TABLE 4.
PAR% of diabetes incidence with consumption of eggs adjusted for non-egg cholesterol as an exposure in the WHI1
| Exposure variables | Covariates | At risk/referent | Follow-up, 10,000 person-years (at risk/referent) | Diabetes cases per 10,000 person-years (at risk/referent) | Partial PAR% (95% CI) |
|---|---|---|---|---|---|
| Egg intake adjusted for non-egg cholesterol intake, full PAR%: 68.2 (50.2, 80.5) | |||||
| Egg intake, 5th quintile | ≥3/wk/<1.6/mo | 13/11 | 116/81 | 4.2 (2.3, 6.1) | |
| Egg intake, 4th quintile | 6.6/mo to <3/wk/<1.6/mo | 12/11 | 95/81 | 1.2 (−0.3, 2.7) | |
| Egg intake, 3rd quintile | 4 to <6.6/mo/<1.6/mo | 11/11 | 86/81 | 0.3 (−1.0, 1.6) | |
| Obesity, kg/m2 | ≥30/18.5 to <25 | 12/25 | 174/56 | 25.0 (22.3, 27.6) | |
| Overweight, kg/m2 | 25 to <30/18.5 to <25 | 21/25 | 93/56 | 12.8 (11.1, 14.5) | |
| Alcohol, drinks/wk | <1/≥1 | 33/36 | 106/76 | 9.6 (6.0, 13.3) | |
| Advanced age, y | ≥70/<60 | 17/15 | 96/88 | 7.8 (6.0, 9.7) | |
| Self-reported race/ethnicity | Non-white/white | 6/73 | 144/87 | 6.1 (4.8, 7.4) | |
| Old age, y | 60–69/<60 | 27/15 | 93/88 | 5.3 (2.6, 8.1) | |
| Smoking | Ever/never | 28/43 | 95/90 | 4.4 (1.9, 6.8) | |
| Non-egg cholesterol—5th quintile, mg/d | ≥210.4/<95.5 | 10/11 | 121/76 | 4.2 (1.8, 6.6) | |
| Regular coffee, cups/d | 0/≥0 | 31/39 | 97/88 | 4.1 (1.3, 7.0) | |
| Non-egg energy, kcal/d | ≥1255/<1255 | 21/53 | 93/93 | 3.6 (1.4, 5.9) | |
| Exercise, MET-h/wk | <6/≥6 | 18/56 | 114/83 | 3.5 (1.2, 5.8) | |
| AHEI-2010 | <52.3/≥52.3 | 22/51 | 108/83 | 3.1 (0.5, 5.7) | |
| Non-egg cholesterol—4th quintile, mg/d | 162.4 to <210.4/<95.5 | 12/11 | 99/76 | 2.8 (0.7, 4.9) | |
| 3-y BMI decrease, kg/m2 | <−0.9/−0.9 to <1.48 | 9/39 | 101/82 | 2.5 (1.3, 3.7) | |
| 3-y BMI increase, kg/m2 | ≥1.48/−0.9 to <1.48 | 11/39 | 123/82 | 2.3 (0.6, 3.9) | |
| Non-egg cholesterol—3rd quintile, mg/d | 128.6 to <162.4/<95.5 | 13/11 | 87/76 | 1.1 (−0.8, 3.0) | |
| Non-egg cholesterol—2nd quintile, mg/d | 95.5 to <128.6/<95.5 | 13/11 | 83/76 | 0.6 (−1.1, 2.3) | |
| Education level | Less than high school grad/more than high school grad | 9/68 | 106/90 | 0.6 (−0.8, 1.9) | |
PAR%s were calculated for the exposure and covariate variables separately (partial PAR%) and combined (full PAR%). The techniques and SAS macro developed by Spiegelman et al. (30) were used for this purpose. The egg consumption exposure variable was the mean of the year 0 and year 3 values at year 3 (follow-up baseline), when the WHI FFQ was readministered. The 2nd quintile of egg consumption was not included in the analysis because it did not yield an HR >1.0 (30). Incident diabetes was the first occurrence of self-reported diabetes treated with pills or injections between the year 3 baseline survey and March 31, 2018. Our analytic sample included observational study participants without 1) implausible FFQ energy intakes (defined as mean intakes <600 or >5000 kcal/d), BMI (<15 or >50), or height <122 cm (4 ft); 2) any of the following pre-existing major chronic conditions at follow-up baseline: diabetes, angina, myocardial infarction, stroke, heart failure, emphysema, kidney dialysis, coronary artery bypass graft, percutaneous coronary intervention, or cancer; or 3) missing values on any pre-existing major chronic condition or any exposure, outcome, or confounder variables. AHEI-2010, Alternative Healthy Eating Index 2010; MET, metabolic equivalent; PAR%, population attributable risk percentage; WHI, Women's Health Initiative.