Methodology Review

You have full text access to this OnlineOpen article

Strategies to improve retention in randomised trials

  1. Valerie C Brueton1,*,
  2. Jayne Tierney2,
  3. Sally Stenning3,
  4. Seeromanie Harding4,
  5. Sarah Meredith3,
  6. Irwin Nazareth5,
  7. Greta Rait5

Editorial Group: Cochrane Methodology Review Group

Published Online: 3 DEC 2013

Assessed as up-to-date: 13 JUN 2013

DOI: 10.1002/14651858.MR000032.pub2


How to Cite

Brueton VC, Tierney J, Stenning S, Harding S, Meredith S, Nazareth I, Rait G. Strategies to improve retention in randomised trials. Cochrane Database of Systematic Reviews 2013, Issue 12. Art. No.: MR000032. DOI: 10.1002/14651858.MR000032.pub2.

Author Information

  1. 1

    MRC Clinical Trials Unit, Meta-analysis Group, London, UK

  2. 2

    MRC Clinical Trials Unit at UCL, Meta-analysis Group, London, UK

  3. 3

    MRC Clinical Trials Unit, London, UK

  4. 4

    Medical Research Council, Social and Public Health Sciences Unit, Glasgow, UK

  5. 5

    University College London, Research Department of Primary Care and Population Health, London, UK

*Valerie C Brueton, Meta-analysis Group, MRC Clinical Trials Unit, 125 Kingsway, London, WC2B 6NH, UK. vcb@ctu.mrc.ac.uk.

Publication History

  1. Publication Status: New
  2. Published Online: 3 DEC 2013

SEARCH

 
Characteristics of included studies [ordered by study ID]
Ashby 2011

Methods2-arm randomised trial


DataAdults aged 18-65 years who provided email or mobile telephone contact details for receiving electronic reminders for follow-up in a migraine prevention trial


Comparisons1. Electronic reminder: either SMS text message, email message, or both sent after the 4-week follow-up study questionnaire sent

2. No electronic reminder sent


OutcomesPrimary: questionnaire response rate defined as proportion of questionnaires returned by participants at final analysis at 40 days

Secondary: time to response


NotesRetention trial embedded in a randomised trial evaluating the effectiveness of food elimination diet based on the ELISA test for food sensitivity for prevention of migraine. Primary outcome for the migraine prevention trial (host trial): change in the number of headache days over 12 weeks using the migraine disability assessment questionnaire (MIDAS). Retention trial identified through mail out to UK clinical trials units


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?YesAuthors response "an independent data manager at the trials unit was responsible for generating the allocation sequence and assigning participants"

Adequate sequence generation?Unclear"Randomly generated numbers used to list all participants by ID [identification] number who had provided a mobile phone number and/or and email address. The first half of listed participants were allocated to the intervention group the remaining participants were allocated to the control group"

Blinding?UnclearNo reference to blinding of either participants or outcome assessors in the study report

Free of selective outcome reporting?YesAll defined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement

Avenell 2004

MethodsRandomised trial


DataAdults aged ≥ 70 years with a low trauma osteoporotic fracture in the past 10 years recruited in 1 centre of the RECORD trial


Comparisons1. An open version of the RECORD trial otherwise identical in design

2. RECORD trial, a randomised double-blind placebo-controlled factorial design of oral calcium 1 g daily and or vitamin D 800 IU/20 µg supplementation


OutcomesProportion of eligible participants recruited

Proportion remaining in the trial at 1 year

Proportion compliant on pill counts at 8 months


NotesOpen version of the Randomised Evaluation of Calcium or Vitamin D (RECORD) Trial treated as the intervention group in the analysis. Proportion retained at 4, 8 and 12 months were reported. Primary outcome for the randomised double-blind placebo-controlled version of the RECORD trial was all new low-energy fractures


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Yes"Nurse used a pre programed laptop computer"

Adequate sequence generation?Yes"A pre programed laptop computer to generate random allocation"

Blinding?YesDouble-blind randomised trial design compared with an open trial design. For the double-blind randomised trial "allocation remained concealed until the final analyses". "All outcomes were reported or verified by people who were masked to the allocation scheme"

Free of selective outcome reporting?YesAll defined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement

Bailey 1

MethodsRandomised trial embedded in the sex unzipped website feasibility trial


DataUK English speaking people aged 16-20 years


Comparisons1. Offer of GBP20 voucher

2. Offer of GBP10 voucher


OutcomesRetention of participants at 3-month follow-up, i.e. completion of sexual health survey


NotesRetention trial identified through personal correspondence with the author

Sexunzipped website evaluated in an online trial


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?YesAuthors response "the trial statisticians generated the randomisation sequence with participants identified by ID [identification] number only and the trial manager implemented it manually"

Adequate sequence generation?YesAuthors response "participants were randomised after recruitment but before follow-up to a GBP10 or GBP20 incentive. Randomisation to increased incentive was through simple permutation of the list of remaining recruits"

Blinding?UnclearAuthors response "allocation sequences were generated without participants' knowledge". "For those allocated to the increased amount of GBP20, this was revealed in a 3 month follow-up email. Those allocated to GBP10 were not aware that others were offered GBP20 (unless friends had enrolled and had discussed the study). Since the trial recruited participants online from all over the UK, this will have reduced the chance of bias due to contamination"

Free of selective outcome reporting?YesReports the primary outcome

Bailey 2

MethodsRandomised trial


DataUK English speaking people aged 16-20 years


Comparisons1. Offer of GBP20 voucher: GBP10 in advance and GBP10 on receipt of questionnaire and chlamydia kit

2. Offer of GBP10 voucher: GBP5 in advance and GBP5 on receipt of questionnaire and chlamydia kit


OutcomesRetention of participants at 3-month follow-up, i.e. completion of sexual health survey and return of chlamydia kit


NotesStudy identified through personal correspondence with author

Sex unzipped website evaluated in an online trial


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?YesAuthors response "the trial statisticians generated the randomisation sequence (with participants identified by ID [identification] number only), and the trial manager implemented it manually"

Adequate sequence generation?YesAuthors response "participants were randomised after recruitment but before follow-up to a GBP10 or GBP20 incentive. Randomisation to increased incentive was through simple permutation of the list of remaining recruits"

Blinding?UnclearAuthors response "allocation sequences were generated without participants' knowledge". "For those allocated to the increased amount of GBP20, this was revealed in a 3 month follow-up email. Those allocated to GBP10 were not aware that others were offered GBP20 (unless friends had enrolled and had discussed the study). Since the trial recruited participants online from all over the UK, this will have reduced the chance of bias due to contamination"

Free of selective outcome reporting?YesReports the primary outcome

Bauer 2004a

Methods3-arm randomised trial (first incentive vs. no incentive)


DataPilot study of 15 randomly selected participants from each of 20 communities participating in COMMIT trial


ComparisonsEnclosed with mouthwash swish collection kits sent to participants "subjects were further randomised to receive" either:

  1. USD10 cheque told to keep the cheque whether or not they participated, sent with covering letter and prepaid envelope
  2. No incentive, kit sent with covering letter and prepaid envelope


All sent 2 weeks after an advance letter with a professionally rendered brochure


OutcomesPercentage of mouthwash kits returned reported. No time point given


NotesStudy embedded in the cluster randomised Community Intervention Trial for Smoking Cessation (COMMIT) trial. Primary outcome: quit rate among heavy smokers. Several attempts to contact authors regarding allocation sequence. Data extracted from Edwards Cochrane review on response to postal questionnaires


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?UnclearNo reply from author

Adequate sequence generation?Unclear"Subjects were further randomised to receive an incentive of....."

Blinding?UnclearInsufficient information to make a judgement about this, no reply from author

Free of selective outcome reporting?UnclearNo reply from author

Other sources of bias?UnclearInsufficient information to make a judgement

Bauer 2004ab

Methods3-arm randomised trial (combined incentive vs. no incentive)


DataPilot study of 15 randomly selected participants from each of 20 communities participating in COMMIT trial


ComparisonsEnclosed with mouthwash swish collection kits sent to participants "subjects were further randomised to receive" either:

  1. USD10 cheque told to keep the cheque whether or not they participated sent with covering letter and prepaid envelope or
  2. USD2 cheque with covering letter and prepaid envelope or
  3. No incentive, kit sent with covering letter and prepaid envelope


All sent 2 weeks after an advance letter with a professionally rendered brochure


OutcomesPercentage of mouthwash kits returned reported. No time point given


NotesStudy embedded in the cluster randomised Community Intervention Trial for Smoking Cessation (COMMIT) trial. Primary outcome: quit rate among heavy smokers. Several attempts to contact authors regarding allocation sequence. Data extracted from Edwards Cochrane review on response to postal questionnaires


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?UnclearNo reply from author

Adequate sequence generation?Unclear"Subjects were further randomised to receive an incentive of....."

Blinding?UnclearInsufficient information to make a judgement about this no reply from author

Free of selective outcome reporting?UnclearNo reply from author

Other sources of bias?UnclearInsufficient information to make a judgement

Bauer 2004b

Methods3-arm randomised trial (second incentive vs. no incentive)


DataPilot study of 15 randomly selected participants from each of 20 communities participating in COMMIT trial


ComparisonsEnclosed with mouthwash swish collection kits sent to participants, "subjects were further randomised to receive" either:

  1. USD2 check with covering letter and prepaid envelope
  2. No incentive, kit sent with covering letter and prepaid envelope


All sent 2 weeks after an advance letter with a professionally rendered brochure


OutcomesPercentage of mouthwash kits returned reported. No time point given


NotesStudy embedded in the cluster randomised Community Intervention Trial for Smoking Cessation (COMMIT) trial. Primary outcome: quit rate among heavy smokers. Several attempts to contact authors regarding allocation sequence. Data extracted from Edwards Cochrane review on response to postal questionnaires


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?UnclearNo reply from author

Adequate sequence generation?Unclear"Subjects were further randomised to receive an incentive of....."

Blinding?UnclearInsufficient information to make a judgement about this

Free of selective outcome reporting?UnclearNo reply from author

Other sources of bias?UnclearInsufficient information to make a judgement

Bowen 2000a

MethodsRandomised 2 x 2 factorial trial (first incentive vs. no incentive)


DataAdults aged 50-69 years recruited from 2 CARET trial sites, participating in the Participant Retention Item Distribution Evaluation (PRIDE) trial


Comparisons
  1. Certificate of appreciation preprinted on an 8.5 x 11 inch (21.59 x 27.94 cm)bond off-white paper with gold trim and bold, black lettering. The certificate read "[participant's name] in recognition of your contribution to an important national study for the prevention of lung cancer, CARET, CancerPrevention Study, sponsored by The National Cancer Institute". The participant's name was computer printed in an attractive font on the certificate. Each certificate had the signatures of the Co-ordinating Center's principal investigator, study centre investigator and CARET's project officer from the National Cancer Institute, given during a visit for randomisation or follow-up
  2. No incentive


OutcomesPrimary: time of first inactivation (stop taking vitamins or placebos) during the 2-year follow-up period of PRIDE


NotesPrimary disease outcome for the Carotene and Retinol Efficacy Trial (CARET): lung cancer


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?UnclearAuthor response "the Coordination Center supplied one date-labelled envelope for each day in the enrolment period to the two participating study centres. The envelopes contained the randomisation assignment for the given day. At the end of each day during the randomisation period, study centre staff opened the envelope containing the intervention assignment for the next day. Study centre staff members were otherwise blinded to the allocation sequence" 

Adequate sequence generation?YesAuthor response "we used a block randomisation approach (stratified by study centre) with a 1:1:1:1 intervention arm allocation ratio, where the randomisation unit was the date of enrolment". Note: treated this as quasi-randomised in the analysis

Blinding?YesAuthor response "with IRB [Institutional Review Board] approval, the study was conducted without participants' knowledge of this research. Thus, participants were blinded to their own intervention only in the sense that they were unaware they were randomised to receive particular item(s)"

Free of selective outcome reporting?YesCumulative incidence of individuals who became inactive during 2-year follow-up reported

Other sources of bias?UnclearInsufficient information to make a judgement

Bowen 2000abc

MethodsRandomised 2 x 2 factorial trial (combined incentive vs. no incentive)


DataAdults aged 50-69 years recruited from 2 CARET trial sites, participating in the Participant Retention Item Distribution Evaluation (PRIDE) trial


