1We have already explained the form of epidemiological investigation as well as the discovery of the geographical area in which the disease was concentrated in a previous paper [1] in this journal in 1984. In that paper 14 départements were studied. "The area of concentration" of the disease was specifically investigated (in the départements of Ain and Jura) as well as adjoining areas ; also included were the two "control" départements of Meurthe-et-Moselle and Deux-Sèvres. The probability of there being a concentration of this disease in the last département encouraged us to carry out further geographical exploration in order to study the geographical distribution of this inherited disease.
2An epidemiological investigation in 52 départements enables us to approach the study of the prevalence of this inherited disease.
I – Mail survey procedures
1 – An empirical approach
3The method used was naturally adapted to the clinical profile of Rendu-Osler disease, as in all suggested procedures in the epidemiological investigations of genetic diseases. For example, J. Sutter’s work on the dysplasia of the hip and the agenesis of upper lateral incisors [2], J. Feingold on cystic fibrosis [3], E. Dubois on cystinosis [4], C. Bonaiti on glaucoma [5], and E. Shields on cleft palate [6], as well as a number of investigations conducted by Unit 155 of INSERM (epidemiological genetics).
2 – Practical procedures
Mailing
4All physicians likely to come across cases of Rendu-Osler disease in their practices were contacted. A questionnaire was sent to them together with a circular which showed how to diagnose the disease. The circular stressed that even negative answers would be useful for analysis. A self-addressed stamped envelope was enclosed to encourage physicians to participate.
5For each département, lists of practitioners were provided either by the Direction Départementale de l’Action Sanitaire et Sociale (DDASS) or by the départment ’s medical association or failing that, Rosenwald’s directory of the medical profession was used. The telephone directory was also used to complete the lists. Eventually, specialists who had no reasons for seeing patients suffering from Rendu-Osler disease were excluded from this list (for example, paediatricians, psychiatrists, or orthopaedic surgeons), as were some who were eliminated because of the nature of their practice : junior medical staff, retired or non-practising physicians.
6Lastly, those whose contact with patients suffering from Rendu-Osler disease was very limited were also excluded, for example anaesthetists, surgeons, except those specializing in cardio-pulmonary and neuro-surgery who might diagnose Rendu-Osler symptoms while treating arterio-veinous complications [7]. Unfortunately, it is only in hospitals that surgeons’ specialization is officially recorded.
7Altogether nearly 60 % of practitioners were contacted. Some 23,435 circulars were mailed in the 52 départements studied.
Processing of replies
8The replies were analyzed on the basis of their contents, those giving information on one or several patients were processed in a specific manner. With the help of multiple-entry lists (patronymic, geographical location of family) efforts were made to see whether the patient mentioned was related to a family previously identified as containing sufferers from the disease. Among the patients identified in the mail survey, cross-checking of initial data helped avoid starting new files on patients whose families were already known to us through other members.
9In fact one member in each family was selected. His case notes included his personal data, as well as data on those related to him who had been identified in the epidemiological study, and by the different procedures used. Thus a family genealogy was established, indicating the numbers of both live and deceased patients as well as the degree of seriousness of their illness. A report on the patient’s disease-related experiences was also enclosed.
10If the physician replying to our questionnaire had obtained the consent of one or several of his patients, we wrote to him or them to ask for additional information about his/their family tree. In fact, the detail provided by different patients differed, and so did the value of information collected from them.
3 – The choice of départments
11Analysis carried out in 14 départements had revealed the likelihood of a second concentration of Rendu-Osler disease in the département of Deux-Sèvres.
12To verify this hypothesis, we investigated adjoining départements and tried to determine whether there existed a "transmission line" linking the two concentrations, whilst completing our investigation of North-Eastern France, which has been traditionally linked with the Ain-Jura concentration. Poitou-Charentes, Centre, Limousin, Auvergne, Bourgogne, Rhône-Alpes, Franche-Comté, Alsace, Lorraine, Champagne-Ardenne and Picardie were totally covered. Our investigation also included the départements of Eure, Manche, Maine-et-Loire and Vendée. Ile-de-France was omitted because of problems specific to the area (population density and medical organization chart). The investigation in the South-West began with two départements, Ariège and Aude.
13The départements under investigation cover a surface area of 313,026 sq.km or 57 % of metropolitan France with a population of 23,074,000 or 42 % of the total population (see Map 1). The population in these départements varies from 132,000 inhabitants (in the Territoire de Belfort) to 1,445,000 (in Rhône) and between these two départements, the number of physicians contacted varies between 150 and 1,023. The population density shows great contrasts, from 450 inhabitants per sq.km in Rhône to 25 in Creuse. On the other hand, the ratio between the number of physicians contacted and the population in a département is less variable, one doctor for every 435 inhabitants in Bas-Rhin to one for every 1,648 in Haute-Savoie.
Extension of the survey to 52 départements

