## Abstract

Background Previous studies on the association between childhood infections and childhood leukaemia have produced inconsistent results, likely due to the recall error/bias of infection data reported by the parents. The current study used a population-based and record-based case–control design to evaluate the association between childhood leukaemia and infections using the National Health Insurance Research Database of Taiwan.

Methods In all, 846 childhood acute lymphoblastic leukaemia (ALL) and 193 acute myeloid leukaemia (AML) patients newly diagnosed between 2000 and 2008, aged >1 and <10 years, were included. Up to four controls (3374 for ALL and 766 for AML) individually matched to each case on sex, birth date and time of diagnosis (reference date for the controls) were identified. Conditional logistic regression was performed to assess the association between childhood leukaemia and infections.

Results Having any infection before 1 year of age was associated with an increased risk for both childhood ALL (odds ratio = 3.2, 95% confidence interval 2.2–4.7) and AML (odds ratio = 6.0, 95% confidence interval 2.0–17.8), with a stronger risk associated with more episodes of infections. Similar results were observed for infections occurring >1 year before the cases’ diagnosis of childhood leukaemia.

Conclusions Children with leukaemia may have a dysregulated immune function present at an early age, resulting in more episodes of symptomatic infections compared with healthy controls. However, confounding by other infectious measures such as birth order and day care attendance could not be ruled out. Finally, the results are only relevant to the medically diagnosed infections.

## Introduction

Leukaemia is the most common cancer among children, accounting for approximately one-third of all childhood cancers.1 The two major histological subtypes of childhood leukaemia are acute lymphoblastic leukaemia (ALL) and acute myeloid leukaemia (AML).2 Approximately 10% of childhood leukaemia cases can be explained by sex, age, race, exposure to ionizing radiation and congenital diseases (e.g. Down syndrome, neurofibromatosis);3,4 however, the causes of most childhood leukaemia remain unknown. Greaves’5 ‘delayed infection’ hypothesis states that microbial exposures during early childhood are critical for the normal development of immune function. A lack of sufficient early-life microbial challenges may result in a dysregulated immune response to infections when encountered later in childhood, leading to the development of leukaemia.5 Kinlen’s6 ‘population mixing’ hypothesis states that an elevated risk of childhood leukaemia may result from increasing contact between susceptible and infected individuals coming from different geographical populations. The ‘population mixing’ hypothesis suggests the existence of a specific leukaemia-causing agent(s), whereas the ‘delayed infection’ hypothesis does not; however, the common factor in both hypotheses is the abnormal immune responses to infections, which may lead to the development of childhood leukaemia.

Studies of the association between infections and childhood leukaemia have been conducted with proxy measures of infections such as day care attendance and birth order and with direct measures of infections. In a meta-analysis of 14 studies, Urayama et al.7 reported an inverse association between day care attendance and childhood ALL [combined odds ratio (OR) = 0.76, 95% confidence interval (CI) 0.67–0.87]. The association between childhood ALL and birth order is less consistent than that for day care attendance, with some showing an inverse association,812 whereas others reported either a positive association13,14 or a null association.15–19

Thirteen studies have examined the association between early childhood infections and risk of childhood leukaemia, mostly focusing on infections occurring before 1 year of age and ALL.10,12,13,1517,2026 These studies have produced inconsistent results on the relationship between childhood leukaemia and early childhood infections, with seven reporting an inverse association,10,12,13,15,17,21,26 four reporting a null association20,2325 and two reporting a positive association.16,22 Eleven10,12,13,1517,20,21,23,25,26 of the 13 studies relied on self-report by the parents to obtain the child’s infection history, which can be subject to recall error/bias. Of the two studies that used medical records for the child’s infection history, one reported a null association24 and the other reported a positive association.22 Though 12 of the 13 studies were population-based studies with cases and controls recruited from the same source population, the participation of cases and controls was not 100%. This could have affected the accuracy of the results if participants and non-participants had different characteristics (e.g. socio-economic status) that may be associated with both childhood infections and leukaemia.

To avoid the potential problems associated with recall error/bias and non-participation, the current study used the population-based and record-based case–control design to evaluate the association between childhood infections and childhood leukaemia using data of 1039 childhood leukaemia cases and 4140 controls from the National Health Insurance Research Database (NHIRD) of Taiwan.

## Materials and Methods

This study was approved by the Institutional Review Board of the National Health Research Institutes, Taiwan.

### Data source

The data used for the current analysis came from the NHIRD, which is a population-based database generated for medical research using the administrative and health claims data recorded by Taiwan’s National Health Insurance (NHI) programme. Taiwan’s NHI program is a single-payer programme launched on 1 March 1995 and covers ∼99% of the 23 million Taiwanese citizens. Health-care facilities contracted under the NHI provide the insurees of the NHI with inpatient care, ambulatory care, dental care and prescription drugs. The claims data of the NHI are routinely monitored by the Bureau of the NHI for their accuracy and completeness.27

### Subject selection

Two data sets of the NHIRD were used for subject selection28: (i) Catastrophic Illness Dataset, which is a data set containing health claims data associated with serious illnesses, including cancer; and (ii) The Longitudinal Health Insurance Database 2005, which is a database containing the health claims data of 1 million enrollees randomly sampled from the 2005 NHIRD enrolment file. This 1 million-person random sample is representative of the entire insured population of Taiwan.

From the Catastrophic Illness Dataset, we identified children newly diagnosed with leukaemia [ALL, International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code: 204.0 and AML, ICD-9-CM code: 205.0] between January 1, 2000 and December 31, 2008, aged >1 year and <10 years. Patients diagnosed with malignant tumours are eligible to apply for a certificate of catastrophic illness to be exempted from all co-payments. Because of this financial incentive, almost everyone with cancer in Taiwan has a certificate of catastrophic illness; thus, the identification of cancer cases from the Catastrophic Illness Dataset should be nearly 100%. In addition, to apply for the certificate of catastrophic illness, one is required to have an official certificate of diagnosis from the hospital; thus, the cancer cases identified from the Catastrophic Illness Dataset should be accurate. The date recorded in the Catastrophic Illness Dataset or the first date of inpatient claims for childhood leukaemia, whichever came first, was considered the date of diagnosis. Infant leukaemia cases (leukaemia diagnosed at age 1 year or younger) were excluded because their pathogenesis is distinctly different from non-infant childhood leukaemia.3 To capture the full infection history from birth, leukaemia cases diagnosed at age 10 years or older were excluded (the earliest available data in our database are from January 1, 1999). For every leukaemia case, up to four controls, with no history of cancer, individually matched to the case on date of birth, sex and time of case diagnosis (reference date for the control) were identified from The Longitudinal Health Insurance Database 2005.