Comparisons
  1. Certificate of appreciation preprinted on an 8.5 x 11 inch (21.59 x 27.94 cm)bond off-white paper with gold trim and bold, black lettering. The certificate read '[participant’s name] in recognition of your contribution to an important national study for the prevention of lung cancer, CARET, CancerPrevention Study, sponsored by The National Cancer Institute'. The participant's name was computer printed in an attractive font on the certificate. Each certificate had the signatures of the Co-ordinating Center's principal investigator, study centre investigator, and CARET's project officer from the National Cancer Institute, given during a visit for randomisation or follow-up (arm a)
  2. Lapel pin 1 inch (2.5 cm) in size and designed in cloisonne. Choice between a pin with 6 colours with inscription 'CARET NCI prevention study' and an orange carrot in the middle of the pin or a pin with 5 colours with inscription 'PARTICIPNAT CARET Cancer Prevention Study' and 'Sponsored by NCI' and given during a visit for randomisation or follow-up (arm b)
  3. Certificate of appreciation (details as before) and lapel pin (details as before) and given during a visit for randomisation or follow-up (arm c)
  4. No incentive


OutcomesPrimary: time of first inactivation (stop taking vitamins or placebos) during the 2-year follow-up period of PRIDE


NotesPrimary disease outcome for the Carotene and Retinol Efficacy Trial (CARET): lung cancer


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?UnclearAuthor response "the Coordination Center supplied one date-labelled envelope for each day in the enrolment period to the two participating study centres. The envelopes contained the randomisation assignment for the given day. At the end of each day during the randomisation period, study centre staff opened the envelope containing the intervention assignment for the next day. Study centre staff members were otherwise blinded to the allocation sequence"

Adequate sequence generation?YesAuthor response "we used a block randomisation approach (stratified by study centre) with a 1:1:1:1 intervention arm allocation ratio, where the randomisation unit was the date of enrolment"

Blinding?YesAuthor response "with IRB [Institutional Review Board] approval, the study was conducted without participants’ knowledge of this research. Thus, participants were blinded to their own intervention only in the sense that they were unaware they were randomised to receive particular item(s)"

Free of selective outcome reporting?YesCumulative incidence of individuals who became inactive during 2 year follow-up reported

Other sources of bias?UnclearInsufficient information to make a judgement

Bowen 2000b

MethodsRandomised 2 x 2 factorial trial (second incentive vs. no incentive)


DataAdults aged 50-69 years recruited from 2 CARET trial sites, participating in the Participant Retention Item Distribution Evaluation (PRIDE) trial


Comparisons
  1. Lapel pin 1 inch (2.5 cm) in size and designed in cloisonne. Choice between a pin with 6 colours with inscription 'CARET NCI prevention study' and an orange carrot in the middle of the pin or a pin with 5 colours with inscription 'PARTICIPNAT CARET Cancer Prevention Study' and "Sponsored by NCI" and given during a visit for randomisation or follow-up
  2. No incentive


OutcomesPrimary: time of first inactivation (stop taking vitamins or placebos) during the 2-year follow-up period of PRIDE


NotesPrimary disease outcome for the Carotene and Retinol Efficacy Trial (CARET): lung cancer


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?UnclearAuthor response "the Coordination Center supplied one date-labelled envelope for each day in the enrolment period to the two participating study centres. The envelopes contained the randomisation assignment for the given day. At the end of each day during the randomisation period, study centre staff opened the envelop containing the intervention assignment for the next day. Study centre staff members were otherwise blinded to the allocation sequence"

Adequate sequence generation?YesAuthor response "we used a block randomisation approach (stratified by study centre) with a 1:1:1:1 intervention arm allocation ratio, where the randomisation unit was the date of enrolment"

Blinding?YesAuthor response "with IRB [Institutional Review Board] approval, the study was conducted without participants’ knowledge of this research. Thus, participants were blinded to their own intervention only in the sense that they were unaware they were randomised to receive particular item(s)"

Free of selective outcome reporting?YesCumulative incidence of individuals who became inactive during 2-year follow-up reported

Other sources of bias?UnclearInsufficient information to make a judgement

Bowen 2000c

MethodsRandomised 2 x 2 factorial trial (third incentive vs. no incentive)


DataAdults aged 50-69 years recruited from 2 CARET trial sites, participating in the Participant Retention Item Distribution Evaluation (PRIDE) trial


Comparisons
  1. Certificate of appreciation (details as before) and lapel pin (details as before) and given during a visit for randomisation or follow-up
  2. No incentive


OutcomesPrimary: time of first inactivation (stop taking vitamins or placebos) during the 2-year follow-up period of PRIDE


NotesPrimary disease outcome for the Carotene and Retinol Efficacy Trial (CARET): lung cancer


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?UnclearAuthor response "the Coordination Center supplied one date-labelled envelope for each day in the enrolment period to the two participating study centres. The envelopes contained the randomisation assignment for the given day. At the end of each day during the randomisation period, study centre staff opened the envelop containing the intervention assignment for the next day. Study centre staff members were otherwise blinded to the allocation sequence"

Adequate sequence generation?YesAuthor response "we used a block randomisation approach (stratified by study centre) with a 1:1:1:1 intervention arm allocation ratio, where the randomisation unit was the date of enrolment"

Blinding?YesAuthor response "with IRB [Institutional Review Board] approval, the study was conducted without participants’ knowledge of this research. Thus, participants were blinded to their own intervention only in the sense that they were unaware they were randomised to receive particular item(s)"

Free of selective outcome reporting?YesCumulative incidence of individuals who became inactive during 2-year follow-up reported

Other sources of bias?UnclearInsufficient information to make a judgement

Chaffin 2009

Methods2-arm randomised trial


DataAll parents referred for parenting services at a small inter-city non-profit community agency operating a parenting programme


Comparisons1. Initial preparenting orientation condition self motivation group

2. Initial preparenting orientation condition standard informational group


OutcomesDropout from the parenting group at 12 weeks


NotesA second randomisation was performed after completion of the orientation programme to parent child interactive therapy vs. standard didactic parenting condition. Dropout recorded at 2-week intervals up to 12 weeks


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?No"Unblinded randomisation list"

Adequate sequence generation?Yes"Computer generated randomisation list"

Blinding?YesAuthor response "they (parents) were informed only in general terms that we were interested in which types of services helped". Participant interviews were conducted by computer. Blinding of personnel: observational parent-child interaction coding was done by personnel who were not informed about intervention condition assignment

Free of selective outcome reporting?YesAll defined outcomes reported

Other sources of bias?No"It is possible that therapist effects may have played a role in the outcomes"

Cockayne 2005

Methods2-arm randomised trial


DataWomen aged ≥ 70 years randomised at 1 centre for a fracture prevention trial due to receive their final follow-up questionnaire in March 2004


Comparisons1. Final follow-up questionnaire additional question offering results of the trial delivered by post

2. Final follow-up questionnaire no offer of study results delivered by post

Both groups received a personalised cover letter showing university sponsorship, along with a business reply envelope. Non-responders within 3 weeks were sent up to 2 reminder letters, questionnaires and business reply envelopes, 3 and 6 weeks after the initial mailing          


OutcomesReturn of final follow-up questionnaire by participants. Time point not specified


NotesAuthors contacted to confirm numbers randomised to each arm

The Fracture Prevention Trial: calcium 1000 mg plus vitamin D3 800 IU plus information sheet on dietary calcium intake and falls prevention vs. information sheet. Primary outcome for the fracture prevention trial: all clinical fractures excluding those of the digits, rib, face and skull


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Yes"An independent researcher from the York trials unit randomised eligible women"

Adequate sequence generation?Yes"Randomised eligible women in a 3:1 ratio by computer"

Blinding?Unclear"Administration of the questionnaire was not blind to the group allocation"

Free of selective outcome reporting?YesAll outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement

Couper 2007

Methods2-arm randomised trial (incentive plus postal questionnaire vs. telephone survey)


DataAdults ≥ 18 years, BMI of ≥ 25 participating in an Internet-based weight management trial who did not respond to the 12-month questionnaire


Comparisons1. Telephone call and survey by trained interviewers. Repeated up to 15 times. Attempts made on various days and at various times of the day

2. Postal questionnaire with return address and covering letter signed by directors plus a USD5 bill


OutcomesQuestionnaire response. No time point given


NotesInternet-based weight management trial compared Internet-based tailored weight management materials with Internet-based non-tailored user navigated weight management materials. Primary outcome for the Internet-based trial: percentage of baseline weight lost


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?UnclearNot described

Adequate sequence generation?UnclearAuthor response "randomly assigned"

Blinding?UnclearNot described

Free of selective outcome reporting?UnclearUnclear at the outset what is to be reported

Other sources of bias?UnclearInsufficient information to make a judgement

Cox 2008

Methods2-arm randomised trial


DataAll sedentary women aged 50-70 years participating in the SWEAT2 trial


Comparisons1. 12 work sheets with strategies for goal setting, time management and overcoming barriers to attendance, mini workshops, received worksheets to complete at home - after 6 months received newsletters only. Intervention delivered by a trained facilitator, before an exercise session

2. Information sheets about programme requirements exercise techniques, safety plus 9 newsletters and a report on fitness at 6 months delivered by principle investigator plus 9 newsletters


OutcomesProgramme retention at 6 and 12 months


NotesContact with authors to clarify if withdrawals occurred before or after randomisation

SWEAT2 compared moderate walking programme vs. swimming programme. Primary outcome adherence to the programme


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Yes"Computer generated random numbers by a statistician"

Adequate sequence generation?Yes"Computer generated random numbers"

Blinding?Unclear"Not practical to blind the participants or the research staff to the group assignment" (see other sources of bias below)

Free of selective outcome reporting?YesAll defined outcomes reported

Other sources of bias?Yes"Walk and swim sessions were not separated according to the behavioural intervention participants asked not to discuss written materials in the practical sessions"

Dorman 1997

Methods2-arm randomised trial


DataAll UK participants entered into the International Stroke Trial between 2 March 1993 and 31 May 1995 who were still alive


Comparisons1. Short EuroQol posted with personalised letter and reply paid envelope. 1 reminder sent after 2 weeks

2. Long SF36 questionnaire posted with personalised letter and reply paid envelope. 1 reminder sent after 2 weeks


OutcomesFrequency of response after first and reminder mail out. Data for response to first mail out used


NotesInternational Stroke Trial compared heparin 125,000 IU twice daily + aspirin 300 mg vs. heparin 125,000 IU twice daily; heparin 5000 IU twice daily + aspirin 300 mg daily vs. heparin 5000 IU twice daily vs. aspirin 300 mg daily vs. no heparin or aspirin. Primary outcome: death within 14 days or dependency at 6 months


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?YesAuthors response "generated by computer"

Adequate sequence generation?UnclearAuthor response "generated by computer"

Blinding?UnclearAuthors report "there was no blinding for either study staff or participants"

Free of selective outcome reporting?YesAll defined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement

Edwards 2001

Methods2-arm randomised trial


DataAdults aged ≥ 16 years with head injury in the UK CRASH trial


Comparisons1. 1-page, 7-question functional dependence questionnaire with covering letter and stamped return envelope. Reminders after 4 and 8 weeks

2. 3-page, 16-question functional dependence questionnaire with covering letter and stamped return envelope. Reminders after 4 and 8 weeks


OutcomesNo of questionnaires returned within 3 months


NotesAuthors provided numbers randomised and responded

Primary outcome for the CRASH: death from any cause within 2 weeks of injury and death or disability at 6 months


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Yes"Central computer"

Adequate sequence generation?Unclear"Central computer"

Blinding?UnclearAuthor response "questionnaires were packaged and sent to patients by personnel who were independent of the study"

Free of selective outcome reporting?YesReported non-response

Other sources of bias?UnclearInsufficient information to make a judgement

Ford 2006

Methods2-arm quasi-randomised trial


DataAfrican American men aged 55 years enrolled in the intervention screening arm of the PLCO trial


Comparisons1. Indepth case management. Case management monthly telephone calls to participants, assisted making medical appointments, helped participants obtain health insurance information, legal aid, transportation services, food programmes, financial support, medication assistance, free medical care, information related to health risks facing African Americans. Provision of PLCO Cancer Screening Trial screening information and the scheduling of annual screening appointments.
2. Regular trial screening procedures. Participants called annually to schedule screening examinations  


OutcomesNumber completing the next scheduled PLCO cancer screen at 3 years


NotesPLCO trial compared digital rectal examination, transvaginal ultrasound and chest x-ray at baseline and 5 years to usual follow-up. PSA and cancer antigen CA125 at baseline, and annually for 5 years. Primary outcome mortality from 4 PLCO cancers


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?UnclearNot described

Adequate sequence generation?No"Randomised by participant id number"

Blinding?UnclearInsufficient information to make a judgement about this

Free of selective outcome reporting?YesAll defined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Gates 2009

Methods2-arm quasi-randomised trial


DataAll participants enrolled in the MINT trial due a follow-up questionnaire at 4 or 8 months after whiplash injury