Extension of the survey to 52 départements
II – The replies
1 – Percentage of replies received
1423,435 circulars were sent out in 52 départements, 5,745 of them in the 14 départements investigated in 1984, and 17,690 in the 38 new ones. The response ratio has fallen slightly from 15.7 % in 1984 to 14.2 % today. If we were to take into account only those physicians who were actually contacted, i.e. excluding those who had moved away or died, the response ratio would come to 14.7 % (see Table 1).
15The ratio differed considerably in different départements. In the first part of the investigation, it varied between 24.7 % in Haute-Savoie and 9.0 % in Doubs. In the 38 new départements, it varied between 21.6 % in Loir-et-Cher and 5.5 % in Cantal. It would be tempting to conclude that the physicians in the départements that were most affected would be more likely to participate in the investigation than others. In fact, no correlation was observed between the number of those who replied and the fact that they had had patients suffering from the disease. Out of 3,331 replies received 84.6 % came from those who had never had such patients (1,761 + 1,103, Table 1).
16There was also no relation between the reponse ratio and the availability of medical facilities, the density of physicians or the existence of a teaching hospital in the département. However, the response ratio was lowest in those départements in which information about specialists and the conduct of medical practice was least precise.
2 – Practitioners who responded
17An earlier investigation based on 324 individuals showed that Rendu-Osler disease often went untreated [8]. The families knew that the disease was hereditary, and therefore inevitable, and that there was no real cure. Provided the symptoms were not serious, which was true of the majority of cases, patients consulted a general practitioner who prescribed long-term palliative treatment, and showed the family simple ways of stopping nose bleeding, the most troublesome symptom of this condition.
18A fraction of the patients with more severe symptoms, depending on the nature of the complications, saw the specialist most suitable for them : a dermatologist for those with telangiectasia problems, an ear, nose and throat (ENT) specialist for nasal haemorrhage and angioma of the mucous membranes of the nose and mouth, a gastro-enterologist for angioma in the digestive system.
Statistics of the mail survey

Statistics of the mail survey
N.B. :- Category 1 : Practitioners having already treated such patients.
- Category 2 : Practitioners ready to participate in the survey in the future.
- Category 3 : Other respondents.
19An analysis of the replies by practitioners who had treated such patients confirmed this approach. 512 replied that they had treated patients suffering from Rendu-Osler disease (see Table 2). The majority were general practitioners : they numbered 331, or nearly 65 % of all responding practitioners. Among the specialists, dermatologists (43) most frequently treated such patients, followed by ENT specialists (43), while those who treat digestive complications were next on the list (gastro-enterologists, specialists of the digestive system and internal organs).
Distribution of responses by category of practitioners (answers of type 1)

Distribution of responses by category of practitioners (answers of type 1)
20Some patients were reported several times : the same name was often reported by a general practitioner and a specialist. Dermatologists and ENT specialists often saw several patients during the course of their practice. This is especially true of well-known specialized units in teaching hospitals which often have up to ten patients on their files.
21Replies from practitioners of occupational medicine play a major role in investigations of this type. They screen patients systematically and negative replies from them can be significant ; five such doctors had seen Rendu-Osler patients.
3 – Reported cases of Rendu-Osler disease
The size of the mail survey
22Up to now 627 cases have been identified throughout France. Mail surveys in 52 départements showed 392 individuals residing in 62 départements [9]. Some practitioners, particularly those attached to hospitals, often see patients who have travelled a long way for a consultation.
Place of residence and place of reporting in the mail survey

