| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
Correspondence: Address correspondence to Denise Orwig, PhD, Division of Gerontology, Department of Epidemiology and Preventive Medicine, University of Maryland School of Medicine, 660 West Redwood Street, Suite 200, Baltimore, MD 21201. E-mail: dorwig{at}epi.umaryland.edu
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Key Words: Medications • Compliance • Adherence • Community dwelling • Assessment
Research has long established that drugs may cause or contribute to negative adverse events, such as significant morbidity and increased costs of care, by increasing health care utilization (Atkin & Shenfield, 1995; Bates et al., 1995; Hanlon et al., 1997; Johnson & Bootman, 1995; Walker & Wynne, 1994). Researchers have identified noncompliance with a medication regimen as a preventable reason for adverse drug events; medication noncompliance is the second largest health problem in the United States in terms of resources consumed. Tafreshi, Melby, Kaback, and Nord (1999) found that 70.4% of medication-related emergency room visits assessed were judged to be mostly due to noncompliance issues. Investigators have found that drug-related admissions secondary to noncompliance occur in 6% to 28% of elderly patients (Col, Fanale, & Kronholm, 1990; Courtman & Stallings, 1995; Dartnell et al., 1996; Grymonpre, Mitenko, Sitar, Aoki, & Montgomery, 1988; Schneider, Mion, & Frengley, 1992), and the mean cost per admission associated with medication noncompliance is $2,150 (Col et al.).
Studies have reported noncompliance or nonadherence rates in the elderly population that range from 26% to 59% (van Eijken, Tsang, Wensing, de Smet, & Grol, 2003). Medication management is a complicated process, involving selecting the right drug, ensuring that the patient can obtain it and take it correctly, and evaluating its action on the patient (Garrard, Harms, & Hanlon, 1998). One study by Ruscin and Semla (1996) showed that, in a sample of elders with an average age of 80, 22% of those who were physically dependent and one third of those who had cognitive impairment were still responsible for taking their own medications. Compounding this situation, a study by Royall, Cordes, and Polk (1997) showed that half of residents in the community they studied had trouble comprehending medication information.
Medication compliance assessments in the literature include performance-based measures (observation of medication-taking behavior, reading prescription labels, opening safety caps, removing pills, etc.; Edelberg, Shallenberger, & Wei, 1999; Fitten, Coleman, Siembieda, Yu, & Ganzell, 1995; Meyer & Schuna, 1989; Patterson et al., 2002; Raehl, Bond, Woods, Patry, & Sleeper, 2002; Ruscin & Semla, 1996), questionnaires (Bennett et al., 2001; de Klerk, van der Heijde, Landewe, van der Tempel, & van der Linden, 2003; Yamada, Sugiyama, Ashida, Ohwaki, & Fujii, 2001), home-based measures (Brown et al., 1998; Stewart, Pearson, Luke, & Horowitz, 1998; West et al., 1997), and pharmacy-type interventions (patient education, pharmacist-developed discharge plans; Ali, Laurin, Lariviere, Tremblay, & Cloutier, 2003; Al-Rashed, Wright, Roebuck, Sunter, & Chrystyn, 2002; Grant, Devita, Singer, & Meigs, 2003; Hoffman et al., 2003; Jensen, 2003; Lowe, Raynor, Purvis, Farrin, & Hudson, 2000; Nazareth et al., 2001; Sturgess, McElnay, Hughes, & Crealey, 2003). A majority of the measures are limited by either their use by clinic patients or with patients using a specific drug type, narrow coverage of compliance issues, little utility for use with an entire drug regimen, lack of validation testing, or testing outside the United States.
Currently there is no comprehensive standardized instrument that looks at knowledge, administration, and procurement that can be used by untrained care providers, such as family members, to assess the older adult's ability to self-medicate in the home environment. Informal care providers who assist with activities of daily living (ADLs) are often the ones who have firsthand contact with the older adult on a regular basis and are in a position to identify medication management problems; yet, they may feel unable to identify such issues. Children and other relatives of aging adults provide about 80% of all care, and these informal care providers provide three times as much direct care for elders as do all nursing homes, hospitals, and other institutions combined (Perlman, 1983). Medication assistance is becoming a more prevalent aspect of caregiving (Brodaty & Green, 2002; Conn, Taylor, & Messina, 1995; Smith, Francis, Gray, Denham, & Graffy, 2003). A recent survey of long-term caregivers found that 48% of caregivers helped with medications (Donelan et al., 2002). These types of care providers are an often overlooked resource to assist with medication compliance problems among elderly adults.
