The development of chronic pain and limited motion secondary to recurrent joint and muscle bleeding causes significant morbidity in persons with haemophilia. Overall, patients with the musculoskeletal complications of haemophilia are more likely to have had early and recurrent ankle, knee or elbow haemarthroses; hip abnormalities, when they occur, generally manifest at an older age [1]. Although less common, hip abnormalities serve as a significant source of impaired mobility. Joint and muscle bleeding around the hip could pose a particularly significant problem in the paediatric population as prior studies indicate that arthropathy developing in childhood results in more severely destructive hip disease than that occurring after 20 years of age [2].

Although the risk factors for hip dysfunction in haemophilia remain poorly understood, ancillary factors known to generally affect haemophilic joint morbidity may impact this phenomenon [3,4]. Prophylactic factor administration has been shown to prevent joint destruction and to decrease the frequency of all joint haemorrhage [5–8]. However, among patients with severe haemophilia who develop high-titre inhibitors of ≥5 Bethesda Units (BU), effective prophylaxis is more difficult to achieve and overall morbidity, including haemophilic arthropathy, is often more pronounced [4,9]. Furthermore, in the US children and adults exhibit an increasingly sedentary lifestyle with an associated increased likelihood of obesity [10]. The potential influence of this national trend on range of motion (ROM), and secondarily, on predisposition to haemorrhage, in joints is concerning. Recently published studies demonstrated that obese males with haemophilia A had lower ROM and a faster rate of loss of joint mobility in the lower limbs compared with those with a normal body mass index (BMI) [3,11]. The impact of knee and ankle arthropathy on hip disease in haemophilia has not been documented to date.

Based on recent institutional experience with hip haemarthroses in several children aged 1–10 years of age, we sought to examine the national prevalence of hip abnormalities in US children and adults with severe, moderate and mild haemophilia A and haemophilia B who are currently enrolled in the Center for Disease Control and Prevention’s Universal Data Collection (UDC) database. The UDC project is a surveillance database of over 14 000 patients with haemophilia treated at federally funded Hemophilia Treatment Centers (HTC), and includes joint ROM on 96% of enrolled subjects [12,13]. We also sought to evaluate the prevalence of hip abnormalities relative to potentially important comorbidities including advancing age; a history of high-titre inhibitor; concomitant prevalence of abnormal ROM in knee and ankle and elevated BMI. A better understanding of this haemophilic complication and the identification of its potential risk factors could allow for targeted therapy and lifestyle adjustments aimed at reducing risk of hip joint abnormality and associated morbidity.

The UDC project, initiated in 1998, is a surveillance database containing demographic and clinical information on over 14 000 patients with haemophilia treated at a network of specialized HTC, and includes joint ROM measurements on 96% of enrolled subjects [12,13]. All male UDC participants with severe, moderate and mild haemophilia A and haemophilia B, enrolled in the UDC database between 1998 and 2008, and aged 2–69 years at the time of their last visit, were eligible for inclusion in the study. The study examined the relationship between demographic and clinical characteristics of the population and the prevalence of documented or suggested hip abnormalities. Hip abnormality was defined by: (i) a deficit in hip ROM at the last recorded UDC visit, and/or (ii) designation of the hip as a target joint at the last recorded visit, and/or (iii) joint replacement/fusion as recorded at any UDC visit during the study period. Deficit in hip ROM was determined for each subject and defined as a ROM that was at least 3 standard deviations below the mean normal ROM for hip flexion and/or extension for his age group individuals [14]. Joint-specific bleeding data, including bleeding site and frequency, are not captured in the UDC data set. To better define hip abnormality in the absence of such data, subjects with bilateral hip ROM deficits were excluded from the analysis unless the deficit was corroborated by concomitant target joint designation or hip replacement in at least one hip joint. In this way, we attempted to minimize the likelihood of inappropriately including bilateral deficits in hip ROM not due to haemorrhage, (e.g. muscle tightness, increased body mass, inactivity), in the analysis. Similarly, the lack of site-specific bleeding data precluded the exclusion from analysis of residual deficits in hip ROM from prior hip flexor or extensor muscle and connective tissue haemorrhage. Therefore, we chose to use the term ‘hip abnormality’ to encompass both the intra-articular and extra-articular complications of bleeding into the hip area.

Hip abnormality was examined relative to the following comorbidity variables: advancing age; history of high-titre inhibitor; presence of knee and ankle arthropathy and BMI. In addition, haemophilia type and severity were examined. Finally, the prevalence of hip abnormalities was examined in subjects with severe haemophilia A and haemophilia B and no history of an inhibitor who were receiving continuous prophylaxis at the first UDC visit and compared with those subjects who received only on-demand (episodic) treatment.

