" Continuous update literature review on diet and cancer Protocol for systematic review on nutrition, physical activity and health outcomes in breast cancer survivors – version 2. The Continuous Update Project on breast cancer survivors is an extension of the Continuous Update Project (CUP) on diet, nutrition, physical activity, and cancer prevention. The current protocol for the Continuous Update on breast cancer survivors should ensure consistency of approach to the evidence used in the literature reviews for the WCRF/AICR Second Expert Report for cancer incidence and in the CU P. The starting points for this protocol are: • The convention for conducting systematic reviews developed by WCRF International for the Second Expert Report. 1 • The recommendations of the Cancer Survivors Protocol Development Committee (Appendix 1) • The protocol developed by the SLR group on cancer survivors for the Second Expert Report (SLR centre: University of Bristol) (Appendix 2) The peer-reviewed protocol will represent the agreed plan for the Continuous Update on breast cancer survivors. Should departure from the agreed plan be considered necessary at a later stage, this must be agreed by the WCRF/AICR Secretariat and the reasons documented. CANCER SURVIVORS PROTOCOL DEVELOPMENT COMMITTEE MEMBERS. Larry Kushi (LK), Chair Kaiser Permanente, Oakland CA, USA Marie-Christine Boutron-Ruault INSERM, E3N-EPIC Group, Institut Gustave Roussy, Villejuif (France) Bas Bueno-de-Mesquita National Institute for Public Health and the Environment (RIVM), The Netherlands Josette Chor School of Public Health and Primary Care, Chinese University of Hong Kong Wendy Demark-Wahnefried University of Alabama at Birmingham Comprehensive Cancer Center, USA Michelle Harvie University of Manchester, UK WCRF Executive
88
Embed
Breast Cancer Survivors Protocol - Food, Nutrition, and the
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
! "!
Continuous update literature review on diet and cancer
Protocol for systematic review on nutrition, physical activity and health outcomes in breast
cancer survivors – version 2.
The Continuous Update Project on breast cancer survivors is an extension of the Continuous Update
Project (CUP) on diet, nutrition, physical activity, and cancer prevention.
The current protocol for the Continuous Update on breast cancer survivors should ensure consistency
of approach to the evidence used in the literature reviews for the WCRF/AICR Second Expert Report
for cancer incidence and in the CU P.
The starting points for this protocol are:
• The convention for conducting systematic reviews developed by WCRF International for the
Second Expert Report.1
• The recommendations of the Cancer Survivors Protocol Development Committee (Appendix
1)
• The protocol developed by the SLR group on cancer survivors for the Second Expert Report
(SLR centre: University of Bristol) (Appendix 2)
The peer-reviewed protocol will represent the agreed plan for the Continuous Update on breast
cancer survivors. Should departure from the agreed plan be considered necessary at a later stage, this
must be agreed by the WCRF/AICR Secretariat and the reasons documented.
CANCER SURVIVORS PROTOCOL DEVELOPMENT COMMITTEE MEMBERS.
lifestyles in older cancer survivors to improve health and preserve function. J Am Geriatr Soc. 2009
Nov;57 Suppl 2:S262-4.
6. Kushi LH, Kwan ML, Lee MM, Ambrosone CB. Lifestyle factors and survival in women with
breast cancer. J Nutr. 2007 Jan;137(1 Suppl):236S-242S
7. Patterson RE, Cadmus LA, Emond JA, Pierce JP. Physical activity, diet,adiposity and female
breast cancer prognosis: a review of the epidemiologic literature. Maturitas. 2010;66(1):5-15
8. Sanderson S, Tatt ID, Higgins JP. Tools for assessing quality and susceptibility to bias in
observational studies in epidemiology: a systematic review and annotated bibliography. Int J
Epidemiol. 2007 Jun;36(3):666-76.
