eTable 2a. Instrument for evaluating the quality of studies included in the systematic review.

Major risk of bias domains*	Risk	Criteria	Hints/ notes
1. Recruitment procedure & follow-up (in cohort studies): For cohort studies HINT: We are looking for selection bias: – Was the cohort representative of a defined population? # – Was everybody included who should have been included? # –If response on day care center level is slightly <50% but does not indicate selection bias, it will be listed as a demerit in extraction table. Preliminary ruling: –If the cohort recruitment is based on a convenient/self-reported sampling OR if response is <10% or if response was not reported/calculable, the study will be excluded from analysis.	low	Cohort recruitment was acceptable.# Baseline response on both educator and day care center level is acceptable (50% or more) OR is <50% and >30%, but substantial differential selection could be excluded (e. g. by a non-responder analysis). Loss to follow-up is below 20% in total and not different between the two groups (up to 10% difference).*
	high	Cohort recruitment was not acceptable.# Response not reported/ not calculable. Total loss to follow-up is larger than acceptable (20% or more)* OR drop out differs between the groups by more than 10%* OR the reasons for drop out considerably differ between exposed and non-exposed groups.*
For case-control studies HINT: We are looking for selection bias: – Were the cases and control subjects representative of the same defined population (“study base”; geographically and/or temporally)? # – Was there an established reliable system for selecting all the cases? # – The same exclusion criteria are used for both cases and controls. # – Comparison is made between participants and non-participants to establish their similarities or differences. # – If response on day care center level is slightly <50% but does not indicate selection bias, it will be listed as a demerit in extraction table. Preliminary ruling: – If the recruitment is based on a convenient/self-reported sampling OR if response is <10% or if response was not reported/calculable, the study will be excluded from analysis.	low	Case selection and recruitment was acceptable.# Control subjects’ selection and recruitment was acceptable.# Non-response was less than 50% for cases and/or control subjects OR it was >50% and <70%, but substantial differential selection of cases and control subjects could be excluded (e.g. by a non-responder analysis)*
	high	Case selection and recruitment was not acceptable.# Control subjects’ selection and recruitment was not acceptable.# Non-response was >70% for cases or control subjects OR it was >50% and<70%, but substantial differential selection of cases and control subjects could not be excluded.* Response not reported/ not calculable
For cross-sectional studies HINT: We are looking for selection bias: – Was the study population representative of a defined population? # – Was everybody included who should have been included? # – If response on day care center level is slightly <50% but does not indicate selection bias, it will be listed as a demerit in extraction table. Preliminary ruling: – If the recruitment is based on a convenient/ self-reported sampling OR if response is <10% or if response was not reported/calculable, the study will be excluded from analysis.	low	Recruitment of the study population was acceptable.# Non-response was less than 50% OR it was >50% and <70%, but substantial differential selection of the study population could be excluded (e.g. by a non-responder analysis).*
	high	Recruitment of the study population was not acceptable.# Non-response was >70% OR it was >50% and <70%, but substantial differential selection of the study population could not be excluded.* Response not reported/ not calculable.
2. Exposure definition and ‧measurement	low	Exposure definition included at least basic job characteristics (e.g., job tasks, length of employment). Exposure was accurately measured to minimize bias. Adequate comparison group of non-exposed workers (e.g. office workers) included.
	high	Exposure does not cover basic job characteristics. Exposure was not accurately measured.# Different methods were used to measure exposure in different groups/cases and control subjects (in case-control studies). No adequate comparison group of non-exposed workers included (only for outcome 1b)
	unclear	Not reported.
3.1a Outcome “infection rate”. Source and validation	low	Outcome was accurately/ objectively measured to minimize bias (positive serology, medical diagnosis).# Measurement methods were similar in the different groups.#
	high	Outcome was not accurately or subjectively measured (self-reported).# Measurement methods were different in the groups.#
	unclear	Not reported.
3.1b Outcome “risk of infection”. Source and validation	low	Outcome was accurately/ objectively measured to minimize bias (e.g. positive serology, medical diagnosis).# Measurement methods were similar in the different groups.#
	high	Outcome was not accurately or subjectively measured (e.g. self-reported).# Measurement methods were different in the groups.#
	unclear	Not reported.
3.2 Outcome “Immunization coverage of the nursery-school teachers”. Source and validation Only applicable to vaccine-preventable diseases.	low	Outcome was accurately/ objectively measured to minimize bias (e.g. checked by certificates of vaccination, use of validated instruments).# Measurement methods were similar in the different groups.#
	high	Outcome was not accurately or subjectively measured (e.g. self-reported).# Measurement methods were different in the groups.#
	unclear	Not reported.
3.3 Outcome “Immunity status of the nursery-school teachers”. Source and validation	low	Outcome was accurately/ objectively measured to minimize bias (e.g. positive serology).# Measurement methods were similar in the different groups.#
	high	Outcome was not accurately or subjectively measured.# Measurement methods were different in the groups.#
	unclear	Not reported.
4. Confounding and effect modification HINT: If the immunity status of the children in care is not being considered, it will be listed as a demerit in extraction table.	low	If risk estimators were calculated, major confounding factors (at least age, sex, SES and for vaccine-preventable diseases, immunization status of child care workers) & effect modifiers were considered. If only prevalence or incidence was assessed, at least sex, age and immunization coverage (at least mean values for the study population) are described.
	high	Major confounding factors or effect modifiers were not considered.
	unclear	Not reported.
5. Analysis method: methods to reduce research specific bias HINT: If the prevalence of serology is very high, we will not accept Prevalence Odds Ratios as adequate.	low	Authors used adequate statistical models to reduce bias (e.g., standardization, matching, adjustment in multivariate model, stratification, propensity scoring). For prevalences, matching/stratification may not be required as long as a good description of the age structure and immunization status of the population is given.
	high	Authors did not use adequate statistical models to reduce bias.
	unclear	□
6. Chronology	low	Incident diseases were included.# Temporal relation may be established (exposure precedes the outcome).# Negative serology known at baseline (career entry, baseline of study) AND was accurately/ objectively measured. For outcomes 2 and 3, cross-sectional studies are appropriate.
	high	Prevalent diseases were included OR prevalent diseases of baseline were not excluded (in cohort studies).# Temporal relation cannot be established. Serology is unknown at baseline. Cross-sectional studies without basic information about temporal course (not applicable to outcomes 2 or 3)
	unclear	Not reported.
7. Funding	low	Grant/ non-profit-organizations* Study was clearly not affected by sponsors.*
	high	Sponsoring organization participated in data analysis. Study was probably affected by sponsors.
	unclear	Industry, combined industry+grant*, unclear if study was affected by sponsors. Not reported.
8. Conflict of interest	low	Reported not having conflict of interest or clear from report/ communication that study was not affected by author(s) affiliation.*
	high	Conflict of interest exists (at least one author).*
	unclear	Not reported.

*according to Ijaz et al. (2013), with modifications; # SIGN/CASP