Differences between self-reported and interviewer-measured height, weight and BMI using HSE 2011-2016 data

This section of the report describes patterns in the differences between self-reported and interviewer-measured height and weight (and BMI and BMI status derived from these), using six years of HSE data from 2011 to 2016. Differences between self-reported and interviewer-measured values are presented by survey year, age and sex.

This analysis used HSE data collected in each survey between 2011 and 2016 (years in which participants were asked in the interview to report their own height and weight as well as these being measured by the interviewer later in that visit).

Early in the interview, participants were asked to report how tall they were without shoes, and how much they weighed without clothes or shoes. Participants could give measurements in either imperial or metric units, and these were converted to metric units if necessary (metres for height, kilograms for weight). If the participant did not know their height and/or weight or they were unwilling to provide the information, the interviewer recorded this. Participants were not told at the time of interview that their height and weight would be measured; however, given their informed consent, it is likely that they might have anticipated being measured subsequently. Participants may also have observed that the interviewer carried measuring equipment: stadiometer and scales.

Height was measured using a portable stadiometer with a sliding head plate, a base plate and connecting rods marked with a measuring scale. Participants were asked to remove their shoes. One measurement was taken, with the participant stretching to the maximum height and the head positioned in the Frankfort plane. The reading was recorded to the nearest millimetre. Participants who were pregnant, unable to stand or were unsteady on their feet were not measured.

For the weight measurement, participants were asked to remove their shoes and any bulky clothing. A single measurement was recorded to the nearest 100g. Participants who were pregnant, unable to stand, or unsteady on their feet were not weighed. Participants who weighed more than 200kg were asked for their estimated weight because the scales are inaccurate above this level. No adjustment was made for the weight of clothing.

In order to define overweight or obesity, a measurement is required that allows for differences in weight due to height. A widely accepted measure of weight for height, BMI, defined as weight in kilograms divided by the square of the height in metres (kg/m²), is used for this purpose in the HSE series.

Participants were classified into the following five mutually exclusive BMI groups according to the World Health Organization’s BMI classification (World Health Organization, 2010).

BMI categories of overweight and obese have frequently been combined to show the proportion of participants who are either overweight or obese (BMI 25kg/m² or more). In this report, a BMI threshold of 30kg/m² has been used to define participants as obese. These definitions were applied to all participants aged 16 and over, as this is how adults are defined in the HSE series.

In the analysis of height and weight, data were excluded from those participants who were considered by the interviewer to have unreliable measurements, for example those who were stooped or wore excessive clothing. As this section of the report focuses on the differences between self-reported and interviewer-measured values, the analyses were restricted to the 38,940 participants in the HSE 2011-2016 with valid values for both self-reported and interviewer-measured height and weight.¹

All analyses were sex-specific, because previous evidence suggests that the factors associated with the difference between self-reported and measured height and weight differ between men and women (Flegal et al., 2019).

First, mean values of self-reported and interviewer-measured height and weight (and BMI derived from these) were calculated, as well as the proportion of participants in each BMI category. Second, to compare self-reported and interviewer-measured values, the differences (self-reported minus measured) were calculated between means (height, weight, BMI) and between prevalence levels of overweight or obese, and obese. Positive values for the difference indicate overestimation (for example over-reporting of height); negative values indicate underestimation (for example, under-reporting of weight).

These two sets of analyses were calculated for each survey year to examine change over time in misreporting. Statistical tests were performed to assess linear trends in misreporting (that is, whether the differences between self-reported and measured values increased or decreased over the time period at a steady rate). As our results showed no linear trend in misreporting in either sex, the analyses were repeated on HSE data combined from survey years 2011 to 2016. 

Third, BMI status from self-reported and interviewer-measured height and weight were cross-tabulated. Using BMI status from interviewer-measured height and weight as the gold standard, estimates of sensitivity (the percent of true positives) and specificity (the percent of true negatives) were calculated to quantify the classification accuracy of BMI status derived from self-reported height and weight. For example, a sensitivity of 60% for the obese category indicates that the classification of BMI derived from self-reported height and weight correctly identified 60% of participants who were obese according to BMI derived from measured height and weight (these are true positives: participants classified as obese based on both self-reported and measured height and weight). A specificity of 80% for the obese category indicates that the classification correctly identified 80% of participants who were not obese according to measured BMI (these are true negatives: participants classified as not obese according to both self-reported and measured height and weight). 

Significance tests have been used in this report to determine whether differences between self-reported and interviewer-measured height and weight (and BMI derived from these) are genuine differences (statistically significant) or the result of random natural variation.

The significance testing methodology presented in this report uses linear regression modelling to test the relationship between the outcome variable and explanatory variables. For example, change over time in misreporting (presented in Tables 1 and 2) is examined using models where the outcome variable is the difference between self-reported and measured data and the explanatory variables are sex and survey year. To assess whether misreporting increased or decreased at a steady rate over the time period (a linear trend), survey year is included as a continuous explanatory variable.

A first test of significance is for the main effects only using a Wald test. The Wald test is a statistical test used to calculate the significance of parameters in a statistical model. For example, the test might examine whether there is a statistically significant relationship between misreporting and sex (after controlling for survey year) and between misreporting and survey year (after controlling for sex).

The test seeks to establish whether the variation in the outcome between groups that is observed could have happened by chance or whether it is likely to reflect some 'real' differences in the population.

P-values for comparisons are shown in parentheses. A p-value is the probability of the observed result occurring due to chance alone. A p-value of less than 5% is conventionally taken to indicate a statistically significant result (p<0.05). It should be noted that the p-value is dependent on the sample size, so that with large samples differences or associations which are very small may still be statistically significant.

Using this method of statistical testing, differences that are significant at the 5% level indicate that there is sufficient evidence in the data to suggest that the differences in the sample reflect a true difference in the population.

A second test of significance looks at the interaction between sex and the other explanatory variable under consideration. If the interaction is statistically significant (p<0.05) this indicates that there is likely to be an underlying difference in the pattern of results for men and women, and this will normally be commented on in the report text.

This analysis is based only on those participants who had both measured and self-reported height, and both measured and self-reported weight. Therefore the estimates of the differences between self-reported and interviewer-measured height and weight differ slightly from previously published tables (Sutton, 2012) which were based on those with valid height measurement or valid weight measurement.

Self-reported mean height was higher than interviewer-measured mean height. Men consistently over-reported their height more than women. In 2011, men and women estimated on average that they were 1.6cm and 1.0cm taller respectively than the measurement taken at the interviewer visit. The equivalent figures for 2016 were 1.5cm and 1.0cm respectively. Among both sexes, there was no linear trend (i.e. no steady rate of increase or decrease) in the gap between self-reported and interviewer-measured mean height. (Year: p=0.427; sex: p<0.001; year*sex: p=0.659)

Self-reported mean weight was consistently lower than interviewer-measured mean weight. Women consistently under-reported their weight more than men. In 2011, women and men under-reported their weight by an average of 2.1kg and 1.5kg respectively. The equivalent figures for 2016 were 2.0kg and 1.5kg respectively. Among both sexes, there was no linear trend in the gap between self-reported and interviewer-measured mean weight. (Year: p=0.731; sex: p<0.001; year*sex: p=0.506)

Due to the absence of any linear trends in the gap between self-reported and interviewer-measured height, weight, BMI, obesity and overweight (including obesity) the following analyses are presented using HSE data combined from survey years 2011 to 2016.