9. DerSimonian R, Laird N (1986) Meta-analysis in clinical trials. Control Clin Trials 7:177–188
10. N Orsini, R Bellocco and S Greenland, Generalized least squares for trend estimation of
summarized dose-response data, Stata J 6 (2006), pp. 40–57
11. Chêne G, Thompson SG. Methods for summarizing the risk associations of quantitative variables
in epidemiologic studies in a consistent form.Am J Epidemiol. 1996;144(6):610-21.
12. JP Higgins and SG Thompson, Quantifying heterogeneity in a meta-analysis, Stat Med 21
(2002), pp. 1539–1558.
13. A Tobias, Assessing the influence of a single study in meta-analysis, Stata Tech Bull 47 (1999),
pp. 15–17.
14. Bekkering GE et al. How much of the data published in observational studies of the association
between diet and colorectal or bladder cancer is usable for meta-analysis? Am J Epidemiol
(2008);167(9):1017-26.
Appendix 1: The recommendations of the Cancer Survivors Protocol Development
Committee
Population characteristics and subject eligibility to consider
Inclusion criteria
• In situ cancer
• Record histological type; DCIS/LCIS/other;
• Report separately from invasive cancer if possible
• Historical studies
• Historical studies with data before treatment were considered valuable. DG said that
an assessment of heterogeneity would be useful to determine whether historical
studies reported different results from more recent studies.
• Stage of diagnosis – Include all studies with data from diagnosis onwards
Exclusion criteria
• Male breast cancer eg mammographic density
• Pre-cancerous conditions – Exclude for breast cancer
o May be important for other cancers
• Studies that report on a combination of different cancers ie. Breast and other cancers,
without separate analyses on breast cancer survivors
o Record these papers as they may be used in the future but do not include in
this analysis
Information to extract if reported in papers
• Stage at diagnosis
o Include information on stage and grade of cancer
• Hormone receptor status
o Extract information on hormone receptor status
• Treatment status
o To include type of treatment. Need to develop broad groupings e.g. hormone
therapy, radiotherapy, chemotherapy (adjuvant and neoadjuvant) and surgery,
and combinations.
o Duration of treatment and timing in relation to exposure assessment
• How cancer was detected (eg, screening)
o People attending screening may be different and how cancer is detected
affects prognosis so these details should be extracted.
• BRCA1, BRCA2 and molecular subtypes
• Menopausal status
• Age
• Country study carried out in
o BMI range was considered important as heavier women might have different
effects; therefore studies in Asia might get effect at lower BMI
• Pre-diagnosis factors
o Include HRT use, diabetes status and treatment
Study designs to include (exclude everything else)
• Follow up of cases from case-control studies
• Follow up of cases from cohort studies
• Cancer survivor cohorts (explicitly designed)
• Randomized controlled trials (RCTs)
Exposures
To include
• List of exposures used in previous SLRs (including body size measures and change in
BMI)
• Supplements of substances found in diet e.g. vitamins (to include high dose vitamins)
and soy
• Biomarkers of dietary intake
• Physical activity
• Metabolic biomarkers
o These biomarkers might be important for interpretation of the data so extract
these if reported in included papers eg IGF
To exclude
• Herbal products
• Complementary and alternative medicine (CAM)
o There are a lot of agents tested in small populations. It was decided to restrict
the products to those than are found in ordinary diets.
o These products need to be taken into account in analyses so information on
use of such products should be extracted if reported in papers.
• Nutrition support (e.g. enteral/parenteral nutrition)
o Exclude as an exposure but extract if papers report on nutrition support as it
may be a modifier
Timing of exposure to include
• Pre-diagnosis
• Post-diagnosis
• Multiple time points
o Extract information on timing of exposure in relation to time before/after
diagnosis
Endpoints/outcomes to include
• All-cause or cancer (all types) mortality
• Breast-cancer specific mortality
• Disease-free survival
o It was questioned how we would define this. It was decided that anything
survival related that could be understood in papers should be extracted.
o Recurrence-free survival
• Recurrence
• New primary cancer (breast or other)
o Breast cancer survivors are at increased risk of second primary breast and
other cancers. This is important information; however will be unlikely to be
able to analyse by specific cancers.
