Utilization of primary health care workers for early detection of oral cancer and precancer cases in Sri Lanka

/Research Recherche Bulletin ofthe World Health Organization, 62(2): 243-250(1984) © World Health Organization 1984 Utilization of primary health care workers for early detection of oral cancer and precancer cases in. Sri Lanka K. A. A. S. WARNAKULASURIYA,l A. N. I. EKANAYAKE,2 S. SIVAYOHAM,3 J. STJERNSWARD,4 J. J. PINDBORG,5 L. H. SOBIN,6 & K. S. G. P. PERERA Oral cancer presents a serious public health problem in south-east Asian countries. In Sri Lanka and India 35-40% of all cancers are reported to be oral cancers, which are curable ifdetected in the early stages. The idea that, in developing countries, one of thefew practical approaches to early detection of these cases could be through the utilization of primary health care workers was tested in afield study carried out in Sri Lanka. In a control area the subjects with oral lesions were identified by medical/dental officers. In the study, 34primary health care workers were able, alongside their routine duties, to examine the oral cavity of 28 295 subjects during a period of 52 weeks; 1220 subjects were detected with lesions needing re-examination. Theperformance of theseprimary health care workers was very satisfactory in terms of the number of cases detected and the accuracy of their diagnoses, which were verified by re-examination at a specially designated referral centre. The clinical diagnoses of the three categories of lesions detected were as follows: stage 1 lesionsfor observation (homogeneous or ulcerated leukoplakia), stage 2 lesionsfor investi- gation (speckled leukoplakia, erythroplakia, or submucousfibrosis), and stage 3 lesionsfor treatment (cancer). There was 89% correspondence between the stage 1-3 diagnoses by the health workers and the clinical diagnoses made at the referral centre. However, compliance by patients was low because only 50% ofthe cases detected in thefield attended the referral centre. The cost-effectiveness of this approach to cancer control has been demonstrated. The high frequency of oral cancer in south-east Asian countries, which has been well documented (1-3), makes it a serious public health problem in this region. In Sri Lanka and India, approximately 35-40% of all cancers occur in the oral cavity (1, 2), ' Head, Department of Oral Medicine, Dental School, University of Sri Lanka, Peradeniya, Sri Lanka. Requests for reprints should be sent to this author. 2 Head, Department of Community Dental Health, Dental School, University of Sri Lanka, Peradeniya, Sri Lanka. 3Director, National Cancer Control Programme, Colombo, Sri Lanka. Chief, Cancer Unit, World Health Organization, Geneva, Switzerland. 5Professor of Oral Pathology, Royal Dental College, Copen- hagen, Denmark. 6 Pathologist, Department of Gastrointestinal Pathology, Armed Forces Institute, Washington, DC, USA. 7Assistant Lecturer, Dental School, University of Sri Lanka, Peradeniya, Sri Lanka. compared with only 2-3% in the United Kingdom and the USA (1). This difference is also reflected in the annual incidence per 100 000 population, which is approximately 5 times higher in India compared to the Scandinavian countries (2). Furthermore, various studies in India have indicated a high prevalence of oral premalignant lesions (4) such as leukoplakia (0.2-4.9%) and oral submucous fibrosis (0.2-0.4%). The high frequency of oral cancer and precancer is known to be related to high tobacco consumption (5), and in south-east Asian countries this is seen with particular reference to betel quid chewing with tobacco as an ingredient. If detected at an early stage, oral cancers can be treated and cured. The great majority of patients, however, come forward only when the disease is too advanced for curative therapy. The natural history of oral cancer demonstrates that often a precancer stage precedes the development of cancer, and it is known 4397 -243- K. A. A. S. WARNAKULASURIYA ET AL. that intervention at this stage may result in regression of the lesion. Because of the need for appropriate national cancer policies and new approaches to cancer control in the developing countries, the present study was designed to test the utilization of primary health care (PHC) workers for the early detection of oral precancerous and cancerous lesions. In Sri Lanka, PHC workers are engaged in various aspects of community health care, such as maternal and child health and immuniz- ation, and their services are