Characteristics Associated With Postdischarge Medication Errors

ORIGINAL ARTICLE

For editorialcomment,see page 1027;for a related articlsee page 1116

From the DepartmentVeterans Affairs, TenneValley Healthcare SysteGeriatric Research Educand Clinical Center (A.SJ.F.S., T.S.), and SectionHospital Medicine, DivisGeneral Internal MedicinPublic Health, DepartmMedicine (A.S.M., S.K.),ter for Health ServicesResearch (A.S.M., C.L.L.J.M.L.J., C.C., K.M.G., T.S

Affiliations continthe end of this

1042

Characteristics Associated With PostdischargeMedication Errors

e,

ofsseemation.M.,ofion ofe andent ofCen-

,., S.K.),

ued atarticle.

Amanda S. Mixon, MD, MS, MSPH; Amy P. Myers, PharmD; Cardella L. Leak, BA;J. Mary Lou Jacobsen, BA; Courtney Cawthon, MPH; Kathryn M. Goggins, MPH;

Samuel Nwosu, MS; Jonathan S. Schildcrout, PhD; John F. Schnelle, PhD;Theodore Speroff, PhD; and Sunil Kripalani, MD, MSc

Abstract

Objective: To examine the association of patient- and medication-related factors with postdischargemedication errors.Patients and Methods: The Vanderbilt Inpatient Cohort Study includes adults hospitalized with acutecoronary syndromes and/or acute decompensated heart failure. We measured health literacy, subjectivenumeracy, marital status, cognition, social support, educational attainment, income, depression, globalhealth status, and medication adherence in patients enrolled from October 1, 2011, through August 31,2012. We used binomial logistic regression to determine predictors of discordance between the dischargemedication list and the patient-reported list during postdischarge medication review.Results: Among 471 patients (mean age, 59 years), the mean total number of medications reported was12, and 79 patients (16.8%) had inadequate or marginal health literacy. A total of 242 patients (51.4%)were taking 1 or more discordant medication (ie, appeared on either the discharge list or patient-reportedlist but not both), 129 (27.4%) failed to report a medication on their discharge list, and 168 (35.7%)reported a medication not on their discharge list. In addition, 279 participants (59.2%) had a misun-derstanding in indication, dose, or frequency in a cardiac medication. In multivariable analyses, highersubjective numeracy (odds ratio [OR], 0.81; 95% CI, 0.67-0.98) was associated with lower odds of havingdiscordant medications. For cardiac medications, participants with higher health literacy (OR, 0.84; 95%CI, 0.74-0.95), with higher subjective numeracy (OR, 0.77; 95% CI, 0.63-0.95), and who were female(OR, 0.60; 95% CI, 0.46-0.78) had lower odds of misunderstandings in indication, dose, or frequency.Conclusion: Medication errors are present in approximately half of patients after hospital discharge andare more common among patients with lower numeracy or health literacy.

ª 2014 Mayo Foundation for Medical Education and Research n Mayo Clin Proc. 2014;89(8):1042-1051

F requently, the discharge process is rushedand disjointed, despite the critical impor-tance of communicating with patients

about postdischarge medications. Health careprofessionals may not effectively counsel pa-tients regarding medications on the dischargeinstructions.1 Likewise, patients may have diffi-culties understanding the changes to their medi-cation regimen because of limitations in healthliteracy, numeracy, and other patient factors.2-4

Postdischarge medication errors are common,2

but the patient-related factors associated withsuch errors are not well understood.

Health literacy, the ability to understand andact onmedical information,5 and numeracy, “theability to use and understand numbers in dailylife,”6 have been associated with medication

Mayo Clin Proc. n August 2014;89www.mayoclinicproceedings.org n

understanding and adherence.7-9 In addition,other patient factors, such as cognitive impair-ment,10 poor social support,11 and depression,12

have been associated with postdischarge out-comes, such as unscheduled health care use oradverse events in patients with cardiovasculardisease. However, the independent associationof these factors with postdischarge medicationerrors has not been examined in this population.

Postdischarge medication errors are impor-tant because they significantly contribute toadverse drug events (ADEs) or harm due tomedications.13-15 Medication errors includeomissions, commissions, and misunderstand-ings in indication, dose, or frequency.10-13

Errors can be due to differences between med-ications the patient thinks he/she should be

(8):1042-1051 n http://dx.doi.org/10.1016/j.mayocp.2014.04.023ª 2014 Mayo Foundation for Medical Education and Research

http://dx.doi.org/10.1016/j.mayocp.2014.04.023

http://www.mayoclinicproceedings.org

MEDICATION ERRORS AFTER HOSPITAL DISCHARGE

taking and what is prescribed, often due to poorphysician-patient communication or patient-related factors, as mentioned above.16,17

Prior studies have found that 30% to 70%of patients have medication errors between thedischarge list and the patient-reported regimenafter discharge,17-21 although few studies havefocused on patients with cardiovascular dis-ease. Because multiple types of medicationsare prescribed and cardiac medications cancause serious harm, patients with cardiovascu-lar disease are at higher risk for errors andADEs after discharge.13,15,17,22-24

This article describes predictors of medica-tion errors among patients recently hospitalizedfor cardiovascular disease. On the basis of ourconceptual model of factors associated withpostdischarge outcomes,25 we hypothesizedthat low health literacy and numeracy, moremedications on discharge, more changes tomedications during hospitalization, impairedcognition, poor social support, low preadmis-sion medication adherence, and depressionwould be associated with postdischarge medi-cation errors.

METHODS

Study Setting and DesignTheVanderbilt InpatientCohort Study (VICS) isa prospective study of patients admitted withcardiovascular disease to Vanderbilt UniversityHospital. The purpose of VICS is to investigatethe effect of patient and social factors on postdi-scharge health outcomes, such as medicationsafety, quality of life, unplanned hospital utiliza-tion, and mortality. The rationale and design ofVICS are detailed elsewhere.25 Briefly, partici-pants completed a baseline interview while hos-pitalized, and follow-up telephone calls wereconducted at approximately 2 to 3, 30, and 90days after discharge. We conducted an interimanalysis of patient- and medication-related fac-tors associated with medication errors after hos-pital discharge. The study was approved bythe Vanderbilt University Institutional ReviewBoard.

