Comparaison du rapport coOt/eff icacit6 de deux tests de d6pistage des tumeurs colo-rectales

dans un centre d'examen de sant6

C. G U L D E N F E L S *, S. B R I A N ( ~ O N * **, O. H O U O T *, B. J U N O D **, J .P. D E S C H A M P S * **

�9 Centre de M~decine Preventive - Avenue du Doyen J.-Parisot, 54500 Vandteuvre-lds-Nancy �9 * Ecole Nationale de la Sant~ Publique - Avenue pr L.-Bernard, 35043 Rennes Cedex

�9 ** D~partement de Sant~ Publique - Facult~ de M~decine, BP 184, 54505 Vandteuvre-lds-Nancy Cedex

Cost-effectiveness of two screening tests for colo-rectal tumors used in a health examination center

Comparacibn de la relaci6n costo/eficacia de dos test de despistage de tumores colorectales en un centro de exdmenes de salud

RI~SUMI~

Dans une 6tude effectu6e au Centre de M6decine Pr6ventive de Nancy et portant sur 2 186 personnes, les auteurs se proposent de comparer deux tests de d6pistage des tumeurs colorectales : l'Hemoccult II | et le Fecatwin Sensitive Feca EIA | La va|eur pr6dictive positive des deux tests est sensiblement 6quivalente, mais le test EIA d6piste trois lois plus de tumeurs et pr6sente aussi plus de faux positifs que i'Hemoccult | En terme 6conomique, le coot total du d6pistage est 2,1 fois plus 61ev6 avec Feca EIA mais le coot unitaire de la tumeur d6pist6e est plus faible (7 178 F vs 12 793 F, p < 0,001). Cet argument 6conomique est discut6 selon le point de vue des Centres d'Examens de Sant6, de la S6curit6 Sociale et de la Sant6 Publique

S U M M A R Y

A study was conducted at the Center for Preventive Medicine in Vandveuvre-l~s-Nancy in order to compare the performances o f two screening tests for colorectal cancer : Hemocult 11 ~ and fecatwin sensitive Feca EIA | The positive predictive value is approximately the same for both tests. The Feca EIA | test has been able to detect 3 times more tumors than Hemocult 11 ~, but the Feca test gives more false positive results. In economical te~ms, the total cost o f screening is 2.1 times higher for Feca EIA, but the unit cost for each discovered tumor is lower for Feca EIA than it is for Hemocult 11 ~ (7 178 francs vs 12 793 francs, p < 0.0(91). These results are briefly discussed with respect to three different issues : the role o f a screening center (i), the budget o f the social security system (ii), and public health (iii).

R E S U M E N

En un estudio realizado en el Centro de Medecine Preventiva de Nancy en una poblaci6n de 2 186 personas, los autores se proponen de comparar dos test de despistage de tumores colorectales : el ~ Hemoccult H | ~ y el ~ Fecatwin Sensitive Feca E IA | ~,. El valor predictivo positivo de los dos test es sensiblemente equivalente, pero el test EIA detecta tres veces mds tumores y presenta tambi~n m6s falsos positivos que el Hemoccult H | En tdrminos econ6micos, el costo total des despistage es 2,1 veces m6s elevado con Feca EIA | pero el costo unitario del tumor detectado es menor (7 178 F vs 12 793 F, p < 0,001). Este argumento econ6mico es discutido segtin los diferentes punto de vista de los Centros de Exdmenes de Salud, de la Seguridad Social y de la Salud Pablica.

I N T R O D U C T i O N

Le d6pistage des tumeurs colo-rectales par la recherche de sang occulte dans les selles int6resse depuis quelques ann6es les centres d'examen de sant6 (CES). Le test actuellement utilis6 est l'He-

moccult II | test au gaiac, r6agissant ~t la pr6sence de substances ~ activit6 peroxydasique. G. Du- bois [5], dans une 6tude effectu6e dans trois CES a montr6 l'excellente acceptabilit6 de ce test qui permet une d6couverte pr6coce de l~sions ad6no- mateuses et donc son adaptation parfaite aux objectifs d'un CES.

Tir6s ~t part : B. JUNOD, ENSP, avenue du Professeur L.- Bernard, F-35043 Rennes Cedex (France)

Mots-cl~s : d6pistage, tumeur colo-rectale, co0t/efficacit6.