Comparisons1. GBP5 voucher redeemable at shops www.highstreetvouchers.com plus questionnaire, cover letter included a sentence explaining that the voucher is to thank participants for their time and effort, delivered by post

2. No voucher and a standard covering letter with the questionnaire


OutcomesNumber of questionnaires returned after first contact with participants


NotesNumber of questionnaires returned in the incentive arm checked with authors

Primary outcome for the Managing Injuries of the Neck Trial (MINT) was returned to normal after whiplash injury measured using the Neck Disability Index (NDI)


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?UnclearAuthor response "lack of concealment of allocations before randomisation was not a major concern because it would have been very difficult for the staff in the study office who were sending out the questionnaires to have selectively allocated systematically different patients to the trial arm"

Adequate sequence generation?NoAuthor response "allocation to study arms was according to whether a specific digit of the patients study number was odd or even"

Blinding?UnclearAuthor response "trial office staff were unblinded, they had no influence over any participant's decision to return the questionnaire, and postal and telephone follow-up contacts were performed in a standardised way for all participants, without any reference to whether or not they were participating in the incentive study"

Free of selective outcome reporting?YesAll defined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Hughes 1989

Methods2-arm randomised trial


DataAll smokers for > 1 year in the nicotine gum vs. placebo gum trial


Comparisons1. A letter promising a free reprint of the study results in return for sending in the questionnaire

2. No offer of results


OutcomesQuestionnaire response. No time point given


NotesAuthor contacted regarding numbers randomised

Nicotine gum trial was double blind. The primary outcome was alleviation of signs and symptoms of tobacco withdrawal measured using a rating list and POMS (profile of mood states) questionnaire and DSM III criteria for tobacco withdrawal


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?YesAuthor response "by a non involved researcher sequence in sealed envelope, never opened during study"

Adequate sequence generation?YesAuthor response "using a table of random numbers"

Blinding?UnclearAuthor response "participants fully aware of each condition and which they were in"

Free of selective outcome reporting?YesResponse rates reported

Other sources of bias?UnclearInsufficient evidence to make a judgement on this

Kenton 2007a

MethodsRandomised 2 x 2 factorial trial (incentive vs. offer of incentive)


DataNew mothers with Edinburgh postnatal depression scale score > 9 participating in postpartum depression peer support trial


Comparisons1. USD2 coin mailed with questionnaire

2. Offer of entry into prize draw for USD50 gift certificate


OutcomesQuestionnaire response. No time point given


NotesThe postpartum depression peer support trial enrolled women < 2 weeks' postpartum at high risk for postnatal depression. New proactive individualised telephone-based peer support with standard postpartum care. The primary outcome was postnatal depression measured using the Edinburgh postnatal depression scale and Structured Clinical Interview Depression (SCID)


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?YesAuthor response "allocation sequence stored in a password protected file only accessed by an external researcher, computer generated"

Adequate sequence generation?UnclearAuthor response "computer generated"

Blinding?YesAuthor response "participants were not aware of the sub-study and didn't know other participants were receiving different types of mailings. The trial coordinator was not blinded to the study group"

Free of selective outcome reporting?YesDefined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Kenton 2007a-d

MethodsRandomised 2 x 2 factorial trial


DataNew mothers with Edinburgh, postnatal depression scale score > 9 participating in postpartum depression peer support trial


Comparisons1. USD2 coin mailed with questionnaire

2. Offer of entry into prize draw for USD50 gift certificate

3. USD2 coin with questionnaire sent with high-priority postage stamp

4. Offer of entry into lottery draw for USD50 gift certificate with high-priority postage stamp


OutcomesQuestionnaire response. No time point given


NotesThe postpartum depression peer support trial enrolled women < 2 weeks' postpartum at high risk for postnatal depression. New proactive individualised telephone-based peer support with standard postpartum care. The primary outcome was postnatal depression measured using the Edinburgh postnatal depression scale and Structured Clinical Interview Depression (SCID)


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?YesAuthor response "allocation sequence stored in a password protected file only accessed by an external researcher, computer generated"

Adequate sequence generation?UnclearAuthor response "computer generated"

Blinding?YesAuthor response "participants were not aware of the sub-study and didn't know other participants were receiving different types of mailings. The trial coordinator was not blinded to the study group"

Free of selective outcome reporting?YesDefined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Kenton 2007b

MethodsRandomised 2 x 2 factorial trial (incentive + priority mail vs. offer of incentive + priority mail)


DataNew mothers with Edinburgh postnatal depression scale score > 9 participating in postpartum depression peer support trial


Comparisons1. USD2 coin with questionnaire sent with high-priority postage stamp

2. Offer of entry into lottery draw for USD50 gift certificate with high-priority postage stamp


OutcomesQuestionnaire response. No time point given


NotesThe postpartum depression peer support trial enrolled women < 2 weeks' postpartum at high risk for postnatal depression. New proactive individualised telephone-based peer support with standard postpartum care. The primary outcome was postnatal depression measured using the Edinburgh postnatal depression scale and Structured Clinical Interview Depression (SCID).


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?YesAuthor response "allocation sequence stored in a password protected file only accessed by an external researcher, computer generated"

Adequate sequence generation?UnclearAuthor response "computer generated"

Blinding?YesAuthor response "

participants were not aware of the sub-study and didn't know other participants were receiving different types of mailings. The trial coordinator was not blinded to the study group"

Free of selective outcome reporting?YesExpected outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Kenton 2007c

MethodsRandomised 2 x 2 factorial trial (priority mail vs. no priority mail)


DataNew mothers with Edinburgh postnatal depression scale score > 9 participating in postpartum depression peer support trial


Comparisons1. USD2 coin sent by high-priority postage stamp plus questionnaire

2. USD2 coin plus questionnaire


OutcomesQuestionnaire response. No time point given


NotesThe postpartum depression peer support trial enrolled women < 2 week' postpartum at high risk for postnatal depression. New proactive individualised telephone-based peer support with standard postpartum care. The primary outcome was postnatal depression measured using the Edinburgh postnatal depression scale and Structured Clinical Interview Depression (SCID)


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?YesAuthor response "allocation sequence stored in a password protected file only accessed by an external researcher, computer generated"

Adequate sequence generation?UnclearAuthor response "computer generated"

Blinding?YesAuthor response "participants were not aware of the sub-study and didn't know other participants were receiving different types of mailings. The trial coordinator was not blinded to the study group"

Free of selective outcome reporting?YesExpected outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Kenton 2007d

MethodsRandomised 2 x 2 factorial trial (offer of entry into prize draw plus high-priority postage stamp vs. offer of entry into prize draw)


DataNew mothers with Edinburgh postnatal depression scale score > 9 participating in postpartum depression peer support trial


Comparisons1. Offer of entry into prize draw for USD50 gift certificate plus high-priority postage stamp

2. Offer of entry into prize draw for USD50 gift certificate


OutcomesQuestionnaire response. No time point given


NotesThe postpartum depression peer support trial enrolled women < 2 weeks' postpartum at high risk for postnatal depression. New proactive individualised telephone-based peer support with standard postpartum care. The primary outcome was postnatal depression measured using the Edinburgh postnatal depression scale and Structured Clinical Interview Depression (SCID)


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?YesAuthor response "allocation sequence stored in a password protected file only accessed by an external researcher, computer generated"

Adequate sequence generation?UnclearAuthor response "computer generated"

Blinding?YesAuthor response "participants were not aware of the sub-study and didn't know other participants were receiving different types of mailings. The trial coordinator was not blinded to the study group"

Free of selective outcome reporting?YesExpected outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Kenyon 2005

Methods2-arm randomised trial


DataWomen in ORACLE 1 and  2 participating in the evaluation of health and development study


Comparisons1. GBP5 voucher mailed with questionnaire redeemable at many high street shops

2. No incentive


OutcomesQuestionnaire response rate. No time point given


NotesORACLE trial: women with preterm prelabour rupture of fetal membranes and women with intact membranes in preterm labour, randomised to erythromycin 250 mg, co-amoxiclav 325 mg, erythromycin 250 mg plus co-amoxiclav 325 mg or placebo x 10 days or until birth. Primary outcome composite of neonatal death, chronic lung disease or major cerebral abnormality before discharge from hospital


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Yes"Randomly assigned by concealed computer generated allocation"

Adequate sequence generation?Yes"Randomly assigned by "computer" "

Blinding?YesAuthor response "both the participants and the study personnel were blinded to the allocation"

Free of selective outcome reporting?YesExpected outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Khadjesari 2011

MethodsKhadjesari 1: 4-arm randomised trial

Khadjesari 2: 2-arm randomised trial


DataSee 'Table of characteristics' for Khadjesari 2011 1abc; Khadjesari 2011 2


ComparisonsSee 'Table of characteristics' for Khadjesari 2011 1abc; Khadjesari 2011 2


OutcomesSee 'Table of characteristics' for Khadjesari 2011 1abc; Khadjesari 2011 2


NotesSee 'Table of characteristics' for Khadjesari 2011 1abc; Khadjesari 2011 2


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Yes"Allocation was concealed", "randomisation could not be subverted by the study team"

Adequate sequence generation?YesComputer-generated randomisation sequence

Blinding?UnclearUnclear

Free of selective outcome reporting?YesAll outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement about this

Khadjesari 2011 1a

Methods4-arm randomised trial (offer of incentives vs. no offer)


DataNon-responders at 1 week to the pilot DYD trial 3-month online follow-up questionnaire. Adults aged ≥ 18 scoring 5 or more on the AUDIT-C test


Comparisons1. Offer of GBP5 Amazon.co.uk voucher. Emailed voucher code on receipt of response (arm a)

2. Email with no offer of incentive (control)


OutcomesProportion completing questionnaire after 40 days


NotesPrimary outcome for DYD: total past week alcohol consumption measured using the TOT-AL measure of total alcohol consumption. Trial identified through personal communication


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Yes"Allocation was concealed randomisation could not be subverted by the study team"

Adequate sequence generation?YesComputer-generated randomisation sequence

Blinding?UnclearUnclear

Free of selective outcome reporting?YesExpected outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement about this

Khadjesari 2011 1abc

Methods4-arm randomised trial


DataNon-responders at 1 week to the pilot DYD trial 3-month online follow-up questionnaire. Adults aged ≥ 18 scoring 5 or more on the AUDIT-C test


Comparisons1. Offer of GBP5 Amazon.co.uk voucher. Emailed voucher code on receipt of response (arm a)

2. Offer of GBP5 donation to Cancer Research UK. Emailed hyperlink to charity's website showing donation when response received (arm b)

3. Offer of entry into GBP250 prize draw emailed confirmation of entry when response received (arm c)

4. Email prompt for completion of questionnaires with no offer of incentive (control arm)


OutcomesProportion completing questionnaire after 40 days


NotesPrimary outcome for DYD: total past week alcohol consumption measured using the TOT-AL measure of total alcohol consumption. Trial identified through personal communication


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Yes"Allocation was concealed randomisation could not be subverted by the study team"

Adequate sequence generation?YesComputer-generated randomisation sequence

Blinding?UnclearUnclear

Free of selective outcome reporting?YesExpected outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement about this

Khadjesari 2011 1ac

Methods4-arm randomised trial (incentives combined vs. no incentive)


DataNon-responders at 1 week to the pilot DYD trial 3-month online follow-up questionnaire. Adults aged ≥ 18 scoring 5 or more on the AUDIT-C test


Comparisons1. Offer of GBP5 Amazon.co.uk voucher. Emailed voucher code on receipt of response (arm a)

2.Offer of entry into GBP250 prize draw emailed confirmation of entry when response received (arm c)

3. Email with no offer of incentive (control arm)


OutcomesProportion completing questionnaire after 40 days


NotesPrimary outcome for DYD: total past week alcohol consumption measured using the TOT-AL measure of total alcohol consumption. Trial identified through personal communication

Separate comparison for arm b of attrition trial see Khadjesari 2011 1b


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Yes"Allocation was concealed randomisation could not be subverted by the study team"

Adequate sequence generation?YesComputer-generated randomisation sequence

Blinding?UnclearUnclear

Free of selective outcome reporting?YesExpected outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement about this

Khadjesari 2011 1b

Methods4-arm randomised trial (offer of donation to charity vs. no offer)


DataNon-responders at 1 week to the pilot DYD trial 3-month online follow-up questionnaire. Adults aged ≥ 18 scoring 5 or more on the AUDIT-C test


Comparisons1. Offer of GBP5 donation to Cancer Research UK. Emailed hyperlink to charity's website showing donation when response received (arm b)

