Mortality trend for multiple sclerosis in Italy (1974±1993) T. Tassinari 1 , S. Parodi 2 , R. Badino 1 & M. Vercelli 2,3 1 Azienda Ospedaliera, `Ospedale S. Corona', 17027 Pietra Ligure, Savona; 2 S.A. Registro Tumori, Istituto Nazionale Ricerca sul Cancro; 3 Dipartimento di Oncologia Biologia Genetica, Universita Á di Genova, Genova, Italy Accepted in revised form 11 April 2001 Abstract. The prevalence of multiple sclerosis (MS) in Italy is one of the highest in Europe, estimated proportions ranging between 32 and 69 cases per 100,000. An update of mortality time trend analyses can help trace a picture of the evolution of the disease and contribute to the interpretation of dierences related to latitude and gender. Standardized mortal- ity ratios (SMR) were computed for the 1974±1993 calendar period, using the pooled data set as stan- dard. World population was the standard used to estimate age-adjusted rates for all Italy, while for selected regions the 1991 Italian population was uti- lized. Time trend analysis was performed using the Poisson regression model and the variation of mor- tality rates in time was expressed as mean dierence per cent per year. Age-adjusted rates per million in- habitants were 4.1 for males and 5.0 for females. Northern Italian regions showed higher MS mortality rates than Central and Southern regions and Sicily, particularly in females; the eect was less evident in Sardinia. Time trend analysis by area showed a mortality decrease in Northern Italy (1.15%) that reached statistical signi®cance for men. In central regions a slight increase of borderline statistical sig- ni®cance (1.03%) was observed only in women, while in the South and Sicily a statistically signi®cant increase was seen in both males (2.14%) and fe- males (3.09%). The analysis of the time trend for all Italy did not reveal signi®cant variations in male mortality; by contrast, a slight, but statistically sig- ni®cant, increase (0.88%) was observed in female mortality. Analysis of age-adjusted MS mortality rates in each region suggested the presence of an ef- fect due to latitude. The occurrence of an increasing trend in Southern Italian regions and Sicily deserves further investigation. Key words: Mortality trend, Multiple sclerosis, Geographic distribution Introduction Several epidemiological studies conducted during this century have attempted to describe the distribution of multiple sclerosis (MS) in Europe and throughout the world. Despite these eorts, many uncertainties still surround the disease [1], with recent evidence even challenging the validity of the long-accepted direct relationship between incidence of MS and the distance from the equator [2]. Descriptive epidemiological studies have, however, eectively shown a dierent geographical distribution and time trends of MS incidence in various areas of the world [3]. In particular, Lauer [4], on the basis of epidemiological literature, included Italy in the high frequency countries. In fact, the prevalence studies carried out in Italy since 1980 have utilized more accurate survey techniques and have estimated prevalence of between 32 and 69 cases per 100,000 inhabitants [5]. Mortality data, despite their intrinsic limits, con- stitute a fundamental tool of any epidemiological survey and provide the starting points for the elabo- ration of new research eorts [6]. We therefore eval- uated the geographic distribution and time trend of MS mortality in Italy during the period 1974±1993. Material and methods Mortality ®gures, registered as code 340 of ICD-8 and ICD-9 classi®cations, were drawn from Italian National Institute of Health (ISS) and Institute for Statistics (ISTAT) electronic publications for the 1974±1993 calendar period. Mortality for MS was analyzed for both the 20 Italian Regions and four broad geographic areas (North, Central, South and Sicily, Sardinia; Figure 1). These areas were selected to take into account the latitude eect on MS mortality as described by Meco et al. [7]; Sardinia was consid- ered separately given its known very high MS mortality rate. The standardized mortality ratios (SMR) were computed using the pooled data set as standard, while the related 95% con®dence intervals (CIs) were assessed using the Poisson probability distribution function [8]. Age-adjusted rates for the selected areas European Journal of Epidemiology 17: 105±110, 2001. Ó 2001 Kluwer Academic Publishers. Printed in the Netherlands.