4364.0.55.001 - National Health Survey: First Results, 2017-18  
Latest ISSUE Released at 11:30 AM (CANBERRA TIME) 30/09/2019   
   Page tools: Print Print Page Print all pages in this productPrint All RSS Feed RSS Bookmark and Share Search this Product

Research Paper: Explaining the accuracy of self-reported height and weight data


Introduction

Overweight and obesity, particularly when linked with other lifestyle factors such as poor diet, sedentary lifestyle and smoking is a risk factor for many serious health conditions such as [1]:
    • heart disease
    • high blood pressure
    • diabetes
    • musculoskeletal problems.

Overweight and obesity have risen over recent decades, so collecting accurate data has become increasingly important for effective public health monitoring. At the same time, the number of people who decided to provide their physical dimensions has declined.


Measuring overweight and obesity

Measuring overweight and obesity at the population level is most often done by collecting height and weight measurements to calculate Body Mass Index (BMI), a standard unit that allows comparisons between people of different sizes on the same scale.
Height and weight data used to assess overweight and obesity can be gathered from survey respondents by:
    • direct physical measurement (the method used in the NHS to derive BMI)
    • asking respondents to self-report their height and weight.

Three ABS National Health Surveys (1995, 2007-08 and 2017-18) collected height and weight using both methods. While self-reported height and weight is logistically simpler to collect, this method is less accurate because of the tendency for people to over-report their height and under-report their weight [2-4].

This article compares self-reported and measured data (from the surveys where both were collected) to quantify this effect and identify factors associated with accuracy.


Persons aged 18 years & over - Proportion of overweight and obese persons, by collection method and survey year(a)
Bar Graph: Overweight and obesity increase over time, self-reported BMI shows lower prevalence than measured.
Footote(s): (a) Based on adults providing both self-reported and measured height and weight.
Source(s): NHS 2017-18, NHS 2007-2018, NHS 1995


Calculation methods
  • Self-reported and measured data were used to calculate BMI, and estimate the prevalence of overweight and obesity in the population. Survey weights were applied to data to produce prevalence rates that were comparable to NHS published data.
  • Reporting error for height, weight and BMI was calculated by subtracting measured data from self-reported data. This is expressed in the measurement unit (cm, kg or kg/cm2) and a percentage. These data are unweighted.
  • Average height, weight, BMI as well as prevalence and reporting error from self-reported and measured data were compared.
  • Outlier data points were identified as points where the difference between self-reported and measured (expressed as a percentage) was greater than 4 standard deviations from the group mean, and were removed from subsequent analysis. See Appendix 3 for more details.

    Prevalence of overweight and obesity according to self-reported data

    Classifying overweight and obesity using self-reported data generally underestimates overweight and obesity. In 2017-18:
      • Four in five (80.0%) adults’ self-reported data classified them into the correct BMI category.
      • Self-reported data suggests that only 57.6% of the population were overweight or obese, compared to 65.8% when using measured data.
      • Specifically, only 65.8% of males and 48.9% of females were classified as overweight or obese based on their self-reported dimensions, but 74.0% of males and 57.2% of females were actually overweight or obese based on physical measurements.

    Relying on self-reported data also fails to accurately show the increasing rates of overweight and obesity observed over time.
      • Between 2007-08 and 2017-18, the proportion of overweight or obese persons rose by 4.9 percentage points according to physical measurements, but only 2.4 percentage points according to the self-reported data.


    Persons aged 18 years & over - Proportion of overweight and obese persons, by collection method, age group & sex(a)
    Line graph: Overweight and obesity increase with age and levels off at 45-54 years. Females show lower prevalence than males.
    Footnote(s): (a) Based on adults providing both self-reported and measured height and weight.
    Source(s): NHS 2017-18


    Sources of error in self-reported BMI calculations

    The accuracy of self-reported height and weight each influence the estimation of overweight and obesity. On average the population has a propensity to over-report their height and under-report their weight, but some additional patterns are seen within population groups as outlined below.


    Self-reported height

    In general, self-reported height is overestimated compared to measured height.
    • In 2017-18, the majority (62.4%) of people self-reported their height to be within 3.0 cm of their measured height.
    • Males over-reported by a median of 2.2 cm (1.3%) while females were slightly more accurate with a median over-report of 1.6 cm (1.0%).
    • Older people self-reported their height less accurately than younger people with those aged 75 years or over having a mean over-report of 2.6%, compared to 1.0% for those aged 18-24 years.
    • While the overall accuracy of self-reported height data improved between 1995 and 2007-08, self-reported height data in 2017-18 was less accurate than previous surveys for people aged less than 55-64 years.


    Persons aged 18 years & over - Accuracy of self-reported height by age group, as percentage error(a)
    Line graph: Height reporting error increases with age. 2007-18 data had the least error, followed by 1995 then 2017-18 data.
    Footnote(s): (a) Based on adults providing both self-reported and measured height and weight.
    Source(s): NHS 2017-18


    Self-reported weight

    In general, self-reported weight is underestimated compared to measured weight.
    • The majority (78.8%) of people self-reported their weight to be within 3.0 kg of their measured weight.
    • The median under-report for weight was 1.0 kg (1.5%) for females, 0.8 kg (0.9%) for males.
    • Persons who were measured as being obese or overweight underestimated their weight by a median of 1.7% and 1.3% respectively, while those measured as being normal weight under-reported by 0.6%.
    • Unlike height self-reporting, there was no association between reporting accuracy and age.
    • Weight reporting accuracy improved from 1995 to 2007-08, with a further smaller improvement in 2017-18, as shown in the graph below.


