Acquired (autoimmune) hemophilia: demographics, outcomes, and readmissions

Acquired hemophilia (AHA) is a rare acquired bleeding disorder with an incidence of 1.5 cases per million population per year. It is most commonly caused by the production of inhibitory autoantibodies against factor VIII of the coagulation pathway.¹ It is a disease of the older population, with a median age of onset of >65 years. A small percentage of cases have also been reported in young women during pregnancy or the postpartum period. More than half of AHA cases are idiopathic. In the remaining half, most have a history of autoimmune disorder or malignancy.² Other causes that can lead to AHA, as reported in the literature, include infections, drugs (commonly reported with penicillins, interferons, and clopidogrel), and recently coronavirus disease 19 (COVID-19) infection and vaccination.^3-5 

Treatment of AHA mainly consists of 2 pathophysiological principles. The first is achieving hemostasis with the use of bypassing agents such as activated prothrombin complex concentrate, recombinant factor VIIa, and, most recently, emicizumab. The second is the use of immunosuppression to eradicate neutralizing autoantibodies. Treatment of the underlying etiology, if identified, also forms 1 of the most important components of the treatment compendium of AHA.^6,7 Awareness of the treatment-related complications, including but not limited to thrombosis, recurrent bleeding diathesis, and infections, is also of paramount importance, especially in older populations with multiple other medical comorbidities. Although there have been significant developments in the treatment aspects, not much literature exists on the epidemiology and outcomes of these treatments on a larger population basis. Through this nationwide population-based retrospective study, we aimed to shed light on the epidemiology of AHA, including readmission rates and causes of readmission.

We conducted a retrospective cohort study using the Nationwide Readmissions Database (NRD) from 1 January 2016 to 30 November 2019, excluding December data each year to estimate 30-day readmissions. The NRD is an administrative database developed by the Agency for Healthcare Research and Quality for the Healthcare Cost and Utilization Project.⁸ This large, publicly available database contains various patient- and hospital-level information on ∼35 million hospital discharges after applying weights. However, the NRD does not include granular clinical details, such as medications administered, laboratory findings, or vital signs, nor does it provide the temporal precision needed to differentiate between complications that occur during hospitalization and chronic conditions present at admission. These characteristics of the data set are important to consider when interpreting study findings. The database used in the study contains deidentified information, thus this study was exempt from review by the institutional review board. This study was conducted in accordance with the Declaration of Helsinki.

AHA admissions were identified using the ICD-10-CM (International Classification of Diseases, 10th Revision, Clinical Modification) code D68.311. Descriptive analysis was performed to report on patient demographics (age, sex, and primary expected payer), hospital characteristics (teaching status and urban-rural designation), Charlson Comorbidity Index, various comorbid conditions (autoimmune disease, hematologic malignancy, solid malignancy, coronary artery disease, diabetes mellitus, liver disease, dyslipidemia, tobacco use disorder, and obesity), pregnancy, and complications (disseminated intravascular coagulation, ischemic stroke, intracranial hemorrhage, venous thromboembolism, septic shock, acute kidney injury, acute myocardial infarction, acute respiratory failure, and gastrointestinal bleeding) in AHA admissions. The use of blood transfusion and invasive mechanical ventilation was also identified, in addition to reporting length of stay, deaths during hospitalization, disposition of patients, and total charges incurred during admission. Comorbidities and complications were identified using ICD-10-CM codes and the Charlson Comorbidity Index.

Hospitalizations were categorized into 2 age groups: those aged <65 years and those aged ≥65 years, to study differences in in-hospital mortality, sex distribution, underlying comorbidities, and complications. We used the Pearson χ² test to obtain percentages for categorical variables and the epctile (estimation and inference for percentiles) command for calculating weighted medians of continuous variables with an interquartile range (IQR). The all-cause 30-day readmission rate was calculated, and the primary diagnoses at readmission were also identified. The analysis was conducted after applying weights and using the methodology provided by the Healthcare Cost and Utilization Project.

Of 1450 admissions for AHA between 2016 and 2019 in the United States, 803 (55.4%) were male, and 857 (59.1%) belonged to the age group of 61 to 80 years, with a median age at admission of 73 years. Furthermore, most were admitted to a teaching hospital (83.5%), with the highest quintile of AHA-specific hospital volumes (43.8%) in a large metropolitan area with at least 1 million residents (59.8%).

