The routine examination of asymptomatic and symptomatic patients can lead to detection of earlier stage cancers and premalignant lesions. There is no definitive evidence, however, to show that this screening can reduce oral cancer mortality, and there are no randomized controlled trials (RCT) in any Western or other low-risk populations.[9,12,13,14,15]
In a single RCT of screening versus usual care, 13 geographic clusters in the Trivandrum district of Kerala, India, were randomly assigned to receive systematic oral visual screening by trained health workers (seven screened clusters, six control clusters) every 3 years for three screening rounds during the period of 1996 to 2004.[16,17,18] Cluster size varied from 5,177 to 12,147 (mean 8,815) participants. The number of interviewed (and analyzed) participants in the screening and control clusters were 87,829 and 80,086, respectively. During the 9-year period of the clinical trial, there were 77 deaths from oral cancer in the screening arm and 87 in the control arm. The cause-specific mortality rates were 16.4 versus 20.7 per 100,000 person-years (relative risk [RR] = 0.79; 95% confidence interval [CI], 0.51–1.22). In a subset analysis restricted to tobacco or alcohol users, the cause-specific mortality rates were 29.9 versus 45.4 per 100,000 person-years (RR = 0.66; 95% CI, 0.45–0.95). In male tobacco or alcohol users, the RR was 0.57 (95% CI, 0.35–0.93) and in female users, the RR was 0.78 (95% CI, 0.43–1.42). There was a qualitative difference in screening effect for nonusers: 3.0 versus 0.9 per 100,000 person-years in the screened and control arms (RR = 3.47; 95% CI, 0.12–96.51). Although there was a higher number of early-stage (I and II) cases in the screened arm versus the control arm (85 vs. 37), the number of late-stage (III and IV) cancers was similar (104 vs. 105). Therefore, there was not a clear, true stage shift, even though the proportion of cases with early-stage disease was higher in the screened arm.
Aside from the issue of generalizability to other populations and from the overall lack of a statistically significant result in cause-specific mortality, serious methodologic problems make interpretation of the results difficult, including lack of detail about the randomization process, lack of allocation concealment, and lack of adjustment for clustering effect. The total number of clusters randomized was small, and there were different distributions of income and household possessions between the two study arms. Withdrawals and dropouts were not clearly described. In summary, the sole randomized trial does not provide solid evidence of a cause-specific mortality benefit associated with systematic oral cavity visual examination.