complementary to those delivered by the hospital-based health service system. A consideration of the basic problems in population screening for cancer (6) shows that screening for oral precancer and cancer meets the essential requirements of a screening programme. The objectives of the present study were: (a) to assess whether it was feasible to introduce early detection of oral cancer and precancer into the routine work schedule of primary health care workers; (b) to assess the efficiency of these workers and their motivation for this type of field activity; (c) to ascertain the extent to which patients, who were requested to go to a referral or treatment centre, would be motivated to go there; (d) to ascertain the reliability of the PHC workers' diagnoses and "staging" of lesions, based on a simplified classification; (e) to assess the cost-effectiveness of an early detec- tion programme of this kind; (f) to compare the above approach with an alter- native system involving hospital-based case-finding, where screening of the oral cavity was carried out by medical and dental personnel. SUBJECTS AND METHODS This study was conducted in two health areas in the central region of Sri Lanka. In one area (Kadugan- nawa) a field approach was tested utilizing primary health care workers, and in the other (Gampola) a hospital-based approach using medical and dental officers was tested. The Kadugannawa study was conducted in an area of 173.5 km2 with a population of 87 277 adults over 20 years old. Thirty-five PHC workers (consisting of 23 midwives, 9 public health inspectors, and 3 public health nurses) employed by the Ministry of Health in this area were utilized. At the commencement of the study all of them participated in a two-day training programme which provided a clinical demonstration of oral cancer and precancer cases and instructions re- garding the screening method, and dealt with other aspects of programme implementation. A compre- Table 1. Staging of oral precancer and cancer lesions by PHC workers and the corresponding clinical diagnoses Classification by PHC workers Clinical diagnoses Stage 1: 'lesion for Homogeneous leukoplakia observation' Ulcerated leukoplakia Stage 2: 'lesion for Speckled/nodular investigation' leukoplakia Erythroplakia Oral submucous fibrosis Stage 3: 'cancer probable or Overt oral carcinoma already treated for cancer' Treated cancer hensive but simple record-keeping system was de- signed, which took into account the difficulties of working under field conditions. In order to facilitate international standardization, the criteria used for detection of lesions were examined and approved by a panel of WHO consultants, who also participated in the training programme. The PHC workers were requested to carry out an examination of the lining mucosa of the oral cavity of people over the age of 20 years in their area, this activity being incorporated into their routine work schedule which included house-to-house visits. Voters' lists were used to identify and record the number and proportions of persons examined. Dur- ing home visits, oral examinations were carried out in natural daylight using two dental mirrors. On the basis of simple, explicitly stated criteria, the PHC workers were required to classify the detected lesions into three categories consistent with specific clinical groupings, as shown in Table 1. All cases detected in the field were referred to the Department of Oral Medicine, University Dental School, Peradeniya, which was designated as the referral centre; the closest screening area was 0.4 km away from this centre and the furthest screening area was 40 km away. At the referral centre the PHC workers' diagnoses were assessed on the basis of the WHO guide to the diagnosis of oral mucosal diseases (7). This assessment was carried out by project staff who were supervised by WHO consultants. Infor- mation pertaining to betel chewing and smoking habits was obtained from all patients. Biopsies were taken from stage 2 and stage 3 lesions at the referral centre for confirmation of the clinical findings by histopathology . The task of persuading the diagnosed or suspected cases to attend the referral centre was left entirely to each PHC worker who sent the names and addresses of these cases to the referral centre. Subjects who did not report to the centre for re-examination were then contacted by post. An incentive payment for each subsequently confirmed diagnosis was given to PHC workers at the rate of Rs 25 (equivalent