• Quality of life outcomes
o Include only intervention studies with attention control/comparison groups
that follow-up participants for at least 6 months. Short-term studies were seen
as difficult to interpret.
o It was decided not to restrict studies based on sample size
o Includes fatigue, depression, functional/performance status
o There was some concern that searches may pick up huge numbers of potential
and/or relevant papers and that the criteria for inclusion for quality of life
might need to be modified once the searches are under way.
o A specialist in searching quality of life data should be consulted.
• Comorbidity
o Information on other diseases (e.g. cardiovascular disease, osteoporosis,
fractures) should be extracted.
• Lymphoedema
Study quality
• classify studies by extent to which measures of key confounders and covariates are
measured
o Extract information on randomization method, blinding, concealment and
intention to treat analysis for randomized controlled trials.
1
Systematic Review Protocol(Diet, Exercise and Cancer Survival)
1. Research questionDietary modifications and exercise interventions in the management of cancer.
2. ReviewersGeorge Davey Smith, Professor of Clinical EpidemiologyUniversity of Bristol, Department of Social Medicine, Canynge Hall, Whiteladies Road, Bristol, UK.
Anna Davies, Research Associate in EpidemiologyUniversity of Bristol, Department of Social Medicine, Canynge Hall, Whiteladies Road, Bristol, UK.
Roger Harbord, Research Associate in Medical StatisticsUniversity of Bristol, Department of Social Medicine, Canynge Hall, Whiteladies Road, Bristol, UK.
Jonathan Sterne, Senior Lecturer in Medical StatisticsUniversity of Bristol, Department of Social Medicine, Canynge Hall, Whiteladies Road, Bristol, UK.
Steve Thomas, Consultant Surgeon and Senior LecturerDepartment of Oral and Dental Science, Division of Oral and Maxillofacial Surgery, University ofBristol, Bristol, UK.
3. Sources of supportWorld Cancer Research Fund
4. ObjectivesTo investigate whether dietary modification and exercise have a role in the management of, andultimately survival from, cancer.
5. BackgroundIn 1996 there were over 10 million new cases of cancer worldwide and just over 7 million cancerdeaths (WHO, 1997). Traditionally, cancer has been labelled as a disease of developed countries.Data suggests, however, there is also an increasing prevalence of the disease in economicallydeveloping countries, in particular in urban areas. These figures suggest that cancer is a matter forpublic health concern worldwide (WHO, 1997).
The role of diet and exercise in the prevention of cancer has been extensively studied. The literaturein this field was reviewed by a recent WCRF / AICR report: Food, nutrition and the Prevention ofCancer: a global perspective (WCRF and AICR, 1997). By comparison, there is considerably lessscientific research on the role diet and exercise may play in the treatment of cancer. This means thatthere is greater uncertainty, for both health professionals and cancer survivors, surrounding decisionsto include dietary modifications and / or exercise interventions as part of a holistic approach to cancermanagement (AICR, 2002a; AICR, 2002b).
The role of biologically active components of the diet at various stages of the carcinogenic process issummarized in the WCRF/AICR report (WCRF and AICR, 1997). This, together with the reportedevidence from epidemiological studies, has lead to a series of recommendations for changes in dietand activity levels to reduce primary cancer risk (AICR, 2002a). For cancer survivors the guidelinesare less clear: the AICR state that ‘…most experts seem to agree that cancer survivors shouldconsider research results regarding risk reduction for primary cancers as being relevant to theirsituation’ (AICR, 2002a). A systematic review of the literature is needed to further clarify this
2
statement and the association between diet, exercise and cancer survival. To address this we intend tocarry out two reviews:
Review 1: Systematic review of randomized controlled trials (RCTs) investigating the effect ofdietary modification and exercise intervention on cancer survivalReview 2: Finding and updating the most recent epidemiological reviews on the associationbetween diet, exercise and cancer survival for each cancer site.
