Background: Despite progress in describing the problem ofpotentially inappropriate medication (PIM) use, there have beenfew prospective studies demonstrating that interventions with specificmedication criteria can make a difference in decreasing theuse of problematic drugs in older adults.
Objective: To design an intervention study to change physicianbehavior regarding PIM prescribing to older patients.
Study Design and Methods: A prospective randomized blockdesign was used during an 18-month period from January 2001 toJune 2002. The study population was primary care physicians (n =355) in the Medicare + Choice product line of a southeastern managedcare organization and their patients 65 years and older. Therewere 170 physicians in the treatment group and 185 in the controlgroup. Physicians were assigned to the treatment or usual-caregroups using a randomization table, and each group includedphysicians who had and had not prescribed a PIM.
Results: Approximately 71% (84/118) of the physicians in theintervention group who prescribed a PIM completed and faxedback at least 1 potentially inappropriate medication form to themanaged care organization. On 15.4% (260/1692) of the medicationforms, physicians made some change regarding PIM use.
Conclusions: Although many studies have addressed medicationuse among older adults, intervention studies aimed at influencingphysician prescribing in this population are limited. Thisstudy describes a low-cost, replicable method to contact and educatephysicians on drug therapy issues in older adults.
(Am J Manag Care. 2004;10:761-768)
Influence over physician prescribing presents anopportunity and a challenge to managed care organizations(MCOs). As a proportion of total MCOexpenditure, the pharmacy budget represents a significantand increasing amount. The number of prescriptionsdispensed annually in the United States rose from$2 billion in 1992, at a cost of approximately $50 billion,to $3 billion in 1999, at a cost of more than $129billion.1 Older persons are at greater risk for inappropriateprescribing because they are generally takingmore medications, have multiple comorbid diseases,and experience changes in what their aging body doesto the drugs. This article describes a simple and effectivemeans for MCOs to address the challenge of prescribingof potentially inappropriate medications (PIMs)to older Americans.
The prescribing of PIMs, defined as drugs with risksthat outweigh the potential benefits when prescribed toolder patients,2 may be related to preventable problemsin older patients such as depression, constipation,immobility, confusion, and hip fractures.3-5 Other consequencesmay include oversedation, cognitive impairment,increased falls, and fractures.6-8 At worst, PIMprescribing has been associated with significant increasesin mortality and morbidity among older patients.9,10
The prevalence of PIM prescribing is high and hasbeen well documented, but there have been few studiesdescribing PIM use in managed care. A previousdescriptive study10 of PIM use in the managed care settingfound a 23.2% (541/2336) prevalence of PIM use.This is comparable to recent national survey data ofPIM use indicating that approximately 25% (about 6.64million) of community-dwelling older adults were prescribedat least 1 PIM.11 In addition, studies5,12,13 havedemonstrated PIM prescribing and adverse drugevents within nursing home and hospitalized populations.The 1996 Medical Expenditure Panel Surveyfound that 21.3% of community-dwelling older adultsin the United States received at least 1 PIM from a listof 33 PIMs studied.14
Information extracted from the 1995-1996 NationalHospital Ambulatory Care Survey estimated that 251 017visits to outpatient departments in the United Stateswere "drug related," and, of those, the highest visit ratewas seen in patients older than 74 years.15 This reportconcluded that resource use could be reduced withmanagement of drug-related problems. A cohort studyof Medicare enrollees (30 397 person-years of observation)cared for by a multispecialty group practice identified1523 adverse drug events, of which 421 (27.6%)were preventable.3
In older ambulatory patients, inappropriate medicationuse led to reactions that necessitated healthcareservices, physician contact, hospitalization, and emergencydepartment visits.16 Indeed, for many olderadults, these drugs may be linked to increased hospitaladmissions and length of stay.10,12 Other research withinan older managed care population has shown that, asone may expect, PIM use is associated with higher costsand use.10
Last, up to 30% of hospital admissions in older personsmay be linked to medication toxicity and drug-relatedproblems that, in turn, have profound health,safety, and economic consequences for older adults.17 Drug complications are the most common cause ofadverse events in hospitalized patients.18
