Validity of self-assessed waist circumference in a multi-ethnic UK population.
Diabetic medicine : a journal of the British Diabetic Association 2011 ; 29: 404-9.
Khunti K, Taub N, Webb D, Srinivasan B, Stockman J, Griffin SJ, Simmons RK, Davies MJ
DOI : 10.1111/j.1464-5491.2011.03478.x
PubMed ID : 21988535
PMCID : 0
Abstract
To investigate validity of waist circumference measurements obtained by self-report and self-measurement with non-verbal pictorial instructions among a multi-ethnic population.
Five hundred and twenty-six individuals aged 40-75 years (91 South Asian, 430 White European and five other), who attended a screening programme for Type 2 diabetes, estimated their waist circumference and measured their waist with a paper tape measure. Participants were also provided with simple pictorial instructions for measurement of waist circumference in their preferred language and remeasured their waist circumference. We calculated 95% limits of agreement with measures undertaken by a healthcare professional unaware of prior measures.
Mean age was 56.8 years (sd 9.0), mean BMI 30.0 kg/m(2) (sd 5.6) and mean waist circumference 98.4 cm (sd 14.1). Seventy-nine per cent had high waist circumference according to International Diabetes Federation criteria. The mean of participants' self-reported value was 6.8 cm lower than the healthcare professional measure (sd 8.8; 95% limits of agreement -10.4 to 24.0 cm), with significant differences by sex and ethnicity (South Asian men 7.5 cm, South Asian women 0.1 cm, White European men 7.8 cm, White European women 7.0 cm, P < 0.001). Compared with healthcare professional measures, mean self-measured waist circumference was very similar, both with instructions (0.4 cm higher; sd 5.5 cm; -11.1 to 10.4 cm) and without instructions (0.5 cm lower; sd 5.6; -10.4 to 11.4 cm), but with significant differences by sex and ethnicity (P < 0.001).
There was systematic underestimation of self-reported waist circumference in this multi-ethnic UK population. The magnitude of underestimation might reduce the performance of risk scores; however, this can be corrected through self-measurement with pictorial instructions.