Abstract
Dear Sir,
I read with great surprise the article of Ferrante et al. (1), published in Cephalalgia, concerning an observational study in collaboration with general practitioners (GPs) about the prevalence of migraine in the Italian city of Parma. They reported a very high prevalence compared to many previous epidemiological studies performed worldwide. They found a prevalence of 24.7% of definite migraine (DM) and 5.1% of probable migraine (PM): so migraine sufferers are about one third of the population. A previous multicentre study conducted by GPs in Italy found a prevalence of 6.9% of DM and 4.6% of PM (2). The authors discussed that the differences between their results and literature data are likely to be due to different methodological approaches.
The reported high prevalence contrasts with two aspects of migraine management not only at primary care level in Italy, but also worldwide:
The migraine diagnosis registered in the GP’s databases in Italy is rather low. The analysis of about 600,000 patient records, carefully conducted with ‘Health Search’ (a method dedicated to optimize the registration of every problem in the electronic records), shows a percentage of 1.39% of migraine in 2001 and of 4.4% of headache not otherwise specified (3). The same analysis repeated in 2006 reported a percentage of migraine diagnosis slightly higher (1.82%) (4). These data were recently confirmed with the database analysis of a cooperative of GPs, showing 1.6% of ‘migraine’ and 2% of ‘headache’ on a population of 42,710 residents (5). A low percentage (about 0.5%) of triptan users in the general population was reported by some epidemiological studies (6,7) and recently confirmed (0.8%), including the same area of the authors (8).
The enormous discrepancy between both the low percentage of recorded migraine diagnosis and of triptan prescription at primary care level with the high prevalence of migraine detected in epidemiological studies (in particular in that of Ferrante et al.) may be due at least in part, and for various reasons, to the under-diagnosis of migraine by GPs (9).
However, the authors neglected to discuss and to evaluate a factor that may have been decisive in their results, namely, the method of recruitment of study participants. The study participants were recruited by GPs in 2007–2009 from those interested in participating and they arranged an interview at their office, Headache Centre or by phone. This may be the most important factor in determining the outcomes. In fact the sample covered only a proportion of patients, most likely those reaching the office of the GP for examination in the period of study, which are typically those with more diseases, including those migraine patients with more frequent and/or severe attacks and comorbidities. In addition, the GP could more easily be inclined to recruit patients already diagnosed as migraine patients in the database. Instead, the previous Italian epidemiological study covered all patients who visited the GP’s practice, for any reason, on 5 consecutive days (2). Patient selection should be random or involve the entire database of a few GPs.