Key-words: screening, colorectal tumor, cost-effectiveness.

Palabras clave : despistage, tumor colorectal, costo/eficacia.

Acta Endoscoptca Volume 20 - N ~ 1 - 1990 27

ROcemment est apparu sur le march6 un nou- veau type de test associant un test au ga'iac ~t une mOthode immuno-enzymatique (Fecatwin Feca EIA| plus spOcifique de l'hOmoglobine, permet- tant pour un niveau de sensibilit6 donn6 de dimi- nuer le nombre de faux positifs ~ l'origine des principales critiques 6mises h l'encontre de l'He- moccult |

La comparaison de l'intOrOt de deux tests de dOpistage peut se faire /t diffOrents niveaux : taux de positivitO, pouvoir de dOtection, nombre de tumeurs dOpistOes, consOquences des erreurs de classement, coot des tests et des, examens... Ces diffOrents critOres de jugement peuvent aboutir ii des conclusions divergentes. Plusieurs 6tudes ont permis de comparer ces deux tests : Hemoccult II | et Fecatwin Sensitive-Feca EIA | [2, 3, 8]. Elles montrent que le test immunologique offre un taux de positifs plus 61ev6 que l'HOmoccult II | (argu- ment plut6t considOr6 comme dOfavorable /t son encontre) mais dOtecte plus de tumeurs (argument favorable) avec un pouvoir de dOtection trOs voi- sin. Dans cet article, nous nous proposons d'ap- porter des arguments 6conomiques en terme de coot (dOpenses totales engagOes pour le test) et de cofit-efficacit6 (coOt unitaire de la tumeur dO- pistOe).

METHODES

L'Otude a 6t6 effectuOe au Centre de MOdecine PrOventive (CMP) ~ Vandceuvre-l~s-Nancy, entre le 20 avril 1985 et le 30 mars 1986. Nous avons propos6 ~ toutes les personnes de plus de 50 ans, venant au centre pour un examen de santO, d'ef- fectuer ~ domicile 3 jours de suite, les deux tests : Hemoccult | et Fecatwin | qu'ils renvoyaient ensuite au CMP par la poste.

Les personnes dont soit l'Hemoccult, soit le Fecatwin | gaiac 6tait positif, 6taient invitOes contacter leur m6decin traitant. Celui-ci rOalisait un lavement baryt6 ou adressait les patients ~ un gastroentOrologue qui jugeait des investigations pratiquer : une colonoscopie totale d'emblOe a 6t6 d6cidOe dans 51% des cas, une rectosigmoi- doscopie suivie de lavement baryt6 en double con- traste dans 32 % des cas. Un double des rOsultats de ces investigations et de l'examen anatomopa- thologique 6ventuel nous 6tait envoyO.

La mOthode de calcul du prix de revient de ces tests a 6t6 celle exposOe par Dubois [4] et qu'il a appliquOe dans son 6tude prOcOdemment citOe, l'exception de la soustraction du coot des vrais positifs :

coot total = nombre de tests rOalisOs x coot du test + nombre de positifs x coot des examens complOmentaires

coot unitaire = coot total/nombre de tumeurs dOpistOes.

j Consultants de plus de 50 ans

I Tests Hemoccul~

+ Fecatwir~

I

I Analyse du test ] immunologique

Gastro- J / e n t ~ o l ~

J Analyse des r6sultats !

Figure 1 Organigramme de la methode employee

Le coot de l'examen de laboratoire tient compte du prix des rOactifs et du temps de la techni- cienne. Compte tenu du nombre important de tests effectuOs, il ne correspond pas, /t l'Ovidence, au prix de revient d'un test isol6 prescrit par un mOdecin.

Le coot des examens complOmentaires a 6t6 estim6 /i 1 702 F selon G. Dubois [4].

La signification statistique de la diffOrence de coot entre les 2 tests et les intervalles de confiance des coots calculOs se fondent sur l'hypoth~se d'une distribution binomiale du nombre des rOsultats po- sitifs.

RI~S UL TATS

1. Valeurs pr~dictives positives

Durant la pOriode 6tudiOe, 2 186 personnes ont rOalis6 les tests parmi lesquels 257 furent dOclarOs positifs, rOpartis de la fa~on suivante :

28 V o l u m e 2 0 - N ~ 1 - 1 9 9 0 A c t a E n d o s c o p i c a

- - 257 positifs au Fecatwin | gaiac (11,7 %) ;

- - 76 au Feca EIA | (3,5 %) ;

- - 26 h l 'Hemoccult | (1,2 %).

