Sixty individuals responded to the letter of invitation stating that they wished to participate. Forty-five participants (75%) completed round one of the study. Thirty-five participants (58%) completed round two, and 32 participants (53%) completed round three; 16 were ambulance clinicians, and 16 were medical personnel.
A total of 16 new items were added following round one. Raters gained consensus on one hundred and thirty four items (54%) by round three. This figure increased to 164 items (66%) if the items which the raters neared consensus on (those > = 70%) are considered. Almost all the items which reached consensus were viewed as 'important' or 'very important' by participants; only four items on which raters reached consensus on were viewed as 'unimportant' or 'very unimportant' (i.e. rectal thermometer; Clopidogrel (300 mg); Clindamycin; Saline ampule (5 mls)); a further two items 'nearing' consensus, that is reaching 70-79% agreement, were also rated as 'unimportant' or 'very unimportant' (i.e. OPA (Size 000); ET Tube size 10).
There was considerable variation in the percentage of items that gained consensus within the subsets that had been split according to each item's purpose. Subsets with the highest percentage of items reaching consensus were 'Control of Infection' (100%; n = 10); 'Splintage' (73%; n = 8); and 'Circulation' (73%; n = 35). Subsets that had the fewest items that reached consensus were 'Medicine2' (31%; n = 5); and 'Airways2' (35%; n = 11). Consensus for each of these subsets increased when items 'nearing' consensus (i.e. >70%) were also considered, (see Table 2). Items that gained consensus amongst raters as being important or very important are listed in Table 3. A full list of items and levels of consensus reached is also provided, (see Additional file 1).
Round one contained 232 items and rounds two and three contained an additional 16 items, bringing the total number of items to 248 for the last two rounds. The median Likert scores observed for 232 items in each of rounds one, two and three were 4, 5 and 5, respectively, (all with IQR 3 to 5). The non-parametric test for independent events failed to compute, most of the responses were tied in pairs, therefore the Wilcoxon Signed Ranks Test was used to assess whether there was a significant increase in consensus between rounds. A significant difference was found between median Likert scores for the 232 matched pairs of items between round one and round three (Z = -5.26; 151 ties; p < 0.001); but not between round two and round three (Z = -1.79; 215 ties; p = 0.074).
Quantity of Items Required
The median quantities, and Inter Quartile Ranges (IQR), of items that gained consensus by raters are listed in Table 3. Whether the recommended quantities of items between rounds was a statistically significant improvement in participants' consensus was tested using a Wilcoxon Signed Ranks Test. A significant difference was found between the median number of items for the 232 matched pairs of items between round one and round three (Z = -9.83; ties = 80; p < 0.001); and also between round two and round three (Z = -2.39; ties = 160; p = 0.017). Whilst participants suggested similar quantities for many items by round three, other items still had considerably wider recommended quantities. This is clearly evident in the persistently large IQR of some items (e.g. Large Latex Free gloves had a median recommend quantity of 150, but an interquartile range of 107–195).
BMC Emerg Med. 2014;14(5) © 2014 BioMed Central, Ltd.