2. Email with no offer of incentive (control)


OutcomesProportion completing questionnaire after 40 days


NotesPrimary outcome for DYD: total past week alcohol consumption measured using the TOT-AL measure of total alcohol consumption. Trial identified through personal communication


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Yes"Allocation was concealed randomisation could not be subverted by the study team"

Adequate sequence generation?YesComputer-generated randomisation sequence

Blinding?UnclearUnclear

Free of selective outcome reporting?YesExpected outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement about this

Khadjesari 2011 1c

Methods4-arm randomised trial (offer of entry into prize draw vs. no offer)


DataNon-responders at 1 week to the pilot DYD trial 3-month online follow-up questionnaire. Adults aged ≥ 18 scoring 5 or more on the AUDIT-C test


Comparisons1. Offer of entry into GBP250 prize draw emailed confirmation of entry when response received (arm c)

2. Email with no offer of incentive (control arm)


OutcomesProportion completing questionnaire after 40 days


NotesPrimary outcome for DYD: total past week alcohol consumption measured using the TOT-AL measure of total alcohol consumption. Trial identified through personal communication


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Yes"Allocation was concealed randomisation could not be subverted by the study team"

Adequate sequence generation?YesComputer-generated randomisation sequence

Blinding?UnclearUnclear

Free of selective outcome reporting?YesDefined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement about this

Khadjesari 2011 2

Methods2-arm randomised trial


DataAll DYD trial study participants


Comparisons1. Offer of a GBP10 Amazon.co.uk voucher. Email with voucher code sent on completion of questionnaire

2. Reminder email with no voucher offer


OutcomesThe proportion of participants that completed the questionnaire after 40 days


NotesPrimary outcome for DYD: total past week alcohol consumption measured using the TOT-AL measure of total alcohol consumption. Trial identified through personal communication


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Yes"Allocation was concealed", "randomisation could not be subverted by the study team"

Adequate sequence generation?YesComputer-generated randomisation sequence

Blinding?UnclearUnclear

Free of selective outcome reporting?YesAll outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement about this

Land 2007

Methods2-arm randomised trial


DataResearch sites participating in the NSABP B35 trial


Comparisons1. Automated prospective monthly reminder to trial sites of upcoming participant-reported outcome assessments. Reminder listed participants expected to complete Behavioral and Health Outcomes forms in upcoming 3 months

2. No monthly assessment reminder


OutcomesReceipt of questionnaire at any time


NotesB35 Anastrozole vs. tamoxifen for the treatment of ductal carcinoma in situ women aged 55 years plus. Primary outcome: time to first breast cancer reoccurrence


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?UnclearAuthor response to question "describe the methods if any, used to conceal the allocation sequence for the prospective reminder study in B35", "not applicable. Institutions were all randomly assigned before trial initiation, so there was no sequence"

Adequate sequence generation?YesAuthor response "prospective reminder in B-35, three strata were generated, and half in each were randomly assigned to receive the reminder"

Blinding?NoAuthor response "there was no blinding the prospective reminder was received by clinical staff"

Free of selective outcome reporting?YesPrimary outcome reported

Other sources of bias?UnclearInsufficient information to make a judgement

Leigh Brown 1997

Methods2-arm randomised trial


DataParticipants in OMENS due to receive follow-up questionnaires between March and December 1995


Comparisons1. Offer of GBP25 gift voucher monthly prize draw. Postcard reminder after 10 days and 3 weeks with reference to offer

2. No offer. Post card reminders after 10 days and 3 weeks with no reference to offer


OutcomesQuestionnaire response. No time point given


NotesOMENS effectiveness of musculoskeletal medicine vs. care by orthopaedic surgeon-led services for the treatment of non-surgical orthopaedic outpatients Primary outcomes: change in participant reported health (SF-36 and EuroQol) and marginal health


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?NoAuthor response "table of random numbers"

Adequate sequence generation?YesAuthor response "table of random numbers"

Blinding?UnclearAuthor response "those randomised to take part in the prize draw were aware of the intervention and those randomised to be excluded from the draw were unaware. The trial team were aware of the allocation so that they could arrange the monthly draw"

Free of selective outcome reporting?YesDefined outcomes reported outcome reported

Other sources of bias?UnclearInsufficient evidence to make a judgement about this

Letley 2000

Methods2-arm randomised trial


DataParticipants in the UKBEAM feasibility trial, aged 18-65 years with low back pain and a score of 4 or more on the RDQ


ComparisonsReview author (VB) checked which was control and which was intervention

1. RDQ before SF-36in a 26-page questionnaire

2. SF-36 before RDQ in a 26-page questionnaire


OutcomesQuestionnaire response at 3 months


NotesStudy complete, no data available at 8 September 2011. UKBEAM: compared the effectiveness of exercise, manipulation, exercise and manipulation. Primary outcome for UKBEAM scores on the RDQ at 3 and 12 months


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?YesAuthor reported "sequence generation was via remote service ensuring allocation concealment"

Adequate sequence generation?YesAuthor reported "sequence generation was random using randomised permuted blocks"

Blinding?UnclearNo data

Free of selective outcome reporting?YesOverall response rates reported

Other sources of bias?UnclearNo data

MacLennan

Methods2-arm randomised trial


DataParticipants who had not returned questionnaires at 12, 24 and 36 months in the RECORD trial


Comparisons1. Telephone call from RECORD office before first reminder questionnaire sent. Participant asked to complete questionnaire and to try to fill in all questions. Further telephone call from study nurse or RECORD office if not returned after 3 weeks

2. Repeat mailing of usual follow-up letter and questionnaire. Telephone call by study nurse or RECORD office if not returned after 3 weeks (routine follow-up)


OutcomesProportion of first reminder questionnaires returned

Proportion of questionnaires returned at 4 months

Completeness of data


NotesPrimary outcome all new low-energy fractures (self reported) EQ-5D Short form-12. Trial identified through mail out to UK clinical trial units


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?YesAuthor response "central randomisation service"

Adequate sequence generation?YesAuthor response "randomised to receive the "intervention" or not using the central randomisation service"

Blinding?YesAuthor response "blinding was not possible in the "intervention" arm. The control were blinded. Trial staff phoning were not blinded.

Outcome assessment in the sub-study was objective: questionnaire returned yes/no. Trial staff and trial participants were blinded to the RECORD trial allocation"

Free of selective outcome reporting?YesAll defined outcomes reported

Man 2011

Methods2-arm randomised trial


DataAdults 18-65 years sent a 6-month follow-up questionnaire in a trial of yoga for chronic back pain


Comparisons1. SMS text message: "Yoga trial: You should receive a questionnaire any day now. The data is important to us so please return it as soon as you can. Many thanks"

2. Email message: "Thank you for taking part in the Yoga for Low Back Pain Trial. This is an automatic reminder. You should receive your six month questionnaire any day now. The answers you give in the questionnaire are very important to the trial. Therefore, we should be most grateful if you would complete and return the questionnaire (and any other documents) as soon as possible please. Thank you"

3. SMS text message plus email message

4. No SMS test message or email message


OutcomesPostal questionnaire response


NotesYoga for chronic back pain trial: primary outcome functional limitations and disability measured by the RDQ. Setting 13 non-national health service settings in the UK. Participants were recruited through general practices


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?YesComputer-generated sequence to randomly allocate participants

Adequate sequence generation?YesIndependent data manager generated a computer sequence

Blinding?UnclearNo data

Free of selective outcome reporting?YesReported response to questionnaires

Other sources of bias?UnclearNo data

Marson 2007

Methods2-arm randomised trial


DataParticipants in SANAD due a QoL questionnaire at 1 year


Comparisons1. Cover letter with estimate of the length of time required to complete questionnaire

2. Standard cover letter without estimate of the length of time required to complete questionnaire


OutcomesResponse rate. No time point given


NotesSANAD compared the long-term effects of antiepileptic drugs in people with epilepsy. Participants randomised to arm a received carbamazepine, gabapentin, lamotrigine, oxcarbazepine, or topiramate and valproate, lamotrigine, or arm B received topiramate. Primary outcome time to treatment failure and time to achieve a 12-month remission of seizures


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?NoAuthor response "the list was given to the study researcher who worked sequentially through the list from left to right from top to bottom"

Adequate sequence generation?YesAuthor response "list of allocations was generated electronically using random permuted blocks of length 20"

Blinding?YesAuthor response "this was a postal study, blinding was not necessary for participants; they received either a time-framed letter or non time-framed letter. Clinical personnel did not see the letters, as these were sent postally directly to participants from the study research office"

Free of selective outcome reporting?YesAttrition reported

Other sources of bias?UnclearInsufficient evidence to make a judgement about this

McCambridge 2011

MethodsMcCambridge 1: 4-arm randomised trial in the pilot phase of the DYD trial; McCambridge 2: 4-arm randomised trial in the host DYD trial


DataSee 'Table of characteristics' for McCambridge 2011 1; McCambridge 2011 1a; McCambridge 2011 1b; McCambridge 2011 2; McCambridge 2011 2a; McCambridge 2011 2b


ComparisonsSee 'Table of characteristics' for McCambridge 2011 1; McCambridge 2011 1a; McCambridge 2011 1b; McCambridge 2011 2; McCambridge 2011 2a; McCambridge 2011 2b


OutcomesSee 'Table of characteristics' for McCambridge 2011 1; McCambridge 2011 1a; McCambridge 2011 1b; McCambridge 2011 2; McCambridge 2011 2a; McCambridge 2011 2b


NotesSee 'Table of characteristics' for McCambridge 2011 1; McCambridge 2011 1a; McCambridge 2011 1b; McCambridge 2011 2; McCambridge 2011 2a; McCambridge 2011 2b


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?YesSee 'Risk of bias' table McCambridge 2011 1; McCambridge 2011 1a; McCambridge 2011 1b; McCambridge 2011 2; McCambridge 2011 2a; McCambridge 2011 2b

Adequate sequence generation?YesSee 'Risk of bias' table McCambridge 2011 1; McCambridge 2011 1a; McCambridge 2011 1b; McCambridge 2011 2; McCambridge 2011 2a; McCambridge 2011 2b

Blinding?UnclearSee 'Risk of bias' table McCambridge 2011 1; McCambridge 2011 1a; McCambridge 2011 1b; McCambridge 2011 2; McCambridge 2011 2a; McCambridge 2011 2b

Free of selective outcome reporting?YesSee 'Risk of bias' table McCambridge 2011 1; McCambridge 2011 1a; McCambridge 2011 1b; McCambridge 2011 2; McCambridge 2011 2a; McCambridge 2011 2b

Other sources of bias?UnclearSee 'Risk of bias' table McCambridge 2011 1; McCambridge 2011 1a; McCambridge 2011 1b; McCambridge 2011 2; McCambridge 2011 2a; McCambridge 2011 2b

McCambridge 2011 1

Methods4-arm randomised trial in the pilot phase of the DYD trial


DataAdults aged ≥ 18 years scoring ≥ 5 on the AUDIT-C test participating in the pilot phase of the DYD trial 1-month data used


Comparisons1. APQ 23 items

2. AUDIT 10

3. LDQ 10

4. CORE-OM 23/34 items (mental health assessment)


OutcomesResponse to electronic questionnaires


NotesPrimary outcome for DYD: total past week alcohol consumption measured using the TOT-AL measure of total alcohol consumption. Trial identified through personal communication


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Yes"Allocation was concealed randomisation could not be subverted by the study team"

Adequate sequence generation?YesComputer-generated randomisation sequence

Blinding?UnclearUnclear

Free of selective outcome reporting?YesDefined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement about this

McCambridge 2011 1a

Methods4-arm randomised trial ('relevance' of questionnaire: alcohol vs. mental health data)


DataAdults aged ≥ 18 years scoring 5 or more on the AUDIT-C test participating in the pilot phase of the DYD trial 1-month data used


Comparisons1. APQ 23 items

2. CORE-OM 23/34 items (Mental health assessment)


OutcomesResponse to electronic questionnaires


NotesPrimary outcome for DYD: total past week alcohol consumption measured using the TOT-AL measure of total alcohol consumption. Trial identified through personal communication


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Yes"Allocation was concealed randomisation could not be subverted by the study team"

Adequate sequence generation?YesComputer-generated randomisation sequence

Blinding?UnclearUnclear

Free of selective outcome reporting?YesDefined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement about this

McCambridge 2011 1b

Methods4-arm randomised trial (short vs. long questionnaire comparison)


DataAdults aged ≥ 18 years scoring 5 or more on the AUDIT-C test participating in the DYD trial. For this comparison 1-month follow-up data from McCambridge 2011 1 used


Comparisons1. AUDIT 10 items + LDQ 10 items

2. APQ 23 items


OutcomesResponse to electronic questionnaires


NotesPrimary outcome for DYD: total past week alcohol consumption measured using the TOT-AL measure of total alcohol consumption. Trial identified through personal communication


