McMullin ST, Hennenfent JA, Ritchie DJ, Huey WY, Lonergan TP, Schaiff RA, Tonn ME, Bailey TC. A Prospective, Randomized Trial to Assess the Cost Impact of Pharmacist-Initiated Interventions. Arch Intern Med. 1999;159(19):2306-2309. doi:10.1001/archinte.159.19.2306
Hospital pharmacists make many recommendations that improve patients' quality of care and/or reduce drug costs. While the impact of quality-of-care interventions is difficult to quantify, those limited to cost savings could be assessed in a prospective, randomized fashion.
To assess the impact of pharmacist-initiated interventions on cost savings.
Six pharmacists at a large university hospital recorded patient-specific recommendations for 30 days. All quality-of-care interventions were completed by the pharmacists, but those strictly aimed at reducing costs were stratified by drug class and randomized to an intervention or control group. Pharmacists contacted physicians with cost-saving recommendations in the intervention group, while control group patients were simply observed.
Main Outcome Measure
Drug costs after randomization.
Most (n=967 [79%]) of the 1226 interventions recorded were aimed at improving quality of care. The remaining 259 (21%) provided equivalent quality of care, but at less expense. These cost-saving interventions typically involved streamlining therapy to less expensive agents (39%), discontinuing an unnecessary medication (25%), or modifying the route of administration (24%). The group randomized to receive a pharmacist's intervention had drug costs that were 41% lower than those in the control group (mean, $73.75 vs $43.40; P<.001). Interventions involving anti-infective agents had the greatest cost savings (mean, $104.08 vs $58.45; P<.001). For our institution, this extrapolates to an annual savings of approximately $394,000 (95% confidence interval, $46,000-$742,000). As expected, these interventions had no impact on length of hospital stay, in-hospital mortality, 30-day readmissions, or the need to readminister the targeted medication or restart intravenous therapy.
While interventions solely aimed at reducing costs represent a small portion of a pharmacist's activities, they can result in significant savings for an institution.
ALTHOUGH MANY investigators have evaluated the impact of clinical interventions by pharmacists, the economic benefits have never been fully assessed in a prospective, randomized trial.1 Some have compared their interventions with a control group, but most of these studies were not randomized, limited their interventions to specific medications (such as antibiotics or H2-receptor antagonists), or used patient charges rather than actual costs when determining the economic impact of their services.2- 16 We report the results of a large, prospective, randomized evaluation of pharmacist-initiated interventions at our institution using pharmacy acquisition costs to estimate true economic benefits.
Barnes-Jewish Hospital, a member of the BJC Health System, is a 1200-bed teaching hospital affiliated with the Washington University School of Medicine, St Louis, Mo. The Department of Pharmacy has approximately 60 full-time–equivalent pharmacists, including 12 clinical specialists who provide care in such specialized patient areas as solid organ transplantation, bone marrow transplantation, infectious diseases, neurology, oncology, cardiology, internal medicine, and critical care. Six of these clinical specialists participated in a prospective, randomized trial of pharmacist-initiated interventions. Five pharmacists attended physicians' rounds or reviewed patients' medication profiles on a total of 4 internal medicine wards (168 beds) and 5 intensive care units (ICUs) (84 beds). The sixth pharmacist assessed the appropriateness of physicians' requests on an antibiotic approval pager. All of the pharmacists were board-certified pharmacotherapy specialists and had practiced in their respective areas for several years.
Each pharmacist recorded all interventions over a 30-day period. For the purpose of this study, interventions were prospectively characterized as either quality of care or cost saving. While many of the quality-of-care interventions also had cost-saving opportunities (eg, dosage reduction for organ dysfunction or discontinuing a potentially toxic medication), these were not suitable for randomization because there was a potential to compromise care if the patient had been randomized to the control group. Only those interventions that provided equivalent quality of care at less expense were randomized. These interventions included discontinuing an unnecessary medication, switching to an oral formulation of the same medication, recommending a less expensive agent, or reducing the dosage based on clinical indication.
