Amico, K. Rivet PhD*; Fisher, William A. PhD*,†; Cornman, Deborah H. PhD*; Shuper, Paul A. PhD*; Redding, Caroline G. MA*; Konkle-Parker, Deborah J. PhD, FNP‡; Barta, William PhD*; Fisher, Jeffrey D. PhD*
Visual analog scales (VASs) have been extensively used for assessment in a number of health domains (e.g., acute and chronic pain) and have recently been applied to assessment of antiretroviral therapy (ART) medication adherence.1-3 The traditional VAS presents patients with an ordinal scale, where the smallest number represents the least amount of the experience in question (eg, pain) and the highest number represents the most intense amount. When assessing acute pain, these numbers may also include labels or, in the case of pediatric use, pictures depicting relative amounts or degrees of pain or discomfort. The VAS has accumulated substantial support as a valid and efficient tool for use in assessment of pain in regular care and emergency clinics.4-6
Three published works support the validity of VAS assessments of ART adherence.1-3 The VAS for ART medication adherence appears as an ordinal scale representing the percent of medication taken relative to that which has been prescribed, for a given medication, during a 3- to 4-week period. Patients are presented with a line anchored at 0% and 100%; provided with examples of what 0, 50%, and 100% adherence would represent; and asked to assess their own ART medication adherence. Walsh et al3 included this type of VAS adherence assessment in the Medication Self-report Inventory, and when used to assess adherence to a single antiretroviral medication, VAS scores correlated significantly with Medication Events Monitoring (MEMS) data, pill counts, and viral load data. Despite the relatively small sample size (N = 78), these results were generally replicated in the work of Oyugi et al2 with another small sample of 34 ART-naive patients in Uganda starting treatment with a fixed dose combination medication (stavudine + lamivudine + nevirapine) at (1 pill twice daily). The VAS was used to assess adherence during the last 30 days and correlated well with a 3-day previously validated but much more complex AACTG measure,7 MEMS data, unannounced pill counts, and patient viral load. Finally, Giordano et al1 reported substantial and consistent associations between VAS 30-day ART adherence scores and unannounced pill counts and viral load (0.76 and −0.49, respectively) in a sample of 84 predominantly gay, HIV-positive men in San Francisco.
Across each of these studies, results have been supportive of the use of the VAS in assessment of ART adherence, but VAS scores, unlike AACTG measures (e.g., see References 7 and 8), have not been evaluated in relation to their association with specific barriers to ART medication adherence that patients may experience, which would be an important step toward establishing its construct validity. Using the VAS adherence assessment as an efficient brief measure of adherence would be valuable to the extent that it not only identifies levels of ART adherence, which has been established thus far, but that it could also identify patients who are struggling with adherence-related barriers. If poor or suboptimal adherence, according to VAS ratings, is associated with the presence of barriers to ART adherence, the construct validity of this measure would be supported, and its ease of use, low literacy requirements, and straightforward scoring might make it an optimal choice for self-reported adherence assessment within the clinical setting.
The current study assessed the concordance of adherence estimates generated from the 3-day AACTG7 with those generated from a 30-day VAS measure of ART adherence for each medication in a patient's regimen in a sample of HIV-infected patients receiving care in a southeastern US HIV specialty clinic. The association between VAS and AACTG scores and specific self-reported barriers to adherence was also evaluated.
HIV-infected patients attending a large southeastern US HIV specialty care clinic were recruited to complete an anonymous computer-administered survey that included a modified AACTG,7 VAS measures of ART adherence,3 and items assessing information, motivation, and behavior skills-related barriers to ART adherence. All procedures, consent forms, and materials were institutional review board approved, and participants were compensated US$15 for their participation.