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Yes"Allocation was concealed randomisation could not be subverted by the study team"

Adequate sequence generation?YesComputer-generated randomisation sequence

Blinding?UnclearUnclear

Free of selective outcome reporting?YesDefined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement about this

McCambridge 2011 2

Methods4-arm randomised trial in the host DYD trial


DataAdults aged ≥ 18 years scoring 5 or more on the AUDIT-C test participating in the pilot phase of the DYD trial 3-month data used


Comparisons1. APQ 23 items

2. AUDIT 10

3. LDQ 10

4. CORE-OM 23/34 items (Mental health assessment)


OutcomesResponse to electronic questionnaires


NotesPrimary outcome for DYD: total past week alcohol consumption measured using the TOT-AL measure of total alcohol consumption. Trial identified through personal communication


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Yes"Allocation was concealed randomisation could not be subverted by the study team"

Adequate sequence generation?YesComputer-generated randomisation sequence

Blinding?UnclearUnclear

Free of selective outcome reporting?YesDefined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement about this

McCambridge 2011 2a

Methods4-arm randomised trial ('relevance' of questionnaire alcohol vs. mental health)


DataAdults aged ≥ 18 years scoring 5 or more on the AUDIT-C test participating in the DYD trial. For this comparison 3-month follow-up data from McCambridge 2011 2 used


Comparisons1. AUDIT 10 items + LDQ 10 items

2. CORE-10


OutcomesResponse to electronic questionnaires


NotesPrimary outcome for DYD: total past week alcohol consumption measured using the TOT-AL measure of total alcohol consumption. Trial identified through personal communication


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Yes"Allocation was concealed randomisation could not be subverted by the study team"

Adequate sequence generation?YesComputer-generated randomisation sequence

Blinding?UnclearUnclear

Free of selective outcome reporting?YesDefined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement about this

McCambridge 2011 2b

Methods4-arm randomised trial (short vs. long questionnaire)


DataAdults aged ≥ 18 years scoring 5 or more on the AUDIT-C test participating in the DYD trial. For this comparison, 3-month follow-up data from McCambridge 2011 2 used


Comparisons1. AUDIT 10 items + LDQ 10 items

2. APQ 23 items


OutcomesResponse to electronic questionnaires


NotesPrimary outcome for DYD: total past week alcohol consumption measured using the TOT-AL measure of total alcohol consumption. Trial identified through personal communication


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Yes"Allocation was concealed randomisation could not be subverted by the study team"

Adequate sequence generation?YesComputer-generated randomisation sequence

Blinding?UnclearUnclear

Free of selective outcome reporting?YesDefined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement about this

McColl 2003

Methods2-arm randomised trial


Data40 adults aged ≥ 18 years with asthma from 62 general practices participating in the COGENT trial


ComparisonsSee 'Table of characteristics' for McColl 2003 1; McColl 2003 2


OutcomesSee 'Table of characteristics' for McColl 2003 1; McColl 2003 2


NotesSee 'Table of characteristics' for McColl 2003 1; McColl 2003 2


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?NoSee 'Risk of bias' table McColl 2003 1; McColl 2003 2

Adequate sequence generation?UnclearSee 'Risk of bias' table McColl 2003 1; McColl 2003 2

Blinding?YesSee 'Risk of bias' table McColl 2003 1; McColl 2003 2

Free of selective outcome reporting?YesSee 'Risk of bias' table McColl 2003 1; McColl 2003 2

Other sources of bias?UnclearSee 'Risk of bias' table McColl 2003 1; McColl 2003 2

McColl 2003 1

Methods2-arm randomised trial


Data40 adults aged ≥ 18 years with asthma from 62 general practices participating in the COGENT trial


Comparisons1. Asthma specific instruments (NASQ and AQLQ) followed by generic questions in medical outcomes SF-36 version 1 and EQ-5D

2. Generic questions followed by condition specific


OutcomesQuestionnaire response rates, speed and patterns. No time point given


NotesCOGENT before and after cluster randomised, 2 x 2 factorial trial. Evaluated use of computerised support system for decision making for implementing evidenced-based clinical guidelines for the management of asthma (and angina see McColl 2003 2). General practices randomised to computerised guidelines for the management of angina or asthma. Primary outcome adherence to guidelines


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?NoAuthor response "practices provided us with a computerised list of all patients fulfilling eligibility criteria. We drew a simple random sample of patients from that list first 40 from list received version 1, second 40 version 2"

Adequate sequence generation?UnclearAuthor response "practices provided us with a computerised list of all patients fulfilling eligibility criteria. We drew a simple random sample of patients from that list. First 40 received version 1, second 40 version 2. Logistically, truly random allocation of the 80 patients to versions 1 and 2 would have been impractical to implement"

Blinding?YesAuthor response "respondents were not alerted to the fact that we were experimenting with the order of instruments"

Free of selective outcome reporting?YesDefined outcomes reported

Other sources of bias?UnclearInsufficient evidence to make a judgement about this

McColl 2003 2

Methods2-arm randomised trial


Data40 adults aged ≥ 18 years with asthma from 62 general practices participating in the COGENT trial


Comparisons1. Angina-specific instruments (SAQ) followed by generic questions (SF-36 version 1 and EQ-5D) 

2. Generic questions followed by condition-specific questions


OutcomesQuestionnaire response rates, speed and patterns. No time point given


NotesCOGENT before and after cluster randomised, 2 x 2 factorial trial. Evaluated use of computerised support system for decision making for implementing evidenced-based clinical guidelines for the management of angina (and asthma see McColl 2003 1). General practices randomised to computerised guidelines for the management of angina or asthma. Primary outcome adherence to guidelines


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?NoAuthor response "practices provided us with a computerised list of all patients fulfilling eligibility criteria. We drew a simple random sample of patients from that list. First 40 from list received version 1, second 40 version 2"

Adequate sequence generation?UnclearAuthor response "practices provided us with a computerised list of all patients fulfilling eligibility criteria. We drew a simple random sample of patients from that list. First 40 received version 1, second 40 version 2. Logistically, truly random allocation of the 80 patients to versions 1 and 2 would have been impractical to implement"

Blinding?YesAuthor response "respondents were not alerted to the fact that we were experimenting with the order of instruments"

Free of selective outcome reporting?YesDefined outcomes reported

Other sources of bias?UnclearInsufficient evidence to make a judgement about this

Nakash 2007

Methods2-arm randomised trial


DataParticipants aged ≥ 16 years from 5 trial centres randomised to CAST between November 2003 and July 2005


Comparisons1. Trial calendar: monthly customised calendar, included prenotification caption on the months the participant is due to receive the questionnaire with reminder caption the following month

2. No trial calendar


OutcomesQuestionnaire response at 4, 12 weeks and 9 months. Response at 4 weeks used for analysis

Amount of prompting required to return questionnaires at each time point

Percentage of missing data of the core outcome


NotesCAST tubular bandage, below knee cast, Aircast® ankle brace and Bledsoe® boot compared in people with acute severe ankle sprain. Primary outcome FAOS, FLP, SF-12 and EuroQol 5


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Yes"Baseline packs compiled in advance and stored at trial sites. On randomisation the next consecutively numbered pack was taken. Allocation concealed until participant recruited into CAST and pack opened"

Adequate sequence generation?Yes"Computer generated random sequence"

Blinding?Yes"No blinding of participants or clinic staff", "data inputting was blind to allocation"

Free of selective outcome reporting?YesAll defined outcomes reported

Other sources of bias?UnclearInsufficient evidence to make a judgement about this

Renfroe 2002a

Methods2 x 2 x 2 x 2 factorial randomised trial (certificate of appreciation vs. no certificate)


DataAll surviving participants at AVID study termination


Comparisons1. Certificate of appreciation with cover letter signed by physician or co-ordinator sent either by express or standard post, 2-3 weeks after last AVID follow-up visit or 1-4 months after last AVID follow-up visit

2. No certificate with cover letter signed by physician or co-ordinator, sent either by express or standard post, 2-3 weeks after last AVID follow-up visit or 1-4 months after last AVID follow-up visit


OutcomesQuestionnaire response. No time point given


NotesAVID conducted in people resuscitated from ventricular fibrillation or cardioverted for ventricular tachycardia. Compared implanted cardioverter defibrillator vs. antiarrhythmic drugs. Primary outcome overall mortality


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?UnclearThe AVID CTC prepared the resulting 16 sets of distinct participant survey packets. These participant-specific packets were then mailed to the study co-ordinators, who distributed them to participants as instructed

Adequate sequence generation?YesAuthor response "all patients in the study were randomised. Details of the randomisation scheme are no longer extant, but given the factorial design I think it's safe to assume that the randomization for each factor was by permuted blocks of size 2"

Blinding?YesParticipants were instructed to mail the completed surveys directly to the CTC in postage-paid, self-addressed return envelopes. Only their AVID study number identified participants, assuring confidentiality of their survey responses

Free of selective outcome reporting?Yes All defined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Renfroe 2002a-d

Methods2 x 2 x 2 x 2 factorial randomised trial


DataAll surviving participants at AVID study termination


Comparisons1. Certificate of appreciation

2. No certificate of appreciation

3. Overnight express delivery

4. Regular post

5. Cover letter signed by physician

6. Cover letter signed by study co-ordinator

7. Early administration

8. Late administration


OutcomesQuestionnaire response. No time point given


NotesAVID conducted in people resuscitated from ventricular fibrillation or cardioverted for ventricular tachycardia. Compared implanted cardioverter defibrillator vs. antiarrhythmic drugs. Primary outcome overall mortality


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?UnclearThe AVID CTC prepared the resulting 16 sets of distinct participant survey packets. These participant-specific packets were then mailed to the study co-ordinators, who distributed them to participants as instructed

Adequate sequence generation?YesAuthor response "all patients in the study were randomised. Details of the randomisation scheme are no longer extant, but given the factorial design I think it's safe to assume that the randomization for each factor was by permuted blocks of size 2"

Blinding?YesParticipants were instructed to mail the completed surveys directly to the CTC in postage-paid, self addressed return envelopes. Only their AVID study number identified participants, assuring confidentiality of their survey responses

Free of selective outcome reporting?YesAll defined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Renfroe 2002b

Methods2 x 2 x 2 x 2 factorial randomised trial (express vs. regular post)


DataAll surviving participants at AVID study termination


Comparisons1. Overnight express delivery with cover letter signed by physician or co-ordinator with or without a certificate of appreciation 2-3 weeks after last AVID follow-up visit or 1-4 months after last AVID follow-up visit

2. Regular post with cover letter signed by physician or co-ordinator with or without a certificate of appreciation 2-3 weeks after last AVID follow-up visit or 1-4 months after last AVID follow-up visit


OutcomesQuestionnaire response. No time point given


NotesAVID conducted in people resuscitated from ventricular fibrillation or cardioverted for ventricular tachycardia. Compared implanted cardioverter defibrillator with antiarrhythmic drugs. Primary outcome overall mortality


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?UnclearThe AVID CTC prepared the resulting 16 sets of distinct participant survey packets. These participant-specific packets were then mailed to the study co-ordinators, who distributed them to participants as instructed

Adequate sequence generation?YesAuthor response "all patients in the study were randomised. Details of the randomisation scheme are no longer extant, but given the factorial design I think it's safe to assume that the randomisation for each factor was by permuted blocks of size 2"

Blinding?YesParticipants were instructed to mail the completed surveys directly to the CTC in postage-paid, self-addressed return envelopes. Only their AVID study number identified participants, assuring confidentiality of their survey responses

Free of selective outcome reporting?YesAll defined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Renfroe 2002c

Methods2 x 2 x 2 x 2 factorial randomised trial (physician vs. study co-ordinator signed cover letter)


DataAll surviving participants at AVID study termination


Comparisons1. Cover letter signed by physician sent either by express or standard post, with or without a certificate of appreciation, 2-3 weeks after last AVID follow-up visit or 1-4 months after last AVID follow-up visit

2. Cover letter signed by study co-ordinator sent either by express or standard post, with or without a certificate of appreciation 2-3 weeks after last AVID follow-up visit or 1-4 months after last AVID follow-up visit


OutcomesQuestionnaire response. No time point given


NotesAVID conducted in participants resuscitated from ventricular fibrillation or cardioverted for ventricular tachycardia. Compared implanted cardioverter defibrillator vs. antiarrhythmic drugs. Primary outcome overall mortality


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?UnclearThe AVID CTC prepared the resulting 16 sets of distinct participant survey packets. These participant-specific packets were then mailed to the study co-ordinators, who distributed them to participants as instructed