All interventions aimed at improving quality of care were completed by the pharmacists, but those that simply reduced costs were randomized to an intervention or control group. To help ensure equality between the groups, cost-saving interventions were stratified into 6 categories using the American Hospital Formulary Service17 therapeutic classification number. These categories included anti-infective agents, gastrointestinal drugs, cardiovascular drugs, central nervous system agents, autonomic drugs, and miscellaneous agents. Within each stratum, interventions were randomized in blocks of 4 using a computer-generated random number table. After identifying a potential cost-saving intervention, the pharmacists telephoned a separate investigator, who then opened a sealed envelope to reveal group assignments. Pharmacists contacted physicians with cost-saving recommendations in the intervention group, while control group patients were simply observed.
An intention-to-treat approach was used to assess outcome parameters after randomization: pharmacy acquisition drug costs, duration of target drug therapy, length of hospital stay, 30-day readmission rates, in-hospital mortality, and the need for readministration of the targeted medication. Severity of illness was assessed using the All Patient Refined Diagnosis Related Group (APR-DRG) score.18 Labor costs associated with preparing and administering medications were not included in the cost analysis; however, the pharmacists' time spent on rounds with physicians and reviewing medication profiles was recorded. Annual savings were estimated using the mean drug costs after randomization for the patients in the study period extrapolated to all patients for a full year.
Differences between the groups were analyzed using χ2 tests for categorical data, t tests for normally distributed continuous data, and Wilcoxon rank sum tests for nonparametric data and continuous data not following a normal distribution. All analyses were performed with SAS for Windows, version 6.11 (SAS Institute Inc, Cary, NC).
The pharmacists prospectively reviewed 5590 medication profiles and identified 1226 interventions. Of these, 967 (79%) were classified as improving quality of care. These interventions included dosage adjustments for organ dysfunction (15%), recommendations to discontinue (12%) or add (7%) medications, and provision of drug-related information to patients and health-care providers (13%). The remaining 259 interventions (21%) had potential cost-saving implications and were randomized to either an intervention (n=126) or control (n=133) group.
The intervention and control groups were well matched (Table 1). Anti-infective and gastrointestinal agents accounted for 61% and 33% of the cost-saving interventions, respectively. The outcomes after randomization are shown in Table 2. The pharmacists' recommendations were almost universally accepted by the patient's primary physician, as illustrated by the short duration of targeted drug therapy after randomization to the intervention group. The medications targeted for intervention were used for a mean of 2 days longer in the control group compared with the intervention group (2.4 vs 0.4 days; P<.001).
The pharmacy acquisition costs after randomization totaled $9809 in the control group and $5469 in the intervention group. Patients randomized to the pharmacist intervention group had drug costs that were 41% lower than those in the control group (mean, $73.75 vs $43.40; P<.001). The savings were similar whether patients were in an ICU (mean, $70.74 vs $43.33; P=.02) or on a general medicine ward ($76.72 vs $43.47; P<.01) at the time of randomization. The greatest savings were found in patients receiving anti-infective agents (mean, $104.08 vs $58.45; P<.001). Interventions for gastrointestinal agents resulted in less dramatic savings (mean, $16.70 vs $9.54; P=.05).
A total of 292 hours (mean, 1.6 h/d per pharmacist) was spent going on rounds with physicians and reviewing medication profiles. Using the mean difference in drug costs after randomization for the groups and considering only those interventions that were limited to cost savings, we concluded that pharmacists' interventions would have saved approximately $7900 (95% confidence interval [CI], $900-$14,800) if all of the interventions identified in the study period had been made (259 identified interventions×$30.35 savings per intervention). These patient care areas reported approximately 80,000 patient-days in 1997. Assuming the same rate of intervention (eg, 4.6% of patient-days), we estimated that approximately $113,000 (95% CI, $13,000-$212,000) is saved each year by the pharmacists working in these areas. If there were an equal opportunity for interventions in the hospital areas that were not included in the study (eg, oncology, bone marrow transplantation, solid organ transplantation, cardiothoracic ICU, and other medical and surgical floors), expanding the pharmacist's role in our institution could reduce the annual drug budget by approximately $394,000 (95% CI, $46,000-$742,000; based on 280,000 patient-days in 1997).
The randomized interventions, which were strictly aimed at reducing costs, had no impact on the quality-of-care indicators. No differences were found in length of hospital stay, in-hospital mortality, readmissions within 30 days, or the need to readminister the targeted medication or restart intravenous therapy.