    Persons aged 18 years & over - Accuracy of self-reported weight by age group, as percentage error(a)
    Line Graph: Weight reporting error is similar across ages. 2017-18 data had the least error, followed by 2007-08 then 1995.
    Footnote(s): (a) Based on adults providing both self-reported and measured height and weight.
    Source(s): NHS 2017-18


    Cumulative effect of self-reported data on BMI

    The combined effect of over-reporting height and under-reporting weight is amplified when BMI is calculated.
    • Error in height reporting has a greater effect on BMI than the same relative error for weight. For example, under-reporting weight by 2.0% results in 2.0% underestimation of BMI, but over-reporting height by 2.0% results in 3.9% underestimation of BMI.
    • BMI calculated from self-reported data results in an underestimation of 1.1 kg/cm2 (3.8%) for Australian adults. This is consistent between males and females, who underestimate by 1.1 kg/cm2 (3.7%) and 1.1 kg/cm2 (3.8%) respectively.
    • Accuracy of self-reported BMI decreases with age, from 0.8 kg/cm2 (2.9%) in 18-24 year olds to 1.8 kg/cm2 (6.0%) in those over 75 years old. This pattern mirrors the misreporting seen in self-reported height data.
    • Differences in accuracy were also evident within BMI categories, where the underestimation for those categorised as normal was 0.5 kg/cm2 (2.3%) and 1.8 kg/cm2 (5.1%) for those categorised as obese.


    Persons aged 18 years & over - Accuracy of self-reported height, weight, and BMI by category, as percentage error(a)(b)
    Bar graph: Height and weight reporting error produce larger combine error in BMI. This is more obvious in overweight/obese populations.
    Footnote(s): (a) Based on adults providing both self-reported and measured height and weight. (b) Using National Health Survey 2017-18 unweighted sample counts (c) Includes normal and underweight ranges
    Source(s): NHS 2017-18


    Changes over time
      • Self-reporting accuracy improved between 1995 and 2007-08 for both height and weight data. Self-reported weight data in 2017-18 is of similar accuracy to 2007-08, while self-reported height data is less accurate than 2007-08.
      • The accuracy of self-reported data is inconsistent over progressive survey collections; accuracy of self-reported BMI categorisation improved between 1995 and 2007-08, from 77.6% to 83.2% but fell to 80.0% in 2017-18.
      • According to physical measurements the proportion of overweight or obese males rose by 5.9 percentage points between 2007-08 and 2017-18 from 68.1% to 74.0%, but only by 2.8 percentage points according to the self-reported data, from 63.0% to 65.8%.
      • Similarly, physical measurements show that the proportion of overweight and obese females rose by 4.1 percentage points between 2007-08 and 2017-18 from 53.1% to 57.2%, but only by 2.1 percentage points from 46.8% to 48.9% according to self-reported data.


    Persons aged 18 years & over - Accuracy of self-reported height, weight, and BMI by survey year, as percentage error(a)(b)
    Bar graph: Weight reporting-error decreases over survey years. Height- and BMI-reporting error are largest in 1995 and smallest in 2007-08.
    Footnote(s): (a) Based on adults providing both self-reported and measured height and weight. (b) Using unweighted sample counts
    Source(s): NHS 2017-18, NHS 2007-08, NHS 1995


    Children

    Children’s BMI is calculated the same way as adults, but is categorised according to different BMI ranges. For children who participated in the survey, their parent or guardian provided the interviewer with a reported height and weight if the child was under 15 years old.
      • 29.5% of children (aged 2-17 years) had self-reported measured height and weight data collected during NHS 2017-18.
      • On average self-reported height was within 0.4% for measured height (< 1 cm) and 2.3% for measured weight (1.1 kg).
      • Both height and weight tended to be under-reported. Based on self-report data, 79.6% of children were correctly classified as normal, overweight, or obese.


    Summary
      • Self-reported physical dimensions show that overweight and obesity are increasing over time, but underestimate its true prevalence (by 8.2 percentage points).
      • Reporting errors in height have a greater effect on the accuracy of BMI calculations compared to weight.
      • Overall self-reporting accuracy improved between 1995 and 2007-08, however reduced accuracy in height reporting caused self-reported BMI accuracy to decline in 2017-18.
      • Adults have a general tendency to over-report their height and under-report their weight.
      • Older adults report their height less accurately than younger adults.
      • Persons classified as overweight or obese report their weight less accurately.


    References
        1. Australian Institute of Health and Welfare. (2018). Australia's health 2018. Canberra: AIHW.
        2. Gorber, S. C., Tremblay, M. , Moher, D. and Gorber, B. (2007), A comparison of direct vs. self‐report measures for assessing height, weight and body mass index: a systematic review. Obesity Reviews, 8: 307-326.
        3. Stommel and M. Schoenborn, C. A. (2009), Accuracy and usefulness of BMI measures based on self-reported weight and height: findings from the NHANES & NHIS 2001-2006. BMC Public Health, 9: 421.
        4. Taylor, A. W., Grande, E. D., Gill, T. K., Chittleborough, C. R., Wilson, D. H., Adams, R. J., Grant, J. F., Phillips, P. , Appleton, S. and Ruffin, R. E. (2006), How valid are self‐reported height and weight? A comparison between CATI self‐report and clinic measurements using a large cohort study. Australian and New Zealand Journal of Public Health, 30: 238-246.