PatientsEligibility screening shortly after admissionidentified patients with an intermediate or highlikelihood of acute coronary syndrome (ACS)or acute decompensated heart failure (ADHF)

Mayo Clin Proc. n August 2014;89(8):1042-1051 n http://dx.doi.orgwww.mayoclinicproceedings.org

per a physician’s review of the clinical record.Exclusion criteria included age younger than18 years, inability to communicate in English,unstable psychiatric illness, delirium, low likeli-hood of follow-up after discharge, receiving hos-pice care, or otherwise too ill. To be included inthis analysis, patients must have completed themedication review portion of the follow-upinterview.

Baseline AssessmentConsenting patients completed an interviewer-administered baseline assessment of demo-graphic information, including self-reportedrace, ethnicity, educational attainment, andmarital status. Household income was collectedusing the strata from the Behavioral Risk FactorSurveillance System questionnaire.26

Social support was assessed using the 7-item Enhancing Recovery in Coronary HeartDisease Social Support Inventory.27,28 Patientswere asked the questions regarding emotionaland instrumental support. Scores range from 8to 34, with higher scores indicating more so-cial support.

Patients completed the short form of theTest of Functional Health Literacy in Adults,29

a timed test administered in a maximum of 7minutes. Scores may be categorized as inade-quate (score range, 0-16), marginal (scorerange, 17-22), or adequate (score range, 23-36).

We used a 3-item version of the SubjectiveNumeracy Scale (SNS), which quantifies thepatients’ perceived quantitative abilities andpreferences for numerical information.30 The3-item SNS has a correlation coefficient of0.88 with the full-length (8-item) SNS. The in-ternal consistency reliability of the 3-item SNSwas high (Cronbach a¼0.78).30,31 The SNS isreported as the mean on a scale from 1 to 6,with higher scores reflecting better numeracy.

We assessed cognition using the ShortPortable Mental Status Questionnaire, a 10-item measure,32 which is adjusted for educa-tional attainment. Higher scores reflect worsecognitive status and may be categorized asnot impaired (0-2 errors) or impaired (3-10errors).

Self-rated health status was assessed using 5of 10 items from the National Institutes of HealthPatient Reported Outcomes Measurement Infor-mation System global health status question-naire.33 These questions ascertain overall health,

/10.1016/j.mayocp.2014.04.023 1043



MAYO CLINIC PROCEEDINGS

1044

quality of life, physical and mental health, andsatisfactionwith social activities and relationshipson a 5-point Likert-type scale. Scores are reportedas a mean of the 5 items (range, 1-5).

We assessed depression during the 2weeks before the hospitalization using the Pa-tient Health Questionnaire 8.34 Scores arecategorized as none/minimal depression (scorerange, 0-4), mild (score range, 5-9), moderate(score range, 10-14), moderately severe (scorerange, 15-19), and severe (score range, 20-24).

Adherence to the preadmission medicationregimen was measured using a 7-item versionof the Adherence to Refills and MedicationsScale (ARMS) (personal communication, SunilKripalani, MD, MSc, and Kenneth A Wallston,PhD, June 13, 2014). The 7-item ARMS has ahigh internal consistency reliability (Cronbacha¼0.81) and a correlation coefficient of 0.95with the full-length (12-item) ARMS.35 Re-sponses on the 4-point scale are summed for ascore with a possible range of 7 to 28, with lowerscores indicating better adherence. All diagnosesat discharge were retrieved from the medical re-cord and used to compute an Elixhauser score toreflect comorbid conditions.36

Medication-Related DataNet changes to the patient’s medication regimenduring hospitalization were tallied bycomparing the preadmission medication listand discharge medication list from the elec-tronic medical record. We did not count equip-ment or nondrug items (eg, test strips), but weincluded over-the-counter medications. Cardiacmedications included the following classes: anti-anginal, antiplatelet, anticoagulant, antihyper-tensive, cholesterol, diabetes mellitus, anddiuretics. Medications were counted as changedif a dose, frequency, route, or formulationdiffered or if use of a medication was either dis-continued or initiated during hospitalization.We summed the number of all medicationchanges and, separately, the number of changes

TABLE 1. Classification of Medication Errors Identified D

Variable Medication 1

Patient report . ClopDischarge list Lisinopril, 10 mg/d ClopOutcome typea Discordant (omission) Con

aThere were a total of 2 discordant medications (ie, 2 errors).

Mayo Clin Proc. n August 2014;89

to cardiac medications. If dose or frequency in-formation was missing in either the preadmis-sion or discharge medication list, we assumedthe medication was not changed.

Outcome MeasuresFor analyses, the discharge medication list wasorganized into cardiac and noncardiac medica-tions. The total number of medications wasthe sum of the medications on the dischargelist plus additional patient-reported medica-tions. Additional medications were frequentlyprescribed before hospitalization but notmentioned in the discharge list.

We contacted patients by telephone 2 to 3days (range, 1-7) after discharge to assess allprescription and over-the-counter medicationsthe patient reported taking. We compared whatthe patient was taking with the discharge medi-cation list, whichwas given to them at dischargeby the bedside nurse. If patients did not report amedication that appeared on the discharge listor reported a medication not on the dischargelist, it was flagged as a potentially discordantmedication and probed further. If the patientdid not mention the medication initially butcorrectly reported it when prompted or re-ported that use of the medication was stoppedor started by a clinician, no error was recorded.Otherwise, it was judged to be discordant andclassified as an omission or commission.

