Analysis of nationwide hemophilia care: A cohort study using two Japanese healthcare claims databases

Abstract Background and aims In many developed countries, hemophilia care is provided by specialized centers which can offer standardized high‐quality care for patients and collect data for patient registries. However, in countries with less centralized provision of hemophilia care, registry data lacks accuracy and medical care is inconsistent among providers. Claims databases can be an alternative for obtaining nationwide data on hemophilia care, and we applied this approach to evaluate inequalities in hemophilia care in Japan. Methods Medical records of hemophilia A patients were collected by a combination of ICD‐10 code (D66) and prescribed coagulation factors from two major Japanese claims databases (JMDC and Medical Data Vision [MDV]). Patient records with an anti‐inhibitor coagulant complex were excluded. Based on the annual number of hemophilia A patients, medical facilities were categorized into specialized facilities (SP, ≥5 patients) and nonspecialized facilities (N‐SP, <5 patients). Patient age, comorbidities, diagnostic testing, prescribed drugs and their dosages were compared between facility types. Results The JMDC and MDV databases included 274 and 1266 hemophilia A patients, respectively. In the MDV database, SP facilities prescribed extended half‐life factor VIII (FVIII) products for more patients (31.8% vs 24.3%) than N‐SP. The mean annual FVIII consumption per patient was higher in SP facilities (240 333 IU [international units] vs 210 334 IU), and the mean FVIII dosage was higher in SP facilities for all types of FVIII products. The proportion of patients who received diagnostic blood tests was higher in SP (75.7% vs 56.2%). Conclusion The MDV database revealed disparities in hemophilia A care between SP and N‐SP facilities in types of FVIII products prescribed, FVIII consumption, and frequency of the relevant management such as blood tests. Claims databases can be an alternative for the assessment of nationwide hemophilia care patterns in countries without a well‐established registry.


| INTRODUCTION
Hemophilia A is an X-linked hereditary disorder that results from a deficiency or dysfunction of the factor VIII (FVIII) coagulation protein, causing recurrent joint and muscle bleeds and leading to progressive musculoskeletal damage. 1 Estimates of the global prevalence for all severities is 17.1 cases per 100 000 males and 6.0 per 100 000 males for severe cases. 2 In Japan, a survey conducted in 2018 reported an estimated 8751 cases of coagulation disorders, of which 5301 involved hemophilia A. 3 According to current guidelines of the World Federation of Hemophilia, the standard of care for hemophilia A patients is treatment with FVIII concentrates 4 ; both recombinant FVIII (rFVIII) or plasma-derived FVIII (pdFVIII) products are available, and factor replacement therapy may be administered in an episodic or prophylactic manner. 4 The efficacy of regular prophylactic rFVIII infusions for the prevention of joint damage and other hemorrhages is supported with high-quality evidence from a randomized, open-label study. 5 Multiple standard half-life (SHL) rFVIII products are now available, but they need intravenous administration as often as every other day. 6 Protein conjugation, chemical modification or protein sequence modification has been used to develop extended half-life (EHL) rFVIII products to satisfy patient expectations of less frequent injections. [7][8][9][10] Emicizumab-a bispecific antibody that replaces the function of missing activated FVIII thereby restoring hemostasis-is a novel, non-factor treatment for hemophilia A. 11 Emicizumab received licensing approval for treatment of hemophilia A with or without inhibitors in Japan in 2018.
In the majority of high-income countries, patients with hemophilia A have access to specialized hemophilia care centers that offer a full range of options for management of bleeding disorders including prescription of coagulation factors, monitoring of hemostatic parameters, consultation on prophylaxis, medical management of joint health and information on new treatments and clinical trials. 12 In a Japanese survey, 186 out of 290 hospitals provided care for fewer than five blood coagulation disorder patients annually. 3 Although these data are not limited to hemophilia, they imply that many patients are receiving treatment in nonspecialized medical facilities. To expand on these findings, we sought to better understand the current patterns in the treatment and management of hemophilia A in Japan and conducted a retrospective claims-based cohort study of two large patient databases. The specific objectives were to determine whether there are any differences in the provision of care for patients with hemophilia A between specialized and nonspecialized hospitals and clinics (defined as those treating ≥5 and <5 hemophilia A patients per year, respectively), with respect to the therapeutic agent selection and dose.

