Limitations of using a cancer registry to identify incident primary intracranial tumours.

The completeness and accuracy of registration of primary intracranial tumours in the Scottish Cancer Registry was compared with a detailed incidence study performed over a two year period (1989-90). Of 228 patients with any primary intracranial tumour in the incidence study, 124 (54%) were identified as intracranial tumours in the cancer registry. The registry excluded benign tumours (although this was not consistent) and so the sensitivity of the registry varied with tumour type (84% for neuroepithelial tumours, 22% meningeal, 29% sellar, 0% cranial nerve). Of the 31 malignant tumours not found in the registry on our initial search, nine were found to have been included between 1989-90 but using different International Classification of Diseases-9th revision (ICD-9) codes or postcodes, and seven were found registered after 1990. Eleven per cent of cases (18/170) identified in the cancer registry were excluded from the incidence study: 11 had evidence of an intracranial tumour before 1989 whereas four definitely did not have an intracranial tumour. The cancer registry therefore significantly underestimated the incidence of all primary intracranial tumours, and of malignant intracranial tumours. Incidence studies must use additional methods to identify all primary tumours. Cancer registries should consider registering all primary intracranial tumours and may improve case ascertainment by screening neuroradiology data.