A recent paper by Frank R. Ernst, PharmD, RPh, and Amy J. Grizzle, PharmD, has updated a previously published study by Johnson and Bootman. Both of these studies estimated the annual cost of drug-related morbidity and mortality resulting from drug-related problems (DRPs) in the ambulatory care setting in the United States. The original study estimated the annual cost at $76.6 billion. Of the $177.4 billion costs in the new study, hospital admissions accounted for $121.5 billion (69%) per year, and long-term care admissions represented $32.8 billion (18%). Physician visits accounted for another $13.8 billion (8%), and emergency department visits and additional treatment cost more than $5.8 billion (3%) and $3.5 billion (2%), respectively. Among the most significant DRPs identified in the Ernst and Grizzle study are untreated indication, improper drug selection, subtherapeutic dosage, and failure to receive drugs, overdosage, adverse drug reactions, drug interactions, and drug use without indication.
I spoke with Dr. Ernst about the significance of these new data and how the results may affect physician and pharmacist prescribing practices today and in the future. The author of the study, Frank R. Ernst, PharmD, is an Eli Lilly and Company Health Outcomes Fellow, and his coauthor Amy J. Grizzle, PharmD, is Assistant Director, Center for Health Outcomes and PharmacoEconomic Research, College of Pharmacy, University of Arizona, Tucson.
Bernstein: What was the purpose of your paper?
Ernst: Our purpose was to update the 1995 estimate of $76.6 billion for the annual cost of drug-related morbidity and mortality that results from DRPs in the US ambulatory care setting.
Bernstein: Why did you decide to go back at this time to update the earlier figures?
Ernst: The previous study had relied on an expert panel to estimate the probabilities of drug-related treatment outcomes, such as are diagrammed in the article. In addition, the Johnson/Bootman study had used cost data from the early 1990s. We believed that data from the literature would provide empiric evidence of the cost of illness from DRPs. While we could find no literature to support making changes to those probability estimates, we were able to adjust the costs of the various outcomes to reflect year 2000 dollars.
Bernstein: Describe the origins of the cost data -- were they primarily from patient populations who had experienced a DRP, or were they from a more global cohort from which you extrapolated your information?
Ernst: All of the following costs were estimated from statistical data for the United States, and updated using the Consumer Price Index.
The average cost for a physician visit was calculated from the Physicians Fee Guide (Healthcare Consultants).
Average prescription cost was derived from a report on retail drug sales, brand name and generic, published in the journal Drug Topics.
Hospital costs were calculated from data published by the American Hospital Association.
Long-term-care costs were calculated from data pertaining to the number of annual admissions and average length of stay, published by the National Center for Health Statistics.
The average cost of emergency department visits was derived from a study by Dennehy and colleagues. These data were multiplied by the number of physician visits annually, as reported by the Centers for Disease Control and Prevention.
With the exception of the Dennehy and colleagues
data, these costs were primarily from the US population and extrapolated from there. The case mix of the institutions whose mandatory input allows government organizations to formulate these statistics is widely varied. For example, costs for providing treatment to patients with long, complicated, expensive hospitalizations are averaged in with the costs for patients needing only short, uncomplicated, and inexpensive treatment in many cases. Thus, the average costs are expected to be skewed by the expensive treatments, even if they are not as numerous as the inexpensive treatments. How much cost is skewed in this manner cannot be known; therefore, this is one limitation to the precision of the estimates that Johnson/Bootman and we have calculated. Nevertheless, no model is perfect.
Bernstein: Were you able to characterize the kinds of medication problems encountered most often in the study population, and how were you then able to arrive at the final overall cost of drug-related morbidity and mortality?
Ernst: The decision analytical model was used to characterize the probabilities from medication-related problems. Once the 734.5 billion physician office visits were applied to the model, the number of visits was multiplied by the probabilities and then by the costs, resulting in the overall estimate of $177.4 billion spent annually.
Bernstein: How can these data be used to encourage positive changes in prescribing practices among physicians and pharmaceutical care practices among pharmacists?Ernst: An interesting aspect to the model, if one believes in its construct, is that it provides one framework for prescribers to examine the stratified results of decisions after they are made. This type of pharmacoeconomic analysis has only been used since about the mid-1980s, and is the sort of analysis that pharmacoeconomists and health outcomes researchers are trained to evaluate, for just the purpose you mention. That is, it may indeed encourage positive changes in prescribing, once the outcomes and analyses are diagrammed and once cost estimates are applied. It is very exciting to think about how helpful this information can be.
Bernstein: Do you see these cost figures steadily rising in the future? If so, why? What can be done to curb this trend?
Ernst: These cost figures may very well rise in the future. To the extent that healthcare professionals can take pharmacoeconomic analyses and use them in decision making, the trend can be curbed. However, our society demands access to more and more sophisticated technology as it becomes available. The recent reports that new cancer therapies (and others) are on the horizon bear that out. Technology and the innovation research required to supply that demand are crucial to the equation, but technology costs money. The question remains whether we can deny therapy and technology when it is needed, just because it will drive the average cost of treatment upward.
Bernstein: How can pharmacists be part of the solution to help reduce DRPs? Do these study results reinforce the need for expanding the role of pharmacists as part of the medical team?
Ernst: Pharmacists are the medication experts. Intervention by pharmacists has been shown to decrease the costs of illness for many years, and expanding the role of pharmacists can definitely work to our advantage. Since drug-related morbidity and mortality continue to pose a serious medical and economic problem, as we say in our paper, more attention toward solutions is appropriate.
Bernstein: What are future areas of research as they relate to DRPs, and how would you like to improve on study methodologies?
Ernst: In the future, revisions of the model used might be helpful to the body of literature. In addition, more study of the rates and probabilities of the various outcomes would provide a better idea of the precision of the model. We would like to be able to estimate probabilities to reflect the empiric evidence.