People who take antiretroviral medications, which are commonly referred to as highly active antiretroviral therapy (HAART), often take several different medications that require multiple pills and administration times each day. Thus, it is not uncommon for people to take 20 or more pills per day divided into two and three administration times. Some pills are difficult to take because they are large and hard to swallow. For example, a usual dose of Viracept (Pfizer, New York, NY) (625 mg) requires taking two large tablets twice per day. Other medications must be taken under certain conditions or require specific actions to prepare the dose for ingestion. For example, Videx EC (Bristol Myers Squibb, New York, NY) (didanosine) must be taken on an empty stomach, which means ingesting no food 2 hours before and 1 hour after taking the medicine. Depending on the other medications in the regimen, didanosine sometimes must be taken at separate times from other medications. Crixivan (Merck, Whitehouse Station, NJ), when used as an unboosted protease inhibitor, includes drinking 1.5 liters of water each day, and Kaletra (Abbott, Abbott Park, IL) must be refrigerated and is best taken with food having a moderate fat content. In addition, for most medications that require multiple doses per day, the timing of the dose is extremely important to maintain therapeutic levels of the drug in the bloodstream (U.S. Department of Health and Human Services, 2004).
The complexity of the HAART regimen, as measured by the number of medicines within a regimen or the number of doses of medicines taken each day, has been associated with adherence to the regimen. The results of several studies have shown that as regimen complexity increases, adherence declines (Chesney, Morin, & Sherr, 2000; Eldred, Wu, Chaisson, & Moore, 1998; Hinkin et al., 2002; Paterson et al., 2000; Stone et al., 2001). Clinical experience suggests that other components of a regimen can also contribute to ease or difficulty of taking HAART medications (Stone et al, 2001). These components include food restrictions, fluid requirements, fat content of meals, size and taste of pills, and side effects of the medicines. Recent studies suggest that these and other factors that require modifications in daily living can be reasons contributing to less than perfect adherence (Chesney, Morin et al., 2000; Mehta, Moore, & Graham, 1997; Miller, 1997). Although perfect adherence is not always possible, recent studies suggest that 95% or better adherence is required to reduce the possibility of developing resistance to drugs used in HAART (Ickovics et al., 2002; Paterson et al, 2000).
Recent studies and reports of HAART adherence indicate that there is a growing recognition of the importance of measuring the full range of complexity associated with HAART regimens and examining the extent to which various aspects of complexity are associated with adherence (Chesney, Morin, et al., 2000; Stone, Jordan, Tolson, Miller, Pilon, 2004). However, to date there has been no systematic manner in which complexity has been measured. To meet the need of researchers and to improve comparisons among studies, the authors developed a measure of HAART medication complexity. The purpose of this article is to describe the development and testing of the Antiretroviral Medication Complexity Index (AMCI), an instrument developed to quantify the components of antiretroviral medications and medication regimen complexity.
Medication Complexity Indices
The AMCI is modeled after the Medication Complexity Index (MCI) (Kelley, 1988) and the Epilepsy Medication and Treatment Complexity Index (EMTCI) (DiIorio et al, 2003). The MCI was developed by Kelley (1988) to measure the complexity of cardiovascular medication regimens. Regimen complexity as conceptualized by Kelley includes the number of medications and number of doses taken per day, as well as additional directions and mechanical actions necessary to take the medications. For example, an additional direction is taking the medication with food, and a mechanical action is splitting the pill in half. Kelley developed the MCI form to record information about each of these components of complexity.
The MCI was used as the basis for the development of the EMTCI, which was designed to assess the complexity of regimens for people taking antiepileptic drugs. Like the MCI, the EMTCI assesses the number of medications, number of dosages, and special instructions and administrative actions required for pill-taking. Although some of the special instructions and administrative actions for pill-taking on the EMTCI form are similar to those on the MCI, many are unique to antiepileptic drugs. Both the MCI and the EMTCI were assessed for reliability and validity (Conn, Taylor & Kelley, 1991; DiIorio et al, 2003). The MCI was assessed for test-retest reliability and for content and concurrent validity, whereas the EMTCI was assessed for interrater and intrarater reliability and for content and criterion validity. The results of all assessments provide support for the reliability and validity of both indices.
