338:
102:
Diagnostic test performances reported by some studies may be artificially overestimated if it is a case-control design where a healthy population ('fittest of the fit') is compared with a population with advanced disease ('sickest of the sick'); that is two extreme populations are compared, rather
28:
may vary in different clinical settings because each setting has a different mix of patients. Because the performance may be dependent on the mix of patients, performance at one clinic may not be predictive of performance at another clinic. These differences are interpreted as a kind of
442:
Lachs MS, Nachamkin I, Edelstein PH, Goldman J, Feinstein AR, Schwartz JS (1992). "Spectrum bias in the evaluation of diagnostic tests: lessons from the rapid dipstick test for urinary tract infection".
45:, whilst others maintain it is a bias if the true performance of the test differs from that which is 'expected'. Usually the performance of a diagnostic test is measured in terms of its
60:
Generally spectrum bias is considered to have three causes. The first is due to a change in the case-mix of those patients with the target disorder (disease) and this affects the
350:
Leeflang MM, Bossuyt PM, Irwig L., Diagnostic test accuracy may vary with prevalence: implications for evidence-based diagnosis, J Clin
Epidemiol. 2009 Jan;62(1) 5–12.
485:
Rutjes AWS, Reitsma JB, Vandenbroucke JP, Glas AS, Bossuyt PMM, Case-control and two-gate designs in diagnostic accuracy studies, Clin Chem 2005;51(8):1335–41.
360:
Goehring C, Perrier A, Morabia A (2004). "Spectrum bias: a quantitative and graphical analysis of the variability of medical diagnostic test performance".
1111:
1086:
295:"Evidence that disease prevalence may affect the performance of diagnostic tests with an implicit threshold: a cross sectional study"
1116:
54:
53:
and it is changes in these that are considered when referring to spectrum bias. However, other performance measures such as the
64:. The second is due to a change in the case-mix of those without the target disorder (disease-free) and this affects the
513:
154:
Ransohoff DF, Feinstein AR (1978). "Problems of spectrum and bias in evaluating the efficacy of diagnostic tests".
1188:
1106:
916:
1219:
1035:
921:
575:
570:
1224:
693:
198:"Spectrum bias – why clinicians need to be cautious when applying diagnostic test studies"
1029:
555:
1159:
883:
683:
661:
337:
1183:
1096:
978:
743:
723:
619:
107:
92:
134:
80:
as well as theoretical arguments which suggest that it does indeed affect a test's performance.
1209:
1169:
853:
833:
614:
592:
948:
863:
838:
783:
260:
901:
753:
629:
506:
119:
8:
1056:
973:
873:
808:
748:
738:
733:
597:
88:
84:
73:
69:
65:
61:
50:
46:
953:
938:
698:
688:
671:
468:
385:
321:
294:
272:
110:
of subgroups can lead to insights about the test's performance in varying populations.
77:
1066:
1003:
988:
911:
893:
828:
624:
540:
460:
424:
377:
326:
264:
219:
171:
38:
472:
389:
276:
242:"Spectrum bias or spectrum effect? Subgroup variation in diagnostic test evaluation"
1133:
993:
933:
858:
843:
703:
656:
565:
560:
545:
452:
416:
369:
316:
308:
256:
241:
209:
163:
96:
404:
1101:
1091:
868:
848:
763:
666:
641:
636:
609:
587:
499:
25:
456:
312:
167:
1143:
1138:
1128:
1051:
968:
928:
878:
823:
813:
798:
793:
758:
713:
678:
582:
531:
420:
129:
1214:
1203:
1081:
1061:
1024:
998:
963:
943:
906:
818:
778:
773:
768:
646:
550:
124:
34:
17:
214:
197:
68:. The third is due to a change in the prevalence, and this affects both the
1041:
803:
788:
381:
330:
268:
223:
464:
428:
958:
728:
718:
708:
604:
175:
1076:
1071:
1046:
91:
change between different sub-groups of patients may be found with the
1164:
651:
373:
76:. This final cause is not widely appreciated, but there is mounting
1123:
1008:
299:
441:
522:
41:; this has led some authors to refer to the phenomenon as
491:
359:
24:refers to the phenomenon that the performance of a
153:
1201:
103:than typical healthy and diseased populations.
