.3 code at any discharge location were included. Other codes indicating that the thrombocytopenia was due to secondary ITP, such as systemic lupus erythematosus (within the year prior to the diagnosis of ITP or sixth months after), were used to exclude patients who did not have a primary ITP (Supplementary Table 1). We used splenectomy as an exclusion criteria and as a censoring variable because it has been identified as a major risk factor of VTE, both at the time of the surgery, and also after long-term follow up [
      • Ruggeri M.
      • Tosetto A.
      • Palandri F.
      • Polverelli N.
      • Mazzucconi M.G.
      • Santoro C.
      • Gaidano G.
      • Lunghi M.
      • Zaja F.
      • De Stefano V.
      • Sartori R.
      • Fazi P.
      • Rodeghiero F.
      • Gruppo Italiano Malattie E.d.A.A.
      • G.S.I.T.P. Thrombocytopenias Working Party
      Thrombotic risk in patients with primary immune thrombocytopenia is only mildly increased and explained by personal and treatment-related risk factors.
      ,
      • Boyle S.
      • White R.H.
      • Brunson A.
      • Wun T.
      Splenectomy and the incidence of venous thromboembolism and sepsis in patients with immune thrombocytopenia.
      ,
      • Thomsen R.W.
      • Schoonen W.M.
      • Farkas D.K.
      • Riis A.
      • Fryzek J.P.
      • Sorensen H.T.
      Risk of venous thromboembolism in splenectomized patients compared with the general population and appendectomized patients: a 10-year nationwide cohort study.
      ,
      • Rorholt M.
      • Ghanima W.
      • Farkas D.K.
      • Norgaard M.
      Risk of cardiovascular events and pulmonary hypertension following splenectomy - a Danish population-based cohort study from 1996-2012.
      ,
      • Thai L.H.
      • Mahevas M.
      • Roudot-Thoraval F.
      • Limal N.
      • Languille L.
      • Dumas G.
      • Khellaf M.
      • Bierling P.
      • Michel M.
      • Godeau B.
      Long-term complications of splenectomy in adult immune thrombocytopenia.
      ]. Consequently, because we aimed at assessing the impact of baseline risk factors without modification by intervention (demonstrated as linked to thrombosis like splenectomy) we used the surgical procedure code to exclude the follow-up time after splenectomy.

      2.3 Risk factors for thrombosis

      History of VTE and/or AT was searched for using discharge diagnoses in the register (Supplemental Table 3). Solid tumors, chronic kidney disease (stages 3–5), and liver disease were identified through ICD-10 codes for both in-and out patients. Mild, moderate and severe liver diseases were combined into one variable and considered as a single risk factor for VTE (Supplemental Table 4). Diabetes was identified through relevant ICD-codes or by dispensed antidiabetics. Drugs commonly used in the treatment of CVD, were considered as proxies for CVD. Accordingly, information on antiplatelet, antithrombotic, antihypertensive and lipid lowering drugs (AT only) and antiplatelet and antithrombotic drugs (VTE only) were included as time-varying covariates in the multivariable analyses. Only dispensations in the 6 months before the date of ITP diagnosis were considered. (Supplemental Table 4).

      2.4 Outcome definition

      Relevant ICD-10 codes for AT and VTE were searched for in the hospital databases among primary discharge diagnoses. The codes used are indicated in Supplementary Table 3. Only the first event of each type that occurred after the diagnosis of ITP were considered in the analyzes.

      2.5 Ethical approvals

      For the French data, authorizations for use were obtained from the Institut des Données de Santé in March 2012 (no. 40) and the Commission Nationale de l'Informatique et des Libertés in July 2012 (no. DE-2012-076). In Sweden, the Regional Ethics Committee at Karolinska Institutet and the National Board of Health and Welfare approved the study (Record no 2009/4:10, addendum 2009 1597–31/4 and 2017/53–32).

      2.6 Statistical methods

      The cumulative IR of AT and of VTE were calculated by dividing the number of thrombotic events with person-time in years from the date of diagnosis of ITP to time of censoring due to either an outcome (AT or VTE), splenectomy, death or end of follow up (December 31, 2015 France and December 31, 2016 Sweden). AT and VTE were analyzed separately. The cumulative IRs were first calculated in the overall ITP population in each country, and then by baseline risk factors. For the baseline risk factors: cancer, diabetes, chronic liver disease, kidney disease, exposure to antiplatelet, antithrombotic, antihypertensive or lipid lowering drugs, incidence rate ratios (IRR) were estimated. Patients with the risk factor at baseline were compared with patients who did not have the risk factor. In the sensitivity analysis the baseline risk factors were adjusted for age and sex.
      We chose to carry out a cohort study adjusting for age, sex and other relevant covariates [
      • Brazauskas R.
      • Logan B.R.
      Observational studies: matching or regression?.
      ]. In our main analysis we performed a time dependent Multivariable Cox counting process to estimate associations between risk factors of AT and VTE, respectively. All variables were adjusted for, including both time-varying covariates and fixed covariates. The follow-up time was from the date of diagnosis of ITP to the time of censoring, defined as above. The following covariates were considered as time-varying covariates: diabetes, renal disease, liver disease, cancer, and treatments associated with cardiovascular disease, such as antiplatelet drugs and antithrombotic drugs. Fixed covariates were age at ITP diagnosis (categorized in four categories: 18–39 years, 40–59, 60–74 and 75 or older), sex and previous AT and VTE events. We used the counting process approach to estimate hazard ratios (HR) and 95% confidence intervals (CI). The counting process approach is an extended version of Cox regression, which requires that data are split into several records (time intervals) for each patient, measuring time to event and taking all covariates into account [
      • Guo Z.
      • Gill T.M.
      • Allore H.G.
      Modeling repeated time-to-event health conditions with discontinuous risk intervals. An example of a longitudinal study of functional disability among older persons.
      ]. The proportional hazard assumption was tested by plotting Kaplan-curves. SAS statistical software™ version 9.4 (SAS Institute, Cary, North Carolina, USA) was used to analyze the data.

      3. Results

      3.1 Patient selection

      The number of included adult patients with incident primary ITP in the study period was 7225 from France (2009–2015) and 2490 from Sweden (2009–2016). There were slightly more females in both countries: 4046 (56.0%) in France and 1270 (51.0%) in Sweden. Median age at diagnosis of ITP was 61 years in France with interquartile range (IQR) 39–76 and the corresponding figure in Sweden was 64 years (IQR) 39–77 (Table 1).
      Table 1Characteristics of patients with primary immune thrombocytopenia (ITP) in Sweden and France.
      SwedenFrance
      TotalMaleFemaleTotalMaleFemale
      Patients with primary ITP in study period N (%)24901220 (49)1270 (51)72253199 (44)4046 (56)
      Age at ITP diagnosis
       Median

      (IQR)
      64

      (39–77)
      66

      (49–78)
      60

      (34–76)
      61

      (39–76)
      65

      (47–78)
      57

      (35–75)
       Mean

      (SD)
      59

      (22)
      62

      (20)
      56

      (22)
      58

      (23)
      61

      (20)
      55

      (22)
      Number of events
       Arterial
      I740 I741 I742 I743 I744 I745 I748 I749 I20 I21 I22 I24 I255 I256 I513 I630 I631 I632 I634 I635 I638 I639 I65 I66
      13177 (59)54 (41)316214 (68)102(32)
       Venous
      I800 I801 I802 I803 I808 I809 I81 I821 1822 1823 1828 1829 I260 1269 I636 I676 O22.
      5930 (51)29 (49)14865 (44)83 (56)
      Rate of AT

      95% CI per 1000

      Rate of VT
      14.7

      (12.4–17.5)

      6.5
      19. 0

      (15.2–23.7)

      7.1
      11.2

      (8.6–14.6)

      6.0.
      15.0

      (13.4–16.7)

      6.9
      24.1

      (21.1–27.6)

      7.1
      8.4

      (6.9–10.1)

      6.8
      95% CI per 1000 patient-years(5.1–8.4)4.9–10.14.2–8.7(5.9–8.1)(5.6–9.1)(5.5–8.4)
      a I740 I741 I742 I743 I744 I745 I748 I749 I20 I21 I22 I24 I255 I256 I513 I630 I631 I632 I634 I635 I638 I639 I65 I66
      b I800 I801 I802 I803 I808 I809 I81 I821 1822 1823 1828 1829 I260 1269 I636 I676 O22.

