Automated mortality monitoring in Scotland from 2009

Adam P. Wagner, Edward McKenzie, Chris Robertson, J. McMenamin, A. Reynolds, Heather Murdoch

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Mortality monitoring systems are important for gauging the effect of influenza and other wide ranging health threats. We present the daily all-cause mortality monitoring system routinely used in Scotland, which differs from others by using two different statistical models for calculating expected mortality. The first model is an extended Serfling model, which captures annual seasonality in mortality using sine and cosine terms, and is frequently seen in other systems. Serfling models fit to summer seasonality well, but not to the winter peak. Thus, during the winter, there are frequent `excesses’, higher than expected mortality, making it harder to directly judge if winter mortality is higher than in previous years. The second model, a Generalised Additive Model, resolves this by allowing a more flexible seasonal pattern that includes the winter peak. Thus, excesses under the second model directly indicate if winter mortality is higher than in previous years, useful, for example, in judging if a new strain of seasonal influenza is more likely to produce death than previous ones. As common in all-cause mortality monitoring systems, the Scottish system uses a reporting delay correction: we discuss the difficulties of interpretation when such a correction is used and possible avenues for future work that may address these difficulties.

LanguageEnglish
Article number4
JournalEurosurveillance
Volume18
Issue number15
Publication statusPublished - 11 Apr 2013

Fingerprint

Scotland
Mortality
Human Influenza
Statistical Models
Health

Keywords

  • mortality monitoring
  • Scotland
  • automated monitoring

Cite this

Wagner, A. P., McKenzie, E., Robertson, C., McMenamin, J., Reynolds, A., & Murdoch, H. (2013). Automated mortality monitoring in Scotland from 2009. Eurosurveillance, 18(15), [4].
Wagner, Adam P. ; McKenzie, Edward ; Robertson, Chris ; McMenamin, J. ; Reynolds, A. ; Murdoch, Heather. / Automated mortality monitoring in Scotland from 2009. In: Eurosurveillance. 2013 ; Vol. 18, No. 15.
@article{7291996b272141ae8b70da365df96af6,
title = "Automated mortality monitoring in Scotland from 2009",
abstract = "Mortality monitoring systems are important for gauging the effect of influenza and other wide ranging health threats. We present the daily all-cause mortality monitoring system routinely used in Scotland, which differs from others by using two different statistical models for calculating expected mortality. The first model is an extended Serfling model, which captures annual seasonality in mortality using sine and cosine terms, and is frequently seen in other systems. Serfling models fit to summer seasonality well, but not to the winter peak. Thus, during the winter, there are frequent `excesses’, higher than expected mortality, making it harder to directly judge if winter mortality is higher than in previous years. The second model, a Generalised Additive Model, resolves this by allowing a more flexible seasonal pattern that includes the winter peak. Thus, excesses under the second model directly indicate if winter mortality is higher than in previous years, useful, for example, in judging if a new strain of seasonal influenza is more likely to produce death than previous ones. As common in all-cause mortality monitoring systems, the Scottish system uses a reporting delay correction: we discuss the difficulties of interpretation when such a correction is used and possible avenues for future work that may address these difficulties.",
keywords = "mortality monitoring, Scotland, automated monitoring",
author = "Wagner, {Adam P.} and Edward McKenzie and Chris Robertson and J. McMenamin and A. Reynolds and Heather Murdoch",
year = "2013",
month = "4",
day = "11",
language = "English",
volume = "18",
journal = "Eurosurveillance",
issn = "1560-7917",
number = "15",

}

Wagner, AP, McKenzie, E, Robertson, C, McMenamin, J, Reynolds, A & Murdoch, H 2013, 'Automated mortality monitoring in Scotland from 2009' Eurosurveillance, vol. 18, no. 15, 4.

Automated mortality monitoring in Scotland from 2009. / Wagner, Adam P.; McKenzie, Edward; Robertson, Chris; McMenamin, J.; Reynolds, A.; Murdoch, Heather.

In: Eurosurveillance, Vol. 18, No. 15, 4, 11.04.2013.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Automated mortality monitoring in Scotland from 2009

AU - Wagner, Adam P.

AU - McKenzie, Edward

AU - Robertson, Chris

AU - McMenamin, J.

AU - Reynolds, A.

AU - Murdoch, Heather

PY - 2013/4/11

Y1 - 2013/4/11

N2 - Mortality monitoring systems are important for gauging the effect of influenza and other wide ranging health threats. We present the daily all-cause mortality monitoring system routinely used in Scotland, which differs from others by using two different statistical models for calculating expected mortality. The first model is an extended Serfling model, which captures annual seasonality in mortality using sine and cosine terms, and is frequently seen in other systems. Serfling models fit to summer seasonality well, but not to the winter peak. Thus, during the winter, there are frequent `excesses’, higher than expected mortality, making it harder to directly judge if winter mortality is higher than in previous years. The second model, a Generalised Additive Model, resolves this by allowing a more flexible seasonal pattern that includes the winter peak. Thus, excesses under the second model directly indicate if winter mortality is higher than in previous years, useful, for example, in judging if a new strain of seasonal influenza is more likely to produce death than previous ones. As common in all-cause mortality monitoring systems, the Scottish system uses a reporting delay correction: we discuss the difficulties of interpretation when such a correction is used and possible avenues for future work that may address these difficulties.

AB - Mortality monitoring systems are important for gauging the effect of influenza and other wide ranging health threats. We present the daily all-cause mortality monitoring system routinely used in Scotland, which differs from others by using two different statistical models for calculating expected mortality. The first model is an extended Serfling model, which captures annual seasonality in mortality using sine and cosine terms, and is frequently seen in other systems. Serfling models fit to summer seasonality well, but not to the winter peak. Thus, during the winter, there are frequent `excesses’, higher than expected mortality, making it harder to directly judge if winter mortality is higher than in previous years. The second model, a Generalised Additive Model, resolves this by allowing a more flexible seasonal pattern that includes the winter peak. Thus, excesses under the second model directly indicate if winter mortality is higher than in previous years, useful, for example, in judging if a new strain of seasonal influenza is more likely to produce death than previous ones. As common in all-cause mortality monitoring systems, the Scottish system uses a reporting delay correction: we discuss the difficulties of interpretation when such a correction is used and possible avenues for future work that may address these difficulties.

KW - mortality monitoring

KW - Scotland

KW - automated monitoring

UR - http://www.eurosurveillance.org/ViewArticle.aspx?ArticleId=20451

M3 - Article

VL - 18

JO - Eurosurveillance

T2 - Eurosurveillance

JF - Eurosurveillance

SN - 1560-7917

IS - 15

M1 - 4

ER -

Wagner AP, McKenzie E, Robertson C, McMenamin J, Reynolds A, Murdoch H. Automated mortality monitoring in Scotland from 2009. Eurosurveillance. 2013 Apr 11;18(15). 4.