Place of residence and place of reporting in the mail survey
Individuals’ place of residence
23There is often a disparity between the number of Rendu-Osler patients residing in an area and the number reported by practitioners from that area. This is true of Maine-et-Loire (16 compared with 25) and Isère (8 compared with 13). The reverse is true for rural départements (20 compared with 15) in Deux-Sèvres. It was thought preferable to base the analysis on the individuals’ place of residence.
24The numbers of individuals reported in the epidemiogical surveys were highest in the départements of Ain with 31, and Rhone and Deux-Sèvres with 20 each. They were followed by Haut-Rhin, Maine-et-Loire and Savoie. The geographical distribution is rather irregular. But the Ain-Jura concentration with its extension into Rhône, remains preponderant with 63 individuals, and the existence of the disease in the Deux-Sèvres, Maine-et-Loire axis extending into Vienne and Charente, is confirmed with its 58 individuals.
Survey results
25Besides the total number of known cases, two indicators help us to assess the results of the epidemiological investigation.
26• The ratio between the number of practitioners treating sufferers from Rendu-Osler disease and the population of the département is the first (Table 4, left third).
27This ratio varies between 1/18,500 in Loir-et-Cher and 0 in the Territoire de Belfort. In six départements, the ratio is higher than 1/25,000 : Loir-et-Cher has already been mentioned, Deux-Sèvres (1/20,000), Jura (1/22,000), Corrèze (1/24,000), Ain (1/24,600) and Savoie (1/24,800).
28• The ratio between the number of practitioners treating Rendu-Osler disease and the total number of practitioners contacted has also been computed (central part of Table 4).
29This ratio ranges from 7.1 % in Jura to 0 in the Territoire de Belfort. In only four départements is the ratio higher than 5 % : in Jura (7.1 %), Ain (6.7 %), Loir-et-Cher (5.8 %) and Deux-Sèvres (5.7 %).
30None of the three parameters shown in the table is ideal, but by using them together it is seen that the same départements always seem to be at the top of the list : Ain, Jura, Deux-Sèvres, Maine-et-Loire. These are followed by Savoie, Haute-Savoie and Rhône for the Rhône-Alpes region, Maine-et-Loire and Eure-et-Loir in the west, and other "high-risk" départements such as Haut-Rhin and Vosges for the North-East and Corrèze for Limousin.
III – Prevalence estimates
31• Prevalence ratios may be compared by using different results from the epidemiological study. For the 52 départements covered by the mail survey, "minimum levels of prevalence" could be obtained from the survey alone.
The 17 highest response rates in the survey

The 17 highest response rates in the survey
32• In Rhône-Alpes and Deux-Sèvres, "minimum prevalence" could be computed after taking account of the number of patients identified by the mail survey, as well as those obtained by the procedures mentioned below :
- from consulting nosological indexes of the different departments of regional hospitals treating Rendu-Osler patients, regional university hospital centres in Lyons, Besançon, Poitiers and Geneva, and regional hospital centres in Oyonnax, Saint-Claude, Lons-Le-Saunier, Saint Julien, Annemasse and Annecy ;
- from local family history surveys among families whose members were already known to be suffering from this disease ;
- by contacting local practitioners directly ;
- by presenting this research project to associations (like the Lyons ENT Specialists Association or to the French Society of Haematologists) and during postgraduate training sessions ;
- by conducting "door-to-door" medical surveys among all the families living in villages in the heart of the geographical concentrations ;
- by special surveys in two Paris hospitals to find out the exact geographical origin of patients treated in the establishments in the capital ;
- by writing to all families living at a distance that could not be contacted in person ;
- by organizing the prospective recruitment of all new patients coming for treatment or hospitalization in the Lyons area.
33L. Venturini has shown that on an average there were four living patients for every reported case [10]. If all these related patients were residents of the same département, the prevalence ratio could be obtained thus :
1 – Minimum prevalence
35Three départements stand out very clearly with prevalence ratios of 1/3,375 for Ain, 1/4,287 for Deux-Sèvres and 1/5,062 for Jura. In ten others, prevalence ratios vary between 1/6,000 and 1/13,000.
36Three regions which might in the future become areas known for Rendu-Osler disease are singled out : Limousin (Creuse, Corrèze, Haute-Vienne), the eastern slope of the Massif Central (Haute-Loire, Ardèche) and the Massif Vosgien (Vosges, Haut-Rhin, Map 2, Table 5).
Minimum prevalence of Rendu-Osler disease in 52 départements

