Table 2—

Prediction of diabetes by depression symptoms and antidepressant drug exposure

At baseline
During follow-up
BDI <11BDI ≥11No medicationsTaking medicationsBDI always <11BDI ever ≥11No medicationsIntermittent medicationsContinuous medications
PLB arm
    Pyr2,451.6300.42,624.9127.12,316.7435.32,460.0218.873.2
    Cases2703027723245552592813
    IR11.010.010.618.110.612.610.512.817.8
    HRu10.90 (0.61–1.32)11.80 (1.15–2.82)11.11 (0.82–1.50)11.17 (0.78–1.76)1.80 (1.00–3.25)
    HRa10.84 (0.56–1.26)12.25 (1.38–3.66)10.99 (0.72–1.37)11.50 (0.97–2.33)2.60 (1.37–4.94)
MET arm
    Pyr2,584.9289.02,675.7198.22,441.0432.92,549.2224.4100.3
    Cases199252131118737200195
    IR7.78.78.05.57.78.57.88.55.0
    HRu11.14 (0.74–1.75)10.69 (0.37–1.27)11.00 (0.69–1.43)10.96 (0.59–1.57)0.68 (0.28–1.68)
    HRa11.07 (0.68–1.68)10.72 (0.38–1.36)10.91 (0.62–1.33)10.99 (0.59–1.65)0.71 (0.28–1.78)
ILS arm
    Pyr2,711.2299.92,863.5147.62,577.2433.92,733.4178.799.0
    Cases130211341712031123199
    IR4.87.04.711.54.77.14.510.69.1
    HRu11.49 (0.93–2.39)12.68 (1.59–4.52)11.35 (0.90–2.02)11.91 (1.15–3.19)2.51 (1.29–4.89)
    HRa11.15 (0.69–1.93)13.48 (1.93–6.28)10.87 (0.56–1.39)12.07 (1.18–3.62)3.39 (1.61–7.13)
  • Data are hazard ratio(95% CI), unless otherwise indicated. Cases, incident cases of diabetes developing during follow-up. Pyr, person-years of follow-up. HRa, hazard ratio(from Cox proportional hazard model) with 95% CI, adjusted for age, sex, race/ethnicity, education, fasting plasma glucose, weight, and weight change. HRu, hazard ratio(from Cox proportional hazard model) with 95% CI, unadjusted for covariates. IR, incidence rate(cases/100 person-years of follow-up). In addition, depression symptom and antidepressant drug use are adjusted for each other.