| Study design
This study was a retrospective, observational database analysis to evaluate the patterns of treatment in Japanese patients with hemophilia A. Data derived from anonymized reimbursements were extracted from the JMDC (formerly known as Japan Medical Data Centre) and Medical Data Vision (MDV) databases between April 1, 2010 and June 30, 2019 (the selection period). Patients were enrolled into two cohorts with distinct study populations to separately explore Objectives 1 (therapeutic agent selection) and 2 (annual cumulative dose; Figure 1A,B, respectively).
The patient cohort for Objective 1 (ie, source cohort) were those with at least one record of receiving hemophilia A treatment based on an International Classification of Diseases, Tenth Revision (ICD-10) diagnostic code of D66 during the selection period. The index date was the earliest of the first hereditary FVIII deficiency diagnosis or the first recorded claim for prescription of coagulation factors. For Objective 2 (ie, derived cohort), patients from the source cohort with at least 1 year of continuous prescriptions for a hemophilia A treatment without a gap of 90 days or more were selected and the most recent, continuous 1-year treatment episode was analyzed. The index date was the date of the first recorded hemophilia A treatment prescription in the most recent treatment episode.

| Data sources
The JMDC and MDV databases were used to capture patient characteristics and prescription claims data. JMDC is a payer-based administrative database that maintains inpatient and outpatient medical and pharmacy claims, and enrolment information for salaried workers and their families for the Healthcare Insurance Association (Table S1)

| Patients
This study analyzed two distinct cohorts; for Objective 1, the source cohort included male patients with one or more inpatient or outpatient visits for hemophilia A, based on the use of the D66 hereditary  Japan that suggested that non-specialized clinics (defined as fewer than five patients annually) may provide sub-optimal treatment. 13 The possibility that facilities with no specialized physicians may provide care for some patients, and a preliminary analysis of the databases that found a high proportion of patients were treated in clinics with fewer than five patients annually, also informed this decision. For both the JMDC and MDV databases, the number of patients who were prescribed pdFVIII, SHL rFVIII, EHL rFVIII or non-factor treatment by facility type per fiscal year was recorded.

| Outcomes
For Objective 2, patients were evaluated for related comorbidities and hematological tests. Complications specific to hemophilia A were major hemophilia bleeding (major bleeding, joint bleeding, muscular hemorrhage, intracerebral hemorrhage, coagulopathy, purpura, hematoma and other bleeding conditions), liver disease (chronic viral hepatitis, cirrhosis, liver cancer), joint lesions (hemophilia arthropathy and hemophilia arthritis, synovitis), and thrombotic disorders (cerebral infarction, blood deficiency heart disease).

