Association of schistosomiasis with cervical cancer: detecting bias in clinical studies
Riffenburgh, R.H.; Olson, P.E.; Johnstone, P.A.
East African Medical Journal 74(1): 14-16
1997
ISSN/ISBN: 0012-835X PMID: 9145570 Document Number: 482538
To investigate if Schistosoma haematobium infection increases the observed frequency of cervical cancer, data from literature were analysed in meta-analysis fashion. Results reveal that cervical cancer is statistically significantly less frequent in the presence of S. haematobium infection. A protective effect might be inferred; however, underlying bias in data collection and/or analysis is suspected. An approach to detecting bias is given, consisting of: (a) carefully stating and assessing the several components of the scientific method and; (b) identifying and contrasting the populations hypothesized and sampled. The approach is illustrated by searching for bias in the schistosomiasis-and-cancer reports. Sampling discrepancies were detected in patient geography, ages, and states of health, and suspected in disease prevalence sampled and disease prevalence reported. The conclusion is reached that effects of bias in the original studies preclude inference of a "protective" effect of S. haematobium infection against cervical cancer.