Change in 28 HRQL outcomes after open surgery, open surgery combined, and minimally invasive surgery at 3, 6, 9, and 12 months follow-up.

Transcription

1 Supporting information for Meta-analysis shows clinically relevant and long-lasting deterioration in health-related quality of life after esophageal cancer. Online Resource 1 Online Resource 2 Online Resource 3 Online Resource 4 Online Resource 5 Online resource 6 Online Resource 7 Online Resource 8 Online Resource 9 Search Strategy Study Quality HRQL assessment, treatments evaluated, and patients in 15 studies included in the meta-analysis Imputation & sensitivity analysis Change in 28 HRQL outcomes after open, open combined, and minimally invasive at 3, 6, 9, and 12 months follow-up. The direction, clinical relevance, and duration of change of HRQL outcomes The duration of change for 28 HRQL outcomes after main analysis, and after subgroup analysis for open, open combined, and minimally invasive. Sensitivity analysis to assess the influence of transformations/imputations, outliers, study design, and HRQL assessment on the clinical relevance of HRQL change and the amount of heterogeneity across studies. Assessment of publication bias in 22 HRQL outcomes by use of the funnel plot, and the Trimm and Fill, classic Fail Safe N, and Egger Test statistics. 1

2 Online Resource 1: Search Strategy "Esophageal cancer", (Esophagus [MeSH Terms] OR esophageal neoplasms [MeSH Terms] OR esophageal cancer [All fields] OR oesophageal cancer [All fields] OR barrett esophagus [MeSH Terms]); "", (Esophagus resection [All fields] OR esophagectomy [All fields] OR esophagectomy [MeSH Terms] OR [Text word]); "combined therapy", (combined modality therapy [MeSH Terms] OR combined modality therapy [Text Word] OR neoadjuvant therapy [All Field] OR neoadjuvant treatment [All fields]); "radiotherapy", (Radiation therapy [All fields] OR radiotherapy [MeSH Terms] OR radiotherapy adjuvant [Text Word]); "chemotherapy", (chemotherapy [Text word] OR chemotherapy adjuvant [Text word]; "Quality of Life", (quality of life [Mesh Terms] OR quality of life [Text word] OR Health-Related Quality of Life OR qol [Text word] OR hrql [Text word] OR HRQL [ Text word] OR QoL [Text word]; "patient reported outcomes", (patient reported outcome [All fields] OR patient reported outcomes [All fields] OR outcome assessment (health care) [MeSH Terms] OR questionnaire [Text word]). 2

3 Online Resource 2: Study Quality Assessment of study design Oxford Centre for Evidence-Based Medicine (CEBM) 2011 Levels of Evidence # 2.1 Question: What are the treatment harms? Level 1 Systematic review of randomized trials or n-of-1 trial. Level 2 Randomized trial or observational studies with dramatic effect. Level 3 Non-randomized controlled cohort / follow-up study Level 4 Case-series, case-control studies, historically controlled studies, or cross-sectional studies Level 5 Mechanism-based reasoning 3

4 Assessment of risk of bias The Cochrane Collaboration s tool for assessing risk of bias V Criteria specified below are derived from the Cochrane Handbook V5.1.0 Selection Bias (Systematic differences between baseline characteristics of the groups that are compared) Domain Description Risk Requirement Sequence Generation The method used to generate the allocation sequence is described in sufficient detail to allow an assessment of whether it should produce comparable groups Low Unclear The investigators describe a random component in the sequence generation process. Insufficient information about the sequence generation process to permit judgement of Low risk or High risk. "Was the allocation sequence adequately generated?" High The investigators describe a non-random component in the sequence generation process. Other non-random approaches usually involve judgement or some method of non-random categorization of participants. Allocation Concealment The method used to conceal the allocation sequence is described in sufficient detail to determine whether intervention allocations could have been foreseen in advance of, or during, enrolment. Low Participants and investigators enrolling participants could not foresee assignment because one of the following, or an equivalent method, was used to conceal allocation: Central allocation (including telephone, web-based and pharmacy-controlled randomization); Sequentially numbered drug containers of identical appearance; Sequentially numbered, opaque, sealed envelopes. "Was allocation adequately concealed?" Unclear Insufficient information to permit judgement of Low risk or High risk. This is usually the case if the method of concealment is not described or not described in sufficient detail to allow a definite judgement for example if the use of assignment envelopes is described, but it remains unclear whether envelopes were sequentially numbered, opaque and sealed. High Participants or investigators enrolling participants could possibly foresee assignments and thus introduce selection bias, such as allocation based on: Alternation or rotation; Date of birth; Case record number; Any other explicitly unconcealed procedure. Performance Bias (Systematic differences between groups in the care that is provided, or in exposure to factors other than the interventions of interest) Domain Description Risk Requirement Blinding (participants/ personnel) All the measures used, if any, to blind study participants and personnel from knowledge of which intervention a participant received are described. Information is provided relating to whether the intended blinding was effective Low Unclear No blinding or incomplete blinding, but the review authors judge that the outcome is not possible to be influenced by lack of blinding. Blinding of participants and key study personnel ensured, and unlikely that the blinding could have been broken. Insufficient information to permit judgement of Low risk or High risk'. The study did not address this outcome. 4