Review 1: Systematic review of randomised control trials investigating theeffect of dietary modification and exercise intervention on cancer survival.
1. Criteria For Considering Studies For This Review
Type of studiesAll randomized controlled trials (RCTs).
Type of participantsMales and females of any age with a diagnosis of cancer.
Study selection criteria• Dietary modifications as a result of dietary life-style changes, dietary education, micronutrient
supplementation and complementary medicine.• Diet and life-style modifications consequent on the disease or its treatment will NOT be
included. These include peri- and post-operative dietary modifications together with calorieenhancement for cancer cachexia.
• All exercise interventions will be included.• Dietary modifications and exercise interventions will be included regardless of their duration
or the route of dietary intake used. All concomitant interventions will be included.
Types of outcomes• survival / all-cause mortality• cancer mortality• primary cancer recurrence• second primary cancer• quality of life• side effects
2. Search strategy
All search strategies will be generated with the consultation of a medical librarian. The searches willnot be limited to English. We will attempt to get translations of all non-English papers, which appearrelevant. A second party will not independently verify translations. As far as possible, translators willbe ‘blinded’ to the exact nature of the study. Only literature published in peer review journals will beincluded in the review (i.e. no ‘grey literature’ will be searched). Searching will be carried out usingthe following sources and time-periods:
• The Cochrane Library (2003, Issue 2). Searches will include: DARE (Database of Abstractsof Reviews of Effects); CDSR (Cochrane Database of Systematic Reviews); HTA (HealthTechnology Assessment) and CENTRAL (The Cochrane Central Register of ControlledTrials.
• MEDLINE (2000-present). Medline has already been systematically searched - using ahighly sensitive search strategy developed by Carol Lefebvre and colleagues at the UKCochrane Centre (Dickersin K et al, 1994) – to identify all RCTs up to and including the year
3
1999. The identified RCTs are listed in the Cochrane Central Register of Controlled Trials(CENTRAL) (found in the Cochrane Library, listed above as a source to search). For thisreason we intend to only search Medline from 2000-2003.
• EMBASE (1980 – present)• ISI Web of Science (1981 – present)• BIOSIS (Previews) (1985 – present)• AMED (Allied and Complementary Medicine Database) (1981 – present)• Follow-up of references from relevant papers• Personal communication with experts
Search strategy for MEDLINEThis search strategy will be adapted for use in all other databases.
a) Searching for all studies relating to cancer and survival1 exp neoplasms/2 cancer$.tw.3 neoplasm$.tw.4 or/1-35 survivors/6 exp survival analysis/7 surviv$.tw.8 recurrence/9 recur$.tw.10 quality of life/ or quality of life.tw. or qaly$.tw.11 mortality/12 survival rate/13 (manag$ adj3 cancer$).tw14 or/5-1315 4 and 14
c) Searching for all studies relating to exercise modification39 exp Exercise Movement Techniques/40 exp exertion/41 exp sports/42 exp physical fitness/43 exp exercise/44 exercis$.tw.45 (physical$ adj3 activ$).tw.46 (physical$ adj3 fit$).tw.47 (physical$ adj3 train$).tw.48 or/39-47
d) Filter for randomised control trials – developed by Carol Lefebvre at the UK CochraneCentre (Dickersin K et al, 1994)
49 randomized controlled trial.pt.50 controlled clinical trial.pt.51 RANDOMIZED CONTROLLED TRIALS.sh.52 RANDOM ALLOCATION.sh.53 DOUBLE BLIND METHOD.sh.54 SINGLE-BLIND METHOD.sh.55 or/49-5456 (ANIMAL not HUMAN).sh.57 55 not 5658 CLINICAL TRIAL.pt.59 exp CLINICAL TRIALS/60 (clin$ adj25 trial$).ti,ab.61 ((singl$ or doubl$ or trebl$ or tripl$) adj25 (blind$ or mask$)).ti,ab.62 PLACEBOS.sh.63 placebo$.ti,ab.64 random$.ti,ab.65 RESEARCH DESIGN.sh.66 or/58-6567 66 not 5668 67 not 5769 57 or 68
e) Selecting all RCTs on cancer survival that involve dietary or exercise modification
70 15 and 69 and (38 or 48)
5
3. Methods of the review
Data CollectionAn ‘In/Out’ form (appendix 1) will be used to assess each study’s inclusion (or otherwise) into thereview. The inclusion of studies will be assessed independently by two assessors and differencesbetween reviewer’s results will be resolved by discussion and, when necessary, in consultation with athird reviewer.