Development of the Antiretroviral Medication Complexity Index
The MCI and the EMTCI were used as models for the development of the AMCI, which was designed to record self-reported medication complexity. To develop the AMCI, the authors reviewed HAART medications (nonnucleoside reverse transcriptase inhibitors, protease inhibitors, nucleotide/nucleoside reverse transcriptase inhibitors, entry inhibitors) for dosages and special instructions and administrative actions. Based on the review, dosage frequencies, special instructions, and administrative actions were identified. To simplify the instrument, special instructions and administrative actions were combined into one section called administrative actions. Because pill size and side effects of medications are issues with some HAART medications and add to complexity, the pill burden determined by size and number of pills and self-reported severity of side effects was also included in the complexity index. To determine pill size, each pill was measured with a tape measure to determine its length and circumference. The pills were measured twice by two people, and any discrepancies were resolved by measuring the pills again. These data were used to categorize the size of pills as small, medium, and large. Finally, common side effects for each HAART medication were identified. These were compared with the common side effects listed in the medication package inserts and the current “A Guideline for the Use of Antiretroviral Agents in HIV-Infected Adults and Adolescents” issued by the U.S. Public Health Service. (2004).
Description of the Antiretroviral Medication Complexity Index
Using any standard dictionary, complexity can be defined as the quality of having multiple interconnected elements or components that, when combined, serve to make a task more difficult or intricate. The idea of an arrangement or combination of components in relation to defining complexity is noted in an essay on the emergence of complexity in science by Mikulecky (2001). Mikulecky suggests that an important aspect of defining complexity relates to the notion that complexity connotes a situation in which components come together in such a way that the combination or more complicated arrangement of the components gives rise to the complexity of the whole. Based on this conceptualization of complexity, it is important to note that it is the combination of the multiple aspects of actually taking a medication that gives rise to medication complexity.
For the purpose of the development of the AMCI, medication complexity is defined as a set of interconnected components that, when taken together, make the task of taking medications more difficult. Medication complexity components are dose of medication, medication-taking actions, time, and experience. Complexity increases with (a) the number of doses of medication and the number of associated pills, (b) the number of actions required to take a medication properly, (c) the amount of time necessary to prepare and take the medications (e.g., swallow, inject), and (d) negative experiences associated with the medication (e.g., side effects, pill size). HAART medication complexity as measured by the AMCI form is operationalized as the summation of self-reports of pill burden, medication frequency, administration actions, and severity of side effects. The AMCI form consists of five sections: (a) general information, (b) pill burden, (c) medication frequency, (d) administrative actions, and (e) side effects.
In this section, the medication name, dosage, and number of medications in the total regimen are recorded.
Pill burden is defined as the size of the pill multiplied by the number of pills taken each day. For pill burden, the size of each pill and the number of pills taken each day are recorded. Large pills such as Viracept (625 mg) and ritonavir (100 mg) receive a score of 3, medium pills like Sustiva (Bristol Myers Squibb) (600 mg) and Videx EC (400 mg) receive a score of 2, and small pills like Zerit (Bristol Myers Squibb) (40 mg) and Epivir (GlaxoSmithKline, Pittsburgh, PA) (150 mg) receive a score of 1. The score for each pill is multiplied by the number of those pills taken per day. To weight this measure so that pill burden does not overload the total AMCI score, the pill burden score is divided by 10. For example, the pill burden score for a person taking three Kaletra capsules every 12 hours is 1.2 (six pills per day multiplied by 2, for medium pill ranking, divided by 10). The pill burden score for each pill is summed to yield a total pill burden score for the regimen. To determine the pill burden score for a regimen containing Viracept (625 mg), 2 pills twice daily, and Combivir (GlaxoSmithKline, Pittsburgh, PA) (150 mg/300mg), one pill twice daily, Viracept will receive a 3 because it is large and Combivir will receive a 1 because it is small. The score for a person taking two pills of Viracept (625 mg) twice per day and one pill of Combivir (150 mg/300 mg) twice per day is 3 times 4 divided by 10, or 1.2, for Viracept, and 1 times 2 divided by 10, or .2, for Combivir. The total pill burden score for an individual taking both Viracept and Combivir is 1.4.
