Limits...
The use of a chronic disease and risk factor surveillance system to determine the age, period and cohort effects on the prevalence of obesity and diabetes in South Australian adults--2003-2013.

Taylor AW, Shi Z, Montgomerie A, Dal Grande E, Campostrini S - PLoS ONE (2015)

Bottom Line: Cohort years were 1905 to 1995.All variables were treated as continuous.By simultaneously considering the effects of age, period and cohort we have provided additional evidence for effective public health interventions.

View Article: PubMed Central - PubMed

Affiliation: Population Research & Outcome Studies, Discipline of Medicine, The University of Adelaide, South Australia, Australia; Ca' Foscari University, Venice, Italy.

ABSTRACT

Background: Age, period and cohort (APC) analyses, using representative, population-based descriptive data, provide additional understanding behind increased prevalence rates.

Methods: Data on obesity and diabetes from the South Australian (SA) monthly chronic disease and risk factor surveillance system from July 2002 to December 2013 (n = 59,025) were used. Age was the self-reported age of the respondent at the time of the interview. Period was the year of the interview and cohort was age subtracted from the survey year. Cohort years were 1905 to 1995. All variables were treated as continuous. The age-sex standardised prevalence for obesity and diabetes was calculated using the Australia 2011 census. The APC models were constructed with ''apcfit'' in Stata.

Results: The age-sex standardised prevalence of obesity and diabetes increased in 2002-2013 from 18.6% to 24.1% and from 6.2% to 7.9%. The peak age for obesity was approximately 70 years with a steady increasing rate from 20 to 70 years of age. The peak age for diabetes was approximately 80 years. There were strong cohort effects and no period effects for both obesity and diabetes. The magnitude of the cohort effect is much more pronounced for obesity than for diabetes.

Conclusion: The APC analyses showed a higher than expected peak age for both obesity and diabetes, strong cohort effects with an acceleration of risk after 1960s for obesity and after 1940s for diabetes, and no period effects. By simultaneously considering the effects of age, period and cohort we have provided additional evidence for effective public health interventions.

Show MeSH

Related in: MedlinePlus

Multivariable APC-analysis of obesity for males per 100 2002–2013, shown as line graphs representing rates (%) and rate ratios (log scale), with its 95% confidence interval (shaded area).The left line is the estimated age effect, the middle line refers to the estimated birth cohort effect and the short line to the right refers to the estimated period effect.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4414468&req=5

pone.0125233.g004: Multivariable APC-analysis of obesity for males per 100 2002–2013, shown as line graphs representing rates (%) and rate ratios (log scale), with its 95% confidence interval (shaded area).The left line is the estimated age effect, the middle line refers to the estimated birth cohort effect and the short line to the right refers to the estimated period effect.

Mentions: Fig 3 shows the fitted model for the total population with the independent effects of age (proportion of overall obesity), period and cohort (both as the rate ratios for overall obesity). The peak age for obesity is approximately 70 years with a steady increasing rate from 20 to 70 years of age. There are strong cohort effects and no period effects. Figs 4 and 5 show the same fitted models for males and females separately.


The use of a chronic disease and risk factor surveillance system to determine the age, period and cohort effects on the prevalence of obesity and diabetes in South Australian adults--2003-2013.

Taylor AW, Shi Z, Montgomerie A, Dal Grande E, Campostrini S - PLoS ONE (2015)

Multivariable APC-analysis of obesity for males per 100 2002–2013, shown as line graphs representing rates (%) and rate ratios (log scale), with its 95% confidence interval (shaded area).The left line is the estimated age effect, the middle line refers to the estimated birth cohort effect and the short line to the right refers to the estimated period effect.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4414468&req=5

pone.0125233.g004: Multivariable APC-analysis of obesity for males per 100 2002–2013, shown as line graphs representing rates (%) and rate ratios (log scale), with its 95% confidence interval (shaded area).The left line is the estimated age effect, the middle line refers to the estimated birth cohort effect and the short line to the right refers to the estimated period effect.
Mentions: Fig 3 shows the fitted model for the total population with the independent effects of age (proportion of overall obesity), period and cohort (both as the rate ratios for overall obesity). The peak age for obesity is approximately 70 years with a steady increasing rate from 20 to 70 years of age. There are strong cohort effects and no period effects. Figs 4 and 5 show the same fitted models for males and females separately.

Bottom Line: Cohort years were 1905 to 1995.All variables were treated as continuous.By simultaneously considering the effects of age, period and cohort we have provided additional evidence for effective public health interventions.

View Article: PubMed Central - PubMed

Affiliation: Population Research & Outcome Studies, Discipline of Medicine, The University of Adelaide, South Australia, Australia; Ca' Foscari University, Venice, Italy.

ABSTRACT

Background: Age, period and cohort (APC) analyses, using representative, population-based descriptive data, provide additional understanding behind increased prevalence rates.

Methods: Data on obesity and diabetes from the South Australian (SA) monthly chronic disease and risk factor surveillance system from July 2002 to December 2013 (n = 59,025) were used. Age was the self-reported age of the respondent at the time of the interview. Period was the year of the interview and cohort was age subtracted from the survey year. Cohort years were 1905 to 1995. All variables were treated as continuous. The age-sex standardised prevalence for obesity and diabetes was calculated using the Australia 2011 census. The APC models were constructed with ''apcfit'' in Stata.

Results: The age-sex standardised prevalence of obesity and diabetes increased in 2002-2013 from 18.6% to 24.1% and from 6.2% to 7.9%. The peak age for obesity was approximately 70 years with a steady increasing rate from 20 to 70 years of age. The peak age for diabetes was approximately 80 years. There were strong cohort effects and no period effects for both obesity and diabetes. The magnitude of the cohort effect is much more pronounced for obesity than for diabetes.

Conclusion: The APC analyses showed a higher than expected peak age for both obesity and diabetes, strong cohort effects with an acceleration of risk after 1960s for obesity and after 1940s for diabetes, and no period effects. By simultaneously considering the effects of age, period and cohort we have provided additional evidence for effective public health interventions.

Show MeSH
Related in: MedlinePlus