Statistical issues in human papillomavirus testing and screening.

As discussed in the preceding sections, regardless of the research purpose, etiology or prevention, there are critical statistical and study design issues related to measurement of HPV infection status and its cervical lesion outcomes that need to be considered for the appropriate interpretation of the association of HPV infection and its determinants with cervical cancer in epidemiologic studies, and of screening effectiveness by HPV testing in clinical and intervention trials. These statistical issues may affect the validity of epidemiologic and screening studies and have led to inconsistent results in the literature. Meticulous attention to study design and laboratory detection issues helps to minimize the impact of such measurement errors and detection biases. Simple statistical analysis techniques are also available to correct or to control for these biases if they were to be identified and additional information is known concerning the expected test performance.