The aforementioned concerns illustrate the magnitudeof inappropriate prescribing among older peopleand quality improvement opportunities that present inthe ambulatory setting. To date, however, research onPIM prescribing in older patients has focused ondescriptive studies; that is, categorizing specific drugsthat may cause adverse outcomes and describing thesuboptimal prescribing habits of physicians.19-21 Of thetools available to measure PIM use in older adults, perhapsthe most widely used are the Beers criteria. Theseconsensus criteria identify specific medications thatshould be avoided in older persons.2,21 Individual medicationsfrom the Beers criteria have been associatedwith adverse events and drug-related problems (such asthose associated with flurazepam hydrochloride anddiphenhydramine hydrochloride) in several retrospectiveand a few prospective studies.22,23
Despite progress in describing the scale of the problem,there have been few prospective studies demonstratingthat interventions with specific medicationcriteria can make a difference in decreasing the use ofproblematic drugs in older adults. A multidisciplinary2001 study24 to decrease benzodiazepine use in olderadults did not change the baseline rate, but decreasedthe number of inappropriate benzodiazepine prescriptions.Other studies25-27 have addressed the use of academicdetailing to change the practice of antidepressantand antihypertensive use among older adults or havefocused on preventing medication errors. The objectiveof this study was to design and test an intervention tochange physician behavior regarding PIM prescribing toolder patients. This study was part of a larger qualityimprovement effort between an academic medical centerand an MCO to decrease PIM use and improve qualityand medication safety for older adults.
The specific aims of this study were (1) to examinethe effectiveness of a strategy to change physician prescribingbehavior and decrease PIM use in members ofa southeastern MCO, and (2) to study the effect of anintervention performed by a southeastern MCO.
Hypotheses included the following: (1) At least 50%of the physicians will respond to the intervention byfaxing back the medication forms. (2) An interventiondirected toward primary care physicians to decreasePIM use will decrease the overall number of PIMs prescribedin the intervention group.
A prospective randomized block design was used.
The study comprised an 18-month period fromJanuary 2001 to June 2002.
The study population consisted of all primary carephysicians (n = 355) in the Medicare + Choice productline of a southeastern MCO and their patients 65 yearsand older. There were 170 physicians in the treatmentgroup and 185 in the control group. Physicians weregrouped by practice site and then assigned to the treatmentor usual-care groups using a randomization table,and each group included physicians who had and hadnot prescribed a PIM (Figure 1).
Potentially inappropriate medications were definedusing the 1997 Beers2 criteria for medications to avoidin older adults (Table 1). This list is based on expertconsensus developed through an extensive literaturereview with a bibliography and questionnaire evaluatedby nationally recognized experts in geriatric care, clinicalpharmacology, and psychopharmacology using amodified Delphi technique.28
In January 2001, all primary carephysicians in the treatment and controlgroups received a mailing consistingof an educational letter and abrochure entitled "The Challenges ofPrescribing to Seniors" along withthe Beers criteria list noting problematicdrugs for the older (>65years) patient. The mass mailing wasa requirement by the legal office ofthe MCO. In March 2001, the physiciansin the intervention groupreceived an integrated decision supportservice that comprised 3 maincomponents: (1) a detailed educationalbrochure listing PIMs, (2) a listof suggested PIM alternative medications(that were independently suggestedand reviewed by 5geriatricians and pharmacists notaffiliated with the MCO), and (3) apersonally addressed letter from theMCO that described in detail all thephysician's patients who were determinedto be in receipt of 1 or morePIMs as determined by the administrativedatabase of the MCO. A panelof 5 experts (physicians and pharmacists)not affiliated with the MCO orthe academic medical center performeda peer review of the drugs tobe included in the intervention andtheir corresponding alternative medications.This was planned to remove any potential conflictof interest or financial implications regarding themedications included as PIM alternatives.