to US $1.30) 244 CANCER DETECTION BY PRIMARY HEALTH WORKERS for each confirmed stage 2 lesion and Rs 50 (equiva- lent to US $2.60) for each confirmed stage 3 lesion. Furthermore, the travelling costs of all subjects requested to report to the referral centre were reim- bursed. Confirmation of the negative findings (no lesion reported) by the PHC workers was carried out by re- examination of a random sample of about 35 cases in each PHC worker's area by staff from the referral centre during prearranged field visits. Altogether 1212 cases were verified in this way. The study was conducted for a continuous period of 52 weeks during which the project staff maintained regular contact with the PHC workers in the field by means of monthly meetings and field visits. The find- ings relating to each subject who reported to the referral centre were transmitted to the PHC worker at the end of every month; this feedback helped to improve these workers' diagnostic capabilities. Pro- vision was also made for the positively diagnosed cases to receive appropriate treatment at the closest centre. The Gampola study was hospital-based and in- volved 2 medical officers, 3 dental surgeons, and 7 assistant medical officers from a total of 9 hospitals in the Gampola area, which is near Kadugannawa, has closely similar demographic features, and was equi- distant from the referral centre. After a one-day train- ing programme, these doctors and dentists were re- quested to carry out routine screening of the oral cavity of patients presenting at their hospitals for medical and dental care. The methods of examination and instruments used, the diagnostic criteria for classification of lesions into three stages, and the referral forms used were the same as in the Kadugan- nawa study. However, instead of the electoral lists to enumerate the subjects screened, each of these profes- sional workers maintained a record of the number of cases screened. These medical and dental officers in the Gampola area were unsupervised, in contrast to the regular and sustained supervision of PHC workers engaged in the Kadugannawa study, where records were scrutinized at the end of each month. Further, no incentive pay- ments were made to medical/dental officers for undertaking this screening work. RESULTS Cancer detection by PHC workers (Kadugannawa study) Table 2 shows the proportion of the population (both males and females) screened by the three cate- gories of PHC workers. Of the 87 277 persons in the target population, 29 295 (33.3%) were screened by these workers during the study period of 52 weeks; 7.1 0o of the total population was screened within 10 weeks, 18.7%o within 28 weeks, 25.5%7o in 41 weeks, and the balance (up to 33.3%o) thereafter. The mean number of cases screened per worker per week was 17 ±9. Of the 35 PHC workers trained, 34 participated in the study. The efficiency of individual workers in terms of population coverage ranged from 66%o to 307o of the allocated total population; 7 workers in fact screened over 50%7o of their target population. Out of the 29 295 subjects whose oral cavity was examined, 1220 (4.2%) were diagnosed by the PHC workers as having an oral lesion needing a re- examination at the referral centre. Table 3 shows the age and sex distribution of these 1220 individuals. Table 2. Numbers and percentages of the population screened by three groups of PHC workers for oral precancer or cancer lesions in Kadugannawa Number or percentage of population screened by: Section of the Public health Public health population Midwives inspectors nurses All 3 groups (n = 23) (n = 9) (n = 3) (n = 35) Total number of adults 64 280 1 5 932 7065 87 277 ( > 20 years) Total number screened 22 825 4986 1484 29 295 Percentage of total 35.5 31.3 21.0 33.6 population screened Percentage of male 29.7 25.9 14.5 27.7 population screened Percentage of female 41.4 37.3 30.0 39.9 population screened 245 K. A. A. S. WARNAKULASURIYA ET AL. Table 3. Age and sex distribution of 1220 cases with oral lesions detected by PHC workers in Kadugannawa Age group (years) Males Females 20-29 88 22 30-39 161 55 40-49 199 67 50-59 212 85 60-69 142 68 > 70 89 28 Unspecified 4 Total 895 325 Patient's response to referral. Of the 1220 subjects with a suspected lesion, only 614 (50.3 o) were examined by the university project staff either at the referral centre or by prearranged field visits (Table 4). The initial drop-out rate was very high because as many as 837 subjects did not report to the referral centre; 