Table 4

Sensitivity analysis of prior HRH’s impact on HRH risk* (n=342 058)†

Model‡OR (95% CI)P value
Model 1
 A1c coefficient of variation (%) (ref <4)
  4–5.91.19 (1.04 to 1.37)0.010
  6–9.41.28 (1.12 to 1.46)<0.001
  9.5–661.42 (1.24 to 1.63)<0.001
 Prior HRH (ref=no)
  Yes3.12 (2.65 to 3.67)<0.001
 A1c mean (%) (ref=7–7.9)
  <60.67 (0.56 to 0.79)<0.001
  6–6.90.78 (0.71 to 0.86)<0.001
  8–8.91.11 (1.00 to 1.25)0.057
  ≥91.53 (1.33 to 1.75)<0.001
Model 2
 A1c SD (ref <0.25)
  0.25–0.401.26 (1.10 to 1.45)0.001
  0.41–0.681.35 (1.17 to 1.55)<0.001
  0.69–6.461.54 (1.33 to 1.79)<0.001
 Prior HRH (ref=no)
  Yes3.12 (2.65 to 3.67)<0.001
 A1c mean (%) (ref=7–7.9)
  <60.71 (0.60 to 0.85)<0.001
  6–6.90.81 (0.73 to 0.89)<0.001
  8–8.91.09 (0.98 to 1.23)0.117
  ≥91.49 (1.30 to 1.72)<0.001
Model 3
 Adjusted A1c SD (ref <0.62)
  0.62–0.971.08 (0.95 to 1.23)0.253
  0.98–1.561.11 (0.97 to 1.26)0.127
  1.57–17.401.36 (1.19 to 1.56)<0.001
 Prior HRH (ref=no)
  Yes3.12 (2.65 to 3.68)<0.001
 A1c mean (%) (ref=7–7.9)
  <60.67 (0.57 to 0.80)<0.001
  6–6.90.79 (0.71 to 0.87)<0.001
  8–8.91.10 (0.98 to 1.23)0.113
  ≥91.48 (1.29 to 1.70)<0.001
  • *The sensitivity analysis was executed by adding prior HRH from the baseline period as a binary covariate to the logistic regression indicated in the Methods section.

  • †One patient was dropped from logistic regression due to missing service-connected disability.

  • ‡Each model was run with a measure of A1c variability in quartiles, previous HRH event, A1c mean, and the covariates listed in the Methods section.

  • HRH, hypoglycemia-related hospitalization; ref, reference.