The data contained in the letter represented prescribinghabits from the previous calendar quarter. Forexample, in March 2001, physicians in the interventiongroup received information about their PIM claims duringthe fourth quarter of 2000. Physicians with no PIMprescribing data on file received an acknowledgment ofthis. The physician letter was updated each quarter for12 months (ie, each physician received 4 letters).
In addition, all physicians in the intervention groupwere invited to comment on any PIM prescribing changesvia use of a fax-back form (appendix available from theauthor). The faxable form served as a permanentacknowledgment by the physician of patient assessmentor therapy modification and actions taken. The formallowed the physician to indicate his or her PIM action in1 of 5 ways (discontinued medication, assessed patientwith no change indicated, decreased dosage or frequency,prescribed an alternative medication, or did not prescribethis medication). If the patient had not been seensince the prescription was filled, physicians were asked toplace the PIM fax-back form into the patient's medicalchart until the next visit, at which time it would serve asa reminder for patient assessment. Physicians in thecontrol group received no interventions (other than themass mailing in January 2001). The estimated annualcost of the intervention for the MCO was $3456 (appendixavailable from the author).
Data Analysis and Management
To allow for claims run-out, pharmacy claims fromeach quarter were pulled approximately 45 to 60 daysfollowing the end of the quarter. Patient-specific reportswere prepared for the related primary care provider andmailed to the physician, along with a fax-back form forthe physician's documentation of patient assessment(regarding adverse drug effects) or any changes in medicationtherapy. On receipt of the forms in the QualityManagement department, physician responses weredouble-entered into a dedicated database for tabulation.
Statistical significance was assessed at α = .05, and allstatistical analyses were performed in SAS version 6.12(SAS Institute, Cary, NC). A descriptive analysis was performedto determine physician response, documentedaction regarding PIMs, and type of PIMs responded to.Basic χ2 analysis was performed to examine differencesbetween the baseline and study periods in the percentageof members 65 and older and continuously enrolled duringthe measurement period who filled 1 or more PIM prescriptionsbetween January 1 and June 30 of eachmeasurement year.
Physician Response and Change Rate
Only 69.4% (118/170) of the providers in the treatmentgroup prescribed at least 1 PIM. Approximately71% (84/118) ofthose physicianswho prescribed aPIM responded to atleast 1 PIM by completingand faxingthe medication formback to the MCO.Figure 2 illustratesthe physician actionon the fax-backforms. Documentedactions includeddiscontinuing medication,changingthe patient to a more appropriate medication or dosage,documenting evaluation of the patient for PIM effects, orintending to review the therapeutic alternative at thenext patient visit. On 15.4% (260/1692) of the medicationforms, physicians made some change regarding PIMuse. They discontinued the medication in 12.5%(211/1692) of cases, decreased the dosage of the medicationin 1.7% (28/1692), and prescribed an alternativein 1.2% (21/1692) (Figure 2).
Physicians were most likely to send a faxedresponse back regarding antidepressants, musclerelaxants, and analgesics or anti-inflammatory medicationclasses (Figure 3). Table 2 lists the physicianaction on PIMs by medication class. The drug classesmost likely to be discontinued were antihistamines32.2% (55/171), followed by analgesics 16.2% (56/345)and muscle relaxants 13.9% (55/395). Table 3 lists thephysician response by specific PIM medications indescending order of those most likely to be changed bydiscontinuing them, decreasing the dosage, or prescribingan alternative.
Overall Change in Appropriate Prescribing
During the 6 months after the end of the study period(January 1, 2003, to June 30, 2003), the number ofcontinuously enrolled members with at least 1 PIMdeclined significantly (χ2 = 13.20, <.001) to 17.9%(3007/16 818), from a baseline of 19.4% (3364/17 330)at the start of the study period. During our study period,major changes occurred in the primary care physiciannetwork, with 78 primary care providers leavingthe network, 129 joining the network, and the MCO discontinuingtheir contract with the Centers for Medicare& Medicaid Services, so we did not conduct a furtheranalysis of PIM use at the provider level.
