Estimating the Effect of Childhood Socioeconomic Disadvantage on Oral Cancer in India Using Marginal Structural Models

Background: Early life socioeconomic disadvantage could affect adult health directly or indirectly. To the best of our knowledge, there are no studies of the direct effect of early life socioeconomic conditions on oral cancer occurrence in adult life. Methods: We conducted a multicenter, hospital-based, case-control study in India between 2011 and 2012 on 180 histopathologically confirmed incident oral and/or oropharyngeal cancer cases, aged 18 years or more, and 272 controls that included hospital visitors, who were not diagnosed with any cancer in the same hospitals. Life-course data were collected on socioeconomic conditions, risk factors, and parental behavior through interview employing a life grid. The early life socioeconomic conditions measure was determined by occupation of the head of household in childhood. Adult socioeconomic measures included participant's education and current occupation of the head of household. Marginal structural models with stabilized inverse probability weights were used to estimate the controlled direct effects of early life socioeconomic conditions on oral cancer. Results: The total effect model showed that those in the low socioeconomic conditions in the early years of childhood had 60% (risk ratio [RR] = 1.6 [95% confidence interval {CI} = 1.4, 1.9]) increased risk of oral cancer. From the marginal structural models, the estimated risk for developing oral cancer among those in low early life socioeconomic conditions was 50% (RR = 1.5 [95% CI = 1.4, 1.5]), 20% (RR = 1.2 [95% CI = 0.9, 1.7]), and 90% (RR = 1.9 [95% CI = 1.7, 2.2]) greater than those in the high socioeconomic conditions when controlled for smoking, chewing, and alcohol, respectively. When all the three mediators were controlled in a marginal structural model, the RR was 1.3 (95% CI = 1.0, 1.6). Conclusion: Early life low socioeconomic condition had a controlled direct effect on oral cancer when smoking, chewing tobacco, and alcohol were separately adjusted in marginal structural models.