Hereditary haemorrhagic telangiectasia: a population-based study of prevalence and mortality in Danish patients


Dr Anette Drøhse Kjeldsen, Rulkehøjen 44, DK-5260 Odense, Denmark.


Introduction. Hereditary haemorrhagic telangiectasia (HHT) is a dominantly inherited disease characterized by telangiectatic lesions. The disease manifestations are variable and include epistaxis, gastrointestinal bleeding, pulmonary arteriovenous malformations and cerebral arteriovenous malformations. Early death due to these complications has been described.

Design. We report a study on the prevalence and mortality of HHT in a Danish population based on two cross-sectional surveys in combination with a long-term follow-up study.

Settings and subjects. Prevalent cases of HHT as of 1 January 1974 in the County of Fyn, Denmark, were identified. In 1995–97 a follow-up study of mortality was performed of the initial patient sample, and a new point prevalence rate of HHT as of 1 January 1995 was calculated. All live patients and their families were invited to attend a detailed clinical examination.

Results. The prevalence of HHT in the County of Fyn was 13.8 per 100 000 on 1 January 1974 and 15.6 per 100 000 on 1 January 1995. In the HHT group as a whole, we found a slightly increased mortality; however, amongst the HHT patients younger than 60 years at inclusion the mortality of HHT patients was twice the expected. The excess mortality could be fully explained by severe HHT symptoms.

Conclusion. This study suggests that HHT is more prevalent than previously believed. In young patients the disease is associated with an excess mortality which is fully attributable to HHT. Future research should aim at the identification of HHT patients at particular risk of developing severe complications.


Hereditary haemorrhagic telangiectasia, also known as Osler–Weber–Rendu disease, is a rare dominantly inherited disease. It is a distinct clinical entity described by Sir William Osler [ 1] at the turn of the century, in terms of its clinical features, telangiectasis and the familial occurrence.

Only a few studies describing the epidemiology of the disease have been reported. The prevalence of HHT has previously been estimated to be between 2.5 and 19.4 per 100 000 [ 2–6]. Between one- and two-thirds of HHT patients will seek medical advice on treatment [ 7].

Clinical characteristics and genetics of HHT

In HHT patients, telangiectatic lesions are often present in the oral mucosa and the nasal mucosa, and leads to epistaxis in 90% of patients and occasionally also to oral bleeding [ 3, 8]. Epistaxis beginning in the teenage years is the first symptom in the majority of HHT patients [ 3, 7, 8]. Telangiectatic lesions in the gastrointestinal tract cause symptoms in 20–30% of patients [ 9]. With age, this complication may result in severe anaemia, and a high requirement for transfusion in later life is not uncommon [ 10, 11].

Pulmonary arteriovenous malformations (PAVMs) and cerebral arteriovenous malformations (CAVMs) represent other severe manifestations of the disease [ 12, 13]. Patients with one or both of these complications risk early death from rupture of the malformations. In particular, these complications may be fatal during pregnancy [ 14, 15]. However, studies on mortality rates in patients with HHT have been very scarce.

During the last few years, the genetics of HHT have been studied extensively. Currently, at least two different chromosomal loci have been implicated. In HHT type 1, mutations at chromosome 9 alter the coding sequence of endoglin [ 16, 17]. In HHT type 2, mutations at chromosome 12 alter the coding sequence of the activine receptor-like kinase 1 (Alk 1) [ 2, 18]. In both types, the mutations impair blood vessel growth and repair, but it has recently been established that patients with HHT 1 more frequently develop PAVMs than those with HHT 2 [ 19, 20].

Objectives of the present study

The aim of this study was to establish the prevalence of HHT as of 1 January 1995 and to compare this with the prevalence as of 1 January 1974 in the County of Fyn, Denmark. In parallel, a long-term follow-up study of the initial patient sample was carried out in order to study the mortality in HHT patients.