The purpose of this article is to describe an assessment instrument—the Medication Management Instrument for Deficiencies in the Elderly (MedMaIDE)—and to provide results of a reliability and validity study of older adults who self-administered medications and lived independently in the Baltimore area. Nonmedically trained study staff conducted the assessments in the home environment.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
The MedMaIDE consists of 20 items covering three areas considered important for proper medication management: (a) What a person knows about the medication he or she is taking (8 items); (b) Whether a person knows how to take his or her medication (6 items); and (c) Whether a person knows how to get his or her medication from a doctor or pharmacy (6 items). A summary of these domains is shown in Appendix A. In addition to assessing the three areas, the administrator records all medications (prescription, over-the-counter, alternative therapies, etc.; see Appendix B) currently being used by the individual. The instrument is completed for the entire medication regimen and takes approximately 30 min to administer. The measure incorporates both subjective information and direct observations of medication-taking behavior.
|
|
We tested the instrument in a convenience sample of 50 older adults (aged 65 and older) who lived in the community and who were self-medicating. Nonmedical study staff administered the MedMaIDE to participants two times, one week apart. Nonmedical staff were trained research assistants who participated in a standardized training session. A study pharmacist conducted a baseline pill count, followed by another pill count 30 days later. In addition, we collected data on cognitive impairment using the Mini-Mental State Examination (MMSE) (Folstein, Folstein, & McHugh, 1975), ADLs (Katz, Ford, Moskowitz, Jackson, & Jaffe, 1963), instrumental ADLs (IADLs) (Lawton & Brody, 1969), and comorbidities at both time points. We coded all drugs using the American Hospital Formulary Service coding scheme (McEvoy, 1998).
Study Population
We recruited participants from several high-rise apartment complexes in Baltimore; other participants were residents of a local retirement community. Older adults were eligible if they were aged 65 and older, lived independently, were able to understand English, took at least one prescription medication, and were not currently relying on another person to administer medications (i.e., were self-medicating). The University of Maryland, Baltimore, Institutional Review Board reviewed and approved the protocol, and we obtained informed consent from all participants or their proxies. We approached 61 older adults for participation in the study and enrolled 50 (82%).
Measures
Scoring the MedMaIDE
Each of the three areas assessed by the MedMaIDE has a deficiency subscore that together make up a total deficiency score. Critical items (identified with an asterisk in Appendix A) have been identified within each area and are used in calculating the deficiency subscores and the total deficiency score. The noncritical items are still important for assessing overall medication management but are not included in the deficiency scores. A person must be able to answer or do each item correctly for all prescription medications in the regimen to receive a score of yes. If a person receives a score of no for a critical item, that item gets a score of 1. The administrator sums the items in each area to obtain an area deficiency subscore. The administrator then adds the three subscores together to get the total deficiency score. The maximum total deficiency score is 13. Study staff completed 1-week MedMaIDE follow-ups for 97% of the sample.
Reliability Testing
We assessed reliability by test–retest estimates, interrater measurement, and internal consistency. Use of different raters allows for an assessment of interrater reliability by deriving an intraclass correlation coefficient (ICC). We assessed test–retest reliability on the whole sample and report it using an ICC. A different staff person interviewed a subgroup of 15 participants during one of the interviews to determine interrater reliability, which we also report using an ICC. Finally, Cronbach's alpha coefficient assesses the degree to which the items on the MedMaIDE all measure the same underlying construct, which provides a measure of internal consistency.
Validity Testing
Pill Count Compliance
For the purpose of this study, we used the 30-day pill count as a comparative variable (concurrent, concordant validity). We used the pill count method to measure medication compliance; research has shown this method to be reliable and valid (Botelho & Dudrak, 1992; Grymonpre, Didur, Montgomery, & Sitar, 1998; Klein, German, McPhee, Smith, & Levine, 1982; Matsui et al., 1994). It is also a cost-effective and realistic way to assess compliance of all medications in the home setting. This method involves looking at the number of pills dispensed, the date being dispensed, when the individual started taking the medication, and directions for use. We defined the compliance rate as the total number of pills taken divided by the total number of pills that should have been taken; we then multiplied this number by 100. We used the instructions on the bottle, in addition to new prescriptions or other instructions (e.g., hospital discharge records), to determine the number of pills that should have been taken. We derived the percent compliance for each individual prescription medication and then averaged them to obtain the overall compliance for the medication regimen. The optimal compliance was 100%; however, an acceptable level of compliance was greater than 80% (Botelho & Dudrak). In order to minimize the occurrence of a person throwing away pills to defeat the pill count, we told participants only that we were assessing their medications (this was approved by the University of Maryland, Baltimore, Institutional Review Board). The person counting the pills did not state explicitly that they were counting pills. During the initial visits, study staff administered the questionnaire to the older adult while a pharmacist counted and evaluated the pills and pill bottles. There were no problems using this strategy, and the pharmacist completed 30-day pill counts for 88% of the study sample.
