Data quality bias: an underrecognized source of misclassification in pay-for-performance reporting?

Pay-for-performance (P4P) initiatives, in which provider reimbursement is linked to quality assessment, are receiving increasing attention as a possible approach to encouraging and accelerating quality improvement in America's health care systems. The potential of P4P programs, however, is constrained by the quality of data and information resources available for performance reporting. Accurate and reliable appraisal of health care quality is a challenging issue, as achieving recommended processes of care and desired health outcomes is influenced by a diverse range of interrelated factors occurring at multiple levels and arising from multiple sources within the patient encounter, health care system, and larger environment. The challenge of quality assessment is further complicated by the variable quality of data available for reporting by each provider. When data quality varies systematically among providers, a significant risk of inequity in assessment, and therefore reimbursement under P4P programs, may occur. The issue of data quality should be investigated and addressed before widespread implementation of P4P programs is pursued. Significant investment in data collection and reporting mechanisms may be required, especially in resource-limited settings, to achieve the intended effects and avoid increasing disparities in health care quality.