Les 76 Feca EIA | positifs sont, par d6finition, inclus dans les 257 gaiacs positifs ; les 26 sujets Hemoccult positifs sont tous Fecatwin | gaiac posi- tifs.

Seulement 188 personnes effectu~rent les exa- mens compl6mentaires, ce qui permit la d6cou- verte de 5 cancers (1 stade in situ, 2 stades A et 2 stades B selon la classification de Dukes) et 27 ad6nomes. La r6partition des tumeurs d6pist6es et les valeurs pr6dictives positives sont pr6sent6es dans les figures 1 et 2 : l 'Hemoccult II | aurait d6pist6 2 cancers et 4 ad6nomes, le Feca EIA | en aurait trouv6 respectivement 4 et 18. Ces r6sultats sont donn6s uniquement pour les personnes ayant effectu6 les examens compl6mentaires. D'autre part, quelques Hemoccults | n 'ont pu 6tre r6alis6s pour des raisons techniques, en particulier une des personnes porteuse d'un cancer n'a pas pu avoir d'Hemoccult.

188 sulets galac p o s i t i f s ~

4 cancers / 52 immuno + ~ - - 18 adenomes

\ 30 negatifs

1 cancer /

136 immuno - ~ 9 adenornes

~126 negatlfs Figure 2

Repartlt=on des tumeurs deplstees par le Fecatwm EIA

/ 18 Hemoccult + (

182 personnes (

\

/ 164 Hemoccult - ~-- \

VPP = 33 %

2 cancers

4 adenomes

12 negatlfs

2 cancers

23 adenornes

139 nC=gatlfs

Figure 3 Repartition des tumeurs deplst6es par Hemoccult

2. C o m p a r a i s o n des coats

- - H e m o c c u l t H |

Le test revient h 20 Francs, il a 6t6 effectu6 chez 2 186 patients, soit un coot pour le CMP de 43 720 F.

18 personnes ont effectu6 un bilan compl6men- taire h 1 702 F, soit un coot de 30 636 F.

Le coot total est de 43 720 + 30 636 = 74 356 F.

L'intervalle de confiance ~ 95 % de chacun de ces coots est de + 14 094 F.

Si l'on rapporte ce total aux 6 tumeurs d6pist6es par l 'Hemoccult | on obtient un coot de 12 393 F + 2 349 F par tumeur.

- - Feca tw in - Feca E I A |

Le Fecatwin | revient ~ 20 F et le Feca EIA | effectu6 si le Fecatwin gaiac est positif, ~ 100 F, soit un coot pour le CMP d e : 2 1 8 6 F x 20 + 257 x 100 = 69 420 F.

52 personnes effectu~rent un bilan compl6men- taire soit un coot de 88 504 F + 23 768 F.

Le coot total est donc de 157 924 F, soit 2,1 lois plus que pour l 'Hemoccult.

Le coot unitaire de la tumeur d6pist6e est pour ce test de 7 178 F + 1 080 F.

La diff6rence entre coots unitaires est statisti- quement significative (p < 0,001).

D I S C U S S I O N

Ces deux tests ont sensiblement les m6mes valeurs pr6dictives positives. Si le Feca EIA d6piste trois lois plus de tumeurs, par contre le nombre de faux positifs est plus 61ev6 qu'avec l 'Hemoccult. Le choix de l'un ou l 'autre test doit donc reposer 6galement sur d'autres arguments : par exemple, probl~mes psychologiques d 'un faux positif, morbidit6 li6e aux investigations compl6- mentaires, tumeurs non d6pist6es pr6cocement, enfin probl~mes 6conomiques sur lesquels nous avons ax6 notre pr6sentation.

Cependant, si le nombre de publications qui traitent du rapport co0t/efficacit6 de l 'Hemoccult | est important [1, 6, 7, 9], il n'existe pas, ~ notre connaissance, de publication qui compare Hemoc- cult | et Fecatwin-Feca EIA | en termes 6conomi- ques. Ces arguments devront 6tre envisag6s diff6- remment selon que l'on se place dans une optique de sant6 publique ou comme gestionnaire d 'un CES ou de la S6curit6 Sociale.

