Measuring Patient Safety in Ambulatory Care: Potential for Identifying Medical Group Drug"Drug Interaction Rates Using Claims Data

Objective: To evaluate the feasibility of using health-planadministrative data to measure potential drugâ€“drug interaction(DDI) rates in the ambulatory setting at the medical-group leveland to assess the potential use of DDI rates in performance measurement,quality improvement, and research in patient safety.

Study Design: We combined administrative and pharmacyclaims data from 2 large health plans to calculate the rates atwhich member users of selected chronic medications were potentiallyexposed to a second drug known to pose a risk of harmfulinteractions.

Methods: We divided 44 medication combinations with risk ofadverse interactions into those with DDIs of moderate/severe clinicalsignificance and those with DDIs of mild significance. We thencalculated yearly rates of potential DDIs in continuously enrolledmembers aged 19 and older from 1998 through 2001. Rates werecalculated for all members, overall base-medication users, andindividual medical groups responsible for their care.

Results: The analytic data set included 756 047 patient-years ofdata and 110 to 123 medical groups per year. During the 4-yearinterval, one or more unique potential DDIs occurred in 6.2% to6.7% of base-drug users and 2.0% to 2.3% of all adult health-planmembers per year. Medical-group mean user rates were slightlylower (5.33%â€“5.81%), with wide variance (SD = 2.6%â€“3.1%) andhigh stability over time.

Conclusion: Potential DDI rates calculated from health-plandata have promise for measurement in patient medication safety.This readily available and inexpensive evaluation tool has potentialfor monitoring, improvement, and research purposes if furtherstudies validate their relationship to actual adverse events.

(Am J Manag Care. 2004;10:753-759)

Release of the 1999 Institute of Medicine (IOM)report on medical errors created a public furorand stimulated many research and improvementinitiatives.¹ Although the IOM report recommended a50% reduction in errors over 5 years, it did not discussmeasurement of errors or adverse event rates as ameans of monitoring trends and improvement efforts.

The studies cited in the IOM report calculated errorswith methods too expensive and labor intensive forrepeated use in performance monitoring or improvementefforts. Moreover, the studies dealt with hospitalincidents only, with uncertain applicability to ambulatorycare. Thus, we know very little about the nature orepidemiology of errors or adverse events in ambulatorycare settings, and we lack measurement methods foraccountability or improvement.

In the absence of any practical methods of quantifyingoverall errors, measurement of adverse drug events(ADEs) may be a good way to gauge errors and patientsafety in ambulatory care, since studies have found ADEsto be frequent and a significant contributor to emergencydepartment visits and hospitalizations.^2-12 Using multiplemethods in 2003, Gurwitz found an incidence of 50.1ADEs per 1000 person-years in those aged 65; 27.6% weredeemed preventable.¹³ Gandhi concluded from a patientsurvey that during a 3-month period, 25% of outpatientswith a prescription experienced an ADE, 13% of whichwere serious and 11% preventable.¹⁴ Both studies arevaluable in understanding the approximate rates ofADEs but neither is amenable to repeated measures atthe level of individual practices or medical groups.

Although ADEs appear to be relatively frequent,there are practical limitations in relying on chart auditsand surveys to detect and monitor them. Drug claimdata and member information available from healthplans offer an inexpensive and repeatable way to measurepotential ADEs, especially drug—drug interactions(DDIs), in ambulatory care. A DDI is "2 or more drugsinteracting in such a manner that the effectiveness ortoxicity of 1 or more drugs is altered."¹⁵ A potential DDIshould be identifiable from health-plan claims data thatshow 2 such drugs prescribed simultaneously.Researchers have studied DDI rates in special populationsand settings,^16-24 but only 2 recent studies haveused large population-based pharmacy claims databasesto study case rates of specific drug pairs. Peng et al useda drug utilization review program to study potentialDDIs in a population of 3 million, and Hennessy et al didthe same for enrollees in 6 state Medicaid programs.^25,26However, no one has studied DDI rates in individualmedical groups, the level at which accountability andimprovement ability reside. Such analysis can facilitatemeasurement and comparison of performance andprovide medical groups with a basis for qualityimprovement.

