Increasing prevalence of multiple sclerosis in Tuscany: a study based on validated administrative data

BACKGROUND Italy is a high-risk area for Multiple Sclerosis (MS) with a prevalence of around 140-170/100,000 (2005-2009) with the exception of Sardinia, with about 224 cases per 100,000 (2009). Nowadays, in Italy, prevalence is absolutely higher than the above estimates. Indeed, prevalence is rising due to annual incidence that is higher than annual mortality. In Tuscany a population MS register has been founded but, to date, it’s not yet completed. To monitor disease epidemiology, comorbidities and care pathways, but also to describe the disease burden and to plan its prevention, treatment and management strategies and resource allocation, population-based studies are preferable. Administrative data offer a unique opportunity for population-based prevalence study of chronic diseases such as MS. GOALS To update the prevalence of MS in Tuscany using a validated case-finding algorithm based on administrative data and to demonstrate the progressive increment of prevalence. METHODS The prevalence was calculated using an algorithm based on administrative data: hospitalization, MS drug dispensing, disease-specific exemptions from patient copayment, home and residential long-term care and inhabitant registry. To test algorithm sensitivity, we used a true-positive reference cohort of 302 MS patients from the Tuscan MS register. To test algorithm specificity, we used a general population cohort of 2,644,094 individuals who were presumably not affected by MS (who had never effectuated either cranial or spinal cord CT scan or MRI and had never received a neurological outpatient visit within the NHS). We calculated prevalence on three consecutive years (2011, 2012, 2013). RESULTS At prevalence date (31 December), we identified 6,890 cases in 2011, 7,057 in 2012 and 7,330 in 2013 with a rate of 187.9, 191.1 and 195.4/100,000, respectively. The female: male ratio slightly increased from 2.0 in 2011 to 2.1 in 2012-2013. The sensitivity of algorithm was 98% and its specificity was 99.99%. CONCLUSIONS We found a progressive increment of prevalence that confirmed our hypothesis of increasing prevalence. Our algorithm can accurately identify patients and this cohort is suitable to monitor care pathways. Our future aim is to create an integrated dataset with administrative and clinical data from MS register