288:
286:
239:
235:
233:
191:
189:
187:
185:
507:
402:
353:
344:
283:
230:
182:
1170:Heuristics in judgment and decision-making
514:
500:
435:
396:
320:
213:
195:
147:
33:. Mathematically, the spectrum bias is a
261:10.7326/0003-4819-137-7-200210010-00011
57:may also be affected by spectrum bias.
1202:
292:
495:
106:If properly analyzed, recognition of
13:
14:
1236:
336:
240:Mulherin SA, Miller WC (2002).
479:
1:
140:
7:
1036:DĂ©formation professionnelle
457:10.7326/0003-4819-117-2-135
313:10.1136/bmjopen-2011-000746
168:10.1056/NEJM197810262991705
113:
10:
1241:
1030:Basking in reflected glory
521:
421:10.7326/0003-4819-104-1-66
405:"Carcinoembryonic antigen"
1178:
1160:Cognitive bias mitigation
1152:
1017:
892:
529:
744:Illusion of transparency
93:carcinoembryonic antigen
135:Reference class problem
362:Statistics in Medicine
37:and not a traditional
1220:Design of experiments
1112:Arab–Israeli conflict
839:Social influence bias
784:Out-group homogeneity
215:10.1093/fampra/cmn051
754:Mere-exposure effect
684:Extrinsic incentives
630:Selective perception
403:Fletcher RH (1986).
979:Social desirability
874:von Restorff effect
749:Mean world syndrome
724:Hostile attribution
293:Willis, BH (2012).
83:Examples where the
1225:Medical statistics
894:Statistical biases
672:Curse of knowledge
196:Willis BH (2008).
78:empirical evidence
1197:
1196:
834:Social comparison
615:Choice-supportive
120:Simpson's paradox
55:likelihood ratios
1232:
994:Systematic error
949:Omitted-variable
864:Trait ascription
704:Frog pond effect
532:Cognitive biases
516:
509:
502:
493:
492:
486:
483:
477:
476:
445:Ann. Intern. Med
439:
433:
432:
409:Ann. Intern. Med
400:
394:
393:
374:10.1002/sim.1591
357:
351:
348:
342:
341:
340:
334:
324:
290:
281:
280:
249:Ann. Intern. Med
246:
237:
228:
227:
217:
193:
180:
179:
151:
97:urinary dipstick
43:spectrum effects
39:statistical bias
1240:
1239:
1235:
1234:
1233:
1231:
1230:
1229:
1200:
1199:
1198:
1193:
1174:
1148:
1013:
888:
869:Turkey illusion
637:Compassion fade
534:
525:
520:
490:
489:
484:
480:
440:
436:
401:
397:
358:
354:
349:
345:
335:
291:
284:
244:
238:
231:
202:Family Practice
194:
183:
156:N. Engl. J. Med
152:
148:
143:
116:
26:diagnostic test
12:
11:
5:
1238:
1228:
1227:
1222:
1217:
1212:
1195:
1194:
1192:
1191:
1186:
1179:
1176:
1175:
1173:
1172:
1167:
1162:
1156:
1154:
1153:Bias reduction
1150:
1149:
1147:
1146:
1141:
1136:
1131:
1129:Political bias
1126:
1121:
1120:
1119:
1114:
1109:
1104:
1099:
1094:
1089:
1084:
1074:
1069:
1064:
1059:
1057:Infrastructure
1054:
1049:
1044:
1039:
1032:
1027:
1021:
1019:
1015:
1014:
1012:
1011:
1006:
1001:
996:
991:
986:
981:
976:
974:Self-selection
971:
966:
961:
956:
951:
946:
941:
936:
931:
926:
925:
924:
914:
909:
904:
898:
896:
890:
889:
887:
886:
881:
876:
871:
866:
861:
856:
851:
846:
841:
836:
831:
826:
821:
816:
811:
809:Pro-innovation
806:
801:
796:
794:Overton window
791:
786:
781:
776:
771:
766:
761:
756:
751:
746:
741:
736:
731:
726:
721:
716:
711:
706:
701:
696:
691:
686:
681:
676:
675:
674:
664:
662:Dunning–Kruger
659:
654:
649:
644:
639:
634:
633:
632:
622:
617:
612:
607:
602:
601:
600:
590:
585:
580:
579:
578:
576:Correspondence
573:
571:Actor–observer
563:
558:
553:
548:
543:
537:
535:
530:
527:
526:
519:
518:
511:
504:
496:
488:
487:
478:
434:
395:
352:
343:
307:(1): e000746.