Therefore, we combined administrative data from 2large health plans to answer the following questions:

Is it feasible to use health-plan data to identifypotential DDI rates in relation to the individualmedical groups responsible for member care?
Is the frequency of potential DDIs high enough tocharacterize medical groups or clinics?
How much variation in rates exists among medicalgroups?
Are these rates stable enough over time to serve asa credible way to measure trends and changes?

METHODS

We studied the frequency of potential DDIs from1998 through 2001 in health-plan members served bythe medical groups contracting with 2 HMOs. The primarydata sources were pharmacy claims, medical careclaims, and membership files. The institutional reviewboards at both organizations approved the study.

The HMOs participating in this study were HealthPartnersin Minnesota and Lovelace Health Plan in NewMexico. In 2000, HealthPartners covered 657 000 membersthrough contracts with 50 medical groups, andLovelace Health Plan had 240 000 members cared for byapproximately 240 medical groups.

The study included all medical groups that providedprimary care for at least 30 health-plan members from1998 through 2001. Location, based on US Bureau ofthe Census designation, and other descriptive dataabout these medical groups were collected from health-planrecords.

We identified HMO members who were 19 or older asof January 1 in the study year, had continuous membership(with up to a 45-day break) with pharmacy benefitsduring the study year, and had a known affiliationwith an eligible medical group. Descriptive data aboutmembers were obtained from membership and claimsrecords.

We included in our analysis 44 combinations of drugs(a base drug and a conflicting drug or drugs) known tohave potential adverse interactions and in which thebase drug is taken chronically by large numbers of people(Table 1). We used 3 key references to identify druginteraction combinations: (1) Hansten and Horn's DrugInteractions, Analysis and Management, (2) the DRUGREAX®System of MICROMEDEX, and (3) Evaluationsof Drug Interactions™ (EDI).^27-29 Because there is somuch variation among standard texts about what combinationsrepresent a DDI of what severity, we requiredthat each DDI combination be identified as a problem inat least 2 of these 3 references and have a clinical significancerating of 1, 2, or 3 (on a scale of 1 to 5), asassigned by Hansten and Horn.^27,30,31

Severe interaction ("Avoid administration of thecombination. The risk of adverse patient outcomeprecludes concomitant administration.")
Moderate interaction ("Avoid administrationunless it is determined that the benefit of coadministrationoutweighs the risk to the patient.Patients should be monitored closely.")
Mild interaction ("Minimize risk by consideringalternative agents or change dosage or route ofadministration. Patient monitoring is suggested.")

An initial list of DDI combinations was reviewed by apanel of clinical pharmacists, by several primary carephysicians, and by the Pharmacy & TherapeuticsCommittee of one of the two participating plans.Reviewers deleted from the list DDI combinationsdeemed irrelevant to current clinical practice or ofquestionable risk. Deleted combinations includefurosemide and steroids, angiotensin-convertingenzyme (ACE) inhibitors and potassium-sparingdiuretics. For analysis, the combinations were dividedinto potential DDIs of moderate/severe clinical significance(Hansten and Horn ratings 1 and 2) and potentialDDIs of mild significance (Hansten and Horn rating 3).

Individual drug fills were identified from pharmacyclaims data using the National Drug Codes (NDC). Acomprehensive list of NDCs for the study medicationswas gathered. The Master Drug Data Base (MDDB®), acommercial database maintained by Medi-Span,Indianapolis, Ind,³² was used to ensure the inclusion ofall NDCs ever registered with the Food and DrugAdministration.

To determine potential DDIs in this patient population,we identified patients with both a pharmacy fill fora base drug in the year of interest and a fill for a conflictingdrug. The days' supply of the conflicting drugmust have overlapped with that of the base drug by 10or more days. Coprescription of a given DDI combinationwas counted only once per year for a given patient,regardless of the number of fills for either drug.However, if the same patient had overlapping fills for adifferent DDI combination during the same year, thiswas counted as a separate potential DDI.

