Quality and variability of osteoporosis data in general practice computer records: implications for disease registers

To determine the extent to which routinely collected general practitioner computer data could be used to create disease registers of patients with osteoporosis, and to report any improvement in data quality since previous studies.

Audit using anonymized data extracted from general practice computer records from across England.

Morbidity Query Information and Export Syntax (MIQUEST) software was used to extract structured data from the 78 volunteer practices that participated in the study. The data were aggregated and analysed.

There were 100-fold differences in the rates of recording of relevant data. Many patients receiving treatment had no diagnostic codes. Data about secondary causes of osteoporosis and fractures were more consistently recorded than data relating to falls. There were no data to indicate whether fractures were low impact. T-scores, the gold-standard measure of bone density, were very infrequently recorded.

Sufficient data about secondary causes of osteoporosis exist, and these could be searched to identify patients at risk. Meanwhile, fracture recoding could be improved, including likely fragility fractures, and T-scores could be added to computer records. A systematic approach is needed to raise the computer records to a standard where they can be used as valid and reliable disease registers.