The County of Fyn, Denmark, with 470 000 inhabitants, is comparable to the total Danish population. The region comprises 8.1% of the total area and 8.8% of the population of Denmark. The population of the County of Fyn is served by the regional public health care system, in which Odense University Hospital is the main hospital. Another nine smaller hospitals are located in the county. Since 1974, civil registration number, address and diagnosis at discharge have been recorded in all patients. The Regional Patient Registration System therefore identifies all patients with discharge diagnosis after 1 January 1974. The first part of the study was conducted in the period 1 January 1974 to 25 April 1979 [ 7]. The follow-up was conducted in the period 1 January 1995 to 1 January 1997. Informed consent was obtained from all participants. The study fulfilled the Helsinki II declaration and was approved by the Regional Ethic Committee for the Vejle and Fyn Counties.

Diagnostic criteria and disease rating

It is generally accepted that the diagnostic triad of symptoms in HHT consists of (1) presence of multiple telangiectatic lesions, (2) heredity, and (3) epistaxis or other recurrent bleeding. Most authors require at least two of the three criteria to be fulfilled to establish the HHT diagnosis. In the present study we based the diagnosis on (1) presence of multiple, at least 15, telangiectatic lesions, and (2) either a positive family history or recurrent episodes of bleeding. The criterion of heredity was met if the presence of telangiectatic lesions in at least one first-degree relative could be demonstrated. In the case of recurrent bleeding, telangiectatic lesions at bleeding site were required.

The severity of HHT-related symptoms was rated similarly in the two cross-sectional studies. Patients were rated as grade 1 if they had no or only discrete symptoms. Patients with weekly episodes of bleeding leading to slight anaemia and need of medical care on an outpatient basis were rated as grade 2. Finally patients were considered to have grade 3 severity if they had daily or weekly episodes of severe bleeding which had required hospitalization within the previous 5 years or if they had symptomatic PAVMs or CAVMs.

Identification of patients

In the first survey, all HHT patients registered in the County of Fyn were identified. The proband criteria were: (1) fulfilling the diagnostic criteria of HHT; (2) first time of diagnosis before 31 December 1973; and (3) address in the County of Fyn at inclusion. The probands were identified through 219 written inquiries about recorded data on HHT. The inquiries were sent out to all medical institutions (48 departments), all specialists in private practice (26), and all general practitioners (145) in the County of Fyn. All inquiries were answered. In the subsequent survey in 1995, a search in the national patient registration system was performed on the HHT diagnosis (code 448.00, and after 1 January 1994, code I 78.0). This revealed a population of HHT patients comprised of previously identified as well as newly registered patients. The same proband criteria were applied as in the first survey; however, the time period for first diagnosis of HHT was extended until 1 January 1995. In both study periods, all probands alive underwent a clinical evaluation and information about relatives were obtained. All first-degree and more distant relatives with a history of HHT were invited to participate in the study.

Prevalence analysis

In the first cross-sectional survey (1 January 1974) only probands and first-degree relatives were included in the calculation of the disease point prevalence rate, and therefore this estimate represents the minimum prevalence. In the second cross-sectional survey (1 January 1995), all known HHT patients (probands, first-degree and more distant relatives) who were alive with an address in the County of Fyn were included. Confidence limits of 95% were calculated based on the assumption of a Poisson distribution.


All patients included in the calculation of the disease point prevalence rate as of 1 January 1974 were included in the follow-up study. The time of inclusion was the date of the initial clinical examination by one of the investigators (P. Vase). The patients were followed until 1 January 1997 or until death, if it came earlier. In the present survey all patients were reassessed. If a patient had died, hospital records and death certificates were required and reviewed. If the hospital record or results of postmortem examination suggested another more likely cause of death than that stated on the death certificate, the diagnosis on the death certificate was omitted. Untreated PAVM causing dyspnoea in later life was considered as contributing to death. The severity of bleeding, epistaxis and gastrointestinal, was assessed based on the number of blood transfusions received. Patients were rated as having severe bleeding which contributed to death if they had received 6 units of blood or more during the 6 months prior to death. A Kaplan–Meyer plot was calculated and compared to a calculated life table using the statistics of age- and gender-matched controls in Denmark in the period 1979–80 [ 21]. The patients subjected to follow-up were divided arbitrarily in two groups, with age under and above 60 years at entry.


