Aims The purpose of this study was to (1) characterise the procedure of phlebotomy, deconstruct it into its constituent parts and develop a performance metric for the purpose of training healthcare professionals in a large teaching hospital and to (2) evaluate the construct validity of the phlebotomy metric and establish a proficiency benchmark.

Method By engaging with a multidisciplinary team with a wide range of experience of preanalytical errors in phlebotomy and observing video recordings of the procedure performed in the actual working environment, we defined a performance metric. This was brought to a modified Delphi meeting, where consensus was reached by an expert panel. To demonstrate construct validity, we used the metric to objectively assess the performance of novices and expert practitioners.

Results A phlebotomy metric consisting of 11 phases and 77 steps was developed. The mean inter-rater reliability was 0.91 (min 0.83, max 0.95). The expert group completed more steps of the procedure (72 vs 69), made fewer errors (19 vs 13, p=0.014) and fewer critical errors (1 Vs 4, p=0.002) than the novice group.

Conclusions The metrics demonstrated construct validity and the proficiency benchmark was established with a minimum observation of 69 steps, with no critical errors and no more than 13 errors in total.