Associations between intracranial haemorrhage and prescribed prophylaxis in a large cohort of haemophilia patients in the United States
Char Witmer, MD, The Children’s Hospital of Philadelphia, 3501 Civic Center Blvd, Division of Hematology, CTRB 11th Floor, Philadelphia, PA 19104, USA.
Intracranial haemorrhage (ICH) is the most serious type of bleeding for patients with haemophilia. Prior published reports regarding ICH predate the widespread provision of prophylaxis. Our study objectives were to determine risk factors for ICH and whether prophylaxis reduces ICH occurrence. We performed a nested case-control study of persons with haemophilia, ≥2 years of age enrolled in the Centers for Disease Control and Prevention Universal Data Collection project. Of 10 262 patients 199 (1·9%) experienced an ICH for an incidence rate of 390/105 patient years. Head trauma was reported in 44% (88/199). ICH mortality was 19·6% (39/199). Significant risk factors for ICH included a high titre inhibitor [odds ratio (OR) = 4·01, 95% confidence interval (2·40–6·71)], prior ICH [OR = 3·62 (2·66–4·92)] and severe haemophilia [OR = 3·25 (2·01–5·25)]. Prophylaxis was associated with a significant risk reduction for ICH occurrence in patients with severe haemophilia who were negative for human immunodeficiency virus or an inhibitor, with an OR of 0·52 (0·34–0·81) and 0·50 (0·32–0·77) respectively. The most significant risk factors for ICH included the presence of an inhibitor, prior ICH, severity of haemophilia and reported head trauma. This is the first study to demonstrate that prescribed prophylaxis conferred a protective effect against ICH in patients with uncomplicated severe disease.
While joint disease is the most common complication of haemophilia, intracranial haemorrhage (ICH) accounts for the highest number of deaths from bleeding, with 20% mortality (Chorba et al, 1994; Nuss et al, 2001; Stieltjes et al, 2005). ICH is also a significant cause of long term disability with morbidity reported in 29–76% of patients (Andes et al, 1984; de Tezanos Pinto et al, 1992; Klinge et al, 1999; Stieltjes et al, 2005; Traivaree et al, 2007). The morbidities include seizures, psychomotor impairment, statomotor impairment, and a decreased quality of life (Eyster et al, 1978; de Tezanos Pinto et al, 1992; Klinge et al, 1999; Revel-Vilk et al, 2004; Stieltjes et al, 2005). Persons with haemophilia are 20–50 times more likely to develop ICH than those without haemophilia, with a reported prevalence of 2·7–12% and an incidence rate of 290–748/105 patient years (Nuss et al, 2001; Antunes et al, 2003; Stieltjes et al, 2005; Traivaree et al, 2007; Ljung, 2008).
Prior published reports regarding ICH predate the widespread provision of prophylaxis. Prophylaxis, the routine administration of factor concentrate, is recommended for patients with severe haemophilia and effectively prevents joint haemarthrosis, one of the most disabling complications (Manco-Johnson et al, 2007). To date no studies have reported whether prophylaxis decreases the risk for ICH (Ljung, 2008).
Since May 1998, the Division of Blood Disorders of the Centers for Disease Control and Prevention (CDC) has provided assistance to federally funded haemophilia treatment centres (HTCs) in the United States to collect a uniform set of clinical information, including ICH and the use of prophylaxis, in a voluntary surveillance programme called the Universal Data Collection (UDC) project (CDC 2002, 2003, Soucie et al, 2004). To date over 20 000 people with bleeding disorders have participated in this project. Our study objective, using the UDC project, was to identify risk factors associated with ICH, focusing on whether prophylaxis was associated with a reduction in ICH. This study represents the largest and most recent prospective surveillance for ICH in patients with haemophilia.
Patients who participate in the UDC project must receive care at a federally funded HTC and have a congenital or acquired deficiency in a coagulation protein with a functional activity of <50% or have physician-diagnosed von Willebrand disease. HTCs provide specialized multidisciplinary care for patients with bleeding disorders. Patients are enrolled in the UDC project by local HTC staff and participation is entirely voluntary. A uniform set of clinical data is obtained at the time of enrollment and yearly at the participant’s annual comprehensive clinic visit. Annual plasma specimens undergo infectious screening for hepatitis A, B, C, and human immunodeficiency virus (HIV) completed at the CDC laboratory. The UDC project has received approval from the Human Investigation Review Boards of the CDC and all of the participating HTCs.