| Data analysis
Baseline characteristics were analyzed using descriptive statistics, with quantitative variables described according to mean and standard deviation (SD), and categorical variables summarized according to the F I G U R E 1 Study design for Objective 1 (A) and Objective 2 (B). (A) Study design for Objective 1. The prescribed medicine in each index year was counted and categorized. Block arrows represent: (Patient 1) a patient switching from pdFVIII to EHL rFVIII; (Patient 2) a patient treated with EHL with a break in continuous care; (Patient 3) a patient treated continuously with pdFVIII; (Patient 4) a patient receiving non-factor (ie, emicizumab) treatment. (B) Study design for Objective 2. In the selection period 2014-2019, the most recent continuously prescribed hemophilia agent (≥1 year, without 90-day gap) after index date was longitudinally analyzed for annual cumulative dose and the medical history of the patient in the pre-index period was collected. Index date was defined as the first recorded treatment in most recent episode. Patient 4 is excluded because the duration of continuous treatment was less than 1 year. EHL, extended half-life; FVIII, factor VIII; pdFVIII, plasma-derived FVIII; rFVIII, recombinant FVIII; SHL, standard half-life number and percentage of patients in each category. For the JMDC database, an algorithm was developed to handle missing data and consisted of approximating the exact date of a claim using the first date of events attached to a claim when available. The full date of a claim was imputed using the first full date among dates of procedures, prescriptions or admissions attached to the claim. No data imputation was planned for the MDV database, and no data imputation was performed for other study variables in either database. Drug categories were calculated inclusively, that is, patients receiving multiple blood coagulation factors of interest were counted several times. For patients in whom all prescriptions of the most recent treatment episode were recorded within the 1-year period, the annual cumulative dose (Objective 2) was calculated according to the following formula: For patients with prescriptions outside this period (ie, earlier prescriptions), the cumulative dose was adjusted according to prescription duration inside and outside the 1-year period, according to the following formula: The mean dose per treatment was calculated as the mean annual cumulative dose divided by the product of 52 weeks multiplied by the frequency of administration. We set the precondition as for the frequency of administration three times weekly for pdFVIII and SHL rFVIII, and twice weekly for EHL rFVIII here based on general information in the package inserts. Data management was performed using SAS software, version 9.3.

| Ethical considerations
This study was based on anonymized secondary use data; an ethical review was not considered necessary according to the local ethical guidelines, and such a review was not performed.

| Patient disposition
Patient disposition is summarized in Figure 2.

| Objective 1: Therapeutic selection
In  reported only in a small number of patients in specialized medical facilities.  (Table 3). In the MDV data, the similar age distribution was observed in both specialized medical facilities and nonspecialized medical facilities ( Figure S1), suggesting the age distribution between specialized and non-specialized facilities was almost the same, and therefore unlikely to explain differences between the facility types (bodyweight data were not available in the databases).

| Other findings
Data from the derived cohorts were used to evaluate rates of medical  Table 2). The proportion of patients with comorbidities was similar between specialized and nonspecialized medical facilities for most comorbidity types (Tables 1 and 2), except for liver disease and hyperlipidemia, which were higher among patients in specialized medical facilities.

| DISCUSSION
The findings of this claims-based retrospective cohort study suggest that there are opportunities to improve the management of patients with hemophilia A in Japan. Although the "Hemophilia Center" con-  was conducted in fewer than 60% of patients who were treated at nonspecialized medical facilities, compared with 75% of patients at specialized medical facilities. The absence of hematological testing in many patients raises concern as testing is critical, not only for screening and diagnosis of hemophilia A, but also for monitoring response to treatment. 14,15 The 2013 Japanese Society on Thrombosis and Hemostasis guidelines recommend patients undergo hematological testing to ensure plasma FVIII levels meet the required thresholds, set appropriate dosing and frequency of prophylactic treatment, and detect the presence of inhibitors. 16 These deficiencies could be addressed by emphasizing the use of the current Japanese guidelines that promote the provision of care in specialized medical facilities and highlight the importance of regular testing. Awareness and educational programs aimed at physicians in nonspecialized medical facilities who treat hemophilia A patients can also contribute to this goal.

| Continuity of hemophilia care
In evaluating patient disposition, the large proportion of hemo-

| CONCLUSIONS
Our data suggest that many Japanese patients with hemophilia A may receive suboptimal care, with a high proportion of patients not partaking in continuous prophylaxis, and a lack of hematological testing among those who do receive care. Patients are treated predominantly with rFVIII products, but many are treated with pdFVIII, particularly in a nonspecialized setting. We suggest emphasizing the current local guidelines and educational programs, especially in nonspecialized facilities, that highlight the importance of directing patients to receive care in specialized centers. Although this study demonstrates the limitations of using a claims database to monitor hemophilia care, it also shows that such an approach can yield insights into nationwide hemophilia care that would otherwise be unavailable.