En effet, pour le CES, le Fecatwin | en co0tant 11 lois plus cher que l 'Hemoccult | en cas de posi- tivit6 du gaiac, occasionne une d6pense deux lois plus lourde. La totalit6 des d6penses d'un CES est ventil6e sur l 'ensemble des consultants de faqon obtenir un coot unitaire de bilan unique ; le CMP recevant 30 000 personnes par an, on peut estimer

69 420 F/30 000 = 2,31 F le coot de ce test dans le bilan de sant6.

Dans une optique de budget de S6curit6 Sociale, il est sOr que le Fecatwin | ayant un nombre de positifs plus important occasionne plus de bilans compl6mentaires chez le gastro-ent6rologue (157 924 F vs 74 356 F). Cependant, il permet

A c t a E n d o s c o p i c a V o l u m e 2 0 - N ~ 1 - 1990 29

l ' ab l a t i on des a d 6 n o m e s a un s tade p r6coce , 6vi- tan t ~ long t e r m e une hosp i t a l i sa t ion coOteuse p o u r une i n t e rven t i on chi rurg ica le b e a u c o u p plus i m p o r t a n t e .

Enf in , en t e rmes de sant6 pub l ique , le Feca twin E I A pa ra i t sup6r i eu r pu i sque d6pis tan t p lus de t um e ur s p o u r un coo t un i t a i r e m o i n d r e . D e plus , p o u r le pa t i en t l u i -m6me, les b6n6fices sont impor - t a n t s : il y a mo ins de faux n6gatifs donc un t r a i t e m e n t plus p r6coce des t umeur s co lo - rec ta les

6vi tant ainsi les cons6quences phys iques et psycho- logiques tr~s l ou rdes q u ' i m p l i q u e la ch i rurg ie carci- no log ique digest ive.

E n conclus ion , le Feca twin S e n s i t i v e / F e c a E I A | c ompa r6 dans no t re 6 tude ~ l ' H e m o c c u l t l i e pe r - me t une a u g m e n t a t i o n no t a b l e du p o u r c e n t a g e de t ume ur s co lo - rec ta les d6pis t~es , ce qui lui p e r m e t , tout en 6tant p lus on6reux q u ' u n tes t wa ga iac s imple , d ' a v o i r un coo t un i ta i re p a r t u m e u r inf6- r ieur .

REFERENCES

1. ALLISON V.E., FELDMAN R. - - Cost benefits of Hemoccult screening for colorectal carcinoma. Dig. Dis. Scs 1985, 30, 860-865.

2. ARMITAGE N., HARDCASTLE J.D., AMAR S.S. et al. - - A comparison of an immunological foecal occult blood Test Fecatwin Sensitive/Feca EIA with Haemoccutt in po- pulation screening for colorectai cancer. Br. J. Cancer, 1985, 51, 799-804.

3. BRIAN~ON S., HOUOT O., DE SOUZA M. et al. - - Evaluation d'un test immunologique de recherche de sang dans les selles pour le d6pistage des tumeurs colorectales. GastroenteroL Clin. Biol., 1987, 11, 194-200.

4. D U B O I S G . - - Evaluation de la capacit6 de rep6rage d'un 6tat pathologique par un test de d6pistage. Rev. Fr. Santd PubL, 1980, 11, 34-39.

5. DUBOIS G., BREGEVIN B., BERTOIN J. et al. -- La recherche de sang dans les selles par l'Hemoccuit II dans l'examen de sant6. Rev. Med. Assurance Mal., 1985, 4, 46- 48.

6. EDDY D.M., NUGENT F.W., EDDY J.F. et al. -- Screening for colorectal in a high risk population. Results of a mathematical model. Gastroenterology, 1987, 92, 682- 692.

7. FRANK J.W. - - Occult blood screening for colorectal carninoma : the yield and the costs. Am. J. Prey. Med., 1985, 1, 19-23.

8. HARDCASTLE J.D., FARRANDS P.S., BARFOUR T.W. - - Controlled trial of faecal occult blood testing in the detection of colorectal cancer. Lancet, 1983, 2, 1-4.