### Data collection

#### Infections

History of infection during two time periods, before 1 year of age and >1 year before the patient’s diagnosis (reference date for the matched controls), was identified from two NHIRD files: (i) Ambulatory Care Expenditures by Visits and (ii) Inpatients Expenditures by Admission. The infectious conditions included otitis media (ICD-9-CM codes: 381 and 382), acute respiratory infections (ICD-9-CM codes: 460–466), pneumonia and influenza (ICD-9-CM codes: 480–488), unspecified bronchitis (ICD-9-CM code: 490), intestinal infectious diseases (ICD-9-CM codes: 001-009), conjunctivitis (ICD-9-CM code: 372.0–372.3, 771.6) and infections specific to the perinatal period (ICD-9-CM code: 771). Before 2000, diagnoses in the NHIRD were recorded in A-codes or ICD-9-CM codes (only ICD-9-CM codes have been used since 2000). As the A-code system groups infectious and non-infectious conditions under the same code, it was impossible to determine the definite infection status for some subjects, and a ‘possible’ infection status was given to those subjects. Infections diagnosed within 1 year before the patient’s diagnosis of leukaemia were not considered for the analysis to avoid reverse causality, as the development of childhood leukaemia may influence one’s immune status and susceptibility to infections.

#### Covariates

Socio-economic status was measured by the insured payroll-related amount of the primary insuree (mostly either the father or the mother and sometimes a grandparent), with 17 880 New Taiwanese dollars being the lowest insured payroll-related amount (∼623 US dollars). Although the insured payroll-related amount is set according to each individual’s income level, the reimbursement and the co-payment are the same for every insuree regardless of his/her insured payroll-related amount. Therefore, the level of reimbursement should not have an influence on the number of clinical or hospital visits in our study, and was thus not included as a covariate. We included one other covariate, the level of urbanization, which was determined by the city of residence according to a published categorization scheme.29

### Statistical analysis

The distributions of insured payroll-related amount and urbanization levels between cases and controls were compared using chi-squared tests. Conditional logistic regression was performed to generate OR and 95% CI estimating the risk of ALL or AML associated with having specific infections or any infections (Yes/No) and the number of clinical visits for each specific infection or any infections (OR calculated for each visit increment). In addition, the total number of visits for any infections was categorized into four groups (0, 1–5, 6–10 and >10 visits) for analysis. As the number of subjects with a ‘possible’ infection status is small, the interpretation of the ORs comparing the ‘possible infection’ with the ‘no infection’ group may not be meaningful and is not recommended. Subjects aged 2 years or younger were excluded from analysis examining the association between infections occurring >1 year before diagnosis and childhood leukaemia.

Stratified analyses were performed by age groups for ALL only (age 2–5.9 years vs age 6–9.9 years) and by urbanization levels (high urbanization vs low urbanization) and insured payroll-related amount for both ALL and AML. Immune-related factors are thought to have particularly strong contributions to the occurrence of childhood ALL among children aged 2–5.9 years.5 Urbanization levels and socio-economic status may influence the opportunities for infectious exposures. The heterogeneity of the association between infection and childhood leukaemia by age groups, urbanization levels and insured payroll-related amount was evaluated by the log–likelihood ratio test comparing the conditional logistic regression model with the interaction term (infection × age groups, infection × urbanization level or infection × insured payroll-related amount) with the model without the interaction term.

As adjustment for insured payroll-related amount and urbanization levels produced ORs that were similar to those without adjustment, the insured payroll-related amount and urbanization levels were not included in the final statistical models.

## Results

A total of 1039 childhood leukaemia cases and 4140 controls (846 ALL cases and 3374 matched controls; 193 AML cases and 766 matched controls) were identified from the NHIRD for our analysis. Most of the ALL cases (53%) were between 2 and 4.9 years, whereas the 5- to 9.9-year-old patients were the largest group (46%) for AML (Table 1). There were more males than females for both ALL (58% male) and AML (54% male). The cases and controls did not differ in the distribution of insured payroll-related amount or urbanization levels.

An increased risk of childhood ALL was associated with infections occurring before 1 year of age, especially for acute respiratory infection (OR = 3.4, 95% CI 2.3–4.9; Table 2). Having any infections before 1 year of age was associated with about three times the risk of childhood ALL (OR = 3.2, 95% CI 2.2–4.7), with each additional visit conferring a 1.4% increase in the risk of ALL (OR = 1.014, 95% CI 1.006–1.022). Compared with those with no clinical visit for infection, the ORs associated with childhood ALL for those with 1–5, 6–10 and >10 clinical visits for infections before 1 year of age were 2.8 (95% CI 1.8–4.3), 3.3 (95% CI 2.1–5.1) and 3.6 (95% CI 2.4–5.5), respectively. Having acute respiratory infections, intestinal infections or conjunctivitis >1 year before the diagnosis of ALL was associated with an elevated risk of childhood ALL. Having any infections >1 year before the diagnosis of ALL was associated with about four times the risk of childhood ALL (OR = 3.9, 95% CI 2.6–5.8), with each additional visit conferring a 0.3% increase in the risk of ALL (OR = 1.003, 95% CI 1.001–1.005). Compared with those with no clinical visit for infection, the ORs associated with childhood ALL for those with 1–5, 6–10 and >10 clinical visits for infections >1 year before the ALL diagnosis were 2.3 (95% CI 1.4–3.7), 4.3 (95% CI 2.6–7.3) and 4.8 (95% CI 3.1–7.4), respectively.

Table 1

Characteristics of cases and controls by childhood leukaemia subtypes, Taiwan, 2000–08

