Supplementary 2: Detailed Risk Models

Supplementary Table 1: Topography and histology codes of the International Classification of Diseases (ICD) for Oncology - 3rd revision – recorded during the study period for the incident cases of primary tumors of the central nervous system (CNS), leukemia, and lymphoma

Supplementary 2: Detailed risk models

In Poisson disease models, the expected number of cases, noted μjk, in each strata jwith sex s, birth period (p1=1995–2001, p2=2002–2010), attained age a(<3, 3–<5, 5–<7, 7–<9, ≥9 years), and age at exposure (<1, 1–<3, 3–<5, ≥5 years),foran exposure category k, isdefined by

μjk = PYjkλj (1 + ERRjk)

where PYjkis the number of persons-years,λjdenotes the “background” risk (without CT scan exposure), and ERRjkis the excess relative risk in relation to the cumulative organ dose.Exposure categoriesk were defined as <5, 5–24, 25–49, ≥50 mGy to the brain (for CNS cancer), <5, 5–9, 10–14, ≥15 mGy to the red bone marrow (RBM) (for leukaemia), and <5, 5–9, ≥10 mGy to the RBM (for lymphoma).

The “background” risk λj, adjusted for the presence of predisposing factors (PF)specific to the outcome (CNS cancer, leukaemia, or lymphoma), isdefined by

λj =exp(α1sj + α2p2j + α3 + α4+ γPF1j)

where aj is the exact attained age (in years) in the stratum j, tj the exact time since the first exposure (in years), and PFjdenotes the presence of PF (coded 1 if at least one PF, 0 otherwise). Attained age and time since the first exposure are centered on their mean values in the cohort to improve the stability of the models. Interactions between the covariates are not included because such models do not fit the data better than those with only main effects.

The relative risk (RR) associated with the PF is

RRFP = exp(γFP1j)

The excess relative riskERRjkis defined by

ERRjk = β1djk exp(δZj)

where djkis themean dose in the stratum jandthe exposure category k,Zjdenotes alternatively the modifying effect of the following covariates: sex s, presence of PFFP1j, age at the first exposure (<5, 5–9 years), time since exposure (<5, 5–11 years), and attained age (<7, 7–14 years).

1

Supplementary Table 3: Characteristics of CT scan exposures for each predisposing factor considered in the study

Supplementary Table 4: Number of incident cases (N) of primary tumorsof the central nervous system (CNS), leukemia and lymphoma, and incidence rates (IR) per 100,000 person-yearsfor various exclusion periods

Supplementary Table 5: Relative risks (RR) of cancer by categories of CT-scan exposure doses, for exclusion periods of 1, 2, 3, and 4 years

a-RR (95%CI)of central nervous system (CNS) cancers by brain dose category

b-RR (95%CI)of leukemia (including myelodysplasia) by red bone marrow dose category

c-RR (95%CI)of lymphoma by red bone marrow dose category

Supplementary Table 6: Number of cases of primary tumours of the central nervous system (CNS), leukaemia, and lymphoma, incidence rates (IR) per 100000 person-years, excess relative risks (ERR) related to cumulative organ dose (in mGy) from CT scans(adjusted for the presence of predisposing factors (PF) and for a 2-year exclusion period), by patients’ characteristics

PF: factors predisposing specifically to cancer at the site specified; CNS: central nervous system; IR: incidence rates per 100000 person-years; ERR: excess relative risk per mGy;p: p-values for the modifying effect based on likelihood ratio tests **the model does not fit for the defined categories of time since the first scan, although there is a trend of increased risk with increasing time since the first scan (p=0.110) when time is considered as a continuous variable

Supplementary Table 7: Sensitivity analyses for the estimation of excess relative risks (ERR) of CNS cancer, leukaemia, and lymphoma by cumulative organ dose (in mGy) from CT scans, adjusted for the presence of predisposing factors (2-year exclusion period)

a-Number of children, person-years (PY), mean follow-up, and number of cases

b-ERRof CNS cancer, leukemia, and lymphoma

1