Christoph Bidlingmaier, Pediatric Hemophilia Center, Dr von Hauner’s Children’s Hospital, University of Munich, Lindwurmstr. 4, 80337 Munich, Germany. Tel.: +49 89 5160 2811; fax: +49 89 5160 4453. E-mail: email@example.com
Summary. Background: Diagnosing mild bleeding disorders (BDs) in children is difficult. Bleeding scores (BSs) have been proposed for obtaining standardized quantitative histories.
Objectives: To compare the Canadian pediatric bleeding questionnaire (PBQ) with the new ISTH bleeding assessment tool (ISTH BAT) for the determination of BS in a routine pediatric outpatient setting.
Methods: One hundred children with a suspected BD were enrolled in this cross-sectional study. Bleeding scores were calculated for all children and their natural parents. For all children, extensive laboratory investigations were performed.
Results: Based on laboratory tests, 56 children were diagnosed as having no BD, 11 were diagnosed with possible VWD, 12 with VWD 1, 11 with VWD 2, five with possible platelet defects, and five with mild factor deficiencies. Both questionnaires were able to discriminate between no BD and VWD (P = 0.0001), but the area under the receiver characteristics curve to detect any mild BD was only 0.76. Despite the inherited nature of the BD, a family score did not increase the ability to discriminate between no BD and VWD (P = 0.2052). There was no significant difference between the two tools used (P = 0.3253) or simple qualitative criteria, such as yes/no questions regarding bleeding (P = 0.3477).
Conclusions: The two tools translated into German did not differ substantially. Both were able to discriminate between no BD and a possible BD with acceptable accuracy. A BS of < 2 makes a BD unlikely. Simple qualitative criteria were similar; however, to allow comparison of studies and follow-up in patients over time, we recommend the ISTH BAT.
Diagnosing a child with a mild bleeding disorder (BD), namely von Willebrand disease (VWD) type 1, is a difficult task. Children are often referred to a pediatric coagulation center prior to invasive procedures to exclude such BD. However, laboratory testing has been shown to be of little value unless repeated testing has been performed and a correlation with clinical symptoms is possible [1,2]. Because symptoms might be subtle, especially in younger children, a detailed clinical history is crucial. In order to standardize this, two questionnaires, the Canadian Pediatric Bleeding Questionnaire (PBQ)  and the International Society on Thrombosis and Haemostasis bleeding assessment tool (ISTH BAT)  have been proposed. Both tools allow the calculation of a bleeding score (BS) and use the more complex score of the Molecular and Clinical Markers for the Diagnosis and Management of Type 1 VWD study group (MCMDM-1VWD) as a basis . The PBQ scores from −1 to +4 per question, with −1 indicating that the child was at risk of bleeding but did not bleed. The ISTH BAT scores from 0 to 4. The PBQ has been used under study conditions to quantify bleeding symptoms in children with VWD  and platelet function disorders (PDs)  as well as in routine examinations . The MCMDM-1VWD score has also been used under study conditions, mostly in adults, but also more recently in adults and children in the clinical setting .
Our study aimed to evaluate the PBQ and the ISTH BAT in a new German translation in the pediatric clinical setting for evaluation of a suspected BD. We (i) documented the clinical practicability of the scores and then (ii) compared these two scores and (iii) the child and family scores. The rational for calculating a family BS (i.e. the total sum of the BS applied to the child and both natural parents) was that it might serve as a surrogate marker for a pediatric BS, overcoming the problem of a low BS occurring in younger children due to lack of bleeding possibilities. We than (iv) correlated the ISTH child score with the diagnosis and (v) compared the value of this quantitative scoring system with a more simple qualitative yes/no analysis.
Material and methods
We enrolled 100 consecutive patients in this cross-sectional study, who were referred to our specialized pediatric coagulation department (Children’s University Hospital Munich) between December 2009 and June 2010. Patients were eligible if they presented due to: (i) bleeding symptoms, (ii) abnormal laboratory test of the activated partial thromboplastin time (aPTT) or prothrombin time (PT), or (iii) history of bleeding in first degree relatives. Further inclusion criteria were age between 1 and 17 years and availability of the bleeding history of the patient and both natural parents. Children were excluded if they presented with concomitant disease (including recent infections), had recently used drugs such as antibiotics or acetylsalicylic acid, or had been vaccinated in the previous 6 weeks.
Ethical approval was granted by the local ethics committee and written informed consent was obtained from all parents.