5 "Was knowledge of the allocated intervention adequately prevented during the study?" High No blinding or incomplete blinding, and the outcome is possible to be influenced by lack of blinding.blinding of key study participants and personnel attempted, but possible that the blinding could have been broken, and the outcome is possible to be influenced by lack of blinding. Attrition Bias (Systematic differences between groups in withdrawals from a study) Domain Description Risk Requirement Incomplete outcome data - HRQL The completeness of outcome data for each main outcome, including attrition and exclusions from the analysis are described. Attrition and exclusions were also reported as are the numbers in each intervention group (compared with total randomized participants), reasons for attrition/exclusions, and any reinclusions in analyses performed by the review authors. "Were incomplete outcome data adequately addressed? Low No missing outcome data; Reasons for missing outcome data unlikely to be related to true outcome (for survival data, censoring unlikely to be introducing bias); Missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups; For dichotomous outcome data, the proportion of missing outcomes compared with observed event risk not enough to have a clinically relevant impact on the intervention effect estimate; For continuous outcome data, plausible effect size (difference in means or standardized difference in means) among missing outcomes not enough to have a clinically relevant impact on observed effect size; For cross-sectional studies; risk of bias was low if the compliance rate was equal and reasons for missing data was similar. Missing data have been imputed using appropriate methods. Unclear High Insufficient reporting of attrition/exclusions to permit judgement of Low risk or High risk (e.g. number randomized not stated, no reasons for missing data provided);the study did not address this outcome. Any one of the following: Reason for missing outcome data possible to be related to true outcome, with either imbalance in numbers or reasons for missing data across intervention groups; For dichotomous outcome data, the proportion of missing outcomes compared with observed event risk enough to induce clinically relevant bias in intervention effect estimate; For continuous outcome data, plausible effect size (difference in means or standardized difference in means) among missing outcomes enough to induce clinically relevant bias in observed effect size; As-treated analysis done with substantial departure of the intervention received from that assigned at randomization; Potentially inappropriate application of simple imputation Detection Bias (Systematic differences between groups in how outcomes are determined) Domain Description Risk Requirement Blinding of Outcome assessment - HRQL Low All the measures used are described and, if any, the measures used to blind outcome assessors from knowledge of which intervention a participant received. Information is provided Unclear No blinding of outcome assessment, but the review authors judge that the outcome measurement is not possible to be influenced by lack of blinding; Blinding of outcome assessment ensured, and unlikely that the blinding could have been broken. Insufficient information to permit judgement of Low risk or High risk ; The study did not address this outcome 5

6 relating to whether the intended blinding was effective. Do outcome assessors have knowledge of the allocated intervention? High No blinding of outcome assessment, and the outcome measurement is possible to be influenced by lack of blinding; Blinding of outcome assessment, but possible that the blinding could have been broken, and the outcome measurement is possible to be influenced by lack of blinding. Reporting Bias (Systematic differences between reported and unreported findings) Domain Description Risk Requirement Selective outcome reporting The possibility of how selective outcome reporting was examined is stated by the review authors, and what was found. "Are reports of the study free of suggestion of selective outcome reporting?" Low Unclear High If all domains and symptoms of the HRQL questionnaires were mentioned, in either text, graphs or a table. Insufficient information to permit judgement of Low risk or High risk. It is possible that the majority of studies will fall into this category. If a single HRQL domain of symptom was measures, yet not mentioned in the results as either text, graph or a table. Other Sources of Bias Domain Description Risk Requirement Other Sources of Bias Any important concerns about bias not addressed in the other domains in the tool are stated. If particular questions/entries were Low Unclear The study appears to be free of other sources of bias. There may be a risk of bias, but there is either: pre-specified in the review s Insufficient information to assess whether an important risk of bias exists; or protocol, responses should be Insufficient rationale or evidence that an identified problem will introduce bias. provided for each question/entry High There is at least one important risk of bias. For example, the study: Was the study apparently free of other problems that could put it at a high risk of bias? Had a potential source of bias related to the specific study design used Has been claimed to have been fraudulent; or Had some other problem (e.g. no control for multiple testing [P<0.01], inclusion/exclusion criteria not the same for both groups, funding sources) Selection Bias in Observational Studies (Systematic differences between baseline characteristics of the groups that are compared). Criteria specified by the authors. Domain Description Risk Requirement Confounding Did the study describe, consider, measure and include confounding factors in the study analysis? Low Unclear High In comparison studies, if groups were compared for age, sex, disease severity (i.e., TNM) and HRQL at baseline (if baseline was measured) and when discrepancies, if found, were taken into account. In one-group studies, age, sex, and disease severity were taken into account. If it was unclear if potential confounding of age, sec, disease severity and or baseline HRQL was explored or unclear whether discrepancies were taken into account If potential confounding of sex, age, disease severity and HRQL at baseline (if applicable)was not explored and or if differences between groups were not taken into account. 6

7 Assessment of HRQL outcome reporting Criteria as specified by the Efficace Minimum Standard Checklist, unless otherwise stated indicated by *. The answer 'unclear' was specified by the authors. HRQL Issue Answer Requirement Conceptual Is an a priori hypothesis Yes If authors had a predefined HRQL endpoint and/or stated expected changes due to the specific treatment. stated? No If the study was not exploratory and no hypothesis was stated. Unclear When an a priori hypothesis was not clearly reported N/A When a study states an exploratory HRQL evaluation or comparison. Is a rationale for choice of questionnaire(s) reported? Measurement Is the used questionnaire reported to be culturally validated for the specific study population? Is the coverage of questionnaire domains adequate? Methodology Is there an explicit statement who filled out the questionnaire? Yes No Unclear Yes No Unclear N/A Yes No Yes No When authors justified or explained their choice for selecting a specific HRQL questionnaire. When the author did not justify or explain their choice for selecting a specific HRQL questionnaire. When a rationale was not clearly reported If the measure was validated for the specific study population and this was referenced, or explicitly mentioned, in the text. Validated generic cancer instruments were assessed as yes. If the questionnaire was not culturally validated for the specific study population or if an article failed to provide at least a reference regarding the cultural validity of the questionnaire applied. Validated disease-specific instruments were validated as no if a different cancer site was assessed. If it was unclear whether the psychometric properties were applicable to the cultural validity of the used questionnaire. If the HRQL measure is validated in the same population as the one of the study. If a validated esophageal-specific questionnaire was used. If only a HRQL generic questionnaire was used or non-validated esophageal specific items were added. If it was explicitly stated who filled in the questionnaire If the sentence implied that a patient filled in the questionnaire, yet this was not explicitly reported in the text. (i.e., we used a self-report questionnaire) Is baseline compliance Yes If authors reported the numbers, or percentages, of patients that filled in the HRQL questionnaire at baseline (i.e. pre-treatment) reported?! No If no information was reported on the baseline compliance. Unclear If it was unclear how many patients complied at baseline N/A If a study did not implement a baseline HRQL assessment point. Is the timing of the questionnaire assessment during the study specified? Yes No Unclear If the authors reported the specific timing of HRQL assessment (e.g. weeks, months, years). For cross-sectional studies; mean/median and range of follow-up should be specified. If no information was reported with regards to the specific timing of HRQL assessment. If the authors reported when HRQL was assessed, but did not explicitly mention the time frame (i.e., range). For cross-sectional studies, timing of questionnaire assessment was unclear if a range was not provided. Are reasons for missing data Yes If author s report the specific reason(s) for missing data (e.g. death, administrative errors) documented?! No If no information was reported regarding the specific reasons for missing data. Unclear If the reason for missing data was unclearly documented N/A If there were no missing data Interpretation Are clinical significances addressed? Yes If the study addresses the clinical significance of the HRQL outcomes from the patient's perspective (e.g. effect size, minimally important difference) and not simply the statistical significance 7