In the medication frequency section, the number of times the individual takes medications per day is recorded by checking 1 of 10 frequency options. To determine dose frequency, specific questions are asked regarding how often the patient takes the medication. Because this instrument is from the patient's perspective, the questions asked are, “How often do you take or use this medication?” and, “Is it important that you take it at certain times, or were you told to take it at certain times or are there a certain number of hours between the doses?” The item is scored based on the response to the latter question. Frequency options are based on the traditional medication frequencies used when prescribing medication or treatments (e.g., twice daily, four times daily, or every 12 hours). A score is assigned based on the number of times per day the medicine is taken or the number of hours between doses. Medications that must be taken a certain number of hours apart receive a half point higher score than medications that must be taken a certain number of times per day. For example, a medication taken twice per day receives a score of 2, whereas a medication taken every 12 hours receives a score of 2.5.
The administrative action section is composed of a list of 16 actions that are often required for taking HAART medications. These include items such as taking medications with food, on an empty stomach, or with special liquids; refrigerating medications; and taking additional medications to counteract side effects. An other option is also included. Each instruction or action receives a score based on the number of times per day the action is carried out. For example, if a medication must be taken on an empty stomach with each dose and the dose frequency is twice per day, the person would receive a score of 2—one point for every time it is done per day. If the same medication must be taken at times separate from other medications twice during the day, an additional 2 points are given. If a patient has to take a medication to counteract the side effect of diarrhea once a day, this action receives a score of 1.
The side effect section consists of a list of 24 side effects such as diarrhea, nausea, and fatigue that are commonly associated with HAART regimens. Each side effect is rated on a 4-point scale from 1 (no distress) to 4 (a lot of distress). Patients are first asked if they have the side effect, and if yes, how distressful the side effect is. Because side effects can vary, the patient is asked to estimate the degree of severity during the past 2 weeks. If the participant is unsure which medication the side effect is attributable to, all of the medications he or she feels are the cause are recorded. The responses to individual items are summed to yield a total score.
Administration and Scoring
The AMCI is administered in person by a trained interviewer. One form is completed for each medication in the HAART regimen. The interviewer reads each section to the patient and records responses. For the administrative actions section, the interviewer reads from the list of possible actions and asks questions about the action a person takes for the medication and how often these actions are performed each day. Patients are asked what they currently do on a regular basis to take the medication. Many patients simply follow the directions of the health care provider for taking their medications. Others learn from experience that certain actions help them take their medication and consistently use that action to take their pills. Although the action might not be required to take the medicine, it is recorded because for that patient it is a requirement. To determine side effects, the interviewer asks about side effects related to each medication. Using a list, the participant identifies the side effects and amount of distress each causes.
The scores for each item selected on the AMCI are added to yield a score for the four sections. These scores are summed to yield an overall score for each medication. Finally, total scores can be summed across each form for the medications in the regimen to yield a total regimen score. For each section, higher scores indicate higher levels of self-reported medication complexity.
Data for this study were derived from a pilot study and the main study of an intervention designed to promote adherence to antiretroviral medications. The study was conducted between 2000 and 2004. The study, entitled “Get Busy Living,” enrolled 247 participants who were randomly assigned to the intervention and control groups. Before beginning the main trial, the authors conducted a small pilot study in which 22 participants were enrolled, with 17 participants completing the AMCI at the follow-up assessment. The participants in the pilot study completing the AMCI ranged in age from 26 to 53, with a mean age of 42.4 years. Most participants were male (53%), African American (82%) and single (82%). The 247 participants in the main trial ranged in age from 22 to 61, with a mean age of 41.6 years. Most participants were male (68%), African American (87%), and single (80%).
In addition to these two samples, three nurses employed by the infectious disease clinical site from which participants were recruited and four members of the project staff (three nurses and one health educator) assisted with reliability and validity assessments.
