Clinical Pharmacists and Inpatient Medical Care: A Systematic Review

Background  The role of clinical pharmacists in the care of hospitalized patients has evolved over time, with increased emphasis on collaborative care and patient interaction. The purpose of this review was to evaluate the published literature on the effects of interventions by clinical pharmacists on processes and outcomes of care in hospitalized adults.

Methods  Peer-reviewed, English-language articles were identified from January 1, 1985, through April 30, 2005. Three independent assessors evaluated 343 citations. Inpatient pharmacist interventions were selected if they included a control group and objective patient-specific health outcomes; type of intervention, study design, and outcomes such as adverse drug events, medication appropriateness, and resource use were abstracted.

Results  Thirty-six studies met inclusion criteria, including 10 evaluating pharmacists' participation on rounds, 11 medication reconciliation studies, and 15 on drug-specific pharmacist services. Adverse drug events, adverse drug reactions, or medication errors were reduced in 7 of 12 trials that included these outcomes. Medication adherence, knowledge, and appropriateness improved in 7 of 11 studies, while there was shortened hospital length of stay in 9 of 17 trials. No intervention led to worse clinical outcomes and only 1 reported higher health care use. Improvements in both inpatient and outpatient outcome measurements were observed.

Conclusions  The addition of clinical pharmacist services in the care of inpatients generally resulted in improved care, with no evidence of harm. Interacting with the health care team on patient rounds, interviewing patients, reconciling medications, and providing patient discharge counseling and follow-up all resulted in improved outcomes. Future studies should include multiple sites, larger sample sizes, reproducible interventions, and identification of patient-specific factors that lead to improved outcomes.

Clinical pharmacists are uniquely trained in therapeutics and provide comprehensive drug management to patients and providers (includes physicians and additional members of the care team). Pharmacist intervention outcomes include economics, health-related quality of life, patient satisfaction, medication appropriateness, adverse drug events (ADEs), and adverse drug reactions (ADRs). An ADE is defined as “an injury resulting from medical intervention related to a drug,” and an ADR is defined as “an effect that is noxious and unintended and which occurs at doses used in man for prophylaxis, diagnosis, or therapy.”^1(p20) Reviews have been published about clinical pharmacy services in various settings, including ambulatory care,^2-5 geriatrics,⁶ psychiatry,⁷ critical care,⁸ economic outcomes,^9,10 and health-related quality of life,¹¹ and a comprehensive review¹² was published in 1986. To our knowledge, no previous reviews have focused specifically on clinical pharmacist interventions in the inpatient setting. This type of review is of particular importance because most studies reporting medication errors and ADEs were in hospitalized patients, and with the growth of hospital medicine,¹³ there is increased focus on interventions to improve the care of hospitalized patients. Benefits of clinical pharmacists have also been used to support expansion of their scope of practice.¹⁴

Two recent Institute of Medicine reports recognized that pharmacists are an essential resource in safe medication use, that participation of pharmacists on rounds improves medication safety, and that pharmacist-physician-patient collaboration is important.^15,16 In a recent survey, 30% of hospitals (74% of hospitals with >400 beds) reported that pharmacists attend rounds, and the rate is increasing.¹⁷ The role of clinical pharmacists differs from that of traditional pharmacists in that they work directly with providers and patients to provide services not simply associated with dispensing of drugs. Many clinical pharmacists have completed residencies and are board certified in specialty areas such as pharmacotherapy, oncology, nutrition, and psychiatry. This qualitative systematic review evaluates the published literature on the effects of pharmacist interventions in controlled trials in hospitalized patients.

A medical librarian–assisted search of English-language publications from January 1, 1985, through April 30, 2005, was conducted using the following MEDLINE and International Pharmaceutical Abstracts search terms: clinical pharmacy, pharmaceutical care, pharmacy services, pharmacists, inpatient, hospitalization, intensive care unit, treatment outcome, outcome assessment, quality of health care, adverse drug reactions, medication errors, morbidity, and mortality. Hand searches of bibliographies of relevant articles and authors' personal files were performed. Studies presented only in abstracts, letters to the editor, editorials, surveys, reviews, pediatric studies, and studies with the primary intervention occurring in an ambulatory setting were excluded from this review (Figure).

