Application of GRADE: Making evidence-based recommendations about diagnostic tests in clinical practice guidelines

Table 3 Example of evidence profile generated based on systematic review conducted for these guidelines

Question 1, Profile 1: Should skin prick tests be used for the diagnosis of IgE-mediated CMA in patients suspected of CMA? Cut-off ≥3 mm \| All populations
Outcome	No. of studies	Study design	Factors that may decrease quality of evidence					Final quality	Effect per 1000¹	Importance
			Limitations	Indirectness	Inconsistency	Imprecision	Reporting bias
True positives (patients with CMA)	23 studies (2302 patients)	Consecutive or non-consecutive series	Serious²	None	Serious³	None	Unlikely	⊕⊕OO low	Prev 80%: 536 Prev 40%: 268 Prev 10%: 67	CRITICAL
True negatives (patients without CMA)	23 studies (2302 patients)	Consecutive or non-consecutive series	Serious²	None	Serious³	None	Unlikely	⊕⊕OO low	Prev 80%: 108 Prev 40%: 324 Prev 10%: 486	CRITICAL
False positives (patients incorrectly classified as having CMA)	23 studies (2302 patients)	Consecutive or non-consecutive series	Serious²	Serious⁴	Serious³	None	Unlikely	⊕OOO very low	Prev 80%: 92 Prev 40%: 276 Prev 10%: 414	CRITICAL
False negatives (patients incorrectly classified as not having CMA)	23 studies (2302 patients)	Consecutive or non-consecutive series	Serious²	None	Serious³	None	Unlikely	⊕⊕OO low	Prev 80%: 264 Prev 40%: 132 Prev 10%: 33	CRITICAL
Inconclusive⁵	1 study (310 patients)	Non-consecutive series	-	-	-	-	-	-	-	IMPORTANT
Complications	Not reported	-	-	-	-	-	-	-	-	NOT IMPORTANT
Cost	Not reported	-	-	-	-	-	-	-	-	NOT IMPORTANT

Footnotes 1 to 5 provide detailed rationale underlying ratings.
¹Based on combined sensitivity of 67% (95% CI: 64 to 70) and specificity of 74% (95% CI: 72 to 77)
²Most studies enrolled highly selected patients with atopic eczema or gastrointestinal symptoms, no study reported if an index test or a reference standard were interpreted without knowledge of the results of the other test, but it is very likely that those interpreting results of one test knew the results of the other; all except for one study that reported withdrawals did not explain why patients were withdrawn.
³Estimates of sensitivity ranged from 10% to 100%, and specificity from 14% to 100%; we could not explain it by quality of the studies, tests used or included population
⁴There is uncertainty about the consequences for these patients; in some a diagnosis of other potentially serious condition may be delayed
⁵One study in children <12 month of age reported 8% inconclusive challenge tests but did not report number of inconclusive skin prick test

ISSN: 1748-5908