A ‘Data Extraction’ form will be specifically designed for the review (appendix 2). Data abstractionwill be performed independently by two researchers and differences between reviewer’s results willbe resolved by returning to the relevant literature, discussion and, when necessary, consultation with athird reviewer. Data will be collected on the following:
• Target group• Cancer type• Intervention type and details of how many arms• Study Location• Period of intervention• Control group• Compliance• Experimental and control arm participant comparison (number, male:female ratio, age,
ethnicity, non-compliance, number analysed for outcome measures• Details of any switches in treatment condition• Outcomes (whether analysed on intention to treat basis, comparisons between experimental
and control arm)• Quality of the trial (selection bias, performance and detection bias, losses to follow up)
Data AnalysisWhere appropriate differences in outcomes comparing treatment and control groups will be combinedacross studies using standard methods for meta-analysis. Fixed (common) effect methods will beused. Forest plots will be used to display results and to examine possible heterogeneity betweenstudies. In addition to standard tests for heterogeneity I2 statistics will be used measure the amount ofheterogeneity (Higgins et al, 2002). In the presence of heterogeneity we will compare results fromfixed effect and random effects analyses.
Funnel plots will be used to explore “small study effects” (the tendency for smaller studies in a meta-analysis to show larger treatment effects) (Sterne et al, 2000). If funnel plot asymmetry is observedcareful consideration will be given as to its causes as well as the possible impact, on the overallestimate of treatment effect, from any meta-analysis performed (Sterne et al, 2001a).
Where there is sufficient data, sensitivity analyses will be carried out to investigate the impact, on thesummary estimate of effect, of excluding studies with inadequate or unclear allocation concealment,trials in which blinding was not adequate and /or trials in which methods for dealing with loss-to-follow-up were not adequate. (The quality assessment section of the data extraction form will assessthese characteristics).
A preliminary search of the literature suggests that is highly unlikely that there will be enough studiesfor dose-response graphs to be plotted or for meta-regression analyses to be done. The design of RCTsmeans that confounding should not be an issue and so confounding will not be considered.
All analysis will be undertaken using the statistical package Stata (version 8) in which acomprehensive set of user-written commands is available for meta-analysis (Sterne et al, 2001b).
6
Review 2Finding and updating the most recent epidemiological reviews on the associationbetween diet, exercise and cancer survival for each cancer site.
1. Criteria for considering studies for this review
Type of studiesAll case-control, cohort (retrospective and longitudinal) and cross-sectional studies.
Type of participantsMales and females of any age with a diagnosis of cancer.
Study selection criteria• Dietary modifications as a result of dietary life-style changes, dietary education, micronutrient
supplementation and complementary medicine.• Diet and life-style modifications consequent on the disease or its treatment will NOT be
included. These include peri- and post-operative dietary modifications together with calorieenhancement for cancer cachexia.
• All exercise interventions will be included.• Dietary modifications and exercise interventions will be included regardless of their duration
or the route of dietary intake used. All concomitant interventions will be included.