8 No If no information was stated regarding the clinical significance of HRQL results. Is there a presentation of the results in general? Yes No If authors discussed the HRQL outcomes giving any comments regardless of the results (either expected or not). If the authors did not discuss HRQL outcomes in general or at all.! This item is mandatory for robust HRQL reporting for RCT's and longitudinal studies. For cross-sectional studies, reasons for missing data was deemed mandatory Assessment of additional HRQL outcome criteria Items developed by Efficace and identified in Pachler et al. Quality of life after rectal resection for cancer, with or without permanent colostomy HRQL Issue Answer Requirement Is a test of statistical Yes If the statistical test applied is specifically reported in the text. significance for HRQL applied? No Unclear If no statistical test was applied If a p value is reported, yet it is unclear which statistical test was used. Is there a HRQL difference between treatments reported? Is there an exploration of missing data? Yes No N/A Yes No N/A If a study shows at least a significant (statistical or clinical) difference in one HRQL domain, at any given time point, between pre-defined (i.e., described in the aims) treatments arms If no differences between treatments were reported (i.e., not for a domain or symptom, nor for any specific time-point). If there is only one pre-defined treatment arm or if a treatment arm is compared with a reference group. If an exploration for potential differences in characteristics has been reported, between those who participated and those who were lost to follow-up. If there was no exploration for differences between responders and non-responders or if differences were not taken into account. If there were no missing data reported 8

9 Assessment of additional issues of reporting and methodology Criteria based on the Newcastle-Ottowa Quality Assessment Scale en further specified by the authors. Issue Answer Requirement Was the study population Yes When age, sex, disease severity (i.e.,tnm staging), and baseline HRQL (if applicable) are reported for all groups. comprehensively described? No If no information was reported for age, sex, disease severity or baseline HRQL (if applicable). Unclear If age, sex, disease severity and baseline HRQL(if applicable) are reported for the total group and not for the subgroups. Were inclusion and exclusion criteria reported? Were primary (and if applicable) secondary outcomes defined? Was the intervention / treatment clearly described? Yes No Unclear Yes No Unclear Yes No If inclusion and exclusion criteria are reported. If inclusion and exclusion criteria are not reported. If either the inclusion or exclusion criteria are reported. If the HRQL outcome is explicitly defined as either primary or secondary. If HRQL was not mentioned as an outcome. If it was unclear whether HRQL was a primary or secondary outcome. If a clear description of the treatment was reported (e.g. type(s) of, type of chemotherapy/radiotherapy). If a global description of the treatment was reported (e.g. instead of 'transhiatal ') Was the sample size at the first assessment point N 26 for each group? Yes If the number of patients in a group at first assessment was higher or equal to 26. No If the number of patients in a group at first assessment was lower than 26. Unclear If it was unclear how many patients were included in a group at baseline. Was the sample size of a group N 26 for all assessments? Was the first time questionnaire compliance reported? Were follow-up rates / compliances reported? Were the follow-up compliance rates 80%? Was there a sample size calculation for HRQL outcome? Yes No Unclear Yes No Unclear Yes No Unclear N/A Yes No Unclear N/A Yes No Unclear N/A If N 26 at each time point assessed. If N<26 for at least one of the time-points assessed. If it was unclear how many patients were included for HRQL analyses at each time-point specified, or for each group included. If the compliance rate of the first questionnaire assessment is specified. If questionnaire compliance was not reported for the first HRQL assessment If questionnaire compliance is unclearly reported (e.g., no numbers or percentages provided) If the questionnaire compliance was reported for each time point, and each subgroup. If no information regarding follow-up compliance was reported. If the follow-up compliance was not specified for each time point and/or each sub-group. If there was no follow-up assessment If it was reported that for each time point assessed the follow-up compliance rate of those still eligible, (e.g. exclusion of deceased) was 80% for all groups assessed. If at any given time-point, for any specific group, the follow-up rate was below 80% of eligible (e.g., exclusion of deceased) If there was a report for the total follow-up compliance, but not for the specific subgroups. If the total questionnaire compliance was 80%, yet the compliance rate was not specified for each subgroup, this criterion was assessed as unclear. If there was no follow-up assessment If a sample size calculation for HRQL was reported. If sample size calculation for HRQL was not reported If it was unclear if the sample size calculation reported was for HRQL If it was clearly reported that HRQL is not a primary outcome. 9

10 Did studies control for multiple testing? Yes No Unclear N/A If they applied a p value of 0.01, if they applied criteria for clinical significance, or if they used a statistical technique to control for multiple testing If they applied a p value of 0.05, or did not use a statistical technique to control for multiple testing. If they applied a test of statistical significance, but it was unclear which p value was used. If they did not apply significance testing 10

11 Quality assessment results Risk of Bias a First Author Year Sequence generation Allocation concealment Blinding Patient/ Personnel Incomplete outcome data Blinding outcome Selective outcome Reporting Other sources of bias Confounding Included in meta-analysis Scarpa 2012 High Low High High Teoh 2011 Unclear Unclear Unclear Low Unclear High High N/A Nafteux 2011 High High High Unclear High Low High Low Wang 2011 High High High Unclear High Low Low Low Wang 2010 High High High Unclear High Low Low Low Parameswaran 2010 High Low Low High Safieddine 2009 High High Low High v. Meerten 2008 High Low Low High Barbour 2008 High High High Low High Low High High Avery 2007 High High High Unclear High High Low Low Lagergen 2007 High Low High High Tan 2006 High Low High High Reynolds 2006 High High High High High Low Low High de Boer 2004 Low Low Unclear Low Unclear Low Low N/A Brooks 2002 High High High Unclear High Low High High High (%) 7/9 (78%) 7/9 (78%) 7/9 (78%) 1/9 (11%) 13/15 (87%) 3/15 (20%) 7/15 (47%) 9/13 (69%) Low (%) 1/9 (11%) 1/9 (11%) 0/9 (0%) 3/9 (33%) 0/15 (0%) 12/15 (80%) 8/15 (53%) 4/13 (31%) Unclear (%) 1/9 (11%) 1/9 (11%) 2/9 (22%) 5/9 (55%) 2/15 (13%) 0/15 (0%) 0/15 (0%) 0/13 (0%) N/A (%) 6/15 (40%) 6/15 (40%) 6/15 (40%) 6/15 (40%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 2/15 (13%) a Total numbers and percentages for high, low, unclear, and N/A are based on the number of articles included, not treatments (e.g., scores for Brooks were only counted once) 11