Interrater reliability was assessed by asking the project staff and clinical nurses to rate the complexity of medications using the AMCI. The project staff (three nurses and one health educator) were asked to rate seven medications contained within two case scenarios, and the three nurse clinicians were asked to rate nine medications from three scenarios. The scenarios ranged from relatively simple regimens to highly complex regimens. For example, the patient in one scenario took Combivir and Viracept. One pill of Combivir was taken at 8:00 a.m. and one at 8:00 p.m. Five pills of Viracept were taken at 8:00 a.m. and again at 8:00 p.m. with food. There were no special requirements for taking the Combivir and no side effects. However, the patient experienced diarrhea every day about 1 to 2 hours after taking each dose of Viracept. He altered his diet to try to control the diarrhea with rice and bananas and took diarrhea medication as needed.
Scores for each section of the AMCI and overall scores were computed for each medication for each nurse rater. The interrater reliability for the AMCI was estimated using intraclass correlation coefficients. A reliability estimate was computed for four sections of the AMCI—pill burden, medication frequency, administration actions, and side effects, as well as for the total score. Because the scenarios were different for the project staff and the clinical nurses, separate reliability coefficients were estimated for each group.
Data for the test-retest reliability were collected from participants enrolled in the main study. AMCI forms were completed by participants at all three follow-up assessments. For this analysis, AMCI forms completed at the 3-month and 6-month follow-up assessments were selected. The selection of AMCI forms for review was limited to participants who were on the same medication at both assessments. Eighty-one participants were selected for review. Of these, the same interviewer had completed the AMCI form both times for 23 participants and a different interviewer for 58 participants. A test-retest reliability coefficient (Spearman's rho) was computed for the scores for each section and for the AMCI total score. Separate reliability coefficients were estimated for forms completed by the same interviewer and forms completed by different interviewers.
To assess criterion validity, three nurse clinicians were asked to evaluate the AMCI. The nurse clinicians were given three case scenarios containing nine antiretroviral medications within three regimens. The scenarios included information on pill burden, medication frequency, administrative actions, and side effects. The nurses were given a definition of medication complexity and asked to use their clinical judgment to rank the medications individually from most complex (1) to least complex (9) and to rank the complexity of each regimen from most to least complex. After this initial ranking, the nurse clinicians were given copies of the AMCI forms and asked to complete one for each medication using the same scenarios. Medication and regimen scores were calculated and compared with each nurse's original clinical-based ranking. The rankings of the clinical nurses were then compared using Spearman's rho.
Data from the AMCI were available from 17 participants enrolled in the pilot study. These participants completed the AMCI as part of the follow-up assessments in the study. AMCI scores for each medication and for the total regimen were calculated. Correlation coefficients were calculated to assess the relationship between medication complexity and self-reported adherence. Several aspects of self-reported adherence were measured. These were reasons for missing medications, self-management practices, number of missed medications during the past 4 days, general adherence, and adherence rating over the past 2 weeks and the past 30 days.
Reasons for not taking medications were assessed using a 14-item measure from the University of California, San Francisco, Adherence Questionnaire (Chesney, Ickovics, et al., 2000). For each item, participants were asked to indicate how often within the past 30 days the reason accounted for missing medications. The response options for each item were 0 (never), 1 (rarely), 2 (sometimes), and 3 (often). A total score was computed by summing the responses across the 14 items, resulting in a possible score range of 0 to 42 with lower scores representing fewer reasons for missing medications. The alpha coefficient for this scale was .94.
The HIV Self-Management Scale is a 17-item scale that assesses frequency of use of HIV medication self-management practices. Each item is rated on a 5-point scale ranging from 1 (never) to 5 (always). After reverse coding of negatively worded items, responses to individual items were summed. Higher scores indicate more frequent use of self-management strategies. The alpha coefficient for the scale was .80.
The Antiretroviral General Adherence Scale, adapted from the Antituberculosis General Adherence Scale (McDonnell, Turner, & Weaver, 2001) was used to assess the ease and ability of participants to take HAART according to the health care providers' recommendations during the past 30 days. This scale consists of 5 items, each rated according to six categories ranging from 1 (none of the time) to 6 (all of the time). A total score was computed by summing across the five items after recoding the two negatively worded items resulting in a possible score range of 5 to 30. The alpha coefficient for responses for the current sample of participants was .70.