Bleeding assessment tools
Detailed descriptions of the questionnaires used and calculation of the BS are available from Bowman et al.  for the Canadian PBQ and Rodeghiero et al.  for the ISTH BAT. For the ISTH BAT, the version circulating in the ISTH working party on BS in November 2009 was used, which did not differ from the finally published version for the included patients. The questionnaires and the scoring tables were first translated from English into German and then retranslated into English to avoid translation bias and allow validation of these new tools for German speaking patients. In general, both bleeding scores consist of a set of main questions (i.e. ‘Have you ever had spontaneous epistaxis?’), which are then analyzed in more detail by documenting frequency, duration, medical therapy, etc. Then, these answers were assigned to a scoring system by a blinded documentation assistant, ranging from −1 to 4 (PBQ) or 0 to 4 (ISTH BAT). Additionally, main questions (= items) answered ‘Yes’ were documented as positive bleeding symptoms regardless of the severity as qualitative criteria. The total number of positive bleeding symptoms was then used for the ITEM analysis.
All children were interviewed in the presence of their parents using both questionnaires by trained physicians (child score). The physicians were not blinded to the reason for referral; however, the text used for the questionnaire was standardized. The time needed to complete the PBQ was recorded. Both questionnaires were then also completed for both natural parents (family score). Blood was drawn from all children in a standardized way. Physical and emotional stress was minimized as much as possible. Where necessary (e.g. confirmation of VWD type 1), blood tests were repeated. All laboratory tests were performed in all patients, regardless of their answers to the interview or external laboratory investigations. Parents were not tested.
The following standardized laboratory tests were performed for all children: PT (Thromborel S; Siemens Healthcare Diagnostics (Siemens), Eschborn, Germany; reference range 70–100%), aPTT (Pathrombin SL; Siemens, norm 30–42 s), fibrinogen (Clauss method; Multifibren U, Siemens, norm > 160 mg dL−1), platelet count (CBC-Analyser; Sysmex, Norderstedt, Germany; norm > 150 × 109 L−1), lupus-anticoagulant (LA) screening (HemosIL LAC Screen and Confirm; Instrumentation Laboratories (IL), Kirchheim, Germany; norm < 1.2), factor (F) VIII, IX, XI and XII (HemosIL APTT-SP; HemosIL Normal Control and HemosIL Factor Deficient Plasma, IL, norm > 70%), FXIII (Siemens norm > 70%), ristocetin-cofactor activity (VWF: RCo, BC von Willebrand reagent; Siemens, norm > 50%), von Willebrand factor antigen (VWF: Ag, HemosIL von Willebrand factor antigen kit, IL, norm > 50%), collagen binding activity (CBA kit; Haemochrom Diagnostica, Essen, Germany; norm > 70%), analysis of VWF multimers (Prof. U. Budde, Hamburg), closure times of the platelet function analyzer (PFA-100 using ADP and Epinephrin cartridges, PFA kit; Siemens), platelet aggregometry (Multiplate, TRAP, ADP, RISTO and ASPI Test; Dynabyte), Born aggregometry (Aggregometer Apact 4, Rolf Greiner Biochemica), cardiolipin antibodies (ELISA, Phadia, Freiburg, Germany; norm < 11 U L−1), β2 glycoprotein antibodies (ELISA; Orgentec Diagnostika, Mainz, Germany; norm < 11 U L−1) and ABO blood group (bedside test). For patients that showed PT prolongation, additionally FII and V were determined. Normal range values were age and blood group adapted where applicable. C-reactive protein was measured to exclude acute phase reactions.
Definition of diagnoses
No BD: no laboratory abnormalities (including FXII deficiency).
von Willebrand Disease type 1 (VWD 1): VWF:RCo levels of 0.05–0.50 U mL−1 on two occasions and a VWF:Ag of 0.05–0.50 U mL−1 on at least one occasion, a ratio of VWF:RCo/VWF:Ag of > 0.50, a normal multimer pattern, and a positive clinical history (definition according to ).
Possible VWD: low VWF:Ag, FVIII and VWF:RCo levels < 50% on one occasion, and a normal multimer pattern with or without a positive clinical history.
VWD 2: defined according to published laboratory criteria . All patients were subtyped as type 2A by showing a significant loss or relative deficiency of the large multimers and a VWF functional discordance.
Single factor deficiencies: in parallel to the ISTH definition of mild hemophilia, factor levels below 40% were considered relevant.
Platelet defects: suspected if PFA and/or Multiplate were abnormal on at least two occasions. Born aggregometry was performed but the results were not attributable to defined diseases. Therefore these patients were defined as ‘possible PD’. Flow cytometry was not available.