Identification of patients

The inquiries revealed 47 probands representing 22 families. Forty-one were alive on 1 January 1974, and they were all examined in the first study period. First-degree relatives to probands also underwent clinical examination and a total of 35 first-degree relatives with HHT were identified. Twenty lived in the County of Fyn and fulfilled the diagnostic criteria of HHT as of 1 January 1974.

At the clinical re-examination in the second study period, two probands and two first-degree relatives initially classified as HHT patients were excluded from further investigation and analysis, because they did not fulfil the diagnostic criteria of HHT any longer. Three of them belonged to the same family in which no bleeding episodes had occurred in any family member during the follow-up period; furthermore, telangiectatic lesions could no longer be demonstrated in any family member. The fourth individual excluded was a male who in the late teens had suffered from recurrent epistaxis. At follow-up telangiectatic lesions could not be demonstrated and he had epistaxis only twice a year.

The search for HHT diagnosis in the regional in-patient register on 1 January 1995 revealed a total of 107 patients assigned with the HHT diagnosis. Among these, 12 patients were not residents of the County of Fyn. In 30 cases the diagnosis could not be confirmed at clinical examination or by review of hospital records. Among the remaining 65 HHT patients, 28 were already registered as probands in the first survey. Thus, the register search added a total of 37 new probands, and the total number of probands in the two studies was 82; only 29 of all probands were alive on 1 January 1995. The majority of probands were identified as they had been hospitalized due to anaemia at a medical or otorhinolaryngological department.

Prevalence rates

At 1 January 1974, the prevalence of HHT in the County of Fyn was 13.8 (95% confidence limit = 10.5–17.7) per 100 000 population, as compared with 15.6 (95% CI = 12.2–19.6) per 100 000 population as of 1 January 1995. Age-specific prevalence rates were calculated and showed increasing prevalence of the disease with age. Thus the prevalence was highest in the +50 group (28.8 and 26.3 per 100 000 population in 1974 and 1995, respectively) (see Table 1). There was no statistically significant difference between the prevalence rates when stratified for age in 1974 and 1995 (χ2= 0.134, 1 d.f., P= 0.714).

Table 1.   Age-specific prevalence rates of HHT in the County of Fyn, Denmark at 1 January 1974 (1/1/74) and 1 January 1995 (1/1/95) Thumbnail image of

The distribution of patients according to severity of symptoms in 1974 and 1995 did not differ when tested for trend (χ2= 1.5, 2 d.f. P= 0.471) ( Table 2).

Table 2.   Severity of HHT symptoms Thumbnail image of

Follow-up results

The final follow-up population consisted of 57 HHT patients. Two had moved abroad, the dates of their last registration alive in Denmark were considered as withdrawal. Table 3 shows the results of the mortality analysis. Overall, 36 HHT patients had died, whereas only 25.8 were expected (P= 0.045) ( Fig. 1). The increased relative mortality could be attributed to an excessive number of deaths in the subgroup of patients (27 individuals) younger than 60 years at inclusion in 1974 (P= 0.014). In this group, similar relative mortalities were estimated for male and female patients, but in this respect were only statistically significant for males due to a small number of deaths amongst female patients. Amongst the 12 deceased patients younger than 60 years at entry, eight (67%) were considered to have HHT disease grade 3 ( Table 2). In the subgroup of patients aged 60 years or more at inclusion, comprising 30 individuals, the relative mortality was not statistically significantly increased (P= 0.348); this result applied to male as well as female patients. Amongst the 24 deceased patients in this subgroup, 12 (50%) were considered to have HHT disease grade 3 ( Table 2).

Table 3.   Results of mortality analysis Thumbnail image of
Figure 1.

 Kaplan–Meyer survival curve for 57 Danish HHT patients vs. expected. The first clinical examination occurred in the period 1 January 1974 to 25 April 1979. The patients were followed either until 1 January 1997 or until death if it came earlier. The expected survival was calculated from reference values [ 22].

