Effects of cerebral magnetic resonance imaging in outpatients on observed incidence of intracranial tumors and patient survival: a national observational study

Object. It is assumed that the observed increase in brain tumor incidence may at least partially be explained by increased use of MRI. However, to date no direct estimate of this effect is available. The authors undertook this registry-based study to examine whether regional frequencies of cerebral MRI use correlate to regional incidence rates of intracranial tumors and survival of patients with these lesions. Methods. The authors used Norwegian national population registries from January 2002 through December 2007 to conduct this observational study. They obtained information on outpatient MRI scans in Norwegian counties and examined whether the annual regional rates of cerebral MRI scans correlated to regional age- and sex-adjusted brain tumor incidence rates. They also explored whether differences in cerebral MRI use were associated with survival and examined time trends in the study period. Results. Approximately 50,000 cerebral MRI scans are carried out annually in outpatient settings in Norway, and 6363 primary intracranial tumors were diagnosed in Norway during the study period. There was an overall positive correlation between the annual number of cerebral MRI scans per 100,000 capita and age- and sex-adjusted incidence rates of intracranial tumors in the various Norwegian counties (Spearman's rho =0.35, p <0.001). In a linear model, an increase in 1 MRI per 100,000 capita per year results in a 0.004(95% CI 0.002-0.006) increase in diagnosed intracranial tumors per 100,000 capita per year (p <0.001). Subgroup analysis showed a correlation between MRI use and the annual age- and sex-adjusted incidence rates of extraaxial tumors (p = 0.04, Spearman's rho = 0.28) but not intraaxial tumors (p = 0.394). Overall survival for unselected patients with intracranial tumors is longer with increasing number of cerebral MRI scans per capita in the county of residence at the time of the diagnosis (log rank, p = 0.029). However, after adjustment for year of diagnosis and catchment region of the Norwegian neurosurgical centers, the association between MRI scans per capita and overall survival was no longer statistically significant (p = 0.076). Conclusions. Presumably due to the incidental discovery of benign extraaxial tumors, regional differences in the use of cerebral MRI in outpatients affect observed incidence rates of intracranial tumors.