TABLE 2

HRs of use of insulin for cancer in a cohort of 971 insulin new users and 1,935 insulin nonusers matched on age, smoking status, and the likelihood of using insulin

HR*95% CIP
Models for cancer*
    Model 1
        A1C (%)1.160.99–1.360.0747
        Use vs. nonuse of insulin0.180.10–0.33<0.0001
    Model 2
        A1C (%)1.241.03–1.490.0267
        Use vs. nonuse of insulin0.180.10–0.33<0.0001
    Model 3§
        A1C (%)1.261.03–1.550.0230
        Use vs. nonuse of insulin0.170.09–0.32<0.0001
Models for death*
    Model 1
        Use vs. nonuse of insulin1.270.92–1.750.1543
    Model 2
        Use vs. nonuse of insulin1.240.84–1.840.2739
    Model 3
        Use vs. nonuse of insulin1.280.85–1.940.2422
  • *Stratified Cox models on the matching pairs were used.

  • †Not adjusted for other covariates.

  • ‡Adjusted for HDL cholesterol, triglyceride, eGFR, and use of metformin. Other variables had a P value larger than 0.3 and not selected by the stepwise algorithm with P = 0.30. These variables included sex, alcohol drinking (previous and current), duration of diabetes, BMI, systolic blood pressure, LDL cholesterol–related risk (indicator terms for LDL cholesterol <2.8 mmol/l plus albuminuria and ≥3.8 mmol/l), Ln (ACR+1), use of antihypertensive drugs (other than ACEIs or ARBs) at enrollment, and use of drugs from enrollment to the earliest date of cancer, death, or censoring (ACEIs or ARBs, statins, fibrates, and oral antidiabetic drugs listed in Table 1).

  • §Restricted cubic spline was further used to adjust for nonlinear associations between HDL cholesterol and triglyceride with cancer.

  • ‖Adjusted for A1C, BMI, eGFR, Ln(ACR+1), use of antihypertensive drugs (other than ACEIs or ARBs) at enrollment, and use of ACEIs or ARBs, fibrates, gliclazide, and rosiglitazone from enrollment to the earliest date of cancer, death, or censoring (selected by the stepwise algorithm with P = 0.30).

  • ¶Restricted cubic spline was further used to adjust for nonlinear associations among A1C, BMI, ACR, and eGFR with death.