Table 1 andTable 2 indicate howmedicationerrors were counted. In Table 1, a medicationwas coded as an omission if there was no expla-nation for a medication that the patient did notreport but appeared on the discharge list. Omis-sions included instances inwhich the patient didnot fill or refill the prescription, stopped use ofthe medication without a health care profes-sional’s order, or reported that she/he was notaware of the medication. A commission wascoded if there was no explanation for a patient-reported medication not on the discharge list,unless a health care professional had instructed

uring the Postdischarge Telephone Interview

Medication 2 Medication 3

idogrel, 75 mg/d Simvastatin, 40 mg at bedtimeidogrel, 75 mg/d .

cordant Discordant (commission)

(8):1042-1051 n http://dx.doi.org/10.1016/j.mayocp.2014.04.023www.mayoclinicproceedings.org



TABLE 2. Example of How Cardiac Medication Misunderstandings WereCounteda

Drug Indication Dose (mg) Frequency

LisinoprilPatient report “For blood pressure” 10 Twice a dayCorrect response Antihypertensive 10 Once a day

Discrepancy count 0 0 1 Frequencydiscrepancy

ClopidogrelPatient report “For my stomach” 75 Twice a dayCorrect response Antiplateletb 75 Once a day

Discrepancy count 1 Indicationdiscrepancy

0 1 Frequencydiscrepancy

SimvastatinPatient report “Keep stent open” 80 Twice a dayCorrect response Lipid loweringc 40 At bedtime

Discrepancy count 1 Indicationdiscrepancy

1 Dose discrepancy 1 Frequencydiscrepancy

aThere were a total of 1 discordant medication (ie, 1 error) for lisinopril, 2 discrepancies (ie, 2errors) for clopidogrel, and 3 discrepancies (ie, 3 errors) for simvastatin.bExamples of other accepted indication responses for antiplatelet medications are “helps bloodflow,” “for clots,” “blood thinner,” and “for circulation.”cExamples of other accepted indication responses for lipid-lowering medications are “highcholesterol,” “fat in blood,” “high fat,” and “high low-density lipoprotein.”


the patient to take themedicationafter discharge.A commission was also coded if the patient re-ported having taken a medication before hospi-talization and continuing to take it afterdischarge, but it was not on the discharge list.

In addition, we randomly selected onemedication per cardiac type (eg, antihyperten-sives) for more testing, asking the patient toprovide the medication’s indication, dose, andfrequency. Any difference in dose or frequencybetween the discharge list and the patient’sreport was considered a misunderstanding, un-less the patient reported that a health care pro-fessional changed the dose or frequency afterdischarge. Patient responses for indicationswere judged against a physician-created list ofacceptable responses, including common off-label indications and lay terms. An example ofhow cardiac medication misunderstandingswere counted is given in Table 2. A patientcould have more than one outcome (eg, anomission, a commission, and a misunder-standing in indication, dose, or frequency).

Statistical AnalysesWe sought to examine the association betweenpatient- and medication-related factors and thenumber of discordant medications between thedischarge and patient lists, omissions, commis-sions, and misunderstandings in indication,dose, or frequency. For each outcome, a binomiallogistic regression model was built to examinepatient- andmedication-related factors (the inde-pendent variables) associated with the odds ofmedication errors. The independent variablesare consistent with our conceptual framework re-ported elsewhere,25 checked for colinearity, andincluded in all models: age, sex, race (AfricanAmerican, white, or other), marital status, pri-mary diagnosis, educational attainment, income,health literacy, subjective numeracy, cognition,global health status, depression, number of med-ications changed during hospitalization, socialsupport, medication adherence, and Elixhausercomorbidity score. Because each outcome was acount with a different number of possible errors(ie, different denominators) across patients, thelog of the number of possible errors was usedas an offset. Thus, exponentiated parameter esti-mates are odds ratios (ORs) and have the sameinterpretation as those from standard logisticregressionmodels. Because of concerns regardingoverdispersion, robust or sandwich-based SEs


were used to characterize estimator uncertainty.Multiple imputation was used in the multivari-able analyses to avoid deleting any participantswho had missing data for any of the covariates.All scale scores were treated as continuous vari-ables formultivariablemodeling. Finally, becausethe continuous variables operate on heteroge-neous scales, we chose to quantify continuouscovariate effects using ORs associated with inter-quartile range changes in the covariates. Datawere analyzed in R software.37

RESULTSOf the 680 eligible, we enrolled 587 patientsfrom October 1, 2011, through August 31,2012. Ninety-three eligible patients (13.7%)declined enrollment (Figure 1). Of the 587enrolled, 471 patients (80.2%) completed themedication review portion of the postdischargetelephone interview and were included in theseanalyses. The remaining 116 patients did notcomplete the follow-up call (n¼51), did notcomplete the medication review (n¼49), with-drew (n¼11), or died before discharge (n¼5).

Table 3 lists the patients’ baseline character-istics. The mean age was 59 years (SD, 12years), 226 (47.9%) were female, and 90

/10.1016/j.mayocp.2014.04.023 1045



1364 Examined for eligibility

680 Eligible

587 Enrolled

471 Complete follow-up

7637 Potentially eligible patients

6273 Patients excluded:4962 No diagnosis of ADHF or ACS1311 Not available

684 Did not pass screen or not screened:196 Not eligible488 Declined screen

93 Declined participation

116 Incomplete follow-up

FIGURE 1. Flowchart displaying how patient eligibility was determined. Notavailable includes not enough information (n¼166), unstable psychiatricconditions (n¼33), admitted from a nursing home (n¼37), in hospice(n¼4), chronically impaired cognition (n¼39), too ill (n¼201), previouslyenrolled or in a conflicting study (n¼122), unavailable (n¼18), discharged(n¼660), in police custody (n¼1), uncooperative or left against medicaladvice (n¼5), or passed away in hospital before screening (n¼25).Incomplete follow-up includes did not complete follow-up call (n¼51), didnot complete medication review or managing medications (n¼49), with-drew (n¼11), or died (n¼5). ACS ¼ acute coronary syndrome; ADHF ¼acute decompensated heart failure.