9. KRISTEIN M.V. - - The economics of screening for colo- rectal cancer. Soc. Sci. Med., 1980, 14, 275-284.

I N T R O D U C T I O N

Since many years, Health Centers use occult blood screening tests for colorectal tumors detec- tion. Actually, Hemocult I1 ~ is the most commonly used test in French health centers. This gai'ac test reacts to peroxydasic substances. A study conducted by G. Dubois in three French Health centers has shown both the remarquable acceptability of Hemoccult I1 ~ and its ability to help for early discovery of adenomatous lesions. Therefore, he considers this test as adapted for French Health Examination Centers.

Fecatwin Feca EIA | has recently appeared on the market. It relies on the association o f a gai'ac test coupled with an enzymatic method which is more specific for identifying the presence of hemog- lobin. For a given sensibility level, this feature would putatively diminish the number of positive results falsely obtained by Hemocult 11 ~ test.

The main criticisms adressed to Hemocult I1 ~ lie on such false positive results.

The comparison of two screening tests can be done at different levels : the rate of positiveness, the detection power, the number o f detected tumors, the consequences of both false negative and false positive results, the cost of the tests, the cost of colonoscopy induced by the test... Each level is

associated with different ]ugement criteria so that conflicting conclusions may occur. Several studies compare these two tests [2,3,8]. It has been shown, firstly, that Fecatwin Sensitive Feca EIA | tests are more often positive than Hemocult 11 ~ test and, secondly, that Feca EIA | detects more tumors. The first mentioned result has often been considered as a drawback of Feca EIA | whereas the latter is considered as an advantage. The detection power is in fact similar for both tests.

This paper aims to give economical arguments in terms of total cost for the use of each test and for a cost effectiveness estimate, that is the cost for each detected tumor.

M E T H O D S

The study has been conducted from April 20th 1985 to March 30th 1986 at the Center for Preven- tive Medicine in Vandoeuvre-lds-Nancy. Any person aged 50 or more who came for periodic health examination has been asked for testing stools at home on three subsequent days using both Hemoccult 1P 9 and Feca EIA | Each test has been mailed to be analysed at the laboratory of the Center for Preventive Medicine.

30 Volume 20 - N ~ 1 - 1990 Acta Endoscopica

[

Patients over 50

I Hemoccult ~

+ Fecatwm |

tests

Sending by marl I

Negative

General i ~ ~ Practitioner

Gastro- I ~ enterol

[ Results analysis [ Fsgure 1

Organtzatlon of the study

If Hemoccult II ~ and~or Fecatwin | ga~ac test were positive, people were invited to go to their family doctor. He decided the kind of further inves- tigation to be conducted. 5 1 % of the patients undergone a colonoscopy and 32 % of the patients had a restosigmoMoscopy followed by a barium enema. The result of these investigations and a copy of the pathology record, if any, were sent to the Center for Preventive Medicine.

The computation of the cost corresponding to each test follows the method published by G. Dubois [4], except that it does not substract the cost corresponding to correctly identified positive cases. For one test, the total cost equals the product of the number of test times the unit cost of the test, plus the number of positive results multiplied by the mean cost of further investigations.

The cost of the laboratory test takes into account both the reagents and the technical work. Given the large number of realized tests, this unit cost does evidently not fit with the cost for an isolate test prescribed by the family physician.

The cost of complementary examination estimated by G. Dubois amounts to 1 702 francs. The statisti- cal signifiance of the cost difference between the two tests and the confidence intervals of the com- puted costs lie on the hypothesis of a binomial distribution of the number of positive results.

RESULTS

1. Positive Predictive Value

During the study period, 2 186 people have realized the tests. Among them, 257 have been considered as positive :

- - 257 positive to the Fecatwin | ga~ac (11, 7 %) ;

- - 76 positive Feca EIA | (3,5 %) ;

- - 26 positive to Hemoccult 11 ~ (1,2 %).

By definition the 76 Feca EIA | positive tests are included in the 257 gaiac positive. The 26 subjects with positive Hemoccult I1 ~ are all Fecatwin | ga~ac positive.

From these 257 patients, 188 only undergone luther investigation. Among them 27 adenoma and 5 cancers have been detected : 1 in situ, 2 stage A, and 2 stage B according to Dukes classification. The distribution of detected tumors and the positive predictive values are presented in figure 1 and 2 : Hemoccult 11 ~ alone would have detected 2 cancers and 4 adenomas. Feca EIA alone would have detected 4 cancers and 18 adenomas. These results have been given only for people undergoing further examination. Moreover, for iechnical reasons, some of Hemoccult I1 ~ tests have not been completely realized, and one of the cancer patients was in this particular situation.

