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 Page Print All RSS Search this Product|
Appendix 3: Technical details explaining the accuracy of self-reported height and weight data
Data presented in the analysis of self-reported and measured data are from:
A respondent aged 18 years or over.
Body Mass Index
Body Mass Index (BMI) is a simple index of weight-for-height that is commonly used to classify underweight, normal weight, overweight and obesity. It is calculated from height and weight information, using the formula weight (kg) divided by the square of height (m) and is expressed in kilograms per metres squared; kg/m2.
Table 1. Body Mass Index, Adults
Separate BMI classifications were produced for children. BMI scores were created in the same manner described above but also took into account the age and sex of the child. There are different cutoffs for BMI categories (underweight/normal combined, overweight or obese) for male and female children. These categories differ to the categories used in the adult BMI classification and follow the scale provided in Cole TJ, Bellizzi MC, Flegal KM and Dietz WH, Establishing a standard definition for child overweight and obesity worldwide: international survey, BMJ 2000; 320. See Appendix 4 in NHS 17-18 Users' Guide for more details.
A person aged 0-17 years.
Observations which appear inconsistent within the dataset.
Treatment of outliers
Routine processing of the NHS involved conservative editing of the most extreme outliers. However, such editing may not detect all of the inconsistent records (given the number of possible interrelated variables that could be examined). In the analysis of self-reported and measured height and weight data, further data ‘cleaning’ was applied to exclude any records with large disparities, which were likely to be a result of reporting or data entry errors. In the feature article analysis, if the difference between self-reported and measured height or weight (expressed as a percentage) was greater than 4 standard deviations from the group mean, then the record was excluded from analysis.
In 2017-18, 57.5% of respondents agreed to provide both measured and self-reported data and were included in analysis. A comparison between those included and excluded (i.e. those who provided self-reported and measured height and weight versus those who did not) showed no substantial differences between the two groups (see table 1 below).Therefore, for the purpose of the report, the sample analysed was representative of the whole 2017-18 NHS respondent sample (see table 1 below).
Table 2. Sample characteristics from NHS and report subsample, proportion of persons 18 years and over(a)
Source(s): NHS 2017-18, SIH 2017-18,
As standard practise, an imputation (editing) method was applied for respondents that opted not to be measured (33.8% in 2017-18) to derive height and weight data. However, since the feature article focused on measured and self-reported data only, respondents that had imputed height and weight values were excluded from the analysis. See Appendix 2 for more details about imputation methods.
The feature article analysed reporting trends for the 57.5% of survey respondents who agreed to be measured and who self-reported their height and weight. Without data from the remaining survey respondents it is difficult to determine whether those who volunteered and those who did not would differ when it comes to BMI, or reporting accuracy. To this end, there was a third group of respondents of interest who provided self-reported height and weight, but no measured data. When comparing self-reported BMI between the group analysed in the feature article, and the self-reported only group, the latter had a higher average self-reported BMI (see table 2 below). More research is needed to determine potential differences between those who opt to provide self-reported data only, those who also agree to be measured, and those who provide neither.
Table 3. Self-reported BMI by respondent compliance, average BMI
Source(s): NHS 2017-18
These documents will be presented in a new window.