Why did the patient have the test? Considering the clinical pathway and intended role of a test on diagnostic accuracy

Article type
Authors
Dinnes J1, Davenport C1, Chuchu N1, Matin R2, Williams H3, Deeks J1
1Institute of Applied Health and Research, University of Birmingham
2Oxford University Hospitals NHS Foundation Trust, Oxford
3Centre of Evidence-Based Dermatology, Nottingham
Abstract
Background: Disease spectrum (distribution of patient and disease characteristics across a population) varies between and within healthcare settings and affects the accuracy of a diagnostic test. For systematic review findings to be useful for clinicians, policy makers and patients, the added value of a test should be considered in the setting(s) in which it is intended to be used, and in comparison to tests currently used at that point of the clinical pathway.

Objectives: To examine:
1) the extent to which test accuracy studies can be positioned on a clinical pathway;
2) the effect of pathway on test accuracy estimates;
3) resulting impact on the evaluation and communication of the clinical applicability of review findings to users.

Methods: As illustrations, we use recently completed Cochrane reviews of tests used at various points on the clinical pathway for diagnosis of cutaneous melanoma. Similar considerations in other disease areas will be discussed.

Results: For visual inspection of skin lesions, the clinical pathway was clear for 61% (17/28) of studies. Five were in participants earlier in the pathway, where higher sensitivity and lower specificity were observed compared to in 12 studies in participants further along the referral pathway. No similar effect was seen for studies investigating dermoscopy added to visual inspection. The clinical pathway was clear for 41% (11/27) of dermoscopy studies; only two were in participants earlier in the pathway. Reflectance confocal microscopy (RCM) was studied as a replacement for dermoscopy (any suspicious lesion; earlier in the pathway) and added to dermoscopy in difficult to diagnose (equivocal) skin lesions (later in the pathway, following dermoscopy). RCM was more accurate in equivocal (9 studies) than in any suspicious lesion (7 studies). Poor reporting and lack of consistent terminology forced us to make assumptions about where tests were used on the pathway.

Conclusions: The position at which a test is used in a clinical pathway can have a big effect on test accuracy. The ability of a review to establish test accuracy in the target population depends on primary investigators understanding the implications of participant selection and how the test is used, and on the quality of reporting in primary studies.

Patient or healthcare consumer involvement: We have two patient co-authors.