A Windows application for computing standardized mortality ratios and standardized incidence ratios in cohort studies based on calculation of exact person-years at risk

https://doi.org/10.1016/j.cmpb.2013.05.028Get rights and content

Abstract

Standardized mortality ratios and standardized incidence ratios are widely used in cohort studies to compare mortality or incidence in a study population to that in the general population on a age-time-specific basis, but their computation is not included in standard statistical software packages. Here we present a user-friendly Microsoft Windows program for computing standardized mortality ratios and standardized incidence ratios based on calculation of exact person-years at risk stratified by sex, age and calendar time. The program offers flexible import of different file formats for input data and easy handling of general population reference rate tables, such as mortality or incidence tables exported from cancer registry databases. The application of the program is illustrated with two examples using empirical data from the Bavarian Cancer Registry.

Introduction

Rates and ratios are often used in demography and epidemiology to describe the occurrence of a health-related event. Standardization or adjustment of rates and ratios is employed to ensure comparability between heterogeneous populations with differences in their demographic structure regarding the main factors affecting health, most commonly sex, age, race, calendar period. The standardized mortality ratio (SMR) is a summary measure to assess the risk associated with a specific disease, particularly when the population under study is small or the event of interest is rare. SMRs are widely used in analyses of cohort studies, where the individuals of a defined study population are followed through time in order to compare the number of occurred events to the number of events that would be expected based on sex–age–time-specific reference rates for the general population. In typical applications, SMRs are used to compare mortality from certain causes of death in a study population with that in the general population, for example in occupational cohort studies [1]. Similar names are adopted for morbidity, such as standardized incidence ratio (SIR), where the events of interest are new disease cases. The SIR is used, for example, to determine whether the occurrence of cancer is higher or lower than expected in a certain geographical area compared to a reference population. Another application to cancer data is the analysis of the risk of subsequent malignancies following a first cancer diagnosis.

The calculation of SMRs/SIRs in cohort studies is based on the stratification by age and calendar period of the person-years of observation time for the study population. Although SMRs/SIRs are widely used, their computation is not readily available in standard statistical software packages. Publicly available software is scarce and most users have to develop their own macros or programs. Since the implementation is quite complex, the calculations are often simplified, for example approximate numbers of person-years are calculated or the calendar year is ignored and only one reference rate table for the general population is used for several calendar years.

Some software has been published, like SAS macros for computing exact person-years [2], [3], [4]. Using these macros requires not only the availability of the underlying software package but also sound knowledge about it since the input data and the algorithms have to be adjusted to fit each other. A tool for calculating SIRs for the analysis of subsequent cancers is provided by SEER*Stat [5]. However, SEER*Stat is specifically tailored to the analysis of cancer data of the SEER (Surveillance Epidemiology and End Results) Program. Analyzing own data with SEER*Stat requires the input data to be converted to a complex specific file format, which complicates the practical use of the software. PAMCOMP [6], a stand alone Microsoft Windows application for computing exact person-years and SMRs, can import different file formats. The calculation of person-years allows flexible stratification by sex and self-defined calendar periods and age groups. However, the rate file, which holds the mortality rates of the reference population for a specific cause of death, must have the same structure as the person-years table computed by the program for the study population, that is, the same number of age and calendar categories. In consequence, the user has to construct a suitable rate table for each calculation, which can be tedious.

This manuscript presents a user-friendly software application running under Microsoft Windows for computing SMRs/SIRs based on calculation of exact person-years at risk stratified by sex, age and calendar time. It can handle different file formats for input data for the study population and offers a convenient import of general population reference rate tables as, for example, reported from cancer registries.

Section snippets

Statistical methods

The standardized mortality ratio (SMR) was discussed by various authors [1], [7], [8]. The results apply to the standardized incidence ratio (SIR) as well. This section describes briefly the definition of the SMR, how it is calculated and how the confidence limits can be obtained.

The SMR is defined as the ratio of the observed number of deaths in the population being studied and the expected number of deaths in a comparable group of individuals from the general population, matched with respect

Computational realization

The software for the computation of standardized mortality ratios and standardized incidence ratios was realized as an extension of the program SURVSOFT, a stand-alone graphical software application running under Microsoft Windows operating systems (Windows XP, Vista, 7, 8). The program is easy to use and offers flexible data import and convenient handling of general population reference rate files. Online help provides context-sensitive user assistance. Since the program was presented in

Examples

To illustrate the use of the program, we present two applications of the program to empirical data from the Population Based Cancer Registry Bavaria [12], which covers a population of 12.5 million people. Patients diagnosed between 2002 and 2008 and aged 15 or older were included in the analyses, unless they were notified by death certificate only (dataset as of December 2011). The patients were followed up until the end of 2008.

The first example is an analysis of the risk of subsequent primary

Discussion

Utilization of statistical methods in practical applications requires the availability of appropriate software. Standard statistical software packages do not include the calculation of person-years at risk and SMRs/SIRs, and users often need to develop their own macros or programs, which complicates the use of these methods. In this article, we presented a user-friendly Microsoft Windows program for computing SMRs and SIRs with the intention to facilitate their use. The computation of SMRs/SIRs

Conflict of interest statement

The authors declare that there are no conflicts of interest for this work.

References (13)

  • K. Geiss et al.

    SURVSOFT – Software for nonparametric survival analysis

    Computer Methods and Programs in Biomedicine

    (2009)
  • N.E. Breslow et al.

    Statistical Methods in Cancer Research, vol. II: The Design and Analysis of Cohort Studies, IARC Scientific Publications No. 82

    (1987)
  • M. Macaluso

    Exact stratification of person-years

    Epidemiology

    (1992)
  • J. Sun et al.

    An efficient SAS program for exact stratification of person-years

    Computers in Biology and Medicine

    (1996)
  • J. Wood et al.

    A simple program to create exact person-time data in cohort analyses

    International Journal of Epidemiology

    (1997)
  • SEER*Stat statistical software, Surveillance Research Program, NCI, Bethesda, MD, USA....
There are more references available in the full text version of this article.

Cited by (0)

View full text