Does Timeliness of DTaP-IPV-Hib Vaccination Affect Development of Atopic Dermatitis Before 1 Year of Age?

It has been found that the non-live vaccine against Diphtheria, Tetanus, and Pertussis (DTP) in addition to its disease specific effects may have so called non-specific effects with the potential to affect resistance towards vaccine unrelated pathogens, resulting in excess mortality and morbidity(Aaby et al., 2014). These immunomodifying effects may affect development of atopic dermatitis through increased immunological sensitivity(Nilsson, Gruber, Granstrom, Bjorksten, \& Kjellman, 1998). A recent study from Australia found that delayed vaccination with the first dose of Diphtheria, Tetanus, and acellular Pertussis(DTaP)-containing vaccine is associated with reduced eczema (aOR: 0.57; 95% CI: 0.34-0.97, P = 0.04) and reduced use of medication against atopic dermatitis (aOR: 0.45; 95% CI: 0.24-0.83, P = 0.01)(Kiraly et al., 2016). The present study aims to test the above finding and investigate the effect of timeliness of vaccination with DTaP-IPV-Hib (with or without PCV) on development of atopic dermatitis before 1 year of age among Danish children born between 1997 and 2012. From 1997 to 2012 children in Denmark were scheduled to receive the 1st, 2nd, and 3rd dose of DTaP-IPV-Hib at 3, 5, and 12 months respectively.

Primary investigation:

Research question 1.a Is delayed vaccination with the first dose of DTaP-IPV-Hib (vaccination at or after 4 months of age) associated with a reduced risk of development of atopic dermatitis from 4 months and until but not including 12 months of age compared with receiving the first dose of DTaP-IPV-Hib before 4 months of age?

Secondary investigations:

Research question 1.b Among children with a timely 1st dose of DTaP-IPV-Hib (vaccination before 4 months of age), is delayed vaccination with the second dose of DTaP-IPV-Hib (vaccination at or after 6 months of age) associated with a reduced risk of development of atopic dermatitis from 6 months and until but not including 12 months of age compared with receiving the second dose of DTaP-IPV-Hib before 6 months of age?

Research question 2. Is receipt of DTaP-IPV-Hib associated with development of atopic dermatitis from 3 months and until but not including 8 months of age compared with being unvaccinated?

Data assessment Information on development of atopic dermatitis and vaccination status is assessed at baseline (which is 4 months of age (research question 1.a.),6 months of age (research question 1.b.) and 3 months of age (research question 2.)) and through follow up (until 1 year of age). Confounder information is assessed at 3 months of age for all analyses.

Development of atopic dermatitis is assessed according to an algorithm described in detail under "9. Outcome measures", which uses information on prescriptions and hospitalizations up until one year after the last age of outcome assessment (1 year of age in this study) to confirm a case. Children must therefore be alive and living in Denmark until 2 years of age representative of age at end of follow-up plus 1 year.

The Danish national registers contains individual-level information on a broad range of health and social factors. Every child born in Denmark receives a unique personal identification number (ID) at birth, which follows the individual through life. Using the ID it is possible to link information from the different registers at an individual level.

Confounders:

The investigators will seek to retain information on potential confounders from the Danish registers. Potential confounders will be assessed at 3 months of age and included in the adjusted analyses.

Information on sex, season of birth, death, emigration, parental identity, sibling's identity and exact address will be obtained from the Danish Civil Registration System. The investigators will use information on parental identity to link every child to its parents and obtain information on parents' origin of birth, single parenthood and the number of children in the household. The investigators will furthermore derive information on population density based on information on the municipality in which the child lives. Information on parental and sibling's identity will be used to link information on family history of atopic dermatitis to the child. Atopic dermatitis is categorized according to the algorithm described in detail under "9. Outcome measures".

The variables; caesarean section, preterm birth, birthweight, and maternal smoking during pregnancy is obtained from the Danish Medical Birth Register, which contains information on all live and stillbirths in Denmark. Antibiotic use is recorded in the Danish National Prescription Register. Information on chronic disease is obtained from in the Danish national patient registry. Household income is obtained from the Danish registers on personal income and transfer payment.

Maternal highest education is obtained from the Danish education registers.

