Skip to main content
Dryad

A ten-year (2009–2018) database of cancer mortality rates in Italy

Cite this dataset

Di Paola, Arianna et al. (2022). A ten-year (2009–2018) database of cancer mortality rates in Italy [Dataset]. Dryad. https://doi.org/10.5061/dryad.ns1rn8pvg

Abstract

AbstractIn Italy, approximately 400.000 new cases of malignant tumors are recorded every year. The average of annual deaths caused by tumors, according to the Italian Cancer Registers, is about 3.5 deaths and about 2.5 per 1,000 men and women respectively, for a total of about 3 deaths every 1,000 people. Long-term (at least a decade) and spatially detailed data (up to the municipality scale) are neither easily accessible nor fully available for public consultation by the citizens, scientists, research groups, and associations. Therefore, here we present a ten-year (2009–2018) database on cancer mortality rates (in the form of Standardized Mortality Ratios, SMR) for 23 cancer macro-types in Italy on municipal, provincial, and regional scales. We aim to make easily accessible a comprehensive, ready-to-use, and openly accessible source of data on the most updated status of cancer mortality in Italy for local and national stakeholders, researchers, and policymakers and to provide researchers with ready-to-use data to perform specific studies.

Methods

For a given locality, year, and cause of death, the SMR is the ratio between the observed number of deaths (Om) and the number of expected deaths (Em):

SMR = Om/Em (1) 

where Om should be an available observational data and Em is estimated as the weighted sum of age-specific population size for the given locality (ni) per age-specific death rates of the reference population (MRi): 

Em = sum(MRi x ni) (2)

MRi could be provided by a public health organization or be estimated as the ratio between the age-specific number of deaths of reference population (Mi) to the age-specific reference population size (Ni):

MRi = Mi/Ni (3)

Thus, the value of Em is weighted by the age distribution of deaths and population size.

SMR assumes value 1 when the number of observed and expected deaths are equal. Following eqns. (1-3), the SMR was computed for single years of the period 2009-2018 and for single cause of death as defined by the International ICD-10 classification system by using the following data: age-specific number of deaths by cause of reference population (i.e., Mi) from the Italian National Institute of Statistics (ISTAT,  (http://www.istat.it/en/, last access: 26/01/2022)); age-specific census data on reference population (i.e., Ni) from ISTAT; the observed number of deaths by cause (i.e., Om) from ISTAT; the age-specific census data on population (ni); the SMR was estimated at three different level of aggregation: municipal, provincial (equivalent to the European classification NUTS 3) and regional (i.e., NUTS2). The SMR was also computed for the broad category of malignant tumors (i.e. C00-C979, hereinafter cancer macro-type C), and for the broad category of malignant tumor plus non-malignant tumors  (i.e. C00-C979 plus D0-D489, hereinafter cancer macro-type CD). Lower 90% and 95% confidence intervals of 10-year average values were computed according to the Byar method.

Usage notes

The interannual variability of SMR for a given administrative unit might be large under small populations. Indeed, being the SMR a rate standardized over the population size, the expected mortality (i.e., Em) in small populations will result low (say 10-2) and in turn, according to eq. (1), even a few deaths (say 1 or 2) in a year could yield a relatively high SMR as shown in Figure 3.   For this reason, we recommend avoiding using single-year estimates and using the average SMR and/or lower 90% or 95% confidence intervals.

Funding