Quality initiatives and the power of the database: What they are and how they run

The criteria by which healthcare is judged or measured are quality, accessibility, and cost effectiveness. To evaluate these criteria it is important to have a database. There are many strengths and weaknesses to large databases. They can be used as an indicator of the level of performance or quality, for clinical decision making, and as a measurement of cost effectiveness. They can also be useful in the evaluation and development of treatment algorithms and critical pathways for patients with entry level disease. In addition, they can measure patient access to healthcare and the appropriateness of care. It is important for these databases to appropriately adjust for preoperative risk factors that may influence outcome. Outcome in most of the databases is measured by mortality, but morbidity, functional status, quality of life, cost of care, length of stay, return to work, and patient satisfaction are also important outcomes. Factors that can influence the quality of the outcome data are the methods by which the data are collected, standardization of definitions, the currentness of the database, adequate numbers of patients and outcomes, and appropriate analytic techniques. It is important to feed back the data to the healthcare providers in a timely enough fashion so that processes and structures of care can be modified to improve treatment and results. The reliability of the databases and the validity must be substantiated for the healthcare provider to have confidence in the database. There are numerous shortcomings of these large databases, including not capturing the unusual risk factors and thus underestimating the risk of a given patient, gaming (where individual hospitals or groups overreport patient risk factors), lack of homogeneity of varying patient populations, and the accuracy of the data acquisition. It is interesting that all of the large cardiac surgery databases have demonstrated a decrease in risk-adjusted operative mortality since their initiation. Whether this is due to increased attention to processes and structures of care, to more complete reporting of risk factors, or to patient selection is unknown. The perfect risk stratification data system and database probably will never exist. Even with continued refinement it is doubtful that a predictability of greater than 80% to 85% (C-index) will be achieved. Risk-adjusted outcomes should therefore be used only as cues to trigger an assessment of processes and structures of healthcare in an effort to identify problems that can be remedied.