Accuracyofabdominalauscultationforbowelobstruction

摘要： AIM:Toinvestigatetheaccuracyandinter-observervariationofbowelsoundassessmentinpatientswithclinicallysuspectedbowelobstruction.METHODS:BowelsoundswererecordedinpatientswithsuspectedbowelobstructionusingaLittmann®ElectronicStethoscope.Therecordingswereprocessedtoyield25-ssoundsequencesinrandomorderonPCs.Observers,recruitedfromdoctorswithinthedepartment,classifiedthesoundsequencesaseithernormalorpathological.Thereferencetestsforbowelobstructionwereintraoperativeandendoscopicfindingsandclinicalfollowup.Sensitivityandspecificitywerecalculatedforeachobserverandcomparedbetweenjuniorandseniordoctors.InterobservervariationwasmeasuredusingtheKappastatistic.RESULTS:Bowelsoundsequencesfrom98patientswereassessedby53(33juniorand20senior)doctors.Laparotomywasperformedin47patients,35ofwhomhadbowelobstruction.Twopatientsunderwentcolorectalstentingduetolargebowelobstruction.Themediansensitivityandspecificitywas0.42(range:0.19-0.64)and0.78(range:0.35-0.98),respectively.Therewasnosignificantdifferenceinaccuracybetweenjuniorandseniordoctors.Themedianfrequencywithwhichdoctorsclassifiedbowelsoundsasabnormaldidnotdiffersignificantlybetweenpatientswithandwithoutbowelobstruction(26%vs23%,P=0.08).The53doctorsmadeup1378uniquepairsandthemedianKappavaluewas0.29(range:-0.15-0.66).CONCLUSION:Accuracyandinter-observeragreementwasgenerallylow.Clinicaldecisionsinpatientswithpossiblebowelobstructionshouldnotbebasedonauscultatoryassessmentofbowelsounds. （共7頁）