Validity is a term used to describe how well an instrument measures what it purports to measure (e.g., how accurate is the MedMaIDE in detecting an older person's level of deficiency in managing medications?). Ideally, we would establish validity through comparison of the MedMaIDE to a gold standard or to other instruments that measure similar constructs (in this case, pill count compliance). We also evaluated validity by comparing the MedMaIDE to ADL, IADL, and MMSE scores (divergent validity). We also used correlations and response operator characteristic curves to assess validity. In addition, we assessed the unique contribution of the MedMaIDE on pill count compliance through regressions on pill counts controlling for age, gender, ADL score, and MMSE score.
Covariates
In addition to obtaining demographic data, study staff asked questions with respect to the existence of comorbidities and hearing and vision problems. We also assessed cognitive functioning, as well as participants' ability to perform ADLs and IADLs. The MMSE is a well-known assessment instrument used to assess the cognitive status of an individual (Folstein et al., 1975). Previous studies have examined the relationship between cognitive and functional impairment and the capacity for independent medication management and have reported that cognitive impairment is a predictor of poor compliance among older adults who are functionally independent in the community (Botelho & Dudrak, 1992; Okuno, Yanagi, & Tomura, 2001). The ADL index (Katz et al., 1963) evaluates performance in bathing, dressing, toileting, transferring, continence, and feeding. The IADL index (Lawton & Brody, 1969) evaluates more complex behaviors, such as telephoning, shopping, preparing food, keeping house, laundering, using transportation, using medicine, and managing finances. Functional impairment has been shown to affect a person's ability to open medication containers, manipulate small pills, and get to the doctor or pharmacy for a prescription or a refill.
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
|
|
|
80% as compliant and <80% as noncompliant. We used receiver operating characteristic (ROC) curves to choose cut points for the MedMaIDE deficiency score and to identify the sensitivity and specificity of the MedMaIDE. We assessed two cut points for the deficiency score of the MedMaIDE: (a) 0 as compliant and 1 or more as noncompliant, and (b) 0 or 1 as compliant and 2 or more as noncompliant. Table 4 shows the specificity and sensitivity for the two cut points. Specificity of the higher cut point was 83%, while the sensitivity was 68%. This means the MedMaIDE was able to correctly identify participants without compliance problems, and, compared to the pill count, it would identify a modest proportion of people as compliant when they were not. It is important to keep in mind that the definition of compliant with the higher cut point on the MedMaIDE was having one deficiency. A more appropriate assessment is the predictive validity of the MedMaIDE, which is also shown in Table 4. The positive predictive value for the higher cut point was.83. There would be a higher probability compared to the pill count measure that the MedMaIDE would identify deficiencies in medication management; thus, interventions would be implemented when they would not be indicated according to the pill count. Although researchers typically consider 80% or better to be compliant with the pill count measure, it is our belief that some people with less than 20% noncompliance will benefit from interventions as identified through the MedMaIDE.
The Pearson correlation coefficient for the association between the MedMaIDE deficiency score and pill count compliance was –0.52, which suggests a modest measure of concurrent validity. We anticipate that the MedMaIDE may capture more with respect to medication management than just counting pills in a 30-day period. Divergent validity was shown in lower correlations between the MedMaIDE and both the ADL (r = 0.13) and MMSE scores (r = –0.44).
In regression analyses, 27% of the variance in pill count compliance was accounted for by MedMaIDE deficiency score (p <.001), and the MedMaIDE deficiency score remained a significant predictor of compliance even when controlling for age, gender, education, MMSE, and ADLs (p =.033; see Table 5).
|
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
We found the MedMaIDE to be a reliable measure of medication management based on moderate to high test–retest reliability (ICC = 0.93), internal consistency (Cronbach's alpha =.71), and interrater reliability (ICC = 0.73) measures. Compared to a pill count compliance method, the MedMaIDE was highly specific and predictive of compliance and had additional predictive value even when controlling for functional and cognitive status. Although the MedMaIDE had lower sensitivity related to the pill count compliance measure, this is probably due to the fact that the MedMaIDE captured a broader spectrum of issues related to medication-taking behaviors. The strong relationship between the two measures does show concurrent validity. The MedMaIDE's associations with pill count compliance were independent of the effects of age, gender, ADLs, and MMSE status.
