Table 3.

Patient-level and facility-level mortality modelsa

ParameterMortality Models versusNon-TZD OHAs
All-Cause MortalityCardiovascular Mortality
Model 1Model 2Model 1Model 2
HR (95% CI)PHR (95% CI)PHR (95% CI)PHR (95% CI)P
Patient-level: Medications
    Rosiglitazone1.33 (1.02 to 1.73)0.03001.38 (1.05 to 1.82)0.02001.52 (1.06 to 2.18)0.02001.59 (1.14 to 2.22)0.0060
    Pioglitazone1.11 (0.76 to 1.61)0.58001.14 (0.79 to 1.64)0.48001.27 (0.75 to 2.18)0.37001.37 (0.77 to 2.42)0.2800
Facility-level: Prescribing patterns
    Adjusted facility percentage of patients prescribed rosiglitazone, 75th versus 25th percentile (9.4 versus 4.9%)1.14 (1.07 to 1.21)0.00011.16 (1.09 to 1.23)<0.00011.12 (1.02 to 1.24)0.02001.13 (1.01 to 1.26)0.0300
    Above versus below overall median (6.2%)1.36 (1.14 to 1.61)0.00041.36 (1.15 to 1.62)0.00031.47 (1.11 to 1.95)0.00801.42 (1.07 to 1.88)0.0100
  • a Data from US DOPPS I (1999 through 2001) and II (2002 through 2004), among patients reportedly on an OHA (n = 2393). Shown are HRs for the two TZDs, versus the reference category of all other (non-TZD) OHAs (HR = 1; data not shown). Models were stratified by study phase and controlled for the effects of facility clustering. Each facility measure was evaluated as a separate set of the two models. Model 1 covariate adjustments included the two TZD types and insulin; model 2 added covariates for age, gender, race (black versus nonblack), BMI, years with ESRD, comorbid conditions, hemoglobin, serum glucose, total cholesterol concentration, serum albumin, and delivered single-pool Kt/V. Cardiovascular deaths were those caused by the following: Acute MI, atherosclerotic heart disease, cardiomyopathy, cardiac arrhythmia, cardiac arrest, CHF, stroke/cerebrovascular accident, and ischemic brain damage/anoxic encephalopathy.