      3.2 Risk factors

      3.2.1 Incidence rates of arterial thrombosis by country, age, sex and type of diagnosis

      The IR of AT for all included patients with primary ITP was 15.0 (95% CI 13.4–16.7 per 1000 person-years) in France and 14.7 (95% CI 12.4–17.5 per 1000 person-years) in Sweden (Table 1). In female patients, the IR was 8.4 (95% CI 6.9–10.1 per 1000 person-years) in France and 11.2 (95% CI 8.6–14.6 per 1000 person-years) in Sweden. The IR in male patients was 24.1 (95% CI 21.1–27.6 per 1000 person-years) in France and 19.0 (95% CI 15.2–23.7 per 1000 person-years) in Sweden.
      There were 316 events of AT in France and 131 in Sweden, from those 68 (22%) and 40 (31%) were cerebral infarction (I63.0–I63.5, I63.8, I63.9), 50 (16%) and 55 (42%) were acute myocardial infarction (I21.1, I21.2, I21.4, I21.9), 96 (30%) and 24 (18%) were angina pectoris (I20.0, I20.8, I20.9), 102 (32%) and 12 (9%) were other arterial events (I22.1, I22.9, I24.0, I24.8, I24.9, I74.2 I74.5, I74.8, I51.3, I25.5, I25.6, I65.2, I66.9), respectively. The median age at the time of the event was 73 years (IQR 64–82) in France and 214 (68%) of all patients with AT were male. The corresponding figures in Sweden were 81 years (IQR 73–87) and 77 (59%). In total, 171 (54%) of those with AT in France and 17 (13%) in Sweden had a comorbidity (cancer, diabetes or kidney disease) at the time of the event, 79 (25%) and 8 (6%) received antithrombotic drugs and 133 (42%) and 6 (5%) respectively received antiplatelet drugs. In addition, 76 (24%) in France and 32 (24%) in Sweden had a history of AT.

      3.3 Baseline risk factors for arterial thrombosis

      IRs increased with age, and in France, those 60 years or older had an IR of 24.9 (95% CI 20.9–29.7 per 1000 person-years). For the Swedish cohort the IR was 16.1 (95% CI 11.8–22.0 per 1000 person-years), Table 2. The most prevalent risk factors of AT other than male sex and increasing age was a history of AT. A history of an AT at baseline yielded an IRR of 4.8 (95% CI 3.8–6.1) in France, and 2.6 (95% CI 1.8–3.8) in Sweden. A history of VTE at baseline yielded an IR three times higher in France IRR = 2.9 (95% CI 1.0–8.0) and four times higher in Sweden IRR = 4.3 (95% CI 1.9–9.5) than those without a history of VTE at baseline.
      Table 2Arterial event rates, baseline variables.
      Sweden

      Censored on splenectomy, arterial event, death
      France

      Censored on splenectomy, arterial event, death
      Person-time

      (years)
      Number of eventsRateCI0 lowCi highPerson-time

      (years)
      Number of eventsRateCI lowCI high
      All8898.913114.712.417.521,083.131615.013.416.7
      SexMale4064.47719.015.223.78874.521424.121.127.6
      Female4737.75411.28.614.612,208.61028.46.910.1
      Age group (years)18–392649.820.80.23.05719.850.90.42.1
      40–591881.573.71.87.84775.9428.86.511.9
      60–742422.43916.111.8122.05018.012524.920.929.7
      ≤751945.38342.734.452.95569.414425.922.030.4
      Diabetes1No7972.99712.210.014.918,483.324013.011.414.7
      Yes926.03436.726.251.42599.87629.223.336.6
      Cardiovascular disease2No8874.212814.412.117.219,426.222111.410.013.0
      Yes24.73121.339.1376.01656.99557.346.970.1
      Kidney disease3No8845.212714.412.117.120,478.329414.412.816.1
      Yes53.7474.528.0198.5604.72236.424.055.3
      Liver disease4No8821.512714.412.117.120,374.930615.013.416.8
      Yes77.4451.719.4137.7708.11014.17.626.2
      Cancer5No8332.011914.311.917.119,534.028214.412.816.2
      Yes567.01221.112.037.31549.13421.915.730.7
      History of AT6No8489.59911.79.614.220,009.824012.010.613.6
      Yes409.23278.255.3110.61073.37670.856.688.7
      History of VTE7No8823.312614.312.017.021,012.631314.913.316.6
      Yes75.6566.127.5158.970.4342.613.7132.1
      Co-medications
       Antithrombotics8No8297.910913.110.915.918,996.624212.711.214.4
      Yes601.02236.624.155.62086.47435.528.244.5
       Antiplatelets9No8632.511713.611.316.318,295.018910.39.011.9
      Yes266.41452.631.188.72788.112745.638.354.2
       Antihypertensives10No6484.1426.54.88.813,633.01128.26.89.9
      Yes2414.88936.929.945.47450.120427.423.931.4
       Lipid lowering11No7327.07610.48.313.016,709.317810.79.212.3
      Yes1571.95535.026.945.64373.813831.626.737.3
      0CI=Confidence Interval. 1E119 E10 E103-E105 E107 E112 E113-E117 E122 E123 E124 E125 E127 E132- E135 E137 E142-E145 E147 A10BA0 2I110 I130 I132 I50.
      2N183- N185
      3B18 I850 I859 I864 I982 K700-K703 K704 K709 K711 K713 K715 K717 K721 K729 K73 K74 K760 K762-K764 K765 K766 K767 K768 K769 Z944.
      4C00-C26 C30-C34 C37–41 C43 C45-C58 C60-C76 C81-C85 C88 C90-C97.
      5I740 I741 I742 I743 I744 I745 I748 I749 I20 I21 I22 I24 I255 I256 I513 I630 I631 I632 I634 I635 I638 I639 I65 I66.
      6I800I801 I802 I803 I808 I809 I81 I821 1822 1823 1828 1829 I260 1269 I636 I676 O22 7B01AA B01AB B01AE B01AF B01AX 8B01AA B01AB B01AE B01AF B01AX 9B01AC..
      10C02 C03 C07 C08C C08GA C09A C09B C09C C09D 11C10AA C10B.
      Those with a baseline risk factor such as any comorbidity or a drug for CVD had a higher IR of AT compared with not having the comorbidity or the CVD drug e.g. the cumulative incidence was higher in patients with diabetes at baseline compared to not having diabetes. The IR was 29.2 (95% CI 23.3–36.6 per 1000 person-years) for those with diabetes, and 13.0 (95% CI 11.4–14.7 per 1000 person-years) for those without diabetes in France. In Sweden, the IR was 36.7 (95% CI 26.2–51.4 per 1000 person-years) for those with diabetes and 12.2 (95% CI 10.0–14.9 per 1000 person-years) for those without diabetes. Patients who filled a prescription for drugs used for cardiovascular conditions, such as antiplatelet drugs had around four fold higher rates IRR = 4.4 (95% CI 3.5–5.5) in France and twofold in Sweden 2.2 (95% CI 1.3–3.8) compared with patients who did not (Table 2).