In our sample of 50 community-dwelling, self-medicating participants who were 65 years old or older, 70% had one or more deficiencies in medication management on the MedMaIDE. The pill count found approximately 30% of the sample to be noncompliant (i.e., taking less than 80% of prescribed medications over a 30-day period), which is consistent with the literature (van Eijken et al., 2003) even in this relatively high-functioning group of older adults. This small study exemplifies why medication management assessment is so important. The MedMaIDE identified more cases with potential medication problems than did the pill count, cases that could eventually lead to serious adverse events if left undetected.
The study sample was small, and the data are not generalizable to all older adults living in the community. Participants were taking a higher number of medications on average than what has been reported in the literature, but they were responsible for managing these medications on their own. There was little variation in MedMaIDE deficiencies in two of the three domains. This sample was also cognitively intact, on average, yet more than half of the sample was noncompliant with their medication regimens. There could also be other potential confounding variables affecting compliance not controlled for in the regression analysis, although we did control for the five items known to be predictors. A particular strength of the study was the almost equal proportions of White and Black participants.
Another limitation may be the use of pill count compliance. Many methods are available to measure medication compliance among elders, yet there is no gold standard measurement of medication compliance, and all methods (including self-report, serial drug plasma concentration measurement, prescription refill rates, and electronic container monitoring) have potential problems (MacLaughlin et al., 2005; Rudd, Ramesh, Bryant-Kosling, & Guerrero, 1993; Wright, 1993). Frequently, the patient self-reports adherence, which calls into question the accuracy of the data; other modalities such as pillboxes, refill patterns, and electronic monitoring systems have also met with limited success. Assessing blood levels is often not practical and is considered invasive. Investigators have noted that pill counts can be unreliable for assessing compliance (either underestimating or overestimating compliance) and that they have only a moderate correlation with electronic monitoring devices (Choo et al., 1999; Grymonpre et al., 1998; MacLaughlin et al., 2005). However, in a majority of studies, compliance assessment was conducted on only a single type of drug, because many of these modalities are not feasible when assessing an entire drug regimen with multiple medications. Therefore, we chose to use the pill count method as the gold standard because it can be applied to an entire regimen, and we tried to eliminate behavior to defeat the pill count by telling the participants only that their pills were being assessed.
Optimal use of medications in the elderly population remains elusive. The growing number of community-dwelling older adults with functional limitations is likely to further increase the demand for interventions that utilize caregivers. A study of family caregivers' medication administration experiences reported that administration issues and concerns were consistent with the frustrations and stresses of daily hassles found in the caregiving literature (Travis, Bethea, & Winn, 2000). In one study, medication-related tasks accounted for 7.7% of the total caregiving time, yet they contributed substantially to caregivers' stress, with 10 caregivers (32%) reporting problems directly related to medications and 6 (19%) and 16 (52%), respectively, reporting problems in managing the drug regimen currently or within the past year (Ranelli & Aversa, 1994).
Measures are needed to provide older adults and the people who care for them with the necessary tools to minimize adverse events attributable to poor medication compliance. Future studies should investigate whether the MedMaIDE has greater predictive power in preventing adverse events, such as hospital admissions. Future studies should also include a larger sample with a broader range of medication management limitations and a range of cognitive impairments. The MedMaIDE can identify medication management deficiencies surrounding administration, knowledge, and access by non-healthcare professionals, namely the caregivers. The MedMaIDE provides a vehicle for improving medication management by empowering caregivers in the assessment process; providing information on possible interventions needed to improve medication management; reducing caregiver stress and hassles; and facilitating communication between patients, their families, prescribers, and other health care providers through the use of a standardized assessment instrument.
|
Footnotes |
|---|
1 Division of Gerontology, Department of Epidemiology and Preventive Medicine, University of Maryland School of Medicine, Baltimore.
2 Peter Lamy Center, University of Maryland School of Pharmacy, Baltimore.
Decision Editor: Linda S. Noelker, PhD
Received for publication March 16, 2006. Accepted for publication June 2, 2006.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  K. B Farris and B. B. Phillips Instruments Assessing Capacity to Manage Medications Ann. Pharmacother., July 1, 2008; 42(7): 1026 - 1036. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
.