Table 5 Summary of the statistical methods used in the analysis of the ordinal outcome

Descriptive statistics―n (%)^a

- Frequencies and/or proportions/percentages (category specific)

116 (81%)

- Medians (across all categories)

33 (23%)

- Interquartile ranges (across all categories)

31 (22%)

- Means (across all categories)

16 (11%)

- Standard deviations (across all categories)

10 (7%)

- Other methods to summarise by group used

10 (7%)

- Summaries by group not used

2 (1%)

Inferential approach―n (%)

- Frequentist inference

131 (91%)

- Bayesian inference

6 (4%)

- Both frequentist and Bayesian

2 (1%)

- Descriptive analysis only

5 (4%)

Statistical method(s)―n (%)^a

- Proportional odds model

64 (44%)

- Logistic regression model

16 (11%)

- Linear model

16 (11%)

- Cochran-Mantel-Haenszel test

15 (10%)

- Wilcoxon test

14 (10%)

- Fisher’s exact test or chi-square test

12 (8%)

- Binomial regression model

7 (5%)

- Other

24 (17%)

- Could not be determined

6 (4%)

- Descriptive analysis only

6 (4%)

Original statistical method(s) modified―n (%)^b

- No

124 (86%)

- Yes

12 (8%)

- Unknown

2 (1%)

- Not applicable

6 (4%)

Statistical method(s) assumptions checked and clearly described―n (%)

- No

86 (60%)

- Yes

46 (32%)

- Unknown

5 (4%)

- Not applicable (descriptive analysis or only bootstrapping methods used)

7 (5%)

Methods used to check assumptions:^a

46

- Statistical methods

31 (67%)

- Graphical methods

2 (4%)

- Prediction methods

1 (2%)

- Other

7 (15%)

- Not reported

6 (13%)

Methods used to account for repeated measures―n (%)^a

38

- Adjusted for baseline measurement

18 (47%)

- Mixed effects models

14 (37%)

- Generalised estimating equations

4 (11%)

- Other

4 (11%)

- Unknown

2 (5%)

Software package(s) used―n (%)^a

- SAS

81 (56%)

- R

35 (24%)

- Stata

27 (19%)

- SPSS

13 (9%)

- Other

9 (6%)

- Unknown

12 (8%)