Descriptive results were expressed as means (± standard deviation, SD) or medians (± interquartile range [IQR], or minimum and maximum). Continuous data were compared using the Kruskal–Wallis equal population test. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) with exact binomial confidence intervals were calculated on the basis of the children’s true disease status and the ability of the dichotomous BS to define the disease status. Positive likelihood function (receiver operating characteristic curve, ROC) analysis was performed using sensitivity and false positives (1 minus specificity) at all cut-points for a BS. To compare predictability of different BS values, the area under the curve was calculated and tested using the algorithm suggested by DeLong et al. .
Statistical analysis was performed using stata 12.0 (Stata Corp, College Station, TX, USA).
One hundred consecutive children were included in the study. The main referral reason was a positive bleeding history in 44 children. Twenty-nine children were referred due to an isolated aPTT prolongation without previously noted bleeding symptoms, and 27 due to a positive family history of bleeding. The median age of the patients was 5.7 years (range 1.1–16.9), and 46 children were male. Demographic data are summarized in Tables 1 and 2.
Table 1. Patient demographics and laboratory results by diagnosis
VWD Type 1
VWD Type 2
Mild F deficiencies
BD, bleeding disorder; F, factor; VWD, von Willebrand disease; PD, platelet defect; aPTT, activated partial thromboplastin time; VWF, von Willebrand factor; Ag, antigen; RCo, ristocetin cofactor; SD, standard deviation. *Due to the difference in aPTT values, FXII-deficient patients are listed separately in this table. However, based on our definition in this manuscript, these patients are considered as ‘no BD’.
N (Total 100)
Sex male [n]
Median age (range) [years]
Blood Group 0 [n]
Mean aPTT (SD) [sec]
32.3 (± 5.0)
45 (± 24.7)
34.6 (± 5.4)
36.4 (± 5.5)
35.3 (± 4.2)
34.6 (± 3.6)
33.4 (± 11.3)
Median VWF: Ag [%]
Median VWF:RCo [%]
Median FVIII [%]
Table 2. Diagnoses by type of referral
Reason for referral
Bleeding symptoms (n = 44)
Abnormal clotting tests (n = 29)
Family history (n = 27)
BD, bleeding disorder; VWD, von Willebrand disease; PD, platelet defect. *Due to the difference in aPTT values, FXII-deficient patients are listed separately in this table. However, based on our definition in this manuscript, these patients are considered as ‘no BD’.
No BD *[n (%)]
Mild factor deficiency
Fifty-six children were not considered to have a BD, including 12 who were diagnosed with FXII deficiency. Eleven children were classified as having a possible VWD (Table 1). The most frequent definite BD was VWD 1 (n = 12), followed by VWD 2 (n = 11). Mild factor deficiencies were diagnosed in five patients (two patients with FXIII (38% and 40%), and 1 patient with FVII (38%), FX (36%) and FVIII (36%) deficiency each. PD was suspected in five patients because of prolonged closure times in the PFA or abnormalities in Multiplate. However, Born aggregometry did not allow a defined diagnosis. The probability of detecting a BD, including possible VWD, was 57% (95% CI, 41–72%) in children referred due to bleeding symptoms, 35% (95% CI, 18-54%) in children referred due to abnormal clotting tests, and 33% (95% CI, 17–54%) in children referred due to family history (Table 2).
No child presented specific pediatric bleeding symptoms (i.e. bleeding after circumcision, umbilical stump bleeding, and hematoma after vaccination). The most commonly detected bleeding symptom was epistaxis (n = 36). It was combined with other symptoms in 22 patients; in 19 of those a BD was diagnosed. No BD was found in 14 patients with epistaxis as the only symptom. Other bleeding symptoms were hematoma (n = 22), gum bleeding (7), postsurgical bleeding (7), gastrointestinal bleeding (3), menorrhagia (2) and hemarthroses (2). No bleeding symptoms were reported in 42 children. There was no clear association between symptoms and laboratory diagnosis.
Evaluation of the Bleeding Scores
Clinical practicability The duration of the interview to obtain the bleeding history of the patient using the PBQ varied from 5 (no or mild bleeding symptoms) to 30 min (extensive bleeding symptoms). A median time of 10 min was sufficient to complete the questionnaire.
Comparison of the PBQ and the ISTH BAT Table 3 summarizes BS results for the children and their families. The ability to detect any mild BD, including possible VWD, was similar for the PBQ (area under the curve [AUC], 0.768; Fig. 1A) and the ISTH BAT (AUC, 0.764: P > 0.05; Fig. 1B). In one child with VWD 1, the BS computed from the PBQ was 0 because this child had undergone minor surgery without bleeding, therefore scoring ‘−1’and masking a score of ‘1’ due to epistaxis. The ‘−1’ score did not materially alter the results of any other patients. The AUC of the ISTH child score to detect a definite BD was 0.738.