The cumulative mortality rates, according to age at inclusion, are illustrated in Figs. 2 (A) and (B) (age < 60 years and > 60 years at entry, respectively). The curves show the total number of deaths as well as the HHT-related deaths compared with an expected cumulative mortality calculated from reference values. The median follow-up time for the 21 HHT patients who survived was 18.9 years (mean 18.7, range 7.53–22.1), including two patients who had moved abroad.

Figure 2.

 Cumulative mortality amongst 27 HHT patients aged < 60 years at entry (A) and amongst 30 HHT patients aged > 60 years at entry (B). The expected mortality rates were calculated from reference values [ 22].

Causes of death

On reviewing hospital files and death certificates of 36 dead HHT patients, HHT was considered a contributing cause of death in 13 patients (36%). These patients had all required more than 6 units of blood during the 6 months prior to death. In two patients there was a history of epistaxis but no history of gastrointestinal bleeding, and in seven patients gastrointestinal bleeding was the main source of bleeding but epistaxis may have contributed to the anaemia. In the remaining four patients, the anaemia was caused by severe gastrointestinal haemorrhage as only minor episodes of epistaxis were reported ( Table 4).

Table 4.   Causes of death in 36 HHT patients Thumbnail image of

In five patients a diagnosis of PAVM had been established; two had previously had their lung malformation resected, however, both had also suffered from severe progressive dyspnoea after surgery. Based on hospital files the pulmonary insufficiency was considered to contribute to the cause of death; however, clinical examination for recurrence of PAVM had not been performed. Three patients had untreated PAVM and progressive dyspnoea; PAVM as well as bleeding were considered to contribute to death in these patients.

In one patient a diagnosis of CAVM had been established whilst the patient was alive. The patient also had multiple sclerosis and was severely disabled. PAVM was considered an optional diagnosis, but never established. The patient died in 1977 due to a pulmonary abscess and probable malignant disease. In the follow-up study, HHT was not considered to contribute to death.


Identification of patients and estimation of prevalence

Completion of a study such as the present one depends upon an easy access to patient information, in this case derived from a well-organized health care system and a well-defined comparable background population. The geographical setting of this study, the County of Fyn, fulfils these criteria.

The variability in the clinical manifestations of HHT represents a potential source of selection bias. In fact, 42% of prevalent HHT patients at 1 January 1995 had never required medical care, and in 18 of 75 patients the HHT diagnosis was established for the first time at inclusion in the present survey. However, previous studies have demonstrated a large within-family variation in manifestations of HHT [ 3, 13]. Thus it seems unlikely to imagine an HHT family where no family member at all had required medical care for HHT-related symptoms over a study period of more than 30 years. This assessment is supported indirectly from the results of our study. First, the point prevalence estimates of HHT were similar for the two cross-sectional surveys conducted with an interval of more than 20 years. Second, the present cross-sectional survey did not identify any families that fulfilled the criteria for inclusion but were missed by the first survey. It is therefore most likely that the present study material is virtually complete.

The prevalence of HHT in the County of Fyn, Denmark, was 13.8 per 100 000 population as of 1 January 1974 and 15.6 per 100 000 population as of 1 January 1995; this is comparable to studies based on epidemiological designs but rather high compared with other studies. This may reflect differences in study designs, possibly combined with geographical variation in the distribution of HHT. In the present study, all HHT patients were included regardless of severity of disease. If only patients representing grade 3 severity of symptoms had been included, the prevalence rate would have been 5.6 per 100 000 population. In Haut de Jura and Ain, France, a prevalence of 12.5 per 100 000 population was reported by Vincent et al. [ 2] and Plauchu et al. [ 3]. A prevalence of 19.4 per 100 000 population was found in the Dutch Antilles [ 4]. These studies, like the present Danish study, aimed at inclusion of all HHT patients. Porteous et al. [ 22] calculated a prevalence of 2.5 per 100 000 population in the Newcastle area. The patients included in that study were all identified by questionnaire mailed to general practitioners, otorhinolaryngologists and haematologists in the north of England. The authors concluded that the prevalence was the first estimate of HHT in Britain and almost certainly an underestimate. In the state of Vermont, USA [ 5], a minimal prevalence of 6.1 per 100 000 population was estimated, and the authors state that they are confident that the true prevalence is significantly greater, since mostly patients with significant symptoms were included. Therefore, the prevalence of HHT in the present survey is probably a good estimate of the overall prevalence of HHT.

