ca‐Spackova 2010.

Study characteristics
Methods	Retrospective cohort ‐ local health authorities throughout Germany were encouraged to report varicella outbreaks to the Robert Koch Institute on a voluntary basis. Outbreaks were confirmed by public health professionals. At site visits of day‐care centres (DCC), the authors requested self‐administered questionnaires including varicella history and demographic characteristics from the parents of all children. Furthermore, the authors reviewed children’s vaccination records, which are filled in by the healthcare providers who administer the respective vaccine. Besides information on date of injections and vaccine brands, which the authors collected for all varicella vaccinations, the records also contain the lot numbers of the vaccines. Information regarding general characteristics of the respective DCC (number of children and staff present during the outbreak, number of groups in DCC, joint facilities, etc.) was requested. To protect personal information, study identification numbers were used. A reminding letter was sent to non‐responders to ensure maximum participation. Each outbreak investigation was closed as soon as no further case of varicella had occurred for 42 days (twice the maximal incubation period) after rash onset in the last case. The authors also searched for cases in the 42‐day period before disease onset in the index case to ensure that all outbreak‐related cases were included.
Participants	A case was defined as a child attending 1 of the investigated DCC at the time of the respective outbreak with acute onset of clinical varicella symptoms (maculo‐papulo‐vesicular rash with no other apparent cause) as reported by treating physician or parents.
Interventions	Varilrix 1 dose, Priorix‐Tetra 1 dose and 2 doses, Varivax 1 dose
Outcomes	Varicella was classified clinically as mild (< 50 skin lesions), moderate (≥ 50 skin lesions), or severe (any hospitalised case). Breakthrough varicella (BV) was defined as varicella with rash onset > 42 days after vaccination.
Funding Source	Government
Notes	Potential limitations: case definition, case finding, vaccination status ascertainment, and comparability of vaccinated and unvaccinated regarding exposure to the disease during the study period. The degree of exposure to infection and population susceptibility also influences VE estimates. (1) Exclusion criteria to ensure that only susceptible and vaccinated children were included in VE analyses and that vaccination status did not change during the outbreak. (2) All children under investigation had an equal chance of disease exposure. (3) Vaccination status was verified directly from vaccination records. Information bias might have been present if some parts of the questionnaire were not fully understood or remembered (e.g. duration of skin lesions, previous history of varicella, etc.) by the parents, also if the parent would not recognise mild BV. (1) The authors have considered parental case reporting to be reliable. (2) Additionally, 93% of cases in VE analysis were confirmed by a physician. (3) Each DCC was followed actively until outbreaks, all relevant cases were captured. (4) Both information on disease and vaccination status together was available only in 52% of children, and VE, after exclusions, was calculated only amongst 33% of all children (but amongst all responders who were eligible for VE calculation). (5) Responders (providing either vaccination record or questionnaire) and non‐responders differed significantly by age but not by sex. (6) The failure to demonstrate statistically significant differences regarding brand‐specific VE may be due to sample size. (7) The small number of children with BV and the short time intervals since the last dose of vaccination (up to 4.6 years) limited our ability to explore effects of time since vaccination on BV. (8) Some mild BV cases could have been missed as they might not have been recognised by parents, and thus VE might have been overestimated.
Risk of bias
Bias	Authors' judgement	Support for judgement
PCS/RCS ‐ exposed cohort selection	Low risk	Adequate ‐ representative of the exposed
PCS/RCS ‐ non‐exposed cohort selection	Low risk	Adequate ‐ drawn from the same community
PCS/RCS ‐ comparability	Low risk	Adequate ‐ homogeneous age
PCS/RCS ‐ assessment of outcome	Low risk	Adequate ‐ confirmed by physicians
Summary Risk of Bias assessment	Low risk	Plausible bias is unlikely to have seriously altered the results.