Although 20.9% had an underlying solid malignancy, 28.5% had an otherwise high comorbidity burden, with a Charlson Comorbidity Index of >3. Underlying hematologic malignancy was reported in 3.9% of admissions, and 13.5% had a diagnosis of an autoimmune disease. Autoimmune disease was more prevalent in females than in males (22.2% vs 6.4%; P < .05), with no significant sex difference observed for cancer (P = .19).

Regarding complications, acute myocardial infarction occurred in 9.5%, disseminated intravascular coagulation was noted in 1.2%, intracranial hemorrhage in 1.5%, ischemic stroke in 2.5%, and venous thromboembolism in 4.4%, with 30.2% of admissions experiencing some form of bleeding. A total of 27% of admissions received blood transfusions. The proportion of patients who received blood transfusions was not significantly different between those with acute myocardial infarction and those without (29% vs 27%; P = .706). The median length of stay was 7 days and 101 deaths occurred during the index admission, resulting in a 7% inpatient mortality rate.

When admissions were categorized by age group, 1089 were aged ≥65 years, and 361 were aged <65 years. The proportion of females was higher, and underlying autoimmune disease was more prevalent in the older age group compared with the younger age group (47.8% vs 35%; P = .01 and 15.1% vs 8.5%; P = .05, respectively). However, there was no significant difference in the prevalence of cancer between the 2 age groups (P = .14).

The median time to death was 9 days (IQR, 4-19). Mortality was higher in the older age group at 8% compared with 3% in the younger cohort (P = .03). Mortality was also higher in admissions involving ischemic stroke or venous thromboembolic events compared with those without these complications (34.4% vs 6.3%; P < .05 and 19.6% vs 6.4%; P = .02, respectively). However, mortality did not differ with the presence or absence of underlying autoimmune disease (P = .11) or malignancy (P = .96).

Of 1349 alive discharges, 371 were readmitted within 30 days, with a 30-day readmission rate of 27% (Table 1). Readmission rates were higher in those with underlying autoimmune disease than those without (35.5% vs 23.3%; P = .02), whereas the diagnosis of cancer had no impact on readmissions (P = .61). The median time to readmission was 10 days (IQR, 5-18), and the median length of stay during readmission was 7 days (IQR, 3-14), with 10.8% mortality during readmission. The median hospitalization charges were $113 008 for both index admissions and readmissions.

After omitting readmissions coded as hemophilia (a designation lacking sufficient granularity to specify the exact reason for readmission, such as bleeding, treatment-related complications, or other issues), infection (30.8%), followed by bleeding (28.2%), was the most common cause for readmission. Additional reasons for readmissions included cardiovascular events such as arrhythmias and heart failure, observed in 7.6% of cases, followed by malignancy (primarily lung cancer) in 5.7%. Fractures, predominantly of the long bones and vertebrae, accounted for 5.6% of readmissions, whereas neurologic events, including seizures and syncope, were identified in 4.6% of readmissions (Figure 1).

To our knowledge, this study represents the largest collection of AHA admissions in the United States. The median age for admissions was 73 years (IQR, 65-79), which is similar to previously reported results from the largest European Acquired Hemophilia Registry (EACH-2).⁹ The proportion of women was higher in the >65 years age group (47.8% vs 35%), similar to EACH-2 findings.⁹ In our study, only 0.9% of admissions were pregnant patients, which is lower than previously published reports.^10,11 Most of the pregnancy-related AHA occurs in the postpartum period¹² and given the limitation of the database, we could not capture postpartum events separately in our study.

In our study, 13.5% of admissions had autoimmune diseases whereas 24.8% had malignancies. The rate of admissions with malignancies is higher in our study than previously published literature.⁹ The most likely explanation for the higher rate of malignancies in our study is our inability to differentiate whether all these admissions had active malignancies or a history of malignancies.

Bleeding is the most common presentation for AHA admissions. In our study, only 30.2% of admissions were reported to have bleeding. The likely reason for this is a lack of documentation of minor bleeding events as a diagnosis in-hospital admissions. Nevertheless, 27% of admissions required blood transfusions, indicating the severity of bleeding in patients with AHA.

Although bleeding events are a striking feature of this entity, thrombotic events are well-known to occur with the treatment protocols used for these bleeding diatheses.¹³ Both venous and arterial thromboembolic events are an important consideration in the management of these patients. In our study, 17.2% of total admissions reported a thromboembolic event (either arterial or venous). Acute myocardial infarction was the most common thrombotic event, occurring in 10% of admissions. It is beyond the scope of this study to accurately differentiate between acute myocardial infarction resulting from hemorrhagic shock vs a true thrombotic complication. However, the lack of a significant difference in the rate of blood transfusions between patients who had acute myocardial infarction and those who did not suggests that it was likely a thrombotic complication.

