Sara van den Berg

62 Chapter 3 Study Selection (max 4*) Comparability (max 2*) Outcome (max 3*) Overall quality Turner et al, 2014 **** ** *** + Den Elzen et al, 2011 **** ** *** + Derhovanessian et al, 2012 **** ** *** + Nielsen et al, 2015 *** - * - Furman et al, 2015 – study 1 **** ** *** + Furman et al, 2015 – study 2 **** ** *** + Furman et al, 2015 – study 3 *** ** ** +/- Haq et al, 2016 **** ** *** + McElhaney et al, 2015 *** ** *** + Frasca et al, 2015 **** ** *** + Trzonkowski et al, 2003 **** ** ** + Strindhall et al, 2015 *** ** *** + Reed et al, 2016 **** ** ** + Wald et al, 2013 **** ** *** + Arias et al, 2013 *** ** * - Moro-Garcia et al, 2011 *** * * - Guidi et al, 2014 *** - *** - Table 2. Risk of bias and study quality of studies included for systematic assessment Risk of bias was analysed based on the Newcastle Ottawa scale for cohort studies. According to these guidelines, studies were awarded with “stars” for high quality choices in three categories: “selection of cohorts” (max 4*), “comparability of cohorts” (max 2*) and “assessment of outcome” (max 3*). Based on all the acquired information, a study could acquire a maximum of 9 stars and the overall quality of the study was rated as high (+) (≥8 stars), intermediate (+/-) (7 stars) or low (-) (≤ 6 stars) Non-systematic summary of conclusions reported in the different studies First, we summarized the conclusions on a possible effect of CMV infection on the antibody response to influenza vaccination reported in the different studies that we included ( Supplementary Table 3 and Figure 3) for young (<60 years of age) or old (>60 years of age) adults. In young individuals, two studies reported no effect, three a negative effect and four a positive effect of CMV-seropositivity. In old individuals, nine studies reported no effect, six a negative effect and one a positive effect of CMV-seropositivity on the humoral influenza vaccination response. Overall, more studies investigated old individuals than young individuals and a positive effect of CMV-seropositivity was mainly reported in young individuals. When reviewing the results based on strain-type or influenza antibody outcome variable, no clear conclusion could be drawn (data not shown). Assessment of the statistical methods performed by the studies showed that the majority of the studies used appropriate statistics, but different methods were applied for the statistical testing of influenza antibody data (legend Figure 3 ). Normalization of HI data by log transformation and parametrical testing to compare the geometric titer is preferred [24, 46]. However, some studies performed non-parametric testing on raw antibody data to compare the median, which can be different from the geometric mean, especially in cases where the raw data is not natural log-distributed. In addition, influenza antibody outcome variables differed greatly between studies ( Table 3 ). This hampers direct comparison of the results of the different studies in literature and a systematic comparison of the studies is necessary.