1046

(19.0%) were nonwhite. A total of 333 patients(70.7%) had been diagnosed as having ACS,whereas 39 (8.3%) had both ACS and ADHF.A total of 79 patients (16.8%) had inadequateor marginal health literacy. Of note, 341patients (72.3%) reported some level of depres-sion before hospitalization. Participants self-reported taking a mean of 12 medications afterdischarge (range, 1-31). A total of 242 patients(51.4%) had at least one discordantmedication;among them, the median number of discordantmedications was 2 (interquartile range, 1-3).There were 129 participants (27.4%) not takinga medication that was on the discharge list (anomission), and 168 (35.7%) were taking amedication not listed on the discharge list (errorof commission). In addition, 279 participants(59.2%) reported amisunderstanding in indica-tion, dose, or frequency for at least one cardiac


medication tested. Of those participants whohad at least 1 medication misunderstanding,the median number of misunderstandings was2 (interquartile range, 1-3).

In adjusted analyses (Figure 2), higher sub-jective numeracy was associated with lowerodds of having discordant medications (OR,0.81; 95% CI, 0.67-0.98; per 2-point changeon the SNS). Being separated or divorced (OR,0.58; 95% CI, 0.34-0.98) or widowed (OR,0.58; 95% CI, 0.34-0.99) was associated withlower odds of having errors of commission. Incontrast, higher levels of depression were associ-ated with higher odds of errors of commission(OR, 1.36; 95% CI, 1.00-1.85; per 8-unit changeon the Patient HealthQuestionnaire 8). Similarly,race other than white or African American (OR,2.02; 95% CI, 1.13-3.60) and higher educationalattainment (OR, 1.29; 95% CI, 1.06-1.56; per4-year change) were associated with higherodds of having discordant medications, mostlyattributable to errors of commission. No riskfactors were significantly associated (P¼.13-.96)with errors of omission.

For the cardiacmedications tested (Figure 3),higher health literacy (OR, 0.84; 95% CI, 0.74-0.95; per 7-point change on the short form ofthe Test of Functional Health Literacy in Adults)and higher subjective numeracy (OR, 0.77; 95%CI, 0.63-0.95; per 2-point change on the SNS)were associated with lower odds of misunder-standings in indication, dose, or frequency,mostly attributable to errors in indication. Olderage (OR, 1.16; 95% CI, 1.02-1.33; per 10-yearchange) and being single (OR, 1.68; 95% CI,1.04-2.70) were associated with higher odds ofmisunderstandings in indication, dose, or fre-quency, again mostly attributable to errors inindication. Being female (OR, 0.60; 95% CI,0.46-0.78) was associated with lower odds ofmisunderstandings, particularly in indicationand dose. Finally, worse cognitive function(OR, 1.38; 95% CI, 1.05-1.82) was associatedwith higher odds of misunderstandings in fre-quency. Having a primary diagnosis of heartfailure (OR, 0.35; 95%, CI 0.16-0.76) wasassociated with lower odds of a frequencymisunderstanding.

DISCUSSIONWe identified at least onemedication error orun-intentional discrepancy between the dischargelist and patient report in 51.4% of patients. As




TABLE 3. Baseline Characteristics of the StudySamplea

Characteristic

No. (%)of Study

Participantsb

(N¼471)

Mean age (y) (SD) 59.4 (12.5)Female 226 (47.9)Racec

White 380 (80.7)African American 80 (16.9)All other races 10 (2.1)

Marital statusSingle or never married 43 (9.1)Widowed 58 (12.3)Separated or divorced 79 (16.8)Married or living with partner 291 (61.8)

Mean social support score(ENRICHD) (SD) (scorerange, 8-34) 25.7 (4.3)

Educational attainment (y)0-8 23 (4.9)9-11 42 (8.9)12 or GED 135 (28.7)13-15 162 (34.4)16 63 (13.4)�17 46 (9.8)

Income ($)d

0-9999 34 (7.2)10,000-14,999 30 (6.4)15,000-19,999 35 (7.4)20,000-24,999 56 (11.9)25,000-34,999 86 (18.3)35,000-49,000 78 (16.6)50,000-74,999 61 (13.0)75,000-99,999 42 (8.9)�100,000 39 (8.3)

Primary diagnosisACS 333 (70.7)ADHF 99 (21.0)Both ACS and ADHF 39 (8.3)

Health literacy score (s-TOFHLA)e

Inadequate (score range, 0-16) 46 (9.8)Marginal (score range, 17-22) 33 (7.0)Adequate (score range, 23-36) 387 (82.2)

Subjective numeracy score (SNS)(score range, 1-6) 4.3 (1.4)

Cognition (SPMSQ)No impairment, 0-2 errors 436 (92.6)Impaired, 3-10 errors 35 (7.4)

Global health status score(PROMIS) (score range, 1-5) 2.9 (0.8)

Depression score (PHQ-8)No depression (score range, 0-4) 130 (27.6)Mild depression (score range, 5-9) 166 (35.2)Moderate (score range, 10-14) 112 (23.8)

Continued

TABLE 3. Continued

Characteristic

No. (%)of Study

Participantsb

(N¼471)

Depression score (PHQ-8), continuedModerately severe (score range,

15-19) 44 (9.3)Severe (score range, 20-24) 19 (4.0)

Medication adherence (ARMS-7)f

(score range, 7-28) 9.6 (2.6)

aACS ¼ acute coronary syndrome; ADHF ¼ acute decom-pensated heart failure; ENRICHD ¼ Enhancing Recovery inCoronary Heart Disease Social Support Inventory; GED ¼general educational development; PHQ-8 ¼ Patient HealthQuestionnaire 8; PROMIS ¼ Patient Reported OutcomesMeasurement Information System; SPMSQ ¼ Short PortableMental Status Questionnaire; s-TOFHLA ¼ short form of theTest of Functional Health Literacy in Adults.bData are presented as No. (percentage) of study participantsunless otherwise indicated.cMissing data for 1 participant.dMissing data for 10 participants.eMissing data for 5 participants.fMissing data for 8 participants.



hypothesized, higher levels of health literacyand numeracy were associated with lowerodds of medication errors. However, we didnot observe significant associations betweenmedication errors and other potential risk fac-tors: medications changed during hospitaliza-tion, poor social support, or low preadmissionmedication adherence.