The final two outcome measures were based on the participants' responses to two items asking about how often in the past 2 weeks and how often in the past 30 days they had missed taking each of their medications. The response options for these two items were 1 (none), 2 (1–2 times), 3 (3–5 times), and 4 (more than 5 times). These items were completed for each medication taken. A mean score was created for each time frame (2 weeks, 30 days) by summing the ratings across all of a participant's medications and dividing by the number of medications. This score provided a general measure of adherence across all medications.
The results of interrater reliability are presented in Table 1. For project staff, the estimates of interrater reliability by section ranged from .72 to 1.0 and from .33 to 1.0 for nurse clinicians. The lowest reliability for both project staff and nurse clinicians was the rating of side effects (.72 and .33 respectively). The estimates of interrater reliability for the total AMCI score was .89 for the project staff and .57 for nurse clinicians.
Because the rankings of one nurse clinician were somewhat different from the other two, interrater reliability coefficients for each section and the total AMCI were computed using two nurses. Although the coefficient for administrative actions was lower than that of staff, the results for the other three components were comparable.
The results of test-retest reliability are presented in Table 2. As shown, the reliability coefficients ranged from .297 to .988 for forms completed by the same interviewer, and for different interviewers, the coefficients ranged from .200 to .922. Side effects and administrative actions showed the least stability over time, whereas number of pills and pill burden showed the greatest stability. The total AMCI correlation coefficient was .791 for the same rater and .766 for different raters.
Criterion validity results are displayed in Table 3. As expected, there were strong relationships between the rankings of the medications by clinical judgment and the rankings of the medications using the AMCI. For all three nurses, the strongest correlations were for pill burden and administrative actions. Although high, the correlation for side effects was somewhat lower when compared to pill burden and administrative actions. For total score, the overall rankings of two of the three nurses showed high correlation.
The results of the evaluation for construct validity are presented in Table 4. As expected, complexity as measured by the AMCI was associated with some of the measures of adherence. Of the 25 correlations, 16 (64%) were .20 and higher. Of these, 7 were statistically significant. The total AMCI score was associated with reasons for not taking medications, self-management, and frequency of missed doses in the past 30 days. The perceived distress of side effects was associated with the same measures of adherence, and administration actions were associated with frequency of missed doses within the past 2 weeks. Interestingly, the relationships between two components of the AMCI—medication frequency and pill burden—and adherence measures were not as strong as those for other components. These are measures that are used most often to assess complexity in research studies.
The purpose of this study was to evaluate the psychometric properties of the AMCI. Assessment of interrater reliability showed a high level of agreement between the ratings of the two project staff, and fairly consistent agreement between the ratings of two of the three nurse clinicians. The component with the least agreement was that of side effects. The rather low correlation for the clinical nurses on side effects is likely because of confusion about how to complete this section. Rather than selecting side effects listed on the scenario, there was a tendency to select all side effects that might be related to each drug. Agreement for other components of the AMCI tended to be greater among project staff than nurses. These results are likely because of the training on the use of the forms that the project staff received. Clinical nurses, in contrast, were simply given written instructions and asked to record their responses. Because of the low correlations for the responses of one nurse, it is possible that she did not completely understand how to complete the forms. Future assessments and use of the AMCI form should include adequate training to ensure an understanding of how to complete the forms. However, the ratings by the other two nurses were quite high, indicating evidence for the use of the AMCI in both research and clinical settings.
Test-retest reliability was assessed by scores on forms completed at two points in time. Results show that stability varies by component with the number of pills per day and pill burden showing the most stability and side effects showing the least. In addition, the variation apparent in medication frequency and pill burden was primarily because of changes in the number of pills taken per day or the number of times per day they were prescribed. Changes in frequency were also a result of recording twice per day doses with a score of 2 (for twice per day dosing) at one time and a score of 2.5 (for every 12 hour dosing) at the other time. Based on this result, training was modified to emphasize the need to determine if medications are taken twice per day versus every 12 hours.
