Table 3.

Sensitivity analysis for associations of serum folate with cancer mortality, Third National Health and Nutrition Examination Survey Linked Mortality File 1988-2000

Quintiles of serum folate concentration (ng/mL)
P*
<3.03.0-<4.34.3-<6.16.1-<9.4≥9.4
Scenario 1: including participants with cancer (excluding skin cancer) at baseline
No. of cases7211510511093
    HR adjusted for age, sex, and race (95% CI)1.01.37 (0.96-1.96)0.87 (0.60-1.26)1.09 (0.80-1.50)0.65 (0.44-0.98)<0.01
    Fully adjusted HR (95% CI)1.01.55 (1.10-2.17)1.07 (0.72-1.57)1.41 (1.00-1.98)0.90 (0.58-1.41)0.02
Scenario 2: restricting analysis to participants with cancer (excluding skin cancer) at baseline§
    No. of cases1016202422
    HR adjusted for age, sex, and race (95% CI)1.01.00 (0.34-2.91)1.08 (0.44-2.65)1.17 (0.48-2.87)0.80 (0.29-2.20)0.82
    Fully adjusted HR (95% CI)1.01.13 (0.41-3.15)1.21 (0.55-2.66)1.45 (0.67-3.12)1.23 (0.48-3.16)0.90
Scenario 3: restricting analysis to cancer mortality that occurred after 3 y follow-up
    No. of cases5089867967
    HR adjusted for age, sex, and race (95% CI)1.01.74 (1.07-2.83)1.02 (0.62-1.68)1.23 (0.80-1.88)0.81 (0.54-1.21)0.01
    Fully adjusted HR (95% CI)1.01.95 (1.22-3.14)1.22 (0.74-2.04)1.57 (1.01-2.45)1.09 (0.70-1.71)0.03
Scenario 4: using average daily grain consumption as time-varying covariate to account for folic acid fortification beginning in 1998
    No. of cases6299858671
    HR adjusted for age, sex, and race (95% CI)1.01.44 (0.95-2.18)0.82 (0.52-1.29)1.09 (0.72-1.66)0.63 (0.40-0.99)<0.01
    Fully adjusted HR (95% CI)1.01.63 (1.07-2.48)1.02 (0.64-1.61)1.42 (0.92-2.19)0.87 (0.55-1.38)<0.01
Scenario 5: restricting analysis to cancer mortality that occurred before folic acid fortification (January 1, 1998)**
    No. of cases4363576451
    HR adjusted for age, sex, and race (95% CI)1.01.10 (0.71-1.69)0.60 (0.37-0.96)0.90 (0.56-1.45)0.48 (0.29-0.78)<0.01
    Fully adjusted HR (95% CI)1.01.24 (0.82-1.87)0.73 (0.44-1.22)1.16 (0.72-1.88)0.65 (0.38-1.12)0.05
  • * P for trend across quintiles of serum folate concentrations based on adjusted Satterthwaite F-test; all tests two-tailed.

  • Scenario 1 included participants who at baseline had been told by a physician that they had cancer other than skin cancer.

  • Adjusted for age, sex, race or ethnicity, smoking status, alcohol intake, BMI, education, NSAID use, physical activity, HEI, and total energy intake.

  • § Scenario 2 included only participants who at baseline had been told by a physician that they had cancer other than skin cancer, and cancer mortality was restricted to that occurring within 3 y of follow-up.

  • Scenario 3 excluded cancer mortality that occurred within 3 y of follow-up.

  • Scenario 4 included average daily intakes of grain products as a time-varying covariate in the Cox proportional hazards models to adjust for possible effects of folic acid fortification mandated in 1998 in the United States. We used SAS Proc PHREG with the NORMALIZE option after a weight statement to rescale the weighted population to the actual sample size for the time-varying covariate analysis. P value based on Wald χ2 test.

  • ** Scenario 5 restricted cancer mortality to that that occurred before folic acid fortification began in the United States in January 1998.