- Inclusion and exclusion criteria
The following inclusion criteria were used to optimize selection of appropriate articles: articles needed to (1) be written in either English or Chinese; (2) explore the concept of awareness of disease status among cancer patients; (3) explore the impact of disease awareness on patients’ quality of life; (4) be randomized controlled studies, cohort studies, or case control studies. The following exclusion criteria were used: (1) the article was a conference abstract; (2) the full text was unavailable.
- Patient and public involvement
No patients were directly involved in this study.
- Literature retrieval and screening
We searched the following databases, PubMed, CENTRAL (Cochrane Central Register of Controlled Trials), PsycINFO, WEB OF SCIENCE, Embase, CBM (Chinese Biomedical Literature database), WANFANG database (Chinese Medicine Premier), and CNKI (China National Knowledge Infrastructure). The terms used were: neoplasm, cancer, tumor, carcinoma, disclosure, truth telling, breaking bad news, knowledge, knowing, awareness, quality of life, and QOL (Attached file shows the search strategy using PubMed as an example). Reference lists of obtained articles were hand searched and authors were contacted if articles couldn’t be easily obtained the articles. Pairs of reviewers independently screened the literature and the third reviewer resolved any disagreements. The systematic review was registered in 2015 with PROSPERO registration number CRD42017060073. A complementary search using the above terms was performed in February 2018.
- Data extraction and management
Pairs of reviewers independently extracted the following data from included studies: first author, publication year, country, journal, the setting where the research was carried out, the time when the study began and ended, the definition of exposure in the research, study design, financial support, conflicts of interests, patients’ characteristics, and quality of life. The third reviewer resolved any disagreements.
- Primary and secondary outcome measures
The included studies used self-reported participant measures of QoL as primary or secondary end points.
General quality of life;
1) QoL domains:
- physical capability (e.g. ability to perform self-care activities, mobility, and physical activities);
- social capability (e.g. ability to perform work or household responsibilities and social interactions);
- role function (e.g. ability to perform in daily life, amusement, and hobbies);
- emotional wellbeing (e.g. levels of sadness, anxiety, depression, and/or negative affects);
- cognitive capacity (e.g. ability to focus attention and form/retain memories);
- vitality (e.g. overall energy and fatigue);
- economic ability (e.g. financial difficulty)
2) Disease-related symptoms (or both), including fatigue, pain, dyspnea, insomnia, appetite loss, and/or diarrhea.
- Assessment of risk of bias in included studies
Pairs of reviewers independently assessed risk of bias in the included studies by using the ROBINS-I assessment tool  for non-randomized studies, and the Cochrane risk of bias tool for randomized controlled trials. Any disagreements were resolved by discussion or consulting the third reviewer.
- Assessment of publication bias
If we included at least 10 studies in a meta-analysis, we generated funnel plots of effect estimates against their standard errors (on a reversed scale) using Review Manager software (RevMan). We assessed the potential risk of publication bias through a visual analysis of the funnel plots. Roughly symmetrical funnel plots indicated a low risk of publication bias and asymmetrical funnel plots a high risk. One should be aware that this is a rather subjective judgement and that funnel plot asymmetry might also arise from other sources and that publication bias does not always lead to asymmetry. We further attempted to avoid publication bias by searching trials registries and conference proceedings for unpublished studies. We addressed duplicate publication bias by including only one study with more than one publication. If we had doubt about whether multiple publications referred to the same data, we attempted to contact trial authors by email to resolve this issue.
- Grading of the evidence quality
Based on the results of the systematic review, the GRADE system was applied to evaluate the quality of the evidence, with results divided as follows: High quality (or A) - very confident that the real effect value is close to the estimated effect value, Moderate quality (or B) - having a moderate degree of confidence in the estimated value of the effect, and while the real value may be close to the estimated value there is still the possibility of large difference between the two groups, Low quality (or C) - limited confidence in the effect estimate and the true value may be quite different from the estimate, and Very low quality (or D) - little confidence in the effect estimate, with the true value likely to be very different from the estimate. Although evidence based on randomized controlled trails (RCT) is initially classified as high quality, confidence in such evidence may be diminished by five factors: (1) study limitations, (2) inconsistency in research results, (3) use of indirect evidence, (4) inaccurate results, and (5) publication bias. Evidence can be upgraded based on the following three factors; (1) large effect value, (2) existence of a dose-effect relationship, and (3) a possible confounding bias which may reduce efficacy.
Measures of treatment effect: We analyzed continuous outcomes as standardized mean differences (SMD) between groups with 95% CIs. To assess heterogeneity, we determined statistical heterogeneity using theχ2 test. If heterogeneity was low (I2 ＜50%, P ＞ 0. 05), we used the fixed effects model to calculate the combined effect. If heterogeneity was high ( I2≥50%, P≤0. 05), we used the random effects model to combine the studies. To assess reporting biases, we investigated publication and other reporting biases using funnel plots.