For the UDC project, data are collected by HTC staff on three separate standardized collection forms including an initial registration form, an annual comprehensive visit form and a mortality form. With the initial registration, pertinent demographic information is collected including date of birth, haemophilia diagnosis, baseline clotting factor activity and self-reported race/ethnicity. In addition, there are questions regarding whether the subject had a prior history of physician diagnosed ICH and if so, at what age. The annual form records information on the prior year, including bleeding episodes, prescribed treatment, and inhibitor status. Specifically, the date of a physician diagnosed ICH and any associated factors including trauma, thrombocytopenia or ‘other’ are recorded. If a patient dies, a mortality form is completed by the HTC, indicating the primary and contributing causes of death.
We performed a nested case-control study. The study cohort was limited to males ≥2 years of age with haemophilia A or B who enrolled in the UDC and had an initial visit and at least one follow-up event between May 1998 and March 2008. Follow-up events were subsequent annual visits or reported mortality. Patients were followed until their study termination event; death, first reported incident ICH after enrolling in the UDC or last visit during the study period. Risk factors and patient characteristics were taken from information gathered on the last annual visit prior to the termination event.
The following definitions were used in the analysis: severe haemophilia was defined as a factor VIII or IX activity level <1%, moderate as 1–5%, and mild as 6–50% of normal. Factor assays were completed at individual institutions. The prescribed treatment modality (i.e. episodic vs. prophylactic factor infusions) and patient age were determined from information reported on the last annual visit prior to the study termination event. In the UDC, prophylaxis is defined as the recommendation to receive treatment products on a regular schedule to prevent bleeding. The prescribed prophylactic dose or information regarding patient compliance was not collected. HIV, hepatitis B, and hepatitis C status was determined by laboratory analysis completed by the CDC for the last annual visit prior to the study termination event. This infectious testing is serology-based. We were unable to determine if subjects cleared either a hepatitis B or C infection during the study period. Inhibitor status was based on the highest titre value reported during the study period with titre values >5 Bethesda units (BU) classified as high titre and a titre value between 0·5 and 5 BU classified as low titre inhibitor. Inhibitor titre measurement was completed at individual institutions. We were unable to determine if a subject had inhibitor resolution during the study period.
The average incidence rate of ICH was calculated based on cases that occurred in cohort members during the study period. Factors associated with ICH were identified using a nested case control design. Characteristics of patients who reported an ICH or died from ICH after enrolling in UDC (cases) were compared to all others in the cohort (controls). Odds ratios (ORs), with 95% confidence intervals calculated from the standard errors of the estimates were used to evaluate the strength of associations. The 10–15 year age group was used as the reference so that possible associations between young age and ICH could be measured. Interaction effects were assessed using the Breslow-Day Test for homogeneity of the ORs.
The independent association between prophylaxis and ICH was assessed using logistic regression. The model included all variables so that the simultaneous influence of all studied characteristics between prophylaxis and ICH could be assessed. Interaction terms identified during the stratified analysis were included in the multivariate logistic regression model in order to evaluate their statistical significance. Stratum-specific ORs were calculated for combinations of factors with significant interactions. All hypothesis testing was two tailed with ORs and confidence intervals reported. All analyses were performed using SAS version 9.2 statistical software (SAS Institute, Cary, NC, USA).
During the study period, May 1998–March 2008, there were 10 262 subjects who met the inclusion criteria. The mean follow up time for the cohort was 4·9 ± 2·5 years. Table I lists the patient characteristics for the entire cohort. As expected, haemophilia A was more common (78·8%) than haemophilia B and severe disease was predominant (54·4%). The cohort comprised 56·8% paediatric (2–20 years) and 43·2% adult subjects. Prophylaxis was prescribed in 29·7% of the entire cohort, and in 46% of all subjects with severe haemophilia, 17% with moderate and 2% with mild disease. The majority of patients who were prescribed prophylaxis, 79·2% (2419/3052), were younger than 20 years. In subjects classified as having a high titre inhibitor 19·7% (50/254) were prescribed prophylaxis; 62% (31/50) with an activated prothrombin complex concentrate (aPCC), 28% (14/50) with a recombinant factor VIII or IX replacement product and in 10% (5/50) the product was not specified. In subjects classified as having a low titre inhibitor 26·8% (89/332) were prescribed prophylaxis; 19·1% (17/89) with an aPCC, 77·5% (69/89) with either a plasma-derived or recombinant factor VIII or IX replacement product and the product was not specified in 3·4% (3/89).