Table 3. Bleeding scores by diagnosis
VWD Type 1
VWD Type 2
Mild F deficiencies
BD, bleeding disorder; VWD, von Willebrand disease; PD, platelet defect; F, factor; BS, bleeding score; PBQ, pediatric bleeding questionnaire ; ISTH BAT, bleeding assessment tool as proposed by a working party of the International Society on Thrombosis and Haemostasis . *Including patients with factor XII deficiency. †Family score consists of the score for the child and both natural parents.
N (Total 100)
Median BS PBQ child (range)
Median BS PBQ family† (range)
Median BS ISTH BAT child (range)
Median BS ISTH BAT family* (range)
The value of the family score Median family scores were generally higher than child scores, but were not significantly better than the child scores at differentiating between abnormal and normal bleeding in the child (AUC child score of 0.76 vs. AUC family score of 0.82 [P = 0.205]). Therefore, in the following analyses, the results of the ISTH child score are reported.
Bleeding score according to diagnosis The BS varied by final diagnosis, with median ISTH child scores beginning at 0 for no BD and increasing as follows: FXII deficiency (BS 0), possible VWD (BS 0), mild factor deficiencies (BS 1), possible PD (BS 3), VWD 1 (BS 3), and VWD 2 (BS4) (Table 3 and Fig. 2). Relative to children with no BD, the ISTH child scores were similar for children with possible VWD (P = 0.717) or mild factor deficiencies (P = 0.549), but were significantly higher for children with VWD 1 (P < 0.001) or VWD 2 (P < 0.001). The ISTH scores were lower in children with possible VWD than in children with VWD 1 (P = 0.009) or VWD 2 (P = 0.001). The BS also differentiated between VWD 1 and mild factor deficiencies (P = 0.020), and between VWD 2 and mild factor deficiencies (P = 0.050). Children with possible PD had higher scores than children with no BD (P = 0.001) and children with mild factor deficiencies (P = 0.033). There was no significant influence of age or sex on BS in children.
Comparison of the bleeding score with a simple Yes/No questionnaire (ITEM analysis) The sensitivity, specificity and predictive value of the quantitative ISTH child BS and the qualitative ITEM analysis are summarized in Table 4. The PPV of a BS ≥ 2 never exceeded 76.5%. The predictive value to exclude any mild BD was only 60.9%, but reached 90.6% to exclude any VWD. If transferred to a low prevalence setting as proposed by Tosetto et al. , assuming a prevalence of 1%, the predictive value to exclude VWD rose to 99.7% (95% CI, 99.4%, 99.9%).
Table 4. Sensitivity, specificity, positive and negative predictive value (+95% CI) for discriminating bleeding disorders from children without disorders using different criteria and cut-offs
Criteria, number of positive items or value of the bleeding score; SN, sensitivity; SP, specificity; PPV, positive predictive value; NPV, negative predictive value; ROC, receiver operating characteristics; ITEMS YES, number of qualitative bleeding symptoms (i.e. questions answered Yes); BS, quantitative bleeding score; VWD, von Willebrand disease. *Includes possible VWD, mild factor deficiencies, and possible platelet defects.
Any mild bleeding disorder * (n = 44)
Any VWD (n = 23)
VWD 1 (n = 12)
VWD 2 (n = 11)
The AUC value of the quantitative ISTH child score (AUC, 0.764) was not significantly higher than that of the qualitative ITEM analysis (AUC, 0.748; P = 0.3477).
Obtaining a detailed history is crucial to evaluating a potential BD . Because clinical symptoms can be subtle in children with mild BD, a standardized and quantitative approach, using questionnaires to calculate BS values, has been proposed. This also addresses the problem that some bleeding symptoms, such as epistaxis, are common even in healthy children . In our study, the scores used were easy and quick to apply.
As found by others, the BS values were able to discriminate between unaffected patients and those with a BD [3,6,7,9].
It has been debated whether a score of ‘−1’ in the PBQ is necessary. This score emphasizes situations in which a patient at risk did not bleed, and therefore the rate of false-positive results should be lower. In our study, few children actually scored ‘−1’. In one child with definite VWD 1, this led to a BS of 0. We conclude that it is reasonable to omit the ‘−1’ score, as with the ISTH BAT.