The computer-assisted survey that addressed ART adherence included a self-report of ART regimen prescribed, number of doses taken as prescribed during the last 3 days for each ART medication (modified AACTG), and a VAS for each ART medication. First, the participant selected the ART medication(s) he or she was prescribed by clicking on its picture from an on-screen pill chart. For each ART medication selected, the user further indicated the dose times per day and number of pills, liquid doses, or injections prescribed for each dose. For each medication reported, the user then completed the computer-delivered AACTG and VAS, presented in random order. The AACTG was modified from its original format to have participants report, for each medication's prescribed dose during the previous 3 days, whether that dose was taken in full, more than the prescribed dose was taken, or less than the prescribed dose was taken. The VAS item asked participants to mark the point on a line ranging from 0% to 100% that represented "How much of this medication have you taken in the last 3 to 4 weeks?" Consistent with its application in previous research,3 participants could mark either anchor on the VAS line or on any 10% increment in between, were instructed on the meaning of each end point and the midpoint, and were given a statement regarding the normalcy of missing doses from time to time. The modified AACTG and VAS items were presented in random order to control for sensitization or order effects. Self-reported adherence from the AACTG was computed as number of doses taken in full as prescribed over number prescribed for each ART medication, and as a total adherence score obtained by collapsing across all ART medications in the patient's regimen and computing total doses taken in full as prescribed during the 3-day period over total doses prescribed.7 Adherence during the last 3 to 4 weeks from the VAS for each ART medication was also collapsed to create an overall adherence score by adding all percentages and dividing by the number of ART medications reported. Optimal adherence and suboptimal adherence were defined as 90% or greater and lower than 90%, respectively.9,10 Self-reported barriers to adherence to therapy were scored to represent the presence or absence of 1 of 18 different barriers to adherence. Barriers to adherence were assessed using a measure based on the Information Motivation Behavioral Skills model of ART adherence.11 Each item has a critical response region on a Likert-type response scale, and items are combined to create several critical areas, with each area representing a potential barrier reflective of deficits in adherence-related information, motivation, or behavioral skills.
One hundred fifty-one participants completed the computer-assisted survey in the summer of 2005. Of these, 147 provided complete AACTG and VAS score data. Eighty-six percent of the sample was African American, 42% were women, and 60% were heterosexual. Approximately a quarter (27%) of the sample was employed part time or full time, and 76% reported annual family incomes of US$10,000 or less. The predominant self-reported route of HIV acquisition was heterosexual (41%), followed by male-to-male (25%) sexual contact. Twenty-three percent of the patients were uncertain of route of infection. More than half (54%) of the sample had been living with HIV for 8 years or more, and all were currently prescribed between 1 and 5 medications, with some medications representing fixed-dose combinations of 2 to 3 ART medications. The most frequently reported ART medication was efavirenz (40% of the sample), followed by lamivudine + zidovudine (35%), and atazanavir (21%).
Concordance Between VAS and AACTG
Based on scores from the modified AACTG, the average adherence rate during the preceding 3 days was 81% (SD, 0.27; median, 100%) across all ART medications and ranged from 0% to 100%. Averaged VAS score across ART medications for adherence during the last 3 to 4 weeks was 87% (SD, 0.21; median, 98%; range, 0%-100%). Adherence rates estimated by the AACTG and VAS measures correlated (r = 0.585, P < 0.0001), although the mean of ranked adherence scores and SDs produced by the measures significantly differed (z = −3.331, P = 0.001; and F= 1.6782, P = 0.002, respectively). Ranked distributions indicated that, for 50 participants (34%), adherence rates reported on the VAS and AACTG were perfectly matched, whereas 61 participants (42%) reported higher adherence rates on the VAS, and 36 participants (24%) reported higher adherence rates on the AACTG. For each ART medication reported by a participant, we calculated a discrepancy score between adherence estimated by the 2 measures (adherence3-day AACTG − adherence30-day VAS) and averaged these across each participant's regimen. Absolute discrepancies averaged 0.14 (or 14 percentage points), whereas raw discrepancy scores averaged −0.06 (6%), indicating that, on average, VAS scores were somewhat higher than AACTG scores. Raw discrepancy scores and rates of adherence based on the AACTG and VAS measures were not significantly associated with number of ART medication pills per day, number of doses per day, years since HIV diagnosis, or gender. Education level did, however, appear to be associated with adherence levels estimated by the AACTG, such that those with educations beyond high school tended to report higher adherence on the AACTG in comparison to those with lower levels of education (r = 0.18, P < 0.05). Education levels were not, however, associated with adherence estimated by the VAS.
Using the cutoff of 90% adherence, participants were classified as optimally (≥90%) adherent (OA) or suboptimally (<90%) adherent (SA) to their ART regimen. As shown in Table 1, 66% of patients were identically classified as OA or SA by both measures, and 34% had discordant classifications.
