Early findings from a community-based, cluster-randomized, controlled oral cancer screening trial in Kerala, India

Oral cancer satisfies the criteria for a suitable disease for screening, and oral visual inspection is a suitable test for oral cancer screening. The efficacy of screening in reducing mortality from oral cancer has not yet been evaluated. The authors describe a cluster-randomized, controlled oral cancer screening trial in southern India and its early results.Apparently healthy subjects age 35 years or older in 13 clusters called panchayaths were randomized to either an intervention group (n = 7) or a control group (n = 6). Subjects in the intervention group will receive 3 rounds of screening consisting of oral visual inspection by trained health workers at 3-year intervals. The first round of screening was carried out between October 1995 and May 1998. Participants were visited in their homes and interviewed for sociodemographic details, tobacco-smoking and alcohol-drinking habits, and personal medical history. Those with tobacco or alcohol habits were advised to stop those habits. Subjects in the intervention group were offered screening, and those with lesions suggestive of oral leukoplakia, submucous fibrosis, or oral cancer were referred for examination by physicians. Confirmed leukoplakias were excised whenever possible, others were kept on follow-up, and those with confirmed oral cancers were referred for treatment. Data on oral cancer incidence, stage distribution, survival, and mortality in the study groups are obtained by record linkage with the Trivandrum population-based cancer registry and municipal death registration systems.There were 59,894 eligible subjects in the intervention group and 54,707 in the control group; 31.4% of the former group reported no tobacco or alcohol habits, compared with 44.1% of the latter. The distribution of age, education, occupation, income, and socioeconomic status were similar in the two groups. Of 3585 subjects in the intervention group referred, 52.4% were examined by physicians; 36 subjects with oral cancers and 1310 with oral precancers were diagnosed. Of the 63 oral cancers recorded in the cancer registry, 47 were in the intervention group and 16 were in the control group, yielding incidence rates of 56.1 and 20.3 per 100,000 person-years in the intervention and control groups, respectively. The program sensitivity for detection of oral cancer was 76.6% and the specificity 76.2%; the positive predictive value was 1.0% for oral cancer. In the intervention group, 72.3% of the cases were in Stages I-II, as opposed to 12.5% in the control group. The 3-year case fatality rates were 14.9% (7 of 47 patients) in the intervention group and 56.3% (9 of 16 patients) in the control group.Though compliance with referral for confirmatory examination in the first round was lower than the 70% anticipated, intermediate end points, such as stage at diagnosis and case fatality, indicate that the trial is making fairly satisfactory progress.