In Reply:-Dr. Atkin questions whether comparing anesthesia drug and supply cost among anesthesiologists will be productive. In our Table 4, we evaluated “cost per case” and “costs per unit” for anesthesia drugs and supplies among surgical specialties. The relative differences in cost per case between cardiac an neurosurgical anesthesiologists at Duke equaled $207:$63 = 3.3: 1. The relative differences using cost per unit = 3.9: 2.2 = 1.8: 1. Therefore, the use of cost per case suggested that the differences were larger between the surgical specialties than did our method of using cost per unit. Therefore, Dr. Atkin's comments are not referring to the use of cost per unit versus cost per case, but rather the strategy of comparing costs among anesthesiologists by any methodology. We suggest that the relevant question is not whether costs will be compared among anesthesiologists. The questions are (1) whether costs will be compared among anesthesiologists with any form of adjustment for variation in case mix and, if so, (2) how much money will anesthesia groups have to pay to collect the data required to report their cost data. Cost per unit provides some rudimentary form of case-mix adjustment while using data that are already collected by anesthesia groups for the purposes of billing for their services.

Dr. Atkin states that “the authors' work demonstrates the need to commit to building databases of relevant information.” Although we believe that information technologies will contribute to a reduction in perioperative costs, anesthesia groups do not need to purchase new information technologies to use our method. Anesthesia groups have two choices to achieve cost control for drugs and supplies:

Purchase information technologies to track drug and supply usage by individual anesthesiologists. Set quantitative cost goals within the anesthesia group on a cost-per-unit basis. Use information systems to provide quantitative feedback to individual anesthesiologists regarding how well they are satisfying group requirements. The advantage of this strategy is that anesthesiologists continue to have flexibility in their clinical practice. The disadvantage is that anesthesiologists are responsible for the cost outcome of their practice. As a result, individual assignment of cost may lead to inappropriate economic credentialing. In addition, an individual anesthesiologist's efforts to decrease costs associated with a practice may contribute to an increase in costs or a decrease in revenue for the institution. In any case, our manuscript provides a methodology to assist in this option, but it does not require that this option be selected.

Use clinical pathways or practice guidelines, or both, to suggest or regulate (not monitor), or both, appropriate usage within the anesthesia group. The advantage of this strategy is that it does not require the use of an as-sophisticated information system, because only deviation from the clinical pathways or practice guidelines, or both, need to be recorded. The disadvantage of this strategy is that it decreases anesthesiologists' flexibility in clinical practice. In addition, the impact of clinical pathways and practice guidelines on education is unclear. In any case, the use of cost per unit applies equally to this strategy as to the preceding strategy.

Our manuscript did not address the relative benefits of each of these two strategies.

Dr. Stonemetz suggests that the use of more comprehensive methods to adjust for variations in case mix may do a better job at evaluating the “costs of delivering care.” If an anesthesia group has these data available, then using such data is rational. We cannot determine whether the hypothesis is true our data. It is our opinion that it is not known yet whether anesthesia groups will benefit from purchasing more sophisticated information technologies to track these measures of perioperative risk. Studies need to be performed showing that incorporating these measures of risk into assessments of anesthesia drug and supply costs improves the accuracy of predicting these costs and that the improvement in accuracy is cost-effective relative to the cost required to collect the data. An advantage of monitoring cost per at the level of an anesthesia group is that the data are already collected (i.e., quarterly costs divided by quarterly units).

Dr. Stonemetz points out that cost per unit cannot only be used to adjust drug and supply costs for variations in case mix, but can also adjust labor costs (i.e., anesthesiologist and anesthetist salaries). We agree completely. The use of cost per unit is an exceedingly rational, straight forward, and inexpensive approach to measure anesthesia work. The fact that we focused our analysis on drug and supply costs should not mislead others into thinking that we were focusing on the most important issue. Measuring labor costs (salaries) using costs per unit is far more important, because the majority of perioperative costs are accounted for by salaries.

Finally, Table 1does have a typographic error. The correct value is 12 +/− 7.4.

Franklin Dexter, M.D., Ph.D.

Department of Anesthesia; The University of Iowa; Iowa City, Iowa;

(Accepted for publication September 1, 1998.)