Adequate sequence generation?YesAuthor response "all patients in the study were randomised. Details of the randomisation scheme are no longer extant, but given the factorial design I think it's safe to assume that the randomisation for each factor was by permuted blocks of size 2"

Blinding?YesParticipants were instructed to mail the completed surveys directly to the CTC in postage-paid, self-addressed return envelopes. Only their AVID study number identified participants, assuring confidentiality of their survey responses

Free of selective outcome reporting?YesReport all outcomes

Other sources of bias?UnclearInsufficient information to make a judgement on this

Renfroe 2002d

Methods2 x 2 x 2 x 2 factorial randomised trial (early vs. late administration of questionnaire)


DataAll surviving participants at AVID study termination


Comparisons1. Questionnaire sent 2-3 weeks after last AVID follow-up visit by express or standard post with cover letter signed by physician or co-ordinator with or without a certificate of appreciation

2. Questionnaire sent 1-4 months after last AVID follow-up visit, by express or standard post with cover letter signed by physician or co-ordinator with or without a certificate of appreciation


OutcomesQuestionnaire response. No time point given


NotesAVID conducted in people resuscitated from ventricular fibrillation or cardioverted for ventricular tachycardia. Compared implanted cardioverter defibrillator vs. antiarrhythmic drugs. Primary outcome overall mortality


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?UnclearThe AVID CTC prepared the resulting 16 sets of distinct participant survey packets. These participant-specific packets were then mailed to the study co-ordinators, who distributed them to participants as instructed

Adequate sequence generation?YesAuthor response "all patients in the study were randomised. Details of the randomisation scheme are no longer extant, but given the factorial design I think it's safe to assume that the randomisation for each factor was by permuted blocks of size 2"

Blinding?YesParticipants were instructed to mail the completed surveys directly to the CTC in postage-paid, self-addressed return envelopes. Only their AVID study number identified participants, assuring confidentiality of their survey responses

Free of selective outcome reporting?YesAll defined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Severi 2011

MethodsTwo 2-arm randomised trials


DataUK smokers aged ≥ 16 years participating in the Txt2stop trial


ComparisonsSee 'Table of characteristics' for Severi 2011 1 and Severi 2011 2


OutcomesSee 'Table of characteristics' for Severi 2011 1 and Severi 2011 2


NotesSee 'Table of characteristics' for Severi 2011 1 and Severi 2011 2


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Unclear"Concealed from the investigators"

Adequate sequence generation?Yes"Allocated through minimisation using minim software"

Blinding?Yes"Single blind controlled trial, with those recording and assessing outcomes blind to the intervention"

Free of selective outcome reporting?YesAll defined outcomes reported

Severi 2011 1

Methods2-arm randomised trial


DataUK smokers aged ≥16participating in the Txt2stop trial


Comparisons1. Text message "Be proud of yourself for helping medical research! Thank you for filling in the txt 2 stop questionnaire" plus a fridge magnet: the message on the fridge magnet was placed in a sealed envelope, this said: "medical research is important to society" and pointed out that by taking part in TxT2stop the participants are benefiting society. Fridge magnet sent by post 16-20 weeks after randomisation. Text message sent 3 days after TxT2stop postal follow-up questionnaire

2. Text message reminding the participant the follow-up questionnaire was due 3 days after the TxT2stop questionnaire had been sent. The text message said "Thank you for filling in the TxT2stop questionnaire". Sent 3 days after the text to stop postal follow-up questionnaire


OutcomesPrimary outcome: completed follow-up questionnaires at 30 weeks from randomisation

Secondary outcome: completed follow-up questionnaires at 26 weeks from randomisation


NotesTxt2stop UK-based smoking cessation trial evaluating the effectiveness of the Txt2stop mobile phone text messaging smoking cessation programme on biochemically verified continuous smoking abstinence at 6 months. Compared Txt2stop motivational messages vs. behaviour change support to text messages unrelated to quitting


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Unclear"Concealed from the investigators"

Adequate sequence generation?Yes"Allocated through minimisation using minim software"

Blinding?Yes"Single blind controlled trial, with those recording and assessing outcomes blind to the intervention"

Free of selective outcome reporting?Yes All defined outcomes reported

Severi 2011 2

Methods2-arm randomised trial


DataUK smokers aged ≥ 16 years participating in the Txt2stop trial


Comparisons1. Telephone call from female principal investigator (senior clinician and researcher) to participants 6 weeks overdue returning specimen to increase participant follow-up plus standard Txt2stop follow-up procedures

2. Standard Txt2stop follow-up procedures


OutcomesCompleted cotinine sample follow-up at the end of May 2009 for Txt2stop (1 month after a telephone call)


NotesTxt2stop UK-based smoking cessation trial evaluating the effectiveness of the Txt2stop mobile phone text messaging smoking cessation program on biochemically verified continuous smoking abstinence at 6 months. Compared Txt2stop motivational messages vs. behaviour change support to text messages unrelated to quitting


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Unclear"Concealed from the investigators"

Adequate sequence generation?Yes"Allocated through minimisation using minim software"

Blinding?Yes"Single blind controlled trial, with those recording and assessing outcomes blind to the intervention"

Free of selective outcome reporting?YesAll defined outcomes reported

Sharp 2006a

Methods2 x 2 x 2 factorial randomised trial (trial-branded pen vs. no pen: comparison 1)


DataWomen due to receive a TOMBOLA psychosocial questionnaire during June-August 2003 for 12, 18, 24, 30, 34, 36 months' follow-up


Comparisons1. TOMBOLA-branded pen + despatch first-class post + second-class stamped preaddressed return envelope

2. No TOMBOLA-branded pen + despatch first-class post + second-class stamped preaddressed return envelope


OutcomesResponse rates at any time


NotesTOMBOLA compared 2 management policies (colposcopy vs. 6-monthly smears) in women aged 20-59 years with low-grade abnormal cervical smear. Primary outcome: cumulative incidence of cervical intraepithelial neoplasia grade ≥ 2


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Unclear2 authors did randomisation

Adequate sequence generation?Unclear"Computer randomised"

Blinding?UnclearOutcome not influenced by lack of blinding

Free of selective outcome reporting?YesExpected outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Sharp 2006a-h

Methods2 x 2 x 2 factorial randomised trial


DataWomen due to receive a TOMBOLA psychosocial questionnaire during June-August 2003 for 12, 18, 24, 30, 34, 36 months' follow-up


Comparisons1. TOMBOLA-branded pen + despatch first-class post + second-class stamped preaddressed return envelope

2. No TOMBOLA-branded pen + despatch first-class post + second-class stamped preaddressed return envelope

3. TOMBOLA-branded pen + despatch first-class post + free post business-reply envelope

4. No TOMBOLA-branded pen + despatch first-class post + free post business-reply envelope

5. TOMBOLA-branded pen + despatch second-class post + second-class stamped preaddressed return envelope

6. No TOMBOLA-branded pen + despatch second-class post + second-class stamped preaddressed return envelope

7. TOMBOLA-branded pen + despatch second-class post + free post business-reply envelope

8. No TOMBOLA-branded pen + despatch second-class post + free post business-reply envelope


OutcomesResponse rates at any time


NotesTOMBOLA compared 2 management policies (colposcopy vs. 6-monthly smears) in women aged 20-59 years with low-grade abnormal cervical smear. Primary outcome: cumulative incidence of cervical intraepithelial neoplasia grade ≥ 2


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Unclear2 authors did randomisation

Adequate sequence generation?Unclear"Computer randomised"

Blinding?UnclearOutcome not influenced by lack of blinding

Free of selective outcome reporting?YesExpected outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Sharp 2006b

Methods2 x 2 x 2 factorial randomised trial (trial-branded pen vs. no pen: comparison 2)


DataWomen due to receive a TOMBOLA psychosocial questionnaire during June-August 2003 for 12, 18, 24, 30, 34, 36 months' follow-up


Comparisons1. TOMBOLA-branded pen + despatch first-class post + free post business-reply envelope

2. No TOMBOLA-branded pen + despatch first-class post + free post business-reply envelope


OutcomesResponse rates at any time


NotesTOMBOLA compared 2 management policies (colposcopy vs. 6-monthly smears) in women aged 20-59 years with low-grade abnormal cervical smear. Primary outcome: cumulative incidence of cervical intraepithelial neoplasia grade ≥ 2


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Unclear2 authors did randomisation

Adequate sequence generation?Unclear"Computer randomised"

Blinding?UnclearOutcome not influenced by lack of blinding

Free of selective outcome reporting?YesDefined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Sharp 2006c

Methods2 x 2 x 2 factorial randomised trial (trial branded pen vs. no pen: comparison 3)


DataWomen due to receive a TOMBOLA psychosocial questionnaire during June-August 2003 for 12, 18, 24, 30, 34, 36 months' follow-up


Comparisons1. TOMBOLA-branded pen + despatch second-class post + second-class stamped preaddressed return envelope

2. No TOMBOLA-branded pen + despatch second-class post + second-class stamped preaddressed return envelope


OutcomesDefined outcomes reported


NotesTOMBOLA compared 2 management policies (colposcopy vs. 6-monthly smears) in women aged 20-59 years with low-grade abnormal cervical smear. Primary outcome: cumulative incidence of cervical intraepithelial neoplasia grade ≥ 2


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Unclear2 authors did randomisation

Adequate sequence generation?Unclear"Computer randomised"

Blinding?UnclearOutcome not influenced by lack of blinding

Free of selective outcome reporting?YesDefined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Sharp 2006d

Methods2 x 2 x 2 factorial randomised trial (trial-branded pen vs. no pen: comparison 4)


DataWomen due to receive a TOMBOLA psychosocial questionnaire during June-August 2003 for 12, 18, 24, 30, 34, 36 month follow-up.


Comparisons1. TOMBOLA-branded pen + despatch second-class post + free post business-reply envelope

2. No TOMBOLA-branded pen + despatch second-class post + free post business-reply envelope


OutcomesResponse rates at any time


NotesTOMBOLA compared 2 management policies (colposcopy vs. 6-monthly smears) in women aged 20-59 years with low-grade abnormal cervical smear. Primary outcome: cumulative incidence of cervical intraepithelial neoplasia grade ≥ 2


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Unclear2 authors did randomisation

Adequate sequence generation?Unclear"Computer randomised"

Blinding?UnclearOutcome not influenced by lack of blinding

Free of selective outcome reporting?YesDefined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Sharp 2006e

Methods2 x 2 x 2 factorial randomised trial (first vs. second-class post comparison 1)


DataWomen due to receive a TOMBOLA psychosocial questionnaire during June-August 2003 for 12, 18, 24, 30, 34, 36 months' follow-up.


Comparisons1. TOMBOLA-branded pen + despatch first-class post + second-class stamped preaddressed return envelope

2. TOMBOLA-branded pen + despatch second-class post + second-class stamped preaddressed return envelope


OutcomesDefined outcomes reported


NotesTOMBOLA compared 2 management policies (colposcopy vs. 6-monthly smears) in women 20-59 years with low-grade abnormal cervical smear. Primary outcome: cumulative incidence of cervical intraepithelial neoplasia grade ≥ 2


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Unclear2 authors did randomisation

Adequate sequence generation?Unclear"Computer randomised"

Blinding?UnclearOutcome not influenced by lack of blinding

Free of selective outcome reporting?YesDefined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Sharp 2006f

Methods2 x 2 x 2 factorial randomised trial (first vs. second-class post comparison 2)


DataWomen due to receive a TOMBOLA psychosocial questionnaire during June-August 2003 for 12, 18, 24, 30, 34, 36 months' follow-up


Comparisons1. No TOMBOLA-branded pen + despatch first-class post + second-class stamped preaddressed return envelope

2. No TOMBOLA-branded pen + despatch second-class post + second-class stamped preaddressed return envelope


OutcomesResponse rates at any time


NotesTOMBOLA compared 2 management policies (colposcopy vs. 6-monthly smears) in women 20-59 years with low-grade abnormal cervical smear. Primary outcome: cumulative incidence of cervical intraepithelial neoplasia grade ≥ 2


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Unclear2 authors did randomisation

Adequate sequence generation?Unclear"Computer randomised"

Blinding?UnclearOutcome not influenced by lack of blinding

Free of selective outcome reporting?YesDefined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Sharp 2006g

Methods2 x 2 x 2 factorial randomised trial (first vs. second-class comparison 3)


DataWomen due to receive a TOMBOLA psychosocial questionnaire during June-August 2003 for 12, 18, 24, 30, 34, 36 months' follow-up


Comparisons1. TOMBOLA-branded pen + despatch first-class post + second-class stamped preaddressed return envelope

2. TOMBOLA-branded pen + despatch first-class post + free post business-reply envelope


OutcomesResponse rates at any time


NotesTOMBOLA compared 2 management policies (colposcopy vs. 6-monthly smears) in women aged 20-59 years with low-grade abnormal cervical smear. Primary outcome: cumulative incidence of cervical intraepithelial neoplasia grade ≥ 2


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Unclear2 authors did randomisation

Adequate sequence generation?Unclear"Computer randomised"

Blinding?UnclearOutcome not influenced by lack of blinding

Free of selective outcome reporting?YesDefined outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Sharp 2006h

Methods2 x 2 x 2 factorial randomised trial (first vs. second-class comparison 4)


DataWomen due to receive a TOMBOLA psychosocial questionnaire during June-August 2003 for 12, 18, 24, 30, 34, 36 month follow-up.