Self-Reported Adherence Barriers
The self-reported barriers to ART adherence of patients whose VAS scores placed them in categories of optimal and suboptimal adherence were evaluated. To explore the performance of VAS scores in identifying patients with barriers to adherence in relation to the AACTG, this strategy was repeated for those classified as OA and SA according to the AACTG. As can be seen in Table 2, patients classified as SA by either measure appeared to generally report ART adherence barriers with greater frequency than OA patients. On average, participants reported 8 barriers or problems in critical areas (SD, 4.24; median, 8), whereas SA-classified patients by either adherence measure reported more barriers in comparison to their OA cohort. Specifically, VAS-classified SA patients reported significantly more deficits than VAS-classified OA patients (t145 = 5.84, P < 0.001, M = 11.42 vs 7.20), and AACTG-classified SA patients reported significantly more deficits than AACTG-classified OA patients (t145 = 3.17, P = 0.002, M = 9.48 vs 7.32). Over the 18 different ART adherence barriers assessed, VAS-classified SA patients reported problems in 13 barriers significantly more than VAS-classified OA patients. In contrast, significantly more AACTG-classified SA patients than AACTG-classified OA patients reported problems in only 6 barrier areas.
Adherence rates in this sample of patients on ART, as estimated by the modified AACTG or VAS adherence scores, were relatively high, although well within the range of adherence values reported in recent research with the VAS (see References 1, 2, and 3). The mean discrepancy score between the modified AACTG and VAS measures used in this research was small (−6%), and the measures identically classified two thirds (66%) of participants as optimally or suboptimally adherent. In general, the VAS appeared to produce higher estimates of adherence, with fully 61 participants (42% of the sample) reporting higher rates of ART adherence on this measure in comparison to their AACTG reports. It is possible that the VAS may be more prone to favorable estimation than reports of adherence that are anchored in recall of very recent adherence events-such as during the preceding 3 days. Alternatively, the coarseness of the VAS scale in comparison to the AACTG may have resulted in higher VAS estimates of adherence. Moreover, the differences in the time frames used by each measure may have contributed to the discordance in their estimates of adherence. Future research may provide greater details regarding the potential causes of differences in ART adherence rates generated by the VAS and AACTG by allowing for greater sensitivity in VAS response options and using identical time frames for each. It is also important for future research to explore the extent to which discrepancies in the adherence estimates generated by the VAS and AACTG relate to discrepancies in other measures or biologic indices of adherence.
The concordance rates between the VAS and AACTG adherence scores and correlation between these 2 measures found in the current study are somewhat lower than what have been found in previous research,1-3 which may be due to differences in study design, the format of the computer-delivered VAS, or a peculiarity of the current sample. The current study presented the VAS and AACTG measures in random order, as opposed to a set order where the AACTG is followed by the VAS, which may have affected the results. Our study was also unique in that the adherence measures were presented via a computer software program. In either case, replication of the present results using computer delivery of the VAS and AACTG measures presented in random or counterbalanced order is needed.
In terms of the relation of VAS scores with other study variables, the associations reported in the current research were consistent with the general literature8 and support the divergent construct validity of the VAS self-report measure of adherence, in that adherence estimated by the VAS was not related to sex, education level, or years of HIV diagnosis. Moreover, a substantial number of self-reported barriers to adherence differentiated patients classified as OA versus SA on the VAS, and the VAS appeared to differentiate OA and SA individuals across more barriers in this respect compared with a classification based on AACTG scores. Bearing in mind that VAS scores may produce an overestimation of adherence or underestimation of nonadherence, VAS scores did appear to identify SA patients who reported numerous adherence-related barriers, which does provide support for this measure's construct validity. On the basis of the VAS categorization, significantly more SA than OA patients reported lacking ART regimen general information and information specific to the mechanics of ART medications, challenges in integrating one's ART regimen into one's daily life, difficulties with effectively communicating with one's health care provider regarding one's medications, trouble getting support from others for taking medications, problems getting accurate HIV treatment and medication information, difficulty getting refills of medications on time, problems with pill size or taste, difficulty taking medications when using alcohol or street drugs, missing doses when experiencing strong negative emotions, difficulty taking doses when feeling healthy, difficulty taking doses when feeling unhealthy or sick, and difficulty remembering to take one's medication. The behavior of the VAS in relation to barriers strongly suggests that using the VAS to classify patients as OA and SA may prove to be clinically useful. Specifically, the VAS is a brief and practical measure that does appear to identify SA patients facing a substantial number of adherence-related obstacles.