The study included 8192 subjects aged 2–69 years (51% aged 2–19 years) with haemophilia A (78%) and haemophilia B of all severities (54% are severe) enrolled in the UDC database between 1998 and 2008. Hip abnormality was found in 1372/8192 (17%) of all evaluated haemophilic subjects (Table 1). The prevalence of hip abnormality in children aged 2–8 years was 7% (94/1328), and was defined solely by abnormal hip ROM in 87/94 (93%) of the youngest subjects. Hip abnormality defined on the basis of hip replacement or target joints was much less frequent and occurred almost exclusively in subjects over the age of 20 years.

Associations between subject demographic and clinical characteristics and hip abnormality are shown in Table 2. A significant increase in the prevalence of hip abnormality was noted in association with increasing age and haemophilia severity (P < 0.0001), in subjects with haemophilia A as compared with haemophilia B (18% vs. 14% P = 0.0003), among subjects with a positive history of a high-titre inhibitor (22% vs. 16%, P = 0.0007), among those who were overweight or obese as compared with non-overweight (20%, 24% vs. 16% P < 0.0001) and among those with concomitant knee (25% vs. 11%, P < 0.0001) and ankle arthropathy (24% vs. 11%, P < 0.0001) (Table 2). However, there was no significant difference in hip abnormalities among those subjects receiving continuous prophylaxis (17%) when compared with those who had received only episodic (on-demand) therapy (16%, P = 0.47).

To evaluate the independent associations between the studied demographic and clinical characteristics and hip abnormality, a logistic regression model was utilized. We found statistical interaction (effect modification) between concomitant knee arthropathy and both age and BMI. This meant that the strength of the associations between age and hip abnormality and separately, BMI and hip abnormality, was different depending on the presence or absence of knee arthropathy. Therefore, we present stratum-specific odds ratios for both age and BMI in Table 3.

Haemophilia A (aOR = 1.3, 1.0–1.4), severe haemophilia (aOR = 1.3, 1.0–1.5), a high-titre inhibitor (aOR = 1.4, 1.1–1.7) and concomitant ankle arthropathy (aOR = 1.7, 1.4–1.9) were each independently associated with the presence of hip abnormality (Table 3). Among subjects without concomitant knee arthropathy, subjects who were 9–19 years old were more likely than those who were 2–8 years old to have hip abnormalities (aOR = 1.7, 1.2–2.5). In contrast, among those with knee arthropathy only the 45–69-year-old subjects were more likely than the youngest subjects to have hip abnormalities (aOR = 3.4, 1.9–5.9).

Concomitant knee arthropathy also influenced associations between BMI and hip abnormality (Table 3). Overweight (aOR = 1.4, 1.1–1.8) and obese subjects (aOR = 2.1, 1.6–2.8) without knee arthropathy were 1.4 times and 2 times more likely than their normal weight counterparts to have hip abnormalities. However, there was not a significantly increased likelihood of hip abnormalities among either overweight or obese subjects who also had knee arthropathy.

The prevalence of hip abnormalities in haemophilia has not been characterized in the context of the availability of more intensive coagulation factor replacement therapy. In this study, we performed such an evaluation in a US haemophilia cohort that has undergone surveillance for the prevalence of disease- and treatment-related complications of the disorder. However, as discussed in the Methods section of this manuscript, the definitions of hip dysfunction required modification to compensate for the lack of site-specific bleeding data. The term ‘hip abnormality’ was therefore chosen to encompass both the intra-articular and extra-articular morbidity that could contribute to unilateral loss of ROM.

Within the context of the methodological limitations, we report an overall prevalence of hip abnormality in this US haemophilia population of 16.7%. This is a lower prevalence than that observed in this cohort for both knee (42%) and ankle (45%) arthropathy, but possibly more frequent than anticipated in persons with haemophilia. Furthermore, the long-term morbidity associated with overt hip arthropathy appears to be significant. Among older adults in the study sample, 5% had experienced target joint bleeding in a hip and 7% required a replacement of at least one hip.

Our initial clinical observation of hip abnormality in very young children was corroborated by the observed prevalence of 7% in the youngest study sub-cohort. However, the low hip abnormality prevalence was somewhat reassuring. Nonetheless, as developing joints may be more prone to destructive arthropathy with minimal bleeding [17], these data also suggest that caregivers should be alert to symptoms of local pain and unilateral decreased hip ROM as possible signs of hip haemarthrosis or haemorrhage into large muscle groups in the area (e.g. iliopsoas), even in a very young child without other lower extremity arthropathy.

