Information about a group or individual coupled with suspicions or prejudices of medical staff could influence how diagnoses are made, by affecting what examinations are performed and how quickly people are investigated, which can affect rates of diagnosis. This can be termed diagnostic suspicion bias.
As an example, if a group of workers in the industry find out that one of the chemicals they have been exposed to is a carcinogen, then these workers might present to a medical facility sooner, or be more likely to attend screening, than a non-exposed population. Also, medical staff might more readily suspect these individuals than others to have cancer, because of the knowledge of their exposure to the carcinogen, and this might influence what tests are done and how quickly they are ordered.
Diagnostic test accuracy studies that include selected patients because they are more likely to have the condition based on clinical suspicion typically overestimate the accuracy of the test. Studies that used non-consecutive inclusion of patients were associated with an overestimation of the diagnostic odds ratio by 50% compared with those that used a consecutive series of patients.
A study of inpatient care for severe mental illness in the US found that researchers noticed that African Americans were three times (45% vs 19%) more likely to be diagnosed with schizophrenia than whites. Medical staff interviewing patients viewed African Americans as less honest about their symptoms, with less insight into their condition. These views were associated with higher diagnosis rates with diagnostic suspicion bias being responsible for some of the disparity in diagnosis rates.
Prospective studies, with consecutive recruitment of patients and with uniform assessment and measurement throughout the study, can help avoid problems from diagnostic suspicion bias. Where retrospective studies are set up, care must be taken to avoid the effects of diagnostic suspicion bias, checking how diagnostic procedures take place, and if necessary adjusting for disparities.