Lessons from the unexpected: the importance of data infrastructure, conceptual models, and serendipity in health services research.

In examining the importance of data systems, conceptual models, and serendipity in understanding health services, the case is made for a vigorous and responsive data infrastructure and more emphasis on conceptual development. Particularly important is the development of data systems that can keep pace with changes in health care organization and patterns of care. Three examples--from managed care, deinstitutionalization, and physician remuneration--demonstrate the need to empirically examine seemingly obvious assumptions about health patterns and trends, and the lessons to be learned when assumptions are proved incorrect. Major future challenges include incorporating patient preferences into outcomes research, meaningful communication about treatment options and health plan choices, and understanding how organizational culture and norms affect decision processes.