Reflective testing.

T he processing of chemical pathology requests includes pre-analytical, analytical, and post-analytical phases. At its most basic, the postanalytical process is the reporting and delivery of validated results. One of the ways in which the chemical pathology laboratory provides ‘‘added value’’ over a purely analytical service is at this postanalytical stage—that is, further processing of results related to the request— either before issuing the report, simultaneously, or at a later stage. Indeed, such post-analytical processing is integral to the provision of a quality service, and as such is a requirement for laboratory accreditation in the UK. The simplest form of processing is to compare the result with an appropriate, predefined ‘‘action range’’ (range checking), or to compare changes in sequential results from the same individual to predefined expected ranges of change (delta checking). Historically, this would have been carried out visually by the person authorising paper reports, but now such checks are usually performed by laboratory information systems, which automatically authorise results and generate reports for results that do not fail the checks. Although in many cases the results that fail these tests will be deemed to be acceptable in the clinical circumstances and so no further action would be required, others will require action. This may be as simple as telephoning unexpected results to facilitate earlier intervention, but processing can involve adding comments or requesting appropriate further tests on the sample (reflex testing). Some laboratory information systems will perform these actions automatically but, although expert systems are becoming available, such automated processing currently remains fairly basic. In this issue, Paterson and Paterson examine an alternative form of processing, for which they use the term ‘‘reflective testing’’. This is the process by which a laboratory clinician might add on further tests using their clinical judgement. Although this is common practice, the process does not appear to have been evaluated previously, and deserves further scrutiny, particularly because different individuals take different approaches. The question of best practice with regard to adding on of tests has been the matter of some debate. The authors make some assessment of the clinical effectiveness of their strategy, although cost effectiveness and ethical issues also have to be considered. With regard to clinical effectiveness, the authors use the term NND—the number of add on tests that are required to obtain a specific diagnosis. This would seem a logical way to compare the effectiveness of different strategies of investigation, based simply on the proportion of add on tests that yield definitive results, although the sensitivity and specificity (and thus negative and positive predictive values) of the approach are not fully examined. The quality of clinical details would have a major influence on NND—the more information, the more added value can be provided. Furthermore, variability of clinical insight and experience will also have an influence, hence NND would vary between individuals.