282:
255:(7): 598–602.
229:
181:
162:(17): 926–30.
145:
144:
142:
139:
138:
137:
132:
130:Reporting bias
127:
122:
115:
112:
9:
6:
4:
3:
2:
1237:
1226:
1223:
1221:
1218:
1216:
1213:
1211:
1210:Biostatistics
1208:
1207:
1205:
1190:
1187:
1185:
1181:
1180:
1177:
1171:
1168:
1166:
1163:
1161:
1158:
1157:
1155:
1151:
1145:
1142:
1140:
1137:
1135:
1132:
1130:
1127:
1125:
1122:
1118:
1115:
1113:
1110:
1108:
1107:United States
1105:
1103:
1100:
1098:
1095:
1093:
1090:
1088:
1085:
1083:
1082:False balance
1080:
1079:
1078:
1075:
1073:
1070:
1068:
1065:
1063:
1060:
1058:
1055:
1053:
1050:
1048:
1045:
1043:
1040:
1038:
1037:
1033:
1031:
1028:
1026:
1023:
1022:
1020:
1016:
1010:
1007:
1005:
1002:
1000:
997:
995:
992:
990:
987:
985:
982:
980:
977:
975:
972:
970:
967:
965:
962:
960:
957:
955:
954:Participation
952:
950:
947:
945:
942:
940:
937:
935:
932:
930:
927:
923:
922:Psychological
920:
919:
918:
915:
913:
910:
908:
905:
903:
900:
899:
897:
895:
891:
885:
882:
880:
877:
875:
872:
870:
867:
865:
862:
860:
857:
855:
852:
850:
847:
845:
842:
840:
837:
835:
832:
830:
827:
825:
822:
820:
817:
815:
812:
810:
807:
805:
802:
800:
797:
795:
792:
790:
787:
785:
782:
780:
777:
775:
772:
770:
767:
765:
762:
760:
757:
755:
752:
750:
747:
745:
742:
740:
737:
735:
732:
730:
727:
725:
722:
720:
717:
715:
712:
710:
707:
705:
702:
700:
697:
695:
692:
690:
689:Fading affect
687:
685:
682:
680:
677:
673:
670:
669:
668:
665:
663:
660:
658:
655:
653:
650:
648:
645:
643:
640:
638:
635:
631:
628:
627:
626:
623:
621:
618:
616:
613:
611:
608:
606:
603:
599:
596:
595:
594:
591:
589:
586:
584:
581:
577:
574:
572:
569:
568:
567:
564:
562:
559:
557:
554:
552:
549:
547:
544:
542:
539:
538:
536:
533:
528:
524:
517:
512:
510:
505:
503:
498:
497:
494:
482:
474:
470:
466:
462:
458:
454:
451:(2): 135–40.
450:
446:
438:
430:
426:
422:
418:
414:
410:
406:
399:
391:
387:
383:
379:
375:
371:
368:(1): 125–35.