Each member was linked with themedical group responsible for their careas of January 1 in the year of interest.After determining that there was no significantdifference in rates between the2 health plans, data were combined for asingle analysis. Medical group-specificrates were then calculated for only thosemedical groups with at least 30 chronicusers of measured base drugs or classes.Stability in rates over the 4 study yearswas evaluated by calculating Pearsoncorrelation coefficients between individualmedical group rates for consecutiveyears. Potential DDI error rates werecalculated for each year from 1998through 2001 among (1) base-drug usersand (2) all HMO members who could beconnected with a medical group in theanalysis. A Cochran-Armitage trend testanalysis of rates across the 4 years wasalso performed.

RESULTS

The analytic data set included 756047 patient-years of data, with 199 373in 1998, 181 955 in 1999, 189 955 in2000, and 194 084 in 2001. The numberof medical groups eligible each yearfrom 1998 through 2001 was 109, 107,123, and 110, respectively. Table 2 listscharacteristics of the medical groups in2001 only, since there was relatively littledifference from one year to the next.Medical groups varied greatly in numberof clinics, number of clinicians, andnumber of HMO members. Users of thebase drugs were predominantly femaleand middle-aged.

During this period, there was a significantincrease ( <.001) in the percentageof the HMO population usingany of the base drugs (33.04% in 1998 to34.80% in 2001). In addition, the meannumber of medications used by patientsincreased significantly ( <.001) amongpatients using base drugs of moderate/severe significance (8.14 in 1998 to8.58 in 2001) and among patients usingbase drugs of mild significance (7.07 in1998 to 7.74 in 2001).

Table 3 shows the percentage of HMO members whohad at least 1 potential DDI from 1998 through 2001.Percentages are given for all HMO members and forchronic users of base drugs. In each case (except for thepercentage of base-drug users with risk of moderate orsevere reactions) the rates increase gradually over the4-year time interval ( < .001). Base drugs with themost users were selective serotonin reuptake inhibitors,oral contraceptives, 3-hydroxy-3-methylglutaryl coenzymeA (HMG-CoA) reductase inhibitors, ACEinhibitors, and steroids. The rate of individual potentialDDIs varied from 0% for phenytoin plus phenylbutazone,digoxin plus cholestyramine, and propranolol plusrifampin to 16.5% for ACE inhibitors plus nonsteroidalanti-inflammatory drugs/cyclooxygenase-2 inhibitor.

Table 4 shows the potential DDI rates of medicalgroups for their HMO panel members and for their usersof the base drugs. It illustratesthe variation in DDI rates amongmedical groups in any year (eg,0% to 17%). At the same time, itshows how little DDI rateschange from one year to thenext, since Pearson correlationcoefficients for consecutiveyears are rather high. The 10 to13 largest medical groups (servingat least 5000 HMO members)had potential DDI rates 5% to10% higher than those of thesmaller groups (data not shown).The rates for the larger groupswere also much more stable overtime, probably as a result of largersample sizes.

Table 5 provides medical-grouprates and variances forusers of base drugs in variousdrug classes by body system.The data show that, when limitedto those medical groups withat least 30 base-drug users in adrug class, the numbers dropsubstantially. Only half of theclasses of base drugs(antilipemics, cardiovascular,hormonal, and psychoactive)have enough medical groupswith an adequate number ofusers to permit inclusion in suchan analysis, and 2 classes haveespecially high mean rates (cardiovascularand hematologic).

DISCUSSION

These results confirm that it is feasible to calculaterates of potential DDIs from HMO administrative dataregarding individual medical groups. The rates are highenough (6.2%-6.7% of base-drug users on average) andstable enough over time to permit their use for accountabilityand improvement. Even if restricted to moderateor severe potential DDIs, 3.5% of base-drug users are atrisk in any given year. At the same time, there is quitea bit of variation in rates among the medical groups providingcare for these members (SD = 2.6-3.0), as well asamong the individual drugs and drug classes. However,shifting focus from overall potential DDIs to specificclasses of drugs loses 35% to 54% of medical groups for4 classes, and the remaining 4 classes lose much more.It would be difficult to apply these rates to individualclinic sites or physicians.