Examination of side effects showed that differences in side effects from the first to second administration were a result of the presence of a side effect at one time but not the other—most often diarrhea, nausea, and fatigue. The perception of side effects and the degree of distress may change over time for several reasons. Acute side effects may become less noticeable as patients adapt to their regimens, and long-term side effects may become more evident the longer a patient is on his or her regimen. Likewise, the degree of distress may change as patients adapt lifestyles or learn to live with them or as newer more worrisome side effects emerge.
With regard to administration actions, differences were most often because of food restrictions and refrigeration of medications recorded at one time but not the other. This result is similar to that noted by DiIorio et al (2003) for the EMTCI. The test-retest coefficients for special instruction and administrative actions sections were lower than other components of the EMTCI instrument. Upon further examination, the investigators found that participants changed their use of the medications with food or antacids, their carrying of the medication, or their taking of the medications at separate times from other medications. Reasons for these different results for the EMTCI could be because of an actual change in the way the person took the medication, error on the part of the interviewer in recording the information, or a combination of factors. The same is true for the AMCI, in which people may change the way they take medications over time. Also, actual side effects or the distress they cause may lesson or increase overall with time and are expected to be variable. Because of the low reliability for administrative actions and side effects, it is recommended at this time not to use these components of the AMCI to assess change related to interventions.
Criterion validity was assessed by comparing the complexity rankings of expert nurses' clinical judgment to their numerical score ratings of medications using the AMCI forms. The overall correlation of .888 to .940 for two clinical nurses show relatively high association between the two methods of rating medication complexity for pill burden, special administrations sections, and the total AMCI scores.
Finally, the analysis conducted to assess construct validity demonstrates that the scores for the individual sections and the overall AMCI scores correlate in the expected direction with measures of adherence. To assess construct validity, the relationship between AMCI scores and five measures of medication adherence was examined. All of the correlations were in the predicted direction, and most were over .20. The presence and degree of distress associated with side effects showed the strongest correlation with three of the five adherence measures. Medication frequency and administrative actions were associated (> .20) with three measures of adherence. Although these results also show that people who take medications more frequently are likely to skip doses and display a lower tendency to adhere, the correlations are not as strong as the correlations with other aspects of the AMCI. One reason might be the current trend to use medications with twice per day dosing and regimens with few medications, thus limiting the variability in medication frequency across participants. This lack of variability in medication frequency could account for the low correlations. Overall, the psychometric tests provide evidence to partially support reliability and validity of the AMCI measure of medication complexity.
Based on the results of this study, some areas were noted in which improvements could be made. The authors' research shows that it is important for nurses to have a good understanding of the components of the AMCI and its scoring. Specific instructions were developed on how to complete the form and recommend training and assessment of interrater reliability when the AMCI is used for research studies. Participants who were taking several medications often found it difficult to identify the specific medication that caused a certain side effect such as diarrhea. Sometimes participants were taking more than one medication that could cause this symptom. Because each medication is evaluated separately, such a side effect would be rated more than once, thus inflating scores within a regimen. The use of a regimen-related measure of side effects that would reduce redundant side effect ratings is being considered.
Because of the changing nature of the side effects and administrative actions, it is premature to use these aspects of the AMCI in research studies assessing change over time. One exception may be when the focus of the intervention is to reduce side effects or to encourage consistent adherence to food restrictions and proper management of medications (e.g., refrigeration). The current focus of the AMCI is on pill management. Because antiretroviral medications come in different forms including liquids and injections, modifications to include these forms of administration are being considered.
This work focused on the development of the AMCI for research. Because clinical nurses are involved in counseling patients about medication adherence, they might find it to be an objective tool for assessing the complexity of regimens. This information might be useful in identifying those patients who need more counseling and support. Further work is necessary to evaluate the usefulness of the tool in the clinical setting.
This report details initial reliability and validity assessment for the AMCI. Some samples including those for test-retest and criterion related validity were small, and the clinicians included in the analysis were nurses working in an HIV clinic. Additional work is necessary to further test the AMCI with larger sample sizes and among other groups of clinicians working in more diverse settings.