Minimum prevalence of Rendu-Osler disease in 52 départements
37Both concentrations found in Ain-Jura and Deux-Sèvres are surrounded by départements with low prevalence ratios of between 1/13,000 and 1/21,000. This fall is accentuated by the low prevalence ratio in Haute-Marne, Haute-Saône and in the Territoire de Belfort which separates Vosges from Jura. The same phenomenon is observed between Jura and Limousin. Finally at the département level, we observe what seems to be a break rather than a progressive tapering off as the disease spreads from one or two original areas of geographical concentration.
38The salient point is the identification of a large concentration of patients in Ain-Jura and Deux-Sèvres and the emergence of three areas characterized by a greater number of patients in Limousin, Vosges and the eastern Massif Central.
Minimum prevalence of Rendu-Osler disease by département

Minimum prevalence of Rendu-Osler disease by département
39Naturally, mobility interferes with the analysis of the geographical distribution of patients by place of residence. A number of patients from rural areas have become assimilated in Paris and Bordeaux. The case of Limousin is especially ambiguous. The lowest prevalence noted was in Creuse : 1/8,750 and in Corrèze : 1/10,062, rather high for the "postal epidemiological survey". These two départements have been particularly hard-hit by the rural exodus. It is difficult to estimate the prevalence ratio a century ago. Could it be an old focus of infection which is about to disappear owing to the rural exodus ? Or, does the higher proportion of old people among whom the chances of the disease being diagnosed are higher suggest an exaggerated view of the situation ?
40Obviously we are trying to trace inter-regional links that would help us follow the spread of genes from a suspected focus. Only a systematic genealogical approach would make it possible to confirm or reject the theory of families’ regional roots [11]. Moreover, an analysis of the distribution of patients by age should be carried out.
Minimum and approximate prevalence in 8 départements

Minimum and approximate prevalence in 8 départements
41Since there is no evidence to show the progressive spread of the disease from one focus to another, we must admit that several mutations have occurred in different places at different times. However, recent mutations cannot be held responsible for a significant number of cases.
2 – Approximate prevalence
42The Rhône-Alpes region in which the original concentration of the disease was discovered and the region of Deux-Sèvres have been particularly well surveyed and all possible methods of investigation were used to account for all sufferers from Rendu-Osler disease [12]. In the départements of Ain, Jura and Rhône, the number of cases reported in the mail survey was multiplied by three and in five other départements by a factor varying between 1.8 and 2.8.
43The next stage in our epidemiological research would be to use our different procedures for Limousin, the Massif Vosgien and the eastern Massif Central.
44We stress that there have also been considerable variations within départements. Thus in the département of Ain, the arrondissements of Bourg and Belley were relatively little affected by the disease with a prevalence ratio of 1/3,506, whereas in the arrondissements of Nantua and Gex, prevalence ratios are very high at 1/327. For regions with high ratios, an analysis at the arrondissement or even at the canton level is needed.
Prevalence of Rendu-Osler disease by arrondissement

Prevalence of Rendu-Osler disease by arrondissement

45Both the concentrations become apparent when we look at the situation beyond départements. The original focus covers the north-eastern part of Ain (arrondissements of Gex and Nantua) and southern Jura (arrondissements of Saint-Claude and Lons) ; it extends east into the arrondissement of Saint-Julien-en-Genevois (in the département of Haute-Savoie) and north-east into the arrondissement of Louhans (in the département of Saône-et-Loire).
46Thus the arrondissements with the highest prevalence ratios (Gex 1/300, Nantua 1/346, Saint Claude 1/530) are surrounded by a zone in which there is a steeply declining trend of the disease. Beyond a 50 km radius from the epicentre, the prevalence ratios come to 1/10,000, a very considerable change from the outlying départements.
47The arrondissements affected by the Deux-Sèvres concentration show several different elements. Nowhere is the concentration as strong as in the previous case : the maximum prevalence ratio is of 1/1,193 in the Pathenay arrondissement ; in two other arrondissements in Deux-Sèvres prevalence rates are rather high : Bressuire (1/2,430) in the north and Niort (1/3,835) in the south. The concentration seems to extend north-east into the département of Maine-et-Loire (in Saumur arrondissement the prevalence ratio is 1/6,440). In the outlying départements the ratios are very small, especially in Vendée to the west (1/30,187). Hence, the effects of concentration and spread are lower here than those observed for the Ain-Jura concentration. But it leads us to speculate whether this can be considered as a more recent form of the mutation.
48As regards the département of Loir-et-Cher which seems to occupy the highest position (Table 4), the prevalence ratio varies from 1/4,144 (for the Vendôme arrondissement) to 1/6,787 (for that of Blois). Prevalence ratios are lower in the surrounding départements, especially the département of Indre-et-Loire which separates Loir-et-Cher from Maine-et-Loire and Deux-Sèvres, with only a very low prevalence rate of 1/31,625.
Conclusion
49It should not be forgotten that the prevalence ratio of 1/80,000 computed by McKusick is still largely accepted. The minimum figure we obtain for all 52 départements is 1/8,345. The method used and the results of the surveys on the spot lead us to suppose that this prevalence ratio could be two or three times higher.
50We now intend to extend the postal survey to cover the whole country. It is presently being extended to 21 new départements, and the national survey could be completed in 1989. We know that in the areas which have not yet been surveyed, cases of Rendu-Osler disease are found. Some family histories have been published in Lille, Bordeaux and Toulouse. We hope to be able to present a map with the prevalence ratios for the whole of France.
51If there are other geographical concentrations we shall be able to spot them more clearly and note the degree of concentration attained in each. We shall then be able to study how defective genes spread out from each of these concentrations, while keeping track of the demographic structure and population migrations in each département [13].