Types of outcomes• survival / all-cause mortality• cancer mortality• primary cancer recurrence• second primary cancer• quality of life• side effects
2. Search strategy
All search strategies will be generated with the consultation of a medical librarian. The searches willnot be limited to English. We will attempt to get translations of all non-English papers, which appearrelevant. As far as possible, translators will be ‘blinded’ to the exact nature of the study. Onlyliterature published in peer review journals will be included in the review (i.e. no ‘grey literature’ willbe searched). Searching will be carried out using the following sources and time-periods:
• The Cochrane Library (2003, Issue 2). Searches will include: DARE (Database of Abstractsof Reviews of Effects); CDSR (Cochrane Database of Systematic Reviews); HTA (HealthTechnology Assessment)
• MEDLINE (1966-present)• EMBASE (1980 – present)• ISI Web of Science (1981 – present)• BIOSIS (Previews) (1985 – present)• AMED (Allied and Complementary Medicine Database) (1981 – present)• Follow-up of references from relevant papers• Personal communication with experts
The nature of this review requires two search strategies:
1) The first search strategy aims to identify the most recent epidemiological review (if oneexists) on the association between diet and/or exercise and cancer survival for each of thecancer sites: breast, mouth, pharynx, nasopharynx, larynx, oesophagus, lung, stomach,
Search Strategy for MEDLINEThis search strategy will be adapted for use in all other databases
a) Searching for all studies relating to cancer and survival1 exp neoplasms/2 cancer$.tw.3 neoplasm$.tw.4 or/1-35 survivors/6 exp survival analysis/7 surviv$.tw.8 recurrence/9 recur$.tw.10 quality of life/ or quality of life.tw. or qaly$.tw.11 mortality/12 survival rate/13 (manag$ adj3 cancer$).tw14 or/5-1315 4 and 14
d) Filter for selecting only review papers: adapted from a high sensitivity search filter forsystematic reviews (CRD, 2001)
49 (review or review,tutorial or review, academic or review, literature).pt50 (systematic adj review$).tw51 (data adj synthesis).tw.52 (published adj studies).ab.53 (data adj extraction).ab.54 meta-analysis/55 meta-analysis.ti.56 comment.pt.57 letter.pt.58 editorial.pt.59 animal/60 human/61 59 not (59 and 60)62 (or/49-55) not (56 or 57 or 58 or 61)
e) Selecting all reviews on diet and exercise modification in cancer survival63 15 and 62 and (38 or 48)
9
2) The second search strategy aims to identify all relevant epidemiological studies, whichhave been published since the last review. Since the date of the last review will be differentfor each type of cancer it is necessary to create individual search strategies for each cancersite. The searches will be limited according to the date of the last review. If no review existson a particular cancer site then the databases will be searched from all years.
The search strategy listed below is for breast cancer. However, similar strategies will becreated for each cancer site and adapted for use in all other databases.
Search strategy for BREAST CANCER in MEDLINE
a) Searching for all studies relating to breast cancer and survival1 exp Breast Neoplasms/2 breast$ adj3 neoplasm$.tw3 ((breast$ adj3 cancer$) or (breast$ adj3 carcino$)).tw.4 or/1-35 survivors/6 exp survival analysis/7 surviv$.tw.8 recurrence/9 recur$.tw.10 quality of life/ or quality of life.tw. or qaly$.tw.11 mortality/12 survival rate/13 (manag$ adj3 cancer$).tw14 or/5-1315 4 and 14
c) Searching for all studies relating to exercise modification39 exp Exercise Movement Techniques/40 exp exertion/41 exp sports/42 exp physical fitness/43 exp exercise/44 exercis$.tw.45 (physical$ adj3 activ$).tw.46 (physical$ adj3 fit$).tw.47 (physical$ adj3 train$).tw.48 or/39-47
d) Filter for selecting only observational studies: case-control, cohort and cross-sectional studies(Scottish Intercollegiate Guidelines Network, June 2003).