543 of them were reminded by post and only 91 (177o) of these responded. Another 141 subjects were re-examined during field visits. Reliability of the PHC workers' diagnoses. Of the 614 cases re-examined at the referral clinic, 49 were excluded from analysis because of inadequate infor- mation. Among the remaining 565 cases, 400 were referred by the PHC workers with a stage 1 lesion, 150 with a stage 2 lesion, and 15 with a stage 3 lesion. Table 5 gives the detailed diagnoses made at the referral centre. From these results, it appears that the PHC workers detected 3 oral cancers (histologically Table 4. Follow-up of subjects referred by the PHC workers for re-examination at the referral centre Number Subjects referred by PHC workers 1220 Subjects arriving directly after referral 383 (31.4)b Subjects arriving after a postal remindera 91 Subjects re-examined during field visits 140 Total re-examined 614 (50.3)b e Number of postal reminders: 543. b Figure in parentheses is a percentage. confirmed), 116 high-risk stage 2 precancers, and 222 low-risk stage 1 precancers; 64 subjects (11 lo) had no demonstrable oral lesion when re-examined. Assessment of negative findings, excluding false negatives, i.e., among the subjects not referred by the PHC workers, was done on an ad-hoc basis by re- examining 1212 of the screened cases from the study area. Altogether 21 subjects (1.7%o) with an oral lesion were detected; all were relatively small in size (< 2 cm in diameter) except for one non- precancerous oral mucosal lesion, 5 were stage 2 lesions, and 15 were stage 1 lesions. No cancers were present among the negative sample. Cancer/precancer detection by medical/dental officers (Gampola study) Information supplied by the medical/dental officers indicated that a total of 21 318 subjects were screened. Of this number, the medical officers Table 5. Clinical diagnoses, after re-examination at the referral centre, of subjects with stage 1 -3 lesions detected by PHC workers in Kadugannawa Clinical diagnosis Homogeneous Speckled Cancer or Oral lesion No oral Totals or ulcerated leukoplakia, treated cancer but not lesion Staging as given leukoplakia erythroplakia premalignantb present by the PHC worker or oral sub- mucuous fibrosis Stage 1: lesion for observation 174 55 1 (10) 119 51 400 Stage 2: lesion for investigation 48 60 0 1 )a 29 1 2 150 Stage 3: Cancer or treated cancer 0 1 2 (10)a 1 1 15 Totals 222 116 14 149 64 565 (39%) (21 %) (2%) (26%) ( 1 %) (100%) a Figures in parentheses indicate the number of subjects with treated cancers. b Diagnoses such as lichen planus, leukoedema, preleukoplakia, fibroepithelial polyp, and denture hyperplasia. 246 CANCER DETECTION BY PRIMARY HEALTH WORKERS Table 6. Clinical diagnoses, after re-examination at the referral centre, of subjects with stage 1- medical/dental officers in Gampola -3 lesions detected by Clinical diagnosis Homogeneous Speckled Cancer or Oral lesion No oral Totals or ulcerated leukoplakia, treated cancer but not lesion Staging as given by leukoplakia erythroplakia premalignant"b present the medical/dental officer or oral sub- mucuous fibrosis Stage 1: lesion for observation 17 4 3 (O)1 11 3 38 Stage 2: lesion for investigation 3 2 1 (0)" 4 0 10 Stage 3: Cancer or treated cancer 1 2 5 (O)a 2 0 10 Totals 21 8 9 17 3 58 (36%) (14%) (1 5%) (30%) (5%) (100%) a Figures in parentheses indicate the number of subjects with treated cancers. b Diagnoses such as lichen planus, leukoedema, preleukoplakia, fibroepithelial polyp, and denture hyperplasia. screened 1141 subjects, dental surgeons 11 318 sub- jects, and assistant medical practitioners 8859. Of the 15 persons who received the one-day training, 3 did not participate in the study. Five officers screened less than 500 subjects each. A total of 133 cases with an oral lesion were detected in the Gampola area, but only 66 reported to the referral centre; 58 of them were diagnosed as follows -38 as stage 1, 10 as stage 2, and 10 as stage 3. Table 6 shows how these diag- noses compare with the final clinical diagnosis made at the referral centre, where 9 histologically con- firmed cancers, 8 high-risk stage 2 precancers, and 21 low-risk stage 1 precancers were identified; 3 subjects (5.2%) had no oral lesion to be seen when examined. Table 7 compares the total number of cases from the two study areas based on the final diagnoses made by staff at the referral centre. DISCUSSION This study represents a pioneering effort to test a primary health care approach to oral