Our results must be placed into the contextof prior studies that specifically looked at post-discharge medication errors. The frequency ofomission (27.4%) and commission (35.7%) er-rors is similar to what we have reported in thePharmacist Intervention for Low Literacy inCardiovascular Disease study, which enrolleda comparable inpatient population.38 Further-more, 192 patients (40.8%) in the current studycorrectly reported the indication, dose, and fre-quency for tested cardiac medications. Priorstudies have reported each outcome individu-ally: 64% to 79% of patients reported the cor-rect indication,39,40 56% the correct dosage,39

and 68% the correct frequency.39 The fre-quency of errors in our sample was similar toprior studies18,19,41,42 but much higher thanthe 14% observed by Coleman et al.17 In theirstudy, medication errors were assessed by ageriatric nurse practitioner during in-home

/10.1016/j.mayocp.2014.04.023 1047



0.2 0.5 1 2 5

Age (per 10 year change) 1.04 (0.91–1.18)

1.17 (0.87–1.57)

0.96 (0.68–1.36)

2.06 (1.15–3.72)

0.91 (0.60–1.38)

0.82 (0.52–1.29)

0.94 (0.55–1.60)

1.28 (1.06–1.56)

0.86 (0.69–1.09)

0.86 (0.62–1.18)

0.95 (0.55–1.66)

0.93 (0.81–1.07)

0.92 (0.80–1.06)

0.98 (0.82–1.18)

1.21 (0.96–1.52)

0.98 (0.79–1.20)

1.04 (0.83–1.30)

0.98 (0.82–1.16)

0.84 (0.67–1.06)

0.82 (0.67–0.99)

Female (vs male)

Black/AA (vs white)

Other race (vs white)

Separated/divorced (vs married)

Widowed (vs married)

Single/never married (vs married)

Education (per 4 year change)

s-TOFHLA (per 7 point change)

Subjective numeracy (per 2 point change)

Cognition (per 1 point change)

PROMIS (per 1 point change)

Depression (per 8 point change)

Med adherence (per 3 point change)

Elixhauser (per 12 point change)

Total med changes (per every 6 med changes)

Social support (per 6 point change)

Income (per 3 category change)

Heart failure/no ACS (vs ACS/no heart failure)

Heart failure/ACS (vs ACS/no heart failure)

0.2 0.5 1 2 5

1.16 (0.96–1.40)

1.09 (0.75–1.59)

1.02 (0.59–1.75)

1.54 (0.65–3.66)

1.51 (0.88-2.59)

1.30 (0.70-2.45)

1.66 (0.78-3.55)

0.99 (0.75-1.30)

0.90 (0.66-1.22)

0.99 (0.62-1.57)

0.89 (0.36-2.17)

0.91 (0.75-1.10)

0.90 (0.74-1.09)

1.17 (0.88-1.55)

0.99 (0.71-1.37)

1.16 (0.90-1.50)

1.02 (0.77-1.35)

1.18 (0.95-1.47)

0.82 (0.58-1.16)

0.88 (0.67-1.15)

0.2 0.5 1 2 5

0.97 (0.83-1.12)

1.21 (0.80-1.83)

0.92 (0.59-1.42)

2.68 (1.21-5.96)

0.58 (0.34-0.97)

0.56 (0.33-0.97)

0.60 (0.32-1.13)

1.51 (1.18-1.93)

0.79 (0.58-1.09)

0.78 (0.51-1.18)

1.00 (0.59-1.71)

0.96 (0.81-1.13)

0.94 (0.78-1.12)

0.89 (0.71–1.11)

1.34 (0.98–1.82)

0.88 (0.66-1.15)

1.05 (0.76-1.44)

0.86 (0.69-1.06)

0.86 (0.66-1.11)

0.79 (0.62-1.01)

A Discordance

Odds ratio

B Omissions

Odds ratio

C Commissions

Odds ratio

FIGURE 2. Factors associated with discordant medications (A), errors of omission (B), and commission (C) reported as odds ratioswith 95% CIs. AA ¼ African American; ACS ¼ acute coronary syndrome; PROMIS ¼ Patient Reported Outcomes MeasurementInformation System; s-TOFHLA ¼ short form of the Test of Functional Health Literacy in Adults.


1048

interviews of recently discharged older adults.This method may have allowed the geriatricnurse practitioner to synthesize multiple sour-ces of information in determining the patient’scorrect medications.

Similar to our findings, Maniaci et al40

found no association between regimen knowl-edge and educational attainment. In addition,several studies have documented an associa-tion between the number of medications andmedication errors.17,38,43 However, we didnot observe that associations in these analyses.

Notably, Lindquist et al2 found a similarprevalence (56%) of postdischarge medicationerrors in a sample of older adults. Lowerhealth literacy was found to be a significantrisk factor for unintentional nonadherence,which is congruent with our results.

To our knowledge, ours is the first study tofind an association between low numeracy andpostdischargemedication errors. Low numeracyhas been linked to other health outcomes: poorself-efficacy and self-care in diabetes and asthmamanagement,44,45 poorer glycemic control indiabetes,46 poorer quality of life in asthma(mediated by self-efficacy),47 and poorer


medication management in chronic diseases.48

Interestingly, in our analyses, numeracy wasnot associated specifically with misunderstand-ings in dose or frequency, which are the morenumerical aspects of medication instructions.Rather, numeracy was associated with having adiscordant medication (errors of omission orcommission) and having a misunderstandingin indication.

