Intervention studies that use cancer patients from high-risk clinics may be subject to bias
Many studies have tried to evaluate the effects of interventions to reduce cancer risk, such as tamoxifen therapy or prophylactic surgery, among patients that carry a mutation that increases their risk of cancer. However, some of the studies that use subjects from clinics that care for people at high risk of the disease may be subject to serious selection bias, according to a commentary that appears in the August 18 issue of the Journal of the National Cancer Institute.
Researchers who want to study the efficacy of interventions meant to reduce cancer risk in people who carry cancer-predisposing mutations, such as the BRCA1 and BRCA2 mutations for breast and ovarian cancers, face several challenges. Recruiting enough patients for a study can be difficult and expensive. In addition, patients who carry these mutations are rare and are often identified only after someone in their family has developed the cancer. Clinics that serve patients at high risk of developing these cancers can therefore be a tempting source of study participants.
However, the characteristics of the patients that are recruited from high-risk clinics can bias a study's results, according to Sholom Wacholder, Ph.D., of the National Cancer Institute. In his commentary, Wacholder considers the possible sources of selection bias, which could lead to an overly optimistic perception of the value of the interventions being evaluated.
To obtain sufficient numbers of patients for a clinic-based study, researchers sometimes include as case patients people who had been diagnosed with cancer before they came to the clinic. However, the control group consists only of people previously seen at the clinic, whose chance of receiving the intervention is increased simply by virtue of receiving care at the clinic. Even if the intervention has no effect, these controls are, therefore, more likely to have received the intervention than the case patients who were diagnosed before being seen at the clinic. This can lead, generally, to a false impression of intervention efficacy since controls are more likely to receive interventions than some of the case patients, Wacholder says. Thus, researchers might be misled to either overestimate the protection of an intervention or underestimate its harm.
This bias may render clinic-based studies too flawed to serve as a reliable guide for critical management decisions for people at increased genetic risk of cancer, Wacholder writes.
Source: Eurekalert & othersLast reviewed: By John M. Grohol, Psy.D. on 21 Feb 2009
Published on PsychCentral.com. All rights reserved.