If this difference is statistically significant, which form of bias may be affecting the results?

A randomized controlled trial is conducted investigating the effects of different diagnostic imaging modalities on breast cancer mortality. 8,000 women are randomized to receive either conventional mammography or conventional mammography with breast MRI. The primary outcome is survival from the time of breast cancer diagnosis. The conventional mammography group has a median survival after diagnosis of 17.0 years. The MRI plus conventional mammography group has a median survival of 19.5 years. If this difference is statistically significant, which form of bias may be affecting the results?

    1. Selection bias
    1. Misclassification bias
    1. Lead-time bias
    1. Recall bias
    1. Because this study is a randomized controlled trial, it is free of bias

0 voters

EXP

Lead-time bias occurs when an increased ability to detect disease increases the survival time after diagnosis without changing the course of the disease.

Despite the fact that randomized controlled trials (RCTs) are considered the gold standard in determining clinical efficacy, they are not completely immune to bias. Lead-time biases are especially problematic in RCTs that are investigating the efficacy of new diagnostic procedures. The hope of new diagnostic procedures is that detecting disease earlier will improve prognosis. The potential for lead time bias clouds this picture, as the detection of disease before it becomes symptomatic will result in an apparent prolongation of survival after diagnosis, even if early detection has no impact on natural history of disease.

Rothman formally defines the period of the lead-time. In the course of any disease, the first event to occur is the biologic onset of disease, e.g. when the very first cell becomes malignant. At this point the disease is asymptomatic, which is why diagnostic procedures performed during this time are considered screening tests. Eventually the disease will manifest with symptoms. The period between a positive screening and the onset of symptoms is the lead time. Rothman cautions that one should not count lead-time towards the survival time because it does not represent any real benefit in and of itself. In the absence of any benefit from early treatment of the disease the lead-time is simply a period during which the patient is aware of his disease, but the date of his death is unchanged. If early detection cannot postpone his death than an increased lead-time is not only unhelpful, but possibly psychologically detrimental.

Gordis expands on the concept of lead-time bias further. Increasing lead-time seem to cause increases in survival, but the patient is not any better off because death has not truly been delayed. Furthermore, lead-time bias can affect the results of trials whose outcomes are not mortality. The same bias can occur when investigating morbidity, recurrence of disease, quality of life, or patient satisfaction.

Illustration A depicts the natural progression of disease from biologic onset of breast cancer to death. The difference between the apparent lengths of survival between MRI and conventional mammography is the lead-time bias.