Our results imply that providing patient with a disease-specific 3D heart model for surgical consent is equivalent or superior to conventional 2D diagram in helping patients and guardian understand the anatomy and treatment options of the disease. To the best of our knowledge, this is the first randomized controlled trial that has attempted to quantify the benefit of 3D print models in surgical consent. Three-dimensional printed heart replicas have been widely used in medical education[12, 13], training and planning  as well as clinical communication. However, these studies were either descriptive or quantitative, but without control group studies, or involving patients with a heterogeneity of diseases. Though many studies of these demonstrated benefit from this technique, they seemed to be less powerful in terms of study design. Lim et al. reported a randomized controlled trial comparing 3D prints versus cadaveric materials. This study was aimed at medical students and only involved normal cardiac anatomy. The merits of disease- or patient-specific 3D print have been confirmed in studies abovementioned. It results better understanding of 3D spatial and anatomic relationship of an organ or a structure, especially in a lesion with complex anatomy, compared to using conventional diagram or images. Understanding of the surgical procedure is critical to quality of care and compliance.
The impact of education level of patients and guardians has been frequently studied previously. Thought some authors have found variances in perception or preference of communication[18, 19], most of them concluded the subjects’ understanding of information conveyed was not affected by different education levels [18–20]. In the current study, we also found there was no difference in ratings between guardians with lower and higher education levels, indicating 3D prints could be used in all patients/guardians regardless of their education levels.
As for consent duration, one study  has shown 5 minutes on average longer communication time using 3D prints compared to our insignificant result in this aspect. This is possibly due to study variance, such as wording in explanation. In our study, same durations produced better understanding comparing to using 2D diagrams. This highlighted the efficiency of 3D prints in showing the spatial relationship in the heart with a defect.
We only used pure white material to simply show the structure of the ventricular septal defect. Studies[22, 23] have demonstrated color coding models may result in more intuitive representations. For example, blue and red rendering for veins and arteries, respectively, are often used in clinical medicine. We did not use multiple colors for economic issue. Further studies may incorporate multi-color labeling to confirm if this method produces better communication results.
The cost of patient-specific 3D print has dropped dramatically overtime, but it is still significant. Since we in this study only included subjects with perimembranous ventricular septal defect and only used one print, it was cost effective. But we should keep in mind the complexity and individuality of patient with CHD. Therefore, the cost remains a major obstacle for conventional application when this technique is brought to communication with patients or guardians with wider range of congenital cardiac defects.
As abovementioned, the study population is relatively small and only included perimembranous ventricular septal defect subjects, limiting the generalizability of the results. Usually, CT or MRI scan is not used in diagnosing an isolated VSD for echocardiography is almost always enough and also to avoid unnecessary radiation or sedation, making data source for VSD 3D printing difficult to obtain. Since every lesion is unique, even for the same type of heart defect, the single 3D printed model obviously could not completely represent specific characters of each patient’s heart defect, such as variance in sizes, locations of the ventricular septal defect. Moreover, the model could not ideally show the relationship between tricuspid valve and the defect due to poor image definition of valvular structures obtained from CT scan. We chose perimembranous ventricular septal defect for study as it was the most common type of CHD. In view of the low incidence and wide variance of most of the other CHDs, it is difficult to use one replica for all patients of a specific lesion. The survey focused on young guardians’ response to the 3D models in a hospital setting, without any further follow-up, so the impact on post-discharge satisfaction of medical/surgical care or improved quality of life cannot be obtained. The consent process was also performed by the same surgeon who performed the operations, so there was inevitably a conflict of interest while filing the questionnaires for the family, though we consider it could have exert the same effect in the two groups. In addition, to increase the response rate of the survey we did not objectively assess the impact of the models on learning and knowledge acquisition.