Characteristics ALL

AML

Case Control P-value Case Control P-value
N = 846 N = 3374 N = 193 N = 766
n (%) n (%) n (%) n (%)
Age (years)
1.01–1.99 84 (9.9) 331 (9.8) a 38 (19.7) 148 (19.3) a
2.00–4.99 445 (52.6) 1776 (52.6)  66 (34.2) 263 (34.3)
5.00–9.99 317 (37.5) 1267 (37.6)  89 (46.1) 355 (46.3)
Sex
Male 487 (57.6) 1943 (57.6) a 105 (54.4) 417 (54.4) a
Female 359 (42.4) 1431 (42.4)  88 (45.6) 349 (45.6)
Income-related insured amount (NT$)b <17 881 201 (23.9) 810 (24.0) 0.20 53 (27.5) 175 (22.9) 0.17 17 881–30 000 383 (45.5) 1421 (42.1) 90 (46.6) 335 (43.7) 30 001–40 000 84 (10.0) 443 (13.1) 13 (6.7) 91 (11.9) 40 001–50 000 94 (11.1) 407 (12.1) 18 (9.3) 94 (12.3) 60 001–70 000 43 (5.1) 150 (4.5) 8 (4.2) 40 (5.2) 70 001–80 000 27 (3.2) 97 (2.9) 10 (5.2) 24 (3.1) >80 000 10 (1.2) 45 (1.3) 1 (0.5) 7 (0.9) Urbanization level 1 (high) 255 (30.1) 959 (28.4) 0.70 54 (28.0) 200 (26.1) 0.77 2 226 (26.7) 898 (26.6) 55 (28.5) 209 (27.3) 3 157 (18.6) 652 (19.3) 32 (16.6) 168 (21.9) 4 104 (12.3) 459 (13.6) 32 (16.6) 109 (14.2) 5 20 (2.4) 69 (2.1) 4 (2.1) 9 (1.2) 6 22 (2.6) 118 (3.5) 6 (3.1) 26 (3.4) 7 (low) 34 (4.0) 127 (3.8) 7 (3.6) 30 (3.9) Unknown 28 (3.3) 92 (2.7) 3 (1.5) 15 (2.0) Characteristics ALL AML Case Control P-value Case Control P-value N = 846 N = 3374 N = 193 N = 766 n (%) n (%) n (%) n (%) Age (years) 1.01–1.99 84 (9.9) 331 (9.8) a 38 (19.7) 148 (19.3) a 2.00–4.99 445 (52.6) 1776 (52.6) 66 (34.2) 263 (34.3) 5.00–9.99 317 (37.5) 1267 (37.6) 89 (46.1) 355 (46.3) Sex Male 487 (57.6) 1943 (57.6) a 105 (54.4) 417 (54.4) a Female 359 (42.4) 1431 (42.4) 88 (45.6) 349 (45.6) Income-related insured amount (NT$)b
<17 881 201 (23.9) 810 (24.0) 0.20 53 (27.5) 175 (22.9) 0.17
17 881–30 000 383 (45.5) 1421 (42.1)  90 (46.6) 335 (43.7)
30 001–40 000 84 (10.0) 443 (13.1)  13 (6.7) 91 (11.9)
40 001–50 000 94 (11.1) 407 (12.1)  18 (9.3) 94 (12.3)
60 001–70 000 43 (5.1) 150 (4.5)  8 (4.2) 40 (5.2)
70 001–80 000 27 (3.2) 97 (2.9)  10 (5.2) 24 (3.1)
>80 000 10 (1.2) 45 (1.3)  1 (0.5) 7 (0.9)
Urbanization level
1 (high) 255 (30.1) 959 (28.4) 0.70 54 (28.0) 200 (26.1) 0.77
2 226 (26.7) 898 (26.6)  55 (28.5) 209 (27.3)
3 157 (18.6) 652 (19.3)  32 (16.6) 168 (21.9)
4 104 (12.3) 459 (13.6)  32 (16.6) 109 (14.2)
5 20 (2.4) 69 (2.1)  4 (2.1) 9 (1.2)
6 22 (2.6) 118 (3.5)  6 (3.1) 26 (3.4)
7 (low) 34 (4.0) 127 (3.8)  7 (3.6) 30 (3.9)
Unknown 28 (3.3) 92 (2.7)  3 (1.5) 15 (2.0)

aMatching variable.

bNT\$ = New Taiwanese dollar.

An increased risk of childhood AML was associated with infections occurring before 1 year of age, specifically for acute respiratory infection, pneumonia and influenza and intestinal infectious diseases (Table 3). Having any infections before 1 year of age was associated with six times the risk of childhood AML (OR = 6.0, 95% CI 2.0–17.8). Compared with those with no clinical visit for infection, the ORs associated with childhood AML for those with 1–5, 6–10 and >10 clinical visits for infections before 1 year of age were 3.5 (95% CI 1.1–10.8), 5.8 (95% CI 1.9–17.8) and 6.4 (95% CI 2.2–18.5), respectively. Having acute respiratory infections or pneumonia and influenza >1 year before the diagnosis of AML was associated with an elevated risk of childhood AML. Having any infection >1 year before the diagnosis of AML was associated with about five times the risk of childhood AML (OR = 5.3, 95% CI 2.0–14.0). Compared with those with no clinical visit for infection, the ORs associated with childhood AML for those with 1–5, 6–10 and >10 clinical visits for infections >1 year before the AML diagnosis were 2.8 (95% CI 0.9–8.9), 6.4 (95% CI 1.9–21.7) and 5.7 (95% CI 1.9–17.3), respectively.

Table 2

The association between childhood ALL and infections diagnosed during different time periods, Taiwan, 2000–08

Medically diagnosed infections Before 1 year of age

>1 year before the case’s diagnosis

Case Control OR (95% CI)a Caseb Controlb OR (95% CI)a
N = 846 N = 3374 N = 762 N = 3043
n (%) n (%) n (%) n (%)
Otitis media
No 795 (94.0) 3165 (93.8) Referent 563 (73.9) 2321 (76.3) Referent
Yes 51 (6.0) 205 (6.1) 0.99 (0.71–1.36) 197 (25.8) 694 (22.8) 1.19 (0.98–1.45)
Possible 0 (0.0) 4 (0.1) – 2 (0.3) 28 (0.9) 0.29 (0.07–1.23)
Each additional visitc   1.008 (0.930–1.092)   0.999 (0.974–1.024)
Acute respiratory infections
No 308 (36.4) 1463 (43.4) Referent 35 (4.6) 415 (13.7) Referent
Yes 507 (59.9) 1802 (53.4) 3.35 (2.30–4.87) 693 (90.9) 2463 (80.9) 4.10 (2.79–6.03)
Possible 31 (3.7) 109 (3.2) 2.83 (1.59–5.02) 34 (4.5) 165 (5.4) 2.06 (1.19–3.54)
Each additional visitc   1.015 (1.006–1.024)   1.003 (1.001–1.006)
Pneumonia and influenza
No 739 (87.4) 2987 (88.5) Referent 476 (62.5) 2003 (65.8) Referent
Yes 93 (11.0) 333 (9.9) 1.14 (0.88–1.46) 252 (33.1) 925 (30.4) 1.16 (0.97–1.40)
Possible 14 (1.6) 54 (1.6) 1.04 (0.56–1.93) 34 (4.4) 115 (3.8) 1.25 (0.83–1.88)
Each additional visitc   1.040 (0.986–1.097)   1.008 (0.991–1.026)
Unspecified bronchitis
No 833 (98.5) 3319 (98.4) Referent 712 (93.4) 2808 (92.3) Referent
Yes 7 (0.8) 32 (0.9) 0.88 (0.39–1.98) 38 (5.0) 161 (5.3) 0.93 (0.65–1.35)
Possible 6 (0.7) 23 (0.7) 1.03 (0.42–2.56) 12 (1.6) 74 (2.4) 0.63 (0.34–1.18)
Each additional visitc   0.870 (0.569–1.330)   0.889 (0.763–1.036)
Intestinal infectious diseases
No 698 (82.5) 2846 (84.3) Referent 453 (59.4) 1959 (64.4) Referent
Yes 131 (15.5) 444 (13.2) 1.23 (0.98–1.54) 275 (36.1) 949 (31.2) 1.29 (1.08–1.54)
Possible 17 (2.0) 84 (2.5) 0.80 (0.45–1.40) 34 (4.5) 135 (4.4) 1.09 (0.73–1.61)
Each additional visitc   1.036 (0.960–1.117)   1.008 (0.976–1.041)
Conjunctivitis
No 727 (85.9) 2987 (88.5) Referent 438 (57.5) 1918 (63.0) Referent
Yes 103 (12.2) 348 (10.3) 1.24 (0.97–1.59) 289 (37.9) 1014 (33.3) 1.29 (1.07–1.54)
Possible 16 (1.9) 39 (1.2) 1.74 (0.94–3.21) 35 (4.6) 111 (3.7) 1.38 (0.92–2.06)
Each additional visitc   1.034 (0.914–1.169)   1.043 (1.002–1.085)
Perinatal infections
No 814 (96.2) 3265 (96.8) Referent 736 (96.6) 2948 (96.9) Referent
Yes 32 (3.8) 104 (3.1) 1.24 (0.82–1.86) 26 (3.4) 90 (2.9) 1.16 (0.74–1.82)
Possible 0 (0.0) 5 (0.2) – 0 (0.0) 5 (0.2) –
Each additional visitc   1.153 (0.970–1.370)   1.135 (0.959–1.344)
Any infectionsd
No 305 (36.0) 1429 (42.3) Referent 33 (4.3) 368 (12.1) Referent
Yes 510 (60.3) 1831 (54.3) 3.18 (2.17–4.66) 696 (91.4) 2503 (82.2) 3.90 (2.61–5.81)
Possible 31 (3.7) 114 (3.4) 2.67 (1.48–4.79) 33 (4.3) 172 (5.7) 1.83 (1.05–3.17)
Each additional visitc   1.014 (1.006–1.022)   1.003 (1.001–1.005)
Number of clinical visits for infection
0 visits 305 (37.4) 1429 (43.8) Referent 33 (4.5) 368 (12.8) Referent
1–5 visits 109 (13.4) 441 (13.6) 2.77 (1.81–4.26) 63 (8.6) 350 (12.2) 2.26 (1.38–3.69)
6–10 visits 101 (12.4) 369 (11.3) 3.27 (2.10–5.09) 56 (7.7) 191 (6.7) 4.33 (2.58–7.28)
>10 visits 300 (36.8) 1021 (31.3) 3.63 (2.40–5.48) 577 (79.2) 1962 (68.3) 4.78 (3.08–7.39)
Medically diagnosed infections Before 1 year of age