      3.3.1 Incidence rates of venous thromboembolism by country, age, sex and type of diagnoses

      The IR of VTE was 6.9 (95% CI 5.9–8.1 per 1000 person-years) in France and 6.5 (95% CI 5.1–8.4 per 1000 person-years) in Sweden. The IR of VTE was similar in men and women in both countries (Table 1).
      There were 148 events of VTE in France and 59 in Sweden, from those 80 (54%) and 37 (63%) had pulmonary embolism (I26.9, I26.0), 55 (37%) and 14 (24%) had phlebitis and thrombophlebitis (I80.0-I80.3, I80.8, I80.9), 13 (9%) and 8 (13%) had other venous embolism and thrombosis (I82.2, I82.8, I82.9), portal venous thrombosis (I81, I81.9) or other cerebral venous thrombosis (I63.9 and I67.6), respectively. The median age at the time of the event was 72 (IQR 56–81) in France and 65 (44%) of all patients with VTE were male. The corresponding figures in Sweden were 78 years (IQR 69–83) and 30 (54%). In total 43 (29%) of those with VTE in France and 8 (14%) in Sweden had a comorbidity at baseline (cancer, diabetes, liver or kidney disease), 51 (34%) and 8 (14%) had a co-medication with antiplatelet or antithrombotic drugs. In addition, 2 (1%) in France and 8 (26%) in Sweden had a history of VTE.

      3.4 Baseline risk factors for venous thromboembolism

      The most prevalent risk factors were increasing age, a history of VTE, and cancer. In both countries the cumulative incidence was twofold higher IR = 12.1(95% CI 7.7–19.0) in France, 12.3 (95% 5.9–25.7 CI in Sweden) in the presence of cancer at baseline compared with absence of cancer, for which the IRs were 6.5 (95% CI 5.5–7.7) and 6.1 (95% CI 4.7–8.0) respectively. All investigated baseline risk factors and drugs used as a proxy for CVD increased the risk of VTE when studied separately (Table 3.)
      Table 3Venous event rates, baseline variables.
      Sweden

      Censored of splenectomy, venous event, death
      France Censored of splenectomy, venous event, death
      Person-time

      (years)
      Number of eventsRateCI3 lowCI highPerson-time

      (years)
      Number of eventsRateCI lowCi high
      All9057596.55.18.421,379.61486.95.98.1
      SexMale4243.2307.14.910.19160.0657.15.69.1
      Female4814.5296.04.28.712,219.6836.85.58.4
      Age group (years)18–392649.831.10.43.55682.3183.22.05.0
      40–591891.863.21.47.14816.4255.23.57.7
      60–742496.5166.43.910.55208.1377.15.29.8
      ≤752019.63416.812.023.65672.76812.09.515.2
      Diabetes1No8076.9475.84.47.718,664.21286.95.88.2
      Yes980.91212.27.021.52715.4207.44.811.4
      Cardiovascular disease2No9031.4596.55.18.419,563.91256.45.47.6
      Yes26.301815.62312.78.419.1
      Kidney disease3No900.8586.45.08.320,478.329414.412.816.1
      Yes57.0117.52.5124.5604.72236.424.055.3
      Liver disease4No8979.4596.44.98.220,662.01416.85.88.1
      Yes78.3225.56.4102.1717.579.84.720.5
      Cancer5No8486.9526.14.78.019,807.31296.55.57.7
      Yes570.9712.35.925.71572.31912.17.719.0
      History of VT6No8990.7515.74.37.520,166.41376.85.88.0
      Yes67.18119.359.6238.51213.2119.15.016.4
      History of AT7No8587.1556.44.98.320,166.41376.85.88.0
      Yes470.648.53.222.71213.2119.15.0216.4
      Co-medications
       Antithrombotics8No8439.1475.64.27.419,231.51156.05.07.2
      Yes618.71219.410.034.22148.13315.410.921.6
       Antiplatelets9No8780.9546.24.78.018,359.61226.65.67.9
      Yes276.8518.17.543.43020.0268.65.912.6
       Antihypertensives10No6529.9385.84.28.013,680.4775.64.57.0
      Yes2527.8218.35.412.77699.1719.27.311.6
       Lipid lowering11No7407.0456.14.58.116,809.71016.04.97.3
      Yes1650.8148.55.014.34569.94710.37.713.7
      0CI=Confidence Interval 1 E119 E10 E103-E105 E107 E112 E113-E117 E122 E123 E124 E125 E127 E132- E135 E137 E142-E145 E147 A10BA0 2I110 I130 I132 I50.
      2N183- N185.
      3B18 I850 I859 I864 I982 K700-K703 K704 K709 K711 K713 K715 K717 K721 K729 K73 K74 K760 K762-K764 K765 K766 K767 K768 K769 Z944.
      4C00-C26 C30-C34 C37–41 C43 C45-C58 C60-C76 C81-C85 C88 C90-C97
      5I740 I741 I742 I743 I744 I745 I748 I749 I20 I21 I22 I24 I255 I256 I513 I630 I631 I632 I634 I635 I638 I639 I65 I66
      6I800I801 I802 I803 I808 I809 I81 I821 1822 1823 1828 1829 I260 1269 I636 I676 O22
      7B01AA B01AB B01AE B01AF B01AX
      8B01AA B01AB B01AE B01AF B01AX
      9B01AC.
      10C02 C03 C07 C08C C08GA C09A C09B C09C C09D 11C10AA C10B.

      3.5 Sensitivity analysis

      Stratifying by age and sex the estimates were imprecise, but indicated that higher age contributed to a higher IR. Male patients 60 years or older had a more than twofold higher IR for AT than females of the same age (Supplemental Table 1). For VTE, the IR for females and males 75 years or older were similar. There were no recordings of certain comorbidities and CVD drugs in the lowest age group, e.g. cancer and kidney diseases (Supplemental Table 2).

      3.6 Cox multivariable analyses

      In the multivariable Cox models, risk factors associated with an increased risk of AT were: age, in age-group 60–74 years, with an HR of 19.7 (95% CI 7.5–51.3) in France, and 23.4 (95% CI 5.5–101.2) in Sweden; a history of AT, in France the HR was 2.5 (95% CI 1.8–3.5) and in Sweden it was 3.1 (95% CI (2.0–5.0). In France, exposure to antiplatelet drugs showed an increased risk, with an HR of 1.6 (95% CI 1.2–2.1) and in Sweden, a history of VTE showed an increased risk, with an HR of 2.6 (95% CI 1.1–6.5). In Sweden, chronic kidney disease was also shown to increase the risk of AT, with an HR of 2.6 (95% CI 1.2–6.0), Table 4.
      Table 4Hazard ratio with confidence interval (CI) for the association between potential risk factors and arterial thrombosis.
      VariablesSwedenFrance
      Person-time