12 HRQL outcome reporting a First Author Year A priori hypothesis Rationale Questionnaire Cultural validity Coverage domains Explicit statement Baseline compliance Timing assessment Reasons missing data Clinical Sign. Results in general Robust Reporting b Test Sign. HQQL difference Included in meta-analysis Scarpa 2012 No No Yes Yes Yes No Yes No No Yes No Yes N/A No Nafteux 2011 N/A No Yes Yes Yes Yes Yes No No Yes Yes Yes Yes No Teoh 2011 N/A No Yes Yes No No Yes No No Yes No Yes Yes No Wang 2011 N/A No Yes Yes Yes No Yes No No Yes No Yes Yes No Wang 2010 N/A No Unclear Yes Yes No Yes No No Yes No Yes Yes No Parameswaran 2010 N/A Yes Yes Yes Yes Yes Yes Unclear Yes Yes No No N/A No Safieddine 2009 N/A No Yes Yes No Yes Yes No No Yes No Yes N/A No Barbour 2008 N/A No Yes No No Yes Yes Yes Yes Yes Yes No Yes No v. Meerten 2008 N/A No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N/A No Avery 2007 N/A No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Lagergen 2007 N/A Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N/A No Reynolds 2006 N/A Yes Yes Yes Yes Yes Yes Yes No Yes Yes Yes Yes No Tan 2006 N/A No Unclear No No Yes Yes No No Yes No Yes N/A No de Boer 2004 N/A No Yes Yes No Yes Yes Yes Yes Yes Yes Yes Yes Yes Brooks 2002 N/A Yes Yes Yes Yes Yes Yes Yes No Yes Yes Yes No Yes Exploration missing data No (%) 1/1 (100%) 11/15 (73%) 0/15 (0%) 2/15 (13%) 5/15 (33%) 4/15 (27%) 0/15 (0%) 7/15 (47%) 9/15 (60%) 0/15 (0%) 7/15 (47%) 2/15 (13%) 1/9 (11%) 12/15 (80%) Yes (%) 0/1 (0%) 4/15 (27%) 13/15 (87%) 13/15 (87%) 10/15 (66%) 11/15 (73%) 15/15 (100%) 7/15 (47%) 6/15 (40%) 15/15 (100%) 8/15 (53%) 13/15 (87%) 8/9 (89%) 3/15 (20%) Unclear (%) 0/1 (0%) 0/15 (0%) 2/15 (13%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 1/15 (6%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 0/9 (0%) 0/15 (0%) N/A (%) 14/15 (93%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 6/15 (40%) 0/15 (0%) a b Total numbers and percentages for high, low, unclear, and N/A are based on the number of articles included, not the number of treatments included (e.g., scores for Brooks were only counted once) According to criteria specified by the minimum standard checklist for evaluating HRQL outcomes 12

13 Additional issues of reporting and methodology a First Author Year Included in meta-analysis Description population Inclusion/ exclusion criteria Primary/ secondary outcomes Description treatment reported Initial questionnaire compliance reported Follow-up compliance reported Follow-up compliance 80% N 26 First point N 26 All points Sample size calculation Scarpa 2012 No No No Yes No No Unclear Unclear Unclear No No Wang 2011 Yes Yes No Yes No No Yes Yes Yes No No Nafteux 2011 Yes Yes No Yes Yes Yes No Yes Yes No No Teoh 2011 Yes Yes Yes Yes No Unclear Unclear Yes Unclear N/A No Parameswaran 2010 Yes Yes No Yes Yes Unclear Unclear Yes Yes No Yes Wang 2010 No Yes No Yes No No Yes Yes Unclear No No Safieddine 2009 Yes Yes Unclear Yes Yes Yes Unclear Yes Yes No No Barbour 2008 Yes Yes No No Yes Yes Yes No No No Yes v. Meerten 2008 Yes Yes No Yes Yes Yes Yes Yes Yes No No Lagergen 2007 Yes Yes Unclear No Yes Unclear Yes Yes Yes No Yes Avery 2007 Yes Yes No Yes Yes Unclear No Yes Unclear No Yes Reynolds 2006 Yes Yes No Yes Yes Yes No Yes Yes No No Tan 2006 Yes Yes Unclear Yes Yes Yes Yes No No No No de Boer 2004 No Yes Yes Yes Yes Unclear No Yes Yes N/A Yes Brooks 2002 Yes Yes No No Yes Unclear No No No No No No (%) 3/15 (20%) 1/15 (7%) 10/15 (67%) 3/15 (20%) 4/15 (27%) 3/15 (20%) 5/15 33%) 3/15 (20%) 3/15 (20%) 13/15 (87%) 10/15 (67%) Yes (%) 12/15 (80%) 14/15 (93%) 2/15 (13%) 12/15 (80%) 11/15 (73%) 6/15 (40%) 6/15 (40%) 11/15 (73%) 8/15 (53%) 0/15 (0%) 5/12 (42%) Unclear (%) 0/15 (0%) 0/15 (0%) 3/15 (20%) 0/15 (0%) 0/15 (0%) 6/15 (40%) 4/15 (27%) 1/15 (7%) 4/15 (27%) 0/15 (0%) 0/15 (0%) N/A (%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 0/15 (0%) 2/15 (13%) 0/15 (0%) a Total numbers and percentages for high, low, unclear, and N/A are based on the number of articles included, not the number of treatments included (e.g., scores for Brooks were only counted once) Control Multiple Testing 13

14 Identification of robust study methodology and high-quality randomized controlled trials a Author Sequence generation Allocation concealment Blinding Participants / personnel Incomplete outcome data Other sources of bias Robust study Methodology b Robust HRQL outcome reporting c Included in meta-analysis Teoh Unclear Unclear Unclear Low High No No No De Boer Low Low Unclear Low Low Yes Yes Yes High (%) 0/3 (0%) 0/3 (0%) 0/3 (0%) 0/3 (0%) 2/3 (66%) 2/3 (66%) 2/3 (66%) 2/3 (66%) Low (%) 2/3 (66%) 1/3 (33%) 0/3 (0%) 2/3 (66%) 1/3 (33%) 1/3 (33%) 1/3 (33%) 1/3 (33%) Unclear (%) 1/3 (33%) 2/3 (66%) 3/3 (100%) 1/3 (33%) 0/3 (0%) 0/3 (0%) 0/3 (0%) 0/3 (0%) Highquality study a b c High and Low meaning No and Yes, respectively A RCT was identified as methodologically robust if at least four out of five criteria were met, with allocation concealment being mandatory According to criteria specified by the minimum standard checklist for evaluating HRQL outcomes 14