The challenge of suboptimal prescribing is multifaceted,including overuse of medications (ie, polypharmacy),inappropriate use, and underuse. Theprescribing process extends beyond the physician,involving associated healthcare providers, pharmacists,and the patient; therefore, education and awareness ofPIMs are vital for providers and patients.
Although many studies have addressed medicationuse among older adults, intervention studies aimed atinfluencing physician prescribing in this older populationare limited. This study describes a low-cost, replicablemethod to contact and educate physicians on drugtherapy issues in older adults. This project was notdesigned, and was never intended, to supercede theclinical expertise and judgment of the physician. Byproviding the physician with this information, the MCOsupports the clinical expertise of the physician to assessthe medications forpatient safety versusbenefit. This replicablelow-technology interventionappears to be effectivein the short term inchanging PIM prescribingamong physicians ina southeastern MCO.
Our response rate of71.2% is better than thatin previous studies. Astudy by Monane et al29 used educational interventionsby pharmacistsand demonstrated thepharmacists' ability to reach physicians via telephone56% of the time. Fifteen percent of the physicians whomthe pharmacists reached agreed to change PIMs immediately,while another 9% agreed to review and changePIMs at the next visit.29 The resources of a pharmacist,however, were required to contact physicians in theseinterventions. Although a change rate of 15.4% is higherthan in some studies, a significant percentage ofphysicians did not change their behavior. Possible barriersto this include physician awareness of geriatricdrug problems, overuse of medication faxes and lettersto physicians, an inability to control who actuallyreceives the fax, potential delay in when the physiciancontacts the patient, and the retrospective nature of theclaims notification (the member has already filled theprescription). An in-person or personal telephone contactmay have resulted in a higher response and changerate. Future studies targeting inappropriate medicationsshould recognize that drug use is a complex issue andconsider a multicomponent approach targeting thephysician, patient, and other healthcare professionals.This is the focus of our future research and includesnurse assessment, primary care physician and pharmacyconsultation, and patient and physician education.
Intervention studies of the type we conducted areimportant because it is well known that the number ofmedications that older persons consume is high, and asthe number of drugs prescribed increases, the risk ofdrug-related problems increases. The intervention wedeveloped prompts the physician or office staff to assessmedications that patients are taking. Figure 2 illustratesthat 78.4% of the physicians assessed the patient even ifthey did not make changes in PIMs. Some of these maybe medications that the physician did not realizepatients were taking or that were initially prescribed byanother physician for short-term use but were neverdiscontinued.
Second, this is an important and growing populationto target for interventions in managed care. Olderadults use most health services in the United States andhave multiple physical, social, and psychological conditions,making the assessment, planning, and delivery ofhealthcare more complex than in younger adults.However, most physicians and nurses have little to nobackground in geriatrics. Therefore, interventions thattarget the care of older persons are increasinglyvital.30,31 Future studies should use the best evidenceavailable to change prescribing. The Beers criteria haverecently been revised and published.32 Criteria such asthese should be regularly updated when used inresearch and practice. Although the alternative medicationsmay be more expensive (because many ofthese drugs are newer medications), future studiesshould measure whether the benefits from the potentialimprovement in outcomes and healthcare cost savingsoutweigh the extra expense related to the cost of newermedications and the cost of the intervention.
This study has some limitations. First, the use of fax-backforms may not be the best means of communicatingwith the prescribing physicians. Because it is faxedcommunication, we cannot be sure whether it was thephysician or another office staff member who respondedto the intervention. In addition, the use of fax-backforms, although convenient and low-cost, may havekept the physician from immediately having questionsor concerns discussed with the pharmacist. Second, theuse of population consensus criteria may not be generalizablefor all individual patients. Third, there is apotential bias with regard to the initial educationalbrochure given to all physicians 3 months before thetargeted intervention. However, the minimal effect ofmass mailings to healthcare providers is noted, and bothgroups in our study received the mailing. Fourth, Avorn8 reported that the role of the patient is crucial and thatprescribing improvements directed solely at educatingthe physician or prescriber are doomed. Our study didnot involve the patient in the intervention. Last, it is possiblethat discontinuing the PIM and prescribing a costliermedication may increase prescription costs. Thisstudy has the limitations of other MCO studies, includingfluctuating benefits during the study period, patientdisenrollment, and changes in the physician panel.