Table I. Cohort demographics (N = 10 262).
|Haemophilia A||8087 (78·8)|
|Haemophilia B||2175 (21·2)|
| Severe||5581 (54·4)|
| Moderate||2398 (23·4)|
| Mild||2283 (22·2)|
|Age group (years)|
| 2–9||2114 (20·6)|
| 10–15||2095 (20·4)|
| 16–20||1620 (15·8)|
| 21–40||2585 (25·2)|
| >41||1847 (18)|
| Black (non-Hispanic)||1270 (12·4)|
| White (non-Hispanic)||7043 (68·6)|
| White (Hispanic)||1229 (12)|
| Other||720 (7)|
|Prior ICH||1157 (11·3)|
| Total cohort||3052 (29·7)|
| Severe||2591 (84·9)*|
| Moderate||412 (13·5)*|
| Mild||48 (1·6)*|
| Low titre||332 (3·2)|
| High titre||254 (2·5)|
|Hepatitis C||4107 (40)|
|Hepatitis B (chronic)||90 (1)|
Of the 10 262 subjects in the cohort 199 (1·9%) experienced an ICH during the study period. Table II lists patient demographics for those who experienced an ICH event and those who died from an ICH. Based on patient follow up time the incidence rate for ICH was 390/105 person years. The majority of patients with ICH (74·4%) had severe haemophilia. Thirty nine of the 199 patients died from ICH, resulting in a mortality of 19·6%. The mean age of death from ICH was 44·2 ± 16·1 years with a range of 13·8–81·5 years. Of the 150 ICH events in which associated factors were recorded, head trauma was the most common in 58·7% (88/150) followed by spontaneous ICH in 22·7% (34/150).
Table II. Demographics for subjects with ICH events and deaths from ICH.
|Haemophilia A||159 (79·9)||33 (84·6)|
|Haemophilia B||40 (20·1)||6 (15·4)|
| Severe||148 (74·4)||28 (71·8)|
| Moderate||32 (16·1)||3 (7·7)|
| Mild||19 (9·5)||8 (20·5)|
|Age group (years)|
| 2–9||48 (24·1)||0|
| 10–15||26 (13·1)||1 (2·5)|
| 16–20||24 (12·1)||1 (2·5)|
| 21–40||52 (26·1)||15 (38·5)|
| >41||49 (24·6)||22 (56·5)|
| Black (non-Hispanic)||43 (21·6)||5 (12·8)|
| White (non-Hispanic)||117 (58·8)||26 (66·7)|
| White (Hispanic)||23 (11·6)||7 (18)|
| Other||16 (8)||1 (2·6)|
|Prior ICH||61 (30·7)||6 (15·4)|
|Prophylaxis||52 (26·1)||6 (15·4)|
| Low titre||12 (6)||0|
| High titre||17 (8·5)||6 (15·4)|
|HIV||34 (17·1)||17 (43·6)|
|Hepatitis C||106 (53·3)||33 (84·6)|
|Hepatitis B (chronic)||4 (2)||1 (2·5)|
The results of univariate analyses to assess the relationships between clinical factors and ICH are presented in Table III. The following factors were associated with ICH: prior ICH, severe haemophilia, a high titre inhibitor, hepatitis C, Black race, and age (2–9 years or >41 years). Haemophilia type (A versus B) was not a statistically significant factor OR = 1·07 (0·75–1·92). The association between ICH and prophylaxis was not statistically significant with an OR = 0·83 (0·61–1·15). The incidence of ICH was twice as common among patients with severe haemophilia (2·65%) compared to those with moderate haemophilia (1·33%). Less than 1% of patients with mild haemophilia experienced ICH.
Table III. Clinical factors associated with ICH for all patients in the cohort, N = 10 262 (univariate analysis).
|Prior ICH||3·62 (2·66–4·92)||<0·001|
|Severe Haemophilia||3·25 (2·01–5·25)||<0·001|
|High Titre Inhibitor||4·01 (2·40–6·71)||<0·001|
|Hepatitis B||2·38 (0·86–6·55)||0·083|
|Hepatitis C||1·73 (1·30–2·29)||<0·001|
|Black (non-Hispanic)*||2·07 (1·46–2·96)||<0·001|
|Age group (years)†|
| 2–9||1·85 (1·14–2·99)||0·01|
| 16–20||1·20 (0·68–2·09)||0·53|
| 21–40||1·63 (1·02–2·63)||0·04|
| >41||2·17 (1·34–3·50)||0·001|
We restricted the analysis of prophylaxis and ICH to patients with severe disease because prophylaxis prescription and the risk of ICH were highest in this group. Factors that were independently associated with ICH in patients with severe disease after adjusting for the effects of other study variables are shown in Table IV. Children in the 2–9 year age group had the highest OR = 1·92 (1·05–3·51) for ICH. There was a statistically significant interaction between prophylaxis and both inhibitor and HIV status. Prophylaxis prescription in patients with severe haemophilia without HIV and in patients without an inhibitor was associated with a significant risk reduction [OR = 0·52 (0·34–0·81) and 0·50 (0·32–0·77) respectively]. The association between prophylaxis and ICH was not significant among patients with an inhibitor or among those who were HIV infected.