The purpose of this study was to assess the psychometric properties of the AMCI. The AMCI, which consists of four components, was developed to evaluate the complexity of antiretroviral medications and regimens. Several psychometric tests were conducted to assess interrater and test-retest reliability and criterion and construct validity. The results of these tests provide initial evidence for both the reliability and validity of the AMCI when used for research purposes. Additional work is necessary to modify the tool and to assess its usefulness in clinical settings.
Funded by a grant from the National Institute of Nursing Research R01 NR0485701A1.
Chesney, M., Morin, M., & Sherr, L. (2000). Adherence to HIV combination therapy. Social Science and Medicine, 50
Chesney, M. A., Ickovics, J. R., Chambers, D. B., Gifford, A. L., Neidig, J., & Zwickl, B., et al. (2000). Self-reported adherence to antiretroviral medications
among participants in HIV clinical trials: The AACTG adherence instruments. Patient Care Committee & Adherence Working Group of the Outcomes Committee of the Adult AIDS Clinical Trials Group (AACTG). AIDS Care, 12
Conn, V. S., Taylor, S. G., & Kelley, S. (1991). Medication regimen complexity and adherence among older adults. Image, 23
DiIorio, C., Yeager, K., Shafer, P. O., Letz, R., Henry, T., & Schomer, D. L., et al. (2003). The epilepsy medication and treatment complexity index: Reliability and validity testing. Journal of Neuroscience Nursing, 35
Eldred, L. J., Wu, A. W., Chaisson, R. E., & Moore, R. D. (1998). Adherence to antiretroviral and pneumocystis prophylaxis in HIV disease. Journal of Acquired Immune Deficiency Syndromes & Human Retrovirology, 18
Hinkin, C. H., Castellon, S. A., Durvasula, R. S., Hardy, D. J., Lam, M. N., & Mason, K. I., et al. (2002). Medication adherence among HIV+ adults: Effects of cognitive dysfunction and regimen complexity. Neurology, 59
Ickovics, J. R., Cameron, A., Zackin, R., Bassett, R., Chesney, M., & Johnson, V. A., et al. (2002). Consequences and determinants of adherence to antiretroviral medication: results from Adult AIDS Clinical Trials Group protocol 370. Antiviral Therapy, 7
Kelley, S. O. (1988). Measurement of the complexity of medication regimens of the elderly. Unpublished master's thesis
. Columbia: University of Missouri.
McDonnell, M., Turner, J., & Weaver, M. T. (2001). Antecedents of adherence to antituberculosis therapy. Public Health Nursing, 18
Mehta, S., Moore, R. D., & Graham, N. M. (1997). Potential factors affecting adherence with HIV therapy. AIDS, 11
Mikulecky, D. C. (2001). The emergence of complexity: Science coming of age or science growing old? Computers and Chemistry 25:341–348.
Miller, N. H. (1997). Compliance with treatment regimens in chronic asymptomatic diseases. American Journal of Medicine, 102
Paterson, D. L., Swindells, S., Mohr, J., Brester, M., Vergis, E. N., & Squier, C., et al. (2000). Adherence to protease inhibitor therapy and outcomes in patients with HIV infection. Annals of Internal Medicine, 133
Stone, V. E., Hogan, J. W., Schuman, P., Rompalo, A. M., Howard, A. A., & Korkontzelou, C., et al. (2001). Antiretroviral regimen complexity, self-reported adherence, and HIV patients' understanding of their regimens: Survey of women in the HER study. Journal of Acquired Immune Deficiency Syndromes, 28
Stone, V. E., Jordan, J., Tolson, J., Miller, R., & Pilon, T. (2004). Perspectives on adherence and simplicity for HIV-infected patients on antiretroviral therapy: Self report of relative importance of multiple attributes of highly active antiretroviral therapy regimes in predicting adherence. Journal of Acquired Immune Deficiency Syndromes, 36
U.S. Department of Health and Human Services. (2004). A guideline for the use of ARV in HIV-1 infected adults and adolescent. HIV/AIDS Treatment Information Service Online
. Retrieved October 29, 2004, from http://aidsinfo.nih.gov/guidelines