The proportion of actual adverse reactions from these"potential" rates must be studied because these ratesappear to be good candidates for measures of ambulatory-care medication safety, at least for medium to largemedical groups. They seem to provide a useful way tocompare the quality of care among medical groups andmay provide a means for those groups to target and monitorquality improvement actions. Because determinationof these rates requires sufficiently large health-planmembership, however, they cannot be used to compareindividual clinicians or small medical groups.

While this is the first use of health-plan data for suchmeasurement at the medical group level, rates for entirepopulations (eg, a city or state) have been calculated withother types of computerized databases. Using Medi-Calclaims, Laventurier found an overall potential-DDI rate of7.57% for 6 base-drug groups.²⁰ Jinks used Medicaidrecords to find that 11.3% of users of 3 drugs had potentialDDIs.²¹ Three recent studies using large pharmacypopulation databases in Scandinavia found that 1.4% to3% of all prescriptions presented a risk of a major DDI.Overall potential DDI rates were 13.6% and 31% in 2 ofthose studies.^18,23-24 Peng performed retrospective drug-utilizationreview of pharmacy claims for 3 million peopleand found that 245 000 potential DDI cases (0.8% ofclaims) could be identified using coprescription of any of51 DDI pairs.²⁵ More sophisticated automated screens(like the ones we used) reduced Peng's rates by 70%, andpharmacist review of this group reduced the number byan additional 80% to 6.3 per 1000 members.

In our study, both the mean number of medicationsused per patient and the proportion of base-drug userswith potential DDIs increased between 1998 and 2001.This is consistent with the recent findings of Hennessyet al, who found no reduction in the rate of potentialprescribing errors or hospitalization in their retrospectivedrug utilization review with alerts for exceptionssent to prescribing physicians.²⁶ It is well established inthe quality improvement field that, with few exceptions,change in clinician behavior requires local systemchange rather than exhortation, and that is the type ofchange that broader performance measurement andfeedback could provide.

One of the reasons that individual case alerts areineffective is that the coprescription of 2 drugs thatinteract with potentially adverse consequences is notnecessarily erroneous. Physicians must weigh both thebenefit and the risk of drug combinations when decidingwhat to prescribe. Additionally, many drug interactionscan be managed through appropriate laboratoryand symptom monitoring or by instructing the patientto adjust the timing or dosage of a drug. Only a fractionof patients with a potential DDI actually suffer adverseconsequences. Mitchell et al found that, of 160 ambulatorypatients with potential for a DDI, only 8 of thepotential DDIs were of major clinical significance.¹⁹Herr and colleagues found that 7.5% of 201 emergencydepartment (ED) patients at risk for a DDI, due to eitherED medication or medication taken prior to the EDvisit, had clinically relevant adverse interactions within4 weeks after the ED visit.²² These researchers alsofound that medications prescribed during the ED visitput 26% of ED patients at risk for a DDI, and 3% of thisgroup experienced a clinically relevant adverse interactionwithin 4 weeks. These rates are higher than thosein hospital patient studies reviewed by Jankel, butunfortunately there are no published studies of actualadverse events for large populations identified throughpharmacy records or claims data.³³

This study has several limitations. In order to get arate high enough to characterize individual medicalgroups, we were forced to combine mild with moderate—severe DDIs. Consequently, we included many DDIsthat could be managed by close monitoring or bychanges in dosage or route of administration. This complicationhighlights the need to focus on comparison ofDDI rates among medical groups rather than eliminationof each potential DDI.