One of the most important aspects of this study was the high readmission rates. More than one-quarter of total hospital discharges were readmitted, with a median time to readmission of 10 days. Infection was the most common cause of readmission, followed by bleeding complications. These results highlight the complex challenges in managing this disease. High infection rates emphasize the toxic effects of immunosuppressive therapy, whereas higher bleeding rates demonstrate inadequate inhibitor clearance

It was also found that patients with autoimmune diseases have higher chances of readmission. This could be secondary to increased risk of complications from immunosuppression in this already immunocompromised patient population. Another significant finding in our study is that 5.6% of readmissions involved bone fractures. This may be explained by the increased risk of osteoporosis because of prolonged use of immunosuppressive medications, particularly in older patients and those who are frail, as well as an elevated fall risk caused by neurocardiovascular events during the index admission and deconditioning from extended hospitalizations. For patients who require longer durations of steroid therapy, initiating bone-strengthening agents may help mitigate the risk of fractures over time. Additionally, early consultations for physical and occupational therapy should be considered to address fall risk and improve overall mobility. Future studies should investigate the effectiveness of such preventive strategies in this patient population.

The crude mortality rate in this study is 7% during the index hospital admission, with an additional 10.8% mortality for 30-day readmissions. Mortality was higher with older age, ischemic stroke, and venous thromboembolic complications, suggesting the importance of finding a balance with hemostatic agents that is sufficient to minimize bleeding without increasing thrombotic complications. Interestingly, a history of autoimmune disease and malignancy were not associated with increased mortality. Multiple previous studies have reported higher mortality in AHA admissions with malignancies.^14,15 The most likely reason for this discrepancy in our study is the inability to differentiate active malignancy from admissions with a history of malignancy now in remission.

The observed mortality rates in our study align with data from the China Acquired Hemophilia Registry, which reported a 6.7% mortality rate at a median follow-up of 205 days.¹⁶ However, the EACH2 registry reported a higher mortality rate of 26.3% at a median follow-up of 258 days. Notably, the 1-year mortality rate in the EACH2 cohort was <10%, as illustrated by the survival curves.⁹ The variation in mortality rates across studies likely reflects differences in follow-up durations and study designs. Our study specifically focuses on in-hospital mortality during the index admission and within 30 days after discharge, whereas the EACH2 and China Acquired Hemophilia Registry cohorts examined long-term outcomes over extended follow-up periods. It is also important to acknowledge potential ascertainment errors that may contribute to variations in reported mortality rates. Patients who die without a confirmed diagnosis of AHA are not captured in the data set, potentially leading to an underestimation of mortality rates in our study.

Although ICD-10 codes were meticulously used to identify these admissions, inherent flaws in using an administrative database, such as misclassification bias, remain. Furthermore, ICD-10 codes specific to AHA have not undergone external validation, either nationally or at our institution. Although the use of administrative databases such as the NRD is well established in the medical literature, particularly for rare diseases for which institution-specific studies may not be feasible, this lack of validation emphasizes the need for caution when interpreting our findings. It also highlights the importance of future efforts to validate diagnostic codes for rare conditions to enhance the reliability of such studies. Other significant limitations include a lack of detailed treatment information, such as administered medications, laboratory values, and vital signs, which are not captured in the data set. Additionally, the data set does not allow for the ascertainment of time-to-event data, making it challenging to precisely differentiate between complications arising during hospitalization and chronic conditions present at admission. This limitation may affect the interpretation of valuable findings, especially for thrombotic and hemorrhagic complications. Despite these constraints, this large collection of AHA admissions and readmissions on a nationwide scale has generated a valuable pool of data, helping to characterize the clinical outcomes and health care use of this otherwise rare hematological entity.

The study highlights a high rate of thrombotic complications in AHA, necessitating a careful approach to factor replacement for achieving hemostasis. Moreover, bleeding and infections are the primary reasons for the high 30-day readmission rate of 27%, with an additional 10.8% mortality during readmissions. This indicates a significant incidence of treatment-related toxicity and disease relapse.

Contribution: A.S., N.V., and V.S. made substantial contributions to the acquisition, analysis, or interpretation of data for this study; drafted the manuscript or revised it critically for important intellectual content; provided final approval of the version to be published; and have agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Aditi Sharma, Barbara Ann Karmanos Cancer Institute, 4100 John R, Detroit, MI 48201; email: [email protected].