Table 1 Details of included studies

Akl et al., 2011 [41]

Assess the effect of 1) tracking responses and 2) day of mailing (Monday vs Friday) on physician survey response rate

Directors of Family Medicine residency programs (ie GPs)

Training of residents in implementation of clinical guidelines

Two by two factorial randomized design

456

No tracking n = 228, Monday mailing n = 228

Tracking n = 228, Friday mailing n = 228

USA

Asch et al., 1998 [26]

To evaluate the effect of incentive size by providing physicians either a $5 bill or a $2 bill.

Primary care physicians identified from the American Medical Association Physician Master File

Attitudes about cost containment in cancer screening.

RCT

1000

$2 cash incentive received in initial mailing n = 500

$5 cash incentive received in initial mailing n = 500

USA

Baron et al., 2001 [29]

To evaluate the cost effectiveness of a lottery on physicians response rates to a mail survey.

Family Doctors identified from membership list of the Quebec Federation of GPs

The determinants of influenza immunization among family physicians in Quebec.

RCT

1000

No lottery n = 500

Lottery n = 500

Canada

Bonevskiet al., 2011 [40]

To examine the efficacy of two strategies for improving general practitioner response to a survey and a secondary aim was to assess GPs self reported preferred mode of survey administration.

GPs practising in NSW selected from the Australasian Medical Publishing Company database.

The knowledge, attitudes and practices of GPs around vitamin D.

RCT

334 (Cover letter trial), 1166 (Telephone reminder trial)

Standard cover letter N = 167, No telephone reminder n = 576

Standard letter plus local division of general practice cover letter n = 167, Telephone reminder n = 590

Australia

Deehan et al., 1997 [11]

To explore the effects of financial and non-financial inducements on response rates to increase the overall response rate.

A random sample of all GP’s in England and Wales.

Survey on GP’s work and attitudes with alcohol misusing patients.

Five arm parallel trial design with randomization

3584

No inducement n = 1188

£5 charity donation n = 607, £10 charity donation n = 578, £5cash payment n = 613, £10 cash payment n = 598, (Total for all interventions n = 2396)

UK

Drummond et al., 2008 [34]

To investigate the effect of two low cost interventions (pre-contact and questionnaire order) on response to a primary care physician postal survey

Primary care physicians working in Ireland identified from various sources.

A national survey assessing views and practices of physicians regarding prostate specific antigen (PSA) testing.

Two by two factorial randomized design

1599

No pre-contact n = 743, Version 2 survey (Topic specific questions first) n = 744

Pre-contact n = 715, Version 1 survey (general questions first) n = 714

Ireland

Everett et al., 1997 [28]

To examine the effectiveness of $1 incentive on response rates

Family Physicians

Firearm-safety counseling beliefs

RCT

600

No incentive n=300

$1 bill n=300

USA

Gattellari et al., 2012 [13]

Assess effectiveness of two response-aiding strategies in a postal survey

GPs known to be in current practice

Management of nonvalvular atrial fibrillation

Two by two factorial randomized control trial

1000

Mail prompt n = 500, Coloured seal without text n = 500

Fax prompt n = 500, Coloured seal + text n = 500

Australia

Gupta et al., 1997 [39]

To determine the effectiveness of a telephone prompt by a medical researcher compared to a nonmedical research assistant in improving survey response rate of GP’s, and to compare personnel costs.

A national random sample of Australian GPs

A national survey assessing GPs views of clinical practice guidelines

RCT

404

Telephone prompt by a medical researcher n = 202

Telephone prompt by experienced nonmedical research assistant n = 202

Australia

Hocking et al., 2006 [35]

To compare GP response to a telephone interview with response to a postal survey with three reminders.

A random sample of Victorian GPs

Survey to assess knowledge and diagnostic and management practices of genital chlamydia infection.

RCT

867

Postal survey n = 451

Telephone interview n = 416

Australia

James et al., 2011 [27]

To study the effects of payment timing, form of payment and requiring a social security number (SSN) on survey response rates.

Practising US physicians ages 65 and under representing all specialties (including primary care), selected from the AMA Physician Masterfile

Survey on ethical and moral beliefs.

Four arm parallel design with randomization

443

Promised $US25 check requiring SSN n = 102

Immediate $US25 cash n = 129, Immediate $US25 check n = 97, Promised $US25 check not requiring SSN n = 115

USA

McLaren and Shelley 2000 [30]

To compare effect of a primer telephone call and postcard on GP response rate; and to compare the effect of informing GPs about a prize for participating in survey compared to not informing GPs of the prize.

