The reputations of certain clinicians and institutions cause individuals with specific disorders or exposures to gravitate toward them.
Centripetal bias as defined by Sackett in his 1979 indicates that the number of cases recorded at different institutions may be biased because patients might seek out clinicians or institutions with a good reputation in that clinical area, thus artificially inflating the numbers recorded. This may impact estimates of prevalence and the population under investigation may not reflect the characteristics of the wider population. (see Spectrum bias)
Sackett uses the example “e.g. the striking rate of posterior fossa cerebral aneurysms reported from the University of Western Ontario.” (Sackett 1979)
The impact of centripetal bias has not formally been evaluated. In prognostic studies, if a particular centre has more severe cases (due to its good reputation for treating the condition, for example), this may affect the results of the prognostic study. Economic factors might also influence these effects if more well-off patients can access these centres, but poorer patients are unable to do so. (See selection bias)
Surveys should use sampling techniques that aim to account for known and unknown differences in the distribution of factors affecting the condition of interest.
Catalogue of Bias Collaboration. Carl Heneghan. Centripetal bias. In: Catalogue Of Bias 2019. https://catalogofbias.org/biases/centripetal-bias/