>1 year before the case’s diagnosis

Case Control OR (95% CI)a Caseb Controlb OR (95% CI)a
N = 846 N = 3374 N = 762 N = 3043
n (%) n (%) n (%) n (%)
Otitis media
No 795 (94.0) 3165 (93.8) Referent 563 (73.9) 2321 (76.3) Referent
Yes 51 (6.0) 205 (6.1) 0.99 (0.71–1.36) 197 (25.8) 694 (22.8) 1.19 (0.98–1.45)
Possible 0 (0.0) 4 (0.1) – 2 (0.3) 28 (0.9) 0.29 (0.07–1.23)
Each additional visitc   1.008 (0.930–1.092)   0.999 (0.974–1.024)
Acute respiratory infections
No 308 (36.4) 1463 (43.4) Referent 35 (4.6) 415 (13.7) Referent
Yes 507 (59.9) 1802 (53.4) 3.35 (2.30–4.87) 693 (90.9) 2463 (80.9) 4.10 (2.79–6.03)
Possible 31 (3.7) 109 (3.2) 2.83 (1.59–5.02) 34 (4.5) 165 (5.4) 2.06 (1.19–3.54)
Each additional visitc   1.015 (1.006–1.024)   1.003 (1.001–1.006)
Pneumonia and influenza
No 739 (87.4) 2987 (88.5) Referent 476 (62.5) 2003 (65.8) Referent
Yes 93 (11.0) 333 (9.9) 1.14 (0.88–1.46) 252 (33.1) 925 (30.4) 1.16 (0.97–1.40)
Possible 14 (1.6) 54 (1.6) 1.04 (0.56–1.93) 34 (4.4) 115 (3.8) 1.25 (0.83–1.88)
Each additional visitc   1.040 (0.986–1.097)   1.008 (0.991–1.026)
Unspecified bronchitis
No 833 (98.5) 3319 (98.4) Referent 712 (93.4) 2808 (92.3) Referent
Yes 7 (0.8) 32 (0.9) 0.88 (0.39–1.98) 38 (5.0) 161 (5.3) 0.93 (0.65–1.35)
Possible 6 (0.7) 23 (0.7) 1.03 (0.42–2.56) 12 (1.6) 74 (2.4) 0.63 (0.34–1.18)
Each additional visitc   0.870 (0.569–1.330)   0.889 (0.763–1.036)
Intestinal infectious diseases
No 698 (82.5) 2846 (84.3) Referent 453 (59.4) 1959 (64.4) Referent
Yes 131 (15.5) 444 (13.2) 1.23 (0.98–1.54) 275 (36.1) 949 (31.2) 1.29 (1.08–1.54)
Possible 17 (2.0) 84 (2.5) 0.80 (0.45–1.40) 34 (4.5) 135 (4.4) 1.09 (0.73–1.61)
Each additional visitc   1.036 (0.960–1.117)   1.008 (0.976–1.041)
Conjunctivitis
No 727 (85.9) 2987 (88.5) Referent 438 (57.5) 1918 (63.0) Referent
Yes 103 (12.2) 348 (10.3) 1.24 (0.97–1.59) 289 (37.9) 1014 (33.3) 1.29 (1.07–1.54)
Possible 16 (1.9) 39 (1.2) 1.74 (0.94–3.21) 35 (4.6) 111 (3.7) 1.38 (0.92–2.06)
Each additional visitc   1.034 (0.914–1.169)   1.043 (1.002–1.085)
Perinatal infections
No 814 (96.2) 3265 (96.8) Referent 736 (96.6) 2948 (96.9) Referent
Yes 32 (3.8) 104 (3.1) 1.24 (0.82–1.86) 26 (3.4) 90 (2.9) 1.16 (0.74–1.82)
Possible 0 (0.0) 5 (0.2) – 0 (0.0) 5 (0.2) –
Each additional visitc   1.153 (0.970–1.370)   1.135 (0.959–1.344)
Any infectionsd
No 305 (36.0) 1429 (42.3) Referent 33 (4.3) 368 (12.1) Referent
Yes 510 (60.3) 1831 (54.3) 3.18 (2.17–4.66) 696 (91.4) 2503 (82.2) 3.90 (2.61–5.81)
Possible 31 (3.7) 114 (3.4) 2.67 (1.48–4.79) 33 (4.3) 172 (5.7) 1.83 (1.05–3.17)
Each additional visitc   1.014 (1.006–1.022)   1.003 (1.001–1.005)
Number of clinical visits for infection
0 visits 305 (37.4) 1429 (43.8) Referent 33 (4.5) 368 (12.8) Referent
1–5 visits 109 (13.4) 441 (13.6) 2.77 (1.81–4.26) 63 (8.6) 350 (12.2) 2.26 (1.38–3.69)
6–10 visits 101 (12.4) 369 (11.3) 3.27 (2.10–5.09) 56 (7.7) 191 (6.7) 4.33 (2.58–7.28)
>10 visits 300 (36.8) 1021 (31.3) 3.63 (2.40–5.48) 577 (79.2) 1962 (68.3) 4.78 (3.08–7.39)

aOR and 95% CI were calculated using conditional logistic regression with age, sex and the time of diagnosis (reference date for the controls) as the matching variables.

bSubjects diagnosed with leukaemia at age 2 years or younger, and the matched controls were excluded from analysis of allergy occurring >1 year before diagnosis.

cRisk for each additional physician visit for infection.

dAny infection includes otitis media, acute respiratory infections, pneumonia and influenza, unspecified bronchitis, intestinal infectious diseases, conjunctivitis and perinatal infections.