Methods of identified studies were reviewed for required criteria, including pharmacy service or intervention described, control group used, and objective patient-specific health outcomes reported. Pharmacist interventions consisted of cognitive services not routinely associated with activities of dispensing or compounding medications. Order review or order clarification was considered to accompany the act of dispensing. Table 1 gives the elements of clinical pharmacist services acceptable for this review. Services or interventions occurring as a component of guideline or protocol implementation or provider education were excluded. Observational studies, descriptions of pharmacy interventions, and studies without a comparison or control group were also excluded. Studies reporting only pharmacoeconomic outcomes were excluded because these have been previously reviewed.⁹ A single pharmacist (A.B.H.) performed the initial systematic search as described, with subsequent review by a physician (P.J.K.) of all included and excluded studies. One year later, the process was repeated by another pharmacist-physician (B.J.M.) with physician review (P.J.K.). The senior author (J.L.S.) reviewed the final selection and process for completeness.

Of 346 publications identified, 164 were excluded because of publication type or study population, 52 did not use an intervention unique to clinical pharmacy, 82 lacked a control or comparison group, and 12 did not report outcomes necessary for inclusion (Figure). The 36 studies^18-53 reviewed are categorized according to the primary type of clinical pharmacist service: patient care unit pharmacist participation on rounds, admission or discharge medication reconciliation, and drug class–specific pharmacist services (Table 2A and Table 2B).

Two studies involved the intensive care unit (ICU). Leape et al¹⁸ implemented a trial of pharmacist participation in a medical ICU, comparing ADE rates before and after intervention and with a control ICU. Preventable ADEs decreased by 66%, from 10.4 per 1000 patient-days before the intervention to 3.5 patient-days after the intervention (P<.001), with no change in the control ICU, from 10.9 per 1000 patient-days before the intervention to 12.4 per 1000 patient-days after the intervention (P = .76). Actual ADEs also decreased in the study ICU, from 33.0 to 11.6 per 1000 patient-days (P<.001), with an increase in the control ICU, from 34.7 to 46.6 per 1000 patient-days (P<.001). In a medical progressive care unit, Smythe et al¹⁹ implemented a clinical pharmacist–structured evaluation of 131 patients during 8 weeks and reported fewer ADRs compared with baseline (1 vs 8 events; P = .03); ICU transfer, readmission rate, and hospital length of stay (LOS) did not differ between baseline and intervention.

Eight studies^20-27 assessed clinical pharmacists on general medicine, surgery and psychiatry services. Bjornson et al²⁰ evaluated a clinical pharmacist intervention involving medication reconciliation, drug therapy plans, and discharge counseling. Intervention teams had fewer patients transferred for more intensive care and their patients had shorter LOS, but hospital readmissions and mortality did not differ. There were more ADRs in the intervention group (1.7%) compared with the control group (0.5%), but no P value was reported. The authors attribute this to a higher propensity for pharmacists to document ADRs. Scarsi et al²¹ compared results in patients when a pharmacist participated on rounds with an inpatient medicine team compared with patients who received pharmacist services only on the first day of hospitalization or when requested. These authors reported reductions in medication errors, number of patients without a medication error during hospitalization, and duration that an error persisted once it occurred.

One of the first intervention trials of clinical pharmacists on patient care units by Clapham et al²² involved regular interaction with physicians, patients, and nurses compared with the more traditional role of centralized pharmacy drug monitoring. The intervention reduced total average cost ($1293; P<.05) and produced nonsignificant reductions in LOS and drug costs, and pharmacists found working on the patient care area more professionally rewarding. In similar studies by Haig and Kiser²³ and Boyko et al,²⁴ inclusion of pharmacists on general medical teams resulted in reductions in LOS and in hospital and pharmacy costs. Kucukarslan et al²⁵ found that a clinical pharmacist on the medicine team reduced preventable ADEs by 78%, but the number of events was small (2 vs 9; P = .02). The intervention was well accepted by physicians, with 98% of pharmacist recommendations accepted.

Owens et al²⁶ assessed a geriatric team pharmacist and found that the intervention resulted in fewer medications by day 3 (P<.05), with the greatest reduction in patients in nursing homes. Medication use was increased by day 3 in 40% of subjects in the control group vs 18% of patients in the intervention group (P<.005), and control subjects received more medications without indications (19% vs 11%; P<.025) and inappropriate medications (37% vs 20%; P<.005), with no difference in number of medications at 6 weeks and 3 months. In the 1 inpatient psychiatry study, Canales et al²⁷ showed significant improvements in clinical response (measured by psychiatric scales) and extrapyramidal symptoms, with no difference in medication costs and LOS.