Comparisons1. No TOMBOLA-branded pen + despatch first-class post + second-class stamped preaddressed return envelope

2. No TOMBOLA-branded pen + despatch first-class post + free post business-reply envelope


OutcomesResponse rates at any time


NotesTOMBOLA compared 2 management policies (colposcopy vs. 6-monthly smears) in women 20-59 years with low-grade abnormal cervical smear. Primary outcome: cumulative incidence of cervical intraepithelial neoplasia grade ≥ 2


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?Unclear2 authors did randomisation

Adequate sequence generation?Unclear"Computer randomised"

Blinding?UnclearOutcome not influenced by lack of blinding

Free of selective outcome reporting?YesExpected outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement on this

Subar 2001

Methods2-arm randomised trial


DataParticipants aged 55-74 years from 3 centres participating in PLCO trial (control arm)


Comparisons1. Diet history questionnaire DHQ (36-page food frequency questionnaire) plus 1-page questionnaire on time taken to complete questionnaire

2. PLCO food frequency questionnaire (16-page food frequency questionnaire) plus 1-page questionnaire on time taken to complete questionnaire


OutcomesResponse rate. No time point given


NotesPLCO trial compared digital rectal examination, transvaginal ultrasound and chest x-ray at baseline and at 5 years vs. usual follow-up. PSA and cancer antigen CA125 at baseline, and annually for 5 years. Primary outcome: mortality from PLCO cancers


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?NoAuthor response "none known"

Adequate sequence generation?UnclearAuthor response "at each centre half were randomised to received". Authors contacted "we would have likely used some computer generated randomisation scheme"

Blinding?UnclearAuthor response "the respondents were not blinded. Not known if personnel were 'blinded'"

Free of selective outcome reporting?YesExpected outcome reported

Other sources of bias?UnclearInsufficient information to make a judgement about this

Sutherland 1996

Methods2-arm randomised trial


Data226 women taking part in the feasibility study for the Canadian diet and breast cancer prevention feasibility trial


Comparisons1. Total design method for postal follow-up: white envelope with hospital logo and commemorative stamp; headed notepaper; reply self addressed stamped envelope enclosing contents, hand signature on letters. Postcard sent after 7 days, reminders sent twice

2. Customary method for postal follow-up: brown envelope with return address stamped on, computer-printed labels, no signature on letter, reply self addressed stamped envelope folded and inserted behind forms, no reminder


OutcomesTime to return of questionnaire at 70 days


NotesCanadian diet and breast cancer prevention feasibility trial compared teaching women aged over 30 years of age with at least 50% of breast volume occupied by radiological changes of dysplasia how to reduce dietary fat to a level of 15% of calories vs. no teaching. Primary outcome recruitment and retention


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?UnclearNot clear

Adequate sequence generation?YesAuthor response "computer generated random numbers"

Blinding?UnclearAuthor response "personel knew the allocation a subject had received but their only contact with subjects was the follow-up phone call in some allocated to the "customary" method"

Free of selective outcome reporting?YesExpected outcome reported

Other sources of bias?UnclearInsufficient information to make a judgement about this

Svoboda 2001

Methods2-arm randomised trial


DataAdults ≥ 16 years with head injury in the CRASH trial (Czech)


Comparisons1. 1-page, 7-question functional dependence questionnaire sent with a covering letter and a stamped return envelope. Reminders sent after 4 and 8 weeks

2. 3-page, 16-question functional dependence questionnaire sent with a covering letter and a stamped return envelope. Reminders sent after 4 and 8 weeks


OutcomesNumber of questionnaires returned within 3 months


NotesNumbers randomised and responded provided by authors. Primary outcome for the CRASH trial death from any cause within 2 weeks of injury and death or disability at 6 months


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?YesAuthor response "central computer"

Adequate sequence generation?YesAuthor response "random allocation central computer"

Blinding?UnclearAuthor response "the questionnaires were packaged and sent to patients by personnel who were independent of the study"

Free of selective outcome reporting?YesExpected outcome reported

Other sources of bias?UnclearInsufficient information to make a judgement about this

Tai 1997

Methods2-arm randomised trial


DataAll host study participants lost to follow-up in the evaluation of general practice computer templates trial


Comparisons1. Recorded delivery reminder letter with QoL questionnaire included, sent once

2. Telephone calls repeated up to 3 times at 10-12 a.m. or 2-5.30 p.m. Message left on answering machine after third call


OutcomesNumber of questionnaires returned. No time point given


NotesEvaluation of general practice computer templates cluster randomised trial compared computerised templates for asthma and diabetes management in general practice. Primary outcome frequency of use of computer templates assessed by examining computerised records of those who responded to QoL questionnaires


Risk of bias

ItemAuthors' judgementDescription

Allocation concealment?YesAuthor response "computer generated by statistician independent to trial manager"

Adequate sequence generation?UnclearAuthor response "computer generated by statistician"

Blinding?UnclearStudy personnel were not blind to the intervention the participants received

Free of selective outcome reporting?YesExpected outcomes reported

Other sources of bias?UnclearInsufficient information to make a judgement about this

 
Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion

Arnevik 2009This retention trial was not embedded in a randomised trial

Atherton 2010Comparison of Internet vs. postal questionnaires not randomised

Barry 1996Retention trial compared distribution of scores for subjects completing different questionnaire versions. Author confirmed retention/questionnaire return was not an outcome measure

Bednarek 2008Retention trial outcome is continuation of treatment

Cox 2003Retention trial outcome treatment compliance

Day 1998Retention trial measured adherence to treatment. Authors do not have retention data

Eaker 2004Retention trial embedded in a cohort

Edelstein 2005Retention study is not a randomised trial. Incentives not randomised. Author confirmed these were not instituted to help with retention but with adherence to pill taking and life style modification requirements

Grabowski 1995Substudy aim is retention in treatment comparing different follow-up schedules for addiction treatment trial

Hall 1975Not a randomised/quasi-randomised retention trial

Hall 1978Not a randomised/quasi-randomised retention trial

Hoffman 1998Retention trial embedded in a blood bank cohort

Hopkins 1983Retention trial embedded in a survey

Iglesias 2000Retention trial embedded in a cohort of general practitioner practice participants

Iglesias 2001Retention trial embedded in the recruitment phase of the host trial

Johnson 2004Retention study not embedded in a randomised trial

Katz 2001Retention study is not a randomised trial. Authors confirmed the effectiveness of gift incentives was not evaluated in a substudy for the Pride in Parenting trial

Leidy 2000Retention study appears to be a randomised trial but no response from authors to establish if retention was an outcome. For the substudy, trial sites randomised to 1 of 2 orders of administration of quality of life questionnaires. Response rates not reported. Missing data, internal consistency reliability, mean score values, relationship between the 2 measures evaluated

Marsh 1999Host study was not a randomised trial. "Practices were randomly allocated to the intervention group using random number tables. Each intervention practice matched with one control practice"

McAuley 1994Retention study is not a randomised trial. There is a single randomisation stratified by classes in the morning and early evening. No response from authors regarding randomisation to class times

McBee 2009Retention study not a randomised trial. Authors confirm strategies to improve retention were not evaluated in an Age-Related Eye Disease Study 2 (AREDS2) substudy

Poling 2006Substudy aim is about diagnostic compliance. 4-arm trial comparing contingency management with or without active bupropion and voucher control with or without active bupropion. Here contingency management and voucher control are aimed at getting information on the disease condition/response to treatment for the primary outcome of the host trial i.e. negative urine sample for cocaine and opioids. Contingency management and voucher control are not related to retention in the host trial but related to diagnostic compliance

Puffer 2004Retention RCT was embedded in a survey. Authors confirmed that the 2 x 2 factorial study testing four different questionnaire designs was embedded in a survey.

Rhoades 1998Substudy retention in treatment. 2 x 2 trial of dose and visit frequency of attending a clinic either 2 or 5 days per week. Primary outcome was retention in treatment for all randomizations. Similar to Grabowski 1995 trial

Roberts 2000Retention trial embedded in a survey about menopause services

Schmitz 2005Substudy about compliance to treatment and pill taking behaviour rather than trial retention

Smeeth 2001abSubstudy about response to baseline assessment

Stoner 1998Retention study was not a randomised trial. Host study was a cluster randomised trial. Effectiveness of vouchers not evaluated in a substudy

Tassopoulos 2007Not a retention randomised trial

Wu 1997Substudy designed to evaluate whether scores are different using 3 modes of questionnaire administration, rather than retention

 
Characteristics of ongoing studies [ordered by study ID]
Land 2007 2

Trial name or titleRandomised Evaluation of NSABP BAHO Compliance Initiatives

MethodsRandomised trial embedded in Protocol B36

DataParticipants or trial sites/institutions participating in the B36 trial. Pre- and postmenopausal women aged > 18 years, with histologically confirmed invasive breast adenocarcinoma

ComparisonsMenstrual history calendar

Usual procedures

OutcomesTo improve form submission compliance in Behavioural and Health Outcomes (BAHO) protocols

Starting dateUnclear

Contact informationLand@nsabp.pitt.edu

NotesStudy identified through personal correspondence with author

B36: US-based comparison of 2 combination chemotherapy regimens to treat women with breast cancer

Mitchell

Trial name or titleA Randomised Controlled Trial of Combined Pre-Contact and Participant Update for Increasing Questionnaire Response Rates in Older Women

MethodsRandomised trial embedded in the SCOOP screening trial

DataWomen aged 70-85 years not receiving treatment for osteoporosis

Comparisons1. Prenotification to complete the 2-year follow-up questionnaire plus a study update

2. No study update

OutcomesResponse to postal questionnaires at 2-year follow-up

Starting date2010

Contact informationnatasha.mitchell@york.ac.uk

NotesStudy identified through mail out to UK clinical trials units

SCOOP: UK-based pragmatic randomised controlled trial of the effectiveness of screening for osteoporosis in older women for the prevention of fractures. 10-year absolute risk of fracture calculated from a World Health Organization algorithm based on screening questionnaire data, x-ray and dual-energy x-ray absorptiometry scan results compared with usual care

 
Comparison 1. Addition of incentive vs none: main analysis

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Retention14Risk Ratio (M-H, Fixed, 95% CI)Subtotals only

    1.1 Addition of monetary incentive
33166Risk Ratio (M-H, Fixed, 95% CI)1.18 [1.09, 1.28]

    1.2 Addition of offer of monetary incentive/prize draw
23613Risk Ratio (M-H, Fixed, 95% CI)1.25 [1.14, 1.38]

    1.3 Addition of non-monetary incentive
66322Risk Ratio (M-H, Fixed, 95% CI)1.00 [0.98, 1.02]

    1.4 Addition of offer of non-monetary incentive
21138Risk Ratio (M-H, Fixed, 95% CI)0.99 [0.95, 1.03]

    1.5 Addition of offer of monetary donation to charity
1815Risk Ratio (M-H, Fixed, 95% CI)1.02 [0.78, 1.32]

 
Comparison 2. Addition of incentive: sensitivity analysis: quasi-randomised trials removed

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Retention7Risk Ratio (M-H, Fixed, 95% CI)Subtotals only

    1.1 Addition of monetary incentive
21022Risk Ratio (M-H, Fixed, 95% CI)1.31 [1.11, 1.55]

    1.2 Addition of non-monetary incentive
51594Risk Ratio (M-H, Fixed, 95% CI)1.00 [0.93, 1.08]

 
Comparison 3. Addition of incentive: separating research arms of non-factorial trials (three-/four-arm trials)

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Retention14Risk Ratio (M-H, Fixed, 95% CI)Subtotals only

    1.1 Addition of monetary incentive
33066Risk Ratio (M-H, Fixed, 95% CI)1.17 [1.09, 1.27]

    1.2 Offer of monetary incentive
34224Risk Ratio (M-H, Fixed, 95% CI)1.24 [1.13, 1.37]

    1.3 Addition of non-monetary incentive
810793Risk Ratio (M-H, Fixed, 95% CI)1.00 [0.98, 1.01]

 
Comparison 4. Addition of telephone follow-up vs incentive

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Retention1Risk Ratio (M-H, Fixed, 95% CI)Subtotals only