cancer control in a high-risk area. In the Sri Lanka health care system, there are more PHC workers than medical doctors and the former are more evenly distributed through- out the country (8). Consequently the successful utilization of PHC workers for cancer screening would not only result in good population coverage but also bring health care, including cancer control, closer to where the people live and work. In Kadugannawa, these workers demonstrated a high degree of motivation to participate in the study, only one out of the 35 trained workers abstaining Table 7. Distribution of subjects referred from Kadugan- nawa and Gampola, according to the clinical diagnoses after re-examination at the referral centre Final clinical diagnosis Kadugannawa Gampola Overt cancer 3 9 Treated cancer 11 0 Speckled leukoplakia, erythroplakia, or oral submucous fibrosis 11 6 8 Homogeneous or ulcerated leukoplakia 222 21 Benign oral lesions, not premalignant 149 1 7 No lesion (false + ve) 64 3 Total 565 58 from this work. The good population coverage achieved shows that cancer control activities, which are an important part of a country's health system, can be carried out by paramedical workers who are already frontline public health workers in the state health sector. The successful inclusion of cancer detection into their work schedule is consistent with present trends in Sri Lanka to expand these workers' responsibilities over a wider range of diseases, in- cluding cancer. Despite the encouraging rate of population cover- age, a disproportionately larger number of females (39.9%) were screened compared to males (27.7Gb). This was true for all 3 categories of PHC worker (Table 2) and may partly be due to the fact that the numerically largest category were the midwives (all 247 K. A. A. S. WARNAKULASURIYA ET AL. females) whose routine duties usually involve mothers and children. On the other hand, it is likely that since a great deal of screening was carried out during home visits, many males were away at work and were there- fore excluded. Additional screening of people at their work places would be desirable in other programmes of this kind. On the question of whether the screened population was representative of the total population with respect to disease level in the community, or whether there was a tendency to exclude individuals at greater risk (such as older males), our prevalence figure of 4.207o for oral premalignancy is consistent with the findings for other comparable populations (4). An investigation carried out subsequently on a random sample of 3100 previously unscreened subjects, in which more males than females were screened, yielded a prevalence of 3.9% with a closely similar range of oral lesions in terms of severity. This confirmed that the screening method utilizing house- to-house visits is appropriate for case detection. The results with respect to the compliance rate were disappointing; only 50.3% of referred cases were actually re-examined, even though provision was made for travel costs to be reimbursed. Owing to this low acceptance rate, the benefits of case-finding are reduced. Poor compliance may have been due to lack of awareness in the community about the serious nature of oral cancers and the value of this screening programme, and perhaps inadequate dialogue be- tween the health worker and the families on this subject. Further studies have been started to see whether a health education programme to promote increased awareness of the risks of a late cancer, which should be carried out during the screening pro- gramme, would significantly improve the follow-up of referred subjects. Successful implementation of a programme of this kind depends not only on adequate population coverage but also on the ability of health workers at the primary level to carry out screening with suf- ficiently high diagnostic accuracy. Consequently, an important objective of this study was to develop an appropriate diagnostic methodology for the early detection of oral cancer and precancer, which could be used by paramedical persons with limited training and experience. The classification of staging of lesions devised for this study was found to be simple enough for use in field conditions; it was also possible to differentiate the lesions in a way that was clinically significant and related to their prognosis and treat- ment needs, i.e., as low-risk stage I lesions with a much lower potential for malignant transformation than high-risk stage 2 lesions with