Although the database did not allow us to examine predisposing events or risk factors, we were able to evaluate potentially associated comorbid conditions, related and unrelated to haemophilia. Not surprisingly, haemophilia severity and positive high-titre inhibitor status, variables known to be associated with overall increased orthopaedic morbidity, were also associated with significantly higher hip abnormality prevalence in this study. The increased association of hip abnormality with underlying haemophilia A when compared with haemophilia B corroborates the Italian observations of Iorio et al. [18]. However, prophylaxis intervention was not a significant modifying factor in our study. This observation likely relates to the nature of a surveillance data collection limited by nomenclature adopted prior to the widespread use of prophylaxis in the United States. Although a more precise classification was unavailable for inclusion in our analysis, it remains likely that much of the ‘continuous’ prophylactic use of factor in our data set reflected the late initiation of secondary prophylaxis similarly described in the international orthopaedic study by Aledort et al. [7].

In this US cohort, 20% of subjects were overweight and 12% were obese, according to well-defined criteria. In excluding bilateral decreased hip ROM that may have been due to elevated BMI alone, we were able to examine and ascertain a very strong independent association between hip abnormality and this variable, particularly obesity. This association remained highly significant, except in the presence of concomitant knee arthropathy. These results would suggest that the impact of knee arthropathy on the development of hip abnormalities is more significant than overweight and obesity for patients in whom both conditions coexist. Consequently, the association between excess weight and hip abnormalities is best observed in patients without knee arthropathy as a complication of their haemophilia.

The relationship between elevated BMI and hip abnormality is likely complex; each condition has the potential to influence the other. Nonetheless, the growing rates of obesity among the general population demonstrate an alarming trend [19,20]. Based on previously published data, this trend may be even more problematic among those individuals with haemophilia [11]. Several of the comorbidities associated with hip abnormality cannot be easily manipulated; however, altering lifestyle choices to maintain a normal BMI for age could represent an important means of maintaining overall orthopaedic health in persons with haemophilia.

The significant direct correlation observed between hip abnormality and both advancing age as well as concomitant lower extremity arthropathy was also expected. However, we did not anticipate the strength of the association between hip abnormality and knee arthropathy. Concomitant knee arthropathy appeared to be primarily responsible for the significant age-related association with hip abnormality in the adult subgroups in our study. The interpretation of these findings is again limited by the nature of the data. Nonetheless, these data do suggest the potential importance of preventing knee arthropathy in the maintenance of normal hip ROM.

The subgroup of haemophilia subjects aged 9–19 years warrants further discussion. The prevalence of hip abnormality in this age group was 15%, double that observed in the youngest cohort. As with the youngest subjects, the majority of the hip abnormality in this sub-group (96%) was defined by significant unilateral loss of hip ROM. Importantly, absent knee or ankle arthropathy did not appear to modify this significant age effect. Given that our definition of hip abnormality encompassed limitations in both intraarticular hip and extra-articular muscle and connective tissue morbidity, these subjects may have experienced altered hip mechanics from more muscle (including iliopsoas) and groin tissue bleeding associated with greater sports-related activity, parameters that we could not measure. In addition, in a prior recent study of the UDC cohort, subjects aged 2–20 were noted to have a 15% and 17.4% prevalence of overweight and obesity, respectively [11]. Both states of elevated BMI were noted in that study to independently and statistically correlate with loss of ROM over time [11]. Obesity prevalence in the younger subjects trends higher than that for this study population as a whole (12%). Although bilateral loss of hip ROM was excluded in this study to minimize the effect of BMI alone, increased BMI compared with the cohort as a whole could have been partially responsible for the results observed in this age group.

We conclude that among subjects enrolled in the UDC database, hip abnormality accounts for an appreciable amount of orthopaedic morbidity. As might have been anticipated, advancing age, haemophilia severity and type, positive high-titre inhibitor status and elevated BMI were all factors associated with increased prevalence of hip abnormality, as defined in this study. Surprisingly, concomitant knee arthropathy proved to be a strongly associated comorbidity. However, given the limitations of this study, prospective longitudinal data collection that includes joint-specific haemorrhage as well as more robust markers of arthropathy will be required to both confirm the prevalence of specific hip arthropathy and further define its risk factors, particularly in the younger population in whom early intervention would minimize long-term morbidity.