Table 2 Expertise-based design methodology and related reporting (n = 43 unless otherwise stated)

Expertise-based design type

 Pure

38

88

 Hybrid

3

7

 Unclear

2

5

Name used (n = 24)

 Expertise-based

2

8

 Double randomisation

2

8

 Randomised to surgeon

2

8

 Non-randomised surgeon design

1

4

 Randomised-surgeon

1

4

 Surgeon-randomised

2

8

 None

15

63

Reporting of expertise-based design in abstract

 Design name

6

14

 Deliverers of interventions stated to be different

9

21

 Details regarding health professionals delivering one intervention

7

16

 Details regarding health professional delivering both interventions

6

14

 No details

15

35

Reported advantages (n = 20)

 Ensuring intervention was delivered by someone with expertise to avoid criticism of the study

2

5

 Balance of health professionals (e.g., interest, commitment, and prior knowledge of intervention)

4

9

 Randomisation of health professional ‘consistent with efficacy trial’

1

2

 Following preference will reduce non-compliance

1

2

 Using randomisation of health professional strengthens generalisability of findings

1

2

 Eliminates learning of the intervention

3

7

 Eliminates ethical concerns with intervention deliverer not doing what they would do outside of the trial

1

2

 Delivery of intervention maximised (and may reduce adverse events)

1

2

 Ensures experience in control group

1

2

 Reduces cross-over between group compared with conventional study

2

5

 Avoid non-compliance with allocation because of non-familiarity

1

2

 Health professionals delivering their preferred intervention

1

2

 Following usual practice reduces non-compliance with allocation

1

2

Reported disadvantages (n = 9)

 Health professionals delivering interventions may not be representative of practice

1

2

 Health professionals delivering interventions may not be balanced (e.g., motivation and prior experience) unless selected

4

9

 Delivery may vary in other ways between groups because of different health professionals delivering the interventions

2

5

 Disagreement between recruiter and health professional delivering the intervention regarding eligibility led to the intervention not being performed in some cases

1

2

 Addition of new intervention and deliverer may create expectation bias

1

2

Allocation of intervention deliverers

 Randomised

8

19

 Usual practice

5

12

 Preference

4

9

 Defined by research question

1

2

 Not stated

25

58

Criteria for delivering intervention 1

 Number of prior cases

2

4

 Number of years of experience and prior cases

1

2

 Number of years of experience and training in intervention

1

2

 ‘Qualified’ intervention deliverer

1

2

 Training of therapy and group supervision

1

2

 Profession qualification

3

6

 Prior training and experience of intervention

1

2

 Trained in delivering intervention

4

8

 Recommendation by colleagues as expert

1

2

 Experience of working with patient group

1

2

 Willingness to learn new intervention

1

2

 Without prior experience of intervention (training then provided)

1

2

 None (training/supervision provided as part of the study)

3

7

 Not stated

22

51

Criteria for delivering intervention 2

 Number of years of experience and specific outcome levels to be achieved

1

2

 Years of experience

1

2

 Number of years of experience and prior cases

1

2

 Recommendation by colleagues as expert

1

2

 Experience of working with patient group

1

2

 Professional qualification

2

4

 Preference and no training in alternative intervention

1

2

 Willingness to learn new intervention

1

2

 Interest in patient group

1

2

 None (trained as part of the study)

2

5

 None stated

31

72

 Criteria provided for both intervention 1 and 2 deliverers

12

28

Number of health professionals delivering intervention 1

 Reported

30

69

 Median (interquartile range), range

Number of health professionals delivering intervention 2

 Fully reported

23

53

 Median (interquartile range), range