Trends in the prevalence of obesity and overweight in English adults Denise Howel
Objectives Using Health Survey for England obesity and overweight data to describe variation in time trends across age range changes in the rates of increase variation between different pseudo birth cohorts and whether that agrees with patterns in real birth cohort studies
Health Survey for England I All surveys covered adults aged 16+ & allowed calculation of prevalence of obesity (BMI30 kg/m 2 ) & overweight (BMI25 kg/m 2 ) (+ International cutoffs for those aged 16-17yrs)
Health Survey for England Health Survey for England II Households identified by multistage sampling & all adults invited for interview. Non-response weights were provided from 2003 onwards – these werent used since impossible to apply across all surveys. Chose restricted age range ( m & f) - over which obesity/overweight is increasing - avoids feature that those who survive longer are atypical
Age – period- cohort effects Age effect – obesity more common at different ages Period (survey year) effect – obesity affected by population-wide changes over time e.g. rise in fast food Cohort (birth year) effect – obesity varies between those born at different times e.g. change in infant feeding practices
However… Cant disentangle separate effects of age, period and cohort Birth year (cohort) + Age = Period ( Survey year) (Any model with all three factors has infinitely many possible estimates)
Main comparisons Since 16 years of surveys on an age range 16-65, relatively little information on any cohort effect Primary analysis – models with age and period Secondary analysis – age and cohort
Primary comparison Binomial regression -Can estimates differences between groups rather than odds ratios (logistic regression) Outcomes: Prevalence obesity/overweight M & F Explanatory variables: Period / / / /06 Age – polynomial terms (linear, quadratic & cubic) Interaction between age & period (linear& quadratic)
Changes in % obese over 4 yr periods
Trends in obesity Obesity has increased in each successive period and across the age range - similar prevalence in males and females - increase across periods higher for older men Rate of increase has slowed down (2003/6 – 1999/02) < (1999/02 – 1995/8)
Changes in % overweight over 4 yr periods
Overweight has increased in each successive period and across the age range - higher prevalence in males than females - increase across periods similar across ages Rate of increase has slowed down (2003/6 – 1999/02) < (1999/02 – 1995/8)
Comparison with other studies McPherson et al. (1) used HSE data on children and found some evidence that rate of growth is slowing down Zaninotto et al (2) used HSE data on adults and found that a model to predict obesity in 2012 which assumed rate was accelerating was slightly better fit.
Pseudo cohorts I Divide data into those born in different decades (only use those for which subjects were aged throughout ) HSE surveys are cross-sectional, so are comparing average experience of those born in different decades (not individual experience) Has been done for General Household Survey which has longer period of data collection (3)
Pseudo cohorts II Those born in had mean age of 50 in 1991 mean age of 51 in 1992 … Only possible to compare cohort over short portions of the age range t hose born in aged yrs in those born in aged yrs in
Real cohort studies 1946 National birth cohort MRC National survey of health and development n=5362 (4) 1958 British birth cohort National child development study n~17000 (5) Using published obesity/overweight prevalence data collected on 5 or 6 occasions between 15-53yrs
Comparison of trends in obesity/overweight with age in pseudo-cohorts and cohorts Plot prevalence of obesity/overweight against (mean) age for pseudo cohorts (smooth curves) and cohort study data ( at ages data collected)
Findings I Pseudo cohorts: prevalence is consistently higher in cohort born more recently (for a given mean age) e.g. male obesity at mean age of 35yrs 24% in cohort born BUT 13% in cohort born
Findings II Real cohort studies Little difference in early adulthood (and childhood), but by middle age, obesity & overweight higher in later cohort (6) Fair agreement between pseudo-cohort and cohort studies (except for overweight females)
Interpretation Other series of cross-sectional surveys in USA, Switzerland, Denmark & Finland have reported this pattern over different time periods Cant decide if genuine birth cohort effect Possible that childhood diet differed e.g. children born in 1946 had food rationing BUT little difference between 1946 &1958 cohorts in early life
Conclusions Prevalence of obesity and overweight still rising Some evidence that rate of increase is slowing Prevalence consistently higher for a given age in cohorts born more recently
References 1. McPherson K, Brown M et al. Obesity trends for children aged 2-11 and yrs: analysis from the Health Survey for England National Heart Forum Zaninotto P, Head J et al. Trends in obesity among adults n England from to 2004 by age and social class and projections of prevalence to J Epi Comm Health 2009: 63; Davy M. Time and generational trends in smoking among men and women in Great Britain /5 Health Statistics Quarterly 2006: 32; Wadsworth M, Kuh D et al. Cohort profile: the 1946 National Birth cohort (MRC National survey of health and development) Int. J Epidem. 2006: 35; Power C & Elliott J. Cohort profile : 1958 British birth cohort (National child development study. Int. J Epidem. 2006: 35; Li, L, Hardy R et al Child-to-adult body mass index and height trajectories : a comparison of 2 British birth cohorts. Am J Epid. 2008: 168; Howel D. Trends in the prevalence of obesity and overweight in English adults by age and birth cohort, Public Health Nutrition (available on First View)