Table 3

Comparison of combinations of criteria over individual criteria.

Author (year)NIndividual Criteria
Per cent correctly classified
Combined—2 criteria
Per cent correctly classified
Combined—3 criteria
Per cent correctly classified
Age at diagnosisBMI (or equivalent)Insulin treatment/
Time to insulin (TTI)
Age at diagnosis and BMIAge at diagnosis and Insulin/TTIBMI and Insulin/TTIAge at diagnosis, BMI and Insulin/TTIRegression equation or algorithm using all 3 criteria
Boyle (1999)3613 (1807†)925863909393
Laakso (1987)17167737561616756
Prior (1993)57582788589***8089
Shields (2012)7281686982
Welborn (1981)203797788**
Welborn (1983)12185698693**93
  • Data presented as overall percentage correctly classified according to C-peptide category (below or above cut-off for insulin deficiency) using cut-offs of individual criteria and combinations of criteria, for the six studies where comparison within studies was possible. Results in bold are those where the addition of another clinical feature provides better classification within studies.

  • **p<0.01, ***p<0.001, by McNemar's test.

  • †Regression equations/algorithms tested on a separate data set, so a two sample χ2 test is used to determine statistical significance.

  • BMI, body mass index.