Clinical science in transplant is commonly driven by studying big data. In our field, there is unparalleled access to large, comprehensive data registries. When used well, these registries capture the full clinical trajectory of transplant care. Organ Procurement and Transplantation Network–mandated registries (United Network for Organ Sharing and Scientific Registry of Transplant Recipients datasets) track transplant recipients from initial waitlist registration through transplant. In addition, transplant-specific registries can be augmented by other national datasets not specific to transplant (United States Renal Data System, National Inpatient Sample, Agency for Healthcare Research and Quality), as well as public and private claims data and pharmaceutical data. Big data is readily available and very powerful. Despite the benefits of easy access, large sample sizes, and many variables, there are instances in which big data falls short. Large registry databases are built from many data-points. The quality out is only as good as the data that goes in. Missing or inaccurate data can create a big problem for researchers trying to use these registries to inform clinical decision-making. The level of granularity is often inadequate to answer some clinical questions.1 This results from the fact that these registries are built and maintained for purposes other than clinical research. Clinical research on rare diseases has been hindered by an inability to identify patients with the disease in large registry data and by a lack of granularity to capture clinically relevant variables. One example of this is Fontan-associated liver disease (FALD).
FALD is the consequence of the Fontan surgical procedure used to treat congenital single ventricle heart disease. The Fontan procedure is the definitive repair for congenital heart disease that results in single ventricle physiology. A series of procedures are staged from as early as infancy through early childhood, whereby a functional single ventricle is recruited to support pulsatile systemic circulation. Venous return from the inferior vena cava is diverted to the pulmonary arteries bypassing the heart. At its completion, the Fontan procedure creates a total cavopulmonary connection (Figure 1).2 The loss of pulsatile propulsion of blood from the right heart into the pulmonary circulation results in elevated central venous pressure and congestion within the liver. Over time, this congestion can progress to cause hepatic fibrosis and, in the worst cases, cirrhosis. It is likely that FALD develops in all post-Fontan patients, although it may be clinically silent and undiagnosed in the absence of biopsy. Hepatic bridging fibrosis is estimated to develop in 38% of post-Fontan patients by adolescence. By adulthood, many of these patients eventually require heart transplant, liver transplant, or both.3 Little is understood about the optimal timing and approach for transplantation.
In this issue of Transplantation, Kim et al,4 a multidisciplinary group from the American Society of Transplantation Community of Practice, describe their efforts to use big data in the study of FALD. Motivated by a paucity of literature on this topic, the group set out to harness the power of big data to inform clinical decision-making for adolescents and adults with FALD. They cast a broad net, querying several administrative datasets to build a national and international sample of patients with FALD. Within the United States, queries included Scientific Registry of Transplant Recipients, the Society of Thoracic Surgeons Congenital Heart Surgery Database, Vizient, Cerner Health, the Pediatric Heart Network, the Alliance for Adult Research in Congenital Cardiology, and The American College of Cardiology Impact Registry, and Cardiac Networks United. Outside of the United States, they queried the Australian and New Zealand Fontan Registry. Their search showed an overwhelming lack of Fontan- and FALD-specific diagnostic codes within administrative data. Even within Fontan-specific registries, the diagnosis of FALD is rarely captured. Fontan-specific registries also do not contain variables to indicate liver-related complications such as fibrosis, cirrhosis, or clinical decompensations of liver disease. Within transplant-specific registries, history of Fontan is not collected, and it is difficult to impossible to isolate which liver recipients underwent transplantation for the sequelae of FALD.
To better leverage the power of big data, the authors suggest that in the next iteration of the International Classification of Diseases by the World Health Organization, scheduled for 2022, codes specific to history of Fontan and FALD be developed. In addition, the authors urge that transplant registries create a heart-specific variable for “history of Fontan” and a liver-specific variable for FALD. An ideal registry for this disease would need to include the following: diagnostic coding to identify all individuals post-Fontan, longitudinal data that spans childhood to adulthood, multicenter national/international data collection, granular and clinically relevant variables pertaining both to cardiac and liver complications, and unique patient identifiers that can enable linkage with other large databases (claims, pharmaceutical, etc.). Creation of a comprehensive database would also require a centralized and universal entity for collection, storage, and maintenance of data. Collaboration between multiple stakeholders will be essential and might include the American Society of Transplantation, the American College of Cardiology, the Society of Thoracic Surgeons, and the National Institute of Health.
The study by Kim et al4 is important as it identifies a need for proper coding of novel diseases and disease complications within existing databases. The authors have shown that for FALD appropriate coding does not exist. Clinical research thus far lacks the power to draw meaningful conclusions for patient care. It is not because post-Fontan patients do not exist, but rather, there is no reliable way to identify these patients or the clinical consequences of their disease within existing registries. Transplantation has served as the model for using registry data to innovate clinical practice. There are likely many other rare diseases that could greatly benefit from coordinated efforts to improve coding so that the power of big data can be realized. Improving the building blocks of big data will increase its potential, power, and applicability.
1. Massie AB, Kucirka LM, Segev DL. Big data in organ transplantation: registries and administrative claims. Am J Transplant. 2014;14:1723–1730.
2. Gewillig M. The Fontan circulation. Heart. 2005;91:839–846.
3. Gordon-Walker TT, Bove K, Veldtman G. Fontan-associated liver disease: a review. J Cardiol. 2019;74:223–232.
4. Kim MH NA, Lo M, Kumar R, et al. Bid datain transplantation practice - the devil is in the detail - Fontan associated liver disease. Transplantation. In press