a 25%7o rate of malignant transformation over several years (10). The results (Table 5) show that among the 400 stage I cases detected by the PHC workers there was only one cancer case; false postives amounted to 13%o. Of the 150 stage 2 cases, 60 (40%) were correctly diag- nosed, the remainder being either innocent leuko- plakia or benign mucosal lesions; one treated cancer was misdiagnosed as a stage 2 lesion. Of the 15 stage 3 cases, 801o were correctly diagnosed, with only one false positive. In the total of 565 cases the proportion of false positives was Ilo; in other words, (not counting over- or underdiagnoses) the sensitivity of diagnosis achieved was 89%7o. Re-examination of a sample of negative subjects, following screening by the PHC workers, revealed a very low level of false negatives; not a single cancer had been missed, an important requirement in cancer screening. Only five of these non-referred subjects (0.4% out of 1212) had stage 2 lesions. It is thus encouraging to find that these workers were able to detect almost all the oral lesions in the population screened. The overall performance of the PHC workers was very satisfactory in terms of diagnostic accuracy, con- sidering their level of education, the very brief training given to them in oral cancer detection, and the field conditions in which the diagnoses were made. Furthermore, their reliability in diagnosing the stage 1 lesions may justify the proposal that these lesions could in practice be diagnosed and followed up in the field by these workers themselves without the need for referrals, except when necessary. The Gampola study must be viewed as ancillary to the main project in Kadugannawa, the two parts being operationally different in several respects. A conspicuous finding from Gampola was that 5 out of the 12 medical/dental officers screened less thar. 500 subjects each, when in fact an estimated 10 000 people would have been seen by each of them while attending to routine duties during the study period. This may be a reflection of the heavy workload re- lated to routine duties and the consequent inability to cope with the added burden of a cancer screening programme. A total of 133 cases (0.607o) for referral after 21 318 examinations appears to be very low in terms of expected prevalence rates. There is no reason to believe that the prevalence of these precancer cases in Gampola should be substantially lower than that in Kadugannawa, the two areas being geographically adjacent and with socioeconomically comparable populations. On the other hand, it may be argued that the medical and dental officers were more selective in their referrals. Further, if one were to apply the rate of new cancer cases detected as a measure of success of the two methods, Gampola with 9/21318 (42.2 cases per 100 000) scores more than Kadugannawa with 3/29295 (10.2 cases per 100 000). However, the 9 new cases detected in Gampola were among the patients who presented themselves at the institution complaining of oral symptoms rather than through screening of the population. This may explain the 248 CANCER DETECTION BY PRIMARY HEALTH WORKERS 249 low prevalence of symptom-free precancers in the Gampola study. A comparison of the results from Kadugannawa and Gampola shows that the PHC workers were as reliable as their medical/dental counterparts. With regard to stage 1 referrals, the reliability of the two groups was almost identical. For stage 2 referrals, 400/o (60 out of 150) of the PHC workers' diagnosis were exact compared with only 20%7o (2 out of 10) among medical and dental officers. Further, the PHC workers demonstrated a better diagnostic reliability in the detection of overt cancer. A significant finding was the very modest cost of implementing this study. Excluding the cost for three visits by foreign consultants, the total cost for both the Kadugannawa and the Gampola programmes was less than US $2000. Cost-effectiveness is a relevant consideration especially in the context of a developing country, and the economy of the approach tested was mainly due to successful integration of the cancer control activity into the normal work schedule of the PHC workers, thus obviating the need for new re- cruitment of manpower. The small incentive payment per detected case was justified because it helped in motivation during the screening of almost 30 000 persons by primary health care workers. In Sri Lanka this task would have taken an estimated 10 years utilizing the existing dental care