25 2 In PubMed, combinations of MeSH Terms and title/abstract were used. Embase and Cochrane had adapted search strategies based on the PubMed search strategy. Grey literature was not included. The full strategies for each database are presented in Appendix A. Study selection All records were imported in reference manager Endnote X9 (Clarivate Analytics 2013). After manual removal of duplicates, the eligibility assessment based on title and abstract was independently conducted by two authors (JV and AP). Afterwards, disagreements between reviewers were resolved by discussion. Full-text articles were obtained and independently assessed on inclusion and exclusion criteria. Conflicts between reviewers were resolved in all cases. If needed, a third author (CS) was available to resolve any disagreements. Data collection process, data items and summary measures A data extraction sheet was used, which included clinical and demographic characteristics of patients as well as study related details relevant to our review. One author (AP) extracted the data from included studies and the other author (JV) verified the extracted data. Any seemingly incorrect data were discussed. If needed, a third author (CS) was available to resolve any disagreements. The following information was extracted from each included study: (1) study characteristics (study design, number of included patients); (2) patients characteristics (sex, age, tumor site, tumor stage); (3) type of cancer treatment; (4) assessment (evaluation frequency, assessment moment, follow-up); and (5) UW-QoL outcome regarding masticatory ability (mean with standard deviation (SD) or standard error (SE)). When authors clearly defined different subgroups in their study, composed data as well as specified data were extracted. Risk of bias in individual studies The risk of bias was evaluated with the validated methodological index for non-randomized studies (MINORS).22 This instrument contains eight items for non-comparative studies and four additional items for comparative studies. The grading of each item is done by appointing one of three grades: not reported (0); reported but inaccurate (1); reported and adequate (2). The studies were independently assessed by two authors (JV and AP). Criteria for scoring each item were discussed by the two reviewers before as well as during the assessment of the publications. Any disagreements were resolved by discussion. A third