In their 1999 paper, Calvert and colleagues examined diabetes, serum glucose, and thyroid function (as determined by interviews and physical examinations) in 279 workers and 258 neighborhood referents with no occupational dioxin exposure, again matched for age, race, and sex. Workers were divided into four roughly equal-sized categories based on their lipid-adjusted serum TCDD level (<20, 20 ≤ TCDD < 75, 75 ≤ TCDD < 238, and 238 ≤ TCDD < 3,400 pg/g lipid). The glucose regression model included age, race, sex, body mass index (BMI) (stratified), and current medications that can increase serum glucose. The diabetes regression model added history of diabetes among parents or siblings. Diabetes was defined as a fasting serum glucose concentration greater than or equal to 7.8 mmol/l (140 mg/dl) on two days or the participant's reporting a history of diabetes diagnosed by a physician. Medical records were not obtained to validate self-reported diabetes.

The authors used three exposure indices, each in a separate regression analysis: (1) a dichotomous comparison of workers and referents; (2) serum TCDD concentration at the 1986 or 1992 examination, adjusted for serum lipid concentration; and (3) calculated half-life extrapolated, lipid-adjusted serum TCDD concentration. ⁶ Using measures from the second index, the authors split the workers into four equal-sized groups before beginning analyses. In the statistical analysis, each of the four groups was compared with the unexposed referent group. As expected, the workers had significantly increased mean current serum lipid-adjusted TCDD concentration (220 versus 7 pg/g; p < .001) and increased mean half-life extrapolated lipid-adjusted serum TCDD concentrations compared to neighborhood referents.

The worker and referent groups did not differ on serum glucose overall; neither was there a dose–response trend. The authors note, however, that the workers with the highest half-life extrapolated serum TCDD concentrations had significantly increased adjusted mean serum glucose concentration compared to referents. Overall, 9.3 percent of workers (26 individuals) and 7.0 percent of the referent group (18) met one of the definitions for diabetes. This translated to an increased, but not statistically significant, odds ratio of 1.49 (0.77–2.91) for diabetes, adjusted for race, age, BMI, family history of Type 2 diabetes, and current use of medications that can increase serum glucose concentrations. No dose– response trend was observed with serum TCDD or half-life extrapolated lipid-adjusted serum TCDD concentration. However, of the 10 workers with the highest current serum TCDD concentrations (>1,500 pg/g lipid), ⁷ 6 (60 percent) had diabetes mellitus. The authors also reanalyzed their data using the newer American Diabetes Association (ADA, 1997) serum glucose concentration diagnostic