      years
      EventsUnivariate

      95% CI
      Multivariable

      95% CI
      Person-time

      years
      EventsUnivariate

      95% CI
      Multivariable

      95% CI
      Overall30,452131NANA21,083316NANA
       Male14,64877ReferenceReference8875214ReferenceReference
       Female15,472530.6 (0.4–0.8)0.7 (0.5–1.0)12,2091020.35 (0.28–0.45)0.5 (0.4–0.7)
      Age at diagnosis of ITP
       18–3970142ReferenceReference57205ReferenceReference
       40–59637575.8 (1.2–28.5)6.2 (1.3–30.6)4776429.9 (3.9–25.0)8.4 (3.2–22.2)
       60–7495953925.1 (6.0–105.0)23.4 (5.5–101.2)501812527.8 (11.4–67.8)19.7 (7.5–51.3)
       ≥7574688375.4 (18.1–314.4)67.2 (15.7–287.8)556914428.5 (11.7–69.5)18.6 (7.1–48.7)
      Diabetes1109061.6 (0.7–3.7)1.0 (0.4–2.2)3323882.3 (1.8–2.9)1.3 (1.0–1.7)
      Kidney disease265273.9 (1.8–8.4)2.6 (1.2–6.0)1776392.3 (1.6–3.1)1.2 (0.9–1.7)
      Cancer3201440.6 (0.2–1.6)0.3 (0.1–0.8)2916441.3 (0.9–1.8)0.8 (0.5–1.0)
      History of AT41563327.1 (4.7–10.7)3.1 (2.0–5.0)1073765.6 (4.3–7.3)2.5 (1.8–3.5)
      History of VTE530154.9 (2.1–11.7)2.6 (1.1–6.5)7032.9 (0.9–8.9)1.8 (0.5–6.4)
      Antithrombotics6311880.6 (0.3–1.2)0.5 (0.2–0.9)4855790.7 (0.5–1.1)0.9 (0.6–1.1)
      Antiplatelets7142061.0 (0.5–2.3)1.4 (0.6–3.2)37731332.5 (1.8–3.5)1.6 (1.2–2.1)
      Antihypertensive8289980.7 (0.3–1.4)0.7 (0.4–1.4)10,3282151.3 (0.9–1.9)0.9 (0.7–1.1)
      Lipid lowering9130130.5 (0.2–1.7)0.4 (0.1–1.4)51551371.7(1.3–2.4)1.0 (0.7–1.3)
      1E119 E10 E103-E105 E107 E112 E113-E117 E122 E123 E124 E125 E127 E132- E135 E137 E142-E145 E147 A10BA02.
      2N183- N185.
      3 C00-C26 C30-C34 C37–41 C43 C45-C58 C60-C76 C81-C85 C88 C90-C97.
      4 I740 I741 I742 I743 I744 I745 I748 I749 I20 I21 I22 I24 I255 I256 I513 I630 I631 I632 I634 I635 I638 I639 I65 I66.
      5 I800 I801 I802 I803 I808 I809 I81 I821 1822 1823 1828 1829 I260 1269 I636 I676 O22.
      6B01AA B01AB B01AE B01AF B01AX.
      7 B01AC.
      8 C02 C03 C07 C08C C08GA C09A C09B C09C C09D.
      9C10AA C10B.
      For VTE, the risk factors were being above 60 years of age, in France the HR was 2.2 (95% CI 1.2–3.8) and in Sweden it was 5.6 (95% CI 1.6–19.8). In France, cancer was shown to increase the risk, with an HR of 1.8 (95% CI 1.2–2.7). In Sweden a history of VTE increased the risk, with an HR of 13.5 (95% CI 5.5–33.4), Table 5.
      Table 5Hazard ratio with confidence interval (CI) for the association between potential risk factors and venous thromboembolism.
      VariablesSwedenFrance
      Person-time

      years
      EventsUnivariate

      95% CI
      Multivariable

      95% CI
      Person-time

      years
      EventsUnivariate

      95% CI
      Multivariable

      95% CI
      Overall31,61359NANA21,380148NANA
       Male15,51030ReferenceReference916065ReferenceReference
       Female16,103290.8 (0.5–1.4)0.9 (0.5–1.6)12,220831.0 (0.7–1.4)1.2 (0.9–1.7)
      Age at diagnosis of ITP
       18–3970243ReferenceReference568218ReferenceReference
       40–59643762.8 (0.7–11.4)2.5 (0.6–10.4)4816251.6 (0.9–2.9)1.9 (1.1–3.4)
       60–7410,029165.7 (1.6–19.7)5.6 (1.6–19.8)5208372.2 (1.2–3.8)2.3 (1.3–4.0)
       ≥7581243416.1 (4.9–52.7)15.0 (4.2–54.0)5673683.6 (2.1–6.1)3.5 (2.1–6.0)
      Diabetes1116821.2 (0.3–4.8)0.8 (0.2–3.2)3485251.3 (0.9–2.0)1.0 (0.7–1.5)
      Kidney disease27931859151.9 (1.2–3.2)1.5 (0.9–2.5)
      Liver Disease387432.1 (0.7–6.4)2.3 (0.7–7.3)1412111.5 (0.8–2.7)1.3 (0.7–2.4)
      Cancer4201431.0 (0.3–3.1)0.6 (0.2–2.0)2989272.2 (1.5–3.2)1.8 (1.2–2.7)
      History of AT5192141.4 (0.5–3.9)0.9 (0.3–2.4)1213111.3 (0.7–2.4)0.9 (0.5–1.7)
      History of VT6236821.1 (9.5–46.7)13.5 (5.5–33.4)7024.3 (1.1–17.3)3.5 (0.8–14.2)
      Antithrombotics7329271.2 (0.6–2.6)1.6 (0.6–4.56)4795270.5 (0.2–0.9)0.6 (0.4–1.0)
      Antiplatelets8179820.6 (0.2–2.6)0.3 (0.1–2.0)4318320.7 (0.4–1.2)1.0 (0.7–1.4)
      1E119 E10 E103-E105 E107 E112 E113-E117 E122 E123 E124 E125 E127 E132- E135 E137 E142-E145 E147 A10BA02.
      2N183–N185.
      3B18 I850 I859 I864 I982 K700-K703 K704 K709 K711 K713 K715 K717 K721 K729 K73 K74 K760 K762-K764 K765 K766 K767 K768 K769 Z944.
      4 C00-C26 C30-C34 C37–41 C43 C45-C58 C60-C76 C81-C85 C88 C90-C97.
      5I740 I741 I742 I743 I744 I745 I748 I749 I20 I21 I22 I24 I255 I256 I513 I630 I631 I632 I634 I635 I638 I639 I65 I66.
      6I800 I801 I802 I803 I808 I809 I81 I821 1822 1823 1828 1829 I260 1269 I636 I676 O22.
      7B01AA B01AB B01AE B01AF B01AX.
      8B01AC.