Despite these limitations, this intervention is simpleand has shown a positive physician response rate(71.2%) and a decrease in overall PIM prescribing in asample of older adults. The transfer of knowledge andscientific advances into meaningful healthcare decisionmaking is an important challenge for all researchers.Thoughtful application using the Beers criteria andother tools for identifying PIM use can enable providersand insurers to plan interventions aimed at decreasingdrug-related problems in older persons and improvingquality of care.
We would like to acknowledge Wayne Hoffman, MD; Michael Kinstler,MD; Gary Tadlock, RPh; Marc Gottlieb, MPA; and Cheryl Harris, RN, BS, ofBlueCross and BlueShield of Georgia for contributions to the study interventionand data access and management.
From the Departments of Medicine (DMF, JRM) and Biostatistics (JLW) and Center forHealthcare Improvement (DMF), Medical College of Georgia, and Department ofPsychology, Augusta State University (RLR), Augusta, and Georgia and Kerr L. WhiteInstitute (JRM) and BlueCross and BlueShield of Georgia (NAR, LS, RVH), Atlanta, Ga; andSchool of Nursing, College of Health and Human Development, The Pennsylvania StateUniversity, University Park (DMF).
This study was funded by the Medical College of Georgia and BlueCross andBlueShield of Georgia Center for Healthcare Improvement.
Address correspondence to: Donna M. Fick, PhD, RN, School of Nursing, College ofHealth and Human Development, The Pennsylvania State University, 201 Health andHuman Development, University Park, PA 16802. E-mail: firstname.lastname@example.org.
1. National Association of Chain Drug Stores. Industry facts, 1999: 1999 retail Rxsales projected to rise 18 percent, surpass $121 billion on volume of nearly 3 billionprescriptions [press release]. Alexandria, Va: National Association of ChainDrug Stores; August 29, 1999.
Arch Intern Med.
2. Beers MH. Explicit criteria for determining potentially inappropriate medicationuse by the elderly. 1997;157:1531-1536.
3. Gurwitz JH, Field TS, Harrold LR, et al. Incidence and preventability of adversedrug events among older persons in the ambulatory setting. 2003;289:1107-1116.
4. Cumming RG. Epidemiology of medication-related falls and fractures in the elderly. 1998;12:43-53.
J Am Geriatr Soc.
5. Beers MH, Fingold SF, Ouslander JG, et al. Characteristics and quality of prescribingby doctors practicing in nursing homes. 1993;41:802-807.
N Engl J Med.
6. Thapa PB, Gideon P, Cost TW, Milam AB, Ray WA. Antidepressants and therisk of falls among nursing home residents. 1998;339:875-882.
7. Bates DW, Spell N, Cullen DJ, et al, Adverse Drug Events Prevention StudyGroup. The costs of adverse drug events in hospitalized patients. 1997;277:307-311.
Int J Technol Assess HealthCare.
8. Avorn J. The prescription as final common pathway. 1995;11:384-390.
J Am Geriatr Soc.
9. Hanlon JT, Schmader KE, Ruby CM, Weinberger M. Suboptimal prescribing inolder inpatients and outpatients. 2001;49:200-209.
J Manag Care Pharm.
10. Fick DM, Waller JL, Maclean JR, et al. Potentially inappropriate medicationuse in a managed care population: association with higher costs and utilization. 2001;7:407-413.
11. Wilcox SM, Himmelstein DU, Woolhandler S. Inappropriate drug prescribingfor the community-dwelling elderly. 1994;272:292-296.
Int J Geriatr Psychiatry.
12. Furniss L, Craig S, Burns A. Medication use in nursing homes for elderly people. 1998;13:433-439.
J Clin Epidemiol.