Identifying prescription drug use from computerizedclaims records provides only an indirect picture of actualpatient drug intake. Some drugs may have been purchasedbut not taken by the patient, or the patient maynot have taken both drugs simultaneously. Also, we wereunable to ascertain occurrence of actual symptoms oradverse events as a result of taking 2 interacting drugs.Neither could we determine whether the prescribingphysician performed appropriate laboratory monitoringor instructed the patient to change the dosage or timingof a medication to avoid an adverse interaction. Thus,these rates represent only a theoretical maximum risk ofinteractions. Before these rates can be used to monitorperformance, additional research must ascertain howoften adverse outcomes actually occur in patients withpotential DDIs and whether appropriate monitoring orpatient instruction is being provided. It is also importantto learn whether there is a relatively constant ratiobetween potential DDIs and actual adverse events.

To Err is Human

Once this clarifying information is obtained, potentialDDI rates determined from claims data could be animportant tool for improving patient safety. With theirrates of potential DDIs identified, medical groups canbegin to work on improvement. On a larger scale,reduction of these rates may help us to know whetherwe have made progress toward the 50% reduction inmedical errors proposed in .¹

Acknowledgments

This project was supported by Task Order #290-00-0015-04 from theAgency for Healthcare Research and Quality, Rockville, Md. The studyteam is grateful to Sally Beaton, PhD, and Michael Shainline, MS, MBA, ofthe Lovelace Clinic Foundation for assistance with collection of medicalgroup data and to Kelly Fillbrandt, BS, Karen Engebretson, BA, KristenPokela, BS, and Hans Petersen, MS, for assembling and programming data.

From HealthPartners Research Foundation, Minneapolis, Minn (LIS, WWN, ALC);Lovelace Respiratory Research Institute, Albuquerque, NM (JSH, MHR, FJF); and LovelaceClinic Foundation, Albuquerque, NM (MJG, LRY).

Address correspondence to: Leif I. Solberg, MD, HealthPartners Research Foundation,PO Box 1524, MS 21111R, Minneapolis, MN 55440-1524. E-mail: leif.i.solberg@healthpartners.com.

To Err isHuman: Building a Safer Health System.

1. Kohn LT, Corrigan J, Donaldson MS for the Institute of Medicine. Washington, DC: National AcademyPress; 2000.

JGen Intern Med.

2. Gandhi TK, Burstin HR, Cook EF, et al. Drug complications in outpatients. 2000;5:149-154.

Am J Med.

3. Chyka PA. How many deaths occur annually from adverse drug reactions in theUnited States? 2000;109:122-130.

Arch Intern Med.

4. Bedell SE, Jabbour S, Goldberg R, et al. Discrepancies in the use of medications:their extent and predictors in an outpatient practice. 2000;160:2129-2134.

Am J Emerg Med.

5. Goldberg RM, Mabee J, Chan L, Wong S. Drug-drug and drug-disease interactionsin the ED: analysis of a high-risk population. 1996;14(5):447-450.

Arch Intern Med.

6. Miwa LJ, Jones JK, Pathiyal A, Hatoum H. Value of epidemiologic studies indetermining the true incidence of adverse events: the nonsteroidal anti-inflammatorydrug story. 1997;157:2129-2136.

JGen Intern Med.

7. Petty BG. Identifying and reducing complications of outpatient medications. 2000;15:207-208.

Ann Pharmacother.

8. Tafreshi MJ, Melby MJ, Kaback KR, Nord TC. Medication-related visits to theemergency department: a prospective study. 1999;33:1252-1257.

J Res Pharm Econ.

9. Sullivan SD, Kreling DH, Hazlet TK. Noncompliance with medication regimensand subsequent hospitalizations: a literature analysis and cost of hospitalizationestimate. 1990;2:19-33.

Med J Aust.

10. Roughead EE, Gilbert AL, Primrose JG, Sansom LN. Drug-related hospitaladmissions: a review of Australian studies published 1988-1996. 1998;168:405-408.

BMJ.

11. Pouyanne P, Haramburu F, Imbs JL, Begaud B. Admissions to hospital causedby adverse drug reactions: cross sectional incidence study. 2000;320:1036.