    1.1 Telephone survey vs. monetary incentive and questionnaire
1700Risk Ratio (M-H, Fixed, 95% CI)1.08 [0.94, 1.24]

 
Comparison 5. Addition of monetary incentive to both study arms

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Retention2902Risk Ratio (M-H, Fixed, 95% CI)1.12 [1.04, 1.22]

    1.1 Addition of GBP10 plus offer of GBP10 vs. addition of GBP5 plus offer of GBP5
1485Risk Ratio (M-H, Fixed, 95% CI)1.16 [1.04, 1.30]

    1.2 Addition of GBP20 voucher offer vs. addition of GBP10 voucher offer
1417Risk Ratio (M-H, Fixed, 95% CI)1.08 [0.97, 1.21]

 
Comparison 6. Addition of monetary incentive vs offer of incentive

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Retention2297Risk Ratio (M-H, Fixed, 95% CI)1.04 [0.91, 1.19]

    1.1 Addition of monetary incentive vs. offer of entry into prize draw
2297Risk Ratio (M-H, Fixed, 95% CI)1.04 [0.91, 1.19]

 
Comparison 7. Enhanced letter versus standard letter: main analysis

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Retention2Risk Ratio (M-H, Fixed, 95% CI)Subtotals only

    1.1 Enhanced letter vs. standard letter
22479Risk Ratio (M-H, Fixed, 95% CI)1.01 [0.97, 1.05]

 
Comparison 8. Communication strategies letter: total design method

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Retention1Risk Ratio (M-H, Fixed, 95% CI)Subtotals only

    1.1 Total design method for postal questionnaires vs. customary method
1226Risk Ratio (M-H, Fixed, 95% CI)1.43 [1.22, 1.67]

 
Comparison 9. Communication strategies post: main analysis

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Retention7Risk Ratio (M-H, Fixed, 95% CI)Subtotals only

    1.1 Priority vs. regular post
71888Risk Ratio (M-H, Fixed, 95% CI)1.02 [0.95, 1.09]

 
Comparison 10. Communication strategies: additional reminder vs usual follow-up: main analysis

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Retention6Risk Ratio (M-H, Fixed, 95% CI)Subtotals only

    1.1 Additional reminder vs. usual follow-up procedures
63401Risk Ratio (M-H, Fixed, 95% CI)1.03 [0.99, 1.06]

 
Comparison 11. Communication strategies additional reminder to trial site vs usual reminder (ICC 0.054)

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Retention1Risk Ratio (Fixed, 95% CI)Subtotals only

    1.1 Monthly reminder of upcoming assessments to trial site vs. usual reminders
1272Risk Ratio (Fixed, 95% CI)0.96 [0.83, 1.11]

 
Comparison 12. Communication strategies: questionnaire administered early vs late

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Retention1Risk Ratio (M-H, Fixed, 95% CI)Subtotals only

    1.1 Early vs. late administration
1664Risk Ratio (M-H, Fixed, 95% CI)1.10 [0.96, 1.26]

 
Comparison 13. Communication strategies: type of reminder: main analysis

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Retention1Risk Ratio (M-H, Fixed, 95% CI)Subtotals only

    1.1 Recorded delivery vs. telephone reminder
1192Risk Ratio (M-H, Fixed, 95% CI)2.08 [1.11, 3.87]

 
Comparison 14. Questionnaire strategies: new vs standard questionnaire: main analysis

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Retention10Risk Ratio (M-H, Fixed, 95% CI)Subtotals only

    1.1 Short vs. long questionnaire
57277Risk Ratio (M-H, Fixed, 95% CI)1.04 [1.00, 1.08]

    1.2 Long and clear vs. short and condensed questionnaires
1900Risk Ratio (M-H, Fixed, 95% CI)1.01 [0.95, 1.07]

    1.3 Question order: condition first vs. generic first questions
29435Risk Ratio (M-H, Fixed, 95% CI)1.00 [0.97, 1.02]

    1.4 Questionnaire: relevant vs. less relevant to condition
23893Risk Ratio (M-H, Fixed, 95% CI)1.07 [1.01, 1.14]

 
Comparison 15. Questionnaire strategies: new vs standard questionnaire: sensitivity analysis quasi-randomised trial McColl

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Retention8Risk Ratio (M-H, Fixed, 95% CI)Subtotals only

    1.1 Short vs. long questionnaire
57277Risk Ratio (M-H, Fixed, 95% CI)1.04 [1.00, 1.08]

    1.2 Long and clear vs. short and condensed questionnaires
1900Risk Ratio (M-H, Fixed, 95% CI)1.01 [0.95, 1.07]

    1.3 Questionnaire: relevant vs. less relevant to condition
23893Risk Ratio (M-H, Fixed, 95% CI)1.07 [1.01, 1.14]

 
Comparison 16. Behavioural strategies: main analysis

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Retention2Risk Ratio (M-H, Fixed, 95% CI)Subtotals only

    1.1 Motivation vs. information
2273Risk Ratio (M-H, Fixed, 95% CI)1.08 [0.93, 1.24]

 
Comparison 17. Case management

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Retention1Risk Ratio (M-H, Fixed, 95% CI)Subtotals only

    1.1 Case management vs. usual follow-up
1703Risk Ratio (M-H, Fixed, 95% CI)1.00 [0.97, 1.04]

 
Comparison 18. Methodology strategies

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Retention1Risk Ratio (M-H, Fixed, 95% CI)Subtotals only

    1.1 Open vs. blind trial design
1538Risk Ratio (M-H, Fixed, 95% CI)1.37 [1.16, 1.63]

 
Table 1. Trials evaluating incentive strategies

Trial or trial comparisonIncentive groupsControl groupOutcome type

Addition of incentive vs. none

Bauer 2004aba) USD10 cheque

b) USD2 cheque

Arms combined for main analysis
No incentiveDNA specimen kit return plus postal questionnaire response 

Gates 2009GBP5 voucherNo incentivePostal questionnaire response

Kenyon 2005GBP5 voucherNo incentivePostal questionnaire response

Khadjesari 2011 1aca) Offer GBP5 voucher

c) Offer of entry into GBP250 prize draw, groups combined for main analysis
No incentiveInternet-based questionnaire response

Khadjesari 2011 2Offer of GBP10 Amazon.co.uk voucherNo incentiveInternet-based questionnaire response

Bowen 2000abca) Certificate

b) Pin

c) Pin and certificate groups combined for main analysis
No incentiveParticipants retention

Renfroe 2002a Certificate of appreciationNo certificate of appreciationPostal questionnaire response

Sharp 2006aPenNo penPostal questionnaire response

Sharp 2006bPenNo penPostal questionnaire response

Sharp 2006cPenNo penPostal questionnaire response

Sharp 2006dPenNo penPostal questionnaire response

Cockayne 2005Offer of study resultsNo offerPostal questionnaire response

Hughes 1989Offer of free reprint of resultsNo offerPostal questionnaire response

Khadjesari 2011 1bOffer of GBP5 charity donationNo offerInternet-based questionnaire response

Addition of monetary incentive to both groups

Bailey 1 unpublishedOffer of GBP20 shopping voucherOffer of GBP10 shopping voucherPostal questionnaire response

Bailey 2 unpublishedShopping voucher: GBP10 in advance and GBP10 on data returnShopping voucher: GBP5 in advance and GBP5 on data returnQuestionnaire response and chlamydia kit return

Addition of monetary incentive vs. offer of incentive

Kenton 2007aUSD2 coinDraw for USD50 gift voucherPostal questionnaire response

Kenton 2007bUSD2 coinDraw for USD50 gift voucherPostal questionnaire response

Offer of prize draw vs. no offer

Leigh Brown 1997Aware of monthly prize draw of GBP25 gift voucherNo offer of drawPostal questionnaire response

 
Table 2. Trials evaluating communication strategies

Trial or trial comparisonCommunication strategyControl armOutcome type

Enhanced letter vs. standard letter

Renfroe 2002cCover letter signed by physicianCover letter signed by co-ordinatorPostal questionnaire response

Marson 2007Letter explaining the approximate length of time to complete questionnaireStandard letterPostal questionnaire response

Total design method vs. customary method

Sutherland 1996Total design method for postal follow-upStandard method for postal follow-upPostal questionnaire response

Priority vs. regular post

Renfroe 2002bExpress deliveryStandard deliveryPostal questionnaire response

Sharp 2006eDespatch first-class stampDespatch second-class stampPostal questionnaire response

Sharp 2006fDespatch first-class stampDespatch second-class stampPostal questionnaire response

Sharp 2006gSecond-class return envelopeFree post return envelopePostal questionnaire response

Sharp 2006hSecond-class return envelopeFree post return envelopePostal questionnaire response

Kenton 2007cPriority mailStandard mailPostal questionnaire response

Kenton 2007dPriority mailStandard mailPostal questionnaire response

Additional reminder vs. usual follow-up

Ashby 2011Electronic reminder No electronic reminder                          Postal questionnaire response

MacLennan unpublishedTelephone reminderNo telephone reminderPostal questionnaire response

Nakash 2007Trial calendar given at recruitment with questionnaire due datesNo calendarPostal questionnaire response

Severi 2011 1Text message and fridge magnet both emphasising social benefits of study participation.Text message reminder sent 3 days after questionnairePostal questionnaire response

Severi 2011 2Telephone reminder from principle investigatorStandard procedures.Return of cotinine samples

Man 2011SMS text message as follow-up questionnaire sent outNo SMS text messagePostal questionnaire response

Additional trial site reminder vs. usual reminder

Land 2007Prospective monthly reminder of upcoming assessments to trial sitesNo extra reminder to trial sitesPostal questionnaire response

Early vs. late administration of questionnaire

Renfroe 2002dQuestionnaire sent 2-3 weeks after last AVID follow-up visitQuestionnaire sent 1-4 months after last AVID follow-up visitPostal questionnaire response

Recorded delivery vs. telephone reminder

Tai 1997Recorded delivery reminderTelephone reminderPostal questionnaire response

Addition telephone follow-up vs. incentive

Couper 2007Telephone survey by trained interviewerPostal questionnaire and USD5 billPost and questionnaire response

 AVID: Antiarrhythmics Versus Implantable Defibrillators; SMS: short message service.
 
Table 3. Trials evaluating new questionnaire strategies

Trial or trial comparisonQuestionnaire strategyControl armOutcome type

Short vs. long

Dorman 1997Short EuroQolLong SF-36 questionnaire.Postal questionnaire response

Edwards 2001 unpublished1-page, 7-question functional dependence questionnaire 3-page, 16-question functional dependence questionnairePostal questionnaire response

Svoboda 2001 unpublished1-page, 7-question functional dependence questionnaire 3-page, 16-question functional dependence questionnairePostal questionnaire response

McCambridge 2011 1bAUDIT Short

+

LDQ
APQInternet-based questionnaire response

McCambridge 2011 2bAUDIT Short

+

LDQ
APQInternet-based questionnaire response

Long and clear vs. short and condensed

Subar 2001DHQ (36-page food frequency questionnaire)PLCO (16-page food frequency questionnaire)Postal questionnaire response and onsite completion  

 

Question order

McColl 2003 1Asthma condition-specific questions first followed by generic

 
Generic questions followed by condition specificPostal questionnaire

response

McColl 2003 2Angina condition-specific questions followed by genericGeneric questions followed by condition specificPostal questionnaire response

Letley 2000 unpublishedRDQ at front and SF-36 at backSF-36 at front RDQ at backPostal questionnaire response

Relevance of questionnaire

McCambridge 2011 1aAPQ23 itemsCORE-OM Mental health assessment 23/34 itemsInternet-based questionnaire response

McCambridge 2011 2aAUDIT Short

+

LDQ
CORE-OM Mental health assessment 10 itemsInternet-based questionnaire response

 APQ: Alcohol Problems Questionnaire; AUDIT: Alcohol Use Disorders Identification Test; LDQ Leeds Dependency Questionnaire; PLCO: Prostate, Lung, Colorectal, Ovarian; SF-36: Short Form 36 item.
 
Table 4. Gain in number of questionnaires returned per 1000 questionnaires sent

Example of proportion of questionnaires

returned in control arm 
  30%40%50%60%70%80%90%

Strategy to improve retention  RR1

RR
       

Addition of monetary incentive versus none1.180.847107927661532

Addition of offer of monetary incentive/prize draw versus none1.250.80014012010080604020

Addition of higher value monetary incentive versus addition of lower amount1.120.89077665544332211

 RR: risk ratio.