Table 3

The association between childhood AML and infections diagnosed during different time periods, Taiwan, 2000–08

Medically diagnosed infections Before 1 year of age

>1 year before the case’s diagnosis

Case Control OR (95% CI)a Caseb Controlb OR (95% CI)a
N = 193 N = 766 N = 155 N = 618
n (%) n (%) n (%) n (%)
Otitis media
No 178 (92.2) 710 (92.7) Referent 112 (72.3) 454 (73.5) Referent
Yes 14 (7.3) 55 (7.2) 0.99 (0.52–1.91) 42 (27.1) 155 (25.1) 1.11 (0.73–1.70)
Possible 1 (0.5) 1 (0.1) 4.00 (0.25–63.98) 1 (0.6) 9 (1.4) 0.45 (0.06–3.60)
Each additional visitc   0.898 (0.722–1.118)   0.987 (0.934–1.042)
Acute respiratory infections
No 77 (39.9) 361 (47.1) Referent 7 (4.5) 89 (14.4) Referent
Yes 112 (58.0) 380 (49.6) 6.61 (2.22–19.64) 137 (88.4) 480 (77.7) 5.78 (2.25–14.84)
Possible 4 (2.1) 25 (3.3) 2.49 (0.54–11.46) 11 (7.1) 49 (7.9) 2.82 (0.97–8.17)
Each additional visitc   1.012 (0.992–1.032)   1.003 (0.998–1.008)
Pneumonia and influenza
No 154 (79.8) 683 (89.2) Referent 84 (54.2) 420 (67.9) Referent
Yes 36 (18.6) 72 (9.4) 2.49 (1.55–4.01) 61 (39.4) 176 (28.5) 1.87 (1.24–2.82)
Possible 3 (1.6) 11 (1.4) 1.35 (0.36–5.09) 10 (6.4) 22 (3.6) 2.21 (1.00-4.90)
Each additional visitc   1.193 (1.064-1.338)   1.017 (0.981–1.054)
Unspecified bronchitis
No 190 (98.5) 754 (98.4) Referent 147 (94.8) 553 (89.5) Referent
Yes 3 (1.5) 7 (0.9) 1.71 (0.44–6.63) 7 (4.5) 42 (6.8) 0.61 (0.27–1.40)
Possible 0 (0.0) 5 (0.7) – 1 (0.7) 23 (3.7) 0.16 (0.02–1.21)
Each additional visitc   1.118 (0.410–3.048)   0.733 (0.469–1.147)
Intestinal infectious diseases
No 153(79.3) 659 (86.0) Referent 93 (60.0) 406 (65.7) Referent
Yes 34 (17.6) 94 (12.3) 1.75 (1.08–2.85) 53 (34.2) 178 (28.8) 1.37 (0.90–2.08)
Possible 6 (3.1) 13 (1.7) 2.15 (0.78–5.97) 9 (5.8) 34 (5.5) 1.13 (0.52–2.47)
Each additional visitc   1.121 (0.986–1.275)   1.043 (0.963–1.129)
Conjunctivitis
No 169 (87.6) 689 (90.0) Referent 88 (56.8) 370 (59.9) Referent
Yes 20 (10.3) 69 (9.0) 1.21(0.69–2.12) 58 (37.4) 217 (35.1) 1.15 (0.76–1.75)
Possible 4 (2.1) 8 (1.0) 2.18 (0.60–7.94) 9 (5.8) 31 (5.0) 1.23 (0.56–2.72)
Each additional visitc   0.926 (0.695–1.235)   1.049 (0.970–1.135)
Perinatal infections
No 186 (96.4) 743 (97.0) Referent 151 (97.4) 601 (97.2) Referent
Yes 7 (3.6) 23 (3.0) 1.22 (0.51–2.91) 4 (2.6) 17 (2.8) 0.94 (0.31–2.86)
Possible 0 (0.0) 0 (0.0) – 0 (0.0) 0 (0.0) –
Each additional visitc   1.055 (0.494–2.257)   0.838 (0.319–2.201)
Any infectionsd
No 77 (39.9) 357 (46.6) Referent 6 (3.9) 79 (12.8) Referent
Yes 112 (58.0) 386 (50.4) 5.99 (2.02–17.75) 137 (88.4) 492 (79.6) 5.31 (2.02–13.96)
Possible 4 (2.1) 23 (3.0) 2.66 (0.57–12.29) 12 (7.7) 47 (7.6) 3.22 (1.13–9.19)
Each additional visitc   1.014 (0.997-1.030)   1.002 (0.998-1.007)
Number of clinical visits for infection
0 visit 77 (40.8) 357 (48.0) Referent 6 (4.2) 79 (13.8) Referent
1–5 visits 18 (9.5) 93 (12.5) 3.48 (1.12–10.83) 14 (9.8) 73 (12.8) 2.77 (0.86–8.90)
6–10 visits 24 (12.7) 77 (10.4) 5.82 (1.91–17.75) 14 (9.8) 38 (6.7) 6.35 (1.86–21.67)
>10 visits 70 (37.0) 216 (29.1) 6.35 (2.18–18.53) 109 (76.2) 381 (66.7) 5.72 (1.89–17.27)
Medically diagnosed infections Before 1 year of age