12. Beard K. Adverse reactions as a cause of hospital admission in the aged.Drugs Aging. 1992;2:356-367.

JAMA.

13. Gurwitz JH, Field TS, Harrold LR, et al. Incidence and preventability ofadverse drug events among older persons in the ambulatory setting. 2003;289:1107-1116.

N Engl J Med.

14. Gandhi TK, Weingart SN, Borus J, et al. Adverse drug events in ambulatorycare. 2003;348:1556-1564.

Handbook of Drug Interactions.

15. Hartshorn EA. Hamilton, IL: DrugIntelligence Publications; 1973.

J Fam Pract.

16. Davidson KW, Kahn A, Price RD. Reduction of adverse drug reactions bycomputerized drug interaction screening. 1987;25:371-375.

Hosp Pharm.

17. Stanaszek WF, Franklin CE. Survey of potential drug interaction incidence inan outpatient clinic population. 1978;13:255-257, 261, 263.

Dan Med Bull.

18. Rosholm JU, Bjerrum L, Hallas J, Worm J, Gram LF. Polypharmacy and therisk of drug-drug interactions among Danish elderly. A prescription databasestudy. 1998;45:210-213.

Am J Hosp Pharm.

19. Mitchell GW, Stanaszek WF, Nichols NB. Documenting drug-drug interactionsin ambulatory patients. 1979;36:653-657.

Cal Pharm.

20. Laventurier MF, Talley RB. Incidence of drug-drug interactions in a Medi-Calpopulation. 1972;20:18-22.

Am J Hosp Pharm.

21. Jinks MJ, Hansten PD, Hirschman JL. Drug interaction exposures in an ambulatoryMedicaid population. 1979;36:923-927.

Ann Emerg Med.

22. Herr RD, Caravati EM, Tyler LS, Iorg E, Linscott MS. Prospective evaluationof adverse drug interactions in the emergency department. 1992;21:1331-1336.

Pharm World Sci.

23. Bergendal L, Friberg A, Schaffrath A. Potential drug-drug interactions in 5125mostly elderly out-patients in Gothenburg, Sweden. 1995;17:152-157.

BMJ.

24. Merlo J, Liedholm H, Lindblad U, et al. Prescriptions with potential druginteractions dispensed at Swedish pharmacies in January 1999: cross sectionalstudy. 2001;323:428-429.

J Manag Care Pharm.

25. Peng CC, Glassman PA, Marks IR, et al. Retrospective drug utilization review:incidence of clinically relevant potential drug-drug interactions in a large ambulatorypopulation. 2003;9:513-522.

JAMA.

26. Hennessy S, Bilker WB, Zhou L, et al. Retrospective drug utilization review,prescribing errors, and clinical outcomes. 2003;290:1494-1499.

Hansten and Hornâ€™s Drug Interactions, Analysis andManagement.

27. Hansten PD, Horn JR. St. Louis: Facts and Comparisons; 2003.

28. Klasco RK, ed. DRUG-REAX®System. Greenwood Village, Colo: ThomsonMicromedex; edition expires 12/2003.

Evaluations of Drug Interactions® (EDI)

29. Shinn AF, Shrewsbury RP, eds. . San Bruno, CA: First Databank, 2002.

Curr Ther ResClin Exp.

30. Fulda TR, Valuck RJ, Zanden JV, Parker S, Byrns PJ. Disagreement amongdrug compendia on inclusion and ratings of drug-drug interactions. 2000;61:540-548.

CurrTher Res Clin Exp.

31. Valuck RJ, Byrns PJ, Fulda TR, Zanden JV, Parker S. Methodology for assessingdrug-drug interaction evidence in the peer-reviewed medical literature. 2000;61:553-568.

32. Master Drug Data Base (MDDB®). Vol. 2002. Zionville, Ind: Facts andComparisons/Medi-Span; 2002.

DICP Ann Pharmacother.

33. Jankel CA, Speedie SM. Detecting drug interactions: a review of the literature. 1990;24:982-989.