Medication review and reconciliation was the primary target of 11 studies.^28-38 In the 2 admission interventions, Nester and Hale²⁸ found that medication histories taken by pharmacists, as opposed to nurses, resulted in more accurate medication and allergy information, identified allergy history errors more frequently, and entered allergy information into the computer more quickly, with no difference in drug interactions or ADRs. In a study from Australia, Stowasser et al²⁹ implemented a medication liaison service to improve communication between outpatient physicians and pharmacists and the inpatient team at admission and discharge. The intervention group was more likely to have a pharmacist intervene or change at least 1 medication during hospitalization, with no effect on LOS or mortality. At 30 days, the intervention group had fewer health care visits, nonsignificant reduction in readmissions, and no overall change in health status.

Another 9 studies focused on discharge counseling. Smith et al³⁰ performed home visits and assessed pharmacist discharge counseling on patient medication-taking behavior and found significantly better levels of medication adherence (P<.01), although 75% of patients in the intervention group and 96% of patients in the control group were not taking medications as prescribed. Bolas et al³¹ compared standard discharge planning with pharmacist discharge counseling coupled with a discharge letter from the inpatient physician to the patient's general practitioner. Significant improvement was noted in the correlation between discharge and home medications 10 to 14 days after discharge, as well as knowledge of drug name, dosage, and frequency, with no difference in readmission rates. In a Veterans Administration hospital discharge counseling intervention, Williford and Johnson³² reported that patients were no more knowledgeable or compliant at the 6-week follow-up.

In a study by Lipton and Bird,³³ pharmacists reviewed hospital records, consulted with physicians, provided discharge counseling, and made 4 follow-up telephone calls after discharge. At 2 months, patient medication knowledge was higher in the intervention group. At 3 months, patients in the intervention group compared with those in the control group received fewer medications (5.16 vs 6.75; P<.001) and fewer daily doses (8.30 vs 12.04; P<.001), and reported fewer missed doses (8% vs 22%; P<.001); resource use was not affected. From the same study, Lipton et al³⁴ evaluated a 236-patient sample in 6 domains of medication appropriateness. Patients in the intervention group were less likely to have one or more prescribing problems in any category, in appropriateness or in dosage.

Johnston et al³⁵ evaluated the role of pharmacist discharge counseling on medication knowledge in older patients. An evaluation immediately before discharge and a recall questionnaire found that the percentage of critical items correct for the pharmacist-counseled group was 93% compared with 77% in the control group (P = .02). Nazareth et al³⁶ reported no differences in hospital readmissions, outpatient visits, or mortality at 3 or 6 months for a discharge pharmacist intervention to coordinate care with outpatient pharmacists and providers in patients older than 75 years. In a similar study, Al-Rashed et al³⁷ enrolled 83 elderly patients at discharge and reported improvements in knowledge, compliance, outpatient visits, and hospital readmissions. In the most recent study of pharmacist counseling at discharge with telephone follow-up after 3 to 5 days, Schnipper et al³⁸ reported fewer preventable ADEs (1% vs 11%; P = .01) and fewer preventable medication-related emergency department visits or hospital readmissions (1% vs 8%; P = .03) at 30 days in the intervention group compared with the control group, with no difference in medication compliance.

Of the 15 drug class–specific pharmacist services,^39-53 4 studies^39-42 evaluated inpatient anticoagulation services. In a pharmacist-managed anticoagulation service in patients with venous thromboembolism, Mamdani et al³⁹ found no difference in time to therapeutic partial thromboplastin time (PTT), percentage of patients with supratherapeutic PTT levels, or number of blood samples drawn to measure PTT. However, the intervention group had a greater proportion of therapeutic PTT levels, shorter time from blood drawing to adjustment (2.8 vs 4.4 hours; P<.001), earlier initiation of warfarin sodium therapy, and shorter LOS. Dager et al⁴⁰ evaluated a clinical pharmacist anticoagulation service providing daily consultation and follow-up to patients beginning warfarin therapy compared with matched historical control subjects. The intervention was associated with reductions in excessive anticoagulation, major warfarin drug interactions, inpatient days receiving warfarin therapy, and less time spent at supratherapeutic international normalized ratios (INRs). In a study by Ellis et al,⁴¹ consultation was provided to patients receiving warfarin therapy to evaluate laboratory results, warfarin dosages, drug interactions, and outpatient anticoagulation follow-up. Patients in the intervention group had fewer INR and PTT measurements, improved discharge INR stability, rate of therapeutic INRs, and fewer supratherapeutic INRs at clinical follow-up; bleeding and thromboembolism rates did not differ. Finally, Tschol et al⁴² compared warfarin management by pharmacists and physicians after prosthetic valve insertion and found no difference in days in the therapeutic range, subtherapeutic INRs, or major bleeding, but pharmacists had 5.9% fewer days with an INR higher than 4.0 (P<.001).