The dataset is expected to contain close to complete information on confounders. Hence, it is expected that children with missing confounder information constitute an ignorable proportion of the complete study population, wherefore the investigators aim to conduct complete case analyses only.

Statistical model for research question 1.a +1.b; investigation of the effect of timing of vaccination.

For investigation of the effect of timing of the first and second dose of DTaP-IPV-Hib with or without PCV, a Binary regression model will be used to estimate the adjusted relative risk of developing atopic dermatitis from baseline (age for categorized timely vaccination) until one year of age, among delayed vaccinated compared to timely vaccinated children. The analysis will include all abovementioned potential confounders. Absolute risk differences will be reported if relevant.

Intended subgroup and sensitivity analysis for the primary investigation (research question 1.a)

The following presents a priory identified relevant subgroup and sensitivity analyses. If relevant, additional analysis will be included with the aim to pursue potential tendencies revealed in the data analysis.

1. Sex differential effect Data will be analyzed for effect modification by sex because of evidence of sex-differential non-specific effects of vaccines (Aaby et al., 2014) and also sex-differential atopic sensitization associated with delayed vaccination (Kiraly et al., 2016).
2. Further subgroup analyses

• Effect modification by PCV vaccination. PCV was enrolled into the Danish vaccination schedule during the years of this investigation. Hence, it will be investigated if potential NSEs of DTaP-IPV-Hib vary according to receipt of PCV.
• Exploratory analysis of effect modification.
• Subgroup analysis, which only includes first born children; Parents of children with older siblings may be more familiar with AD, which may result in a less frequent health care seeking behavior. Parents of children with older siblings, who have had AD themselves, may further have prescription medicines in the home from treatment of prior cases of AD among the siblings. Hence, cases of AD categorized based on disease specific prescriptions may be misclassified if the parents do not get a new prescription for that child.

1. Sensitivity analyses

• Sensitivity analysis to investigate if the results are affected by inclusion of children with no DTaP-IPV-Hib before 1 year of age.
• Investigation of reverse causation: A Cox regression will be conducted to investigate if presence of atopic dermatitis is associated with subsequent delayed vaccination.
• Inclusion of negative control outcome/exposure; In similar studies, it has been suspected that there may be a risk of ascertainment bias, whereby parents of children who are delayed vaccinated, represent parents with a general lower health care seeking behavior, which could result in lower levels of AD diagnosis among their children. In order to accommodate this potential unmeasured confounder, the investigators will seek to identify a relevant negative control exposure or outcome, which will help indicate the strength of a potential interference from this unmeasured confounder.

Statistical model for investigation of research question 2; investigation of the effect of being vaccinated with DTaP-IPV-Hib with or without PCV compared with being non-vaccinated.

For investigation of the effect of being vaccinated with DTaP-IPV-Hib compared with being non-vaccinated a Cox proportional hazards regression with age as the underlying time will be used to estimate the adjusted HR of developing atopic dermatitis among children vaccinated with DTaP-IPV-Hib compared with unvaccinated children. Vaccination status will be assessed from time of scheduled vaccination (3 months of age) and set as a time dependent variable with the categories unvaccinated and vaccinated.

As vaccination status is included as a time dependent variable, the group of unvaccinated children will inherently become smaller with time as they get vaccinated. In order to assure sufficient group sizes, children will only be followed until 8 months of age.

The model will include all identified potential confounding factors listed above. All co-variates will be included as fixed variables and assessed at baseline (3 months of age).

Proportional hazard assumption will be examined.

Sensitivity analysis for the Cox regression

• Delayed diagnosis; Atopic dermatitis does not become apparent until the child has scratched the affected skin. Furthermore, it may take weeks to months before the parents become aware that this is not something which is passing, and bring the child to a doctor. Therefore, it is very likely that actual onset of AD is weeks/months prior to date of diagnosis by the healthcare professionals or receipt of disease specific prescriptions. Due to this delay, there is a risk of differential misclassification, whereby cases of AD may be falsely ascribed to the vaccinated exposure group. A sensitivity analysis will account for this delay by setting the date for onset of AD prior to onset according to the AD algorithm. Primary health care facilities will be contacted in order to gain expert insight on approximate delay from onset of symptoms to receipt of prescription medicines or referral to a hospital.