>1 year before the case’s diagnosis

Case Control OR (95% CI)a Caseb Controlb OR (95% CI)a
N = 193 N = 766 N = 155 N = 618
n (%) n (%) n (%) n (%)
Otitis media
No 178 (92.2) 710 (92.7) Referent 112 (72.3) 454 (73.5) Referent
Yes 14 (7.3) 55 (7.2) 0.99 (0.52–1.91) 42 (27.1) 155 (25.1) 1.11 (0.73–1.70)
Possible 1 (0.5) 1 (0.1) 4.00 (0.25–63.98) 1 (0.6) 9 (1.4) 0.45 (0.06–3.60)
Each additional visitc   0.898 (0.722–1.118)   0.987 (0.934–1.042)
Acute respiratory infections
No 77 (39.9) 361 (47.1) Referent 7 (4.5) 89 (14.4) Referent
Yes 112 (58.0) 380 (49.6) 6.61 (2.22–19.64) 137 (88.4) 480 (77.7) 5.78 (2.25–14.84)
Possible 4 (2.1) 25 (3.3) 2.49 (0.54–11.46) 11 (7.1) 49 (7.9) 2.82 (0.97–8.17)
Each additional visitc   1.012 (0.992–1.032)   1.003 (0.998–1.008)
Pneumonia and influenza
No 154 (79.8) 683 (89.2) Referent 84 (54.2) 420 (67.9) Referent
Yes 36 (18.6) 72 (9.4) 2.49 (1.55–4.01) 61 (39.4) 176 (28.5) 1.87 (1.24–2.82)
Possible 3 (1.6) 11 (1.4) 1.35 (0.36–5.09) 10 (6.4) 22 (3.6) 2.21 (1.00-4.90)
Each additional visitc   1.193 (1.064-1.338)   1.017 (0.981–1.054)
Unspecified bronchitis
No 190 (98.5) 754 (98.4) Referent 147 (94.8) 553 (89.5) Referent
Yes 3 (1.5) 7 (0.9) 1.71 (0.44–6.63) 7 (4.5) 42 (6.8) 0.61 (0.27–1.40)
Possible 0 (0.0) 5 (0.7) – 1 (0.7) 23 (3.7) 0.16 (0.02–1.21)
Each additional visitc   1.118 (0.410–3.048)   0.733 (0.469–1.147)
Intestinal infectious diseases
No 153(79.3) 659 (86.0) Referent 93 (60.0) 406 (65.7) Referent
Yes 34 (17.6) 94 (12.3) 1.75 (1.08–2.85) 53 (34.2) 178 (28.8) 1.37 (0.90–2.08)
Possible 6 (3.1) 13 (1.7) 2.15 (0.78–5.97) 9 (5.8) 34 (5.5) 1.13 (0.52–2.47)
Each additional visitc   1.121 (0.986–1.275)   1.043 (0.963–1.129)
Conjunctivitis
No 169 (87.6) 689 (90.0) Referent 88 (56.8) 370 (59.9) Referent
Yes 20 (10.3) 69 (9.0) 1.21(0.69–2.12) 58 (37.4) 217 (35.1) 1.15 (0.76–1.75)
Possible 4 (2.1) 8 (1.0) 2.18 (0.60–7.94) 9 (5.8) 31 (5.0) 1.23 (0.56–2.72)
Each additional visitc   0.926 (0.695–1.235)   1.049 (0.970–1.135)
Perinatal infections
No 186 (96.4) 743 (97.0) Referent 151 (97.4) 601 (97.2) Referent
Yes 7 (3.6) 23 (3.0) 1.22 (0.51–2.91) 4 (2.6) 17 (2.8) 0.94 (0.31–2.86)
Possible 0 (0.0) 0 (0.0) – 0 (0.0) 0 (0.0) –
Each additional visitc   1.055 (0.494–2.257)   0.838 (0.319–2.201)
Any infectionsd
No 77 (39.9) 357 (46.6) Referent 6 (3.9) 79 (12.8) Referent
Yes 112 (58.0) 386 (50.4) 5.99 (2.02–17.75) 137 (88.4) 492 (79.6) 5.31 (2.02–13.96)
Possible 4 (2.1) 23 (3.0) 2.66 (0.57–12.29) 12 (7.7) 47 (7.6) 3.22 (1.13–9.19)
Each additional visitc   1.014 (0.997-1.030)   1.002 (0.998-1.007)
Number of clinical visits for infection
0 visit 77 (40.8) 357 (48.0) Referent 6 (4.2) 79 (13.8) Referent
1–5 visits 18 (9.5) 93 (12.5) 3.48 (1.12–10.83) 14 (9.8) 73 (12.8) 2.77 (0.86–8.90)
6–10 visits 24 (12.7) 77 (10.4) 5.82 (1.91–17.75) 14 (9.8) 38 (6.7) 6.35 (1.86–21.67)
>10 visits 70 (37.0) 216 (29.1) 6.35 (2.18–18.53) 109 (76.2) 381 (66.7) 5.72 (1.89–17.27)

aOR and 95% CI were calculated using conditional logistic regression with age, sex and the time of diagnosis (reference date for the controls) as the matching variables.

bSubjects diagnosed with leukaemia at age 2 years or younger and the matched controls were excluded from analysis of allergy occurring >1 year before diagnosis.

cRisk for each additional physician visit for infection.

dAny infection includes otitis media, acute respiratory infections, pneumonia and influenza, unspecified bronchitis, intestinal infectious diseases, conjunctivitis and perinatal infections.

The relationship between childhood ALL and infections did not differ by age (Table 4). In addition, urbanization levels and socio-economic status did not modify the association between infections and childhood ALL or AML (Supplementary Tables S1 and S2, available at IJE online).

Table 4

The association between childhood ALL and having any infections during different time periods by age groups, Taiwan, 2000–08

Having any infectiona by time periods (yes vs no) ALL

2–5.9 years 6–9.9 years
OR (95% CI)b OR (95% CI)b
Before 1 year of age 2.99 (1.86–4.79) 4.44 (1.52–13.02)
Interaction P-value = 0.51
>1 year before the case’s diagnosis 4.27 (2.54–7.16) 3.31 (1.63–6.76)
Interaction P-value = 0.57
Having any infectiona by time periods (yes vs no) ALL

2–5.9 years 6–9.9 years
OR (95% CI)b OR (95% CI)b
Before 1 year of age 2.99 (1.86–4.79) 4.44 (1.52–13.02)
Interaction P-value = 0.51
>1 year before the case’s diagnosis 4.27 (2.54–7.16) 3.31 (1.63–6.76)
Interaction P-value = 0.57

aAny infection includes otitis media, acute respiratory infections, pneumonia and influenza, unspecified bronchitis, intestinal infectious diseases, conjunctivitis and perinatal infections.

bOR and 95% CI were calculated using conditional logistic regression with age, sex and the time of diagnosis (reference date for the controls) as the matching variables.

## Discussion

In the current analysis based on computerized health insurance claims data, we observed a positive association between childhood leukaemia and childhood infections, which is consistent with one22 of the two previous studies that used medical records. The other record-based study reported a null association,24 whereas 710,12,13,15,17,21,26 of the 1110,12,13,1517,20,21,23,25,26 studies using parental report of the child’s infection history observed an inverse association.

The major strength of our study is that the infection data of the study subjects were obtained from computerized health insurance claims data, which are not subject to recall errors/bias. Parents are likely to recall the infection history of their children inaccurately, especially for those that occurred more remotely in the past. In a previous case–control study of childhood leukaemia, the authors observed that parents tended to under-report infectious events of their children, even for those with visits to general practitioners.30 Contrary to the popular notion that the case parents might ruminate more about the exposure history of their children compared with the control parents, the authors found that the case parents were more likely to under-report their children’s infectious events than the control parents.30 The inaccurate reporting by the parents could have easily produced the weak inverse association between childhood infection and childhood leukaemia observed by the previous studies.10,12,13,15,17,21,26 Opposite results have been observed between those generated with medical record data and parental report. Among children who attended regular social groups outside the home, an increased risk of childhood ALL (OR = 1.7, 95% CI 1.1–2.5) with at least one infectious episode in the first year of life was observed using medical record data, whereas data reported by the parents produced an inverse association (OR = 0.7, 95% CI 0.5–0.9).30 Another potential factor leading to the inverse association between childhood leukaemia and early childhood infection by parental report is that the date of leukaemia diagnosis, which is used as the reference date for the controls, may not be as clear for the control parents, so they may tend to report exposure after the reference date; this phenomenon has been observed with parental report of the child’s allergy status.31

