| Abstract | Cilj istraživanja: zasebno odrediti dijagnostičku točnost kolestatskih biljega, ultrazvuka abdomena, endoskopskog ultrazvuka, kolangkopankreatografije magnetskom rezonancom (MRCP), endoskopske retrogradne kolangiopankreatografije (ERCP) i intraoperativne kolangiografije koristeci metode Cochrane sustavnog pregleda i meta-analize. Nadalje, cilj rada je usporediti dijagnostičku točnost parova dijagnostičkih postupaka prema njihovom mjestu u postupku dijagnostike koledokolitijaze. Treći cilj je na osnovu dokaza dobivenih meta-analizom razviti postupnik dijagnostike kod bolesnika koji se prezentiraju simptomima i znakovima suspektnim za koledokolitijazu, a koji je temeljen na dokazima.
Materijal i metode: u rad su uključeni ispitanici u studijama dijagnostičke točnosti analiziranih postupaka koje zadovoljavaju kriterije uključenja. Iz uključcnih studija izvadeni su podaci o indeksnom postupku i referentnom standardu na način koji omogućuje izračunavanje ukupnog broja stvarno pozitivnih, stvarno negativnih, lažno pozitivnih i lažno negativnih bolesnika koji su dijagnosticirani indeksnim postupkom, a u ovisnosti o rezultalu referentnog standarda. Kao referentni standard uzeta je potvrda prisutnosti iIi odsutnosti koledokolitijaze endoskopskom ili kirurškom eksploracijom zajedničkog žučnog voda. U slučaju negativnog rezultata indeksnog postupka, kada je neetično raditi invazivnu eksploraciju, kao potvrda odsutnosti koledokolitijaze prihvaćeno je asimplomatsko praćenje u trajanju najmanje šest mjeseci. Za meta-analizu korišten je bivarijatni model statističke analize i METADAS makro za SAS statistički program. Također su izračunate post-testne vjerojatnosti postojanja koledokolitijaze nakon pozitivnog i negativnog rezultata svakog od istraživanih dijagnostičkih postupaka s pripadajućim 95%-tnim intervalima pouzdanosti (CI od engl. confidence interval. Za konstrukciju forest plotova bivarijatnih ROC krivulja korišten je računalni program Review Manager (Rev Man) 5.2 razvijen od strane Cochrane kolaboracije.
Rezultati: pretraživanjem literature zadanom strategijom pretraživanja dobiveno je 22790 studija. Nakon eliminacije dvostrukih rezultata preostale su 16923 studije. Nadalje je isključeno 16405 studija zbog nepovezanosti s tematikom dijagnostičke točnosti koledokolitijaze. Prema kriterijima uključenja ocijenjeno je 518 studija, od kojih je 36 studija uključeno u sustavni pregled i meta-analizu. Za analizu dijagnostičke točnosti kolestatskih biljega nema dovoljno studija u literaturi; pronađena je samo jedna studija koja zadovoljava kriterije uključenja. Ultrazvuk abdomena pokazao je post-testnu vjerojatnost nakon pozitivnog rezultata od 84,5% (95% Cl 75,0% do 90, I %), dok je post-testna vjerojatnost nakon negativnog rezultata 17,2% (95% CI 8,0% do 33,2%). Ukupno 327% bolesnika koji se podvrgnu ultrazvuku abdomena s ciljem procjene prisutnosti ili odsutnosti koledokolitijaze bit će pogrešno dijagnosticirani. Endoskopski ultrazvuk pokazuje post-testnu vjerojatnost za pozitivan rezultat od 96,0% (95% CI 91 ,8% do 98,1 %) , dok je vjerojatnost za negativan rezultat 3,6% (95% C1 2,2% do 5,8%). Endoskopskim ultrazvukom 7,6% bolesnika dobit ce pogrešnu dijagnozu. MRCP pokazuje post-testnu vjerojatnost za pozitivan rezultat od 93,9% (95% CI 86,7% do 97,3%), dok je vjerojatnost za negativan rezultat 4,9% (95% Cl 2,6% do 8,9%). Ukupno 11 % bolesnika imat ce pogrešno postavljenu dijagnozu nakon učinjenog MRCP-a. Nema statistički značajne razlike između dijagnostičke točnosti endoskopskog ultrazvuka i MRCP-a, medutim MRCP pokazuje nesto bolje rezultate kod prethodno kolecistektomiranih bolesnika. ERCP je pokazao post-testnu vjerojatnost za pozitivan rezultat od 99,0% (95% CI 91 ,0% do 99,9%), dok je vjerojatnost za negativan rezultat 7 4% (95% C1 3 2% do 16 3%). Nakon dijagnostike ERCP-om 8,4% bolesnika bit ce pogrešno dijagnosticirano. Intraoperativna kolangiografija pokazuje post-testnu vjerojatnost za pozitivan rezultat od 98,0% (95% Cl 94,6% do 99,3%), dok je vjerojatnost za negativan rezultat 1,1% (95% CI 0,0% do 31 ,0%). Ovom metodom 3,1% bolesnika imat će pogrešnu dijagnozu. Nema statistički značajne razlike između dijagnostičke točnosti ERCP-a i intraoperativne kolangiografije, međutim kako su ovo vrlo invazivni postupci ne preporučuju se u rutinskoj dijagnostici koledokolitijaze. U bolesnika koji nisu kolecistektomirani preporučuje se učiniti intraoperativnu kolangiografiju u slučaju da bolesnik ima simptome i znakove a rezultati prethodnih dijagnostičkih postupaka su negativni, u centrima koji imaju ovakvu mogućnost. Na osnovi rezultata ovog rada razvijen je dijagnostički postupnik temeljen na dokazima kod bolesnika sa suspektnom koledokolitijazom.
