Table 2.
Predictors of log urinary 11-dehydo thromboxane B2 from stepwise generalized linear model
| Baseline factor | Estimate (SE) | P-value | Model degrees of freedom | Per cent of total sum of squares | Stepa | |
|---|---|---|---|---|---|---|
| Factor | Model | |||||
| Urine volume assayed 0.5 mL (vs. 1.0 mL) | −0.18 (0.02) | 2.5×10−16 | 1 | 1.12% | 1.12% | 0 |
| Female (vs. male) | 0.12 (0.02) | 1.4 × 10−06 | 2 | 0.42% | 1.55% | 0 |
|
Age group
<55 |
0.00 | 0.00045 | 7 | 0.33% | 1.88% | 0 |
| ≥55, <60 | 0.08 (0.03) | |||||
| ≥60, <65 | 0.05 (0.03) | |||||
| ≥65, <70 | 0.09 (0.03) | |||||
| ≥70, <75 | 0.15 (0.03) | |||||
| ≥75 | 0.24 (0.04) | |||||
| Allocated aspirin (vs. placebo) | 0.02 (0.02) | 0.15 | 8 | 0.03% | 1.91% | 0 |
|
Smoker
Current |
0.39 (0.03) | 1.4 × 10−32 | 11 | 2.47% | 4.38% | 1 |
| Former | 0.06 (0.02) | |||||
| Unknown | 0.07 (0.07) | |||||
| Never | 0.00 | |||||
| High-density lipoprotein cholesterol (per mmol/L) | −0.14 (0.03) | 1.6 × 10−14 | 12 | 0.94% | 5.33% | 2 |
|
Diabetes group
Type 2 diet only |
0.05 (0.05) | 3.7 × 10−07 | 14 | 0.47% | 5.80% | 3 |
| Type 2 oral hypoglycaemic or insulin | 0.13 (0.04) | |||||
| Type 1 | 0.00 | |||||
| Body mass index (per 5 kg/m2) | 0.04 (0.01) | 2.4 × 10−05 | 15 | 0.28% | 6.08% | 4 |
| Urinary albumin-creatinine ratio ≥ 3 (vs. <3 mg/mmol) | 0.09 (0.02) | 0.0014 | 16 | 0.16% | 6.24% | 5 |
| Estimated glomerular filtration rate (per 10mL/min/1.73 m2) | 0.01 (0.00) | 0.0021 | 17 | 0.15% | 6.39% | 6 |
| Aspirin use before screening (vs. not) | −0.05 (0.02) | 0.0026 | 18 | 0.14% | 6.54% | 7 |
| Systolic blood pressure (per 10mmHg) | −0.02 (0.01) | 0.0026 | 19 | 0.14% | 6.68% | 8 |
Table shows the 19 predictors used as adjustment factors in the outcome analyses. P-value is from the F test in the final model. Factors available for inclusion in the model included main effects and first-order interactions for continuous, grouped, and data availability indicators as appropriate for the factors urine volume assayed, age (grouped factor has three levels only for interaction term), sex, treatment allocation, smoking (current smoker vs. not for interaction term), diabetes type and treatment, duration of diabetes, body mass index, systolic and diastolic blood pressures, reported treated hypertension, diabetic retinopathy, ethnicity, Townsend Index, aspirin use before screening, statin use at baseline, angiotensin-converting enzyme inhibitor use at baseline, baseline 5-year risk of serious vascular event, days sample spent in the post, total cholesterol, high-density lipoprotein cholesterol, non-high-density lipoprotein cholesterol, apolipoprotein A1, apolipoprotein B, glycated haemoglobin, urinary albumin-creatinine ratio, and estimated glomerular filtration rate.
aThe factors listed as Step 0 were forced into the model.