In HHT patients, the penetrance of the disease is age-dependent and at this point no diagnostic blood test is available. The diagnostic criteria were a tool used at clinical examination in order to establish a correct diagnosis. The fact that four patients previously met the diagnostic criteria of HHT but not at the follow-up examination illustrates the problem in establishing an exact diagnosis of HHT based upon clinical findings at a single outpatient examination. It seems reasonable to apply the Dutch policy for patients in whom the HHT diagnosis is uncertain, namely to examine the patients again after a 3–5 year period [ 12].


Fatal complications of HHT have been reported in a number of studies, such as pulmonary haemorrhage during pregnancy [ 14, 15] and cerebrovascular haemorrhage due to CAVM [ 4]. Clinical evidence also suggests that severe gastrointestinal bleeding may cause fatal complications [ 10, 11, 23]. These severe complications have led to the suggestion of an increased mortality amongst HHT patients [ 19]. However, this has never been documented in previous studies.

We found that HHT is associated with an excess mortality. This was most pronounced amongst the HHT patients below the age of 60 years at inclusion. In this group the mortality rate was twice as high as expected and the excess mortality was fully attributable to HHT. Among the 13 deceased patients in whom HHT was considered to contribute to death, severe bleeding (notably gastrointestinal) was always present, and in three patients there was additionally a history of untreated PAVM causing respiratory symptoms. We did not identify any death due to CAVM; one patient had CAVM diagnosed but died due to malignant disease. Severe symptoms of HHT at a young age seemed to predict early death although this was not significant due to small numbers. Age at onset of symptoms did not seem to be indicative of the course of disease as the average age at debut was the same in the three severity groups (data not shown). The rating of severity of HHT supports the fact that in many HHT patients the symptoms of disease tend to aggravate with age, since patients with mild disease rated as grade 1 are considerably younger than patients with severe disease rated as grade 3 (data not shown).

The increased mortality amongst HHT patients younger than 60 years at inclusion may reflect a survivor bias due to the fact that the present study population consists of both HHT type 1 and HHT type 2 patients. As reports indicate that HHT 1 patients have a more severe disease [ 19, 20], it is possible that patients aged over 60 at inclusion predominantly consist of HHT type 2 patients. However, at the time of inclusion genetic typing of patients was not possible and at the end of the study period two-thirds of the patients had died.

Presently no treatment can prevent the development of telangiectatic lesions in HHT patients. However, serious complications of PAVM and CAVM may be prevented by early detection and treatment [ 4, 24–27]. Treatment of patients with severe gastrointestinal bleeding with oestrogen-progesterone has shown promising results in one study [ 23]. In another study bleeding control was obtained with laser therapy [ 28]. Furthermore, epistaxis may be treated with laser or septal dermatoplasty procedures [ 29, 30]. Effective methods to prevent severe bleeding are, however, not available.

Thus, we encourage physicians dealing with HHT patients to recognize the complexity of the disease, especially the pulmonary and gastrointestinal lesions. The aim of future studies should be to identify patients at particular risk, and treatment strategies should address HHT complications with a substantial morbidity as severe gastrointestinal bleeding and PAVMs.


The study was supported by: the patients and their families with their continuing interest and cooperation; the Foundation of Medical Research in the County of Fyn; the Carla Cornelius Storch Møllers Foundation; the Danish Medical Association Research Foundation; the Research Foundation of Congenital Diseases (forskningsfonden vedrørende medfødte sygdomme); the Ingemann O. Buchs Foundation; and the Højbjerg Foundation. ADK was supported by a research fellowship at Odense University.

Received 17 September 1997; accepted 3 February 1998.