Four studies^43-46 focused on antibiotic therapy and infectious disease consultation. Fraser et al⁴³ evaluated a clinical pharmacist–infectious disease fellow team receiving designated parenteral antibiotic therapy and reported that 62 (49%) of 127 patients in the intervention group received 74 suggestions, of which 85% were implemented. Changes in antibiotic choice, dosing, or administration route resulted in mean antibiotic cost savings of $400 (P = .05). Mortality, clinical response, antibiotic toxic effects, and LOS were similar. Gentry et al⁴⁴ evaluated the effects of an antimicrobial control program in which a pharmacist approved restricted and nonformulary antimicrobial agents and assisted the primary team with changes in therapy and culture report interpretation. A significant reduction in hospital mortality, LOS, and antimicrobial costs was observed, with no change in hospital readmissions. Bailey et al⁴⁵ studied a pharmacist intervention to contact the patient's physician with antibiotic recommendations and reported significant reductions in days with intravenous antibiotic therapy and antibiotic cost, no difference in mortality and LOS, but more readmissions. Finally, Gums et al⁴⁶ evaluated a team including a pharmacist, a microbiologist, and an infectious disease specialist for patients receiving intravenous antibiotic therapy. In 127 patients in the intervention group, LOS was shorter (5.7 vs 9.0 days; P = .0001), with no difference in mortality; 89% of pharmacist recommendations were accepted.

One of the first clinical services established by pharmacists involved therapeutic drug monitoring of aminoglycosides, vancomycin, anticonvulsant drugs, and theophylline. Seven studies^47-53 addressed the clinical value of providing therapeutic drug-monitoring services evaluating toxic effects, therapeutic effects, health care use, and appropriate drug concentration testing. Destache et al⁴⁷ reported shorter febrile periods, faster return to normal vital signs, and shorter LOS with aminoglycoside therapeutic drug monitoring.⁴⁸ Nephrotoxicity occurred less often in pharmacist-dosed patients in several studies, but was not statistically significant. In the vancomycin monitoring trial by Welty and Copa,⁴⁹ fewer patients developed vancomycin-related renal insufficiency. In other aminoglycoside trials, improvements were seen in peak concentrations and LOS,⁵⁰ appropriate aminoglycoside concentrations,⁵¹ and improved pharmacokinetic parameters.⁵² Wing and Duff⁵³ evaluated phenytoin therapeutic drug monitoring and reported significant reductions in number of assays performed or not indicated and blood samples drawn incorrectly and used inappropriately; the number of seizure-related readmissions was also reduced (all P<.05).

The results were further categorized by specific outcomes or process measures of interest. Twelve of 36 studies^{18-21,25,38,40,42,47,49,50,52} evaluated ADEs, ADRs, or medication errors as an end point: preventable and actual ADEs,¹⁸ preventable ADEs,^25,38 ADRs,^19,20 anticoagulation-associated complications,^40,42 medication errors,²¹ and nephrotoxicity.^47,49,50,52 This lack of a uniform definition of ADEs, ADRs, and medication errors prevents systematic generalization to clinical practice or meta-analysis. However, 7 reported a reduction,^{18-21,25,38,48} 5 reported no difference,^{40,42,47,50,52} and no studies reported statistically higher rates compared with controls.

Health care use was assessed in 24 of 36 trials^{19,20,22-25,29,31,33,36-41,43-51,53} by LOS, costs, laboratory testing, readmissions, ICU transfers, or health care visits. Hospital LOS was reduced in 9 of 17 trials, and readmission rates, ICU transfers, test use, and costs were either reduced or not affected. Only 1 trial⁴⁵ demonstrated an increase in use (hospital readmissions). Drug monitoring and process measures were evaluated in 15 trials,^{28,29,39-43,45,47-53} with significant improvements in anticoagulation, medication history and allergy documentation, antibiotic use, treatment response, and therapeutic drug levels. In 4 trials, clinical pharmacist recommendations led to reductions in the number of unnecessary medications and number of daily doses,^26,33 improved medication appropriateness and medications lacking an indication or known ADRs,^26,34 and fewer drug interactions.⁴⁰

