Selection biases can distort findings from cohort studies. Self-selection into a study cohort can yield results that are not representative of the target population. Screening a population to detect persons with high blood pressure, for example, can lead to misleading estimates of the effects of antihypertensive medication, in the absence of a control group, because subsequent blood pressure measurements will tend to be lower solely by virtue of the selection process. Studying a selected group of workers who were healthy before hiring can obscure the effects of occupational exposures in comparison to a study of the general population. Incomplete follow-up of cohort members can also lead to biased estimates of cumulative risk and relative risk. Some time-related biases can affect cohort studies. Left truncation arises when one only studies subjects who have survived to a certain time; the analysis must take left truncation into account to avoid bias. Prevalent cohorts consist of subjects who have a prevalent disease at the time of enrollment. Because the age at disease onset is often unknown in prevalent cohorts, estimates of cumulative and relative risks are subject to various biases.