system, compared with 52 weeks by the newly tested approach. These results demonstrate an important point, namely, that primary health care workers in Sri Lanka, despite their limited experience and back- ground, can be trained to carry out systematic screening for precancers and cancers in the oral cavity at a level of diagnostic reliability consistent with what might be expected from skilled professional persons with dental or medical training. These PHC workers are more aware of the local customs that put certain individuals at risk, as a result of dangerous cancer- related chewing habits, than hospital-based profes- sional health workers. The direct, first-level contact between PHC workers and the local population is also of fundamental importance in this approach to the identification of high-risk groups. In conclusion, it can be stated that (1) the feas- ibility of utilizing primary health care workers for the early detection of oral cancer and precancer lesions, as well as their ability to diagnose these lesions accur- ately, has been established; (2) the problem of a gen- eral lack of compliance by the referred cases to pre- sent themselves for follow-up has been exposed; and (3) the study shows how primary-level health care re- sources can be utilized to detect previously undiag- nosed cancer and precancer cases in the community. RESUME INTERVENTION DES AGENTS DE SOINS DE SANTE PRIMAIRES DANS LE DEPISTAGE PRECOCE DU CANCER ET DES ETATS PRECANCEREUX DE LA BOUCHE A SRI LANKA Le cancer de la bouche constitue un probleme grave de sante publique dans les pays de I'Asie du Sud-Est. A Sri Lanka et en Inde, 35% A 40% des cancers seraient des cancers de la bouche, qui peuvent etre gueris s'ils sont depistes A temps. L'idee que, dans les pays en developpe- ment, l'un des rares moyens pratiques de depistage precoce pourrait faire intervenir les agents de soins de sante pri- maires a e testee sur le terrain a Sri Lanka. Dans une zone test, les sujets presentant des 1lsions buccales ont e identi- fies par des medecins et des dentistes. Trente-quatre agents de soins de sante primaires se sont occupes, outre leurs travaux habituels, d'examiner la cavite buccale de 28 295 personnes pendant 52 semaines; ils ont decele 1220 sujets porteurs de 1lsions exigeant un nouvel examen. Le travail des agents a et efficace en ce qui concerne le nombre de cas depistes et l'exactitude du diagnostic, qui a et verifie par un nouvel examen dans un centre d'orientation/recours spe- cial. Le diagnostic clinique des trois categories de 1lsions etait le suivant: premier stade -lsions a observer (leuco- plasie homogene ou ulcereuse); deuxieme stade - 1sions a analyser (leucoplasie tachetee, erythroplasie ou fibrose sous-muqueuse); troisieme stade - lesions a traiter (cancer). II y avait un degre de concordance de 89% entre le diagnostic des stades 1 a 3 porte par les agents de sante et le diagnostic clinique effectue au centre d'orientation/recours. Toutefois, la docilite des malades etait faible, 50% seule- ment des cas depistes sur le terrain s'etant presentes au centre. La rentabilite de ce mode de lutte contre le cancer est demontree. REFERENCES 1. BINNIE, W. H. Oral cancer. In: Dolby, A. E. ed., Oral mucosa in health and disease. Oxford, Blackwell, 1975, pp. 301-333. 2. PINDBORG, J. J. Epidemiological studies of oral cancer. International dental journal, 27: 172-178 (1977). 250 K. A. A. S. WARNAKULASURIYA ET AL. 3. PINDBORG, J. J. Oral cancer and precancer. Bristol, John Wright, 1980. 4. MEHTA, F. S. ET AL. Oral cancer and precancerous conditions in India. Copenhagen, Munksgaard, 1971. 5. HIRAYAMA, T. An epidemiological study of oral and pharyngeal cancer in central and south-east Asia. Bulletin of the World Health Organization, 34: 41-69 (1966). 6. COLE, P. & MORRISON, A. S. Basic issues in population screening for cancer. Journal of the National Cancer Institute, 64: 1263-1272 (1980). 7. WORLD HEALTH ORGANIZATION. Guide to epidemiology and diagnosis of oral mucosal diseases and conditions. Community dentistry and oral epidemiology, 8: 1-26 (1980). 8. Alma-Ata 1978. Primary health care. Geneva, World Health Organization, 1978. 9. SIMEONOV, L. A. Better healthfor Sri Lanka. Report on a health manpower study. New Delhi, WHO Regional Office for South-East Asia, 1975. 10. GUPTA, P. C. ET AL. Incidence rates of oral cancer and natural history of oral precancerous lesions in a 10-year follow-up study of Indian villagers. Community den- tistry and oral epidemiology, 8: 287-333 (1980).