BACKGROUND: To explore to what extent measurement error can explain the variation of mean patient HbA(1c) between clinics. METHODS: For each year 2005-2010 data from 5380-6985 children, age <18 years, in 35-43 Swedish pediatric clinics was analyzed. Each year 13,000-19,000 HbA(1c) analyses were evaluated. Year mean HbA(1c) for each patient was calculated for HbA(1c) values when insulin dose was ≥0.5 U/kg. In Sweden HbA(1c) values were during the study period standardized to the Mono S level, HbA(1c)(Mono S)%, but are given also in the international unit HbA(1c)(IFCC), mmol/mol. Performance of locally measured HbA(1c) is monitored by Equalis through monthly external quality assessment (EQA) schemes. RESULTS: The yearly mean bias term for each clinic varied from -0.54 to 0.41 HbA(1c)(Mono S)%. The bias between clinic HbA(1c) and target value improved during the 6 years and the mean bias was for 79%-88% of clinics within the recommended level ±0.14 HbA(1c)% the last 2 years. Inter-clinic mean HbA(1c) had a wide interquartile range, 0.30-0.43 HbA(1c)(Mono S)% [3.2-4.5 HbA(1c)(IFCC)mmol/mol]. CONCLUSIONS: Regular participation in EQA schemes is necessary when comparing HbA(1c) values. The measurement error decreased during the 6-year period and explained from 28% to <10% of the inter-clinic variation in year mean clinic HbA(1c).
BACKGROUND: To explore to what extent measurement error can explain the variation of mean patient HbA(1c) between clinics. METHODS: For each year 2005-2010 data from 5380-6985 children, age <18 years, in 35-43 Swedish pediatric clinics was analyzed. Each year 13,000-19,000 HbA(1c) analyses were evaluated. Year mean HbA(1c) for each patient was calculated for HbA(1c) values when insulin dose was ≥0.5 U/kg. In Sweden HbA(1c) values were during the study period standardized to the Mono S level, HbA(1c)(Mono S)%, but are given also in the international unit HbA(1c)(IFCC), mmol/mol. Performance of locally measured HbA(1c) is monitored by Equalis through monthly external quality assessment (EQA) schemes. RESULTS: The yearly mean bias term for each clinic varied from -0.54 to 0.41 HbA(1c)(Mono S)%. The bias between clinic HbA(1c) and target value improved during the 6 years and the mean bias was for 79%-88% of clinics within the recommended level ±0.14 HbA(1c)% the last 2 years. Inter-clinic mean HbA(1c) had a wide interquartile range, 0.30-0.43 HbA(1c)(Mono S)% [3.2-4.5 HbA(1c)(IFCC)mmol/mol]. CONCLUSIONS: Regular participation in EQA schemes is necessary when comparing HbA(1c) values. The measurement error decreased during the 6-year period and explained from 28% to <10% of the inter-clinic variation in year mean clinic HbA(1c).
Authors: U Samuelsson; B Lindblad; A Carlsson; G Forsander; S Ivarsson; I Kockum; Å Lernmark; C Marcus; J Ludvigsson Journal: Diabetes Metab Res Rev Date: 2013-01 Impact factor: 4.876