One drawback of the tools used is that, due to the age of the patient population studied and the associated lack of risk situations, the overall BS values are low, even in affected children. BS values in our study are lower than those described under study conditions by other authors, probably because the patients in our study were unselected (i.e. previously unknown to our department) and had a younger median age than in other studies , thereby resulting in lower BS values. Accordingly, the AUC values produced by ROC analysis were generally lower in this study compared with others .
We tried to address the problem of low BS values in young children by calculating a family score as a surrogate marker. Assuming that most definite BDs are inherited, we expected higher BS values in the parents. Indeed, while the family BS was higher than the child BS alone, this did not reach statistical significance and did not improve the accuracy of the tools in discriminating between no BD and BD for a child.
Interestingly, as found in a previous study , using only qualitative Yes/No criteria (ITEM analysis), the NPV and PPV of the questionnaires did not differ notably from the BS. This leads us to the suggestion that simple qualitative Yes/No questionnaires might be quick and helpful tools in routine clinical practice (e.g. prior to surgery). However, to allow comparison of studies and also to enable follow-up of the bleeding rate over time, we recommend using a qualitative scoring system under study conditions.
This study has some limitations. First, the inconsistent laboratory definition of VWD type 1 in the literature. To enable comparability, we used the original definition of the PBQ . Recently, a more strict definition has been used (< 0.3 U mL−1 for VWF:RCo and VWF:Ag) . Applying these criteria to our patients, the median BS for children with VWD type 1 was still 3, but the child with VWD 1 who scored a 0 in the PBQ would be redefined as possible VWD. However, we believe that, in the pediatric population, it might be safer to accept higher values because VWF levels could rise due to the stress of drawing blood. This is also the reason why we in general define possible VWD in routine clinical practice as BD, especially prior to surgery. Because VWF acts as an acute phase protein, single measurements might show VWF values above 50% and do not discount VWD.
Second, the study lacks a predefined control group. Patients defined as no BD in our study are not rated healthy by their clinical symptoms, but by their normal laboratory values. However, especially in children with repeated infections, the laboratory values might vary between measurements and therefore some normal values might be in fact false-negative. As in most of the cited literature, our group of normal patients (no BD) showed low BS values (mean 0.6), while a recent study in a comparable setting reported a higher mean (2.51) .
This issue might be attributable to the third limitation, which is generally applicable to BS studies carried out in specialized centers. Although we consecutively enrolled unselected patients, they had in effect already been selected by their primary care physicians. This bias probably leads to higher BS values than in the general population because children with no abnormalities are not referred. This might also explain the unusually high rate of patients diagnosed with VWD in our study. While it appears that more severe BDs, such as VWD 1 and 2, can be excluded by the tools examined in our study, it is unclear if these tools will be valuable for detection of milder BD or BD with very low prevalence. The low values of the BS for children with mild factor deficiencies can be explained by the relatively high factor levels, between 30 and 40%, generally not associated with substantial bleeding .
Our results are similar to those published by Tosetto et al. . These authors evaluated the use of a similar BAT in a prospective cohort of children and adults referred for evaluation of bleeding symptoms, abnormal laboratory tests or family investigations. The probability of detecting a BD was especially high in patients referred due to bleeding symptoms, which was confirmed by our study. We found fewer BD in children referred for family investigation, probably because in our cohort family studies were conducted due to a clinical bleeding history and not a laboratory-defined BD.
We conclude that the ISTH BAT could be useful in the clinic for deciding whether extended laboratory work-up is reasonable. It correlated well with previously published tools, such as the PBQ. To improve comparability of studies, we encourage the use of the ISTH BAT in further studies for quantification of bleeding symptoms in different cohorts.
C. Bidlingmaier designed the study, recruited and interviewed participants, collected blood, reviewed and interpreted data, and wrote the manuscript. V. Grote analyzed and interpreted data. U. Budde performed the analysis of VWF multimers, provided clinical information, and critically revised the manuscript. M. Olivieri recruited and interviewed participants, collected blood, reviewed and interpreted data, and revised the manuscript. K. Kurnik supervised research, helped to design the study, reviewed and interpreted data, and critically revised the manuscript. All the authors approved the final version of the manuscript.
This work was supported by a research grant from CSL Behring (Hattersheim, Germany). The investigators analyzed and interpreted all data; analysis was not influenced by the funding source. The editorial assistance of Trilogy Writing and Consulting GmbH, Frankfurt am Main, is acknowledged. Costs were covered by CSL Behring. We thank B. Nahr and S. Jenkins from our team for documentation assistance.
Disclosure of Conflict of Interests
The authors state that they have no conflict of interest.