Healthy User Bias

Healthy-User Bias

Healthy-user (or healthy-adherer) bias may arise when patients who receive a therapy tend to more actively to seek out preventative care and engage in other healthy behaviors than patients who do not receive therapy. Health-seeking behaviors may include eating a healthy diet, exercising regularly, and wearing a seat belt. Incomplete adjustment for such behaviors can lead to spurious inferences regarding the safety of the device of interest because healthy behaviors are associated with a reduced risk of a number of poor health outcomes. For example, a study comparing insulin pumps with real-time continuous glucose monitoring to self-injecting insulin regimens may overstate the effect of insulin pumps on outcomes such as mortality, stroke, or myocardial infarction if the patients receiving insulin pumps sought out this treatment with the aim of obtaining tighter glucose control. Controlling for healthy-user bias in PASS tends to be quite challenging when using secondary data sources, as healthy behaviors and personality traits they represent are difficult to measure in such data sources. Proxies may include evidence of health screening such as yearly physicals, mammography, or colonoscopy; or the receipt of preventative therapies such as influenza vaccine.

Additionally, patients who adhere to their treatment are more likely to engage in healthier behaviors than nonadherent patients. This source of bias may be of concern in PASS where a device may be preferentially used to treat patients with difficulty adhering to their treatments. In the earlier example, if insulin pumps are reserved for patients with poor adherence to their self-injecting insulin regimens, estimates may be biased against the pumps because nonadherent patients may be engaging in more risk-taking and less health-seeking behaviors than patients remaining on the self-injecting insulin regimens. Failing to account for the behaviors that correlate with treatment adherence could lead to erroneous conclusions regarding the safety of a device.

• Science Direct