Notes
-
[*]
Translated by Nita Lery.
-
[**]
Jean Sutter : La Luxation congénitale de la hanche, Paris, INED-PUF 1972, 240 p. (Travaux et Documents, Cahier n° 62).
-
[***]
Alain Bideau : Centre P. Léon, UA 223 CNRS ; Henri Plauchu and Guy Brunet : Institut Européen des génomutations, 86 rue F. Locard, 69005 Lyon ; Jacques-Michel Robert : Université de Lyon II.
-
[1]
Henri Plauchu and Alain Bideau, "Epidemiologie et constitution d’un registre de population à propos d’une concentration géographique d’une maladie héréditaire rare", Population, 4-5, 1984, pp. 765-786.
-
[2]
J. Sutter, L’atteinte des incisives latérales supérieures. Paris, INED-PUF, 1966, 148 p. (Travaux et Documents, Cahier n° 46). La luxation congénitale de la hanche. Paris, INED-PUF, 1972, 242 p. (Travaux et Documents, Cahier n° 62).
-
[3]
J. Feingold, A. Hennequet, M. Jehanne, J. Feigelson, L. Toudic, O. Quiniou, M.L. Briard, "Fréquence de la fibrose kystique du pancréas en France", Annales de Génétique, 17, n° 4, 1974, pp. 257-259.
-
[4]
E. Bois, J. Feingold, P. Frenay, M.L. Briard, "Infantile cystinosis in France : genetics, incidence, geographic distribution", Journal of Medical Genetics, 13, 1976, pp. 434-438.
-
[5]
C. Bonaiti, F. Demenais, M.L. Briard, J. Feingold, J. Frezal, "Congenital glaucoma : genetic models", Human Genetics, 46, 1979, pp. 305-317.
-
[6]
E.D. Shields, D. Bixler, P. Fogh-Andersen, "Cleft palate : a genetic and epidemiologic investigation", Clinical Genetics, 20, 1981, pp. 13-34.
-
[7]
A. Petrequin : La maladie de Rendu-Osler. II – Profil clinique et vécu de la maladie : à propos de 324 observations, Medical thesis, Lyon, 1983, 110 p. typed.
-
[8]
See ref. in fn. 7.
-
[9]
These are individuals reported as suffering from the disease ; starting with them, a genealogical history is established with, on an average, 6 Rendu-Osler patients in the family, four of whom are still alive (the average is based on more than 500 individual histories).
-
[10]
L. Venturini : La Maladie de Rendu-Osler. I – Etude épidémiologique et étude génétique : à propos de 324 observations, Medical thesis, Lyon, 1983, 75 p. typed.
-
[11]
M.J. Wehrlen : Utilisation de la mutation Rendu-Osler comme marqueur génétique dans le suivi des populations, Medical thesis, Lyon, 1984, 168 p. typed.
-
[12]
J.Y. Lehy : L’angiomatose de Rendu-Osler : étude démographique et génétique dans le département des Deux-Sèvres, Medical thesis, Nantes, 1983, 164 p. typed.
-
[13]
McKusick, Mendelian inheritance in man, Baltimore and London, The Johns Hopkins University Press, seventh edition, 1986.