15 Identification of robust study methodology and high-quality non-randomized prospective studies a Author Year N 26 per group at first assessment point N 26 per group for all assessment points Follow-up compliance 80% Control for Confounding Control for multiple testing Robust study methodology Robust HRQL outcome Reporting Included in meta-analysis Wang 2010 Yes Yes Yes Yes No Yes No No Wang 2011 Yes Unclear Yes Yes No No No No Nafteux 2011 Yes Yes No Yes No No Yes No Avery 2007 Yes Unclear No Yes Yes No Yes No v. Meerten 2008 Yes Yes Yes No No No Yes No Lagergen 2007 Yes Yes Yes No Yes Yes Yes Yes Barbour 2008 No No Yes No Yes No Yes No Tan 2006 No No Yes No No No No No Parameswaran 2010 Yes Yes Unclear No Yes No No No Safieddine 2009 Yes Yes Unclear No Unclear No No No Scarpa 2012 Unclear Unclear Unclear No No No No No Reynolds 2006 Yes Yes No No No No Yes No Brooks 2002 No No No No No No Yes No No (%) 3/13 (23%) 3/13 (23%) 4/13 (31%) 9/13 (69%) 8/13 (62%) 11/13 (85%) 6/13 (46%) 12/13 (92%) Highquality study Yes (%) 9/13 (69%) 7/13 (54%) 6/13 (46%) 4/13 (31%) 4/13 (31%) 2/13 (15%) 7/13 (54%) 1/13 (8%) Unclear (%) 1/13 (8%) 3/13 (23%) 3/13 (23%) 0/13 (0%) 1/13 (8%) 0/13 (0%) 0/13 (0%) 0/13 (0%) N/A (%) 0/13 (0%) 0/13 (0%) 0/13 (0%) 0/13 (0%) 0/13 (0%) 0/13 (0%) 0/13 (0%) 0/13 (0%) a Total numbers and percentages for high, low, unclear, and N/A are based on the number of articles included, not the number of treatments included (e.g., scores for Brooks were only counted once) 15

16 Online Resource 3: HRQL assessment, treatments evaluated, and patients in 15 studies included in the meta-analysis HRQL assessment in 15 studies included in the meta-analysis. Author HRQL aim a Sample size eligible Sample size baseline b Handling of missing data c HRQL questionnaire(s) used to evaluate treatment Timing of baseline assessment c Time window baseline assessment Time window follow-up assessments Timing of HRQL assessment d Included in meta-analysis Scarpa Primary * Unclear QLQ-C30, -OES18 Time of diagnosis Unclear 14-days Yes Yes No Yes Teoh Secondary 45 34* Intention to treat QLQ-C30, -OES24 Pre-treatment Unclear Unclear Yes Yes No No Nafteux Primary * Unclear QLQ-C30, -OES18 Before 1 day 14 days Yes Yes Yes Yes Parameswaran Primary EORTC manual QLQ-C30, -OES18 Before 3 weeks 14 days No Yes No Yes Wang Primary Unclear QLQ-C30, -OES18 Before Unclear Unclear Yes Yes No No Wang Primary Unclear QLQ-C30, -OES18 Before Unclear Unclear Yes Yes No No Safedienne Primary Unclear FACT-E Pre -treatment Unclear Unclear Yes No No Yes Van Meerten Secondary EORTC manual QLQ-C30, -OES18 Pre -neoadjuvant Unclear Unclear Yes Yes Yes Yes Barbour Primary Unclear QLQ-C30 Before 4 weeks 7 weeks No Yes No No Avery Primary 81 63* EORTC manual QLQ-C30, -OES18 Pre-treatment 4 weeks 14 days Yes Yes No No Lagergren Primary 47 45* EORTC manual QLQ-C30, -OES18 Before 6 weeks days Yes Yes Yes Yes Tan Unclear Unclear SF-36, PAIS Before Unclear Unclear Yes Yes No Yes Reynolds Primary Unclear QLQ-C30, -OES24 Before 1-2 weeks 14 days Yes Yes No Yes de Boer Secondary Group mean MOS-SF-20, A-RSCL After randomization Unclear Unclear Yes Yes Yes Yes Brooks Primary Unclear FACT-E, POMS At diagnosis Unclear Unclear Yes Yes Yes No QLQ-C30 quality of life questionnaire cancer-30, QLQ-OES18 quality of life questionnaire esophagus-18, SF-36 medical outcomes study short-form general health survey, GERD- HRQL gastroesophageal reflux disease-health-related quality of life, a-rscl adapted Rotterdam symptom checklist, QLQ-OES24 quality of life questionnaire esophagus-24, MOS-SF-20 medical outcomes study short-form general health survey, FACT-E functional assessment of cancer therapy scale esophageal cancer, PAIS psychosocial adjustment to illness scale, POMS profile of mood states. a Primary aim: HRQL was the only outcome mentioned and / or HRQL was the first outcome mentioned in the main text. Secondary aim: HRQL was not the first outcome mentioned in the main text and / or the study referenced a previous study, based on the same data set, reporting on different outcomes b Number of patients assessed for HRQL at the first assessment point as reported in the study report c As explicitly stated in the study report d Yes if a study explicitly stated in the study report that HRQL was assessed at either 3, 6, 9,or 12 months. Studies that used timeframes to cover several months, for example 1-3 months were assessed as unclear * Based on the additional HRQL outcome data provided We imputed the standard deviation We transformed the timing of assessment The study from Lagergren et al. evaluated the same patient group as Avery et al., and reported outcome data for 9 months follow-up 3 months 6 months 9 months 12 months 16

