Review article: review of behavioral operations experimental studies of newsvendor problems for operating room management.

Operating room (OR) managers must plan staffing in the face of uncertain demand for OR time. Planning too much staffing results in underutilized OR time. Planning too little staffing causes overutilized time, which is approximately twice as expensive as underutilized time. Deciding how much staffing to plan for an OR is analogous to the classic newsvendor problem in operations research. A newsvendor must decide how much product to order based on its cost c and sales price p, plus estimates of the uncertain future demand for the product. The newsvendor problem has a simple mathematical solution. The correct amount of product to order is the (p - c)/p quantile of the demand for the product. This optimal order quantity is analogous mathematically to the number of hours of OR time for which staffing should be planned. We performed a systematic review of the behavioral operations experimental literature on newsvendor problems relevant to OR management. Student volunteers participating in experimental studies have great difficulty knowing how much product to order, given c, p, and the demand distribution. Decision making is only modestly improved by more frequent feedback. Even scores of rounds of ordering are insufficient for much learning to occur. Suboptimal decisions result from innate psychological biases. Students anchor on mean demand, make insufficient adjustments, and rely disproportionately on the most recent demand values. The behavior of OR managers who plan staffing for the OR is analogous to that of students participating in a newsvendor experiment. Month after month, an OR manager will plan too little staffing for the surgeon who consistently ends the day late and too much staffing for the surgeon who consistently does not fill an OR. Experimental studies of the newsvendor problem provide mechanistic insights into the reasons that OR managers make poor decisions when planning OR staffing. The students face no organizational factors or personality issues that potentially influence OR managers, strongly suggesting that observed psychological biases are innate and play an important role in influencing the decisions of OR managers. Results suggest that only the use of computerized decision-support systems can effectively circumvent these biases.