Mortality was evaluated in 8 studies. One study⁴⁴ showed a significant reduction; of the other 7 studies, 3 demonstrated lower mortality^20,46,50 and 4 demonstrated higher mortality in the intervention group^36,43,45,47 but these differences were not significant (P>.05). Nine studies^29-33,35-38 used patient measures of medication adherence and knowledge as outcomes, with improvements demonstrated in 5 studies.^{30,31,33,35,37} Only 1 study²⁹ evaluated health-related quality of life, with no improvement in the 36-Item Short-Form Health Survey scores, and psychiatric scales showed improvement in 1 psychiatry intervention.²⁷

Our review supports the use of clinical pharmacists in the inpatient setting to improve the quality, safety, and efficiency of care. The Institute of Medicine report Crossing the Quality Chasm⁵⁴ proposes that clinical pharmacists have a significant role in addressing quality issues in hospitalized patients, and the Joint Commission on Accreditation of Healthcare Organizations mandates medication reconciliation at the time of hospital admission and discharge.⁵⁵ By further developing collaborative health care, the clinical pharmacist can be an integral part of the inpatient care team. Our findings are supported by a large observational study by Bond et al⁵⁶ that identified 17 clinical pharmacy services in hospitals associated with improvement in mortality, drug costs, cost of care, and LOS. In a follow-up study, Bond et al⁵⁷ reported lower medication error rates as the number of clinical pharmacists increased per occupied bed.

Implementing new hospital programs is difficult, especially if they require allocation of new resources. One fundamental advantage to the pharmacist interventions discussed is that most can be implemented through reallocation of existing resources to increase clinical pharmacist services. Published studies evaluating the cost of incorporating clinical pharmacists have generally demonstrated a net hospital cost benefit in terms of cost avoidance and use.^9,58-60 In some settings, new pharmacy positions (eg, technicians) have been created to fill expanded clinical roles and pharmacist duties have been reorganized to enable more direct interaction with patients and physicians on rounds.

There are many limitations to this systematic review and included studies. Many pharmacist intervention studies have small sample size, and most are single-institution studies, limiting generalizability. Interventions are costly, limiting sample size and increasing the chance of a type II error. Different study designs were used (ie, randomized, cohort, case-control, preintervention vs postintervention), and each has limitations. It is difficult to standardize interventions; thus, reproducing them is challenging, limiting comparisons, and it is impossible to combine results, as in meta-analysis. Determining outcome measurements in pharmacist intervention trials is difficult. Many definitions for outcomes such as ADEs, ADRs, and medication errors are confusing and not consistent. Process measures (eg, drug levels) are frequently used and may not be related to outcomes. Health care use is often measured because it is easily quantified, is generalizable, and can be used to justify increased pharmacist personnel costs. Systematic reviews are subject to publication bias, although it is unlikely that the addition of a clinical pharmacist to a medical care team would adversely affect outcomes. This review is retrospective and observational, and, therefore, subject to systematic and random error, and it did not include studies before 1985 because of significant changes in pharmacy services that may make earlier studies less relevant.

More research is needed to better understand the role of clinical pharmacists, clinical areas most likely to benefit, and patient-specific factors associated with improvements. Cost-effectiveness can also be improved by identifying pharmacist duties most beneficial to patients and determining whether less skilled and costly personnel can perform other duties. Future studies should describe interventions in sufficient detail that they can be reproduced, and outcomes such as medication appropriateness and adherence should be measured using validated instruments. Last, larger, multicenter, randomized, controlled trials should be conducted to prove that benefits of pharmacist interventions are generalizable across institutions and to quantify the value to the health care system.

Correspondence: Peter J. Kaboli, MD, MS, Division of General Internal Medicine, Department of Internal Medicine, University of Iowa Carver College of Medicine, University of Iowa Hospitals and Clinics, SE615GH, 200 Hawkins Dr, Iowa City, IA 52246 (peter-kaboli@uiowa.edu).

Accepted for Publication: November 22, 2005.

Financial Disclosure: None.

Funding/Support: This study was supported by the Department of Veterans Affairs, Veterans Health Administration, Health Services Research and Development Service (VA HSR&D) SAF 98-1521. Dr Kaboli is supported by Research Career Development Award RCD 03-033-1 from the VA HSR&D. Dr Schnipper is supported by Mentored Clinical Scientist Development Award HL072806 from the National Heart, Lung, and Blood Institute, National Institutes of Health.

Disclaimer: The views expressed in this article are those of the authors and do not necessarily represent the views of the Department of Veterans Affairs.