IL TASSO DI MORTALITA E LE SUE PROPRIETA STATISTICHE

Translated title of the contribution: The mortality rate and its statistical properties

C. Zocchetti, D. Consonni

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

The rate is an epidemiologic measure which has a widespread use in describing the occurrence of diseases. In this paper, with a didactical approach, the definition of the mortality (morbidity) rate is introduced following two ways of reasoning: firstly, in the context of survival analysis, as an instantaneous conditional probability of failure (either disease or death) (instantaneous risk) and, secondly, as a traditional measure of rapidity of change in time. We then proceed to highlight the differences, in terms of definition, interpretation, and application, between the concepts of rate and risk. As a next step the statistical properties of the rate are explored and it is explained why the variability of the measure is simply associated with the numerator (events) and not with the denominator (person-times) of the rate. In this context the Poisson distribution is commonly considered the probability distribution which better describes the statistical variability of the observed events, and examples of such a distribution are presented. When the number of deaths is sufficiently elevated the Poisson distribution can be adequately approximated by the Gauss distribution, which is simpler and in common use in occupational medicine, and formulas are presented to compute mean and variance of the rate in this situation. When the number of deaths is small a suggestion is made of making a log transformation of the rate (or of the deaths) before using the Gauss distribution: formulas are proposed for this situation, too. As a practical application of the statistical properties presented and as a concluding example, a confidence interval for the rate is computed. Numerical and graphical comparisons of the results deriving from the use of different formulas are described.

Original languageItalian
Pages (from-to)327-343
Number of pages17
JournalMedicina del Lavoro
Volume85
Issue number4
Publication statusPublished - 1994

Fingerprint

Poisson Distribution
Mortality
Occupational Medicine
Survival Analysis
Confidence Intervals
Morbidity

ASJC Scopus subject areas

  • Public Health, Environmental and Occupational Health

Cite this

IL TASSO DI MORTALITA E LE SUE PROPRIETA STATISTICHE. / Zocchetti, C.; Consonni, D.

In: Medicina del Lavoro, Vol. 85, No. 4, 1994, p. 327-343.

Research output: Contribution to journalArticle

Zocchetti, C. ; Consonni, D. / IL TASSO DI MORTALITA E LE SUE PROPRIETA STATISTICHE. In: Medicina del Lavoro. 1994 ; Vol. 85, No. 4. pp. 327-343.
@article{351a22eb4064475e9793a14c21d37ad7,
title = "IL TASSO DI MORTALITA E LE SUE PROPRIETA STATISTICHE",
abstract = "The rate is an epidemiologic measure which has a widespread use in describing the occurrence of diseases. In this paper, with a didactical approach, the definition of the mortality (morbidity) rate is introduced following two ways of reasoning: firstly, in the context of survival analysis, as an instantaneous conditional probability of failure (either disease or death) (instantaneous risk) and, secondly, as a traditional measure of rapidity of change in time. We then proceed to highlight the differences, in terms of definition, interpretation, and application, between the concepts of rate and risk. As a next step the statistical properties of the rate are explored and it is explained why the variability of the measure is simply associated with the numerator (events) and not with the denominator (person-times) of the rate. In this context the Poisson distribution is commonly considered the probability distribution which better describes the statistical variability of the observed events, and examples of such a distribution are presented. When the number of deaths is sufficiently elevated the Poisson distribution can be adequately approximated by the Gauss distribution, which is simpler and in common use in occupational medicine, and formulas are presented to compute mean and variance of the rate in this situation. When the number of deaths is small a suggestion is made of making a log transformation of the rate (or of the deaths) before using the Gauss distribution: formulas are proposed for this situation, too. As a practical application of the statistical properties presented and as a concluding example, a confidence interval for the rate is computed. Numerical and graphical comparisons of the results deriving from the use of different formulas are described.",
keywords = "confidence interval, mortality rate, Poisson, probability",
author = "C. Zocchetti and D. Consonni",
year = "1994",
language = "Italian",
volume = "85",
pages = "327--343",
journal = "Medicina del Lavoro",
issn = "0025-7818",
publisher = "Mattioli 1885 S.p.A.",
number = "4",

}

TY - JOUR

T1 - IL TASSO DI MORTALITA E LE SUE PROPRIETA STATISTICHE

AU - Zocchetti, C.

AU - Consonni, D.

PY - 1994

Y1 - 1994

N2 - The rate is an epidemiologic measure which has a widespread use in describing the occurrence of diseases. In this paper, with a didactical approach, the definition of the mortality (morbidity) rate is introduced following two ways of reasoning: firstly, in the context of survival analysis, as an instantaneous conditional probability of failure (either disease or death) (instantaneous risk) and, secondly, as a traditional measure of rapidity of change in time. We then proceed to highlight the differences, in terms of definition, interpretation, and application, between the concepts of rate and risk. As a next step the statistical properties of the rate are explored and it is explained why the variability of the measure is simply associated with the numerator (events) and not with the denominator (person-times) of the rate. In this context the Poisson distribution is commonly considered the probability distribution which better describes the statistical variability of the observed events, and examples of such a distribution are presented. When the number of deaths is sufficiently elevated the Poisson distribution can be adequately approximated by the Gauss distribution, which is simpler and in common use in occupational medicine, and formulas are presented to compute mean and variance of the rate in this situation. When the number of deaths is small a suggestion is made of making a log transformation of the rate (or of the deaths) before using the Gauss distribution: formulas are proposed for this situation, too. As a practical application of the statistical properties presented and as a concluding example, a confidence interval for the rate is computed. Numerical and graphical comparisons of the results deriving from the use of different formulas are described.

AB - The rate is an epidemiologic measure which has a widespread use in describing the occurrence of diseases. In this paper, with a didactical approach, the definition of the mortality (morbidity) rate is introduced following two ways of reasoning: firstly, in the context of survival analysis, as an instantaneous conditional probability of failure (either disease or death) (instantaneous risk) and, secondly, as a traditional measure of rapidity of change in time. We then proceed to highlight the differences, in terms of definition, interpretation, and application, between the concepts of rate and risk. As a next step the statistical properties of the rate are explored and it is explained why the variability of the measure is simply associated with the numerator (events) and not with the denominator (person-times) of the rate. In this context the Poisson distribution is commonly considered the probability distribution which better describes the statistical variability of the observed events, and examples of such a distribution are presented. When the number of deaths is sufficiently elevated the Poisson distribution can be adequately approximated by the Gauss distribution, which is simpler and in common use in occupational medicine, and formulas are presented to compute mean and variance of the rate in this situation. When the number of deaths is small a suggestion is made of making a log transformation of the rate (or of the deaths) before using the Gauss distribution: formulas are proposed for this situation, too. As a practical application of the statistical properties presented and as a concluding example, a confidence interval for the rate is computed. Numerical and graphical comparisons of the results deriving from the use of different formulas are described.

KW - confidence interval

KW - mortality rate

KW - Poisson

KW - probability

UR - http://www.scopus.com/inward/record.url?scp=0027944359&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0027944359&partnerID=8YFLogxK

M3 - Articolo

C2 - 7808349

AN - SCOPUS:0027944359

VL - 85

SP - 327

EP - 343

JO - Medicina del Lavoro

JF - Medicina del Lavoro

SN - 0025-7818

IS - 4

ER -