17 Treatments evaluated in15 studies included in the meta-analysis Author a Teoh Nafteux Surgical procedure(s) b Right two phase (75%) Three phase (16%) Other (9%) Right two right (18%) Transthoracic left (82%) Percentage neo-adjuvant treatment Neo-adjuvant treatment(s) Adjuvant therapy Lymphadenectomy Anostomosis RT (9%) CT (5%) Two-field Cervical Inthrathoracic Two-field Cervical Inthrathoracic Replacement tube Gastro Colon Gastro? Wang Three-phase (100%) Two-field Cervical Gastro 44,8 % Reynolds Tan Left-sided approach (13%) Transhiatal (3%) Right two phase (51%) Three phase (34%) Two-phase (100%) - transthoracic left (57%) - transthoracic right (43%) Post-operative complications 39% Unclear Unclear Unclear 51,7 % Two-field Cervical Inthrathoracic de Boer Transhiatal (100%) Extended en bloc Cervical Gastro Unclear de Boer Two phase (100%) Extended en bloc Cervical Gastro Unclear Brooks Unclear Unclear Unclear Unclear Unclear combined Scarpa Right two phase Three phase Total (77%) CT (43%) CRT (57%) Two cycles of Cis + 5-FU Daily fractions 1,8 Gy to Gy & 2 cycles Cis + 5-FU Unclear (70%) En-bloc (?%) Cervical (25%) Inthrathoracic (75%) Gastro Gastro (93%) Colon (4%) Jejuno (3%) Unclear 37 % Safieddine Left-sided approach Right two phase Three phase CRT (87%) Cis 30 mg/m 2 + Irino 65m/mg Gy + boost (to 50 Gy) Two-field Cervical Inthrathoracic Unclear Unclear Van Meerten Transhiatal (88%) Two phase (12%) Avery Two-phase (100%) Total (100%) CT (59%) Lagergren Transhiatal (6%) Right two phase (89%) Three phase (4%) Reynolds Transhiatal (1%) Two phase (77%) Three phase (21%) CRT (100%) Pacl 50 mg/m 2 + carb + 23 fractions to 41,4 Gy CRT (41%) Total (62%) CT CRT Brooks Unclear Unclear Two-field Cervical Unclear 73% Two-field Unclear Unclear 47% Cis 80 mg/m FU (1g per m 2 per day) or four cycles of Epi 50mg/m 2 + Cis 60mg/m FU 200 mg/m 2 per infusion Four cycles of CIS 60 mg/m weeks 5-FU + 45 Gy total in 25 fractions for 5 weeks Unclear Two-field Unclear Unclear Unclear CRT Unclear Unclear Unclear Unclear 52% Minimally Invasive Surgery Nafteux Minimally Invasive (100%) Two-field Cervical Gastro Unclear Wang Three-phase (100%) Two-field Cervical (retro) Gastro 35% Wang Three-phase (100%) Two-field Cervical (pre) Gastro 33% Minimally invasive combined Parameswaran Minimally invasive (100%) Total (77%) CTx (100%) Cis + 5-FU Two-field Cervical Gastro 50% Gastrectomy Barbour Gastrectomy CTx (10%) One-field 55% CT: chemotherapy, CRT: chemoradiotherapy, RT: ratiotherapy, Cis: cisplatin, 5-FU: 5-fluorouracil, Gy: Gray, Pacl: paclitaxel, Carb: carboplatin, Irino: irinotecan, VATS: Video-Assisted Thoracoscopic Surgery, Epi: epirubicin, : unclear 17

18 a Studies that evaluated more than one patient group were entered more than once (e.g. the study from de Boer evaluated two patients groups and thus was entered twice). b Left-sided approach included left thoracotomy, thoracoabdominal, thoracolaparotomy, and "transabdominal". Right two phase included Ivor-Lewis, transthoracic, and two-phase esophagectomy. Three phase included threephase esophagectomy and McKeown esophagectomy". Minimally invasive approach included "minimally invasive ", "Minimally invasive esophagectomy", and " VATS". Any techniques not within these categories were classified as other.! data extracted from retrieved patient database * data extracted from previous report on clinical outcomes 18

19 Patients' demographic and clinical characteristics of the 15 studies included in the meta-analysis Author Age Sex Tumour site Histology Tumour stage Teoh Nafteux Wang Reynolds Tan de Boer de Boer Brooks Mean ± SD (range) (M / F ratio) Adeno SCC HGD 62 ± 9,4! (47-75) 64,1 ± (29-82) 58,2 ± m ± (29-79) 60.3 m ± 8.7 (51-69) 62 ± (23-78) 62 ± (35-78) 68,2 ± 10.1 (43-82) combined 60,5 ± Scarpa (53-67) Safieddine van Meerten Avery Lagergren Reynolds 60 ± (33-79) 59 m ± (40-75) 62.4 ± ± (44-79) 58 m ± (44-74) Brooks 57,1 ± 8.7 (42-71) Minimally invasive 63.1 ± Nafteux (41-82) Staging method pcr Stage 0 / I Stage Stage IIA +IIB III Stage IV 88% / 12%! O (100%) 100% unclear unclear unclear unclear unclear 82%/18% O (92%) OGJ (7%) Cardia (1%) 69%! 25%! 1%! ptnm 79%! 14%! 3%! 66% / 34% O 7% 93% ptnm 17% 66% 17% unclear O 62% 38% ptnm 11% 31% 56% 2% 81% / 19% Unclear 100% ptnm unclear unclear unclear unclear 88% / 12% 86% / 14% O/OGJ (84%) Cardia (16%) O/OGJ (84%) Cardia (16%) 100% ptnm 13%* 30%* 50%* 7%* 100% ptnm 16%* 15%* 54%* 15%* 89% / 11% O 67% 17% 11% ctnm 22% 44% 33% 81% / 19% O 60% 40% unclear unclear unclear unclear unclear 77% / 23% 91% / 9% O (71%) OGJ (29%) Thoracic (9%) Lower (91%) 74% 26% c-tnm 40% 39% 21% 76% 22% unclear unclear unclear unclear unclear 70% / 30% O 79% 21% ctnm 17% 83% 66% / 34% O (51%) OGJ (49%) 74% 19% 6% ptnm 13% 25% 49% 13% unclear O 72% 28% ptnm 21% 16% 36% 25% 1% 95% / 5% O 95% 5% ctnm 10% 90% 80% / 20% O(92%) OGJ (8%) 78%! 19%! 2%! ptnm 94%! 6%! Wang Wang 60.3 ± ± % / 33% O 4% 96% ptnm 23% 64% 13% 61% / 39% O 2% 98% ptnm 25% 67% 8% Minimally invasive combined Parameswaran 67m ± 90% / 10% O 89% 8% 3% unclear 16% 34% 44% 6% (49-80) Gastrectomy 19