Although identifying those with barriers to ART adherence is clinically important and speaks to convergent construct validity, self-report measures of adherence must also demonstrate a strong relation with biologic indices or other proxies of adherence. Previous research has supported the validity of the VAS in relation to viral load, MEMS data, and unannounced pill counts1-3 (and, similarly, the AACTG has an extensive body of support for its associations with disease progress and levels of antiretroviral medication blood concentration12-15), but the ultimate utility of VAS measures of ART adherence rests upon more rigorous evaluations of its concurrent and predictive validity with indices of disease progression.
The current results suggest that using the VAS to identify those who may benefit from additional assistance or intervention with their ART adherence may be an effective strategy. In particular, the VAS format may provide a rapid measure that patients can complete before medical appointments and may be used as a tool to identify patients who could benefit from a more detailed evaluation of adherence barriers from their medical care providers. It is important to note, however, that the current results were obtained through an anonymous survey, and whether patients may alter their response to the VAS items if they are aware that their providers will review their responses remains to be investigated. Future research using the VAS measure in clinic settings can further establish the utility of this measure as a quick screening device of ART adherence and can seek to calibrate it (perhaps by adding an empirically derived constant) with objective measures of ART adherence such as MEMS assessments. Given that many clinics currently lack a viable brief method to track and quantify patient adherence,16 the VAS may be a critical addition to providing patients with the adherence support they need.
The authors thank Jo Ann Lewis for her assistance with data collection and the staff and patients at the Adult Special Care Clinic at the University of Mississippi Medical Center.
1. Giordano TP, Guzman D, Clark R, et al. Measuring adherence to antiretroviral therapy in a diverse population using a visual analogue scale. HIV Clin Trials. 2004;5(2):74-79.
2. Oyugi JH, Byakika-Tusiime J, Pharm B, et al. Multiple validated measures of adherence indicate high levels of adherence to generic HIV antiretroviral therapy in a resource-limited setting. J Acquir Immune Defic Syndr. 2004;36:1100-1102.
3. Walsh JC, Mandalia S, Gazzard BG. Responses to a 1 month self-report on adherence to antiretroviral therapy are consistent with electronic data and virological treatment outcome. AIDS. 2002;16:269-277.
4. Coll AM, Jamal RM, Mead D. Postoperative pain assessment tools in day surgery: literature review. J Adv Nurs. 2004;46(2):124-133.
5. Ducharme J, Barber C. A prospective blinded study of emergency pain assessment and therapy. J Emerg Med. 1995;13:571-575.
6. Howard RF. Current status of pain management in children. JAMA. 2003;290:2464-2469.
7. Chesney MA, Ickovics JR, Chambers DB, et al. Self-reported adherence to antiretroviral medications among participants in HIV clinical trials: the AACTG adherence instruments. AIDS Care. 2000;12:255-266.
8. Chesney MA, Morin M, Sherr L. Adherence to HIV combination therapy. Soc Sci Med. 2000;50:1599-1605.
9. Bangsberg DR, Moss AR, Deeks SG. Paradoxes of adherence and drug resistance to HIV antiretroviral therapy. J Antimicrob Chemother. 2004;53:696-699.
10. Sethi AK, Celentano DD, Gange SJ, et al. Association between adherence to antiretroviral therapy and human immunodeficiency virus drug resistance. Clin Infect Dis. 2003;37:1112-1118.
11. Fisher JD, Fisher WA, Amico KR, et al. An information-motivation-behavioral skills model of adherence to antiretroviral therapy. Health Psychol. In press.
12. Kleeberger CA, Phair JP, Strathdee SA, et al. Effect of computer-assisted self-interviews on reporting of sexual HIV risk behaviours in a general populations sample: a methodological experiment. J Acquir Immune Defic Syndr. 2001;26:82-92.
13. Knobel H, Alonso J, Casado JL, et al. Validation of a simplified medication adherence questionnaire in a large cohort of HIV-infected patients: the GEEMA study. AIDS. 2002;16:605-613.
14. Moatti JP, Spire B. Living with HIV/AIDS and adherence to antiretroviral treatments. In: Moatti J-P, Souteyrand Y, Prieur A, et al eds. AIDS in Europe: New Challenges for the Social Sciences. New York: Routledge, 2002:57-73.
15. Nieuwkerk PT, Sprangers MA, Burger DM, et al. Limited patient adherence to highly active antiretroviral therapy for HIV-1 infection in an observational cohort study. Arch Intern Med. 2001;161:1962-1968.
16. Harman JJ, Amico KR, Johnson BT. Standard of care: promoting antiretroviral adherence in clinical care. AIDS Care. 2005;2:237-251.
© 2006 Lippincott Williams & Wilkins, Inc.