Victorian GPs

Management issues surrounding early pregnancy bleeding and miscarriage

Two by two factorial design with randomization

621

Telephone n=305 No prize notification N=310

Postcard n=316Prize notification n=311

Australia

Maheux et al., 1989 [42]

To determine the response rates of Quebec physicians by sending a handwritten thank you letter and by sending a more personalized mailout package.

Quebec Physicians

To determine the level of physicians support for a number of patient care issues.

RCT

Second mailing N = 356, Third mailing N = 253

No handwritten postscript n = 186, Non-personalised mailout n = 127

Handwritten postscript n = 170, Personalised mailout n = 126

Canada

Olivarius and Andreasen, 1995 [43]

To test a possible day the week effect on doctors response rate to a postal questionnaire.

GPs

A nationwide survey on doctors’ attitude on the relative importance of general practice and the specialties in the treatment of general diseases and diseases commonly regarded as belonging to the specialties.

RCT

200

Dispatch of survey on Thursday (for receipt before weekend) n = 98

Dispatch of survey on Saturday (for receipt after weekend) n = 102

Denmark

Pedrana et al., 2008 [44]

To compare response rates to postal surveys sent by normal post and registered mail and to assess the cost implications of the two mailing methods.

General practitioners in Victoria.

To assess antenatal screening practices in Australia

RCT

1550

Normal mail n = 775

Registered mail n = 775

Australia

Pirotta et al., 1999 [38]

To measure the effect of a primer postcard to improve mailed survey response rates.

Victorian GPs who had at least 1500 consultations in 1995-96

General Practitioners attitudes to complementary therapies.

RCT

800

No primer postcard n = 400

Primer postcard n = 400

Australia

Pit et al., 2013 [25]

To assess the effectiveness of small non-conditional non-financial incentive (ie attractive pen) on survey response rates

Practicing GPs in Northern Rivers region of NSW

How to improve sustainable employment of ageing GPs in rural Australia

RCT

125

No pen incentive n = 62

Pen incentive n = 63

Australia

Robertson et al., 2005 [31]

To test the effect of a $AUD2 scratch lottery ticket on response rates

A random sample GP’s and specialists from the medical directory of Australia.

To explore the uptake of 19 new drugs into the clinical practices of Australian GPs and specialists.

RCT

464

No lottery n = 232

AUD$2 scratch lottery n = 232

Australia

Scott et al., 2011[24]

To compare effects and costs of three different modes of survey administration in a national survey of Doctors

A stratified random sample of doctors undertaking clinical practice from a national directory of all doctors in Australia. Stratification was by doctor type and rural/remote category.

Study of the dynamics of the medical labour market in Australia, focusing on workforce participation and its determinants among Australian doctors.

3-arm parallel trial design with randomisation

1091

Online survey, followed by reminder letter with login details n = 369

Sequential mixed mode -online survey, followed by reminder letter with paper survey n = 388, Simultaneous mixed mode -paper survey and login details sent together, with reminder letter with login details n = 334

Australia

Seguin et al., 2004 [37]

To compare email with regular mail for conducting surveys of family physicians.

A random sample of physicians listed in the college of family physicians of Canada’s database who had an email address

Survey on use of sildenafil citrate (Viagra)

RCT

2397

Survey delivered by post to the physicians without email n = 800

Survey delivered by email n = 798, Survey delivered by post to the physicians with e mail n = 799

Canada

Shosteck and Fairweather, 1979 [36]

Comparison of mail and personal administration of questionnaire among physicians.

Office-based primary care physicians who were in active practice within the States of Washington DC and Maryland, and treating upper respiratory or lower urinary tract infections selected from the AMA database.

Physician antibiotic prescription practices.

RCT

543

Mail technique n = 296

Personal interview technique n = 247

USA

Thomson et al., 2004 [32]

To maximise the response to a postal questionnaire and to test the most effective form of incentive.

Practicing GP’s selected from the Lothian Primary Care NHS Trust database.

GP’s attitudes to the management of ingrowing toenails.

RCT

568

Lottery of one chance to win six bottles of champagne n = 286

Lottery of six chances to win one bottle of champagne n = 282

UK

Ward et al., 1998 [33]

To evaluate response aiding strategies feasible in large surveys

Randomly selected sample of general practitioners

Cancer screening

4-arm parallel trial design with randomisation

1550

-

Doctor advance phone prompt: n = 249, Pen prompt: n = 261, Letter prompt: n = 260, Non-doctor advance phone prompt: n = 501

Australia