20 Barbour 69 ± 10 80% / 20% OGJ 100% ptnm 5% 40% 45% 10% SD-standard deviation, Adeno-adenocarcinoma, SCC-Squamous cell carcinoma, HGD-high grade dysplasia, O-oesphagus, OGJ-oesophageal-gastric junction, ptnm-pathological staging, ctnm-clinical staging m median unclear! data extracted from patient database received from the first author * data extracted from a previous report on clinical outcomes 20

21 Online Resource 4: Imputation & sensitivity analysis Four methods that calculate an imputation value of the standard deviation (SD) at 6 and 12 months for the "open " treatment arm evaluated by Renolds et al. (2006) Reynolds open 6 Months 12 Months SD borrowed from Teoh 2011 at 6 months SD at baseline carried forward Weighted SD Weighted change SD SD borrowed from Nafteux 2011 at 12 months SD at baseline carried forward Weighted SD a Weighted change SD b a b c d SD at 3 months * inflation factor c SD Reynolds at 6 months Inflation factor d SD at 6 months * inflation factor c HRQL outcomes Global Quality of Life Physical functioning Role functioning Emotional functioning Cognitive functioning Social functioning Fatigue Nausea & Vomiting Pain Dyspnoea Insomnia Appetite Loss Constipation Diarrhea Financial Difficulties Dysphagia The numbers reported within the bold black border are based the standard deviations as reported by Nafteux (2011) at 12 months follow-up We calculated the "weighted change SD by pooling the SD change (from 6 to 12 months follow-up) of all studies that assessed the same HRQL outcome as Reynolds and multiplying that number with the standard deviation reported by Reynolds at 3 months follow-up The numbers reported within the red border are the imputations used in the final meta-analysis The inflation factor represents the pooled estimate of the weighted change in SD from 6 to 12 months follow of all studies that assessed the same HRQL outcome as Reynold 21

22 Four methods that calculate an imputation value of the standard deviation (SD) at 6 and 12 months for the "open " treatment arm evaluated by Renolds et al. (2006) Reynolds open combined a b c HRQL outcomes SD borrowed from Avery at 6 months 6 Months 12 Months SD at baseline carried forward Weighted SD Weighted change SD SD at 3 months * inflation factor ab SD borrowed from Lagergren at 12 SD at baseline carried forward Weighted SD c SD at 6 months Weighted change SD Inflation factor b SD (6 months) * inflation factor a Global Quality of Life Physical functioning Role functioning Emotional functioning Cognitive functioning Social functioning Fatigue Nausea & Vomiting Pain Dyspnoea Insomnia Appetite Loss Constipation Diarrhea Financial Difficulties Dysphagia The numbers reported within the red border are the imputations used in the final meta-analysis Based on the SD as reported by van Meerten, Scarpa, and Lagergren at 12 and 6 months follow-up The numbers reported within the bold black border are based the standard deviations as reported by Lagergren (2007) at 12 months follow-up 22

23 Sensitivity analysis Sensitivity analysis to compare four imputation methods with the main analysis for "open " at 6 months Reynolds "open " 6 Months Main analysis (i.e., no imputation) Max. 576 patients (3-6 studies) SD borrowed from Teoh at 6 months SD baseline Max. 710 patients (4-7 studies) Weighted SD Max. 710 patients (4-7 studies) Weighted change SD (SD base) Max. 710 patients (4 studies) a HRQL outcomes MD / SMD [95%CI] I 2 (%) MD / SMD [95%CI] I 2 (%) MD / SMD [95%CI] I 2 (%) MD / SMD [95%CI] I 2 (%) MD / SMD [95%CI] I 2 (%) Global Quality of Life 0.10 [-0.17, 0.36] [-0.13, 0.30] [-0.13, 0.30] [-0.13, 0.30] [-0.13, 0.30] 49 Physical functioning [-0.72, 0.28] [-0.72, 0.12] [-0.71, 0.12] [-0.71, 0.12] [-0.70, 0.13] 86 Role functioning [-0.38, 0.03] [-0.31, 0.01] [-0.31, 0.02] [-0.31, 0.02] [-0.31, 0.02] 13 Emotional functioning 2.51 [-4.75, 9.76] [-2.43, 12.71] [-2.64, 12.05] [-2.35, 13.08] [-2.72, 11.81] 51 Cognitive functioning [-7.71, 6.20] [-4.13, 5.74] [-3.95, 5.73] [-3.72, 5.72] [-3.95, 5.73] 47 Social functioning 0.18 [-0.20, 0.55] [-0.17, 0.44] [-0.17, 0.44] [-0.17, 0.44] [-0.17, 0.44] 74 Fatigue [-20.70, -4.83] [-17.22, -6.15] [-17.18, -6.16] [-16.92, -6.24] [-17.18, -6.16] 33 Nausea & Vomiting [-12.94, 4.29] [-8.41, 5.63] [-8.28, 5.72] [-12.94, 4.29] [-8.22, 5.77] 73 Pain [-15.32, -5.39] [-15.32, -6.73] [-15.31, -6.83] [-15.32, -6.64] [-15.31, -6.83] 0 Dyspnoea [-19.78, -5.65] [-15.40, -5.43] [-15.40, -5.43] [-16.32, -5.44] [-15.40, -5.43] 0 Insomnia [-12.46, 10.78] [-8.45, 7.74] [-8.45, 7.74] [-8.14, 7.51] [-8.41, 7.71] 51 Appetite Loss [-25.41, 7.86] [-18.92, 3.40] [-19.76, 3.99] [-19.23, 3.62] [-19.86, 4.06] 72 Constipation [-8.71, 2.66] [-6.18, 5.33] [-6.38, 4.72] [-6.18, 5.33] [-6.44, 4.57] 16 Diarrhea [-19.83, -3.53] [-16.10, -5.71] [-16.20, -5.66] [-16.30, -5.61] [-16.15, -5.69] 28 Financial Difficulties [-13.30, 1.20] [-11.50, 1.24] [-11.23, 1.24] [-10.93, 1.23] [-11.50, 1.24] 0 Dysphagia [-11.67, 38.59] [-5.61, 30.88] [-6.31, 31.59] [-5.48, 30.75] [-6.03, 31.31] 95 MD: weighted mean difference, SMD: standardized mean difference, SD: standard deviation a The numbers reported within the red border are the imputations used in the final meta-analysis 23