The results of the current study suggest that children who develop leukaemia may have dysregulated immune function that is already present in early childhood, causing them to react strongly to infections and thereby requiring clinical attention. This is supported by a study that showed that among children with ALL, the number of clinically diagnosed infectious episodes increased with increasing indices of infectious exposure (birth order, regular social activity outside the home and deprivation level), whereas the infectious levels among controls did not differ by the indices of infectious exposure.30 Another study reported that children with ALL had a lower neonatal level of interleukin 10, a key regulator for the intensity and the duration of immune response to infections, compared with healthy children, suggesting that the dysregulated immune function of children with ALL could be present already at birth.32 The degree to which infections influence leukaemia risk in a positive rather than a protective manner may relate to the severity of response. In our current study, this risk induction appears to be shared by both AML and ALL diagnoses, whereas the protective risk modulation impact from exposure to infections (e.g. from other childhood contacts at day care) appears to be exclusively relevant for lymphocytic diagnoses.7,10 Strong reactions to infection include high fever, swelling and redness and recruitment of immune effector cells to the site of infection. The resulting cytokine ‘storm’ may result in inflammation, tissue and cell damage and release of excessive reactive metabolites that may contribute broadly to leukaemogenesis in the same manner that chronic inflammation is known to contribute to carcinogenesis.33,34 Rather than contributing only to immune modulation, normal infections that result in enough severity to warrant a physician visit may have the opposite impact and contribute to risk of AML and ALL. This contribution to risk of AML may be larger because of the capacity of myeloid cells to generate highly reactive oxygen metabolites.34

The results of the current study do not negate the consistent inverse association between childhood leukaemia and day care attendance, which supports Greaves’ ‘delayed infection’ hypothesis, as medically diagnosed infection and day care attendance are measuring different attributes. When studying the association between infection and childhood leukaemia, it is important to distinguish between strongly symptomatic infections and weakly or asymptomatic infections.35 Measures of medically diagnosed infections reflect the underlying immune dysregulation of children with leukaemia, whereas weakly or asymptomatic infections, which may be better assessed by proxy measures such as day care attendance and birth order, are likely the major players in modulating the normal immune development. For example, commensal bacteria in the intestine can modulate immune function by driving regulatory T-cell differentiation and promote the differentiation of TH1 cells to induce the normal shift of the TH2-dominant immune profile to the TH1-dominant immune profile during early childhood.36 Composition of the intestinal microbiota in early infancy is influenced by the mode of birth delivery (caesarean section vs vaginal birth), breastfeeding and having older siblings, which are all factors that have been associated with the risk of childhood leukaemia.812,3739

The results of the current study should be interpreted in the context of several limitations. It was not possible to analyse the data by molecular subtypes because the NHIRD does not have information on the molecular subtypes of childhood leukaemia. In addition, the NHIRD does not contain data on other infection-related factors, including birth order and day care attendance. Finally, only infectious events with visits to the health providers are recorded by the NHIRD. Given the universal health coverage and the easy access to health-care providers in Taiwan, the percentage of children with strongly symptomatic infections who did not visit a physician is likely low; however, the study could have missed minor infectious episodes that were not treated by physicians.

In summary, the current study suggests that children with leukaemia may have a dysregulated immune function present at an early age, resulting in more episodes of symptomatic infections compared with healthy controls. However, confounding by other infectious measures such as birth order and day care attendance could not be ruled out because of the lack of information. In addition, the results are only relevant to the medically diagnosed infections and not to asymptomatic infections or infections not requiring medical attention.

## Supplementary Data

Supplementary Data are available at IJE online.

## Funding

The National Science Council of Taiwan, R.O.C. (NSC 99-2314-B-400-004), the Establishment of Cancer Research System Excellence Program (DOH100-TD-C-111-004) and the Leukemia and Lymphoma Society of America (grant 6026-10 to J.L.W.).

## Acknowledgements

This study was based in part on data from the NHIRD provided by the Bureau of NHI, Department of Health, Taiwan, R.O.C., and managed by the National Health Research Institutes. The interpretation and conclusions contained herein do not represent those of the Bureau of National Health Insurance, Department of Health, Taiwan, R.O.C. or National Health Research Institutes.

Conflict of Interest: None declared.

KEY MESSAGES

• Parental report of child’s infections in early childhood may not be reliable and can produce inaccurate risk estimates for childhood leukaemia.

• The current analysis shows that medically diagnosed infections occurring before 1 year of age or >1 year before the diagnosis of leukaemia (reference date for the controls) are associated with an increased risk of childhood leukaemia.

• As the results of the current analysis were based on claims data from the NHIRD of Taiwan, which does not contain data on other infection-related factors, including birth order and day care attendance, possible confounding by these factors cannot be ruled out.

• Children with leukaemia may have a dysregulated immune function present at an early age, resulting in more episodes of symptomatic infections compared with healthy controls.

## References

1
Stiller
CA
Parkin
DM
Geographic and ethnic variations in the incidence of childhood cancer
Br Med Bull
,
1996
, vol.
52
(pg.
682
-
03
)
2
Parkin
DM
Kramarova
E
Draper
GJ
, et al.  .
International Incidence of Childhood Cancer Vol II
,
1998
Lyon, France
International Agency for Research on Cancer
3
Ries
LAG
Smith
MA
Gurney
JG
, et al.  .
Cancer Incidence and Survival Among Children and Adolescents: United States SEER Program 1975-1995