4 cancers

/ 52 ,mmuno + 18 aoenoma30 wi,ho 188 patients tumor

with positive Gaiac t e s t ~ ~/~__ 1 cancer

136 tmmuno - 9 adenoma

126 without tumor

Figure 2 Distribution of tumors detected by Fecatwin Elk,

182 patients

VPP = 33 %

18 Hemoccult + (

164 Hemoccult -

2 cancers

4 adenoma

12 without tumor

2 cancers

23 adenoma

139 without tumor

Figure 3 Distribution of tumors detected by Hemoccult

2. Cost comparison

For Hemoccult 11 ~, the test costs 20 francs. It has been done for 2 186 patients, that is a cost of 43 720 francs for the Health Examination Center.

A c t a E n d o s c o p i c a V o l u m e 2 0 - N ~ 1 - 1990 31

18 people undergone further examination at a mean unit cost of 1 702 francs. -This corresponds to 30 636 francs. The total cost amounts to 74 356 francs. The 95 % confidence interval of each cost amounts to 14 094 francs. The mean estimate of the unit cost of one detected tumor gives 12 393 +_ 2 349 francs.

For fecatwin-Feca EIA | the cost is fractionned as follows. Each fecatwin test costs 20 francs. The Feca EIA | is done for 100 francs only if Fecatwin ga~ac is positive. The corresponding cost for the Health Examination Center is 2 186 x 20 + 257 x 100 = 69 420 francs. 52 people undergone further investigation. The corresponding cost amounts to 88 504 +_ 23 768. The total cost amounts to 157324 francs i.e. 2.1 the cost for Hemoccult I1 ~.

For fecatwin Feca EIA | the mean estimate of the unit cost for each detected tumor amounts to 7 178 + 1 080 francs he difference between the unit cost for Hemoccult II ~ and for Fecatwin-Feca EIA | test is statistically significant (p < .001).

DISCUSSION

These two tests have similar positive predictive values. Feca EIA | detects three times more tumors, whereas the number of false positive results is higher than for Hemoccult 11 ~. The choice of either test should rely also on other arguments : for exam- pie, the psychological problems due to the false announcement of a positive result and the possible side effects of further investigations, the morbidity due to the lack of early removal of precancerous lesions, and, finally, the economical problems in terms of cost.

Some publications deal with the cost-effectiveness of Hemoccult Ii ~ [1, 6, 7, 9], but it seems that no study has compared Hemoccult 11 ~ with Fecatwin-

Feca EIA | from an economical point of view. The above mentionned arguments are to be considered in different ways, according to the kind of perspec- tive in which a decision should be taken : as the manager of the periodic health examination center, as the administrator of the social security system or from a public health point of view.

For the Health Examination Center, in case of a positive result for a gat'ac test, the cost of Fecatwin | is 11 times higher than for Hemoccult I1 ~, so that the global cost is twice as high for Hemoccult I1 ~. The total amount spent by the Center is spread out on the people coming for a health examination in such a way that the unit cost of the screening is the same for everybody. Given that 30 000 people come yearly to the Center, the cost o f Fecatwin | amount to 69420/30 000 = 2.31 francs for one health ecamination.

For the social security budget, Fecatwin | induces a higher cost due to the larger number of positive results followed by further endoscopical and radiological examinations (157924 francs versus 74 356 francs). However, the Fecatwin | test is able to detect more removable adenomas. This could preclude costly hospitalisations for invasive cancer treatment.

Finally, from the public health point o f view, Fecatwin EIA | appears to be better, given that it is able to detect more tumors for a given amount of money. Moreover, the patient's benefits are impor- tant. The probability of being a false negative is lower. This means that colorectal tumors can be treated at an carlier stage. Dramatic physical and psychological consequences of surgery for an extended carcinoma are then possibly avoided.

In conclusion, the comparison between Fecatwin Sensitive EIA | with Hemoccult I1 ~ shows that Fecatwin | notely increase the number of detected tumors. The global cost due to Fecatwin | is higher, but the unit cost by detected tumor is lower than for Hemoccult II ~.

32 Volume 20 - N ~ 1 - 1990 Acta Endoscopica