24 Sensitivity analysis to compare the four imputation methods with the main analysis for "open " at 12 months Reynolds "open " Main analysis (i.e., no imputation) Max. 358 patients (3-6 studies) SD borrowed from Nafteux at 12 months 12 Months SD baseline Max. 576 patients (4-7 studies) Weighted SD Weighted change SD a HRQL outcomes MD / SMD [95%CI] I 2 (%) MD / SMD [95%CI] I 2 (%) MD / SMD [95%CI] I 2 (%) MD / SMD [95%CI] I 2 (%) MD / SMD [95%CI] I 2 (%) Global Quality of Life 0.08 [-0.17, 0.32] [-0.28, 0.24] [-0.27, 0.24] [-0.27, 0.24] [-0.27, 0.24] 49 Physical functioning [-0.63, 0.36] [-0.73, 0.16] [-0.75, 0.16] [-0.70, 0.16] [-0.71, 0.16] 83 Role functioning [-0.60, 0.17] [-0.45, 0.10] [-0.45, 0.10] [-0.45, 0.10] [-0.45, 0.10] 55 Emotional functioning [-10.75, 8.75] N/A 5.65 [-8.05, 19.35] [-8.94, 17.81] [-8.05, 19.35] [-8.68, 18.39] 58 Cognitive functioning [-19.72, -0.28] N/A [-14.98, 6.55] [-14.92, 6.60] [-14.98, 6.55] [-15.03, 6.51] 64 Social functioning [-0.24, 0.16] [-0.17, 0.21] [-0.16, 0.19] [-0.16, 0.20] [-0.16, 0.19] 0 Fatigue [-30.44, -7.56] N/A [-23.15, -5.53] [-23.15, -5.53] [-23.15, -5.53] [-23.15, -5.53] 20 Nausea & Vomiting [-19.78, -4.22] N/A [-17.62, 5.89] [-17.62, 5.89] [-17.62, 5.89] [-17.40, 6.08] 84 Pain [-23.96, -0.04] N/A [-17.51, -1.90] [-16.83, -2.17] [-17.51, -1.90] [-16.66, -2.25] 0 Dyspnoea [-11.33, 9.33] N/A [-11.91, 1.55] [-11.96, 1.44] [-11.91, 1.55] [-12.13, 0.84] 19 Insomnia [-18.99, 6.99] N/A [-15.28, 2.18] [-15.61, 2.59] [-15.28, 2.18] [-15.61, 2.59] 0 Appetite Loss [-29.81, -4.19] N/A [-22.74, 0.78] [-23.37, -0.00] [-22.74, 0.78] [-23.18, 0.27] 33 Constipation [-18.15, 4.15] N/A [-12.52, 10.98] [-12.87, 10.65] [-12.52, 10.98] [-12.81, 10.71] 54 Diarrhea [-15.10, 1.10] N/A [-13.49, -2.76] [-13.73, -2.27] [-13.49, -2.76] [-13.09, -3.52] 0 Financial Difficulties [-18.48, 2.48] N/A [-17.98, -2.71] [-18.18, -1.44] [-17.98, -2.71] [-18.22, -1.16] 0 Dysphagia [-19.88, 3.88] N/A [-14.95, 14.44] [-15.46, 13.92] [-14.95, 14.44] [-15.69, 13.64] 60 MD: weighted mean difference, SMD: standardized mean difference, SD: standard deviation a The imputations within the red border are the numbers imputed and used in the final meta-analyses 24

25 Sensitivity analysis to compare the four imputation methods with the main analysis for "open combined" at 6 months Reynolds "open combined" 6 Months Main analysis (i.e., no imputation) (3 studies) SD borrowed from Avery at 6 months SD baseline carried forward (4 studies) Weighted SD (4 studies) Weighted change SD (SD at baseline) (4 studies) a HRQL outcomes MD / SMD [95%CI] I 2 (%) MD / SMD [95%CI] I 2 (%) MD / SMD [95%CI] I 2 (%) MD / SMD [95%CI] I 2 (%) MD / SMD [95%CI] I 2 (%) Global Quality of Life 0.08 [-0.47, 0.63] [-0.29, 0.50] [-0.29, 0.51] [-0.29, 0.51] [-0.29, 0.50] 68 Physical functioning [-0.97, -0.23] [-0.85, -0.34] [-0.84, -0.32] [-0.85, -0.35] [-0.84, -0.33] 25 Role functioning [-0.83, -0.05] [-0.68, -0.10] [-0.68, -0.09] [-0.68, -0.10] [-0.68, -0.09] 44 Emotional functioning [6.84, 16.54] [1.60, 15.55] [2.11, 15.58] [1.11, 15.56] [1.98, 15.57] 57 Cognitive functioning [-6.91, 6.51] [-5.05, 4.98] [-5.18, 5.07] [-4.98, 4.94] [-4.94, 4.92] 39 Social functioning [-0.37, 0.18] [-0.26, 0.31] [-0.26, 0.28] [-0.26, 0.32] [-0.26, 0.31] 41 Fatigue [-17.07, -6.41] [-17.59, -8.14] [-17.59, -8.14] [-17.65, -8.36] [-17.95, -9.12] 4 Nausea & Vomiting [-7.64, 2.09] [-7.03, 0.71] [-7.02, 0.68] [-7.00, 0.64] [-6.97, 0.55] 0 Pain [-8.45, 8.30] [-5.85, 6.92] [-5.90, 6.94] [-5.68, 6.86] [-5.81, 6.91] 47 Dyspnoea [-23.46, -4.86] [-22.41, -8.71] [-22.39, -8.95] [-22.32, -9.53] [-22.33, -9.46] 42 Insomnia 6.37 [-0.50, 13.23] [-1.97, 10.57] [-1.58, 10.80] [-2.18, 10.43] [-1.89, 10.62] 7 Appetite Loss 3.83 [-12.62, 20.27] [-10.61, 14.56] [-10.83, 15.00] [-10.49, 14.33] [-10.73, 14.78] 79 Constipation 6.03 [-1.53, 13.60] [0.35, 19.46] [0.45, 17.94] [0.27, 19.70] [0.46, 18.93] 75 Diarrhea [-19.96, -4.27] [-19.00, -7.24] [-18.87, -7.93] [-18.98, -7.37] [-18.85, -8.00] 52 Financial Difficulties [-3.98, 2.80] [-3.56, 4.18] [-3.55, 3.11] [-3.64, 5.89] [-3.45, 3.52] 2 Dysphagia 3.53 [-19.40, 26.46] [-9.26, 22.58] [-10.10, 23.20] [-9.14, 22.49] [-9.75, 22.94] 92 MD: mean difference, SMD: standardized mean difference, SD: standard deviation a The imputations within the red border are the numbers imputed and used in the final meta-analyses 25