Strengths of this study include the rela-tively large sample size, high response rates,and the use of both objective and subjectivemeasures of social determinants of health,including health literacy and numeracy. Wealso acknowledge the limitations of our study.Our study included only patients admittedwith ACS or ADHF, limiting the generaliz-ability. We recorded but did not delve intothe cause of differences between the dischargemedication list and patient report, giving pa-tients the benefit of the doubt when reportingthat a clinician had changed the regimen. Pa-tients also could have reported what wasprinted on their list or bottles as opposed towhat they actually were taking, biasing our re-sults toward the null hypothesis. In addition,




A Discrepancy

Odds ratio

B Indication

Odds ratio

C Dose

Odds ratio

D Frequency

Odds ratio

0.2 0.5 1 2 5

Age (per 10 year change) 1.15 (1.01-1.31)

0.60 (0.46-0.79)

1.01 (0.68-1.48)

1.06 (0.53-2.12)

1.14 (0.77-1.69)

1.03 (0.66-1.63)

1.71 (1.05-2.78)

1.01 (0.80-1.29)

0.95 (0.76-1.20)

0.84 (0.58-1.21)

0.65 (0.41-1.02)

0.84 (0.74-0.95)

1.12 (0.95-1.31)

1.10 (0.90-1.33)

1.08 (0.86-1.37)

0.93 (0.79-1.10)

1.06 (0.91-1.24)

0.93 (0.77-1.11)

1.15 (0.91-1.45)

0.77 (0.62-0.95)

Female (vs male)

Black/AA (vs white)

Other race (vs white)

Separated/divorced (vs married)

Widowed (vs married)

Single/never married (vs married)

Education (per 4 year change)

s-TOFHLA (per 7 point change)

Subjective numeracy (per 2 point change)

Cognition (per 1 point change)

PROMIS (per 1 point change)

Depression (per 8 point change)

Med adherence (per 3 point change)

Elixhauser (per 12 point change)

Social Support (per 6 point change)

Cardiac med changes (per every 3 cardiac med changes)

Income (per 3 category change)

Heart failure/no ACS (vs ACS/no heart failure)

Heart failure/ACS (vs ACS/no heart failure)

0.2 0.5 1 2 5

1.27 (1.08-1.50)

0.64 (0.46-0.89)

0.90 (0.60-1.37)

0.67 (0.23-1.91 )

0.86 (0.54-1.37)

0.84 (0.53-1.35)

1.82 (1.00-3.31)

0.91 (0.68-1.21)

0.80 (0.61-1.05)

0.97 (0.63-1.49)

0.71 (0.41-1.25)

0.77 (0.66-0.89)

1.10 (0.91-1.31)

1.16 (0.92-1.47)

0.93 (0.70-1.24)

1.01 (0.82-1.23)

1.31 (1.10-1.57)

0.94 (0.74-1.19)

1.22 (0.93-1.60)

0.65 (0.49-0.85)

0.2 0.5 1 2 5

0.99 (0.81-1.22)

0.62 (0.43-0.91)

1.19 (0.67-2.09)

1.14 (0.28-4.64)

1.25 (0.71-2.19)

1.04 (0.48-2.26)

1.44 (0.69-2.98)

0.94 (0.63-1.40)

1.21 (0.86-1.71)

0.74 (0.41-1.33)

0.45 (0.19-1.06)

0.83 (0.69-1.00)

1.16 (0.93-1.44)

1.03 (0.78-1.37)

1.20 (0.83-1.72)

0.82 (0.64-1.04)

0.78 (0.61-0.99)

0.99 (0.78-1.26)

0.94 (0.63-1.41)

0.81 (0.59-1.11)

0.2 0.5 1 2 5

1.04 (0.82-1.32)

0.82 (0.49-1.38)

0.62 (0.28-1.36)

0.38 (0.04-3.44)

1.44 (0.66-3.15)

1.37 (0.51-3.67)

1.99 (0.82-4.87)

0.93 (0.55-1.58)

1.14 (0.70-1.85)

0.30 (0.14-0.63)

0.49 (0.15-1.61)

0.91 (0.70-1.18)

1.39 (1.05-1.85)

0.93 (0.61-1.41)

1.34 (0.85-2.10)

0.75 (0.51-1.09)

0.83 (0.58-1.19)

0.92 (0.63-1.33)

1.26 (0.81-1.97)

0.71 (0.48-1.07)

FIGURE 3. Factors associated with any misunderstanding (indication, dose, or frequency; A), indication (B), dose (C), and frequency(D). AA ¼ African American; ACS ¼ acute coronary syndrome; PROMIS ¼ Patient Reported Outcomes Measurement InformationSystem; s-TOFHLA ¼ short form of the Test of Functional Health Literacy in Adults.


we relied on the discharge list given to thepatient, which may have had errors due to sub-optimal medication reconciliation practices.Patients who were too ill or refused to completethe telephone interview may have been atincreased risk of having medication errors,whereby our results underestimate the fre-quency of errors in our sample. We did notassess clinical outcomes, such as ADEs; howev-er, health literacy has not previously been asso-ciated with ADEs.41 Although we found thatpatients who self-identified as other than whiteor African American had higher odds of havingdiscordant medications and errors of commis-sion, we note that this group was small. Finally,we did not perform a Bonferroni correction formultiple testing, but we clearly identified a pre-specified set of covariates before building theanalytical models.

CONCLUSIONIn summary, we determined that half of pa-tients with ACS and/or ADHF had a medicationerror in the days after hospital discharge. Theseerrors have the potential to harm patients; thus,we must understand which factors are associ-ated with an increased risk of errors. We foundthat patients with low health literacy andnumeracy are at increased risk of medicationerrors; therefore, identification of patients at


risk can help clinicians provide appropriatedischarge medication education.

ACKNOWLEDGMENTSWethank Joanna Lee for organizing and enteringdata and Dr Jesse Ehrenfeld for providing Elix-hauser scores. The content is solely the responsi-bility of the authors and does not necessarilyrepresent the official views of the National Insti-tutes of Health. The funding agency was notinvolved in the design and conduct of the study;collection, management, analysis, and interpre-tation of the data; and preparation, review, orapproval of the manuscript. Drs Mixon and Kri-palani had full access to all study data and takeresponsibility for data integrity and the accuracyof data analysis. The views expressed in thisarticle are those of the author(s) and do notnecessarily represent the views of the Depart-ment of Veterans Affairs.