Zaključak: ovim radom dobiveni su podaci o dijagnostičkoj točnosti različitih postupaka u dijagnostici koledokolitijaze koji su temeljeni na dokazima dobivenima Cochrane sustavnim pregledom i meta-analizom. Uspoređeni su parovi postupaka koji se primijenjuju na svakom od tri stupnja dijagnostičke obrade bolesnika. Također je predložen i dijagnostički postupnik koji se temelji na rezultatima ovog rada. Daljnja istraživanja trebaju biti provedena u prikladno odabranim populacijama bolesnika uz primjenu najboljih mogućih referentnih postupaka. |
| Abstract (english) | DIAGNOSTIC ACCURACY OF DIFFERENT TESTS FOR DIAGNOSIS OF COMMON BILE DUCT STONES
Objectives: this thesis aimed to determine diagnostic test accuracy of liver funcion tests, abdominal ultrasound, endoscopic ultrasound, megnetic resonance cholangiopancreatography (MRCP), endoscopic retrograde cholangiopancreatography (ERCP) and intraoperative cholangiography by using the methods of Cochrane systematic review and meta-analysis. Further, the aim was to compare diagnostic accuracy of pairs of tests according to their position in the diagnostic pathway. Third aim was to develop a diagnostic algorithm for evaluation of patients suspected of common bile duct stones which is based on evidence aquired by this research.
Material and methods: subjects from diagnostic accuracy studies that fulfilled the inclusion criteria were included. Data on the index test and the reference standard were extracted from included studies in the manner so that number of true positive, true negative, false positive and false negative could be calculated based on the results of the reference standard. Results of positive or negative endoscopic or surgical exploration of the common bile duct were accepted as reference standard. In the case of negative index test result, when it would be unethical to explore the common bile duct, the confirmation of absence of stones by asymptomatic follow-up of at least six months was accepted. Bivariate statistical analysis mode and METADAS macro for SAS were used to meta-analyse the data. Also, post-test probabilities for positive and negative index tests with corresponding 95% confidence intervals (CI) were calculated for each of the tests that were evaluated. For construction of forest plots and bivariate ROC curves we used Review Manager (RevMan) 5.2 developed and provided by the Cochrane Collaboration.
Results: A total of 22790 studies were evaluated for inclusion in the systematic review. After deletion of duplicates 16923 studies were further evaluated. A further 16405 studies were eliminated because of evident lack of connection with diagnostic accuracy of common bile duct stones. The remaining 518 studies were evaluated for inclusion, out of which 36 studies were included in the systematic review and meta-analysis. There were not enough studies for evaluation of liver function tests, only one study met the inclusion criteria. Abdominal ultrasound showed post-test probability for positive test of 84.5% (95% CI 75.0% to 90.1%), while post-test probability for negative test is 17.2% (95% CI 8.0% to 33.2%). A total of 32.7% of patients that undergo abdominal ultrasound for evaluation of presence of absence of common bile duct stones will receive wrong diagnosis. Endoscopic ultrasound showed post-test probability for positive test of 96.0% (95% CI 91.8% to 98.1%). Post-test probability for negative test is 3.6% (95% CI 2.2% to 5.8%). Patients evaluated with endoscopic ultrasound will receive wrong diagnosis on 7.6% of cases. MRCP showed post-test probability for positive test of 93.9% (95% CI 86.7% to 97.3%), while post-test probability for negative test is 4.9% (95% CI 2.6% to 8.9%). A total of 11% of patients evaluated with MRCP will receive the wrong diagnosis. There is not statistically significant difference between diagnostic accuracy of endoscopic ultrasound and MRCP. However, MRCP showed somewhat better results in patients with previous cholecystectomy. ERCP showed post-test probability for positive test of 99.0% (95% CI 91.0% to 99.9%), while the post-test probability for negative test is 7.4% (95% CI 3.2% to 16.3%). After performing ERCP, 8.4% of patients will receive the wrong diagnosis. Intraoperative cholangiography showed post-test probability for positive test of 98.0% (95% CI 94.6% to 99.3%), while the post-test probability for a negative test is 1.1% (95% CI 0.0% to 31.0%). There is a probability that 3.1% evaluated with intraoperative cholangiography will receive the wrong diagnosis. There is no statistically significant difference between diagnostic accuracy of ERCP and intraoperative cholangiography. However, since those are highly invasive procedures they are not recommended in routine diagnostic evaluation for common bile duct stones. In patients that have not been previously cholecystectomysed it is recommended to perform intraoperative cholangiography when a patient has signs and symptoms and results of previous diagnostic test are negative, in centres with such capabilities. A diagnostic algorithm for diagnosis of patients suspected of common bile duct stones was developed. The diagnostic algorithm is based on evidence obtained in this study.
Conclusion: results of diagnostic accuracy of different tests of diagnosis of common bile duct stones are presented in this thesis, based on evidence obtained with a Cochrane systematic review and meta-analysis. Pairs of diagnostic tests that are performed on each of the three steps of diagnostic pathway were compared. Also, a diagnostic algorithm was presented, based on results of this thesis. Further research should be performed in adequately selected populations of patients and best available reference standards should be applied. |