Table IV. Clinical factors associated with ICH among 5581 patients with severe haemophilia (multivariate analysis).
| +High titre inhibitor||1·7 (0·54–5·3)||0·36|
| +Low titre inhibitor||2·44 (0·7–8·53)||0·16|
| No Inhibitor||0·5 (0·32–0·77)||0·002|
| HIV||1·37 (0·6–3·2)||0·45|
| No HIV||0·52 (0·34–0·81)||0·004|
|Prior ICH||3·24 (2·27–4·64)||<0·001|
|Hepatitis B||2·99 (1·03–8·63)||0·04|
|Hepatitis C||1·82 (0·98–3·37)||0·06|
|Black (non-Hispanic)†||1·50 (0·99–2·28)||0·06|
|Age group (years)‡|
| 2–9||1·92 (1·05–3·51)||0·03|
| 16–20||0·98 (0·48–2·02)||0·97|
| 21–40||0·72 (0·33–1·6)||0·43|
| >41||1·18 (0·52–2·68)||0·69|
Intracranial haemorrhage is a serious complication for patients with haemophilia. Studies that have previously addressed ICH in haemophilia were limited in either study design, small patient numbers or include a time period that does not reflect the present routine use of prophylaxis. This study is the largest haemophilia cohort studied in the prophylaxis era and is the first to report on the effect of prophylaxis prescription on the occurrence of ICH.
For the entire cohort of patients with haemophilia, prophylaxis was not associated with a significant reduction in ICH. This is probably secondary to the fact that prophylaxis is not indicated for patients with mild disease and used to a lesser extent among patients with moderate disease. Also, the risk of ICH is much higher in patients with severe disease. Further analysis, restricted to patients with severe haemophilia, demonstrated a protective effect of prescribed prophylaxis on ICH occurrence. Patients with severe disease who were prescribed prophylaxis and were not HIV positive had a 48% risk reduction and patients without inhibitors who were prescribed prophylaxis had a 50% risk reduction for ICH.
There is clear evidence to support the use of prophylaxis in paediatric patients to prevent joint disease. A randomized clinical trial in boys with severe factor VIII deficiency demonstrated that prophylaxis prevented joint damage and decreased the frequency of joint haemorrhage (Manco-Johnson et al, 2007). However, the indications for and efficacy of prophylaxis in adults with haemophilia remains controversial (Hay, 2007). Our data indicates that children aged 2–9 years with severe haemophilia have the highest risk of developing ICH. Our findings suggest that prophylaxis should be considered for paediatric and adult patients with severe haemophilia, especially those who have had a prior episode of ICH. Over one-quarter (26%) of patients with ICH were prescribed prophylaxis prior to the ICH event. As with other therapies, prophylaxis is only effective if the patient is compliant and we were unable to measure the level of adherence or frequency of weekly infusions in this study. While patients receiving prophylaxis may have factor levels in the mild to moderate range depending on the time from last infusion, it does not guarantee protection against ICH, especially from trauma. Prophylaxis probably decreases the risk of spontaneous episodes of ICH in patients with severe haemophilia but traumatic induced bleeding is still possible.
In the era of safe clotting factor concentrates, the most serious complication of haemophilia is the development of an inhibitor, an alloantibody that inhibits or neutralizes the deficient factor. In our study, the presence of a high titre inhibitor was the most significant risk factor for ICH for the entire cohort with an OR of 4·01 (2·40–6·71). Prophylaxis in patients with severe haemophilia and a high or low titre inhibitor was not associated with a statistically significant protective effect for ICH with an OR of 1·7 (0·54–5·3) for a high titre inhibitor and 2·44 (0·7–8·53) for low titre inhibitor. The lack of statistical significance is probably secondary to small patient numbers.
Prior studies have found an increased risk for ICH among patients with HIV infection (de Tezanos Pinto et al, 1992; Nuss et al, 2001). It is hypothesized that the bleeding risk in patients with haemophilia is increased if they develop HIV-related thrombocytopenia. Nuss et al (2001) conducted a nested-case control study of 3269 males with haemophilia in the US from 1995 to 1998 with 88 ICH cases. Twenty-five per cent of the total cohort had HIV and 44% of the subjects with ICH were HIV positive. The increased risk of ICH with HIV was present in whites and not in other race or ethnic groups. In our cohort, a smaller proportion of patients were infected with HIV (12·8%) and only 17·1% of the ICH cases were HIV positive. Therefore, while our data suggested an increased risk of ICH in those patients who were HIV positive [OR = 1·42 (0·98–2·06)], the difference was of borderline statistical significance. In patients with severe haemophilia we found significant interaction between HIV and prophylaxis. The protective effect of prophylaxis was lost when a patient with severe haemophilia had HIV [OR = 1·37 (0·6–3·20)]. This suggests an increased risk of ICH in HIV-positive patients that is not mitigated by prophylaxis.
