Methodological variable	Operational definition/information required from each study
1. Representative spectrum (spectrum bias)	When included patients did not represent the intended targeted population, this may have led to an under‐ or overestimation of diagnostic accuracy depending on the difference between the targeted and included populations. The target spectrum in our review was patients with renal failure who were candidates for kidney transplantation. This was scored 'yes' if study participants included only patients with kidney disease who were considered to be candidates for kidney transplantation
2. Acceptable reference standard	An imperfect reference standard may have resulted in misclassification of disease positives and disease negatives. For the purpose of this review, studies had an acceptable reference standard if they used coronary angiography as the reference standard
3. Acceptable delay between tests (disease progression bias)	Disease may have progressed to a more advanced stage (i.e. greater degree of coronary artery stenosis) if a significant time interval between index and reference tests was observed, thereby leading to disease progression bias. This was scored as ‘yes’ if the delay between test was short (i.e. less than three months)
4. Partial verification avoided (verification bias)	Partial verification bias usually leads to an overestimation of sensitivity, although its effect on specificity varies. This item was scored ‘yes’ if all patients who received the index test were also evaluated by the reference standard
5. Differential verification avoided	This was scored ‘yes’ if no patients were verified with a second or third reference standard
6. Incorporation avoided (incorporation bias)	This bias usually leads to an overestimation of diagnostic test accuracy. Incorporation bias was deemed to have existed if the index test was incorporated in a composite reference standard. Studies were scored ‘yes’ if their classification of disease status did not directly involve the results of the index test
7. Reference standard results blinded (information bias)	When the reference standard was interpreted knowing the index test results, this may have led to the overestimation of diagnostic test accuracy. Studies were scored ‘yes’ if blinding of the reference standard was explicitly stated in the article or if this was acknowledged by authors in subsequent personal communication. Otherwise, the studies were marked ‘unclear’, unless blinding was explicitly stated to be absent
8. Index test results blinded (information bias)	When the index test results were interpreted without the knowledge of results of the reference standard, or with more information than in practice, this may have resulted in bias, usually leading to an overestimation of diagnostic accuracy. This item was scored ‘yes’ if blinding of the index test was explicitly stated in the article or if this was acknowledged by authors in subsequent personal communication. Otherwise, the studies were marked ‘unclear’, unless blinding was explicitly stated to be absent
9. Relevant clinical information (information bias)	The availability of clinical data during interpretation of test results may have affected estimates of test performance. This item was scored ‘yes’ if the data available during the study of diagnostic test accuracy was the same as that which would have been available in normal clinical practice
10. Uninterpretable results explained	This item was scored ‘yes’ if uninterpretable results were explained or if there were no uninterpretable results present. This item was scored 'no' if uninterpretable results were found but not explained
11. Withdrawals explained	Excluding patients from the study may have led to an overestimation of diagnostic accuracy. This item was scored ‘yes’ if withdrawals were explained or if there were no withdrawals from the study. This item was scored 'no' if there were withdrawals from the study, but these were unexplained