This study provides several important insights regarding the value of clinical pharmacists in the hospital. First, clinical pharmacists make many recommendations that affect the care of hospitalized patients. In our study, pharmacists intervened in 22% of the patient profiles they reviewed. Second, the majority of pharmacists' interventions (79%) are aimed at improving quality of care. While our study design would not allow for the randomization of quality-of-care interventions, more than 25% of these suggestions also resulted in lower drug costs through dosage reductions or recommendations to discontinue a medication. These quality-of-care interventions may have affected other costly factors, such as length of hospital stay or readmissions, but we could not measure this with our study design. Third, cost-saving interventions can be made in this patient population without adversely affecting quality of care. This was a critical finding; it supported our assumption that the targeted interventions were providing an equivalent level of care, but at less expense. The fact that we did not find a significant reduction in the length of hospital stay is consistent with the findings of other prospective, randomized intervention studies.2,19,20 Finally, the savings from just those interventions that were aimed at reducing drug costs (eg, 21% of all interventions) are still substantial. The drug cost savings from these types of interventions are estimated to be nearly $400,000 per year at our hospital.
The strengths of our study include its prospective, randomized design; the inclusion of several different pharmacists practicing in a diverse patient population; and the use of pharmacy acquisition costs rather than patient charges in the economic analysis. The potential limitations of our study are its relatively short duration, the inability to include cost savings from quality-of-care interventions or those realized after discharge from the hospital, and the reliance on certain assumptions in the cost analysis—primarily, that there would be an equal opportunity for pharmacist-initiated interventions in the areas of the hospital that were not studied. We believe this is a reasonable assumption, given that most of the pharmacists' interventions involved agents that are used frequently throughout the hospital (eg, antibiotics and gastrointestinal agents) and that there was no difference in the savings between the ICU and non-ICU areas. In fact, we may find even greater savings in the other areas of the hospital by including expensive medications that are not routinely used in the areas we studied (eg, chemotherapeutic agents, ondansetron hydrochloride, granisetron hydrochloride, filgrastim, oprelvekin, immune globulin intravenous pentetate, antithymocyte globulin).
Our study was not intended to be a complete pharmacoeconomic evaluation; hence, we did not consider input costs, such as those associated with providing these services, in the cost analysis. Since our pharmacists have many responsibilities outside direct patient care and only spent a fraction of their time involved in the study, it is difficult to estimate their true labor costs. In addition, the pharmacists tracked only their total amount of time involved in patient care activities, not just that time it took to identify the cost-saving interventions. Nonetheless, using a mean hourly rate of $30 for a clinical pharmacist's time, and assuming that each intervention takes the same amount of time to identify, the labor costs associated with identifying these cost-saving interventions were approximately $1900 for the study period. Using this method to account for labor costs, the net annualized savings would be approximately $86,000 for the areas studied and $301,000 for whole hospital.
We would like to comment further on the fact that our hospital has a long tradition of clinical pharmacy services and a strong program of drug controls already in place that limit the use of many expensive medications (eg, atracurium besylate, cisatracurium besylate, vecuronium bromide, midazolam hydrochloride, propofol, abciximab, bumetanide, esmolol hydrochloride, enalaprilat, diltiazem hydrochloride injection). Only 4 patients in our study received one of these expensive autonomic or central nervous system agents, and only 3 received one of the cardiovascular agents. The use of antibiotics is also tightly controlled at our institution.21,22 An earlier prospective, randomized evaluation of an antibiotic streamlining program failed to document financial benefits beyond those restrictions that were already in place.20 It is likely that these hospital-wide programs positively influenced the prescribing behaviors of physicians in our control group, and this should be considered when extrapolating our findings to other institutions without such controls. We believe that patient-specific recommendations by pharmacists could result in even greater savings at other institutions, where the drug costs in the control group may be higher than what we observed.
Accepted for publication February 22, 1999.
Corresponding author: S. Troy McMullin, PharmD, Department of Pharmacy, Barnes-Jewish Hospital, 216 S Kingshighway, St Louis, MO 63110 (e-mail: firstname.lastname@example.org).