NIH Pub. No. 99-4649. Bethesda, MD: National Cancer Institute, SEER Program, 1999
4
Greaves
MF
Alexander
FE
An infectious etiology for common acute lymphoblastic leukemia in childhood?
Leukemia
,
1993
, vol.
7
(pg.
349
-
60
)
5
Greaves
M
Infection, immune responses and the aetiology of childhood leukaemia
Nat Rev Cancer
,
2006
, vol.
6
(pg.
193
-
203
)
6
Kinlen
LJ
Epidemiological evidence for an infective basis in childhood leukaemia
Br J Cancer
,
1995
, vol.
71
(pg.
1
-
5
)
7
Urayama
KY
Buffler
PA
Gallagher
ER
Ayoob
JM
Ma
X
A meta-analysis of the association between day-care attendance and childhood acute lymphoblastic leukaemia
Int J Epidemiol
,
2010
, vol.
39
(pg.
718
-
32
)
8
Dockerty
JD
Draper
G
Vincent
T
Rowan
SD
Bunch
KJ
Case-control study of parental age, parity and socioeconomic level in relation to childhood cancers
Int J Epidemiol
,
2001
, vol.
30
(pg.
1428
-
37
)
9
Hjalgrim
LL
Rostgaard
K
Hjalgrim
H
, et al.  .
Birth weight and risk for childhood leukemia in Denmark, Sweden, Norway, and Iceland
J Natl Cancer Inst
,
2004
, vol.
96
(pg.
1549
-
56
)
10
Urayama
KY
Ma
X
Selvin
S
, et al.  .
Early life exposure to infections and risk of childhood acute lymphoblastic leukemia
Int J Cancer
,
2011
, vol.
128
(pg.
1632
-
43
)
11
Schuz
J
Kaatsch
P
Kaletsch
U
Meinert
R
Michaelis
J
Association of childhood cancer with factors related to pregnancy and birth
Int J Epidemiol
,
1999
, vol.
28
(pg.
631
-
39
)
12
van Steensel-Moll
HA
Valkenburg
HA
van Zanen
GE
Childhood leukemia and infectious diseases in the first year of life: a register-based case-control study
Am J Epidemiol
,
1986
, vol.
124
(pg.
590
-
94
)
13
Jourdan-Da Silva
N
Perel
Y
Méchinaud
F
, et al.  .
Infectious diseases in the first year of life, perinatal characteristics and childhood acute leukaemia
Br J Cancer
,
2004
, vol.
90
(pg.
139
-
45
)
14
Ou
SX
Han
D
Severson
RK
, et al.  .
Birth characteristics, maternal reproductive history, hormone use during pregnancy, and risk of childhood acute lymphocytic leukemia by immunophenotype (United States)
Cancer Causes Control
,
2002
, vol.
13
(pg.
15
-
25
)
15
Perrillat
F
Clavel
J
Auclerc
MF
, et al.  .
Day-care, early common infections and childhood acute leukaemia: a multicentre French case-control study
Br J Cancer
,
2002
, vol.
86
(pg.
1064
-
69
)
16
Dockerty
JD
Skegg
DC
Elwood
JM
Herbison
GP
Becroft
DM
Lewis
ME
Infections, vaccinations, and the risk of childhood leukaemia
Br J Cancer
,
1999
, vol.
80
(pg.
1483
-
89
)
17
Neglia
JP
Linet
MS
Shu
XO
, et al.  .
Patterns of infection and day care utilization and risk of childhood acute lymphoblastic leukaemia
Br J Cancer
,
2000
, vol.
82
(pg.
234
-
40
)
18
Reynolds
P
Von Behren
J
Elkin
EP
Birth characteristics and leukemia in young children
Am J Epidemiol
,
2002
, vol.
155
(pg.
603
-
13
)
19
Rosenbaum
PF
Buck
GM
Brecher
ML
Early child-care and preschool experiences and the risk of childhood acute lymphoblastic leukemia
Am J Epidemiol
,
2000
, vol.
152
(pg.
1136
-
44
)
20
Schuz
J
Kaletsch
U
Meinert
R
Kaatsch
P
Michaelis
J
Association of childhood leukaemia with factors related to the immune system
Br J Cancer
,
1999
, vol.
80
(pg.
585
-
90
)
21
Chan
LC
Lam
TH
Li
CK
, et al.  .
Is the timing of exposure to infection a major determinant of acute lymphoblastic leukaemia in Hong Kong?
Paediatr Perinat Epidemiol
,
2002
, vol.
16
(pg.
154
-
65
)
22
Roman
E
Simpson
J
Ansell
P
, et al.  .
Childhood acute lymphoblastic leukemia and infections in the first year of life: a report from the United Kingdom Childhood Cancer Study
Am J Epidemiol
,
2007
, vol.
165
(pg.
496
-
504
)
23
Rosenbaum
PF
Buck
GM
Brecher
ML
Allergy and infectious disease histories and the risk of childhood acute lymphoblastic leukaemia
Paediatr Perinat Epidemiol
,
2005
, vol.
19
(pg.
152
-
64
)
24
Cardwell
CR
McKinney
PA
Patterson
CC
Murray
LJ
Infections in early life and childhood leukaemia risk: a UK case-control study of general practitioner records
Br J Cancer
,
2008
, vol.
99
(pg.
1529
-
33
)
25
MacArthur
AC
McBride
ML
Spinelli
JJ
Tamaro
S
Gallagher
RP
Theriault
GP
Risk of childhood leukemia associated with vaccination, infection, and medication use in childhood: the Cross-Canada Childhood Leukemia Study
Am J Epidemiol
,
2008
, vol.
167
(pg.
598
-
606
)
26
Rudant
J
Orsi
L
Menegaux
F
, et al.  .
Childhood acute leukemia, early common infections, and allergy: the ESCALE Study
Am J Epidemiol
,
2010
, vol.
172
(pg.
1015
-
27
)
27
Bureau of National Health Insurance, Department of Health
National Health Insurance in Taiwan 2010

28
Bureau of National Health Insurance
National Health Insurance Research Database

http://w3.nhri.org.tw/nhird//brief_02.htm# (8 June 2011, date last accessed)
29
Liu
CY
Hung
YT
Chuang
YL
, et al.  .
Incorporating development stratification of Taiwan townships into sampling design of large scale health interview survey
J Healthc Manag
,
2006
, vol.
4
(pg.
1
-
22
)
30
Simpson
J
Smith
A
Ansell
P
Roman
E
Childhood leukaemia and infectious exposure: a report from the United Kingdom Childhood Cancer Study (UKCCS)
Eur J Cancer
,
2007
, vol.
43
(pg.
2396
-
403
)
31
Schuz
J
Morgan
G
Bohler
E
Kaatsch
P
Michaelis
J
Atopic disease and childhood acute lymphoblastic leukemia
Int J Cancer
,
2003
, vol.
105
(pg.
255
-
60
)
32
Chang
JS
Zhou
M
Buffler
PA
Chokkalingam
AP
Metayer
C
Wiemels
JL
Profound deficit of IL10 at birth in children who develop childhood acute lymphoblastic leukemia
Cancer Epidemiol Biomarkers Prev
,
2011
, vol.
20
(pg.
1736
-
40
)
33
de Visser
KE
Eichten
A
Coussens
LM
Paradoxical roles of the immune system during cancer development
Nat Rev Cancer
,
2006
, vol.
6
(pg.
24
-
37
)
34
Lonkar
P
Dedon
PC
Reactive species and DNA damage in chronic inflammation: reconciling chemical mechanisms and biological fates
Int J Cancer
,
2011
, vol.
128
(pg.
1999
-
2009
)
35
Greaves
M
Buffler
PA
Infections in early life and risk of childhood ALL
Br J Cancer
,
2009
, vol.
100
pg.
863

36
McLoughlin
RM
Mills
KH
Influence of gastrointestinal commensal bacteria on the immune responses that mediate allergy and asthma
J Allergy Clin Immunol
,
2011
, vol.
127
(pg.
1097
-
107; quiz 108–09
)
37
Penders
J
Thijs
C
Vink
C
, et al.  .
Factors influencing the composition of the intestinal microbiota in early infancy
Pediatrics
,
2006
, vol.
118
(pg.
511
-
21
)
38
Kwan
ML
Buffler
PA
Abrams
B
Kiley
VA
Breastfeeding and the risk of childhood leukemia: a meta-analysis
Public Health Rep
,
2004
, vol.
119
(pg.
521
-
35
)
39
Kaye
SA
Robison
LL
Smithson
WA
Gunderson
P
King
FL
Neglia
JP
Maternal reproductive history and birth characteristics in childhood acute lymphoblastic leukemia
Cancer
,
1991
, vol.
68
(pg.
1351
-
55
)