Use of computerised general practice database for epidemiological studies in italy: a comparative study with the official national statistics

Introduction: The Italian College of General Practitioners (SIMG) set up the Health Search Database (HSD) in 1998, and it was primarily aimed at carrying out studies of incidence and prevalence and studies on drug safety and prescription. Data from general practice databases are generally able to supply the denominator for epidemiological studies in a way that questionnaire based studies cannot. However, when considering numerator data the completeness of the information reported varies, and it could be affected by certain patient characteristics. Objectives: To compare primary care morbidity and drug prescription with official national statistics, and to examine how the differences vary across the diseases, and the different therapeutic classes. Methods: We compared morbidity and prescription information obtained from 432 747 patients from the HSD, with either morbidity data from 119 799 individuals selected for the Sixth Health Interview Survey (HIS), or drug prescription data as assessed from the National Drug Monitoring Centre (OsMed), which routinely collects information on all prescriptions issued within the Italian National Health System. Comparing the age and sex specific prevalences within the same timeframe, and the age standardised prevalence assessed difference in morbidity. Drug consumption derived from the HSD was compared with that estimated from the OsMed, by using the DDD/1000 inhabitants/ day. We then calculated the pattern of use by dividing the number of DDD/1000 inhabitants/day for each therapeutic group with the total number of DDD/1000 inhabitants/day being prescribed. Results: No differences in morbidity for diabetes and hypertension was found among males, while among females, the HSD age adjusted prevalence per 1000 inhabitants was slightly lower compared with the HIS. The comparison between the HSD and the HIS for chronic obstructive pulmonary disease (COPD) (40.3 versus 55.5 females; 26.0 versus 44.5 males) and gastroduodenal ulcer (19.3 versus 39.8 females; 11.9 versus 31.0 males) showed relevant differences. Drug use in the HSD reported an overall understimate in comparison with OsMed. However, the general pattern of use across the different anatomical therapeutic chemical classes did not show any relevant variation. Conclusions: The difference between primary care and population morbidity seems generally low for diseases with clear cut diagnosis, whereas for diseases without clearly established diagnostic criteria (COPD and gastroduodenal ulcer) the discrepancy appears more relevant. For drug prescriptions, the general underestimate could be due to the lack of information about paediatric age. However, the similar prescribing pattern reported allows researchers to use information from the HSD to carry out pharmacoepidemiological studies generalised to the whole Italian general practitioner population.