      4. Discussion

      4.1 Incidence rates of thrombosis in ITP patients

      We observed similar IRs of AT and VTE in France and Sweden in adult patients with primary ITP. The IR of AT was higher than the IR of VTE. The IR of both AT and VTE in our study are consistent with findings reported from other countries. The rates of both AT and VTE are higher than those reported for the general population. A Scandinavian study reported an IR of 16.0 for AT patients with chronic ITP and an IR of 8.8 for the general population [
      • Norgaard M.
      • Cetin K.
      • Maegbaek M.L.
      • Kristensen N.R.
      • Ghanima W.
      • Bahmanyar S.
      • Stryker S.
      • Christiansen C.F.
      Risk of arterial thrombotic and venous thromboembolic events in patients with primary chronic immune thrombocytopenia: a Scandinavian population-based cohort study.
      ]. A meta-analysis of observational studies from both Europe and the US (all stages of ITP were included) reported the IR of AT to be in the range of 10.0–28.0 in patients with ITP and 7.0–18.0 in populations without ITP. For VTE the meta-analysis reported an IR of 4.0–7.0 in patients with ITP, and 1.0–4.0 in populations without ITP [
      • Langeberg W.J.
      • Schoonen W.M.
      • Eisen M.
      • Gamelin L.
      • Stryker S.
      Thromboembolism in patients with immune thrombocytopenia (ITP): a meta-analysis of observational studies.
      ]. The Scandinavian study reported the IR for VTE in chronic ITP to be 9.0 [
      • Norgaard M.
      • Cetin K.
      • Maegbaek M.L.
      • Kristensen N.R.
      • Ghanima W.
      • Bahmanyar S.
      • Stryker S.
      • Christiansen C.F.
      Risk of arterial thrombotic and venous thromboembolic events in patients with primary chronic immune thrombocytopenia: a Scandinavian population-based cohort study.
      ] and for the general population without ITP to be 2.8. In a study from the USA the HR of a VTE event was 7.7 per 1000 person-years relative to an HR of 1.9 per 1000 person-years for those without ITP [
      • Yusuf H.R.
      • Hooper W.C.
      • Grosse S.D.
      • Parker C.S.
      • Boulet S.L.
      • Ortel T.L.
      Risk of venous thromboembolism occurrence among adults with selected autoimmune diseases: a study among a U.S. cohort of commercial insurance enrollees.
      ].
      The clinical relevance of our findings, in the light of these other studies, is the increased risk of AT and VTE in patients with ITP, the similar rates in France and Sweden strengthen this finding. Treating patients with ITP can be complex since they have both a risk of bleeding and of thrombosis, this is especially important in patients with anticoagulation, and in an elderly population it is extremely relevant.

      4.2 Comorbidities and medications for cardiovascular disease at baseline

      We confirmed that having a cardiovascular disease, when diagnosed with ITP, is associated with an increased risk of both arterial and venous thrombosis and that an increased risk of AT was associated with high age, male sex and a previous AT in both countries. Among French patients with ITP, the exposure to antiplatelet drugs, a proxy for baseline CVD, was also associated with AT occurrence. In addition, among the Swedish patients with an ITP diagnosis, a previous venous thrombosis and chronic kidney disease diagnosis increased the risk. Increasing age was associated with VTE in both countries, as well as cancer in France. Cancer is a heterogeneous disease and the slightly higher risk in France may be linked to the fact that some cancers vary in incidence from country to country [
      • Ferlay J.
      • Steliarova-Foucher E.
      • Lortet-Tieulent J.
      • Rosso S.
      • Coebergh J.W.
      • Comber H.
      • Forman D.
      • Bray F.
      Cancer incidence and mortality patterns in Europe: estimates for 40 countries in 2012.
      ]. It could also be due to the larger population in France, relative to Sweden, meaning there is additional power to detect an association. Using the multivariable model where confounders can be simultaneously adjusted for, even though some residual confounding might remain, we confirmed that increasing age and a history of VTE are two important risk factors of VTE. The incidence of ITP increases with age, the median age at diagnosis is 61 years (IQR 39–76) in France and 64 years (IQR 39–77) in Sweden.

      4.3 In comparison to other studies

      Other studies which have investigated CVD in patients with ITP have produced similar results to those found in this study. Chandan et al. found an IRR of 1.4 (95% CI 1.23–1.55) of CVD in patients with incident ITP compared to a comparison cohort without ITP for conditions such as hypertension, diabetes, use of lipid lowering drugs and antiplatelet drugs [
      • Chandan J.S.
      • Thomas T.
      • Lee S.
      • Marshall T.
      • Willis B.
      • Nirantharakumar K.
      • Gill P.
      The association between idiopathic thrombocytopenic purpura and cardiovascular disease: a retrospective cohort study.
      ]. Moulis et al. previously reported data from the clinical CARMEN registry where they found that CVDs are frequent in patients with ITP at the time of ITP diagnosis; 75.3% had at least one CVD risk factor and among them 13.9% had diabetes, 18% had antiplatelet drugs and 39.2% had antihypertensive drugs [
      • Moulis G.
      • Germain J.
      • Comont T.
      • Arrouy A.
      • Lapeyre-Mestre M.
      • Adoue D.
      C.i. group, cardiovascular risk factors in immune thrombocytopenia adults: results from the CARMEN registry.
      ]. Enger et al. found an increased IRR of diabetes and CVD (IRR = 1.7), and kidney failure (IRR = 2.0) in patients with persistent or chronic ITP compared to a comparison cohort without ITP from a US claims database [
      • Enger C.
      • Bennett D.
      • Forssen U.
      • Fogarty P.F.
      • McAfee A.T.
      Comorbidities in patients with persistent or chronic immune thrombocytopenia.
      ]. A multicenter study from the Netherlands also reported an increased risk of AT and VTE in patients with kidney disease [
      • Ocak G.
      • Vossen C.Y.
      • Rotmans J.I.
      • Lijfering W.M.
      • Rosendaal F.R.
      • Parlevliet K.J.
      • Krediet R.T.
      • Boeschoten E.W.
      • Dekker F.W.
      • Verduijn M.
      Venous and arterial thrombosis in dialysis patients.
      ]. The association between cancer and VTE found in our study is in line with the results of other studies [
      • Blom J.W.
      • Vanderschoot J.P.
      • Oostindier M.J.
      • Osanto S.
      • van der Meer F.J.
      • Rosendaal F.R.
      Incidence of venous thrombosis in a large cohort of 66,329 cancer patients: results of a record linkage study.
      ,
      • Figueroa R.
      • Alfonso A.
      • Lopez-Picazo J.
      • Gil-Bazo I.
      • Garcia-Mouriz A.
      • Hermida J.
      • Paramo J.A.
      • Lecumberri R.
      Insights into venous thromboembolism prevention in hospitalized cancer patients: lessons from a prospective study.
      ,
      • Farge D.
      • Bounameaux H.
      • Brenner B.
      • Cajfinger F.
      • Debourdeau P.
      • Khorana A.A.
      • Pabinger I.
      • Solymoss S.
      • Douketis J.
      • Kakkar A.
      International clinical practice guidelines including guidance for direct oral anticoagulants in the treatment and prophylaxis of venous thromboembolism in patients with cancer.
      ]. The risk of a recurrent VTE is well documented [
      • Spencer F.A.
      • Emery C.
      • Joffe S.W.
      • Pacifico L.
      • Lessard D.
      • Reed G.
      • Gore J.M.
      • Goldberg R.J.
      Incidence rates, clinical profile, and outcomes of patients with venous thromboembolism. The Worcester VTE study.
      ,
      • Kyrle P.A.
      • Rosendaal F.R.
      • Eichinger S.
      Risk assessment for recurrent venous thrombosis.
      ] and an increased risk was found in both the French and the Swedish cohorts, with the strongest association found in the latter. There is a known risk of recurrent arterial events after a first arterial thrombosis. In our study, patients with a history of an arterial thrombosis did have an increased risk of recurrent arterial thrombosis. A history of a venous thrombosis also increases the risk of an arterial event [
      • Luxembourg B.
      • Schmitt J.
      • Humpich M.
      • Glowatzki M.
      • Dressler D.
      • Seifried E.
      • Lindhoff-Last E.
      Cardiovascular risk factors in idiopathic compared to risk-associated venous thromboembolism: a focus on fibrinogen, factor VIII, and high-sensitivity C-reactive protein (hs-CRP).
      ,
      • Sorensen H.T.
      • Horvath-Puho E.
      • Pedersen L.
      • Baron J.A.
      • Prandoni P.
      Venous thromboembolism and subsequent hospitalisation due to acute arterial cardiovascular events: a 20-year cohort study.
      ]. VTE and AT shared common risk factors such as high age and diabetes [
      • Franchini M.
      • Mannucci P.M.
      Venous and arterial thrombosis: different sides of the same coin?.
      ,
      • Prandoni P.
      Venous and arterial thrombosis: is there a link?.
      ,
      • Di Minno M.N.D.
      • Tufano A.
      • Ageno W.
      • Prandoni P.
      • Di Minno G.
      Identifying high-risk individuals for cardiovascular disease: similarities between venous and arterial thrombosis in perspective. A 2011 update.
      ].
      The findings of this study and previously published studies on risk of thrombosis in ITP patients suggest that the patients have increased risk of both bleeding and thromboembolic events. Although the mechanisms behind the risk of thromboembolic events in patients with ITP are to a large extent unknown, it is clear that a low platelet count does not protect against thrombosis [
      • Frederiksen H.
      • Maegbaek M.L.
      • Norgaard M.
      Twenty-year mortality of adult patients with primary immune thrombocytopenia: a Danish population-based cohort study.
      ,
      • Balitsky A.K.
      • Kelton J.G.
      • Arnold D.M.
      Managing antithrombotic therapy in immune thrombocytopenia: development of the TH2 risk assessment score.
      ]. In a recent study, Balitsky et al.proposed a “Thrombosis and Thrombocytopenia risk assessment score” to summarize the net risk of thrombosis and bleeding in patients with ITP, which can help clinicians with decision making when the patients have indication for treatment with anticoagulation. The study highlighted the importance of comorbidities, when considering risk of bleeding and thrombosis, in line with the findings of our study [
      • Balitsky A.K.
      • Kelton J.G.
      • Arnold D.M.
      Managing antithrombotic therapy in immune thrombocytopenia: development of the TH2 risk assessment score.
      ].