49 epidemiologic studies/50 exp case control studies/51 case control.tw52 exp cohort studies/53 (cohort adj (study or studies)).tw54 cohort analy$.tw55 (Follow up adj (study or studies)).tw56 (observational adj (study or studies)).tw57 longitudinal.tw58 retrospective.tw59 exp cross-sectional studies/60 cross sectional.tw61 or/49-60
e) Selecting all epidemiological studies on diet and exercise modification in breastcancer survival
62 15 and 61 and (38 or 48)
3. Methods of the review
Data CollectionWhere possible the most up-to-date review will be used to obtain initial data for each cancer site.
Inclusion (or otherwise) of epidemiological studies published since the latest review will be assessedusing an in/out form. The inclusion of studies will be assessed independently by two assessors anddifferences between reviewer’s results will be resolved by discussion and, when necessary, inconsultation with a third reviewer.
Data abstraction forms will be specifically designed. Data abstraction will be performedindependently by two researchers and differences between reviewer’s results will be resolved byreturning to the relevant literature, discussion and, when necessary, consultation with a third reviewer.
11
Data Analysis
Observational studies, unlike RCTs, are susceptible to confounding and selection bias: as a result thefindings from observational studies may be distorted. Simply increasing study numbers inobservational studies will not necessarily reduce bias and confounding. In fact, smaller studies maybe able to better characterise potential confounding factors and, therefore, give a more accuratepicture than larger observational studies. For these reasons the general methods of meta-analysis –combining data and weighting studies according to their statistical size (the larger the study thegreater the weight) – have been suggested inappropriate with regards observational studies (Egger etal, 2001). Meta-analysis of observational studies will, therefore, not be undertaken in this review.
However, whilst meta-analysis is inappropriate a systematic review of the data is both plausible andappropriate. Where possible analysis will first focus on converting the results from each study into asuitable standard format to allow easy comparisons of results between studies. These results will thenbe tabulated and where appropriate ‘Forest Plots’ generated to give a graphical display of comparableresults, together with their 95% confidence intervals, from each study. Any heterogeneity betweenstudy findings will then be investigated using sensitivity analysis to test the stability of findings acrossdifferent study designs, approaches to exposure ascertainment and to selection of study participants.
Using these methods, as outlined by Egger and colleagues (Egger et al, 2001), we will avoidgenerating spurious results and misleading conclusions which are otherwise a common problem whenreviewing data from a number of observational studies which differ in study design. Furthermore, itwill assist considerably in deciding what future RCTs may be usefully initiated.
As already stated above confounding is an important consideration in observational studies and as partof the data collection we will note how confounding has been dealt with.
All analysis will be undertaken using the statistical package STATA, version 8.
12
References
AICR. Dietary Options for Cancer Survivors. Washington: American Institute of Cancer Research;2002a
AICR. Nutrition After Cancer. Washington: American Institute of Cancer Research; 2002b
Centre for Reviews and Dissemination (CRD). Undertaking Systematic Reviews of Research onEffectiveness. CRD’s Guidance for those Carrying Out or Commissioning Reviews. CRD ReportNo. 4. 2nd ed. York: CRD; 2001.
Dickersin K, SchererR, Lefebvre C. Identifying relevant studies for systematic reviews. BMJ 1994;309:1286-91.
Egger M, Davey Smith G, Schneider M. Systematic reviews of observational studies. In: Egger M,Davey Smith G, Altman DG, eds. Systematic Reviews in Health Care. Meta-analysis in context.London: BMJ Books; 2001. p. 211-227.
Higgins JPT and Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med2002;21:1539-1558.
Scottish Intercollegiate Guidelines Network (SIGN). Search Filters. Available from:http://www.sign.ac.uk/methodology/filters.html#obs. Accessed June 23, 2003.
