Aim: development of an algorithm to estimate asthma prevalence by record linkage of different health databases: causes of death (CM), hospital discharges (SDO), drug prescriptions archive (PF), health tax exemptions (ET) in three Italian areas.
Setting: Venezia AULSS 12, city of Torino, Firenze ASL 10.
Participants: residents aged between 0 and 34 years in the above three centres in the three year period 2002-2004, for a total of 661,275 inhabitants on 30 June 2003.
Main outcome: annual crude and standardized prevalence (per 100 inhabitants) with 95% confidence intervals by gender and age groups (0-14, 15-34, total: 0-34 years).
Methods: for each year of interest, we selected the following: cases with asthma as primary cause of death from CM; all persons discharged from hospitals with diagnosis (primary or secondary) of asthma (ICD9-CM code = 493*); selected prescriptions of anti-asthma drugs (ATC code = R03A, R03CC02, R03CC04, R03CK, R03DC01, R03DC03), and health-tax exemptions for asthma (code = 007.493) from ET. We defined as prevalent case a subject who was present every single year in at least one of four health databases. We reported the absolute and relative contribution of each information system by area, age, gender and year of interest. A sensitivity analysis using more restrictive criteria to identify prevalent cases (two or more prescriptions per year) was also performed.
Results: the PF archive is the most important information source in identifying prevalent cases (from 92.5% of Torino to 95.4% of Firenze). The standardized prevalence of asthma in 0-34 years of age is higher in Venezia (6.37%; 2003 year) than in the other two areas, which show similar values (4.01% in Firenze, 3.77% in Torino; 2003 year). In both genders, the standardized prevalence of asthma is, for all centers, clearly higher in the 0-14 age group than in the 15-34 age group. However, Venezia has a prevalence almost twice (11.21%) that of Firenze (6.20%) and Torino (5.60%) in the 0-14years age group. The use of more restrictive criteria in case identification consistently reduces the estimated prevalence; however, in the 0-14 age group the prevalence estimated in Venice (3.3%) is still almost twice as high as those observed in the other centres (1.8% in Florence and 1.6% in Turin).
Conclusions: the algorithm used to estimate asthma prevalence in the 0-34 years age group provides values which differ considerably between the centres that contributed to the study. A validation study is required to evaluate the diagnostic quality of the identified cases, in particular among younger subjects.