The role of trauma in patients with haemophilia who develop ICH is well established, although the proportion of ICH events attributed to trauma has varied across studies. In the study reported by Nuss et al (2001), only 21% of ICH events had a preceding head injury, while other studies have found trauma to be more prominent, with a range of 39·7–67·2% (de Tezanos Pinto et al, 1992; Klinge et al, 1999; Antunes et al, 2003; Stieltjes et al, 2005). In our study, over half (58·7%) of the events with recorded data regarding contributing factors had a preceding history of head trauma. If the remaining 49 ICH events without contributing factor information were assumed to not be associated with trauma then 44·2% (88/199) of cases had a preceding history of head trauma. This significant proportion underscores the importance of this potentially preventable cause of ICH in patients with haemophilia.
The 19·6% mortality rate from ICH in our cohort is consistent with prior published reports (Nuss et al, 2001; Stieltjes et al, 2005). Prior studies have suggested that a delay in presentation may contribute to mortality from ICH in patients with haemophilia (Stieltjes et al, 2005). This area of concern could not be addressed in our study. Ninety-five per cent (37/39) of the deaths from ICH occurred in subjects over 20 years of age. It is striking that the burden of mortality has shifted to the adult population. This has not been previously reported and is in contrast to the results of Stieltjes et al (2005) where 25·9% (7/29) of the mortality was in patients <15 years of age. It is possible that the increasing use of prophylaxis in the paediatric age group may be contributing to the decreased mortality from ICH in children.
This study has several limitations that should be considered. Due to UDC enrollment requirements during the study period, patients under the age of 2 years were not included, so we cannot comment on the rate of ICH in this age group. The lack of data from birth to the age of 2 years means that our estimates are an underestimate of the true burden of ICH in the haemophilia population. The UDC project extended enrollment to children aged 0–2 years as a pilot project in 2003 and nationwide in 2004. Kulkarni et al (2009) reported on this new UDC enrollment data in 580 males with haemophilia age 0–2 years and found that following bleeding from circumcision the cranium was the second most common site of the first bleeding episodes throughout the first 2 years of life. Of the 77 children with cranium bleeding 36% experienced an ICH and 81% of these episodes were associated with delivery.
There are also limitations to the study definitions for hepatitis C and inhibitor status. The study definition for chronic hepatitis C is based on serological testing. We were unable to identify patients who cleared the infection. While this limitation results in a misclassification of patients who cleared the hepatitis C infection this would bias the results toward the null and underestimate the effect of hepatitis C infection on the occurrence of ICH. A patient was determined to have an inhibitor based on the presence of an inhibitor at any visit prior to study termination. The inhibitor may have resolved during the study period but we were unable to account for this with certainty. Secondary to this uncertainty we chose this conservative definition knowing that it may misclassify patients as having an inhibitor which may have resolved prior to the ICH event. We feel that this misclassification would bias the results towards the null and underestimate the effect of an inhibitor on ICH making our results that much stronger.
In addition, there may be selection bias because patients in this study received specialized care at an HTC and were willing to participate in the UDC. However, based on a study of all haemophilia patients living in six US states, approximately 70% of patients with haemophilia receive care at a HTC and an estimated 90% of all eligible haemophilia patients invited to participate in the UDC have enrolled (Soucie et al, 1998, CDC 2002). As mentioned previously, our exposure definition for prophylaxis is limited by the information collected in the database. We do not know the specific schedule or patient adherence to the prescribed regimen. It is important to note that the inclusion of patients that may have been on a less rigorous prophylaxis schedule or those who were non-compliant would bias our results toward the null, making our positive results even stronger.
In summary, in this modern haemophilia cohort the most significant risk factors for ICH included the presence of an inhibitor, a prior ICH and factor level. Patients with severe disease who did not have an inhibitor or HIV experienced a significant risk reduction for ICH when prescribed prophylaxis. We believe that the discussion of therapeutic options for patients with severe haemophilia should consider the risk of ICH. Our results provide evidence that prophylaxis is an important component of ICH prevention.
The findings and conclusions in this report are those of the author(s) and do not necessarily represent the views of the Centers for Disease Control and Prevention.