Abbreviations and Acronyms: ACS = acute coronarysyndrome; ADE = adverse drug event; ADHF = acutedecompensated heart failure; ARMS = Adherence to Refillsand Medications Scale; OR = odds ratio; SNS = SubjectiveNumeracy Scale; VICS = Vanderbilt Inpatient Cohort Study

Affiliations (Continued from the first page of thisarticle.): Department of Biostatistics (S.N., J.S.S., T.S.), Cen-ter for Quality Aging (J.M.L.J., J.F.S.), and Center for ClinicalQuality and Implementation Research (S.K.), Vanderbilt

/10.1016/j.mayocp.2014.04.023 1049




1050

University, and Department of Pharmaceutical Services,Vanderbilt University Medical Center (A.P.M.), Nashville,Tennessee.

Grant Support: This study was supported by grant R01HL109388 from the National Heart, Lung, and Blood Insti-tute (Dr Kripalani) and in part by grant UL1 RR024975-01from the National Center for Research Resources and grant2 UL1 TR000445-06 from the National Center forAdvancing Translational Sciences.

Potential Competing Interests: Dr Kripalani is a consultantto and holds equity in PictureRx, LLC. Dr Mixon is a Veter-ans Affairs Health Services Research and Development Ser-vice Career Development awardee at the NashvilleDepartment of Veterans Affairs. No other authors reportconflicts of interest or financial disclosures.

Data Previously Presented: The study was presented atthe Society of General Internal Medicine 35th AnnualMeeting; April 25, 2013; Denver, Colorado.

Correspondence: Address to Amanda S. Mixon, MD, MS,MSPH, Suite 6000 MCE, North Tower, 1215 21st Ave S,Nashville, TN 37232 ([email protected]).

REFERENCES1. Cua YM, Kripalani S. Medication use in the transition from hos-

pital to home. Ann Acad Med. 2008;37(2):136-141.2. Lindquist LA, Go L, Fleisher J, Jain N, Friesema E, Baker DW.

Relationship of health literacy to intentional and unintentionalnon-adherence of hospital discharge medications. J Gen InternMed. 2012;27(2):173-178.

3. Haynes KT, Oberne A, Cawthon C, Kripalani S. Pharmacists’recommendations to improve care transitions. Ann Pharmac-other. 2012;46(9):1152-1159.

4. Kaboli PJ, Hoth AB, McClimon BJ, Schnipper JL. Clinical pharma-cists and inpatient medical care: a systematic review. Arch InternMed. 2006;166(9):955-964.

5. Selden CR, Zorn M, Ratzan S, Parker RM. Current Bibliographiesin Medicine: Health Literacy. Bethesda, MD: National Library ofMedicine; 2000.

6. Rothman RL, Housam R, Weiss H, et al. Patient understandingof food labels: the role of literacy and numeracy. Am J Prev Med.Nov 2006;31(5):391-398.

7. Campbell NL, Boustani MA, Skopelja EN, Gao S, Unverzagt FW,Murray MD. Medication adherence in older adults with cognitiveimpairment: a systematic evidence-based review. Am J GeriatrPharmacother. 2012;10(3):165-177.

8. Loke YK, Hinz I, Wang X, Salter C. Systematic review of con-sistency between adherence to cardiovascular or diabetesmedication and health literacy in older adults. Ann Pharmac-other. 2012;46(6):863-872.

9. Gellad WF, Grenard JL, Marcum ZA. A systematic review ofbarriers to medication adherence in the elderly: looking beyondcost and regimen complexity. Am J Geriatr Pharmacother. 2011;9(1):11-23.

10. Chaudhry SI, Wang Y, Gill TM, Krumholz HM. Geriatric condi-tions and subsequent mortality in older patients with heart fail-ure. J Am Coll Cardiol. 2010;55(4):309-316.

11. Calvillo-King L, Arnold D, Eubank KJ, et al. Impact of socialfactors on risk of readmission or mortality in pneumoniaand heart failure: systematic review. J Gen Intern Med. 2013;28(2):269-282.

12. Reese RL, Freedland KE, Steinmeyer BC, Rich MW, Rackley JW,Carney RM. Depression and rehospitalization following acute


myocardial infarction. Circ Cardiovasc Qual Outcomes. 2011;4(6):626-633.

13. Forster AJ, Murff HJ, Peterson JF, Gandhi TK, Bates DW. The inci-dence and severity of adverse events affecting patients afterdischarge from the hospital.Ann InternMed. 2003;138(3):161-167.

14. Forster AJ, Clark HD, Menard A, et al. Adverse events amongmedical patients after discharge from hospital. CMAJ. 2004;170(3):345-349.

15. Forster AJ, Murff HJ, Peterson JF, Gandhi TK, Bates DW.Adverse drug events occurring following hospital discharge.J GenIntern Med. 2005;20(4):317-323.

16. Pippins JR, Gandhi TK, Hamann C, et al. Classifying and predict-ing errors of inpatient medication reconciliation. J Gen InternMed. 2008;23(9):1414-1422.

17. Coleman EA, Smith JD, Raha D, Min SJ. Posthospital medicationdiscrepancies: prevalence and contributing factors. Arch InternMed. 2005;165(16):1842-1847.

18. Mesteig M, Helbostad JL, Sletvold O, Rosstad T, Saltvedt I. Un-wanted incidents during transition of geriatric patients fromhospital to home: a prospective observational study. BMCHealth Serv Res. 2010;10:1.

19. Lalonde L, Lampron AM, Vanier MC, Levasseur P, Khaddag R,Chaar N. Effectiveness of a medication discharge plan for tran-sitions of care from hospital to outpatient settings. Am J HealthSyst Pharm. 2008;65(15):1451-1457.

20. Hain DJ, Tappen R, Diaz S, Ouslander JG. Cognitive impairmentand medication self-management errors in older adults dis-charged home from a community hospital. Home HealthcNurse. 2012;30(4):246-254.

21. Schnipper JL, Kirwin JL, Cotugno MC, et al. Role of pharmacistcounseling in preventing adverse drug events after hospitaliza-tion. Arch Intern Med. 2006;166(5):565-571.