      4.4 What this study adds and future studies needed

      With data from two national databases we reported IRs for both AT and VTE which were higher in patients with ITP compared with IRs for the general populations. The results from this study confirms certain risk factors of thrombosis, including increasing age and male sex for AT. This has applications especially when treating the elderly and in patients with many comorbidities.

      4.5 Strengths

      Since the SNDS and the Swedish Health Registers have national coverage, the risk of selection bias should be a minor issue. Hospital visits and prescribed drugs dispensed from pharmacies are obliged to be registered in Sweden, and in France the reimbursement system covers the entire French population. To our knowledge this is the largest ITP study population analyzed using the same methods on harmonized data sets, including 7225 incident patients with ITP from France and 2490 from Sweden. With such a large dataset we were able to use multivariable models to adjust for important covariates yielding less biased estimates.
      There is also good consistency between data from France and Sweden. Both databases contain ICD-10 codes to detect hospital diagnoses and ATC-codes for drugs, and cover the same type of hospital visits and dispensed drugs. The quality of the data is good in both countries: in the French database the codes for ITP have been validated with the most recently reported PPV being 96% [
      • Moulis G.
      • Germain J.
      • Adoue D.
      • Beyne-Rauzy O.
      • Derumeaux H.
      • Sailler L.
      • Lapeyre-Mestre M.
      Validation of immune thrombocytopenia diagnosis code in the French hospital electronic database.
      ] [
      • Mezaache S.
      • Derumeaux H.
      • Ferraro P.
      • Capdepon P.
      • Steinbach J.C.
      • Abballe X.
      • Palas D.
      • Saichi N.
      • Desboeuf K.
      • Lapeyre-Mestre M.
      • Sailler L.
      • Moulis G.
      Validation of an algorithm identifying incident primary immune thrombocytopenia in the French national health insurance database.
      ],The outcomes are also validated separately with medical chart review, with a PPV of 83–99% [
      • Prat M.
      • Derumeaux H.
      • Sailler L.
      • Lapeyre-Mestre M.
      • Moulis G.
      Positive predictive values of peripheral arterial and venous thrombosis codes in French hospital database.
      ,
      • Giroud M.
      • Hommel M.
      • Benzenine E.
      • Fauconnier J.
      • Bejot Y.
      • Quantin C.
      • Study F.
      Positive predictive value of French hospitalization discharge codes for stroke and transient ischemic attack.
      ]. In Sweden there is an overall PPV of 85–95% for myocardial infarction and stroke where the codes have been validated separately with medical chart review [
      • Ludvigsson J.F.
      • Andersson E.
      • Ekbom A.
      • Feychting M.
      • Kim J.L.
      • Reuterwall C.
      • Heurgren M.
      • Olausson P.O.
      External review and validation of the Swedish national inpatient register.
      ]. Including data from separate countries increases the external validity of the study, and the detailed information about patient characteristics to analyze the impact of various risk factors is beneficial. There are consistent results between the two countries and the results are also consistent with clinical data.

      4.6 Limitations

      There were some unmeasured confounders and potential risk factors, which we were not able to measure in this study. Examples are information about known platelet counts, genetic risk factors for AT and VTE, and immobilization (a risk factor for VTE), smoking, and BMI (risk factors for both VTE and AT). Moreover, we adjusted for antihypertensive drugs, lipid lowering drugs and other drugs related to cardiovascular disease. Lack of laboratory data at the time of the event is also an important limitation. However, the focus of this study was to investigate the impact of baseline risk factors in a cohort of ITP patients. Primary ITP is a diagnosis of exclusion and some patients may be misdiagnosed, for example with myelodysplastic syndrome. However, in both France and Sweden, the diagnosis is given after an extensive list of exclusion criteria to rule out other causes of the thrombocytopenia and validation studies showed that a high proportion of records were the correct diagnosis [
      • Ludvigsson J.F.
      • Andersson E.
      • Ekbom A.
      • Feychting M.
      • Kim J.L.
      • Reuterwall C.
      • Heurgren M.
      • Olausson P.O.
      External review and validation of the Swedish national inpatient register.
      ,
      • Mezaache S.
      • Derumeaux H.
      • Ferraro P.
      • Capdepon P.
      • Steinbach J.C.
      • Abballe X.
      • Palas D.
      • Saichi N.
      • Desboeuf K.
      • Lapeyre-Mestre M.
      • Sailler L.
      • Moulis G.
      Validation of an algorithm identifying incident primary immune thrombocytopenia in the French national health insurance database.
      ]. >30% of the patients do not require any treatment for their ITP [
      • Rodeghiero F.
      • Besalduch J.
      • Michel M.
      • Provan D.
      • Grotzinger K.
      • Thompson G.
      Treatment practices in adults with chronic immune thrombocytopenia - a European perspective.
      ]. According to a validation study from France >95% of the patients with incident ITP had a hospital code of D69.3 at diagnosis [
      • Mezaache S.
      • Derumeaux H.
      • Ferraro P.
      • Capdepon P.
      • Steinbach J.C.
      • Abballe X.
      • Palas D.
      • Saichi N.
      • Desboeuf K.
      • Lapeyre-Mestre M.
      • Sailler L.
      • Moulis G.
      Validation of an algorithm identifying incident primary immune thrombocytopenia in the French national health insurance database.
      ] [
      • Moulis G.
      • Palmaro A.
      • Montastruc J.L.
      • Godeau B.
      • Lapeyre-Mestre M.
      • Sailler L.
      Epidemiology of incident immune thrombocytopenia: a nationwide population-based study in France.
      ].

      5. Conclusion

      For patients with primary ITP, the IR of AT and of VTE were of similar magnitude in France and in Sweden, and for both AT and VTE the risk was almost two times higher relative to a comparison population of similar age. As in the general population, the rate of AT was higher than that of VTE. Using multivariable methods to adjust for several risk factors simultaneously, age and male sex remained as the most important risk factors for AT and age for VTE. When treating patients with ITP, the occurrence of risk factors associated with the disease and with thrombosis, should be acknowledged.
      The following are the supplementary data related to this article.