367:
363:
356:
347:
339:
332:
328:
323:
318:
314:
310:
306:
302:
301:
296:
289:
287:
278:
274:
270:
266:
262:
258:
254:
250:
243:
236:
234:
225:
221:
216:
211:
208:(5): 390–96.
207:
203:
199:
192:
190:
188:
186:
177:
173:
169:
165:
161:
157:
150:
146:
136:
133:
131:
128:
126:
125:Biased sample
123:
121:
118:
117:
111:
109:
108:heterogeneity
104:
100:
98:
94:
90:
86:
81:
79:
75:
71:
67:
63:
58:
56:
52:
48:
44:
40:
36:
35:sampling bias
32:
27:
23:
22:spectrum bias
19:
18:biostatistics
1067:In education
1034:
1018:Other biases
1004:Verification
989:Survivorship
983:
939:Non-response
912:Healthy user
854:Substitution
829:Self-serving
625:Confirmation
593:Availability
541:Acquiescence
481:
448:
444:
437:
415:(1): 66–73.
412:
408:
398:
365:
361:
355:
346:
304:
298:
252:
248:
205:
201:
159:
155:
149:
105:
101:
82:
59:
42:
30:
21:
15:
1134:Publication
1087:Vietnam War
934:Length time
917:Information
859:Time-saving
719:Horn effect
709:Halo effect
657:Distinction
566:Attribution
561:Attentional
89:specificity
85:sensitivity
74:specificity
70:sensitivity
66:specificity
62:sensitivity
51:specificity
47:sensitivity
1204:Categories
1097:South Asia
1072:Liking gap
884:In animals
849:Status quo
764:Negativity
667:Egocentric
642:Congruence
620:Commitment
610:Blind spot
598:Mean world
588:Automation
141:References
1165:Debiasing
1144:White hat
1139:Reporting
1052:Inductive
969:Selection
929:Lead time
902:Estimator
879:Zero-risk
844:Spotlight
824:Restraint
814:Proximity
799:Precision
759:Narrative
714:Hindsight
699:Frequency
679:Emotional
652:Declinism
583:Authority
556:Anchoring
546:Ambiguity
95:test and
1062:Inherent
1025:Academic
999:Systemic
984:Spectrum
964:Sampling
944:Observer
907:Forecast
819:Response
779:Optimism
774:Omission
769:Normalcy
739:In-group
734:Implicit
647:Cultural
551:Affinity
473:25381473
390:24636826
382:14695644
331:22307105
300:BMJ Open
277:35752032
269:12353947
224:18765409
114:See also
1184:General
1182:Lists:
1117:Ukraine
1042:Funding
804:Present
789:Outcome
694:Framing
465:1605428
429:3510056
322:3274715
99:tests.
1189:Memory
1102:Sweden
1092:Norway
959:Recall
729:Impact
605:Belief
523:Biases
471:
463:
427:
388:
380:
329:
319:
275:
267:
222:
176:692598
174:
1077:Media
1047:FUTON
469:S2CID
386:S2CID
273:S2CID
245:(PDF)
1215:Bias
461:PMID
425:PMID
378:PMID
327:PMID
265:PMID
220:PMID
172:PMID
87:and
72:and
49:and
31:bias
1124:Net
1009:Wet
453:doi
449:117
417:doi
413:104
370:doi
317:PMC
309:doi
257:doi
253:137
210:doi
164:doi
160:299
20:,
16:In
1206::
467:.
459:.
447:.
423:.
411:.
407:.
384:.
376:.
366:23
364:.
325:.
315:.
303:.
297:.
285:^
271:.
263:.
251:.
247:.
232:^
218:.
206:25
204:.
200:.
184:^
170:.
158:.
515:e
508:t
501:v
475:.
455::
431:.
419::
392:.
372::
333:.
311::
305:2
279:.
259::
226:.
212::
178:.
166::
Text is available under the Creative Commons Attribution-ShareAlike License. Additional terms may apply.