Table 3

Associations of egg consumption with mortality from cardiovascular disease among 4 61 213 participants

EndpointsEgg consumptionP for linear trend*HR for 1 egg/week†
Never/rarely1–3 days/month1–3 days/week4–6 days/week7 days/week
PYs382 204838 6131 970 423442 564551 465
CVD
  Deaths1316223442969351204
 Deaths/PYs (1/1000)3.42.72.22.12.2
  Model 11.000.85 (0.80 to 0.91)0.78 (0.73 to 0.83)0.66 (0.60 to 0.72)0.63 (0.58 to 0.69)<0.0010.87 (0.85 to 0.89)
  Model 21.000.91 (0.85 to 0.98)0.87 (0.82 to 0.93)0.78 (0.71 to 0.85)0.78 (0.71 to 0.84)<0.0010.93 (0.91 to 0.95)
  Model 31.000.91 (0.85 to 0.98)0.88 (0.82 to 0.94)0.79 (0.73 to 0.87)0.82 (0.75 to 0.89)<0.0010.94 (0.92 to 0.97)
IHD
  Deaths3956641462338515
 Deaths/PYs (1/1000)1.00.80.70.80.9
  Model 11.000.86 (0.75 to 0.97)0.84 (0.75 to 0.94)0.72 (0.62 to 0.84)0.71 (0.62 to 0.82)<0.0010.91 (0.88 to 0.95)
  Model 21.000.90 (0.79 to 1.02)0.93 (0.83 to 1.05)0.84 (0.72 to 0.97)0.85 (0.74 to 0.97)0.0190.96 (0.92 to 0.99)
  Model 31.000.90 (0.79 to 1.02)0.94 (0.83 to 1.05)0.85 (0.73 to 0.99)0.88 (0.77 to 1.02)0.1310.97 (0.93 to 1.01)
Haemorrhagic stroke
  Deaths5408421373365315
 Deaths/PYs (1/1000)1.41.00.70.80.6
  Model 11.000.80 (0.72 to 0.89)0.69 (0.62 to 0.76)0.60 (0.52 to 0.69)0.53 (0.46 to 0.62)<0.0010.83 (0.80 to 0.86)
  Model 21.000.87 (0.78 to 0.97)0.79 (0.71 to 0.88)0.73 (0.63 to 0.84)0.67 (0.58 to 0.78)<0.0010.89 (0.85 to 0.93)
  Model 31.000.87 (0.78 to 0.97)0.79 (0.71 to 0.88)0.74 (0.65 to 0.86)0.72 (0.62 to 0.84)<0.0010.91 (0.87 to 0.95)
Ischaemic stroke
  Deaths11323743790126
 Deaths/PYs (1/1000)0.30.30.20.20.2
  Model 11.000.97 (0.78 to 1.22)0.90 (0.72 to 1.11)0.82 (0.62 to 1.09)0.73 (0.56 to 0.95)0.0060.90 (0.84 to 0.97)
  Model 21.001.05 (0.83 to 1.31)1.02 (0.82 to 1.27)1.00 (0.75 to 1.34)0.92 (0.70 to 1.21)0.3430.97 (0.90 to 1.04)
  Model 31.001.05 (0.83 to 1.31)1.02 (0.82 to 1.27)1.00 (0.74 to 1.33)0.93 (0.71 to 1.22)0.3880.97 (0.90 to 1.04)
  • Stratified Cox proportional models were used with stratification on survey site and birth cohort (in 5-year intervals). Multivariate models were adjusted for: model 1: age at recruitment (continuous) and sex (men or women); model 2: additionally included education level (no formal school, primary school, middle school, high school, college, or university or higher), household income (<2500, 2500–4999, 5000–9999, 10 000–19 999, 20 000–34 999, or ≥35 000 yuan/year), marital status (married, widowed, divorced or separated, or never married), alcohol consumption (not weekly; ex-regular; not daily; daily consuming 1–15, 15–29, 30–59, or ≥60 g), tobacco smoking (never or occasional; former; current smoking with 1–14, 15–24, or ≥25 cigarettes/day), physical activity in MET-hours/day (continuous), BMI (continuous), waist to hip ratio (continuous), prevalent hypertension (presence or absence), use of aspirin (presence, absence, or unknown), family history of CVD (presence or absence); model 3: additionally included intake of multivitamin supplementation (presence or absence) and dietary pattern (new affluence, traditional northern, or traditional southern).

  • *Tests for linear trend were conducted by assigning 0, 0.5, 2.0, 5.0, 7.0 to the frequency levels from the lowest to the highest and treating the variable as a continuous variable in the Cox model.

  • †HR for each one egg increment per week were calculated by using the usual amount in the multivariate Cox models.

  • BMI, body mass index; CVD, cardiovascular disease; HR, hazard ratios; IHD, ischaemic heart disease; MET, metabolic equivalent task; PY, person-years.