      Statement of conflict of interest

      Charlotta Ekstrand, Shahram Bahmanyar, Helle Kieler and Marie Linder are employed at the Centre for Pharmacoepidemiology , which receive grants from several entities (pharmaceutical companies, regulatory authorities and contract research organizations) for performance of drug safety and drug utilization studies. None of them were involved in this study at any stage of the process.
      The French research group Bérangère Baricault Margaux Lafaurie Laurent Sailler Maryse Lapeyre-Mestre Guillaume Moulis had no conflict of interest to report regarding this study.

      Acknowledgements

      Sarah Burkill for language editing

      References

        • Rodeghiero F.
        • Stasi R.
        • Gernsheimer T.
        • Michel M.
        • Provan D.
        • Arnold D.M.
        • Bussel J.B.
        • Cines D.B.
        • Chong B.H.
        • Cooper N.
        • Godeau B.
        • Lechner K.
        • Mazzucconi M.G.
        • McMillan R.
        • Sanz M.A.
        • Imbach P.
        • Blanchette V.
        • Kuhne T.
        • Ruggeri M.
        • George J.N.
        Standardization of terminology, definitions and outcome criteria in immune thrombocytopenic purpura of adults and children: report from an international working group.
        Blood. 2009; 113: 2386-2393
        • Cines D.B.
        • Liebman H.A.
        The immune thrombocytopenia syndrome: a disorder of diverse pathogenesis and clinical presentation.
        Hematol. Oncol. Clin. North Am. 2009; 23: 1155-1161
        • George J.N.
        • Woolf S.H.
        • Raskob G.E.
        • Wasser J.S.
        • Aledort L.M.
        • Ballem P.J.
        • Blanchette V.S.
        • Bussel J.B.
        • Cines D.B.
        • Kelton J.G.
        • Lichtin A.E.
        • McMillan R.
        • Okerbloom J.A.
        • Regan D.H.
        • Warrier I.
        Idiopathic thrombocytopenic purpura: a practice guideline developed by explicit methods for the American Society of Hematology.
        Blood. 1996; 88: 3-40
        • Moulis G.
        • Germain J.
        • Comont T.
        • Brun N.
        • Dingremont C.
        • Castel B.
        • Arista S.
        • Sailler L.
        • Lapeyre-Mestre M.
        • Beyne-Rauzy O.
        • Godeau B.
        • Adoue D.
        • Group C.I.
        Newly diagnosed immune thrombocytopenia adults: clinical epidemiology, exposure to treatments, and evolution. Results of the CARMEN multicenter prospective cohort.
        Am. J. Hematol. 2017; 92: 493-500
        • Chandan J.S.
        • Thomas T.
        • Lee S.
        • Marshall T.
        • Willis B.
        • Nirantharakumar K.
        • Gill P.
        The association between idiopathic thrombocytopenic purpura and cardiovascular disease: a retrospective cohort study.
        J. Thromb. Haemost. 2018; 16: 474-480
        • Frederiksen H.
        • Maegbaek M.L.
        • Norgaard M.
        Twenty-year mortality of adult patients with primary immune thrombocytopenia: a Danish population-based cohort study.
        Br. J. Haematol. 2014; 166: 260-267
        • Peerschke E.I.
        • Yin W.
        • Ghebrehiwet B.
        Complement activation on platelets: implications for vascular inflammation and thrombosis.
        Mol. Immunol. 2010; 47: 2170-2175
        • Peerschke E.I.
        • Andemariam B.
        • Yin W.
        • Bussel J.B.
        Complement activation on platelets correlates with a decrease in circulating immature platelets in patients with immune thrombocytopenic purpura.
        Br. J. Haematol. 2010; 148: 638-645
        • Ruggeri M.
        • Tosetto A.
        • Palandri F.
        • Polverelli N.
        • Mazzucconi M.G.
        • Santoro C.
        • Gaidano G.
        • Lunghi M.
        • Zaja F.
        • De Stefano V.
        • Sartori R.
        • Fazi P.
        • Rodeghiero F.
        • Gruppo Italiano Malattie E.d.A.A.
        • G.S.I.T.P. Thrombocytopenias Working Party
        Thrombotic risk in patients with primary immune thrombocytopenia is only mildly increased and explained by personal and treatment-related risk factors.
        J. Thromb. Haemost. 2014; 12: 1266-1273
        • Boyle S.
        • White R.H.
        • Brunson A.
        • Wun T.
        Splenectomy and the incidence of venous thromboembolism and sepsis in patients with immune thrombocytopenia.
        Blood. 2013; 121: 4782-4790
        • Thomsen R.W.
        • Schoonen W.M.
        • Farkas D.K.
        • Riis A.
        • Fryzek J.P.
        • Sorensen H.T.
        Risk of venous thromboembolism in splenectomized patients compared with the general population and appendectomized patients: a 10-year nationwide cohort study.
        J. Thromb. Haemost. 2010; 8: 1413-1416
        • Rorholt M.
        • Ghanima W.
        • Farkas D.K.
        • Norgaard M.
        Risk of cardiovascular events and pulmonary hypertension following splenectomy - a Danish population-based cohort study from 1996-2012.
        Haematologica. 2017; 102: 1333-1341
        • Moulis G.
        • Lapeyre-Mestre M.
        • Palmaro A.
        • Pugnet G.
        • Montastruc J.L.
        • Sailler L.
        French health insurance databases: what interest for medical research?.
        La Revue de medecine interne. 2015; 36: 411-417
        • Bezin J.
        • Duong M.
        • Lassalle R.
        • Droz C.
        • Pariente A.
        • Blin P.
        • Moore N.
        The national healthcare system claims databases in France, SNIIRAM and EGB: powerful tools for pharmacoepidemiology.
        Pharmacoepidemiol. Drug Saf. 2017; 26: 954-962
      1. Socialstyrelsen, (2018).