13. Passaro A, Volpato S, Romagnoni F, Manzoli N, Zuliani G, Fellin R, GruppoItaliano di Farmacovigilanza nell'Anziano. Benzodiazepines with different half-lifeand falling in a hospitalized population: the GIFA study. 2000;53:1222-1229.
14. Zhan C, Sangl J, Bierman A, et al. Potentially inappropriate medication use inthe community-dwelling elderly: findings from the 1996 Medical ExpenditurePanel Survey. 2001;286:2823-2829.
15. Aparasu RR, Helgeland DL. Visits to hospital outpatient departments in theUnited States due to adverse effects of medications. 2000;35:825-831.
16. Hanlon JT, Shimp LA, Semla TP. Recent advances in geriatrics: drug-relatedproblems in the elderly. 2000;34:360-365.
J Am Geriatr Soc.
17. Hanlon JT, Schmader KE, Kornkowski MJ, et al. Adverse drug events in highrisk older outpatients. 1997;45:945-948.
N Engl J Med.
18. Leape LL, Brennan TA, Laird N, et al. The nature of adverse events in hospitalizedpatients: results of the Harvard Medical Practice Study II. 1991;324:377-384.
Am J Med.
19. Hanlon JT, Weinberger M, Samsa GP, et al. A randomized, controlled trial ofa clinical pharmacist intervention to improve inappropriate prescribing in elderlyoutpatients with polypharmacy. 1996;100:428-437.
20. McLeod JP, Huang AR, Tamblyn RM, Gayton DC. Defining inappropriatepractices in prescribing for elderly people: a national consensus panel. CMAJ.1997;156:385-391.
Arch Intern Med.
21. Beers MH, Ouslander JG, Rollingher J, Reuben DB, Brooks J, Beck JC. Explicitcriteria for determining inappropriate medication use in nursing home residents:UCLA Division of Geriatric Medicine. 1991;151:1825-1832.
Arch Intern Med.
22. Agostini JV, Leo-Summers LS, Inouye SK. Cognitive and other adverse effectsof diphenhydramine use in hospitalized older patients. 2001;161:2091-2097.
Arch Intern Med.
23. Ensrud KE, Blackwell T, Mangione CM, et al. Central nervous system activemedications and risk for fractures in older women. 2003;163:949-957.
Intern Med J.
24. Elliott RA, Woodward MC, Oborne CA. Improving benzodiazepine prescribingfor elderly hospital inpatient audit and multidisciplinary feedback. 2001;31:529-535.
Am J Hypertens.
25. Siegel D, Lopez J, Meier J, et al. Academic detailing to improve antihypertensiveprescribing patterns. 2003;16:508-511.
26. van Eijk ME, Avorn J, Porsius AJ, de Boer A. Reducing prescribing of highlyanticholinergic antidepressants for elderly people: randomised trial of group versusindividual academic detailing. 2001;322:654-657.
27. Classen D. Medication safety: moving from illusion to reality. 2003;289:1154-1156.
The Delphi Method, III: Use of Self Ratings toImprove Group Estimates.
28. Dalkey N, Brown B, Cochran S. Santa Monica, Calif: Communications Department,RAND, prepared for a US Air Force Project. November 1969. RM-6115-PR.
29. Monane M, Matthias DM, Nagle BA, Kelly MA. Improving prescribing patternsfor the elderly through an online drug utilization review intervention: a system linkingthe physician, pharmacist, and computer. 1998;280:1249-1252.
Am J Geriatr Psychiatry.
30. Bragg EJ, Warshaw GA. Evolution of geriatric medicine fellowship training inthe United States. 2003;11:280-290.
Online J IssuesNurs.
31. Mion LC. Care provision for older adults: who will provide? 2003;8:e4.
32. Fick DM, Cooper JW, Wade WE, Waller JL, Maclean JR, Beers MH. Updatingthe Beers criteria for potentially inappropriate medication use in older adults. 2003;163:2716-2724.