22. Persell SD, Heiman HL, Weingart SN, et al. Understanding ofdrug indications by ambulatory care patients. Am J Health SystPharm. 2004;61(23):2523-2527.

23. Unroe KT, Pfeiffenberger T, Riegelhaupt S, Jastrzembski J,Lokhnygina Y, Colon-Emeric C. Inpatient medication reconcili-ation at admission and discharge: A retrospective cohort studyof age and other risk factors for medication discrepancies. Am JGeriatr Pharmacother. 2010;8(2):115-126.

24. Wong JD, Bajcar JM, Wong GG, et al. Medication reconciliationat hospital discharge: evaluating discrepancies. Ann Pharmac-other. 2008;42(10):1373-1379.

25. Meyers AG, Salanitro AH, Wallston KA, et al. Determinants ofhealth after hospital discharge: rationale and design of the Van-derbilt Inpatient Cohort Study (VICS) 2013. BMC Health ServRes. 2014;14:10.

26. Centers for Disease Control and Prevention. Behavioral RiskFactor Surveillance System Survey Questionnaire. http://www.cdc.gov/brfss/index.htm. Accessed March 10, 2010.

27. Enhancing Recovery in Coronary Heart Disease (ENRICHD)study intervention: rationale and design. Psychosom Med.2001;63(5):747-755.

28. The ENRICHD Investigators. Enhancing recovery in coronaryheart disease patients (ENRICHD): study design and methods.Am Heart J. 2000;139(1, pt 1):1-9.

29. Nurss JR, Parker RM, Williams MV, Baker DW. Short Test ofFunctional Health Literacy in Adults. Snow Camp, NC: Pepper-corn Books and Press; 1998.

30. Wallston KA, McNaughton C, Storrow A, Cavanaugh K, Roth-man RL. Validation of a short, 3-item version of the SubjectiveNumeracy Scale (SNS-3). Paper presented at: Health LiteracyResearch Conference; October 18, 2011; Chicago, IL.

31. Fagerlin A, Zikmund-Fisher BJ, Ubel PA, Jankovic A, Derry HA,Smith DM. Measuring numeracy without a math test: develop-ment of the Subjective Numeracy Scale. Med Decis Making.2007;27(5):672-680.

32. Pfeiffer E. A short portable mental status questionnaire for theassessment of organic brain deficit in elderly patients. J Am Ger-iatr Soc. 1975;23(10):433-441.


mailto:[email protected]

http://www.cdc.gov/brfss/index.htm

http://www.cdc.gov/brfss/index.htm




33. Hays RD, Bjorner JB, Revicki DA, Spritzer KL, Cella D. Develop-ment of physical and mental health summary scores from thepatient-reported outcomes measurement information system(PROMIS) global items. Qual Life Res. 2009;18(7):873-880.

34. Kroenke K, Strine TW, Spitzer RL, Williams JB, Berry JT,Mokdad AH. The PHQ-8 as a measure of current depressionin the general population. J Affect Disord. 2009;114(1-3):163-173.

35. Kripalani S, Risser J, Gatti M, Jacobson TA. Development andevaluation of the Adherence to Refills and Medications Scale(ARMS) among low-literacy patients with chronic disease. ValueHealth. 2009;12(1):118-123.

36. van Walraven C, Austin PC, Jennings A, Quan H, Forster AJ.A modification of the Elixhauser comorbidity measures into apoint system for hospital death using administrative data. MedCare. 2009;47(6):626-633.

37. R: A language and environment for statistical computing [com-puter program]. Vienna, Austria: R Foundation for StatisticalComputing; 2013.

38. Salanitro AH, Osborn CY, Schnipper JL, et al. Effect of patient-and medication-related factors on inpatient medication recon-ciliation errors. J Gen Intern Med. 2012;27(8):924-932.

39. Maniaci MJ, Heckman MG, Dawson NL. Functional health liter-acy and understanding of medications at discharge. Mayo ClinicProc. 2008;83(5):554-558.

40. Mosher HJ, Lund BC, Kripalani S, Kaboli PJ. Association of healthliteracy with medication knowledge, adherence, and adversedrug events among elderly veterans. J Health Commun. 2012;17(suppl 3):241-251.


41. Hain D, Diaz S, Tappen R, Ouslander JG. Medication Discrep-ancies and Research Implications Among Older Adults DischargedHome From a Community Hospital. Boca Raton: Charles E.Schmidt College of Biomedical Science, Florida Atlantic Univer-sity; 2011.

42. Moore C, Wisnivesky J, Williams S, McGinn T. Medicalerrors related to discontinuity of care from an inpatientto an outpatient setting. J Gen Intern Med. 2003;18(8):646-651.

43. Stitt DM, Elliott DP, Thompson SN. Medication discrepanciesidentified at time of hospital discharge in a geriatric population.Am J Geriatr Pharmacother. 2011;9(4):234-240.

44. Berkman ND, Sheridan SL, Donahue KE, Halpern DJ, Crotty K.Low health literacy and health outcomes: an updated system-atic review. Ann Intern Med. 2011;155(2):97-107.

45. Al Sayah F, Majumdar SR, Williams B, Robertson S, Johnson JA.Health literacy and health outcomes in diabetes: a systematicreview. J Gen Intern Med. 2013;28(3):444-452.

46. Cavanaugh K, Huizinga MM, Wallston KA, et al. Association ofnumeracy and diabetes control. Ann Intern Med. 2008;148(10):737-746.

47. Apter AJ, Wang X, Bogen D, et al. Linking numeracy andasthma-related quality of life. Patient Educ Couns. 2009;75(3):386-391.

48. Waldrop-Valverde D, Osborn CY, Rodriguez A, Rothman RL,Kumar M, Jones DL. Numeracy skills explain racial differencesin HIV medication management. AIDS Behav. 2010;14(4):799-806.

/10.1016/j.mayocp.2014.04.023 1051



Documents

Characteristics Associated With Postdischarge Medication Errors