        • Ludvigsson J.F.
        • Andersson E.
        • Ekbom A.
        • Feychting M.
        • Kim J.L.
        • Reuterwall C.
        • Heurgren M.
        • Olausson P.O.
        External review and validation of the Swedish national inpatient register.
        BMC Public Health. 2011; 11: 450
        • Wettermark B.
        • Hammar N.
        • Fored C.M.
        • Leimanis A.
        • Otterblad Olausson P.
        • Bergman U.
        • Persson I.
        • Sundstrom A.
        • Westerholm B.
        • Rosen M.
        The new Swedish Prescribed Drug Register--opportunities for pharmacoepidemiological research and experience from the first six months.
        Pharmacoepidemiol. Drug Saf. 2007; 16: 726-735
        • Wallerstedt S.M.
        • Wettermark B.
        • Hoffmann M.
        The first decade with the Swedish prescribed drug register - a systematic review of the output in the scientific literature.
        Basic Clin. Pharmacol. Toxicol. 2016; 119: 464-469
        • Mezaache S.
        • Derumeaux H.
        • Ferraro P.
        • Capdepon P.
        • Steinbach J.C.
        • Abballe X.
        • Palas D.
        • Saichi N.
        • Desboeuf K.
        • Lapeyre-Mestre M.
        • Sailler L.
        • Moulis G.
        Validation of an algorithm identifying incident primary immune thrombocytopenia in the French national health insurance database.
        Eur. J. Haematol. 2017; 99: 344-349
        • Thai L.H.
        • Mahevas M.
        • Roudot-Thoraval F.
        • Limal N.
        • Languille L.
        • Dumas G.
        • Khellaf M.
        • Bierling P.
        • Michel M.
        • Godeau B.
        Long-term complications of splenectomy in adult immune thrombocytopenia.
        Medicine (Baltimore). 2016; 95e5098
        • Brazauskas R.
        • Logan B.R.
        Observational studies: matching or regression?.
        Biol. Blood Marrow Transplant. 2016; 22: 557-563
        • Guo Z.
        • Gill T.M.
        • Allore H.G.
        Modeling repeated time-to-event health conditions with discontinuous risk intervals. An example of a longitudinal study of functional disability among older persons.
        Methods Inf. Med. 2008; 47: 107-116
        • Norgaard M.
        • Cetin K.
        • Maegbaek M.L.
        • Kristensen N.R.
        • Ghanima W.
        • Bahmanyar S.
        • Stryker S.
        • Christiansen C.F.
        Risk of arterial thrombotic and venous thromboembolic events in patients with primary chronic immune thrombocytopenia: a Scandinavian population-based cohort study.
        Br. J. Haematol. 2016; 174: 639-642
        • Langeberg W.J.
        • Schoonen W.M.
        • Eisen M.
        • Gamelin L.
        • Stryker S.
        Thromboembolism in patients with immune thrombocytopenia (ITP): a meta-analysis of observational studies.
        Int. J. Hematol. 2016; 103: 655-664
        • Yusuf H.R.
        • Hooper W.C.
        • Grosse S.D.
        • Parker C.S.
        • Boulet S.L.
        • Ortel T.L.
        Risk of venous thromboembolism occurrence among adults with selected autoimmune diseases: a study among a U.S. cohort of commercial insurance enrollees.
        Thromb. Res. 2015; 135: 50-57
        • Ferlay J.
        • Steliarova-Foucher E.
        • Lortet-Tieulent J.
        • Rosso S.
        • Coebergh J.W.
        • Comber H.
        • Forman D.
        • Bray F.
        Cancer incidence and mortality patterns in Europe: estimates for 40 countries in 2012.
        Eur. J. Cancer. 2013; 49: 1374-1403
        • Moulis G.
        • Germain J.
        • Comont T.
        • Arrouy A.
        • Lapeyre-Mestre M.
        • Adoue D.
        C.i. group, cardiovascular risk factors in immune thrombocytopenia adults: results from the CARMEN registry.
        Am. J. Hematol. 2018; 93: E159-E182
        • Enger C.
        • Bennett D.
        • Forssen U.
        • Fogarty P.F.
        • McAfee A.T.
        Comorbidities in patients with persistent or chronic immune thrombocytopenia.
        Int. J. Hematol. 2010; 92: 289-295
        • Ocak G.
        • Vossen C.Y.
        • Rotmans J.I.
        • Lijfering W.M.
        • Rosendaal F.R.
        • Parlevliet K.J.
        • Krediet R.T.
        • Boeschoten E.W.
        • Dekker F.W.
        • Verduijn M.
        Venous and arterial thrombosis in dialysis patients.
        Thromb. Haemost. 2011; 106: 1046-1052
        • Blom J.W.
        • Vanderschoot J.P.
        • Oostindier M.J.
        • Osanto S.
        • van der Meer F.J.
        • Rosendaal F.R.
        Incidence of venous thrombosis in a large cohort of 66,329 cancer patients: results of a record linkage study.
        J. Thromb. Haemost. 2006; 4: 529-535
        • Figueroa R.
        • Alfonso A.
        • Lopez-Picazo J.
        • Gil-Bazo I.
        • Garcia-Mouriz A.
        • Hermida J.
        • Paramo J.A.
        • Lecumberri R.
        Insights into venous thromboembolism prevention in hospitalized cancer patients: lessons from a prospective study.
        PLoS One. 2018; 13
        • Farge D.
        • Bounameaux H.
        • Brenner B.
        • Cajfinger F.
        • Debourdeau P.
        • Khorana A.A.
        • Pabinger I.
        • Solymoss S.
        • Douketis J.
        • Kakkar A.
        International clinical practice guidelines including guidance for direct oral anticoagulants in the treatment and prophylaxis of venous thromboembolism in patients with cancer.
        Lancet Oncol. 2016; 17: E452-E466
        • Spencer F.A.
        • Emery C.
        • Joffe S.W.
        • Pacifico L.
        • Lessard D.
        • Reed G.
        • Gore J.M.
        • Goldberg R.J.
        Incidence rates, clinical profile, and outcomes of patients with venous thromboembolism. The Worcester VTE study.
        J. Thromb. Thrombolysis. 2009; 28: 401-409
        • Kyrle P.A.
        • Rosendaal F.R.
        • Eichinger S.
        Risk assessment for recurrent venous thrombosis.
        Lancet. 2010; 376: 2032-2039
        • Luxembourg B.
        • Schmitt J.
        • Humpich M.
        • Glowatzki M.
        • Dressler D.
        • Seifried E.
        • Lindhoff-Last E.
        Cardiovascular risk factors in idiopathic compared to risk-associated venous thromboembolism: a focus on fibrinogen, factor VIII, and high-sensitivity C-reactive protein (hs-CRP).
        Thromb. Haemost. 2009; 102: 668-675
        • Sorensen H.T.
        • Horvath-Puho E.
        • Pedersen L.
        • Baron J.A.
        • Prandoni P.
        Venous thromboembolism and subsequent hospitalisation due to acute arterial cardiovascular events: a 20-year cohort study.
        Lancet. 2007; 370: 1773-1779
        • Franchini M.
        • Mannucci P.M.
        Venous and arterial thrombosis: different sides of the same coin?.
        Eur. J. Intern. Med. 2008; 19: 476-481
        • Prandoni P.
        Venous and arterial thrombosis: is there a link?.
        Adv. Exp. Med. Biol. 2017; 906: 273-283
        • Di Minno M.N.D.
        • Tufano A.
        • Ageno W.
        • Prandoni P.
        • Di Minno G.
        Identifying high-risk individuals for cardiovascular disease: similarities between venous and arterial thrombosis in perspective. A 2011 update.
        Intern. Emerg. Med. 2012; 7: 9-13
        • Balitsky A.K.
        • Kelton J.G.
        • Arnold D.M.
        Managing antithrombotic therapy in immune thrombocytopenia: development of the TH2 risk assessment score.
        Blood. 2018; 132: 2684-2686
        • Moulis G.
        • Germain J.
        • Adoue D.
        • Beyne-Rauzy O.
        • Derumeaux H.
        • Sailler L.
        • Lapeyre-Mestre M.
        Validation of immune thrombocytopenia diagnosis code in the French hospital electronic database.
        Eur. J. Intern. Med. 2016; 32: e21-e22
        • Prat M.
        • Derumeaux H.
        • Sailler L.
        • Lapeyre-Mestre M.
        • Moulis G.
        Positive predictive values of peripheral arterial and venous thrombosis codes in French hospital database.
        Fundam. Clin. Pharmacol. 2018; 32: 108-113
        • Giroud M.
        • Hommel M.
        • Benzenine E.
        • Fauconnier J.
        • Bejot Y.
        • Quantin C.
        • Study F.
        Positive predictive value of French hospitalization discharge codes for stroke and transient ischemic attack.
        Eur. Neurol. 2015; 74: 92-99
        • Rodeghiero F.
        • Besalduch J.
        • Michel M.
        • Provan D.
        • Grotzinger K.
        • Thompson G.
        Treatment practices in adults with chronic immune thrombocytopenia - a European perspective.
        Eur. J. Haematol. 2010; 84: 160-168
        • Moulis G.
        • Palmaro A.
        • Montastruc J.L.
        • Godeau B.
        • Lapeyre-Mestre M.
        • Sailler L.
        Epidemiology of incident immune thrombocytopenia: a nationwide population-based study in France.
        Blood. 2014; 124: 3308-3315