Table II.

Protocol objectives for the independent classification of study fractures

The primary objective of this study is for two independent shoulder surgeons with training in the trial procedures and Neer’s classification to assess the radiographs of all randomised patients in the proximal fracture of the humerus evaluation by randomisation trial in order to categorise the trial fractures reliably. This will enable completion of the following:

- A description of the study population to inform on the generalisability of the study findings.

- Assessment of the potential difference between the actual and the originally intended trial population in terms of:

- The original categories of fracture types (i.e. 3, 8, 9 and 12; see Fig. 1) versus those resulting from a pragmatic application of the displacement criteria and

- Protocol violations (i.e. non-involvement of the surgical neck; dislocation at the shoulder joint).

# - Quantification of the differences between ‘involvement’ of the tuberosities as recorded by recruiting surgeons on study eligibility forms compared with that of the two independent surgeons and also for the latter surgeons’ assessment of whether (a) the surgical neck is involved and (b) there is ‘displacement’ of parts according to Neer’s categorisation; this will enable some insight as to whether what we did for the trial is a suitable proxy of Neer’s classification.

# - Identification, and thus quantification, of the numbers of fractures in the two subgroup categories (2-part versus 3- and 4-part) for secondary subgroup analyses.

# - Assessment of the correspondence between the actual population in the distribution of the four original fracture categories and that based on the prospective epidemiological study by Court-Brown et al.11