3, Table 2). Evidence on indirect impact in low-coverage (<70%) settings

is mixed, with significant impact seen in some populations and not others. Data on indirect effect of PCV on AT–IPD showed a trend toward increasing impact with time (median decrease: 33%; IQR: 7–42%), though CP-690550 clinical trial with lower overall impact compared to that on VT-IPD (Appendix B.3, Table 3). This impact on AT-IPD was observed in all non-target age-groups (Fig. 5) and is also noted in pneumococcal pneumonia [10] and [29]. Data from mixed target and non-target groups show a greater decrease in VT-IPD rates than that in pure non-targeted groups, reflecting a mix of direct and indirect effect (Appendix B.3, Table 4). However, studies with 1-dose coverage data suggest a vaccine impact on VT-IPD that cannot be entirely accounted for by direct effect. Data were available for six unique populations: Australian aboriginals, Alaska Natives, American Indians, Gambians, Israelis and Portuguese E7080 (Appendix B.3, Table 5). Studies in children were primarily RCTs; those in adults were primarily observational. The median decrease

in VT-carriage prevalence (among either the study sample or, rarely, the subset who were carriers of any pneumococcal strain) was 77% (IQR 64–80%). Data points did not span a sufficient time range to evaluate time-related trends. The majority of carriage data is drawn from high-risk populations. Few additional supporting data points were identified for NP carriage. Supporting data are listed for pre- vs. post-introduction all-type NP in non-target groups and pre- vs. post-introduction VT-carriage in mixed groups in Appendix B.3, Tables 6 and 7; a discussion is provided in Appendix B.4. A relevant data point not eligible for inclusion due to publication

date comes from an observational study including Native American adults shortly after PCV introduction Calpain (2001–2002) and subsequently (2006–2008), finding a relative decrease of 97.5% and an absolute reduction of 4.0% in VT-NP [46]. Most individual data points were categorized as low or very-low quality by GRADE criteria because nearly all data were from observational studies, and over half the primary evidence sources were further downgraded for including only high-risk populations, but few for methodological issues (Appendix B.5). While GRADE methodology categorizes observational studies as ‘low quality’, the GRADE system was designed to assess individual patient treatments, not to assess public health benefit. Furthermore, only observational, or community randomized studies can assess population-level post-introduction effects. An additional 14 studies published after the PCV Dosing Landscape Review search met primary evidence inclusion criteria.