ABC | Volume 112, Nº3, March 2019

Original Article Kiyose et al A systematic review for valvular prostheses Arq Bras Cardiol. 2019; 112(3):292-301 Table 3 – Assessment of the quality of evidence and summary of findings Quality assessment Summary of findings No of studies (No. of participants) Study limitations Inconsistency Indirectness Imprecision Publication bias Relative risk (95% CI) Quality Mortality 4 (1,535) No serious limitations No serious inconsistency Direct No serious imprecision Unlikely 1.07 (0.99, 1.15) ⊕⊕⊕⊕ HIGH Reoperation 4 (1,535) No serious limitations No serious inconsistency Direct No serious imprecision Unlikely 3.60 (2.44, 5.32) ⊕⊕⊕⊕ HIGH Bleeding 4 (1,535) No serious limitations No serious inconsistency Direct No serious imprecision Unlikely 0.64 (0.52, 0.78) ⊕⊕⊕⊕ HIGH Embolism 4 (1,535 No serious limitations No serious inconsistency Direct Imprecision † Unlikely 0.93 (0.66, 1.31) ⊕⊕⊕ O MODERATE Endocarditis 4 (1,535) No serious limitations No serious inconsistency Direct Imprecision* Unlikely 1.21 (0.78, 1.88) ⊕⊕⊕ O MODERATE * Effect estimate compatible with either no effect or harm; † Effect estimate compatible with either substantial benefit or harm. Indeed a number of observational studies have shown the extended durability of biological prostheses, with a decrease in mortality of reoperation. 5 In parallel, use of biological prostheses has increased substantially. 6 However, the evidence provided by observational studies is weak due to the high risk of selection bias. Conversely, observational studies have also suggested increased mortality with biological prosthesis for mitral valve replacement. Our results showed a nonsignificant trend towards increased mortality with biological valves irrespective of position. Quality of evidence (GRADE) The included randomized studies present a low risk of bias and directly evaluate whether differences in clinical outcomes exist between biological and mechanical prostheses. Reporting bias is also unlikely. Regarding the mortality, reoperation and bleeding outcomes, the estimated effect of biological versus mechanical prostheses exhibited good precision and absence of serious inconsistency. We, therefore, consider that the evidence is of high quality (Table 3). For the systemic arterial embolism and endocarditis outcomes, although there was no serious inconsistency, the estimated effect is imprecise (i.e., the 95% CI is compatible with an unfavourable outcome of both the bioprosthesis and the mechanical prosthesis). Strengths and weaknesses Our systematic review has strengths and limitations. The development of the search strategy may be cited as a strength, as it was very sensitive and offered little likelihood of not identifying any relevant evidence. The main databases were searched along with unpublished evidence sources, and a manual evidence search was performed. All systematic review procedures were directed by guidelines and literature specific to this type of study, including all methodological characteristics necessary for proper execution of the review. 10 The included trials conducted extended follow-up of patients (from 2 to 20 years), allowing adequate evaluation of the effect of biological versus mechanical prostheses in clinical outcomes, particularly those with late incidence of outcomes such as the need for reoperation. With regard to limiting factors, the inherent limitations of systematic reviews should be considered, such as slight differences in the populations of trial studies. For example, patients with a small aortic annulus were excluded in the Bloomfield study, 20 those with a small mitral annulus or significant coronary artery disease were excluded from the Veterans study, 17 and patients with aortic valve lesions were excluded from Vallejo's study. 13 A major weakness of our systematic review is the age of available trials. Three of the 4 trials included are old and used first generation biological prostheses and single-disk mechanical prostheses. As both prostheses and ancillary care have evolved, it is possible that the results we have observed would not be currently applicable. Indeed a number of observational studies have shown the higher durability of biological prostheses and a trend towards its use in younger patients. 6 However, the evidence provided by observational studies is weak due to the high risk of selection bias. Furthermore, the results of the randomized trial by Stassano et al. 21 comparing modern biological to mechanical prosthesis are completely consistent with those of previous trials. In special, there was an important increase in the need of reoperation and a decreased risk of bleeding with biological prostheses. Thus, although more evidence from new trials comparing biological to mechanical is urgently needed, the best available evidence does not support the increasing preference for biological prostheses. 299