Sterne JAC, Gavaghan D, Egger M (2000). Publication and related bias in meta-analysis: Power ofstatistical tests and prevalence in the literature. J Clin Epidemiol 2000;53:1119-1129.
Sterne JAC, Egger M, Davey Smith G. Investigating and dealing with publication and othe rbiases inmeta-analysis. BMJ 2001a ;323:101-105.
Sterne JAC, Bradburn MJ, Egger M. Meta-analysis in Stata. In: Egger M, Davey Smith G, AltmanDG, eds. Systematic Reviews in Health Care. Meta-analysis in context. London: BMJ Books; 2001b.p. 347-69.
WHO. The World Health Report. Geneva: WHO; 1997.
WCRF and AICR. Food, Nutrition and the Prevention of Cancer: a Global Perspective. Washing ton:American Institute of Cancer Research; 1997.
13
Appendix 1
In/Out Form
Diet, exercise and cancer survival – review of RCTs.
On RefMan database? ............................................................................................
Study Selection Criteria (please circle):
1. Study design is an RCT, not before/after or other design? Y/N
2. The study concerns individuals diagnosed with cancer Y/N• The study can include men and women of any age
3. The study concerns a dietary modification and/oran exercise intervention Y/N
• Dietary modification includes any life-style dietary modification, dietary education,nutrient supplementation or complementary medicine diet-based therapy.
• There is no restriction on route of dietary intake used.• Diet and life-style modifications consequent on the disease or its treatment will NOT
be included. These are peri- and post-operative dietary modification together withcalorie enhancement for cancer cachexia.
• All exercise interventions are included.• Modifications and interventions are included regardless of their time period.
LOCATION OF POPULATION FREETEXT (e.g. all patients registered at a GP inthe UK who are diagnosed with cancer, all out-patients attending a cancer clinic at aUS hospital): ............................. ...................................................................
TOTAL PERIOD OF ACTIVE INTERVENTION COVERED BY THIS REPORT(please leave blank if not reported):
Mean.........................................
Medium ....................................
Range.......................................
TYPE OF CONTROL CONDITION WAS:
Placebo Personal Contact/ No-intervention-usual diet/ No intervention – usualexercise / General health education materials/ Placebo supplement(NB: select all that apply)
HOW WAS COMPLIANCE CHECKED?
Monitoring of consumption by researcher/ self-report / other’s report on subject (e.g.relative) / weighed food intake diary/ <7 days dietary diary/ 7+ days dietary diary/biochemical levels in the blood or urine / food frequency questionnaire / physicalactivity monitoring machine / 24-hour exercise diary / <7 days exercise diary / 7+days exercise diary / exercise frequency questionnaire / it was not checked (NB select all that apply)
HOW MANY INTERVENTION ARMS WERE THERE IN THIS STUDY?
Number =..................................
DETAILS OF THE INTERVENTIONS IN THIS STUDY FREETEXT:
Method of Randomisation: ALLOCATION COMPUTER GENERATED/CENTRALOFFICE-TELEPHONE SERVICE/SEALED ENVELOPES/OTHER/NOT CLEAR
other (specify) FREETEXT……………………………………………………..
PERFORMANCE AND DETECTION BIAS
Were participants blind to the intervention assigned? YES/NO/NOT CLEAR
Were the clinicians/investigators blind to whichintervention was being provided? YES/NO/NOT CLEAR
Were the assessors of outcome measures blindto the intervention provided? YES/NO/NOT CLEAR
Were the statisticians blind to the interventionprovided? YES/NO/NOT CLEAR
LOSSES TO FOLLOW UP
How many subjects were lost to outcome follow up?• Survival / all-cause mortality...............................................................• cancer mortality..................................................................................• primary cancer recurrence .................................................................• second primary cancer.......................................................................• quality of life .......................................................................................• side effects.........................................................................................
21
Are subjects lost to follow up enumerated with reasons for loss?