Table 4.

Daycare attendance measured by child-hours and the risk of childhood ALL, the NCCLS, 1995 to 2002

Child-hours in thousandTotal ALL
c-ALL
Non-c-ALL
Case/controlOR (95% CI)*Case/controlOR (95% CI)*Case/controlOR (95% CI)*
Censored on the reference date
    Hispanic
        057/671.0039/371.0018/301.00
        >0 and <1533/500.98 (0.53-1.84)19/300.85 (0.35-2.08)14/201.25 (0.50-3.10)
        ≥1530/361.27 (0.62-2.56)13/191.02 (0.37-2.85)17/172.05 (0.68-6.12)
    P for trend0.730.380.54
    Non-Hispanic White
        037/351.0019/221.0018/131.00
        >0 and <1550/560.93 (0.44-1.95)33/350.97 (0.33-2.85)17/210.68 (0.22-2.13)
        ≥1549/810.60 (0.28-1.27)22/400.28 (0.08-0.95)27/410.80 (0.28-2.26)
    P for trend0.030.010.41
Censored 1 y before the reference date
    Hispanic
        066/871.0046/531.0020/341.00
        >0 and <1528/411.22 (0.64-2.32)14/231.14 (0.45-2.87)14/181.37 (0.54-3.48)
        ≥1526/251.81 (0.78-4.18)11/101.87 (0.54-6.39)15/152.27 (0.66-7.74)
    P for trend0.680.370.56
    Non-Hispanic White
        054/561.0034/411.0020/151.00
        >0 and <1536/500.93 (0.47-1.86)21/291.08 (0.44-2.66)15/210.69 (0.21-2.27)
        ≥1546/660.75 (0.36-1.55)19/270.34 (0.10-1.16)27/390.88 (0.30-2.58)
    P for trend0.080.010.46
Censored at age 1 y
    Hispanic
        098/1321.0058/741.0040/581.00
        >0 and <512/141.78 (0.66-4.83)6/81.90 (0.43-8.45)6/62.22 (0.54-9.24)
        ≥510/72.10 (0.70-6.34)7/42.53 (0.60-10.7)3/31.46 (0.26-8.03)
    P for trend0.230.210.81
    Non-Hispanic White
        0102/1161.0055/651.0047/511.00
        >0 and <520/261.27 (0.60-2.69)9/110.95 (0.27-3.39)11/151.52 (0.58-4.01)
        ≥514/300.42 (0.18-0.99)10/210.33 (0.11-1.01)4/90.50 (0.09-2.65)
    P for trend0.040.030.91
  • * ORs and 95% CIs are derived from conditional logistic regression models, adjusting for annual household income and maternal education.

  • “Censored” at a specific point in time indicates that exposure occurred after that point is not considered. The “reference date” is the date of diagnosis for cases and the corresponding date for matched controls.

  • P for trend is obtained by putting child-hours in the conditional logistic regression model as a continuous variable.